PRIVATE TUTORING AND ITS IMPACT ON STUDENTS’ ACADEMIC

ACHIEVEMENT, FORMAL SCHOOLING, AND EDUCATIONAL INEQUALITY

IN KOREA

Ji Yun Lee

Submitted in partial fulfillment of the requirements for

the degree of Doctor of Philosophy
under the Executive Committee of the Graduate School of Arts and Sciences

COLUMBIA UNIVERSITY

2013
 © 2013
Ji Yun Lee
All rights reserved
 ABSTRACT

PRIVATE TUTORING AND ITS IMPACT ON STUDENTS’ ACADEMIC

ACHIEVEMENT, FORMAL SCHOOLING, AND EDUCATIONAL INEQUALITY
IN KOREA

JI YUN LEE

Over the last two decades, private tutoring has emerged as an important issue in

education as its demand has been growing around the world. However, the evidence of the

effectiveness of private tutoring is still mixed. Using the Korean Educational Longitudinal

Survey, this dissertation explores the causal impact of private tutoring in Korea on three

outcomes: students’ academic achievement, the quality of the learning environment in formal

schooling, and educational inequality.

The first set of empirical analyses explores how private tutoring in secondary schools

affects students’ academic achievement in both short-term and long-term aspects using Ordinary

Least Squares, Instrumental Variable, and Propensity Score Matching methods. The results

suggest that private tutoring in middle school, on average, has positive short-term effects on

students’ academic achievement in middle school, but minimal long-term effects on the

university entrance examination scores. By subject area, English and math tutoring are effective

in improving academic achievement in middle school, whereas verbal tutoring is not. Moreover,

private tutoring in grade 7 is most beneficial for students in middle school. In terms of private

tutoring in high school, only math tutoring is beneficial for high school students in improving

scores on the university entrance examination.

 The second set of the analyses employs Ordered Logit, Propensity Score Matching, and

Difference-in-Differences methods to estimate the impact of private tutoring on the quality of

formal schooling. The quality of the learning environment in formal schooling is measured by

students’ attention to lessons in class. On average, private tutoring shows a positive influence on

students’ attention to lessons in grade 8, but the magnitude of its influence is not substantial.

However, when differentiating the results by ability group, positive effects are detected mostly in

the low-ability group, which means that low achievers pay more attention to lessons in schools if

they participate in private tutoring. These results imply that private tutoring improves the overall

learning environment in formal schooling, which in turn increases the overall quality of

schooling.

The third set of the analyses uses Quantile Regression, Two-Stage Least Absolute

Deviation estimator, and Propensity Score Subclassification to estimate the heterogeneous

effects of private tutoring between ability groups, which provides implications on educational

inequality based on academic achievement. The overall results suggest that private tutoring in

middle school exacerbates educational inequality between high and low achievers, which implies

a widening of the achievement gap. In addition, enrolling in tutoring at an earlier grade level

results in greater heterogeneity between high and low achievers in academic performance than

enrolling in tutoring during later grade levels. However, private tutoring in high school

contributes to reducing the achievement gap; low achievers benefit more from private tutoring in

high school compared to high achievers. Moreover, three years of cumulative math tutoring and

receiving a single year of math tutoring in grade 12 contribute to narrowing the achievement gap

between low and high achievers in the university entrance examination scores.
 TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................................... v

LIST OF FIGURES ....................................................................................................................... ix

ACKNOWLEDGEMENTS ............................................................................................................ x

CHAPTER I: INTRODUCTION .................................................................................................... 1

1.1 Background and Motivation ................................................................................................. 1

1.2 Definition of Private Tutoring .............................................................................................. 8

1.3 Research Questions ............................................................................................................. 10

1.4 Endogeneity of Private Tutoring ......................................................................................... 10

1.5 Structure of the Study ......................................................................................................... 12

CHAPTER II: LITERATURE REVIEW ..................................................................................... 14

2.1 Theoretical Background ...................................................................................................... 14

2.1.1 Human Capital Theory................................................................................................. 14

2.1.2 The Standard Microeconomic Theory of Supply and Demand ................................... 17

2.1.3 Educational Production Function ................................................................................. 20

2.2 Determinants of Private Tutoring ....................................................................................... 21

2.2.1 Micro factors ................................................................................................................ 21

2.2.2 Macro factors ............................................................................................................... 23

2.3 Empirical Literature on the Effect of Private Tutoring on Outcomes ................................ 26

2.4 Gaps in the Existing Empirical Literature .......................................................................... 34

CHAPTER III: OVERVIEW OF KOREAN PRIVATE TUTORING ......................................... 37

3.1 Purposes .............................................................................................................................. 37

3.2 Intensity............................................................................................................................... 40

i
 3.3 Providers ............................................................................................................................. 41

3.3.1 Forms of Private Tutoring ............................................................................................ 41

3.3.2 Number of Providers .................................................................................................... 46

3.3.3 Size of Private Tutoring Market .................................................................................. 47

3.3.4 Annual Salary of Tutors ............................................................................................... 48

3.4 Contents of Private Tutoring ............................................................................................... 49

3.5 Expenditure on Private Tutoring ......................................................................................... 51

3.5.1 Total Expenditure......................................................................................................... 51

3.5.2 Expenditure by Region ................................................................................................ 53

3.5.3 Monthly Expenditure by Income Group ...................................................................... 54

CHAPTER IV: THE IMPACT ON STUDENTS’ ACADEMIC ACHIEVEMENT ................... 56

4.1 Data Description ................................................................................................................. 58

4.2 Estimation Models .............................................................................................................. 60

4.2.1 Ordinary Least Square ................................................................................................. 60

4.2.3 Instrumental Variables ................................................................................................. 62

4.2.4 Propensity Score Matching .......................................................................................... 63

4.3 Variables ............................................................................................................................. 66

4.4 Descriptive Statistics of the Sample ................................................................................... 71

4.5 Validity Check for Instrumental Variable........................................................................... 81

4.6 Empirical Results ................................................................................................................ 85

Private Tutoring in Middle School: Short-term Effects ................................................ 86

Private Tutoring in Middle School: Long-term Effects .............................................. 101

Private Tutoring in High School: Short-term Effects ................................................. 106

ii
 4.7 Discussion ......................................................................................................................... 111

CHAPTER V. THE IMPACT ON FORMAL SCHOOLING .................................................... 117

5.1 Estimation Models ............................................................................................................ 118

5.1.1 Propensity Score Matching ........................................................................................ 119

5.1.2 Difference-in-Differences Estimation ........................................................................ 122

5.2 Variables ........................................................................................................................... 124

5.3 Estimation and Results...................................................................................................... 126

5.3.1 Descriptive Statistics .................................................................................................. 126

5.3.2 Diagnostics for Propensity Score Matching .............................................................. 131

5.3.3 Empirical Results ....................................................................................................... 132

Effect of Regulation on Private Tutoring.................................................................... 137

5.4 Discussion ......................................................................................................................... 139

CHAPTER VI. THE IMPACT ON EDUCATIONAL INEQUALITY ..................................... 143

6.1 Estimation Models ............................................................................................................ 145

6.2 Variables ........................................................................................................................... 148

6.3 Empirical Results .............................................................................................................. 149

6.3.1 Private Tutoring in Middle School: Short-term Heterogeneous Effects .................... 149

6.3.2 Private Tutoring in Middle School: Long-term Heterogeneous Effects .................... 173

6.3.3 Private Tutoring in High School: Short-term Heterogeneous Effects ....................... 180

6.4 Discussion ......................................................................................................................... 186

CHAPTER VII: CONCLUSION ................................................................................................ 190

7.1 Summary of Findings ........................................................................................................ 190

7.2 Policy Responses .............................................................................................................. 193

iii
 7.2.1 History of policies on private tutoring in Korea ........................................................ 193

7.2.2 Suggestions to Policy Makers in Education............................................................... 194

7.3 Limitations of the Study and Directions for Future Research .......................................... 198

REFERENCE .............................................................................................................................. 201

APPENDIX ................................................................................................................................. 211

iv
 LIST OF TABLES

Table 1. The scale of private tutoring in selected countries ........................................................... 4

Table 2. Participation rate and monthly expenditure on private tutoring by academic percentile 38

Table 3. Participation rate by education level in Korea ................................................................ 41

Table 4. Participation rate in each type of private tutoring by school level in Korea .................. 45

Table 5. Participation rate in each type of private tutoring by school level in Japan ................... 45

Table 6. Number of private tutoring providers by type in Korea ................................................. 46

Table 7. Annual gross sales of private tutoring by type and education level in 2008 .................. 47

Table 8. Average annual salary by type of tutors in Korea .......................................................... 48

Table 9. Private tutoring participation rate by subject (covered in formal schooling) and by
education level in Korea in 2010 .......................................................................................... 49

Table 10. Private tutoring participation rate by subject (not covered in formal schooling) and
education level in Korea in 2010 .......................................................................................... 50

Table 11. Total household's expenditure on private tutoring ........................................................ 51

Table 12. Total, annual, and monthly expenditure on private tutoring and expenditure by school
level in 2007 .......................................................................................................................... 52

Table 13. Household expenditure in primary and secondary education, 1985-2006.................... 53

Table 14. Monthly expenditure on private tutoring per student by size of region........................ 53

Table 15. Monthly private tutoring expenditure per household by income group ....................... 54

Table 16. Descriptive statistics of control variables ..................................................................... 73

Table 17. Descriptive statistics of dependent variables ................................................................ 76

Table 18. Descriptive statistics of variables related to private tutoring ........................................ 77

Table 19. Correlation coefficients between the instrument and covariates .................................. 82

Table 20. The first-stage equation in 2SLS .................................................................................. 83

Table 21. Effects of one year of private tutoring on academic achievement in grade 9 ............... 87

v
Table 22. Effects of two years of private tutoring on academic achievement in grade 9 ............. 87

Table 23. Effects of three years of private tutoring on academic achievement in grade 9 ........... 88

Table 24. Effects of one year of verbal tutoring on verbal achievement in grade 9 ..................... 90

Table 25. Effects of two years of verbal tutoring on verbal achievement in grade 9 ................... 90

Table 26. Effects of three years of verbal tutoring on verbal achievement in grade 9 ................. 90

Table 27. Effects of one year of English tutoring on English achievement in grade 9 ................ 91

Table 28. Effects of two years of English tutoring on English achievement in grade 9 ............... 92

Table 29. Effects of three years of English tutoring on English achievement in grade 9 ............. 92

Table 30. Effects of one year of math tutoring on math achievement in grade 9 ......................... 93

Table 31. Effects of two years of math tutoring on math achievement in grade 9 ....................... 94

Table 32. Effects of three years of math tutoring on math achievement in grade 9 ..................... 94

Table 33. Effects of private tutoring in grade 7 on academic achievement for 3 years ................ 96

Table 34. Effects of private tutoring in grade 8 on academic achievement for 2 years ................ 97

Table 35. Effects of private tutoring in grade 9 on academic achievement in grade 9 ................. 97

Table 36. Summary of the estimates in single-year effects by academic subject and grade ...... 100

Table 37. Long-term effects of private tutoring in middle school on the CSAT ........................ 102

Table 38. Long-term effects of verbal tutoring in middle school on the CSAT verbal achievement
............................................................................................................................................. 103

Table 39. Long-term effects of English tutoring in middle school on the CSAT English
achievement ........................................................................................................................ 104

Table 40. Long-term effects of math tutoring in middle school on the CSAT math achievement
............................................................................................................................................. 105

Table 41. Cumulative and single-year effects of private tutoring in high school on the CSAT
achievement ........................................................................................................................ 107

Table 42. Cumulative and single-year effects of verbal tutoring in high school on the CSAT
verbal achievement ............................................................................................................. 108

Table 43. Cumulative and single-year effects of English tutoring in high school on the CSAT
English achievement ........................................................................................................... 109

vi
Table 44. Cumulative and single-year effects of math tutoring in high school on the CSAT math
achievement ........................................................................................................................ 110

Table 45. Descriptive statistics of control variables ................................................................... 128

Table 46. Descriptive statistics of dependent variables .............................................................. 130

Table 47. Summary of estimates of the average effect of private tutoring on students' attention to
lessons in formal schools .................................................................................................... 133

Table 48. Heterogeneous effects of private tutoring on students' attention to lessons in formal
schools................................................................................................................................. 136

Table 49. Difference-in-differences estimation: effect of the regulation on private tutoring on

students' attention to lessons in formal schools .................................................................. 139

Table 50. Inequality in the amount of private tutoring in middle school by quantile group ...... 144

Table 51. Heterogeneous effects of 1 year of private tutoring in middle school on overall
academic achievement in grade 9 ....................................................................................... 149

Table 52. Heterogeneous effects of 2 years of private tutoring in middle school on overall
academic achievement in grade 9 ....................................................................................... 150

Table 53. Heterogeneous effects of 3 years of private tutoring in middle school on overall
academic achievement in grade 9 ....................................................................................... 151

Table 54. Heterogeneous effects of 1 year of English tutoring in middle school on English
achievement in grade 9 ....................................................................................................... 154

Table 55. Heterogeneous effects of 2 years of English tutoring in middle school on English
achievement in grade 9 ....................................................................................................... 155

Table 56. Heterogeneous effects of 3 years of math tutoring in middle school on math
achievement in grade 9 ....................................................................................................... 156

Table 57. Heterogeneous effects of private tutoring in grade 7 on overall academic achievement
for 3 years ........................................................................................................................... 159

Table 58. Heterogeneous effects of private tutoring in grade 8 on overall achievement for 2 years
............................................................................................................................................. 160

Table 59. Heterogeneous effects of private tutoring in grade 9 on academic achievement in grade
9........................................................................................................................................... 161

Table 60. Heterogeneous effects of verbal tutoring in grade 7 on verbal achievement for 3 years
............................................................................................................................................. 164

vii
Table 61. Heterogeneous effects of verbal tutoring in grade 8 on verbal achievement for 2 years
............................................................................................................................................. 165

Table 62. Heterogeneous effects of verbal tutoring in grade 9 on verbal achievement in grade 9
............................................................................................................................................. 166

Table 63. Heterogeneous effects of English tutoring in grade 7 on English achievement for 3
years .................................................................................................................................... 167

Table 64. Heterogeneous effects of English tutoring in grade 8 on English achievement for 2
years .................................................................................................................................... 168

Table 65. Heterogeneous effects of English tutoring in grade 9 on English achievement in grade
9........................................................................................................................................... 169

Table 66. Heterogeneous effects of math tutoring in grade 7 on math achievement for 3 years 170

Table 67. Heterogeneous effects of math tutoring in grade 8 on math achievement for 2 years 171

Table 68. Heterogeneous effects of math tutoring in grade 9 on math achievement in grade 9 . 172

Table 69. Long-term heterogeneous effects of private tutoring in grade 7 on overall achievement
on the CSAT ....................................................................................................................... 175

Table 70. Long-term heterogeneous effects of verbal tutoring in grade 9 on the CSAT verbal
achievement ........................................................................................................................ 176

Table 71. Long-term heterogeneous effects of English tutoring in grade 7 on the CSAT English
achievement ........................................................................................................................ 177

Table 72. Heterogeneous effects of English tutoring in grade 8 on the CSAT English
achievement ........................................................................................................................ 178

Table 73. Heterogeneous effects of math tutoring in grade 7 on the CSAT math achievement 179

Table 74. Heterogeneous effects of 2 years of private tutoring in high school on the CSAT .... 181

Table 75. Heterogeneous effects of 3 years of private tutoring in high school on the CSAT .... 182

Table 76. Heterogeneous effects of 3 years of math tutoring on the CSAT math achievement . 184

Table 77. Heterogeneous effects of math tutoring in grade 12 on the CSAT math achievement
............................................................................................................................................. 186

Table 78. Cost-effectiveness ratios of four types of private tutoring ......................................... 200

viii
 LIST OF FIGURES

Figure 1. Demand and supply of education with private tutoring ................................................ 18

Figure 2. Examples of survey questions to measure students' level of self-esteem...................... 69

Figure 3. Structure of empirical analyses ..................................................................................... 86

Figure 4. Common support between the treatment and control groups ...................................... 132

ix
 ACKNOWLEDGEMENTS

This moment of writing these last few words on my dissertation reminds me of the time

when I wrote my statement of purpose for admission into this doctoral program. At that time,

ideas about my research and future career were wide-ranging and somewhat vague, both of

which needed more elaboration. While I was going through the four-year journey as a doctoral

student, my ideas became more sophisticated and sheer by learning the professional skills and

knowledge in education at Teachers College. The doctoral student experience taught me the

importance of persistence, effort, and people around me to succeed in life.

One of the crucial events in my life was meeting my advisor, Professor Henry Levin, at

Teachers College. I learned so much from him while taking his classes and working as his

teaching assistant. Especially, during the dissertation process, his guidance and support were

vital to designing and completing my dissertation. I would like to express my deepest gratitude

to him. I also want to thank Professor Francisco Rivera-Batiz, Professor Mun Tsang, Professor

Douglas Ready, and Professor Jane Waldfogel who taught me important research skills and

knowledge, and provided me valuable comments and suggestions for my dissertation.

I also received tremendous help from my precious friends, Emma García, Brooks

Bowden, and Ilja Cornelisz. They always listened and responded to my concerns and thoughts

about my dissertation every Thursday night for almost two years. Thank you so much!

Especially, I thank Dr. Emma García who is my best friend, motivator, and supporter.

I want to dedicate my dissertation to my parents, Won-hee Lee and Young-ae Lee. I

really appreciate your support, encouragement, and belief in me, which have been the source of

my strength. Dad, you made me enter into the field of education, which I will never regret! A

x
special dedication also goes to my husband, Karl Jaewon Shin, who made all the sacrifices for

me to complete the Ph.D. coursework and dissertation. I cannot thank you enough for your

patience and support. We finally went through a long tunnel of living apart for almost three

years! I am so lucky to have you as my husband. I also thank my sisters, Karl’s parents, and all

my friends who have continuously encouraged me throughout this journey. A special thanks to

my friend, Jane Han, who edited my proposal and lengthy dissertation several times during her

pregnancy.

xi
 1

CHAPTER I: INTRODUCTION

1.1 Background and Motivation

Education has been widely considered as one of the important determinants of an

individual’s productivity and economic growth of countries. Theodore W. Schultz (1961)

claimed in his article, Investment in Human Capital, that education is one of the major activities

that increase human capital, which is tightly linked to individuals’ earnings and growth of

economy. In order to find evidence of the link connecting education to individual’s productivity

and economic growth, researchers have focused largely on formal education, i.e., primary,

secondary, and tertiary schools in both public and private sectors. There is a breadth of studies

that explore how increases in both quantity and quality of schooling are related to students’

academic achievement and labor market outcomes as well as economic development. However,

there is a lack of studies that investigate the effect of private tutoring, a form of supplementary

education where students can acquire more skills and knowledge to increase their human capital.

This is an important area in education as private tutoring becomes a growing phenomenon in

many countries (Dang and Rogers, 2008).

The private tutoring sector has been expanding in many countries, so much so that it can

be considered the third emerging education sector in addition to public and private school sectors

(Dang and Rogers, 2008). Table 1 provides some statistics to show this growing phenomenon of

private tutoring in the world. For example, in Azerbaijan, almost all senior students in secondary

school received private tutoring; 92 percent of senior students in high schools reported that they

received private tutoring. In Cyprus, 86 percent of students in secondary schools received

additional education from private tutoring: 70 percent in Hong Kong, 88 percent in Kenya, 83
 2

percent of students in primary schools in Republic of Korea (hereafter Korea), around 70 percent

of pupils in secondary schools in Japan, and 91 percent in Mauritius.

Table 1 also conveys that private tutoring prevails in both developed and developing

countries. The scale of private tutoring appears to be the highest in East Asian countries.

However, it is also an important phenomenon in many countries of different sizes, level of

economic development, political environment or geographical locations (Dang and Roger, 2008).

For example, in Turkey, spending on private tutoring has approached the level of expenditures

on the formal public system; expenditures on private tutoring are about 1.44 percent of Gross

Domestic Product (hereafter GDP) while public education expenditures are about 2 percent

(Tansel and Bircan, 2006). The situation is more severe in Korea; recent statistics from the

Korean National Statistics Office (hereafter KNSO) showed that expenditures of private tutoring

in Korea were about 3 percent of GDP in 2009. Thus, private tutoring is a widespread

phenomenon in the world that is worth investigating.

It is also important to note that a substantial amount of demand for private tutoring

appears at all levels of education. As Table 1 shows, a large proportion of students in upper-

secondary schools participate in private tutoring to prepare for university entrance exams. In

addition, students in primary and lower secondary schools display increasing demand for their

academic achievement. Moreover, some countries like Japan and Korea reported a substantial

amount of private tutoring expenditures by high school graduates who have not gained admission

to the university that they wish to attend. These people are labeled ronin in Japan and jaesusang

in Korea. For them, it is common to spend an additional year or more in exam preparation by

enrolling in private tutoring institutes such as yobiko in Japan and hakwon in Korea in order to

get into the university that they want. More explanations of different types of private tutoring
 3

will be explained in Chapter III. Therefore, the private tutoring phenomenon is not limited to a

specific level of education.

However, the definition of private tutoring varies in different countries and educational

settings. For example, India and Israel consider tutoring that is financed by government and

non-governmental organizations (hereafter NGO) as well as privately funded tutoring as private,

while Korea does not include those publicly funded as private tutoring. In addition, many

researchers interchangeably use shadow education when describing private supplementary

tutoring because it mimics formal schooling. (Stevenson and Baker, 1992; Bray, 1999b; Lee at

al., 2009). Dang and Rogers (2008) also used shadow education in the meaning of the

dependency of private tutoring on the formal education system, which implies that the private

tutoring industry does not stand alone as an independent educational activity apart from formal

schooling. However, I am hesitant to use the word “shadow” since it carries negative

connotations, such as unauthorized or dependent, which does not accurately describe the Korean

private tutoring in current years. Therefore, it is important to specify the definition or scope of

private tutoring that this dissertation discusses.

Table 1. The scale of private tutoring in selected countries

Country Year Level/Grade/Age Percent of students tutored Comment Source

Secondary school 57%
These first-year university students Silova &
Azerbaijan 2004
University 92% received private tutoring in their Kazimzade (2006)
last year of secondary school.
This study found more boys Ahmed et al.
Bangladesh 2004 Primary School 43% receiving tutoring than girls. (2005)
The proportion of students taking
private tutoring was 60% among
1997-
Cambodia Primary School 31% urban schools and 9% among rural Bray (1999)
1998
schools.

Student age 13 and 5%-17% for students age 13;

1999 CME (2000)
16 8%-20% for students age 16
Canada Over the past 30 years, the number
of formal tutoring business in major Aurini & Davies
1997 School age children N/A
Canadian cities has grown between (2004)
200%-500%.
These students received private Stylianou et al.
Cyprus 2003 College 86% tutoring in lyceum. (2003)
This study used the Egypt
Demographic and Health Survey. Suliman & El-
Egypt 2000 Children age 6-15 71%
Kogali (2002)

80% attended group (cram)

Psacharopoulous
preparatory schools, 50%
&
Greece 2000 University received individual private
Papakonstantinou
tutoring and 33% received both
(2005)
group and individual tutoring.

1995- Grade 6 pupils in two urban and Tembon et al.

Guinea Grade 6 19% four rural schools (1997)
1996

4
 35% of Secondary 1-3 students;
1996- Bray & Kwok
Hong Kong Secondary School 47% of secondary 4-5 students;
1998 (2003)
70% of secondary 6-7 students.
64% of 8th graders received This study used data from the 1995
Japan 1995 Grade 8 weekly tutoring in math and TIMSS NCES (1996)
41% in science.
58% of the students attending
private tutoring paid for it. The
Onsomu et al.
Kenya 2000 Grade 6 88% proportion of pupils who received
(2005)
private tutoring had gone up from
69% in 1998 to 88% in 2000.
Primary School 83% In aggregate, 73% of all Korean
Republic of
2003 Middle School 75% students had private tutoring. Kwak (2004)
Korea
High School 56%
2004- Budiene &
Lithuania University 62%
2005 Zabulionis (2006)
In 1990, 8,420 students in grade 3,
5 and 6 were surveyed and
respective proportions receiving
tutoring were 59%, 53% and 31%. Marimuthu et al.
Malaysia 1990 Grade 3, 5, and 6 83%
About 84% of students had received (1991)
some form of tutoring by the time
they reached upper secondary
school.
91% of these students paid for
private tutoring. The proportion of
Kulpoo &
Mauritius 2001 Grade 6 87% pupils who received private tutoring
Soonarane (2005)
had gone up from 78% in 1995 to
87% in 2001.
A 1993 survey of 1,953 formal
secondary science teachers
indicated that 53% provided after- Caillods et al.
Morocco 1993 Secondary school 78% school tutoring. The lowest (1998)
proportion (27%) was in the first
year of secondary education; but the
figure rose to 78% in the most

5
 senior grade.
32% in rural areas and 58% in
Romania 1994 Grade 12 urban areas received private UNESCO (2000)
supplementary tutoring.
Primary school 49%
Singapore 1992 George (1992)
Secondary school 30%
Glewwe &
Sri Lanka 2003 Grade 5 78% Jayachandran
(2006)
81% of 397 senior secondary
Taiwan 1998 Secondary school Tseng (1998)
school students
The number of private tutoring
centers in 2002 totaled 2,100 (up
from only 174 in 1984) which is Tansel & Bircan
Turkey 2001 High School 35%
close to the number of 2,500 high (2006)
schools in the whole country in the
same year.
These students received private
tutoring in their last year of Hrynevych et al.
Ukraine 2004 University 68% secondary school. (2006)

In aggregate, 27% received private

Years 6 & 11 26%
United tutoring. Ireson &
2003
Kingdom Rushforth (2005)
Year 13 30%
To prepare for the SAT or ACT,
14% -21% took special courses at
1990- high school, 8%-14% took group
High School Briggs (2001)
1992 private tutoring (commercial
United States coaching classes), and 6%-8% took
one-to-one private tutoring.
It is estimated that almost 7 million
2000 Elementary school elementary school students were Gordon & Gordon
likely to get tutoring and that (2003)

6
 tutoring has grown to be a
professional-service industry of
over $5-$8 billion.
Primary school 31% Around 34% of households with
children in school sent their
Lower secondary children to private lessons and the
1997- 56%
Vietnam school majority of them (90%) allocated Dang (2007b)
1998
Upper secondary between 1% and 5% of the total
77% household expenditure on private
school
tutoring.
38% of these students paid for
private tutoring. The proportion of
Nassor et al.
Zanzibar 2000 Grade 6 56% pupils who received private tutoring
(2005)
had gone up from 46% in 1995 to
56% in 2000.
The percentage varied from 36% to Machingaidze et
Zimbabwe 1995 Grade 6 61%
74% across the regions. al. (1998)
Sources: This Table is largely based on Table 1 in Dang and Rogers (2008) and Table 1 in Bray (1999).

7
 8

1.2 Definition of Private Tutoring

In order to help identify the nature of private tutoring in different contexts, it is useful to

set the criteria as other researchers have in their studies. The criteria that this study uses are

based on Bray’s (1999) who adopted several criteria to help readers understand the context of

private tutoring: supplementation, privateness, and academicness.

The first criterion is the matter of supplementation. Most countries where private

tutoring prevails consider tutoring only for subjects that are already covered in formal schooling

(Bray, 1999). In other words, subjects not taught in school, such as language or art, are often

excluded in the category of private tutoring. This is one of the reasons why we encounter studies

of private tutoring that often limit the scope of studies by observing supplementary tutoring. The

other reason why it is common to analyze supplementary tutoring may be that those subjects

covered by supplementary tutoring are tested in schools, which enables observing the effect of

private tutoring. However, tutoring classes for subjects not taught in school are often taken by

people who wish to satisfy their personal interests or development; therefore, we often have

difficulties detecting the outcomes. Following the convention, this study investigates private

tutoring that plays a supplementary role.

The second criterion is the dimension of privateness. Tutoring services are provided by

different entities for different purposes. One of the most common forms of tutoring is the one

that is provided by private entrepreneurs and individuals for profit-making purposes (Bray, 1999).

This form of tutoring called juku and hakwon is common in Japan and Korea, respectively

(Stevenson and Baker, 1992; Kim & Lee, 2010). More detailed information about the forms of

private tutoring will be explained in Chapter III. In addition, there are other types of tutoring

that are provided by governments and NGOs. As mentioned earlier, these forms of tutoring are
 9

available in many countries, such as India and Israel, and they often include these forms into the

category of “private” tutoring (Banerjee et al., 2007; Lavy and Schlosser, 2005). However, this

study only considers tutoring lessons that are provided by private entities, without examining

tutoring supplied by public entities. In addition, unpaid tutoring or voluntary help provided from

family members is also not taken into account as private tutoring.

The third criterion is the academicness of the subjects for tutoring. Tutors are commonly

perceived as people who help students carry the heavy academic load of formal classrooms (Bray,

1999). However, tutoring classes for non-academic subjects, such as musical, artistic or sporting

skills, which are covered in school are also available although demand for these classes is limited.

Due to this commonly accepted concept of tutors, studies of private tutoring often tend to focus

on academic subjects taught in school, without taking non-academic subjects into account.

Another possible explanation is that since academic subjects are more easily examinable than

non-academic subjects, this characteristic facilitates to observe the outcomes of tutoring. This

may be one of the reasons why researchers limit the scope of private tutoring only for academic

subjects, which this study also follows1.

The three criteria used by Bray (1999) help readers have a more concrete concept of

private tutoring. Reflecting on these three criteria, in this paper, private tutoring is defined as

fee-based tutoring provided by private entrepreneurs and individuals for profit-making purposes,

which provides supplementary instruction to children in academic subjects that they study in the

formal primary and secondary education system (Dang and Rogers, 2008). In other words, this

study does not include tutoring activities that cover subjects that are not taught in formal

1
Focus of this study is on private tutoring for academic subjects. There are studies such as Lipscomb (2007) and
Barron et al. (2000) that examined how non-academic extracurricular involvement affects academic achievement.
 10

schooling, are provided by public entities that do not require fees, or teach non-academic

subjects.

1.3 Research Questions

This dissertation addresses the impact of private tutoring on three aspects: students’

academic outcomes, formal schooling, and educational inequality. With respect to each aspect,

this dissertation attempts to answer the following three questions.

(1) The impact on students’ academic achievement:

Does private tutoring have a causal impact on students’ academic achievement? What

are the short- and long-term effects of private tutoring on students’ academic

achievement?

(2) The impact on formal schooling:

Does private tutoring influence students’ behavior in schools that affect the learning

environment in schools? In particular, how is participation in private tutoring

associated with students’ attention to lessons in schools?

(3) The impact on educational inequality:

Does private tutoring exacerbate educational inequality? Are there heterogeneous

effects of private tutoring on student academic achievement? For whom are private

tutoring effects significant, and how large might they be?

1.4 Endogeneity of Private Tutoring

When we estimate the effect of private tutoring on student outcomes, the simplest way of

measuring the returns to private tutoring is to use the Ordinary Least Square (hereafter OLS)
 11

statistical approach, which utilizes a set of covariates in order to control for differences between

students who participate in private tutoring (the treatment group) and their peers who do not

participate (the control group) as follows:

0 1 k k

where Y is a student’s outcome variable, PT is a private tutoring variable, which can be a

student’s participation in private tutoring, the number of hours per week that a student spends on

private tutoring, or expenditure on private tutoring classes. In this study, PT means participation

in any type of private tutoring. X is a vector of other student, household, school, and community

characteristics; and ε is the error term. We are interested in the parameter, α1, which is the

estimated effect of participating in private tutoring.

However, it is risky to rely on this estimate because students who participate in private

tutoring tend to be different in various unobservable ways from their peers who do not

participate. For example, students who are in the private tutoring sector are more likely to be

raised by parents who have higher aspiration or concerns with their children’s education. Those

parents would help their children succeed in school in numerous ways, such as by helping them

with their homework, creating supportive home environments to promote their study, or

spending money on getting supplementary education. In addition, students who receive private

tutoring may have higher academic motivation than those who do not. However, we should also

understand that students who have strong academic motivation and have parents with higher

aspiration for their children’s education tend to develop positive outcomes, such as solid

academic performance without private tutoring. Therefore, these characteristics could affect
 12

both students’ decision to receive private tutoring and their academic performance. There will

not be any problem if we could control these characteristics in the regression estimation.

However, since these types of characteristics are rarely and inaccurately measured in practice,

we cannot properly control for them in the regression analysis. Thus, this selectivity of students

who take private tutoring and these unobserved variables cause biases in estimating the effect of

private tutoring. The bias that is created by the selectivity of treated students is called the

selection bias, and the bias caused by unobserved variables is called the unobserved variable bias.

Without taking care of all the unobserved variables, these variables will end up in the error term,

ε, and due to their correlation with a private tutoring variable, they will make the estimation

results inconsistent and unreliable. Having this threat in mind, this study uses several quasi-

experimental methods that reduce biases generated by the OLS estimation and help calculate

precise estimates of the private tutoring effect.

1.5 Structure of the Study

This dissertation is organized as follows: Chapter II introduces the theoretical

background of this study and a summary of existing literature on private tutoring; Chapter III

introduces an overview of private tutoring in Korea; Chapter IV, V, and VI discuss three

empirical analyses in terms of the effect of private tutoring on students’ academic achievement,

quality of formal schooling, and educational inequality, respectively; Chapter VII summarizes

the findings and concludes with the discussion of policy implications, limitation of the study, and

directions for future research.

As explained above, private tutoring is a worldwide phenomenon, which prevails not

only in East Asian countries but also other regions in the world, and it can be considered the
 13

third education sector as Dang and Rogers (2008) claimed. Therefore, the effectiveness of

tutoring becomes an important area of research as the demand for private tutoring expands. This

study investigates the impact of private tutoring on three aspects: students’ academic

achievement, the learning environment in formal schooling, and educational inequality. Taking

methodological challenges that are caused by the selection bias and the unobserved variable bias

into account, several quasi-experimental methods are introduced to accurately estimate the

impact of private tutoring. Beforehand, Chapter II describes the theoretical background and

existing literature on the topic of private tutoring, and discusses the gaps in the existing literature

on private tutoring.
 14

CHAPTER II: LITERATURE REVIEW

This chapter introduces three theories that are necessary to understand the private tutoring

phenomenon: 1) human capital theory, 2) the standard microeconomic theory of supply and

demand, and 3) educational production function. The next section summarizes the existing

literature on the topic of private tutoring. Many studies have explored the micro and macro

determinants of purchasing private tutoring and its impact on several student-related outcomes.

After a critical overview of the literature, I will explain the gap in the existing literature, which

this dissertation partially fills.

2.1 Theoretical Background

2.1.1 Human Capital Theory

The demand for private tutoring can be interpreted as a form of human decision making

with the aim of increasing knowledge and skills motivated by the desire to build human capital

that yields higher future earnings and success based on the theory of human capital. In order to

understand the underlying context of private tutoring, understanding human capital theory is

basic.

Treating human beings as capital goods was controversial among economists until the

mid-20th century even though a few economists had started viewing human beings as a capital of

nations2 (Schultz, 1961). Due to more humanistic conceptions of human beings, the concept of

human capital had to endure criticisms against it (Schultz, 1961). However, Theodore W.

2
There were three distinguished people who have looked upon human beings as capital: Adam Smith, Johann
Heinrich von Thünen, and Irving Fisher between 18th and mid-20th centuries (Schultz, 1961).
 15

Schultz who was an economist in the 20th century undertook to criticize the classical notion of

labor as simplistic in his article Investment in Human Capital. The classical notion is that labor

should be considered as a capacity to do manual work requiring little knowledge and skill. Also,

he argued that human beings should be treated as a form of capital of the country because of their

important economic role. The improvement in the quality of human capital increases the

productivity of workers, which is linked to real earnings as well as the economic growth of

nations (Shultz, 1961; Becker, 1962; Lucas, 1988; Mankiw, Romer & Weil, 1992). Due to these

important roles of human capital, Schultz (1961) believed that human capital should be promoted

by investing in people through five important activities such as health services, on-the-job

training, formal education, adult education, and migration of individuals and families.

After this formal introduction of human capital by Schultz, Gary Becker (1962)

developed a theory of investment in human capital with an emphasis on empirical implications.

More broadly than Schultz (1961), Becker (1962) defined human capital investments to be any

“activities that influence future real income through the imbedding of resources in people.”

(Becker, 1962, pp. 9). Among various activities, Becker (1962) used on-the-job training and

how it affects the earnings of trained workers. With the distinction between general and specific

on-the-job training3, he showed the age-earning profiles for trained and untrained workers, which

showed different patterns in growth of earnings. For both general and specific on-the-job

training, earnings of trained workers are lower than earnings of untrained workers in the period

of training due to costs that workers bear for training4. However, earnings of trained workers

surpass the earnings of the untrained in later ages due to increased productivity with skills and

3
General training is training that is useful in many firms in addition to the firm providing it whereas specific
training trains workers with skills and knowledge that are valuable in the firm providing it. Therefore, in case of
specific training, productivity is raised more in the firm acquiring the knowledge than in other firms (Becker, 1962).
4
Costs consist of direct outlays of training and foregone income that workers could have earned in other
occupations (Becker, 1962).
 16

knowledge that workers acquired during training. Becker (1962) also argued that the magnitude

of returns to general training during the period of training is smaller compared to the returns to

specific training because workers bear all the costs of general on-the-job training whereas firms

share costs with workers who receive specific on-the-job training5. Because the training process

requires costs and time to complete and because its benefits are born after a considerable time

period, the value of the resources during training period is regarded as investment. Also, the

higher earnings in the later years that result from the training are treated as its yield.

In addition to on-the-job training, other activities could also increase human capital.

Becker (1962) defined a school as an institution specializing in the production of training.

Through schooling, people absorb skills and knowledge (either general or specific) that they

would need in the labor market. Therefore, schooling would have the same kind of implications

as on-the-job training. Moreover, apart from the knowledge and skills that people could learn

from training or schooling, other information such as economic, political, or social systems could

help people choose patterns of consumption, employment, or allegiance to political parties,

which could significantly raise real incomes of workers (Becker, 1962). Activities that promote

emotional and physical health are other ways to improve human capital that have a significant

influence on earnings.

Within the category of investment in human capital, private tutoring can be considered as

one of the activities that help raise the quality of human capital. Similar to schooling, private

tutoring aims to provide the necessary knowledge and skills that are required to succeed in

formal schooling and the labor market in the future. In addition, students with access to private

5
The reason why workers bear all the cost of general on-the-job training is that general training increases workers’
marginal product in the firm providing it as well as other firms. Skills and knowledge that workers learned in
general training can be used in different firms (Becker, 1962). However, specific training is useful in a specific firm,
which affects productivity of the firm instead of workers, so firms are responsible for costs of specific training
partially or entirely.
 17

tutoring institutions often benefit from receiving other information about external academic

resources, universities or future careers that are often provided by tutors or private tutoring

institutions. These resources could strengthen the human capital of students, which is believed to

have a strong connection with their success in the future labor market. Therefore, the theory of

investment in human capital provides a critical background to explain the demand for private

tutoring.

2.1.2 The Standard Microeconomic Theory of Supply and Demand

The standard microeconomic theory of supply and demand can be used to interpret the

private tutoring phenomenon. This theory explains how the quantity of education, including

private tutoring, is determined by the interaction between supply and demand for education in

the market. Dang and Rogers (2008) presented the supply and demand for education for a

typical household in the case where private tutoring is available as shown in Figure 1.

In Figure 1, there are three supply curves that represent the different types of education

provided. The three supply curves S0, S1, and S2 represent 1) private education, 2) public

education, and 3) public education with private tutoring, respectively. S0 is placed farther up in

the left corner compared to S1 and S2 because of the high costs of private education. S0 is also

more inelastic in price because parents who send a child to a private school are less sensitive to

the price of education and more sensitive to the quality of education. S1 shows an upward-

sloping curve ending at point A, and the line becomes vertical rising from point A (perfectly

inelastic supply). The vertical slope of S1 means, regardless of consumer demand, after a certain

point, public schools may reach their capacity limit, preventing them from offering as much
 18

education—in terms of both quantity and quality—as parents or students want6. S2 shares a

common solid upward-sloping curve with S1, but it includes a dashed diagonal line staring from

point A. This dashed line is less steep than the vertical curve of S1, implying that private tutoring

can meet students’ and parents’ demand for education as opposed to public education. In

addition, this dashed line is steeper than the solid part of S2, indicating that the cost of private

tutoring is higher than the cost of public education.

Figure 1. Demand and supply of education with private tutoring

The household demand for education is represented by either the demand curve D1 or D2.

Even though schooling is provided free of tuition, a household always bears certain costs to send

a child to school, such as school fees or foregone earnings by being at school instead of working

6
The capacity constraint of public education may be a better description of the short run than of the long run. In the
long run, the vertical part of S1 would become more elastic because government may invest to increase the quality
of effective public education by expanding school capacity or increasing teacher quality.
 19

(opportunity costs). With these costs of education, D1 is the demand curve for a representative

household, and D2 represents another household that is assumed to have either a higher income,

stronger preferences in education, or higher expectations about a child’s returns to education.

Due to these differences, at each price, a household in D2 would spend more on education than a

household in D1.

The quantity of education is determined by the intersection between the supply and

demand curves (equilibrium). If the representative household’s demand for education is

represented by the demand curve D2, the amount of private education the household consumes is

Q0, and the amount of public education consumed is Q2. In the presence of private tutoring, the

same household can consume Q*2, which is a larger amount of education than Q2.

This standard framework explains a household’s different behaviors in different settings

of education. Availability of private tutoring promotes households to consume more education

than when they have formal education only. However, this framework fails to explain several

other aspects of private tutoring. For example, as Figure 1 shows, parents and students face

different prices in each setting (S0, S1, and S2). The price that a household shoulders in the

setting of public education with private tutoring (P*2) is lower than the price that a household

takes on when only public education is available to them, which may not be true in many cases.

In addition, this framework fails to explain the demand of private tutoring by students in private

schools.

The failure of capturing all aspects of private tutoring using this framework may be due

to certain assumptions that this framework incorporates (Dang & Rogers, 2008). This

framework assumes that 1) the market for private tutoring is perfectly competitive, 2) public

education reaches a strict capacity constraint after a certain point, and 3) an increase in education
 20

units through private tutoring increases students’ human capital. However, these assumptions

may not always be valid in practice and may differ from setting to setting within a country and

among different countries. The market for private tutoring is not always perfectly competitive

because of teacher corruption. In a perfectly competitive market of private tutoring, households

independently make decisions on whether to spend money on private tutoring for their children.

However, in many developing countries, it is often the case that public school teachers offer

private tutoring for their students after regular classroom hours and they make it mandatory by

providing a part of curriculum during tutoring hours (Dang & Rogers, 2008; Buchmann, 1999;

Silova & Bray, 2006). Even though some countries such as Ukraine and Korea prevent teachers

in formal schools from tutoring outside of schools, it is still common in other developing

countries. In addition, public education does not necessarily have a capacity constraint in the

long run because governments can try to increase the quality of public education by allowing

longer school hours or increasing teacher quality. Lastly, private tutoring may aim for test

preparation instead of accumulation of human capital. Even though this framework has several

limitations that do not allow us to explain every aspect of private tutoring, it delivers a broad

picture of an education market with private tutoring and explains how its introduction in the

market can increase households’ consumption of this type of education.

2.1.3 Educational Production Function

In order to estimate the effect of private tutoring on various student outcomes, this

dissertation employs an educational production function approach. An educational production

function is analogous with industry production functions in economics, which explain the

relationship between the inputs, such as labor and capital, into the production process using
 21

existing technology and the resulting output of firms including goods and services (Pindyck &

Rubinfeld, 2000). In education, test scores, graduation rates, or dropout rates are used as outputs,

and inputs are typically factors like students, family, and school. The statistical relation between

these inputs and outputs is mathematically represented as follows (Hanushek, 1986).

(2.1) At = f (Ft, Tt, OSt)

At represents the achievement of a student at period t; Ft represents the family inputs, such as

parental education, income, race, and home language, that affect student achievement; Tt

represents teacher inputs for a student such as the education level of teachers, teaching

experience, and other teacher qualifications; and OSt represents other school inputs such as class

size, location of schools, library, curriculum and so on. Many researchers modify this function

by including other inputs such as students’ innate ability, peer factors, and neighborhood factors.

Using the various types of educational production functions, this dissertation investigates the

relationship between private tutoring participation and student academic outcomes.

2.2 Determinants of Private Tutoring

2.2.1 Micro factors

There are several studies that explored the determinants or drivers of private tutoring.

Based on related studies, Dang and Rogers (2008) accumulated the results in existing studies

regarding what kinds of micro and macro factors influence the demand for private tutoring. In

terms of micro factors, many studies show a consensus that the most influential micro factors are

household income, parental education, and urban location (Assaad & El-Badawy, 2004;
 22

Stevenson & Baker, 1992; Tansel & Bircan, 2006; Dang, 2007b). Specifically, students whose

parents live in an urban area with higher income and education levels have a higher probability

of receiving private tutoring than their peers who live in a rural area with parents with lower

income and education levels. Similar patterns have been found in Korea (Choi et al., 2003; Kim,

2004; Kim & Lee, 2010; Lee, 2003; Won, 2001; Yun, 1997).

In addition, a student’s grade level and household size explain the pattern of demand for

private tutoring in some countries. In Egypt and Vietnam, students in higher grade levels,

especially the ones in the last grade of their current school level or in diploma-granting years,

showed a higher demand for private tutoring (Assaad & El-Badawy, 2004; Dang, 2007b). And

in Korea, Turkey, and Vietnam, the number of children in households showed a negative

relationship with private tutoring expenditures (Kim & Lee, 2010; Tansel & Bircan, 2006; Dang,

2007b). However, the household size variable is likely to be endogenous. For example, parents

who have several children may have a lower level of parental concern for their children’s

education or different educational beliefs. In addition, resources such as government subsidies

and corporate benefits for people with multiple children may affect parents’ decision on how

much they spend money on private tutoring. However, since these different characteristics of

parents with multiple children are difficult to measure and unobserved, it would create bias in

estimating its effect.

Baker et al. (2001) also argued that student academic performance is one of the factors

that determine the private tutoring expenditures of households. However, this factor plays out

differently in different countries. For example, using data from the Third International

Mathematics and Science Study (TIMSS) in 1995 for 41 countries, Baker et al. (2001) showed

that in three-fourths of these countries, low-performing students spent more money on private
 23

tutoring, whereas one-fourth of them exhibited an opposite pattern. This implies that some

countries use private tutoring as a remedial strategy, while other countries consider it as an

enrichment strategy (Baker et al, 2001; Won, 2001). In this study, Baker et al. (2001) used the

OLS estimation to explore the determinants of private tutoring with several control variables

including a composite measure of socioeconomic status of students, their gender, type of

community, at-school remedial resources, and whether there is a difference between the language

used at home and the language used in school instructions. However, there were many

unobserved variables that the authors failed to control such as student academic motivation and

parent academic aspiration for their children, which make estimates biased.

Stevenson and Baker (1992) used a logistic regression equation to find the determinants

of the shadow education in Japan. In addition to the micro factors mentioned above, student’s

gender, student curriculum track (academic track vs. vocational track), and high school

reputation seem to explain the pattern of students’ shadow education in Japan. They found that

male students, students in the academic track, and students in high schools that have a higher

reputation spent more money on shadow education than their counterparts.

2.2.2 Macro factors

In addition to micro factors, the literature on private tutoring has cited several macro

factors that seem to foster the demand for private tutoring. First, Bray (1999) argued that as

some of underdeveloped countries have transitioned to a market economy, the demand and

supply of private tutoring have substantially increased. As the economy improves, increasing the

real income of households, households would start spending more money on their children’s

education and other goods, an income effect in economics. Bray (1999) used cases in China,
 24

Vietnam, and Eastern Europe to show the emergence of supplementary tutoring, which

previously did not exist in these countries.

Stevenson and Baker (1992) introduced another macro factor that would affect people’s

decisions on private tutoring. If there is a tight linkage between education and future success in

the labor market, they argued that this linkage would stimulate the competition for more

education, and thus private tutoring. Given this tight linkage, the existence of high-stakes formal

examination, a prerequisite to acquire more education in higher education institutions, increases

the demand for education even more (Stevenson & Baker, 1992; Bray, 1999). In the case of

Taiwan, the view is pervasive that students participate in shadow education to prepare for

university entrance examinations because getting into prestigious universities determines their

success in the labor market (Lin, 1983). Also, in Hong Kong, there is high demand for shadow

education as a preparation for the secondary-school-certification examinations because

performance on the exam is the most crucial factor that determines their job and salary level

(Mitchell, 1968; Sweeting, 1983). However, Baker et al. (2001) found the opposite result in

terms of the influence of high-stakes testing. They analyzed 41 countries using the 1994-95

TIMSS data and found no relationship between the existence of high-stakes testing and the use

of shadow education.

Cultural values are also emphasized to explain the pattern of private tutoring in many

countries. Bray (1999) argued that supplementary tutoring is especially prevalent in cultures that

stress effort. Many Asian cultures, which show a high demand for tutoring, are influenced by

Confucian traditions that stress effort as a factor for future success, whereas a person’s ability is

more emphasized in European and North-American cultures (Rohlen & LeTendre, 1996). In

addition, “many Asian cultures value disciplined study and are both competitive and status-
 25

conscious,” which influence people’s need for private tutoring (Bray, 1999). In terms of

competition, the intensive competition among students for college entrance in Korea is analyzed

as one of the important determinants of private tutoring (Hyun et al., 2002; Kim, 2004; Lee,

2005).

The ineffectiveness of the public education system is found as one of the determinants of

private tutoring (Kim & Lee, 2010; Buchmann, 1999; Silova & Bray, 2006). In Korea, the

public education system is regulated strictly by the government especially after the introduction

of the Middle School and High School Equalization Policy (Lee & Hong, 2001). Due to this

rigidity, it is completely insulated from the market forces and local parents’ demand for

education (Kim & Lee, 2010). To meet this high demand, parents and students who are not

satisfied with education provided by the public school system find other ways to meet their needs

by participating in the private tutoring sector or by studying abroad (Chun et al., 2003; Kim,

2004). Several Korean researchers have conducted studies to analyze whether the High School

Equalization Policy is one of the factors fanning the increasing demand for private tutoring, and

they argued that this policy has played a significant role (Lee & Hong, 2001; Kim et al., 2003;

Kim et al., 2003). Buchmann (1999) and Silova and Bray (2006) expressed concerns that low

pay levels and weak monitoring of teachers in the public system may create a private tutoring

market for teachers who wish to reap more rewards from teaching outside the public schools.

Private supplementary tutoring may become more necessary in systems that are teacher-centered

rather than child-centered, and are intolerant of slow learners (Bray, 1999). However, most of

these studies are speculative and anecdotal and do not indicate the causal link between the nature

of education systems and demand for private tutoring.

 26

The degree to which mass education is institutionalized within a nation is also related to

the national use of shadow education (Baker et al., 2001). Baker et al. (2001) used two variables

as indicators for the degree of mass education in a nation: public expenditures on education as

percentage of GNP and gross enrollment ratio at elementary and secondary levels from the

UNESCO Statistical Yearbooks. This study found that the expanded provision of mass

education does not generate a higher prevalence of shadow education use. In other words,

families in a less funded and less enrolled system use more shadow education than families in a

fully developed, mass-education system.

As explained above, many researchers have introduced these macro factors that may

determine the demand for private tutoring. However, there is little formal empirical research to

test the above-described hypotheses, so it is still uncertain whether there are causal relationships

between these macro factors and the prevalence of private tutoring.

2.3 Empirical Literature on the Effect of Private Tutoring on Outcomes

Empirical studies have found positive effects of private tutoring on students’ educational

outcomes. Stevenson and Baker (1992) investigated whether the participation in high school

shadow education increased the likelihood of university attendance in Japan. Overall, they found

that students who reported that they have participated in certain types of shadow education

during high school years showed a higher probability of attending universities in their first year

following graduation from high school. For students in the first year out of high school, practice

examinations and correspondence courses improved the probability of entering college by 16

percent and 25 percent, respectively, whereas having a private tutor significantly reduced this

probability, which reflects the remedial character of this form of private tutoring in Japan. In

addition, after-school classes (juku) had only a small and insignificant effect on attendance,
 27

probably because students in better high schools who tend to be high-performing use their high

school’s after-school program instead of juku. For students in the second year out of high school,

attending special tutoring school increased the probability of entering college by 80 percent.

However, using the logistic regression, Stevenson and Baker (1992) failed to remove the bias

caused by selectivity of private tutoring participants. In addition, the student academic

performance variable used in all regression models is problematic because it is likely to be

endogenous, which means that it is correlated with unobserved student innate ability or

motivation.

Buchmann (2002) also found positive impacts of private tutoring on student academic

performance in Kenya. For 13- and 19-year-old students, private tutoring reduced the chance of

grade repetition and increased student academic performance. Similarly, Briggs (2001) looked at

commercial private tutoring courses in the United States and tried to measure its effectiveness

using the OLS method. He found that coaching increased SAT math scores by 14-15 points,

SAT verbal scores by 6-8 points, and ACT math and English scores by 0-0.6 points, but

decreased ACT reading scores by 0.6-0.7 points. Park, Park, and Kim (2001) and Yang and Kim

(2003) also argued that private tutoring contributed to the improvement of the math and science

scores in TIMSS 1999 using Hierarchical Linear Modeling (HLM). Using the multiple

regression, Lee (2001) claimed that private tutoring is one of the important factors for academic

achievement. Using path analysis, Kulpoo (1998) and Polydorides (1986) also found the

positive correlations between private tutoring and academic achievement in Mauritius and

Greece, respectively. However, significant caveats of these studies are that either they do not

control for school characteristics, which may bias the estimation results, or they do not use an

appropriate method to observe the causal relationship.

 28

As opposed to positive effects, several studies found negative or no effect of private

tutoring on student outcomes. Han, Sung, and Gil (2001) and Ban, Jung, and Yang (2005)

provided counter-evidence in which private tutoring is not statistically significant in explaining

academic achievement using regression analysis. Lee et al. (2004) investigated the effect of pre-

class tutoring on academic achievement of students who were enrolled in secondary schools in

Korea. Pre-class tutoring is defined as private tutoring that teaches a school’s curriculum at least

one month ahead of its schedule. This study concluded that pre-class tutoring has no short- or

long-term effects on student academic performance in Korean language, English and

mathematics. However, this study also has several weaknesses in the analysis. The sampling

procedure raises many doubts whether it ensured a perfectly random sample. The authors

seemed to have failed to account for differences in regional characteristics and student

motivation, which affect both participation decision of private tutoring and academic

achievement. In addition, for the long-term analyses, the sample size was too small, which tends

to lead to biased estimates.

Cheo and Quah (2005) and Ha and Harpham (2005) also found insignificant effects of

tutoring on student achievement. Cheo and Quah (2005) conducted an analysis using multiple

regression with students in grade 8 in three schools in Singapore and found negative and

insignificant effects of having a paid home tutor. Similarly, using logistic regression, Ha and

Harpham’s (2005) results showed an insignificant effect of taking extra after-school classes on

writing and numeracy for eight-year-old children in Vietnam, whereas children with extra classes

were more than twice as likely to be able to read correctly compared to children who didn’t take

extra classes. Although the authors controlled for many confounding factors, such as region,
 29

household wealth, parents’ education, household size, child ethnicity, and mental health, they did

not control for school characteristics, which are important confounding factors.

Some studies investigated the effect of private tutoring based on the subjective answers

of perceived academic improvement. Korean parents showed strong trust in the effect of private

tutoring (Paik, 1999; Yun, 1997). In addition, studies have shown that German and Japanese

parents believed that private tutoring plays a positive role in increasing academic achievement of

their children (Kramer & Werner, 1998; MEXT, 1994).

All studies introduced so far have mixed evidence on the effect of private tutoring on

student academic performance. However, the results from these studies should be received with

caution because they are not successful in controlling for endogeneity of private tutoring. By

contrast, there are many studies that have tried to control for endogeneity of private tutoring in

some credible ways as follows.

The ideal setting of the study in order to control for endogeneity is the Randomized

Control Trial (RTC). Banerjee et al. (2007) used a randomized experiment of a remedial

education program called the Balsakhi Program conducted in schools in urban India. This

program provided government schools with a teacher to work with third and fourth graders who

have been identified as falling behind their peers. Typically, a teacher meets with a group of

these students (15-20 students) and holds a class for two hours a day. The participating schools

were randomly divided into two groups. Schools in group A provided remedial education only

to third graders, whereas schools in group B provided it to fourth graders, and vice versa in the

following year. Therefore, children in grade 3 in schools that received the program for grade 4

form the comparison group for children who received the program for grade 3. Given this design

of the program, Banerjee et al. (2007) found a larger improvement in average test scores for the
 30

treated group than the comparison group (0.28 standard deviations increase in average test

scores). This study seems successful in minimizing possible biases of the estimates caused by

selectivity of the program participation. Even though this remedial education program is

somewhat different from private tutoring programs since the Balsakhi Program is financed by a

non-governmental organization and free of charge for families, it is exemplary of a study design

that could control for endogeneity of tutoring.

Using observational data, researchers have also been trying to use quasi-experimental

methods in order to explain the effect of private tutoring. Mischo and Haag (2002) conducted

the study to observe the effect of private tutoring for students in grades 5 to 11 in Luxembourg.

This study used a form of matching, in which 122 students receiving tutoring (treatment) identify

a match using subject matters and performance in the subjects. Using this procedure, 122 non-

tutoring pupils were recruited as a comparison group. This study concluded that receiving

private tutoring “causes” a larger improvement in academic performance. Out of six school

marks in the school system (1=very good to 6=insufficient), school marks for student receiving

tutoring improved by 0.97, 0.77, 1.18, and 0.72 for mathematics, English, Latin, and French,

respectively, whereas school marks for non-tutoring students showed a smaller improvement in

these subjects. These differences were statistically significant. Mischo and Haag (2002),

however, failed to match students in the treatment and control groups in terms of their

unobserved characteristics, such as student academic motivation or family background, which

also have an influence on their decision of participating in private tutoring and their academic

performance.

The instrumental variable method seems popular in the studies of private tutoring in order

to control for endogeneity. Lavy and Schlosser (2005) observed the effect of a remedial
 31

education program on matriculation rate and participation in the matriculation exam in Israel. As

an instrument for the proportion of students participating in the program (treatment intensity),

authors used the interaction terms of schools size with the year dummy variable and treatment

status. Instead of using school size itself as an instrument, which has an independent direct

effect on school achievements, i.e., the matriculation rate and the number of students who

participated in the matriculation exam, the authors overcame the difficulty by estimating the

direct effect of school size on school achievements based on data from the untreated cohorts in

order to partial out the direct effect of school size on school achievements. Also, to allow more

flexibility, they split the instrument into three variables based on the distribution of school size.

Using this instrument, they found that a remedial education program increased the mean

matriculation rate for schools and the number of participating students in the matriculation exam

by 3-4 percent and 11-12 percent, respectively. Suryadarma et al. (2006) used proportion of

classmates taking extra courses as an instrument of private tutoring to observe whether taking

extra courses explains student academic achievement in Indonesia. They found no impacts on

mathematics or dictation scores for the fourth graders. In this study, the authors argued that this

instrument fulfills the requirement of being highly correlated with the instrumented variable, i.e.,

participation in extra courses, and exogenous to a student’s score, which is the dependent

variable. However, the authors did not consider peer effects, which could occur when there are

many classmates who receive tutoring and they have an influence on students who do not take

tutoring. In addition, the authors did not provide tests to check the validity of the instrument. As

opposed to the results in Suryadarma et al. (2006), Dang (2007b) found positive effects on

student academic performance using per hour private tutoring fees charged in the commune as

instruments for participation of private tutoring in Vietnam. The author claimed that the
 32

instrument is exogenous because the fees are regulated by government rules based on local living

standards. However, the study explained that students could also attend other tutoring classes

outside the commune. In this study, the author failed to prove that this instrument is exogenous

to students’ academic performance. He concluded that, for both primary and lower secondary

students, higher spending on private tutoring decreased the probability that the student falls into

either the poor or average performance categories, but increased the probability that the student

enters the good and excellent performance categories. However, the influence was larger for

lower secondary schools.

However, we should be critical about whether instruments meet certain assumptions to be

appropriate. As one of the most important assumptions for the instrumental variable method,

instruments have to fulfill the requirement of being highly correlated with the instrumented

variable, which is the decision on whether to participate in private tutoring in this study, and

exogenous to the dependent variable, which is called exclusion restriction. However, the

proportion of classmates taking extra courses, which was used as an instrument in Suryadarma et

al. (2006), seems to have effects on student performance explained by peer effects and is not

clearly predictive of selection. If the proportion of classmates who take extra courses is high, it

may create positive externalities to students who do not take extra courses, a factor that can have

an impact on their academic achievement.

Besides the instrumental variable method, other quasi-experimental methods were used

as identification strategies. Briggs (2001) applied the Heckman model to correct for the effects

of selection bias and found identical estimates for the couching variable to those produced by the

linear regression. Jacob and Lefgren (2004) investigated summer remedial programs in Chicago

using a regression discontinuity method. From 1997-1998, students who made below the cutoff
 33

scores for mathematics and reading should participate in summer remedial programs, whereas

students who scored above the cutoff scores were not subject to take remedial programs. In

order to identify the effect of these summer remedial programs, they compared students who

scored just below and just above the cutoff assuming that the unobservable characteristics of

students do not vary discontinuously around the cutoff. This study found that the programs

increased math and reading achievement for third graders by about 12 percent of the average

annual learning gains whereas 6 percent for sixth graders. However, this study did not provide

evidence whether other characteristics of students are continuous around the cutoff even though

it is one of the important assumptions for the validity of the regression discontinuity method.

Korean researchers have also put in much effort to uncover the effect of private tutoring

in various ways. In the effort to uncover a causal relationship between private tutoring and

academic achievement, Kang (2007) and Choi (2007) applied the instrumental variable method

to correct for endogeneity. Using the birth order as the exogenous instrumental variable of the

expenditures of private tutoring, they claimed that the effect of private tutoring was not

statistically significant on academic achievement and college attendance. Kang (2007) claimed

that being a first-born significantly and positively affected private tutoring expenditures for a

student; parents spend 30 percent more money on private tutoring for first-born children than for

later-born children. In addition, he argued that being a first-born has no direct association with

students’ academic performance. However, as mentioned earlier, there is a lack of Korean

studies that used causal inference methods; therefore, the true effect of private tutoring is still

unknown.
 34

2.4 Gaps in the Existing Empirical Literature

As listed above, most of empirical studies on the effect of private tutoring focus on

educational outcomes such as student’s academic achievement and college attendance.

Educational outcomes have been considered the focus of interest of students, parents, and

policymakers because the main reason for participating in private tutoring is to improve learning

and educational results. However, private tutoring not only generates an educational impact but

also an impact on 1) formal schooling, 2) society, and 3) economy (Bray, 1999).

In terms of the impact on formal schooling, many qualitative studies have argued that

tutoring is reported to have several effects on formal schooling. For example, tutoring can take

away students’ interest and attention from lessons in schools since they have already covered the

topics with tutors (Hussein, 1987; Nanayakkara & Ranaweera, 1994; Sawada & Kobayashi,

1986). In addition, tutoring can decrease the effectiveness of teachers, especially in a situation

where teachers are allowed to be tutors. This is because teachers might have an incentive to

design the curriculum as too full and might deliberately slow down their pace of delivery in order

to ensure that they have a market for supplementary classes that generate additional income for

themselves (Hargreaves, 1997; Caillods et al., 1998; Bray, 1999).

Private tutoring has several social implications such as consequences of pressure on

students, the impact on social relationships, and the implications for social inequalities (Bray,

1999). Some researchers argued that children who attend both formal and private tutoring

classes deal with more academic pressures and depression than those who only attend formal

classes (Tsukada, 1991; Wijetunge, 1994), and it exacerbates pressures and depression in a

setting where high-stakes tests take place (UNICEF, 1994). De Silva (1994) also pointed out

that participation in private tutoring can weaken family bonds of affection because children are

away from home and their parents most of the time. Adversely, Paiva et al., (1997) and Russell
 35

(1997) claimed that supplementary tutoring would also provide a healthy framework within

which young people can develop and meet many peers, which improve their social relationship.

Moreover, private tutoring may create a mechanism that increases social inequality (Bray, 1999).

Since most forms of private tutoring is more easily accessible to students from high-income

families, it can create greater inequality in access of supplementary education that widens the

achievement gap among income or racial groups. Yang (2007) also argued that private tutoring

widens the education inequality in terms of college entrance, whereas Choi (2007) provided a

counter-argument on the relationship between expenditures on private tutoring and college

entrance.

Lastly, in terms of the economic implications of private tutoring, private tutoring as a

form of education can increase students’ human capital, which increases labor market earnings in

the future, according to the human capital theory (Schultz, 1961; Becker, 1962). Advocates of

the human capital theory might consider private tutoring to be more tightly connected to

individuals’ economic enhancement than formal schooling because it is closely tied to the

demands of the market place and because enhanced economic return is among the chief reasons

why students and their parents invest in it (Bray, 1999). Alternatively, private tutoring might

contribute to suppressing creativity, which can adversely affect future economic productivity,

because most of private tutoring is focused on preparation of traditional examinations that

largely focused on memorization (Bray, 1999). While there is substantial literature on the rates

of returns to formal education (Pschoaropoulous, 1994; Carnoy, 1997; Bennell, 1998), a lack of

empirical literature exists on this subject.

Although some studies take into account the social and economic outcomes of private

tutoring, they are mostly speculative and anecdotal without appropriate quantitative evidence.
 36

Thus, the most significant gap in the existing literature on private tutoring is that there is a lack

of empirical studies to investigate these external returns to investment in private tutoring. This

study seeks to fill this gap and explores these unknown research areas in order to explain the

impact of private tutoring to a greater extent beyond student academic achievements that are

already used in many studies.

Moreover, even though there a number of Korean studies about private tutoring as

previously stated, studies that control for endogeneity of private tutoring are lacking. Thus, it is

still not conclusive whether there is a causal link between private tutoring and outcomes in Korea.

Using quasi-experimental methods, this study aims to uncover the casual effect of Korean private

tutoring on several outcomes. Before empirical analyses, Chapter III will explain a detailed

picture of Korean private tutoring. Chapter III will help understand the situation of private

tutoring in Korea in terms of its purpose, intensity, providers, contents, and expenditures.
 37

CHAPTER III: OVERVIEW OF KOREAN PRIVATE TUTORING

The characteristics of the private tutoring market differ by country. Among many

countries that have a sizable demand for private tutoring, the Korean private tutoring market is

predominant. Therefore, it is worth investigating its characteristics. To understand the detailed

characteristics, Chapter III provides detailed information on the major purposes of taking private

tutoring, participation rate by level of schooling, forms of private tutoring, numbers of providers,

size of the private tutoring market, tutor salaries, contents, and expenditures on private tutoring,

etc. This chapter will help readers understand how important private tutoring is in the Korean

society.

3.1 Purposes

There are several purposes of participating in private tutoring. The most popular purpose

is to use private tutoring as an enrichment strategy (Baker et al., 2001). This strategy is used

among high-achieving students who wish to further increase their performance through private

tutoring. Some of the factors that fulfill this enrichment purpose include the tight linkage

between education and future success, and people’s concern for the deterioration of public

schooling. In Korea, enrichment seems to be one of the major purposes of private tutoring as

reflected in Table 2. According to the KNSO (2010), participation in private tutoring is the

highest among high-achieving students. About 85 percent of students placed higher than the

90th percentile report that they participate in at least one form of private tutoring activities, and

this participation rate is twice the rate for students placed lower than the 20th percentile.
 38

Table 2. Participation rate and monthly expenditure on private tutoring by academic

percentile

Academic Percentile Participation Rate (%) Monthly Expenditure (USD)

Higher than 90th 85.3 317
90th - 70th 83.9 282
70th - 40th 73.8 233
40th - 20th 59.8 182
Lower than 20th 48.8 136
Source: Korea National Statistics Office (2010)
Note: Average rates and expenditures with respect to students in primary and secondary schools

Table 2 also shows a gradual decrease in participation as students’ academic standing

falls. It is consistent with findings from Baker et al. (2001) using cross-national data from the

Third International Mathematics and Science Study (TIMSS). Baker et al. (2001) found that the

role of private tutoring in Korea is classified as predominantly enrichment because almost twice

as many high-scoring students regularly use private tutoring than low-scoring students. However,

Baker et al. (2001) argued that most of countries used private tutoring for a different purpose.

Private tutoring in many countries is also classified as a remedial strategy (Baker et al.,

2001; Dang & Rogers, 2008). Baker et al. (2001) found the dominant remedial strategy in

nations such as Cyprus, Israel, Belgium, and Demark. Compared to students who scored in the

top one-third, two to more than three times as many students used private tutoring among those

in the lowest one-third. Similar but less dominant patterns were revealed in the U.S., Germany

and Kuwait. This indicates that in these nations private tutoring is mostly used for a remedial

purpose among low-achieving students in order to meet certain academic thresholds. Even

though enrichment is the modal trend of its use in Korea as explained, there are a substantial

number of students who use private tutoring as a remedial strategy. As Table 2 shows, on

average, 54.3 percent of students who are placed lower than the 40th percentile participate in

private tutoring. Even though this rate of participation is much lower than the one from high-
 39

achieving students, this proportion is still substantial and meaningful. However, Bray and

Lykins (2012) claimed that private tutoring is much less used for remedial help in current years.

Students also decide to receive private tutoring in order to prepare for examinations that

are required to move to a higher level of schooling. This phenomenon is prevalent in many

Asian countries, such as Japan, Korea, and Taiwan, where high-stakes tests exist (Stevenson &

Baker, 1992; Baker et al., 2001). In Japan, a form of private tutoring that is exclusively focused

on practicing the university entrance exam is the most popular type among high school students

(Stevenson & Baker, 1992). Similarly, in Korea, preparing for the university entrance exam is

one of the most common objectives of private tutoring (KNSO, 2008). According to the 2008

survey from the KNSO, about one-quarter of students reported that preparation for the university

exam is their intention of doing private tutoring.

The survey from the KNSO (2008) also found that about 15 percent of participants

reported that their major purpose of private tutoring is to reduce anxiety from peer pressure.

Since a large number of students are engaged in private tutoring in Korea, as explained in the

following section, students who do not participate tend to feel pressure to be part of it. Several

studies about private tutoring from the Korean Educational Development Institute (2001, 2003,

and 2005) showed that some students participate in private tutoring because most of their

classmates do. Or they report that they would become anxious or fall behind if they don’t

participate.

Lastly, some students use private tutoring for their personal development in subjects that

are not taught in schools such as foreign languages, fine arts, or music. There is a small

proportion of students who wish to develop their personal interests in music, fine arts, athletics,

or skills required for their future career through private tutoring (KNSO, 2010). This purpose of
 40

private tutoring is more linked to students’ leisure and personal development apart from

academic purposes.

3.2 Intensity

Table 3 shows private tutoring participation rates by the level of school based on a

national survey of private tutoring conducted by the KNSO since 2007. On average, in 2010,

about three-quarters of students (73.6 percent) in the formal school system participate in at least

one kind of private tutoring. Specifically, about eight out of 10 elementary school students (86.8

percent) were enrolled in private tutoring activities. Among middle school students, 72.2 percent

reported that they are currently participating in at least one form of private tutoring in 2010. In

terms of high school students, about half (52.8 percent) use private tutoring for their study.

Interestingly, the rate of participation in private tutoring decreases as the level of education

increases. In other words, the intensity of private tutoring participation is the highest at an

elementary school level compared to middle and high school levels. This decreasing trend has

several implications. It may imply parents’ strong belief in the importance of education at the

elementary level. Or it may explain that higher levels of formal schooling are better at meeting

parents’ or students’ academic needs so that students in high schools participate less in private

tutoring and rely more on formal schooling than students in elementary or middle schools.
 41

Table 3. Participation rate by education level in Korea

Elementary School Middle School High School

Average
(Grade 1 – 6) (Grade 7 – 9) (Grade 10 – 12)
2007 77.0 88.8 74.6 55.0
2008 75.1 87.9 72.5 53.4
2009 75.0 87.4 74.3 53.8
2010 73.6 86.8 72.2 52.8
Unit: in percentage
Source: Korea National Statistics Office (2007, 2008, 2009, 2010)

3.3 Providers

3.3.1 Forms of Private Tutoring

Students and families can choose different forms of private tutoring. Generally, there are

five forms of private tutoring based on the situation in Japan and Korea, which have the two

largest private tutoring markets in the world (Dawson, 2010).

First, there is individual tutoring in which one instructor provides private lessons to one

student, typically at a student’s home. In Korea, instructors are usually college or graduate

students, or professional tutors who are not employed at private learning institutes. It is called

gwaoe in Korea. Also in Japan, it is also common that private tutors (katei kyosi) work with

students on a one-on-one basis. Private tutors tend to be more focused on helping students’

regular schoolwork instead of examination preparations in Japan (Stevenson & Baker, 1992).

However, in Korea, tutors teach the academic contents in formal schooling in advance to prepare

students as well as help students’ homework. Like school teachers, tutors teach new subjects at

the same time, providing examples related to the subjects in order to maximize the effectiveness

of tutoring. Tutors also provide homework and help tutees’ schoolwork if needed. This form of
 42

private tutoring was the most popular in the 1970s when competition for university entrance was

fierce in Korea. In the effort to reduce the demand for private tutoring, the military government

issued “The July 30 Education Reform,” which prohibited students and teachers from taking and

giving private one-to-one or one-to-many lessons. Thus, the demand decreased temporarily in

the 1980s.

Second, students receive educational materials such as self-study sheets that are regularly

prepared and delivered by private companies. There are three subcategories under this form of

private tutoring. First, students are provided with answers and additional explanations for

questions in order to help them study independently. Second, after students submit the finished

self-study sheets, they are returned to students with comments and instructions for further study.

Lastly, after students finish the self-study materials by themselves, a tutor who is employed by

private companies that make the educational materials visits a student’s home and provides

additional lessons related to the materials. All these categories are generally called haksupji,

which is popular for students at any level of education in Korea. In Japan, this form is called

correspondence course (tsushin tensaku), which is popular for students who plan to take the

university entrance exam. Students receive mock tests by mail and send them back to private

providers in order to get their grades. There is a practice examination (mogi shiken), which is

similar but slightly different from the correspondence course. Instead of receiving mock tests by

mail, students visit private companies and take the test that the companies generate. After a few

days, students receive a report by mail that informs students and parents of their chances of being

admitted to university by comparing their performance with national norms as well as their

standing in each subject area (Stevenson & Baker, 1992).

 43

Third, the most common and substantial form of private tutoring is the one provided by

private for-profit institutes (hakwon) in Korea (often called cram schools in the U.S.). It has

classrooms and instructors, and the instruction is given in a classroom-like setting. After formal

schooling, students attend hakwon that they choose and review what they learned in formal

schools or learn school materials in advance (learning-in-advance7) similar to one-to-one tutoring.

However, there is a large discrepancy between hakwon in Korea and test preparation centers in

the U.S. While the centers provide mostly test preparation courses for the SATs and ACTs,

hakwon mimics classrooms and provides more advanced study than formal schools in order to

meet the needs of students and parents. This is also why students neglect public school

education (Kim et al., 2003; Stevenson & Baker, 1992).

The Korean government has maintained a strong control over hakwon, requiring those

who want to establish one to acquire a government-issued permit. Instructors at hakwon have to

meet certain academic qualifications, and school teachers are prohibited from participating in

any form of private tutoring including hakwon. Also, there are strict requirements for facilities

in order to establish a hakwon (Kim & Lee, 2001; Kim & Chang, 2010). Similarly, primary and

secondary Japanese students also participate in similar private after-school classes (juku). This

group tutoring comes in five different types: (1) shingaku juku (university entrance exam

preparation), (2) hoshu juku (remedial study), (3) kyosai juku (for school refusers and drop-outs),

(4) doriru juku (“drill” work, e.g. Kumon), and (5) sogo juku (comprehensive type which

includes elements of the other four) (Roesgaard, 2006). Juku is for students in grades 1 through

9 (Lee, Park & Lee, 2009).

7
Private tutoring institutes (hakwon in Korea and juku in Japan) teach formal education curriculum in advance of
public schools. They teach students for 2 months during the vacation before the beginning of the academic year,
during the school term, they teach the curriculum at a more rapid pace than the school, and then they review
materials during the exam period (Kim, 2003). This becomes a problem because school becomes a place for review
and it threatens the authority of public education (Russell, 2002).
 44

Fourth, online tutoring services are rapidly growing as an alternative form of private

tutoring in Korea. Private tutoring institutes not only provide in-house tutoring, which was

already explained, but also make lessons and materials available online. Students can purchase

these online courses and materials provided by instructors employed by private tutoring institutes.

There is also a public online tutoring called the Educational Broadcasting Station (EBS), which

started in the mid-1990s. In order to decrease the demand for private tutoring, the courts

supported a new and less expensive government-funded educational tutoring model. Quality

teachers in formal schools or private tutoring institutions participate and make their lectures

available on EBS, enabling all students to benefit from them without bearing any financial

burden.

Lastly, as briefly explained above, high school graduates who do not gain admission to

any university or the one they wish may participate in full-time preparation institutes. Those

students are called ronin and jaesusang in Japan and Korea, respectively. They often attend a

private examination preparation school, which is typically considered a bigger version of hakwon,

requiring at least a year to prepare solely for the university entrance examination. In Japan, the

private examination preparation school is called yobiko, which is very similar to a formal school

(Stevenson & Baker, 1992). In this study, this form of private tutoring will not be studied

because people in this category are not enrolled in the formal school system. This study will

limit the scope to private tutoring that is available to students who are enrolled in formal schools.

Table 4 shows the participation rate in each form of private tutoring by Korean students

at each level of education. Among forms of private tutoring, participation in private tutoring

institutes (hakwon) is the highest in all levels of education (50.8 percent in elementary schools;

53.8 percent in middle schools; 28.3 percent in high schools). Using self-study materials
 45

(haksupji) is also common for elementary school students (35.6 percent), but participation in this

form decreases dramatically as the level of education increases. Table 5 shows the participation

rate in private tutoring by Japanese students during high school years using a nationally

representative data. In Japan, more than half of high school students (54 percent) in the sample

participate in practice examination (mogi shiken), followed by private tutoring institutes (juku)

(35 percent).

Table 4. Participation rate in each type of private tutoring by school level in Korea
Elementary Middle High
Total
School School School
Private Tutoring 64.8 74.4 68.5 44.9
Private tutor (one-to-one tutoring) 10.6 7.3 12.4 14.4
Private tutor (group tutoring) 11.7 14.7 10.7 7.8
Private tutoring institute (hakwon) 45.5 50.8 53.8 28.3
Self-study materials (haksupji) 19.5 35.6 10.5 1.5
Online tutoring 3.7 2.5 4.2 5.1
Unit: in percentage
Source: Korean National Statistics Office (2010)
Note: This table shows private tutoring participation for core subjects only (reading, writing, English, mathematics, science,
social science, foreign language (besides English), computer)

Table 5. Participation rate in each type of private tutoring by school level in Japan
Total Sample of Students with
Students (%) College Plana (%)
Practice examination (mogi shiken) 54 68
Correspondence course (tsushin tensaku) 30 43
Private tutor (katei kyoshi) 8 11
Private tutoring institute (juku) 35 46
Plans to be a ronin after high school 29 32
N 7,240 5,352
Unit: in percentage
Source: Stevenson and Baker (1992), p.1646
a
students with college plans represent 75 percent of base-year sample.
Data: longitudinal study of high school seniors in Japan conducted by the Youth Research Institute of Tokyo in 1980 and
1982 (nationally representative dataset)
 46

3.3.2 Number of Providers

The number of private tutoring providers by type is shown in Table 6. According to the

Survey of Private Tutoring Providers administered by the Ministry of Education, Science and

Technology (hereafter MEST) in Korea, in 2008, there were 55,834 tutors for one-to-one

tutoring, 40,202 tutors for group tutoring, 192 online tutoring companies, and 81 companies for

self-study materials that are registered at the local ministries of education. The number of

private tutoring institutes (hakwon) has increased tremendously from 14,043 in 2000 to 65,810 in

2008 even though they are under a regulatory environment (Kim & Lee, 2001). In addition, a

large number of graduates from the humanities and social sciences studies enter the private

tutoring sector (Kim & Park, 2012). Table 6 captures the popularity of private tutoring in Korea.

Table 6. Number of private tutoring providers by type in Korea

Number of private tutoring providers

Private tutor (one-to-one tutoring) 55,834
Private tutor (group tutoring) 40,202
Private tutoring institute (hakwon) 65,810
Self-study materials (haksupji) 81(341)*
Online tutoring 192(550)*
Source: Survey of private tutoring providers by Ministry of Education, Science and Technology (MEST) in 2008
* number of private tutoring companies that serve more than one type of private tutoring
 47

3.3.3 Size of Private Tutoring Market

Table 7 shows the annual gross sales of each form of private tutoring in 2008. Private

tutoring institutes’ (hakwon) gross sales was about 10 billion U.S. dollars (hereafter USD) in

2008, which is the highest compared to other forms of private tutoring. About half of gross sales

came from elementary school students, followed by middle and high school students. One-to-

one tutoring generated the second-largest gross sales (about 5 billion USD) among these forms of

private tutoring. The largest proportion of the total sales came from high school students. The

gross sales for self-study material companies is also substantial (about 1.9 billion USD), and as

explained above, students in kindergarten and elementary schools take up the majority of people

who consume this form of private tutoring. Group tutoring with a private tutor represented about

886 million USD of their total sales in 2008, and it is more common for elementary school

students. These total gross sales reflect the size of the private tutoring market in Korea.

Table 7. Annual gross sales of private tutoring by type and education level in 2008

Private
Private tutor Private tutor Self-study
tutoring Online
(one-to-one (group materials
institute tutoring
tutoring) tutoring) (haksupji)
(hakwon)
Total 5,051.9 886.6 10,281.4 1,983.9 411.1
Before elementary - 8.9 261.7 589.2 -
Elementary School 1,744.3 662.3 4,540.6 1,095.1 117.2
Middle School 1,374.2 80.9 3,394.5 190.5 133.6
High School 1,933.4 134.5 1,230.6 111.1 160.3
After secondary - - 854.0 - -
Unit: 1 million USD
Source: Korea National Statistics Office (KNSO), 2009; Ministry of Education, Science and Technology, 2008
Note: Won is translated in USD based on that 1,000 Korean won equal approximately 1 U.S. dollar.
 48

3.3.4 Annual Salary of Tutors

As Table 8 shows, average annual salaries for tutors employed by private tutoring

institutes and self-study material companies are about 18,876 and 18,980 USD, respectively.

These salaries are similar to an average starting annual salary for a school teacher who graduated

from college and has a teacher certificate (19,083 USD)8. Tutors employed by companies are

paid much higher than self-employed tutors. However, average salaries for those tutors are much

lower than average annual salary of school teachers, which is about 29,459 USD9 in Korea.

Table 8. Average annual salary by type of tutors in Korea

Type of tutors Average annual salary

Tutors in hakwon 18,876
Tutors in self-study material companies 18,980
Individual tutors
College or graduate students 5,540
Professional tutors 12,020
Unit: 1 USD
Source: Survey of private tutoring providers administered by Ministry of Education, Science and Technology
(MEST) in 2008
Note: Won is translated in USD based on that 1,000 Korean won equal approximately 1 USD.

8
Monthly salary schedule for K-12 teachers in 2011 is listed in Table A in Appendix
9
Formula = [(sum of all average monthly salaries in Table A in Appendix)*12]/40
 49

3.4 Contents of Private Tutoring

Contents of private tutoring can be divided into two groups. First, there is private

tutoring that utilizes contents covered in formal schooling. In Korea, popular subjects for private

tutoring are reading, writing, English, mathematics, social studies, science (biology, physics,

chemistry, and earth science), foreign languages (besides English), and computer. The

participation rate by each subject is shown in Table 9. On average, among students who

participate in private tutoring activities, English and mathematics are the most popular subjects

for private tutoring. The popularity of the two subjects is largely explained by the importance of

these two subjects in the university entrance exam. This dominant share of private tutoring on

English and mathematics is consistent in each level of education. Tutoring for reading is also

substantial in all levels of education (40.6 percent in elementary school; 34 percent in middle

school; and 13.5 percent in high school). However, private tutoring for writing, foreign

languages, and computer skills is not substantial compared to other subjects. As mentioned

above in terms of intensity, the participation rates for private tutoring decrease as the level of

education increases.

Table 9. Private tutoring participation rate by subject (covered in formal schooling) and by
education level in Korea in 2010

Elementary Middle High

Average
School School School
Reading 31.4 40.6 34.0 13.5
Writing 7.9 13.0 4.7 2.5
English 52.5 60.2 61.9 29.9
Mathematics 53.6 58.1 63.2 36.2
Social Science & Natural Science 20.5 22.4 30.5 7.4
Foreign languages and Computer 10.7 16.7 9.2 1.9
Unit: in percentage
Source: Korea National Statistics Office (2010)
 50

Second, there are private tutoring activities that cover contents not taught in formal

schools. This category can be divided into 1) contents related to students’ personal interests or

leisure and 2) contents related to career preparation. Examples of contents for personal interest

and leisure are fine arts, music, dance, and athletics. Examples for career preparation are classes

for the Test of English for International Communication (TOEIC) or certificates for skills that

employers look for in a recruiting process. Participation rates for these contents are presented in

Table 10. On average, music and athletics are the two most popular subjects (18 percent in

music and 15.5 percent in athletics). Tutoring fine arts is also noticeable in elementary schools.

Students start tutoring for career preparation in their high school years, but its share of tutoring is

minimal.

Table 10. Private tutoring participation rate by subject (not covered in formal schooling)
and education level in Korea in 2010

Elementary Middle High

Average
School School School
Music 18.0 33.9 6.0 3.5
Fine Arts 6.9 12.6 1.8 2.6
Athletics 15.5 28.9 5.8 2.7
Career preparation 0.4 - - 1.5
Other 3.9 6.4 1.8 1.8
Unit: in percentage
Source: Korea National Statistics Office (2010)
 51

3.5 Expenditure on Private Tutoring

3.5.1 Total Expenditure

It is well known that Korean parents spend a substantial proportion of household income

to pay for children’s private tutoring. A study by Choi et al. (2003) estimated the trend of

households’ expenditures on private tutoring. As Table 11 shows, spending on private tutoring

has been substantial since 1998 and amounted to 12.4 billion USD in 2003. Choi et al. (2003)

reported that it is equivalent to about 56 percent of the national budget on education and about 2

percent of GDP in 2003, which is less than 3.3 percent of GDP in 1998. A recent survey on

private tutoring conducted by the KNSO reported that in 2007 total expenditure on private

tutoring added up to about 20 billion USD, which is about 1.9 percent of GDP in Korea (Table

12). The average annual and monthly expenditure per student is about 2,664 and 222 USD,

respectively. More detailed information about household’s private tutoring expenditure by level

of education is presented in Table 12.

Table 11. Total household's expenditure on private tutoring

1998 2001 2003

Current Price 11,132,249,000 9,964,015,000 12,407,731,000
Constant Price 11,427,795,000 10,040,230,000 12,407,731,000
% of National Education Budget 67% 53.2% 55.9%
% of GDP 3.3% 1.9% 1.9%
Unit: 1 USD
Source: Choi et al. (2003)
 52

Table 12. Total, annual, and monthly expenditure on private tutoring and expenditure by
school level in 2007

Average monthly
Average annual
Total Expenditure on expenditure per student
expenditure per
Private Tutoring Among
student Total
participants a
Total 20,040,000,000 2,664 222 288
Elementary 10,209,800,000 2,726 227 256
Middle School 5,612,000,000 2,810 234 314
High School 4,218,100,000 2,368 197 359
Unit: 1 USD
Source: Korea National Statistics Office (2008)
a
: average monthly expenditure per student calculated among the participants of private tutoring
Note: 1,000 Korean won equal approximately 1 U.S. dollar.

Nam (2007) also provided time-series estimates of household expenditure on private

tutoring as well as expenditure on formal education from 1985 to 2006 as shown in Table 13.

Nam’s (2007) estimates on private tutoring are generally higher than estimates presented above

from Choi et al. (2003) and the KNSO (2008). For example, in 2003, according to Choi et al.

(2003), households paid about 12.4 billion USD (1.9 percent of GDP) on private tutoring,

whereas Nam (2007) estimated the figure at 18.3 billion USD (2.5 percent of GDP). In addition,

Nam (2007) reported that households spent about 25 billion USD in 2006, which is about 2.7

percent of GDP. It is also surprising to see that expenditures on private tutoring have been

higher than expenditures on formal schooling since 1995. Even though these estimates from

several studies are slightly different from each other, all show continuously increasing private

tutoring expenditures. Recent statistics from the KNSO reported that spending on private

tutoring again reached 3 percent of GDP in 2009.

 53

Table 13. Household expenditure in primary and secondary education, 1985-2006

Tuitions and fees for

School Supplies Textbooks Private Tutoring
formal schools
1985 1.4 (1.72) 0.1 (0.16) 0.3 (0.31) 0.5 (0.54)
1990 2.8 (1.51) 0.3 (0.14) 0.8 (0.41) 2.7 (1.45)
1995 6.1 (1.54) 0.4 (0.11) 1.9 (0.48) 7.6 (1.91)
2000 10.2 (1.76) 0.6 (0.10) 3.2 (0.56) 10.9 (1.89)
2001 11.3 (1.81) 0.5 (0.09) 3.6 (0.58) 12.3 (1.97)
2002 11.3 (1.65) 0.6 (0.08) 3.8 (0.56) 13.4 (1.96)
2003 11.0 (1.52) 0.6 (0.07) 2.3 (0.31) 18.3 (2.53)
2004 12.1 (1.55) 0.6 (0.07) 1.3 (0.17) 21.7 (2.61)
2005 12.2 (1.50) 0.5 (0.07) 1.3 (0.15) 23.7 (2.77)
2006 13.3 (1.56) 0.5 (0.06) 1.3 (0.16) 25.0 (2.79)
Unit: 1 billion USD; numbers in parentheses are percentages of GDP.
Source: Nam (2007)
Note: Expenditure on textbooks for private tutoring has been reclassified from the "Textbooks" column to
"Private Tutoring" column beginning in 2003.

3.5.2 Expenditure by Region

According to a survey by the KNSO (2010), the average monthly payment for private

tutoring per student is about 240 USD in 2010 as Table 14 shows. However, the average

monthly private tutoring expenditure per student in Seoul is about 321 USD, which is the highest

among all metropolitan and smaller cities. An average expenditure in Seoul is more than twice

the spending in small towns in Korea. Table 14 demonstrates monthly expenditure on private

tutoring per student by size of region.

Table 14. Monthly expenditure on private tutoring per student by size of region

Small and
Metropolitan Eup/
Total Seoul Medium-
Cities Myun a
sized Cities
Private Tutoring Expenditure 240 321 227 244 160
Participation Rate (%) 73.6 77.5 74.1 74.7 65.4
Unit: 1 USD
Source: Korea National Statistics Office (2010)
a
: Eup and Myun are administrative units similar to towns
 54

3.5.3 Monthly Expenditure by Income Group

According to the KNSO (2010), on average, monthly spending on private tutoring per

household is about 240,000 won, which is about 6.5 percent of their monthly income10. When

looking at the monthly expenditure by income group as shown in Table 15, the higher the

family’s income, the more they spend on private tutoring (Lee, 2005). For example, households

in the lowest income group spend about 63 USD per month, whereas households in the highest

income group spend about 484 USD a month on their children’s private tutoring. In other words,

wealthy families spend about eight times more than poor families. In addition, poor families

spend about 12.6 percent of their monthly income on children’s private tutoring, whereas

wealthier families spend about 6 percent of their income, which indicates a heavier burden for

lower income families (Choi et al., 2003; Lee, 2005).

Table 15. Monthly private tutoring expenditure per household by income group

Monthly Private Tutoring Percentage of monthly

Income Group
Expenditure income
Less than 1,000 USD 63 12.6
1,000 - 2,000 USD 103 6.8
2,000 - 3,000 USD 170 6.8
3,000 - 4,000 USD 240 6.8
4,000 - 5,000 USD 298 6.6
5,000 - 6,000 USD 362 6.5
6,000 - 7,000 USD 404 6.2
More than 7,000 USD 484 6.4
Unit: 1 USD
Source: Korea National Statistics Office (2010)

10
Average monthly household’s income in 2010 is about 3,666 USD (KNSO, 2010).
 55

As previously described, private tutoring is prevalent in Korea, which has been an

important social issue. The high demand for private tutoring has been steady for decades with

strong parental beliefs on the effectiveness of private tutoring. In the next chapter, I will explain

how I constructed the estimation models in order to quantify the parental beliefs on how much

private tutoring affects students’ academic achievement. Moreover, I will describe the dataset

and variables, which will be used in the empirical analyses.

 56

CHAPTER IV: THE IMPACT ON STUDENTS’ ACADEMIC ACHIEVEMENT

Students and parents decide to participate in private tutoring for various reasons as

reviewed in Chapter III. However, the overarching goal is to further improve academic

performance through private tutoring, and this goal has not changed. In order to qualitatively

and quantitatively prove whether this goal is achievable through private tutoring and to what

extent private tutoring affects academic performance, many scholars have done research using a

number of statistical methods that were introduced in the literature review. Due to

methodological challenges to control for the selectivity of students who participate in private

tutoring, however, there are only a few studies that successfully observed a causal link between

private tutoring and students’ academic achievement, especially in the Korean context. As

previously explained, students who participate in private tutoring tend to be different in many

unobservable ways from their peers who do not participate; students who take tutoring tend to

have higher academic motivation, a more supportive home environment to promote their study,

or parents who have higher academic aspiration for their children. This selectivity of students

would create problems in estimation since these characteristics are often unobserved, which

make impossible to control for. Since these unobserved characteristics affect academic

achievement of students who participate in private tutoring, the estimates of private tutoring

effects tend to be biased. In order to correct for this selectivity of students in the treatment group,

this study employs the instrumental variable method and the propensity score matching to

explore the casual impact of private tutoring for secondary school students’ academic

achievement in Korea. Under this main research question, I addressed cumulative and single-

year effects of private tutoring to estimate the effect of years of private tutoring and the most
 57

effective time or grade to start private tutoring. In addition, short-term and long-term effects of

private tutoring are observed using different types of variables that indicate academic

achievement.

Before conducting analyses using the instrumental variable method, I also used the

Heckman model (Heckman, 1978; 1979; Greene, 2003) to correct for the problem of selection

bias. Using similar specification functions to those employed in Briggs’s study (2004) on

observing the effect of commercial coaching programs on the SAT, I compared the estimates

obtained from the OLS and Heckman models. In the selection function, I included several

variables that are not used in the response function, such as the educational goal of students and

parents, parental support in education, students’ academic standing, and parental pressure. In

addition, I also included several covariates that were statistically significant in private tutoring

cross-tabulations, such as socio-economic status, gender, students’ intrinsic motivation, and

regional characteristics, into the selection function. The reason why I chose the similar variables

that were used in Briggs’s study (2004) is to check whether I obtain similar results as Briggs did.

With different combinations of these variables in the selection function, I tried to estimate the

effect of tutoring. However, after many trials, I concluded that two approaches produced

relatively similar estimates of the effects of private tutoring as opposed to Briggs’s results, which

found the different estimates with different specification functions. Therefore, I decided not to

report estimation results of the Heckman model in this study because the Heckman model does

not improve the precision of the estimates of private tutoring effects.

Having this goal of uncovering a causal link between private tutoring and academic

achievement in mind, first, I will describe the data that I utilized for this analysis followed by

estimation models, variables, empirical results, and discussion.

 58

4.1 Data Description

The dataset used in this dissertation is the Korean Educational Longitudinal Survey

(KELS) administered by the Korean Educational Research Institute (hereafter KEDI). The

KEDI is the nation’s most prestigious research institute in education, which devotes itself to

developing quality research and innovative solutions for policy-makers and stakeholders in

education. Starting in 2005, 6,908 students in grade 7 (the first year in middle school in Korea)

were selected using stratified and cluster random sampling; schools were randomly selected

within strata11 of cities, and students were randomly selected within schools. In terms of the

Korean school configuration, middle school education starts in grade 7 and finishes in grade 9,

and high school education is from grades 10 to 12. This nationally representative sample of

students has been followed and surveyed each year since 2005. This dissertation used all

students in the KELS data without restricting analyses to a sub-sample of the data.

Students in the data were asked questions about a wide range of topics including

experiences in school and home, educational resources and support, and expectations on their

future life. Moreover, the survey has also been administered to parents, teachers, and

principals12 of students in order to collect a wider range of information about students’ families,

classrooms, and schools.

One of the biggest advantages of this dataset is that it contains the most sophisticated

information about private tutoring relative to other existing datasets in Korea, such as type of

private tutoring that a student participates in, the subject of private tutoring, duration of

participation, hours spent per week, and monthly expenditure devoted to private tutoring by type

11
There are four stratums categorized by the size of cities.
12
Surveys for principals were administered for the first three years only. From the 4th wave, KEDI directly
collected information about schools that respondents attended.
 59

and subject, as reported by parents. Students were also asked whether they are satisfied with

private tutoring if they are participating.

There are also various measures of students’ academic achievement. The KEDI

administered achievement tests to all students in the data in verbal, English, and mathematics for

the first three years of data collection in 2005, 2006 and 2007, and results of these tests—Item

Response Theory (IRT) scores—are available for these three years. In addition, results of the

university entrance examination in the form of decile rank are available in the sixth wave of the

dataset. This university entrance examination is the College Scholastic Aptitude Test (hereafter

CSAT), which is a standardized test for all students during their senior year in high school. The

CSAT is administered by the Korea Institute for Curriculum and Evaluation (KICE), which is a

government-funded educational research institute. These test results allowed me to investigate

the short- and long-term effects of private tutoring that are received in middle school years.

Moreover, there is information about how much students understand in each subject—reported

by both students and teachers—and the degree of students’ attention to lessons in classrooms for

each subject measured by minutes. These measures can be used as alternatives of students’

academic performance.

As mentioned earlier, the KELS is longitudinal data, which has considerable advantages

over cross-sectional data. Whereas cross-sectional data observe a different group of people in

each year, longitudinal studies follow the same people over time. Therefore, in the longitudinal

data, researchers are free from considering different characteristics by different cohorts, which

makes estimates more precise. Also, this type of data enables researchers to collect information

about individual change. Another advantage is that the KELS has low sample attrition. Sample

attrition is a big challenge for many analysts because it may make the interpretation of estimates
 60

problematic. It is more problematic when attrition is high and selective on characteristics (non-

random), which is often the case in practice. However, in the case of the KELS, there is no

sample attrition in the first three waves of data, and about 91 percent, 83 percent, and 77 percent

of the total sample remained in the fourth, fifth, and sixth wave, respectively. Therefore, it is

less likely that sample attrition causes problems in the estimation using the KELS.

4.2 Estimation Models

4.2.1 Ordinary Least Square

As mentioned in Chapter II, this analysis employs the educational production function in

order to estimate the effect of private tutoring on students’ academic achievement. Among the

various estimation methods, OLS with rich controls is applied. The OLS method assumes that

there is a linear relationship between inputs and outputs. The equations of the OLS estimation

are as follows.

Cumulative Effect:

(4.1) ih(a) 0 1 ih(a) 2 ih ih

Equation 4.1 is derived to explain the cumulative effect of taking private tutoring. This

equation represents two different model specifications. The first model uses ih(a), which is an

achievement score of student i in education level h (middle or high school) in a subject a; the

second model observes ih, which is an average achievement score (using scores in three subjects)

of student i in education level h. ih(a), a variable of interest, indicates years of private

tutoring that student i has participated in education level h in subject a or any subjects.
 61

ih(a) indicates a set of dichotomous variables that indicate one, two, or three years of

private tutoring. In equation 4.1, 1 is the main coefficient of interest that is interpreted as the

effect of total years of private tutoring on academic achievement. ih is a vector of control

variables that consist of student background characteristics (e.g. gender, household’s income,

parental education, academic motivation, level of a student’s self-esteem, and previous academic

achievement, etc.), characteristics of a school that a student attends (e.g. public vs. private,

general vs. technical, and student-teacher ratio), and regional characteristics (e.g. urban vs. rural,

and region with a school choice scheme), and ih is an error term that captures unmeasured

variables.

Single-year Effect:

(4.2) ij(a) 0 1 ij(a) 2 ij ij

Unlike equation 4.1, equation 4.2 is designed to observe the single-year effect of private

tutoring. The analysis of single-year effects uncovers the effect of taking private tutoring in each

grade in suggesting implications about when taking private tutoring is most effective to raise

academic achievement. Equation 4.2 follows a similar pattern with equation 4.1, but j represents

school grade of student i. However, instead of using a cumulative variable of private tutoring,

ij(a) is a dichotomous variable of private tutoring taking the value 1 if a student i participates in

any form of private tutoring in subject a or any subjects in grade j, 0 otherwise. The single-year

effect of private tutoring is captured in 1, which is the main coefficient of interest. In order to

isolate the effect of the single-year, I also controlled for participation in private tutoring in

previous years. Again, is a vector of control variables that is identical in equation 4.1, and ij
 62

is an error term. Detailed descriptions of all explanatory variables are provided in the later part

of this chapter.

4.2.3 Instrumental Variables

Relying on results from the OLS estimations may be risky due to the selectivity of

students who participate in private tutoring, as previously mentioned in Section 1.4. Students

who participate in private tutoring may be different from those who do not participate in many

observable and unobservable ways. Even though the OLS estimation uses a set of controls to

account for these observable differences, it is still impossible to control for unobserved

differences between these two groups, which may distort the estimate of the private tutoring

effect. In order to address this selectivity problem and unobserved variable bias, the literature

suggests the use of the instrumental variable approach.

Instrumental variable estimation should enable the assumptions of exogenous variation in

the treatment (Murnane & Willett, 2010). After scrutiny, this analysis uses a student’s monthly

cost of private tutoring as a proportion of monthly household income as an instrument of private

tutoring. The validity of this instrument is tested in several ways as explained in Section 4.5.

Using this instrumental variable, the first stage in the two-stage least square (2SLS) is given by:

Cumulative Effect (first stage):

(4.3) ih(a) 0 1 ih(a) 2 ih ih

Single-year Effect (first stage):

(4.4) ij(a) 0 1 ij(a) 2 ij ij

 63

where ih(a) is a dichotomous variable that indicates total years of private tutoring, and

ij(a) is also a dichotomous variable that represents the private tutoring participation. In

equation 4.3, ih(a) is an instrumental variable, which is a proportion of student i’s costs of private

tutoring on subject a or overall in monthly household income during education level h. Similarly,

ij(a) in equation 4.4 is a proportion of student i’s costs of private tutoring on subject a or overall

in grade j. ih and ij are random error terms associated with the reduced form equation. From

equations 4.3 and 4.4, the predicted values of two variables of private tutoring are calculated.

Instead of using actual dichotomous variables of private tutoring, the predicted values are

used in the second stage as follows.

Cumulative Effect (second stage):

(4.5) ih(a) 0 1 ih(a) 2 ih ih

Single-year Effect (second stage):

(4.6) ij(a) 0 1 ij(a) 2 ij ij

where ^ represent predicted values, and 1 and 1 are the main coefficients of interest that imply

the cumulative effect and single-year effect of private tutoring after accounting for endogeneity

bias.

4.2.4 Propensity Score Matching

This study also used Propensity Score Matching (PSM) method to check whether the

results are consistent with the PSM estimates. PSM is also one of the quasi-experimental

methods, which is used with observational data in order to detect a causal mechanism. In PSM,
 64

rather than controlling for all the covariates, students in a treatment group (i.e., students who

participate in private tutoring) are matched with students in a control group (i.e., students who do

not participate in private tutoring) using the propensity score, e(X), which is a numerical

summary of i,k. We can estimate e(X) using logistic or probit regression where the dependent

variable is the treatment (a dichotomous variable that indicates participation in private tutoring)

and the predictors are all confounding covariates as follows. In the process of matching, simple

one-to-one nearest neighbor matching with replacement is used among various types of matching

(Dehejia &Wahba, 2002). In this method, each treatment unit can be matched to the nearest

comparison unit, even if a comparison unit is matched more than once. Researchers claim that

this method helps in reducing the bias because it minimizes the propensity score distance

between the treatment unit and the matched comparison unit. Using this method of matching, I

constructed a counterfactual to measure the outcome, i.e., academic achievement, which students

would have obtained, had they not participated in private tutoring.

(4.7) e(X) = Pr(PTi=1|Xi,k)

This propensity score, e(X), is a predicted probability for each person that he or she

receives the treatment. Using these two groups that are similar in various ways except for the

treatment, the mean of dependent variable (i.e., students’ academic achievement) for a treatment

group is compared to the one for a control group to observe the treatment effect. This effect is

called the average treatment effect on the treated (ATT). The mathematical representation of

difference-in-means in PSM is as follows.

 65

(4.8) τ = E[Y(1)|Z=1,e(X)] – E[Y(0)|Z=0,e(X)] = YZ=1,e(X) – YZ=0,e(X)

where τ is the treatment effect. In addition to this method of difference-in-means, I also

conducted regression-adjusted matched estimation, which is running a regression of outcome on

treatment indicator and confounding covariates using weights to construct the sample to

represent matched groups. The regression-adjusted matched estimation also provides a

sensitivity check for estimates obtained from difference-in-means.

Two estimation methods derive , which is the average treatment effect on the treated

(ATT). This average treatment effect on the treated represents the effect of private tutoring

among students who participate in private tutoring by comparing with a counterfactual that I

constructed by matching.

It should be noted that correctly calculating standard errors in PSM is a problem for

several reasons. After matching, the observations are no longer independent of each other even

with regular matching. In other words, if there are correlations between matched pairs, standard

errors are subject to be biased. In addition, matching with replacement creates the additional

complication of including some units multiple times. In order to correct for standard errors,

standard errors are calculated by bootstrapping with 1,000 replications.

However, PSM also bears methodological drawbacks that should be stated explicitly.

Several researchers have criticized that this method still has potential problems that are caused

by unobserved variables because only observed characteristics of the sample are used in the

process of matching (Michalopoulos, Bloom & Hill, 2004). In addition, the method is sensitive

to choices of confounding covariates to calculate propensity score, e(X). Even though this
 66

method carries these methodological disadvantages, it can help reduce large biases compared to

OLS.

4.3 Variables

Treatment variables. The purpose of this analysis is to estimate the effect of private

tutoring on students’ academic achievement. As I already mentioned in the section of estimation

models, I used two types of private tutoring variables: dichotomous variables that indicate 1)

total years of private tutoring and 2) private tutoring participation in general. The purpose of

using variables that indicate years of private tutoring is to observe if there is any cumulative

effect of taking private tutoring. I constructed several dichotomous variables, which indicate

years of tutoring using education level (I separated middle school and high school in terms of

years of participation in private tutoring) and subject area (verbal, English, math, or all three

subjects). For example, the variable, one year of private tutoring in middle school (PT1yr_ms)

takes the value of 1 if a student takes one year of private tutoring in middle school and 0 if a

student does not take tutoring in middle school. The variable, two years of private tutoring in

middle school (PT2yr_ms), takes the value 1 if a student takes two years of private tutoring in

middle school and 0 if a student does not take any tutoring in middle school. As mentioned

earlier, this set of dichotomous variables are subdivided by education level (middle school and

high school) and academic subject (verbal, English, and math). Also, there is the other set of

dichotomous variables taking the value 1 if a student participates in any form of private tutoring,

0 otherwise. Several of these dichotomous variables are also constructed using the grade level of

participation and subject area of private tutoring. The use of these variables aims to observe the

single-year effect to detect when taking private tutoring is most effective for increasing students’
 67

academic achievement in middle school and high school, separately. All treatment variables

used are presented in Table B.1 in Appendix.

Dependent Variables. Two types of dependent variables are used in order to estimate the

effect of private tutoring on students’ academic achievement: students’ scores on the KEDI test

and the CSAT. As previously described in the section of data description, the KEDI test was

annually administered by the KEDI to all students in the KELS data during middle school. With

the KEDI test, an average score of three subjects, verbal (Korean language), English, and

mathematics, and a score of each subject is used to observe the private tutoring effects on overall

achievement and achievement by subject. The KEDI tests were administered only during middle

school years (grades 7, 8, and 9) and the KEDI stopped administering the test from the fourth

year of survey. Therefore, the KEDI scores are used to assess the private tutoring effect on

academic achievement in middle school. Since the KEDI scores are available in raw format, I

transformed them into z-scores. Regarding the CSAT scores, an average score and a score in

verbal, English, and math are used for the purpose of looking at the effect of private tutoring on

achievement on the university entrance exam, which is one of the major purposes of taking

private tutoring in Korea. However, since the CSAT results are available in the unit of decile

rank, estimating the effect of private tutoring on the university entrance exam is less precise than

the estimation using percentiles or actual scores. Thus, the decile rank of the CSAT, the only

variable that indicates students’ high school academic achievement in the KELS data, does not

capture small differences in the CSAT scores, which determine admission into the most

competitive institutions. The decile measures are also imprecise when translating the effects at

the tails of the distribution where the distribution is skewed. Moreover, this decile measure will

be more problematic when estimating the tutoring effects for students in the highest decile due to
 68

ceiling effects. Since students in the highest decile cannot move into a higher decile, but can

only stay in the highest or fall to a lower one, this decile measure will understate the effects for

this particular group of students. All the dependent variables used in the analysis are presented

in Table B.2 in Appendix.

Control Variables. To control for observed differences between treatment and control

groups, I chose various explanatory variables in a vector as shown in all the equations. Based

on the existing literature on the determinants and the effects of private tutoring, which were

summarized in Chapter II, I chose variables that were analyzed as important predictors of private

tutoring. The first set of controls includes students’ background characteristics. I included a

student’s gender taking the value 1 if a student is female, 0 male (GENDER) and a student’s

socio-economic status (SES). I constructed a scale for a student’s socio-economic status by

using monthly household income and parental educations levels. First, I standardized the three

variables, monthly household income, father’s education, and mother’s education, and averaged

the three standardized variables to create the SES scale variable. I also included a student’s

hours of self-study per week taking the value 1 if a student spends more than 10 hours per week

for his or her self-study without assistance, 0 otherwise (SELF-STUDY). Self-study would

include homework, reviewing, or previewing academic contents. The reason why I included this

variable as a control is that the variable can be a proxy for students’ academic motivation, which

is unobserved. Students who have higher academic motivation would spend more time on self-

study than their peers with lower academic motivation. In addition, the level of a student’s self-

esteem (SELF-ESTEEM) is also used as one of the student’s background characteristics. Similar

to the SES variable, I constructed a scale that indicates the level of a student’s self-esteem using

15 variables that measure a student’s self-esteem. Among survey questions, there are 15
 69

questions that were designed by the KEDI to measure students’ self-esteem. All students in the

KELS dataset were asked in terms of their self-esteem, and examples of actual questions are

listed in Figure 2. Lastly, average academic score achieved one year prior to the treatment

(PRESCORE) is included to control for students’ academic ability since the influence of private

tutoring on high achievers would be different from the one on low achievers.

Figure 2. Examples of survey questions to measure students' level of self-esteem

There are several variables that indicate school characteristics in the vector X to control

for different school characteristics that could influence students’ academic performance. First, I

included the type of school that a student attends taking the value 1 if a school is private, 0 public

(SCH_TYPE). Research showed that the private school effect on a students’ academic

achievement is different from the public school effect; therefore, different characteristics

between private and public schools need to be controlled. I also included high school track

(SCH_TRACK), taking the value 1 if a high school follows a general track, 0 if a high school
 70

follows a technical track. General track indicates college preparatory high schools, and technical

track is used by vocational schools for students who plan to work after graduation from high

school. Students in a high school that follows a general track are more likely to participate in

private tutoring than their peers in a technical high school; therefore, high school track is an

important predictor of the treatment. Student-teacher ratio (ST_RATIO) is controlled as a proxy

for education quality. The existing literature showed that education quality is one of the

important determinants of private tutoring. Students in schools with lower student-teacher ratio

would have more interaction with their teachers, which allows students to fulfill their educational

needs in schools since they have a low demand for private tutoring compared to students in

schools with higher student-teacher ratio. The student-teacher ratio is calculated by dividing the

total number of students by the number of teachers excluding administrators in a school. The

last set of controls contains regional characteristics. Urban residence (URBAN), which takes the

value 1 if a student lives in urban area, 0 otherwise, is chosen because students who live in an

urban area have better access to private tutoring than students in a rural area. I included a

variable indicating the district with a school choice scheme (SCH_CHOICE), which takes the

value 1 if a student lives in a district with a school choice scheme, 0 otherwise. The existing

literature argued that living in a district with a school choice scheme is one of the macro

determinants of participating in private tutoring. I tried to include characteristics of students,

schools, and regions as controls, which were analyzed as important determinants of private

tutoring in the exiting literature summarized in Chapter II. These control variables are also listed

in Table B.3 in Appendix.

 71

4.4 Descriptive Statistics of the Sample

For the control variables used in estimations, Table 16 shows descriptive statistics for

grade 7 and grade 10 by participation in private tutoring. I chose variables measured in grade 7

and grade 10 for control variables because these two grades are the beginning grades of middle

school and high school, respectively. As already explained, the Korean school configuration is

different from the school configuration in the U.S.; middle school education starts in grade 7 and

finishes in grade 9, and high school education is from grades 10 to 12. Another reason is that

since I divided the analyses by education level (middle school and high school), I used variables

measured in grade 7 and grade 10 as control variables in the analysis of middle school and high

school, respectively. This dissertation used all students in the KELS data without restricting the

analysis to a certain sub-sample. The data is comprised of 48 percent females and 52 percent

males. Female students are less likely to take private tutoring than male students; on average, 46

percent of female students report that they have participated in private tutoring in any subject and

type, while 52 percent of female students among students have not participated in private

tutoring. As previous research has shown, students with higher socio-economic status, academic

motivation (hours of self-study), self-esteem, and previous academic achievement are more

highly represented in private tutoring. Interestingly, among students who take private tutoring,

15 percent of students in grade 7 and 40 percent of students in grade 10 report that they study by

themselves more than 10 hours per week, but only 7 percent of 7th graders and 12 percent of

10th graders report to do so among students who do not take private tutoring. Students who

participate in private tutoring show higher self-esteem than those who do not. The average z-

scores of self-esteem for students who participate in private tutoring are 0.08 in grade 7 and 0.11

in grade 10, but the average scores for students who do not participate in private tutoring are

-0.19 and -0.24 in grades 7 and 10, respectively. Students who participate in tutoring show
 72

higher academic achievement than students who do not participate. Pre-score used in grade 7 is

the academic ranking from 1 to 9 reported by teachers, and there are 0.92 differences in ranking

between two groups. In grade 10, the academic achievement measured in grade 9, which is a

raw score, is used as a pre-score for students in grade 10. Like in grade 7, the average academic

achievement for students who take tutoring is 16.41 points higher than the average achievement

for students who do not take tutoring.

In terms of school characteristics, students assisted by private tutoring are more likely to

attend private schools, come from a school that has a higher student-teacher ratio, and attend a

school that follows a general track. Specifically, there is no statistically significant difference in

school type in grade 7, but in grade 10, tutored students are more likely to attend private schools

than non-tutored students; 49 percent of tutored students report that they attend private schools,

while 44 percent of non-tutored students report as such. On average, schools of tutored students

have about two students more per teacher compared to schools of students who do not take

tutoring. In terms of school track, 85 percent of tutored students come from schools that follow a

general track, whereas 49 percent of non-tutored students come from a general track.

Lastly, regarding regional characteristics, students who receive private tutoring are more

likely to live in urban areas than those who do not receive tutoring. About half of students in the

tutored group report that they live in an urban area, whereas only about 35 percent of students

report as such. Students in the tutored group are less likely to live in a district with a school

choice scheme than students in the control group; only 35 percent of students with private

tutoring can choose their schools, but 49 percent of students without tutoring can select their

schools.
 73

Table 16. Descriptive statistics of control variables

Private Tutoring
Variables Average Difference t-value3
yes1 no2
Gender (1=female) 0.48 0.46 0.52 -0.06 -3.78

Socio-economic status (z-score)

Grade 7 -0.02 0.12 -0.40 0.52 21.05
Grade 10 -0.01 0.11 -0.35 0.46 22.07

Hours of self-study (1= >10hrs)

Grade 7 0.13 0.15 0.07 0.08 7.72
Grade 10 0.31 0.40 0.12 0.28 23.13

Self-esteem (z-score)
Grade 7 -0.00 0.08 -0.19 0.27 14.31
Grade 10 0.00 0.11 -0.24 0.35 20.20

Pre-score
Grade 7 5.96 6.24 5.32 0.92 16.16
Grade 10 54.03 59.49 43.08 16.41 27.27

School Type (1=private)

Grade 7 0.20 0.20 0.18 0.02 1.56
Grade 10 0.47 0.49 0.44 0.05 3.36

Student/Teacher Ratio
Grade 7 20.30 20.74 19.16 1.58 11.44
Grade 10 15.88 16.40 14.75 1.65 21.43

School Track (1=general)

Grade 10 0.74 0.85 0.49 0.36 32.38

Urbanicity
Grade 7 0.46 0.49 0.36 0.13 8.53
Grade 10 0.45 0.49 0.37 0.12 9.35

School Choice (1=choice)

Grade 7 0.40 0.35 0.49 -0.14 -9.49
Grade 10 0.39 0.35 0.49 -0.14 -11.00

1
N=4,884 (Grade 7); 4,228 (Grade 10)
2
N=1,341 (Grade 7); 1,909 (Grade 10)
3
t-value for [diff= mean(1) - mean(0)]

Table 17 shows summary statistics of the dependent variables based on tutoring status.

In all grades in middle school, on average, test scores of students who receive private tutoring in
 74

all subjects are higher than the average scores of the sample as well as scores of students who do

not receive tutoring. The same pattern is shown in the CSAT results.

Table 18 presents summary statistics of variables related to private tutoring. Regarding

participation in private tutoring, more than 70 percent of students report that they receive private

tutoring in middle school, and participation rate decreases after students enter high school. On

average, students take 2.32 years of private tutoring in middle school and 1.67 years of tutoring

in high school. About 87 percent of students report that they receive three years of private

tutoring during middle school, and about 60 percent of students receive three years of tutoring

during high school years. A more detailed description on the participation in private tutoring by

subject is provided in Table 18. On average, there are higher participation rates in English and

math tutoring compared to verbal tutoring. The participation rate decreases as students move to

upper grade levels.

The average monthly cost of private tutoring varies by grade. The average monthly cost

of private tutoring is the highest in grade 9 (last year of middle school) followed by grade 10.

About 250 USD and 240 USD per month are spent for a child’s private tutoring in grades 9 and

10, respectively, whereas about 130 USD per month is allocated to private tutoring in grade 7.

Among the average monthly cost of private tutoring, parents spend more money on English and

math tutoring than verbal tutoring. In addition, parents place similar importance on English and

math tutoring when their child is in middle school; the monthly cost of English and math tutoring

is similar during middle school years. However, parents invest more on math tutoring when their

child is in high school. In terms of the proportion of a household monthly income, four to eight

percent of monthly household income is spent solely on private tutoring for one child.
 75

In order to see the different amounts of private tutoring per grade and per subject, Table

18 also presents the weekly hours spent on private tutoring by grade and subject. I provided the

weekly hours of private tutoring during middle school years only because variables that show

this information during high school years have a large number of missing observations. On

average, students spent about 5.72, 6.87, and 8.11 hours per week for private tutoring lessons in

grades 7, 8, and 9, respectively. Among these hours, about 40 percent is spent on English and

math tutoring each, and about 20 percent is spent on verbal tutoring.

This dataset does not have information about the quality of private tutoring, which is also

an important aspect of private tutoring. By dividing the monthly cost of private tutoring by

monthly hours of private tutoring, I constructed a quality measure for private tutoring by grade

and subject. According to the quality measure, which is the hourly cost of private tutoring, on

average, the quality of private tutoring in grade 9 is the highest compared to the quality of private

tutoring in grades 7 and 8; the hourly cost of private tutoring in grades 7, 8, and 9 is about 7 USD,

8 USD, and 10 USD, respectively. This pattern is also shown in private tutoring in each subject.

Among the three subjects, the hourly cost of English tutoring is the highest, followed by math

and verbal tutoring.

These explorations of variables provide an overview of the data that I dealt with and an

opportunity to understand the data. Also, this description gives a detailed picture on the pattern

of private tutoring in Korea. Having this in mind, I proceed further to an in-depth analysis.
Table 17. Descriptive statistics of dependent variables
Private Tutoring
Variables Description Average Difference t-value3
yes1 no2
Dependent Variables
Grade 7
G7VEM_S KEDI test; average score of Verbal, English and Math in grade 7 59.99 63.77 50.97 12.80 22.70
G7VER_S KEDI test; Verbal score in grade 7 65.49 67.96 60.47 7.49 13.52
G7ENG_S KEDI test; English score in grade 7 56.18 60.62 45.42 15.20 22.70
G7MAT_S KEDI test; Math score in grade 7 58.06 62.65 46.80 15.85 22.51
Grade 8
G8VEM_S KEDI test; average score of Verbal, English and Math in grade 8 55.33 58.86 44.57 14.29 24.24
G8VER_S KEDI test; Verbal score in grade 8 59.49 61.79 52.88 8.91 15.56
G8ENG_S KEDI test; English score in grade 8 55.59 59.84 42.71 17.13 23.72
G8MAT_S KEDI test; Math score in grade 8 50.88 54.90 38.21 16.69 23.02
Grade 9
G9VEM_S KEDI test; average score of Verbal, English and Math in grade 9 54.03 57.66 41.87 15.79 23.24
G9VER_S KEDI test; Verbal score in grade 9 56.96 59.07 49.58 9.49 13.89
G9ENG_S KEDI test; English score in grade 9 53.98 58.19 39.46 18.73 22.32
G9MAT_S KEDI test; Math score in grade 9 51.13 55.78 36.50 19.28 24.12
Grade 12
CSAT_avg CSAT; average decile rank of Verbal, English, and Math in grade 12 5.59 6.08 5.23 0.85 16.94
CSAT_V CSAT; decile rank of Verbal achievement in grade 12 5.81 6.11 5.60 0.51 9.19
CSAT_E CSAT; decile rank of English achievement in grade 12 5.55 6.11 5.15 0.96 16.87
CSAT_M CSAT; decile rank of Math achievement in grade 12 5.40 6.00 4.98 1.02 16.73

76
Table 18. Descriptive statistics of variables related to private tutoring

Variable Description of Variable Obs. Mean Std. Dev. Min Max

Participation in Overall PT
PT_G7 Participation in PT in grade 7 (1=yes; 0=no) 6225 0.78 0.41 0 1
PT_G8 Participation in PT in grade 8 (1=yes; 0=no) 6285 0.78 0.41 0 1
PT_G9 Participation in PT in grade 9 (1=yes; 0=no) 4984 0.73 0.44 0 1
PT_G10 Participation in PT in grade 10 (1=yes; 0=no) 6208 0.68 0.46 0 1
PT_G11 Participation in PT in grade 11 (1=yes; 0=no) 5664 0.59 0.49 0 1
PT_G12 Participation in PT in grade 12 (1=yes; 0=no) 5384 0.35 0.47 0 1
TOTALPT_MS Total years of PT in middle school 4211 2.32 0.99 0 3
TOTALPT_HS Total years of PT in high school 4814 1.67 1.11 0 3
PT1yr_ms 1=1 year of private tutoring in middle school; 0= 0 year 834 0.53 0.49 0 1
PT2yr_ms 1=2 years of private tutoring in middle school; 0= 0 year 1158 0.66 0.47 0 1
PT3yr_ms 1=3 years of private tutoring in middle school; 0= 0 year 3001 0.87 0.34 0 1
PT1yr_hs 1=1 year of private tutoring in high school; 0= 0 year 2077 0.52 0.49 0 1
PT2yr_hs 1=2 years of private tutoring in high school; 0= 0 year 2267 0.56 0.49 0 1
PT3yr_hs 1=3 years of private tutoring in high school; 0= 0 year 2444 0.60 0.49 0 1
Participation in Verbal PT
VERPT_G7 Participation in verbal PT in grade 7 (1=yes; 0=no) 6240 0.55 0.50 0 1
VERPT_G8 Participation in verbal PT in grade 8 (1=yes; 0=no) 6270 0.52 0.50 0 1
VERPT_G9 Participation in verbal PT in grade 9 (1=yes; 0=no) 5203 0.56 0.50 0 1
VERPT_G10 Participation in verbal PT in grade 10 (1=yes; 0=no) 6217 0.43 0.49 0 1
VERPT_G11 Participation in verbal PT in grade 11 (1=yes; 0=no) 5670 0.25 0.43 0 1
VERPT_G12 Participation in verbal PT in grade 12 (1=yes; 0=no) 5384 0.11 0.31 0 1
TOTALVERPT_MS Total years of verbal PT in middle school 4405 1.67 1.14 0 3
TOTALVERPT_HS Total years of verbal PT in high school 4825 0.79 0.88 0 3
VERPT1yr_ms 1=1 year of verbal PT in middle school; 0= 0 year 1909 0.49 0.50 0 1
VERPT2yr_ms 1=2 years of verbal PT in middle school; 0= 0 year 2050 0.53 0.50 0 1
VERPT3yr_ms 1=3 years of verbal PT in middle school; 0= 0 year 2384 0.59 0.49 0 1
VERPT1yr_hs 1=1 year of verbal PT in high school; 0= 0 year 3854 0.43 0.49 0 1
VERPT2yr_hs 1=2 years of verbal PT in high school; 0= 0 year 2927 0.25 0.43 0 1
VERPT3yr_hs 1=3 years of verbal PT in high school; 0= 0 year 2450 0.10 0.30 0 1
Participation in English PT
ENGPT_G7 Participation in English PT in grade 7 (1=yes; 0=no) 6242 0.73 0.44 0 1
ENGPT_G8 Participation in English PT in grade 8 (1=yes; 0=no) 6281 0.72 0.45 0 1
ENGPT_G9 Participation in English PT in grade 9 (1=yes; 0=no) 5744 0.73 0.44 0 1
ENGPT_G10 Participation in English PT in grade 10 (1=yes; 0=no) 6218 0.58 0.49 0 1

77
Variable Description of Variable Obs. Mean Std. Dev. Min Max
ENGPT_G11 Participation in English PT in grade 11 (1=yes; 0=no) 5683 0.46 0.50 0 1
ENGPT_G12 Participation in English PT in grade 12 (1=yes; 0=no) 5385 0.22 0.42 0 1
TOTALENGPT_MS Total years of English PT in middle school 4893 2.25 1.03 0 3
TOTALENGPT_HS Total years of English PT in high school 4837 1.29 1.05 0 3
ENGPT1yr_ms 1=1 year of English PT in middle school; 0= 0 year 1076 0.51 0.50 0 1
ENGPT2yr_ms 1=2 years of English PT in middle school; 0= 0 year 1501 0.65 0.48 0 1
ENGPT3yr_ms 1=3 years of English PT in middle school; 0= 0 year 3364 0.84 0.36 0 1
ENGPT1yr_hs 1=1 year of English PT in high school; 0= 0 year 2790 0.50 0.50 0 1
ENGPT2yr_hs 1=2 years of English PT in high school; 0= 0 year 2681 0.47 0.50 0 1
ENGPT3yr_hs 1=3 years of English PT in high school; 0= 0 year 2176 0.35 0.48 0 1
Participation in Math PT
MATHPT_G7 Participation in math PT in grade 7 (1=yes; 0=no) 6243 0.72 0.45 0 1
MATHPT_G8 Participation in math PT in grade 8 (1=yes; 0=no) 6283 0.72 0.45 0 1
MATHPT_G9 Participation in math PT in grade 9 (1=yes; 0=no) 5772 0.74 0.44 0 1
MATHPT_G10 Participation in math PT in grade 10 (1=yes; 0=no) 6217 0.64 0.48 0 1
MATHPT_G11 Participation in math PT in grade 11 (1=yes; 0=no) 5690 0.53 0.50 0 1
MATHPT_G12 Participation in math PT in grade 12 (1=yes; 0=no) 5385 0.27 0.44 0 1
TOTALMATPT_MS Total years of math PT in middle school 4921 2.26 1.02 0 3
TOTALMATPT_HS Total years of math PT in high school 4844 1.47 1.09 0 3
MATHPT1yr_ms 1=1 year of math PT in middle school; 0= 0 year 1063 0.51 0.50 0 1
MATHPT2yr_ms 1=2 years of math PT in middle school; 0= 0 year 1523 0.66 0.47 0 1
MATHPT3yr_ms 1=3 years of math PT in middle school; 0= 0 year 3377 0.84 0.36 0 1
MATHPT1yr_hs 1=1 year of math PT in high school; 0= 0 year 2439 0.51 0.50 0 1
MATHPT2yr_hs 1=2 years of math PT in high school; 0= 0 year 2518 0.52 0.50 0 1
MATHPT3yr_hs 1=3 years of math PT in high school; 0= 0 year 2281 0.47 0.50 0 1

Monthly cost on private tutoring1

PTcost_G7 Monthly cost on PT in grade 7 6909 13.33 17.62 0 250
PTcost_G8 Monthly cost on PT in grade 8 6909 16.90 21.46 0 240
PTcost_G9 Monthly cost on PT in grade 9 6909 25.22 40.02 0 1460
PTcost_G10 Monthly cost on PT in grade 10 6909 24.39 37.39 0 542
PTcost_G11 Monthly cost on PT in grade 11 6909 21.83 37.54 0 900
PTcost_G12 Monthly cost on PT in grade 12 6909 16.11 33.03 0 630
VERPTcost_G7 Monthly cost on verbal PT in grade 7 6909 2.52 5.33 0 100
VERPTcost_G8 Monthly cost on verbal PT in grade 8 6909 2.97 5.51 0 60
VERPTcost_G9 Monthly cost on verbal PT in grade 9 6909 4.26 9.68 0 300
VERPTcost_G10 Monthly cost on verbal PT in grade 10 6909 3.23 9.22 0 171

78
Variable Description of Variable Obs. Mean Std. Dev. Min Max
VERPTcost_G11 Monthly cost on verbal PT in grade 11 6909 2.50 8.03 0 160
VERPTcost_G12 Monthly cost on verbal PT in grade 12 6909 2.49 9.16 0 210
ENGPTcost_G7 Monthly cost on English PT in grade 7 6909 5.68 8.79 0 140
ENGPTcost_G8 Monthly cost on English PT in grade 8 6909 6.95 10.21 0 130
ENGPTcost_G9 Monthly cost on English PT in grade 9 6909 10.18 17.67 0 730
ENGPTcost_G10 Monthly cost on English PT in grade 10 6909 9.58 16.81 0 290
ENGPTcost_G11 Monthly cost on English PT in grade 11 6909 8.12 17.57 0 500
ENGPTcost_G12 Monthly cost on English PT in grade 12 6909 5.61 14.41 0 210
MATHPTcost_G7 Monthly cost on math PT in grade 7 6909 5.13 8.26 0 150
MATHPTcost_G8 Monthly cost on math PT in grade 8 6909 6.97 10.45 0 116
MATHPTcost_G9 Monthly cost on math PT in grade 9 6909 10.78 18.90 0 730
MATHPTcost_G10 Monthly cost on math PT in grade 10 6909 11.58 18.92 0 354
MATHPTcost_G11 Monthly cost on math PT in grade 11 6909 11.21 20.51 0 400
MATHPTcost_G12 Monthly cost on math PT in grade 12 6909 7.99 18.19 0 300

Proportion of PT cost (=Monthly PT cost/Monthly household income)

G7PT_c Proportion of PT cost in monthly income in grade 7 6195 0.04 0.06 0 2.52
G8PT_c Proportion of PT cost in monthly income in grade 8 5652 0.06 0.07 0 1.14
G9PT_c Proportion of PT cost in monthly income in grade 9 5414 0.08 0.27 0 18.25
G10PT_c Proportion of PT cost in monthly income in grade 10 5197 0.07 0.11 0 1.73
G11PT_c Proportion of PT cost in monthly income in grade 11 5362 0.08 0.40 0 24
G12PT_c Proportion of PT cost in monthly income in grade 12 4605 0.07 0.74 0 50

Hours of private tutoring in middle school2

PThour_G7 Weekly hours of PT in grade 7 6909 5.72 6.77 0 90
PThour_G8 Weekly hours of PT in grade 8 6909 6.87 8.01 0 63
PThour_G9 Weekly hours of PT in grade 9 6909 8.11 13.04 0 450
VERPThour_G7 Weekly hours of verbal PT in grade 7 6909 1.30 2.45 0 60
VERPThour_G8 Weekly hours of verbal PT in grade 8 6909 1.51 2.50 0 24
VERPThour_G9 Weekly hours of verbal PT in grade 9 6909 1.75 3.68 0 90
ENGPThour_G7 Weekly hours of English PT in grade 7 6909 2.21 2.88 0 51
ENGPThour_G8 Weekly hours of English PT in grade 8 6909 2.64 3.25 0 31
ENGPThour_G9 Weekly hours of English PT in grade 9 6909 3.09 5.25 0 225
MATHPThour_G7 Weekly hours of math PT in grade 7 6909 2.20 2.96 0 55
MATHPThour_G8 Weekly hours of math PT in grade 8 6909 2.72 3.40 0 43
MATHPThour_G9 Weekly hours of math PT in grade 9 6909 3.26 5.55 0 225

79
Variable Description of Variable Obs. Mean Std. Dev. Min Max
Quality measure of private tutoring: cost per hour (=monthly cost on private tutoring/(weekly hours of private tutoring*4))
PTq_G7 Cost per hour of PT in grade 7 4055 0.79 0.91 0 15
PTq_G8 Cost per hour of PT in grade 8 4012 0.84 0.88 0 18.75
PTq_G9 Cost per hour of PT in grade 9 4132 1.08 1.17 0 20.28
VERPTq_G7 Cost per hour of verbal PT in grade 7 2584 0.56 0.70 0 12
VERPTq_G8 Cost per hour of verbal PT in grade 8 2627 0.61 0.68 0 8.75
VERPTq_G9 Cost per hour of verbal PT in grade 9 2589 0.79 0.90 0 12.5
ENGPTq_G7 Cost per hour of English PT in grade 7 3651 0.77 0.85 0 8.75
ENGPTq_G8 Cost per hour of English PT in grade 8 3787 0.83 0.86 0 10
ENGPTq_G9 Cost per hour of English PT in grade 9 3806 1.07 1.17 0 20.28
MATHPTq_G7 Cost per hour of math PT in grade 7 3592 0.67 0.71 0 7.5
MATHPTq_G8 Cost per hour of math PT in grade 8 3800 0.77 0.78 0 8.33
MATHPTq_G9 Cost per hour of math PT in grade 9 3891 1.05 1.11 0 20.27

Notes: (1) 1 Unit = 10,000 won (approximately 10 USD); (2) 2 hours of tutoring in high school are not presented due to large missing observations

80
 81

4.5 Validity Check for Instrumental Variable

As mentioned in the estimation model, I used an instrumental variable to correct for the

endogeneity problem in estimating the effect of private tutoring on students’ academic

achievement. I chose the proportion of private tutoring cost in monthly household income as an

instrument. The monthly cost of private tutoring is reflected by the monthly fees of tutoring,

which is determined by the market and regulated by government rules based on local living

standards. In addition, the fee for private tutoring is an important determinant whether to

participate in private tutoring. Due to data limitations, instead of using the average fee of private

tutoring, which is exogenously determined, the cost of private tutoring as a proportion of a

monthly household income is used as an alternative because it is largely determined by the

private tutoring fee. However, since this instrument is subject to many criticisms, I will discuss

the assumptions for the instrumental variable to check whether these are met with this instrument.

In addition to the assumptions, I conducted several tests to check the power of the instrument.

To test whether the instrument for private tutoring is valid for this analysis, it is important

to test several assumptions of the instrument. The first assumption is whether the instrument is

randomly assigned to students (Angrist, Imbens & Rubin, 1996). This assumption is important

because if the instrument is not randomly assigned, it is likely to be correlated with particular

observed and unobserved characteristics of students, which create bias in estimation. This

assumption can be tested by looking at the degree of association between the instrument and

personal characteristics of students to see whether there is any selectivity of the instrument.

Since there are several versions of the instrument depending on grade, education level, and

subject, I present the result of correlation analysis using one of the instruments as an example as

shown in Table 19. The association between a proportion of private tutoring costs in monthly
 82

household income and personal characteristics used in Xk is less than 0.1, which is low. The

correlation between other versions of the instrument and covariates is also minimal (the results

are not provided in the text).

Table 19. Correlation coefficients between the instrument and covariates

Instrument
(Proportion of PT cost in monthly
household income)
URBANICITY 0.076
GENDER -0.003
SES 0.049
SCHOOL TYPE -0.010
STU-TEA RATIO 0.061
SCHOOL CHOICE -0.072
PRESCORE 0.073
SELF-STUDY 0.042
SELF ESTEEM 0.069

Secondly, the instrument must only affect the outcome through the treatment, which is

often called the exclusion restriction. In other words, there should be a non-zero correlation

between the instrument and the treatment variable, and the instrument should be uncorrelated

with the post-treatment outcome and error term to be a valid instrument. Table 20 shows results

from the first-stage equation that indicates the high correlation between the instrument and the

treatment. The coefficient of the instrument is positive and significant at 0.1 percent with a

coefficient of 7.98 and a standard error of 0.27. Even after including the control variables in the

first-stage equation, the coefficient is significant at the 0.1 percent level. This indicates that there

is a statistically significant correlation between the instrument (the proportion of private tutoring

costs in a monthly household income) and the participation in private tutoring. I also checked

the correlation between the instrument and the dependent variable to test whether the instrument

has a small or zero correlation with the dependent variable. I found that there is almost zero
 83

correlation between them (0.09). Therefore these statistics confirm that this instrument meets the

exclusion restriction.

Table 20. The first-stage equation in 2SLS

Without With
covariates covariates

Instrument (PT cost as a

proportion of a monthly 2.23*** 2.06***
household income)
(0.076) (0.074)

R-squared 0.13 0.21

Observations 5,755 5,491
F-statistics 852.44 146.82
*** p<0.001, ** p<0.01, * p<0.05

Moreover, the instrument also should meet monotonicity assumption, which means that

there should be no defiers (Angrist, Imbens & Rubin, 1996). In other words, no students would

have received private tutoring if they cannot afford the costs of private tutoring but would not

have received it if they could afford it. It is impossible to directly test this assumption. However,

this assumption seems to be quite plausible because private tutoring is often categorized as

normal goods that are defined as goods for which demand increases as consumer income rises.

The last assumption is the Stable Unit Treatment Value Assumption (SUTVA). The

SUTVA implies that potential outcomes for each person should not be related to the treatment

status of other people (Angrist, Imbens & Rubin, 1996). A possible scenario would be that a

student’s private tutoring participation would encourage his or her peer’s participation in private

tutoring, which may indirectly affect the peer’s academic achievement (Kim, 2007b). This

possibility would generate problems in estimation when a large number of students in the sample

are gathered in the same classroom. However, it is impossible to test this scenario using the

KELS data because one or two students in a classroom are sampled, on average.
 84

This study performed the Durbin and Wu-Hausman tests to check the superiority of using

2SLS over the OLS method. The null hypothesis is the regressor is in fact exogenous, which

means that the OLS estimator is more efficient. The Durbin chi-square is 27.24 with p-value of

0.000, and the Wu-Hausman F-statistic is 27.32 with p-value of 0.000. As shown, both test

statistics are highly significant, so I reject the null of exogeneity, which means that the regressor

is endogenous. These test confirmed that the IV estimate is more superior than the OLS estimate

in this study.

Tests for a Weak Instrument

To test the power of the instrument variable, I conducted “estat firststage” in STATA,

which provides several statistics to measure the relevance of the excluded exogenous variable.

The R-squared and adjusted R-squared of the first-stage regression13 are 32.9 and 32.7,

respectively, which are large enough. Also, the F-statistics14 is 719.41, which far exceeds 10 for

inference based on the 2SLS estimator to be reliable when there is one endogenous regressor.

The F-statistic is also statistically significant at the 0.1 percent level. The minimum eigenvalue

statistic (Cragg & Donald, 1993) is a further test of a weak instrument. Since the minimum

eigenvalue statistic is 719.41, which exceeds the critical value (16.38), I can reject the null

hypothesis of a weak instrument. Based on these several tests, the instrument, the proportion of

a student’s private tutoring cost in monthly household income, is proved as an appropriate

instrument for this analysis.

13
Higher values purportedly indicate stronger instruments, and instrumental-variables estimators exhibit less bias
when the instruments are strongly correlated with the endogenous variable.
14
The F statistic is the joint significance of α1, α2, α3, …, αk in the equation 4.2, the coefficients on the additional
instruments.
 85

4.6 Empirical Results

This section is broadly divided into two parts: private tutoring in middle school and high

school years. The results on the effects of private tutoring during middle school years consist of

two sub-sections, which are 1) short-term effects (effects within one to three years after private

tutoring participation) using the scores of the KEDI achievement tests and 2) long-term effects

(effects within four to six years after the participation) using the results from the CSAT. In each

sub-section, the results for participating in private tutoring in any subject are reported, followed

by the results of private tutoring by academic subject using two types of private tutoring

variables as described in Section 4.3. The second part regarding the effects of private tutoring in

high school follows a similar structure to the first part, but only the results of short-term effects

are available in the absence of long-term effects because the KELS data used in this study does

not provide information after high school graduation. There are two reasons why I separated

analyses by education level instead of combining tutoring in middle and high schools as a whole.

First, as previously described, the characteristics of private tutoring in middle school are

somewhat different from the characteristics of private tutoring in high school. Private tutoring in

middle school mimics formal schooling, whereas private tutoring in high school is largely

focused on preparation for the CSAT. Secondly, due to data limitation, the CSAT decile rank is

the only achievement measure in high school level in the dataset and there are no other

achievement measures during high school years. Therefore, it is impossible to observe the

cumulative effect (combining middle and high school private tutoring) of private tutoring on

academic achievement in high school. Figure 3 may help understand the structure of this section.

The results of analysis are presented based on this structure.

 86

Figure 3. Structure of empirical analyses

PT in any subject

Verbal PT
Short-term Effects
English PT

Math PT
Middle School PT
PT in any subject

Verbal PT
Long-term Effects
English PT
Private Tutoring
(PT)
Math PT

PT in any subject

Verbal PT
High School PT Short-term Effects
English PT

Math PT

Private Tutoring in Middle School: Short-term Effects

Cumulative Effects

Using a set of dichotomous measures of private tutoring, which represents years of

private tutoring in middle school, I begin by reporting the effects of one year of tutoring on

students’ academic achievement in middle school as presented in Table 21. The result from the

OLS method suggests that taking one year of private tutoring does not have a statistically

significant relationship with academic achievement in grade 9 (the last year in middle school)

after controlling for students’ family, school, and regional characteristics. In other words, there

is no statistically significant difference in average academic achievement between students with

 87

one year of tutoring and those who do not receive tutoring. The results from the IV and PSM

approaches also suggest that one year of private tutoring does not affect students’ academic

achievement. Thus, these results suggest that one year of tutoring is not effective in increasing

students’ academic achievement.

Table 21. Effects of one year of private tutoring on academic achievement in grade 9
PSM
OLS IV Regression-
Difference-in-
adjusted
means
VARIABLES Matching

1 year of PT in middle school -0.03 0.16 0.059 0.06

(0.046) (0.131) (0.079) (0.062)

Covariates ✓ ✓ ✓ ✓
Observations 778 521 778 615
R-squared 0.261 0.294 0.029 0.258
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VEM_Z (average score in grade 9); (2) R-squared for difference-in-means is
Pseudo R-squared. (3) Complete table for OLS and IV are provided in Table C.1 in Appendix.

Table 22. Effects of two years of private tutoring on academic achievement in grade 9
PSM
OLS IV Regression-
Difference-in-
adjusted
means
VARIABLES Matching

2 years of PT in middle school 0.19*** 0.25 0.12 0.15*

(0.048) (0.132) (0.070) (0.060)

Covariates ✓ ✓ ✓ ✓
Observations 1,084 763 1,084 968
R-squared 0.240 0.257 0.082 0.308
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VEM_Z (average score in grade 9); (2) R-squared for difference-in-means is
Pseudo R-squared. (3) Complete table for OLS and IV are provided in Table C.2 in Appendix.

However, taking two years of private tutoring show a positive and significant impact on

academic achievement in grade 9, according to the OLS estimate as presented in Table 22. The

achievement score for students with two years of tutoring is higher by 0.19 standard deviations

than the score for non-tutored students. Using the IV method, even though the magnitude of the
 88

estimate is larger than the OLS estimate, the estimate is no longer statistically significant. The

result from the PSM method (difference-in-means) is consistent with the one from the IV method,

but the effect size is half of the IV estimate. The regression-adjusted matching approach,

however, draws a similar result with OLS; taking two years of private tutoring increases

academic achievement by 0.15 standard deviations, and this estimate is statistically significant at

the 5 percent level. The four estimates are not consistent in terms of both significance and the

magnitude of the estimate; therefore, the impact of two years of private tutoring is inconclusive

using three approaches.

Table 23. Effects of three years of private tutoring on academic achievement in grade 9
PSM
OLS IV Regression-
Difference-in-
adjusted
means
VARIABLES Matching

3 years of PT in middle school 0.38*** 0.54*** 0.29** 0.31***

(0.045) (0.133) (0.109) (0.094)

Covariates ✓ ✓ ✓ ✓
Observations 2,854 2,252 2,854 2,749
R-squared 0.294 0.267 0.260 0.350
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VEM_Z (average score in grade 9); (2) R-squared for difference-in-means is
Pseudo R-squared. (3) Complete table for OLS and IV are provided in Table C.3 in Appendix.

Table 23 presents the effect of three years of private tutoring on student’s academic

achievement. According to the OLS estimate, students with three years of tutoring show higher

achievement by 0.38 standard deviations than non-tutored students. This estimate is statistically

significant at the 0.1 percent level. Also, the size of the effect is almost twice as large as the one

for two years of private tutoring. Taking account of the IV approach, the effect of three years of

private tutoring is still significant with a larger effect size. Taking three years of tutoring

increases academic achievement by 0.54 standard deviations compared to no tutoring. The

 89

estimates from PSM also show positive and statistically significant effects of three years of

private tutoring. However, the effect sizes are much smaller than the ones from the IV and OLS

methods. According to difference-in-means, students with three years of tutoring show higher

achievement score by 0.29 compared to their peers who have similar characteristics but no

tutoring. The regression-adjusted matching shows a similar result. Based on these results, it is

conclusive that taking three years of private tutoring has a positive and significant impact on

students’ academic achievement, but the magnitude of the effect is inconclusive.

Instead of using the overall private tutoring as a treatment, I also explored the effect of

private tutoring in each academic subject (verbal, English, and math). In terms of verbal tutoring,

Table 24, 25, and 26 present the effects of taking one, two, and three years of verbal private

tutoring, respectively. The OLS estimate suggests that taking one year of verbal tutoring has a

negative and significant effect on verbal score in grade 9. Students with one year of verbal

tutoring show 0.17 standard deviations lower verbal achievement than students without verbal

tutoring, and it is statistically significant at the 0.1 percent level. Taking account of the IV

approach, the estimate is no longer significant even though it is still negative. The PSM results

are similar to the OLS results. The results in Table 25 and 26 suggest that two and three years of

verbal tutoring do not have statistically significant influence on verbal achievement in grade 9.

Even though all estimates are not significant, almost all are negative, which indicates a negative

relationship between verbal tutoring and verbal achievement. This result is striking because

verbal private tutoring (reading and writing) is one of the popular subjects for students who take

private tutoring as presented in Table 9.

 90

Table 24. Effects of one year of verbal tutoring on verbal achievement in grade 9
PSM
OLS IV Difference-in- Regression-
VARIABLES means adjusted Matching

1 year of Verbal PT
-0.17*** -0.07 -0.22*** -0.22***
in middle school
(0.043) (0.121) (0.066) (0.058)

Covariates ✓ ✓ ✓ ✓
Observations 1,768 1,304 1,768 1,345
R-squared 0.249 0.259 0.021 0.256
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VER_Z (verbal score in grade 9); (2) R-squared for difference-in-means is Pseudo R-
squared. (3) Complete table for OLS and IV are provided in Table C.4 in Appendix.

Table 25. Effects of two years of verbal tutoring on verbal achievement in grade 9
PSM
OLS IV Difference-in- Regression-
VARIABLES means adjusted Matching

2 years of Verbal PT
-0.05 -0.11 -0.064 -0.062
in middle school
(0.042) (0.087) (0.065) (0.058)

Covariates ✓ ✓ ✓ ✓
Observations 1,900 1,394 1,900 1,497
R-squared 0.215 0.207 0.053 0.227
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VER_Z (verbal score in grade 9); (2) R-squared for difference-in-means is Pseudo R-
squared. (3) Complete table for OLS and IV are provided in Table C.5 in Appendix.

Table 26. Effects of three years of verbal tutoring on verbal achievement in grade 9
PSM
OLS IV Difference-in- Regression-
VARIABLES means adjusted Matching

3 years of Verbal PT 0.02 -0.12 -0.06 -0.01

in middle school
(0.040) (0.083) (0.063) (0.057)

Covariates ✓ ✓ ✓ ✓
Observations 2,207 1,714 2,207 1,821
R-squared 0.203 0.194 0.095 0.198
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VER_Z (verbal score in grade 9); (2) R-squared for difference-in-means is Pseudo R-
squared. (3) Complete table for OLS and IV are provided in Table C.6 in Appendix.
 91

The cumulative effects of English tutoring are different from the ones of verbal tutoring

as presented in Table 27, 28, and 29. One year of English tutoring does not have statistically

significant effects on English achievement in grade 9 based on all three approaches. However,

according to the OLS estimate, two years of English tutoring increases English achievement by

0.23 standard deviations compared to no tutoring. This estimate is statistically significant at the

0.1 percent level. While the estimates from the PSM methods are similar to the OLS estimates in

terms of significance and magnitude, the IV estimate is much smaller and insignificant. Table 29

presents the effects of three years of English tutoring. The OLS estimate for three years of

English tutoring is larger than the estimate for one year and two years of English tutoring.

Students with three years of English tutoring have 0.41 standard deviations higher English

achievement compared to those without English tutoring, and it is statistically significant at the

0.1 percent level. The results from the PSM are similar to the ones from the OLS, while the IV

estimate is more than twice as large as the OLS and IV estimates. The IV results suggest that

three years of English tutoring increases English achievement by 1.06 standard deviations. Even

though the magnitude of the estimates is not consistent using the three methods, it is concluded

that three years of English tutoring has positive and significant effects on English achievement.

Table 27. Effects of one year of English tutoring on English achievement in grade 9
PSM
OLS IV Difference-in- Regression-adjusted
VARIABLES means Matching

1 year of English PT 0.05 0.24 0.04 0.04

in middle school
(0.050) (0.156) (0.082) (0.067)

Covariates ✓ ✓ ✓ ✓
Observations 993 687 993 781
R-squared 0.231 0.274 0.036 0.291
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9ENG_Z (English score in grade 9); (2) R-squared for difference-in-means is Pseudo R-
squared. (3) Complete table for OLS and IV are provided in Table C.7 in Appendix.
 92

Table 28. Effects of two years of English tutoring on English achievement in grade 9
PSM
OLS IV Difference-in- Regression-adjusted
VARIABLES means Matching

2 years of English PT
0.23*** 0.11 0.33*** 0.34***
in middle school
(0.049) (0.293) (0.083) (0.073)

Covariates ✓ ✓ ✓ ✓
Observations 1,377 993 1,377 1,197
R-squared 0.274 0.270 0.111 0.269
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9ENG_Z (English score in grade 9); (2) R-squared for difference-in-means is Pseudo R-
squared. (3) Complete table for OLS and IV are provided in Table C.8 in Appendix.

Table 29. Effects of three years of English tutoring on English achievement in grade 9
PSM
OLS IV Difference-in- Regression-adjusted
VARIABLES means Matching

3 years of English PT
0.41*** 1.06*** 0.37*** 0.42***
in middle school
(0.046) (0.137) (0.101) (0.085)

Covariates ✓ ✓ ✓ ✓
Observations 3,142 2,492 3,142 2,996
R-squared 0.310 0.252 0.249 0.374
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9ENG_Z (English score in grade 9); (2) R-squared for difference-in-means is Pseudo R-
squared. (3) Complete table for OLS and IV are provided in Table C.9 in Appendix.

The cumulative effects of math tutoring are similar to the ones of English tutoring. The

results in Table 30 suggest that there is no statistically significant effect of one year of math

tutoring on math achievement in grade 9. However, the OLS results in Table 31 show that two

years of math tutoring increases math achievement by 0.29 standard deviations and it is

statistically significant at the 0.1 percent level. Similarly, the PSM estimates are 0.21 standard

deviations and they are statistically significant at the 1 percent level. Adversely, the IV estimates

indicate that there is no significant difference in math achievement between students with two

years of math tutoring and non-tutored students. All estimates for the effects of three years of
 93

math tutoring are positive and statistically significant as presented in Table 32. According to the

OLS estimate, students with three years of math tutoring show 0.46 standard deviations higher

math achievement than students with no math tutoring. Taking account of the IV approach, the

effect size is almost twice of the OLS estimate; there are 0.88 standard deviations differences in

math achievement between students with three years of math tutoring and those without math

tutoring. However, the estimates from the PSM methods are similar to the OLS estimates but

much smaller than the IV estimates. Similar to the cumulative effects of English tutoring, it is

evident that there are positive and significant effects of three years of math tutoring on math

achievement, but the magnitude of the effects is inconclusive. In summary, it is concluded that

there are positive cumulative effects of taking three years of English and math private tutoring on

students’ achievement in English and math, respectively, whereas one year and two years of

English and math tutoring do not show consistent results. Moreover, verbal private tutoring does

not show statistically significant cumulative effects.

Table 30. Effects of one year of math tutoring on math achievement in grade 9
PSM
OLS IV Difference-in- Regression-adjusted
VARIABLES means Matching

1 year of Math PT in middle school 0.05 0.21 0.09 0.07

(0.048) (0.128) (0.081) (0.061)

Covariates ✓ ✓ ✓ ✓
Observations 992 693 992 779
R-squared 0.242 0.272 0.032 0.278
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9MAT_Z (Math score in grade 9); (2) R-squared for difference-in-means is Pseudo R-squared.
(3) Complete table for OLS and IV are provided in Table C.10 in Appendix.
 94

Table 31. Effects of two years of math tutoring on math achievement in grade 9
PSM
OLS IV Difference-in- Regression-adjusted
VARIABLES means Matching

2 years of Math PT in middle

0.29*** 0.01 0.21** 0.21**
school
(0.049) (0.307) (0.076) (0.065)

Covariates ✓ ✓ ✓ ✓
Observations 1,423 1,031 1,423 1,253
R-squared 0.244 0.245 0.095 0.253
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9MAT_Z (Math score in grade 9); (2) R-squared for difference-in-means is Pseudo R-squared.
(3) Complete table for OLS and IV are provided in Table C.11 in Appendix.

Table 32. Effects of three years of math tutoring on math achievement in grade 9
PSM
OLS IV Difference-in- Regression-adjusted
VARIABLES means Matching

3 years of Math PT in middle

0.46*** 0.88*** 0.31*** 0.35***
school
(0.048) (0.136) (0.084) (0.081)

Covariates ✓ ✓ ✓ ✓
Observations 3,198 2,531 3,198 3,050
R-squared 0.254 0.215 0.223 0.314
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9MAT_Z (Math score in grade 9); (2) R-squared for difference-in-means is Pseudo R-squared.
(3) Complete table for OLS and IV are provided in Table C.12 in Appendix.

Single-Year Effects

The use of measures of private tutoring participation provides information about the most

effective time to take private tutoring to increase students’ academic achievement. This analysis

may uncover the extent of effect of taking private tutoring at each grade level. Many parents

believe that early exposure to private tutoring helps students understand school curriculum in a

better and faster way and increases academic confidence, which directly affects academic

performance (Lee at al., 2004). In order to quantify this belief, I investigated the effect of private

tutoring at each grade level. In a vector of control variables in the estimation equations, I
 95

included private tutoring experience in a previous year and a year after the participation to isolate

the effect of private tutoring in a particular grade that I estimated.

Table 33, 34, and 35 present empirical results on the effects of private tutoring in grades

7, 8, and 9, respectively. The OLS estimates in column 1, 4, and 7 of Table 33 explain that

taking private tutoring in grade 7 has a positive effect on their overall academic achievement in

the first three years after taking private tutoring. A student who participates in private tutoring

during the first grade in middle school scores 0.22 standard deviations higher than a student who

does not participate in private tutoring in the same grade. Even though the effect remains in

grade 8 as presented in column 4, the extent of the effect is minimal (0.09 standard deviation

increase in overall academic achievement). Moreover, the effect of private tutoring in grade 7

disappears in the second year after the participation. However, the story is different when the IV

estimation method is used. Taking account of the IV approach, the size of the IV estimates is

more than twice as large as the OLS estimates. In the year of participation, a student with private

tutoring scores more than half of a standard deviation (0.53) higher than a student with no private

tutoring. In the first and second year after the participation, the effect sizes decrease but they are

still sizable: 0.33 and 0.22 standard deviations differences between a student with private

tutoring and his or her counterpart in year 1 and 2, respectively. All these IV estimates are

statistically significant. When analyzing with the PSM method (difference-in-means), on

average, the PSM estimates are larger than the OLS estimates, but smaller than the IV estimates.

However, there is a consistent pattern in the IV and PSM estimations, where private tutoring in

grade 7 has positive and statistically significant influence on academic achievement in middle

school, but the magnitude of the influence decreases as time passes. Since there is a large
 96

difference in the size of the estimates between the IV and PSM estimates, the magnitude of the

effects is still inconclusive.

Table 33. Effects of private tutoring in grade 7 on academic achievement for 3 years
Year 0 Year 1 Year 2
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

PT in grade 7 0.22*** 0.53*** 0.21*** 0.09** 0.33*** 0.19*** 0.05 0.22* 0.12*
(0.022) (0.066) (0.035) (0.026) (0.076) (0.044) (0.033) (0.094) (0.059)

Student Characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Family Characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
School Characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Regional Characteristics ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Private tutoring in G8 ✓ ✓ ✓ ✓ ✓ ✓
Private tutoring in G9 ✓ ✓ ✓

Observations 5,895 5,473 5,895 5,380 5,022 5,380 3,990 3,716 3,990
R-squared 0.444 0.419 0.105 0.364 0.351 0.206 0.315 0.307 0.246
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G7VEM_Z for Year 0; G8VEM_Z for Year 1; G9VEM_Z for Year 2; (2) Complete table for OLS and IV
are provided in Table C.13 in Appendix.

In terms of private tutoring in grade 8 in Table 34, taking tutoring also increases the

overall academic achievement in grade 8 by 0.27 standard deviations when estimating with the

IV method, but the extent of effect is much smaller than the one for private tutoring in grade 7.

In addition, the effect does not last in the next year, according to the IV estimation. However,

the PSM estimates suggest that even though the estimate (0.18) is smaller than the IV estimate

(0.27), the significant effect stays for two years and the effect size in year 1 (0.21) is larger than

the one in year 0 (0.18). Lastly, as shown in Table 35, participating in private tutoring in grade 9

increases achievement by about one-fifth of a standard deviation (0.19), according to the IV

estimation, which also implies that its effect is much smaller than the effects of tutoring in earlier

grades. The PSM estimate also suggests that students who take tutoring in grade 9 show 0.22

standard deviations higher academic achievement than those who do not take. Therefore, the
 97

effect size is consistent using three methods. In summary, on average, private tutoring in middle

school has a sizable impact on average academic achievement in middle school. In addition,

both the IV and PSM estimates agree that private tutoring in grades 7 and 9 is effective to

increase an average score in grade 9, which is often considered as a barometer for academic

success in high school. Therefore, parents’ belief of the effectiveness of private tutoring is

somewhat true.

Table 34. Effects of private tutoring in grade 8 on academic achievement for 2 years
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES OLS IV PSM OLS IV PSM

PT in grade 8 0.12*** 0.27*** 0.18** 0.09** 0.13 0.21**

(0.020) (0.066) (0.059) (0.028) (0.097) (0.083)

Student Characteristics ✓ ✓ ✓ ✓ ✓ ✓
Family Characteristics ✓ ✓ ✓ ✓ ✓ ✓
School Characteristics ✓ ✓ ✓ ✓ ✓ ✓
Regional Characteristics ✓ ✓ ✓ ✓ ✓ ✓
Private tutoring in G7 ✓ ✓ ✓ ✓ ✓ ✓
Private tutoring in G9 ✓ ✓ ✓

Observations 5,501 4,953 5,501 4,077 3,657 4,077

R-squared 0.633 0.623 0.231 0.540 0.530 0.318
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G8VEM_Z for Year 0; G9VEM_Z for Year 1; (2) Complete table for OLS
and IV are provided in Table C.14 in Appendix.

Table 35. Effects of private tutoring in grade 9 on academic achievement in grade 9

Year 0 (Grade 9)
(1) (2) (2)
VARIABLES OLS IV PSM

PT in grade 9 0.16*** 0.19* 0.22***

(0.023) (0.090) (0.066)

Student Characteristics ✓ ✓ ✓
Family Characteristics ✓ ✓ ✓
School Characteristics ✓ ✓ ✓
Regional Characteristics ✓ ✓ ✓
Private tutoring in G8 ✓ ✓ ✓

Observations 4,378 3,761 4378

R-squared 0.585 0.579 0.261
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VEM_Z (average score in grade 9); (2) Complete table
for OLS and IV are provided in Table C.15 in Appendix.
 98

I investigated these single-year effects more closely by separating the analyses into

academic subjects. Each dichotomous variable for private tutoring represents whether a student

takes private tutoring, and if so, in which subject and during what grade. The results of single-

year analyses by academic subject are presented from Table 36. First, verbal private tutoring in

each of the grades 7, 8, and 9 is not statistically significant in explaining students’ verbal

achievement during middle school years. Taking account of the IV and PSM approaches, results

remain the same. In the case of English private tutoring in each grade level, however, effects are

striking. Taking English private tutoring in grade 7 has an effect on improving English

achievement in grade 7 by 0.25 standard deviations, according to the OLS estimate in column 1

of English tutoring in Table 36. When estimating with the IV method, the estimate is about three

times larger than the OLS estimate, which means English private tutoring in grade 7 increases a

student’s English achievement by 0.67 standard deviations in the same year of starting English

private tutoring. This effect of English private tutoring does not disappear in the next two years,

and the effect sizes are also steady (0.46 in grade 8 and 0.49 in grade 9). However, these large

sizes of the IV estimates bring concerns in estimation because the IV estimates are much larger

than the OLS and PSM estimates. The next two rows indicate the effect of English tutoring in

grade 8 on English achievement in grades 8 and 9. Starting English tutoring in grade 8 increases

English achievement by 0.31 standard deviations in grade 8, but the effect does not stay in grade

9. However, the PSM estimates suggest that taking English tutoring in grade 8 increases English

achievement for two years with a similar size of effects (0.18 in grade 8 and 0.19 in grade 9).

Similarly, English tutoring in grade 9 does not have an impact on English achievement in the

same grade, according to the IV estimates, while the PSM estimate suggests that it increases

English scores by 0.22 standard deviations, which is larger than the OLS estimate (0.16). Lastly,
 99

results for the effect of math private tutoring in each grade are presented in the bottom part of

Table 36. According to the OLS estimates, math tutoring in grade 7 increases math achievement

by 0.26 and 0.09 standard deviations in grades 7 and 8, respectively, and it does not have any

effect on math achievement in grade 9. The similar pattern is shown when estimating the effects

with the PSM method. However, using the IV estimation method, taking math tutoring in grade

7 boosts math scores by about half of a standard deviation (0.53) in grade 7 and about one-third

of a standard deviation in grade 8 (0.36) and grade 9 (0.34). Therefore, the effects of math

tutoring taken in grade 7 are steady and sizable but smaller than the effects of English private

tutoring. Math private tutoring in grade 8 is also effective in raising math scores in grades 8 and

9. According to the IV estimates, a student with math private tutoring in grade 8 receives a score

of about 0.4 standard deviations higher in both grades 8 and 9 than a student with no math

tutoring in grade 8. The PSM estimates also suggest that math tutoring in grade 8 increases math

scores by 0.31 and 0.19 standard deviations, which are smaller than the IV estimates but larger

than the OLS estimates. Unlike English private tutoring, the effects of math tutoring continue

for two years. Starting math tutoring in grade 9 does not help improve math achievement in the

same grade similar to English tutoring based on the IV estimates, while the PSM estimate

explains that taking math tutoring in grade 9 increases math achievement by 0.30. In summary,

the analyses of single-year effects suggest that private tutoring affects overall academic

achievement in a larger extent when a student takes private tutoring in grades 7 and 9. In

addition, English and math private tutoring in grades 7 and 8 largely contribute to academic

achievement. However, the magnitude of the effects is still questionable because the estimates

from the IV and PSM methods are quite different.

 100

Table 36. Summary of the estimates in single-year effects by academic subject and grade

Grade 7 Grade 8 Grade 9

(1) (2) (2) (3) (4) (4) (5) (6) (6)
Treatment Variable OLS IV PSM OLS IV PSM OLS IV PSM
VERBAL TUTORING
VERBAL PT in G7 -0.01 0.01 0.02 -0.05 -0.08 -0.02 -0.02 -0.12 -0.02
(0.022) (0.056) (0.034) (0.025) (0.071) (0.041) (0.031) (0.087) (0.050)
VERBAL PT in G8 0.04* -0.00 0.04 0.00 -0.11 0.06
(0.021) (0.053) (0.039) (0.028) (0.075) (0.056)
VERBAL PT in G9 -0.00 -0.04 -0.02
(0.027) (0.093) (0.051)
ENGLISH TUTORING
ENGLISH PT in G7 0.25*** 0.67*** 0.24*** 0.12*** 0.46*** 0.11** 0.07* 0.49*** 0.07*
(0.024) (0.077) (0.038) (0.029) (0.090) (0.045) (0.033) (0.105) (0.035)
ENGLISH PT in G8 0.14*** 0.31*** 0.18*** 0.09** 0.18 0.19**
(0.024) (0.073) (0.047) (0.030) (0.102) (0.063)
ENGLISH PT in G9 0.16*** -0.15 0.22**
(0.028) (0.272) (0.067)
MATH TUTORING
MATH PT in G7 0.26*** 0.53*** 0.25*** 0.09** 0.36*** 0.16** 0.04 0.34** 0.07
(0.024) (0.081) (0.039) (0.030) (0.085) (0.047) (0.033) (0.132) (0.056)
MATH PT in G8 0.17*** 0.45*** 0.31*** 0.11*** 0.43*** 0.19*
(0.026) (0.083) (0.054) (0.032) (0.111) (0.080)
MATH PT in G9 0.24*** -0.09 0.30***
(0.031) (0.287) (0.072)
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) all numbers for the OLS and IV methods are the estimates of treatment effect after controlling for all
covariates; (2) the PSM estimate is from difference-in-means method. (3) Complete tables for OLS and IV are provided
in Table D’s in Appendix.
 101

Private Tutoring in Middle School: Long-term Effects

Instead of using achievement scores in middle school years, I used the CSAT results to

observe the long-term effect of private tutoring during middle school years. Due to data

limitation, instead of using a raw score or percentile rank of the CSAT, defile rank is used as a

dependent variable. As noticed, using the decile rank is concerning because it does not have

accurate information on the CSAT achievement, and very small differences in the decile can

make a big difference in getting into the most competitive institutions. I also included variables

that indicate participation in private tutoring during high school years to isolate the impact of

middle school tutoring on the CSAT. However, information on tutoring participation prior to

entering middle school is not included in the models due to data limitation.

In terms of overall private tutoring, there are no cumulative or single-year effects of

taking private tutoring in middle school as presented in Table 37 except for the effect of tutoring

in grade 7. The PSM estimate suggests that taking private tutoring in grade 7 increases average

CSAT scores by 0.33 deciles. As expected, verbal private tutoring is also not statistically

significant in explaining verbal achievement on the CSAT (Table 38). English private tutoring,

however, shows positive and significant cumulative effects as shown in Table 39. The IV

estimate suggests that an English score on the CSAT of a student with three years of English

private tutoring is 1.23 deciles higher than an English score of a student with zero years of

English tutoring. Moreover, taking English tutoring in grade 7 increases English scores on the

CSAT by 0.65 deciles, whereas having English tutoring in later grades does not contribute to the

improvement in English scores. The effects of English tutoring are surprising because taking

three years of English tutoring increases the CSAT English scores by 1.23 deciles and this

increment of score dramatically changes the pool of universities that students can apply to. Since

the prestige of the university that a student attends is regarded as an important determinant of
 102

future success in Korea (Lee, 2006), it seems to be worth investing in private tutoring in English

during middle school years. However, the more surprising result is that math private tutoring

taken in middle school does not help increase CSAT math scores (Table 40) as opposed to the

results of short-term effects of math tutoring presented earlier. Since math is reported as the

most demanded subject of private tutoring as shown in Table 9, this long-term effect of math

private tutoring is astonishing.

Table 37. Long-term effects of private tutoring in middle school on the CSAT
Panel A. Cumulative effects
1 year of PT 2 years of PT 3 years of PT
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.03 0.34 0.09 0.13 0.36 -0.10 0.04 0.48 -0.08
(0.180) (0.482) (0.299) (0.166) (0.410) (0.168) (0.134) (0.456) (0.097)

Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 246 181 246 421 325 421 1,452 1,190 1,434
R-squared 0.275 0.294 0.078 0.181 0.180 0.158 0.318 0.310 0.419

Panel B. Single-year effects

PT in grade 7 PT in grade 8 PT in grade 9
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.14 0.42 0.33* 0.09 -0.18 0.11 -0.07 0.31 0.07
(0.087) (0.274) (0.159) (0.086) (0.293) (0.255) (0.081) (0.385) (0.197)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,905 1,800 1,905 1,934 1,770 1,934 2,021 1,785 2,021
R-squared 0.296 0.287 0.227 0.405 0.403 0.310 0.39 0.378 0.303
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_avg (average CSAT score); (2) Complete tables for OLS and IV are provided in Table E.1 and
E.2 in Appendix.
 103

Table 38. Long-term effects of verbal tutoring in middle school on the CSAT verbal
achievement
Panel A. Cumulative effects
1 year of Verbal PT 2 years of Verbal PT 3 years of Verbal PT
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest -0.17 0.52* -0.32 -0.21* -0.28 -0.20 -0.26** -0.47* -0.29
(0.106) (0.256) (0.181) (0.103) (0.222) (0.178) (0.098) (0.192) (0.169)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 791 635 791 925 722 925 1,114 909 1,114
R-squared 0.254 0.204 0.043 0.222 0.214 0.080 0.212 0.210 0.121

Panel B. Single-year effects

Verbal PT in grade 7 Verbal PT in grade 8 Verbal PT in grade 9
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest -0.04 -0.08 -0.03 -0.04 -0.33 0.05 -0.07 0.07 -0.04
(0.071) (0.188) (0.111) (0.071) (0.176) (0.121) (0.072) (0.290) (0.116)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,032 1,921 2,032 2,064 1,889 2,064 2,044 1,812 2,044
R-squared 0.198 0.194 0.142 0.282 0.278 0.221 0.252 0.243 0.201
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_V (Verbal score on the CSAT); (2) Complete tables for OLS and IV are provided in Table E.3 and
E.4 in Appendix.
 104

Table 39. Long-term effects of English tutoring in middle school on the CSAT English
achievement
Panel A. Cumulative effects
1 year of English PT 2 years of English PT 3 years of English PT
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest -0.18 -0.02 -0.59* -0.13 0.54 -0.53 0.09 1.23** -0.26
(0.180) (0.521) (0.286) (0.163) (0.408) (0.299) (0.132) (0.444) (0.362)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 343 260 343 595 462 595 1,678 1,375 1,678
R-squared 0.203 0.203 0.081 0.204 0.173 0.187 0.257 0.225 0.288

Panel B. Single-year effects

English PT in grade 7 English PT in grade 8 English PT in grade 9
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.10 0.65** 0.2 0.13 0.39 0.15 -0.05 0.51 0.10
(0.085) (0.251) (0.127) (0.090) (0.317) (0.177) (0.092) (0.389) (0.207)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
English PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,302 2,179 2,302 2,339 2,148 2,339 2,319 2,062 2,319
R-squared 0.247 0.229 0.174 0.342 0.342 0.307 0.337 0.326 0.291
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_E (English score on the CSAT); (2) Complete tables for OLS and IV are provided in Table E.5 and
E.6 in Appendix.
 105

Table 40. Long-term effects of math tutoring in middle school on the CSAT math
achievement
Panel A. Cumulative effects
1 year of Math PT 2 years of Math PT 3 years of Math PT
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest -0.04 -0.34 0.01 -0.17 0.13 -0.49 -0.15 0.56 -0.99*
(0.192) (0.475) (0.317) (0.172) (0.442) (0.289) (0.143) (0.461) (0.349)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 344 266 334 594 466 594 1,668 1,369 1,668
R-squared 0.235 0.258 0.045 0.243 0.230 0.148 0.256 0.242 0.263

Panel B. Single-year effects

Math PT in grade 7 Math PT in grade 8 Math PT in grade 9
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.11 -0.13 0.08 -0.07 0.01 -0.14 -0.14 0.44 -0.20
(0.091) (0.450) (0.166) (0.098) (0.362) (0.214) (0.102) (0.437) (0.211)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in high school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,294 2,172 2,294 2,330 2,142 2,330 2,310 2,055 2,310
R-squared 0.238 0.236 0.185 0.298 0.297 0.291 0.296 0.284 0.290
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_M (Math score on the CSAT); (2) Complete tables for OLS and IV are provided in Table E.7
and E.8 in Appendix.
 106

Private Tutoring in High School: Short-term Effects

Instead of using variables indicating private tutoring in middle school, this section

utilizes dichotomous variables of private tutoring in high school in order to observe the

cumulative and single-year effects of private tutoring in high school on the CSAT. Unlike the

analysis of middle school, short-term effects of private tutoring in high school are observed using

the results of the CSAT only because scores of the KEDI test during high school years are not

available. Also, there are no outcome variables after high school graduation to measure a long-

term effect.

Table 41 explains that there are cumulative effects of taking private tutoring in high

school on overall achievement on the CSAT. Even though one year and two years of private

tutoring in high school do not explain the CSAT achievement, three years of private tutoring is

statistically significant in explaining the improvement of the CSAT score. The OLS estimate

suggests that students with three years of private tutoring in high school show higher

achievement on the CSAT by 0.59 deciles compared to non-tutored students. This estimate is

statistically significant at the 0.1 percent level. Moreover, the IV estimate also explains a similar

effect of three years of private tutoring in high school (0.51 deciles) with the 5 percent level

significance. However, there are no consistent single-year effects of high school tutoring as

shown in Panel B in Table 41. These estimates are calculated after controlling for students’

family, school, and regional characteristics, which also include students’ academic achievement

prior to the treatment and private tutoring participation in middle school and high school (for

single-year effects).
 107

Table 41. Cumulative and single-year effects of private tutoring in high school on the
CSAT achievement
Panel A. Cumulative effects
1 year of PT in HS 2 years of PT in HS 3 years of PT in HS
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.06 0.36 0.01 0.29** 0.51 0.26 0.59*** 0.51* 0.30
-0.104 -0.376 (0.177) -0.101 -0.304 (0.239) (0.105) (0.251) (0.324)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 639 465 639 826 627 826 985 782 985
R-squared 0.327 0.319 0.074 0.407 0.402 0.235 0.453 0.442 0.393

Panel B. Single-year effects

PT in grade 10 PT in grade 11 PT in grade 12
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.07 0.20 0.15 0.25*** 0.36 0.26 0.26*** 0.16 0.19
(0.073) (0.335) (0.159) (0.067) (0.283) (0.163) (0.059) (0.145) (0.095)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 10 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 11 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 12 ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,940 1,680 1,940 1,940 1,827 1,940 1,940 1,752 1,940
R-squared 0.421 0.407 0.247 0.421 0.413 0.291 0.421 0.422 0.204
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_avg (average CSAT score); (2) Complete tables for OLS and IV are provided in Table F.1 and
F.2 in Appendix.

In terms of verbal tutoring in high school, there are no significant cumulative effects of

high school tutoring on the CSAT (Table 42). Even though the results suggest that two years of

verbal tutoring in high school increases the CSAT verbal achievement by 0.24 deciles based on

the OLS estimate, the IV and PSM estimates do not show the same result. In terms of the single-

year effects, the OLS and PSM estimates suggest that verbal tutoring in grade 12 leads to an

increase in the CSAT verbal achievement by 0.21 and 0.40 deciles, respectively, whereas the IV

estimate is not statistically significant. Verbal tutoring in grades 10 and 11 is not statistically

significant in explaining the verbal achievement on the CSAT.

 108

Table 42. Cumulative and single-year effects of verbal tutoring in high school on the CSAT
verbal achievement
Panel A. Cumulative effects
1 year of Verbal PT 2 years of Verbal PT 3 years of Verbal PT
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.01 0.54 0.03 0.24** 0.34 0.07 0.23 0.23 0.13
(0.071) (0.464) (0.118) (0.093) (0.240) (0.154) (0.143) (0.258) (0.244)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,596 1,227 1,596 1,208 938 1,208 988 760 988
R-squared 0.256 0.230 0.045 0.273 0.275 0.120 0.290 0.270 0.228

Panel B. Single-year effects

Verbal PT in grade 10 Verbal PT in grade 11 Verbal PT in grade 12
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.00 0.11 0.11 0.13 0.05 0.26 0.21* 0.09 0.40**
(0.064) (0.291) (0.112) (0.075) (0.279) (0.138) (0.094) (0.260) (0.166)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 10 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 11 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 12 ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,070 1,790 2,070 2,070 1,952 2,070 2,070 1,870 2,070
R-squared 0.269 0.261 0.101 0.269 0.259 0.151 0.269 0.272 0.146
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_V (Verbal score on the CSAT); (2) Complete tables for OLS and IV are provided in Table F.3 and
F.4 in Appendix.

In terms of English tutoring in high school in Table 43, the OLS estimates suggest that

two and three years of English tutoring in high school contribute to the CSAT English

achievement by 0.22 and 0.35 deciles, respectively. However, taking account of the IV and PSM

approaches, the effects are smaller and become insignificant. When observing the single-year

effects, taking English tutoring in grades 11 and 12 is statistically significant in increasing the

CSAT English achievement by 0.18 and 0.19 deciles, respectively, based on the OLS estimates,

whereas the PSM method suggests that only English tutoring in grade 12 is beneficial to students

in increasing English scores on the CSAT by 0.34 deciles. However, none of the IV estimates is
 109

statistically significant in all grade levels in high school. Even though the IV estimates are not

statistically significant, it is still noteworthy to mention the size of several coefficients. Taking

three years of English tutoring in high school increases CSAT English scores by 0.29 deciles

compared to no English tutoring. In addition, the coefficients for taking English tutoring in

grades 10 and 11 show increases in CSAT English scores by 0.26 and 0.33 deciles, respectively.

Even though these estimates have large standard errors that make the coefficients statistically

insignificant, sizes of the coefficients are large enough to mention.

Table 43. Cumulative and single-year effects of English tutoring in high school on the
CSAT English achievement
Panel A. Cumulative effects
1 year of English PT 2 years of English PT 3 years of English PT
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest -0.12 0.41 -0.24 0.22* 0.18 -0.06 0.36*** 0.29 0.13
(0.090) (0.349) (0.155) (0.090) (0.221) (0.162) (0.100) (0.164) (0.084)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,157 878 1,157 1,228 956 1,228 963 757 963
R-squared 0.327 0.298 0.061 0.363 0.352 0.180 0.399 0.377 0.234

Panel B. Single-year effects

English PT in grade 10 English PT in grade 11 English PT in grade 12
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.03 0.26 0.01 0.18** 0.33 0.10 0.19** 0.14 0.34***
(0.067) (0.267) (0.123) (0.065) (0.212) (0.115) (0.068) (0.144) (0.112)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 10 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 11 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 12 ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,344 2,032 2,344 2,344 2,212 2,344 2,344 2,114 2,344
R-squared 0.348 0.336 0.159 0.348 0.335 0.191 0.348 0.349 0.142
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_E (English score on the CSAT); (2) Complete tables for OLS and IV are provided in Table F.5 and F.6
in Appendix.
 110

Table 44. Cumulative and single-year effects of math tutoring in high school on the CSAT
math achievement
Panel A. Cumulative effects
1 year of Math PT 2 years of Math PT 3 years of Math PT
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.20 0.56 0.06 0.73*** 1.01*** 0.65** 1.18*** 1.08*** 0.85***
(0.111) (0.374) (0.187) (0.116) (0.283) (0.230) (0.121) (0.246) (0.263)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Observations 910 680 910 1,083 832 1,083 1,081 854 1,081
R-squared 0.259 0.247 0.077 0.324 0.308 0.237 0.401 0.386 0.383

Panel B. Single-year effects

Math PT in grade 10 Math PT in grade 11 Math PT in grade 12
(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES OLS IV PSM OLS IV PSM OLS IV PSM

Coefficient of interest 0.14 0.14 0.01 0.55*** 0.63** 0.44*** 0.43*** 0.42** 0.37***
(0.082) (0.322) (0.193) (0.076) (0.241) (0.132) (0.072) (0.157) (0.118)

Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 10 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 11 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 12 ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,338 2,033 2,338 2,338 2,209 2,338 2,338 2,117 2,338
R-squared 0.345 0.334 0.222 0.345 0.341 0.272 0.345 0.343 0.204
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_M (Math score on the CSAT); (2) Complete tables for OLS and IV are provided in Table F.7 and F.8
in Appendix.

Unlike verbal and English private tutoring, math private tutoring in high school has

positive and significant effects on the CSAT math achievement as shown in Table 44. In terms

of the cumulative effects, taking two and three years of math tutoring in high school significantly

increases the CSAT math achievement by 0.73 and 1.18 deciles, according to the OLS estimation.

Moreover, these effects are fairly consistent with the estimates from the IV and PSM methods.

The IV method implies that students with two and three years of math tutoring in high school

have higher math achievement on the CSAT by 1.01 and 1.08 deciles, respectively. The PSM

estimates show similar results with slightly smaller effect sizes (0.65 deciles for two years of
 111

math tutoring and 0.85 deciles for three years). Therefore, it is evident that there are statistically

significant cumulative effects of math tutoring in high school. In terms of the single-year effects,

the results in Panel B suggest that math tutoring in grades 11 and 12 is statistically significant in

explaining the improvement of CSAT math scores, and this result is consistent in all three

empirical methods. Taking math tutoring in grade 11 contributes to raising math achievement by

0.55, 0.63, and 0.44 deciles based on the OLS, IV, and PSM methods, respectively. Moreover,

math tutoring in grade 12 has similar effects on math achievement on the CSAT. Thus, while the

effects of verbal and English tutoring in high school are not conclusive using the three methods,

the effect of math tutoring in high school is incontrovertible. Based on these empirical results,

more in-depth discussion should take place as follows.

4.7 Discussion

Participation in private tutoring regardless of subject during middle school shows positive

effects on students’ academic achievement in the last grade of middle school as opposed to the

results obtained by Kang (2005) and Choi (2007). Taking three years of private tutoring in

middle school increases an average score on the KEDI tests by 0.3-0.5 standard deviations when

taking account of the PSM and IV approaches, respectively. Even though this positive effect

was anticipated, the magnitude of the effect is surprisingly small if you take the annual costs of

private tutoring into consideration. According to the descriptive statistics presented in Table 18,

the average monthly cost of private tutoring is about 180,000 Won15. So, annually, on average,

parents of middle school students who take private tutoring spend about 2,160,000 Won, which

15
Take an average of monthly cost in private tutoring in grades 7, 8, and 9 [(13.33+16.90+25.22)/3 = 18.48] in the
unit of 10,000 won.
 112

is approximately equivalent to 2,160 USD. Thus, the effect does not seem substantial when

considering the cost of tutoring.

The results also imply that the effect of private tutoring becomes larger as the number of

years in private tutoring increases. For example, if a student takes three years of private tutoring

during middle school, he or she benefits by 0.5 standard deviations while a student who has

similar characteristics and takes one year of private tutoring benefits by 0.2 standard deviations,

according to the IV estimates. This pattern is the same in the OLS and PSM methods. Therefore,

it is evident that benefits become larger as students continue taking private tutoring.

The results also suggest that among the three years of middle school, on average, taking

private tutoring in grade 7 is most effective in raising average achievement, according to the

results of single-year effects. Taking tutoring in grade 7 increases overall achievement by 0.53

in grade 7, 0.33 in grade 8, and 0.22 in grade 9 based on the IV estimation. The PSM estimation

shows 0.21 in grade 7, 0.19 in grade 8, and 0.12 in grade 9, which imply the effects of tutoring in

grade 7. The sizes of these estimates are bigger than the ones for the effects of tutoring in grades

8 and 9. This result indicates that understanding the basics covered in the first grade in middle

school is crucial to understanding advanced curriculum covered in upper grades. Another

interpretation of this result would be that learning-in-advance in grade 7 seems important to

succeed in schools. As mentioned in Section 3.3 in Chapter III, it is common in hakwon for

students to learn materials that will be taught in formal schools in advance. This argument seems

also convincing because if students listen to lectures in schools after having been exposed to the

content already, it would take less time for them to fully understand compared to those who learn

them for the first time in schools.

 113

Separate analyses by academic subject during middle school also suggest several

conclusions. The results of verbal tutoring are especially striking because verbal subject is one

of the popular subjects for private tutoring. As the results suggest, verbal tutoring is not

statistically significant in explaining verbal achievement in any grades. This also means that

there is no statistical difference between a student with verbal tutoring and a student with formal

schooling only. This ineffectiveness of verbal tutoring may be explained by the pedagogy of

verbal tutoring. Verbal tutoring is mostly designed to practice analytical reading using various

literatures chosen by tutors. However, instead of covering each literature as a whole, an excerpt

of each piece of literature is used to cover as many literatures as possible. From the author’s

experience, therefore, verbal tutoring places more importance on quantity of materials rather than

quality of instruction. This aspect of verbal tutoring may explain why it does not play a

significant effect on verbal achievement.

As opposed to verbal tutoring, English and math tutoring show significant and positive

effects on English and math achievement in middle school. Taking three years of English and

math tutoring increases English and math achievement in grade 9 by 0.41 and 0.46 standard

deviations, respectively, according to the OLS estimates. These OLS estimates are similar to the

PSM results. However, the IV estimates suggest that it contributes to English and math

achievement by 1.06 and 0.88 standard deviations, respectively. Similar to the results from

overall private tutoring, having English tutoring in grade 7 is the most effective in increasing

achievement in middle school based on the IV estimates; English tutoring in grade 7 raises

English achievement by 0.67 in grade 7, 0.46 in grade 8, and 0.49 in grade 9. Even though the

OLS and PSM estimates are smaller than the IV estimates, the effect sizes are larger than the

ones for English tutoring in grades 8 and 9. In addition, according to the IV and PSM results, the
 114

effects of English tutoring in grades 8 and 9 on English achievement in grade 9 are none or

minimal. It means that English tutoring in grades 8 and 9 does not largely contribute to students’

English achievement in grade 9, which is critical for success in high school. In the situation

where about 61 percent of total middle school students report that they take English tutoring,

according to the national statistics, this result is quite striking because it implies that many

parents spend so much money without reaping large benefits from it. Fortunately, math tutoring

is more promising than English tutoring. Math tutoring in both grades 7 and 8 is beneficial to

students’ math achievement in middle school. Unlike English tutoring, however, taking math

tutoring in grade 8 is most effective in raising math achievement followed by math tutoring in

grade 7. Math tutoring in grade 9 is not effective like English tutoring, according to the IV

estimate, whereas the OLS and PSM methods suggest that math tutoring in grade 9 is also

effective. These results also imply that understanding the contents in English and math covered

in early grades in middle school is important for understanding advanced contents in the later

grades in middle school. In addition, the instructional strategy of English and math tutoring,

which is mostly drill and practice, seems to be effective in generating higher scores on the tests.

The less significant long-term effects of middle school tutoring on the university entrance

examination also need to be discussed. There are no positive and significant effects of private

tutoring in middle school on the CSAT except for English tutoring. Especially, overall

participation in private tutoring does not show any effect on the CSAT, which is surprising. The

ultimate goal of Korean parents who spend money on private tutoring in middle school is for

their child to succeed in high-stakes tests such as the CSAT, which is closely related to their

child’s future success in Korea. Existing Korean literature also claims that private tutoring at the

middle school level is not only for improving school grades, but for the long-term development
 115

of academic competence for admission to prestigious universities (Kim, Kang, Park, Lee &

Hwang, 2006). In addition, even though the contents of private tutoring in middle school are not

aligned specifically with curriculum contents in high school, which the CSAT usually tests,

curriculum in middle school provides the foundations of curriculum in high school. Therefore, it

is unexpected to find a lack of measurable effectiveness of middle school private tutoring on the

CSAT. It also implies that the CSAT does not place much attention on fundamental knowledge

that is learned in middle school and it requires advanced knowledge to succeed on the CSAT

instead.

However, English tutoring follows the authors’ expectation; three years of English

tutoring in middle school raises 1.23 deciles on the CSAT English scores. This is a huge effect

that can influence admission to universities. Since the KEDI tests were not administered during

high school years, it is impossible to observe the effect of middle school tutoring on academic

achievement during high school years in these analyses. However, even though the OLS and

PSM estimates do not show significant effects, it is still noteworthy that the effects of English

tutoring in middle school last longer than the effects of verbal and math tutoring.

As opposed to middle school tutoring, the OLS results suggest that high school tutoring

seems effective on CSAT achievement. This positive and significant effect of tutoring in high

school on the CSAT may be explained by the characteristics of high school tutoring. Unlike

middle school tutoring, tutoring during high school is more focused on preparing for the CSAT

(Lee, Park & Lee, 2009). In addition, curriculum in high school is more tightly linked to what

the CSAT measures than curriculum in middle school. However, the IV and PSM estimates do

not show similar results, which bring concerns about the accuracy of the OLS estimates. When

observing the effects by academic subject, only math tutoring in grades 11 and 12 shows
 116

consistently significant effects on the CSAT math achievement. Taking two and three years of

math tutoring raise math achievement on the CSAT by about 1 decile, and math tutoring in

grades 11 and 12 increases math score on the CSAT by about 0.4-0.6 deciles, which are quite

large. However, this large effectiveness of math tutoring also raises concerns about math

education in formal schools. It can be posited that math education in formal schools does not

successfully prepare students to achieve high scores on the CSAT. Another perspective on this is

that a student having math education only in formal schools cannot excel a student with math

tutoring. This argument is connected to the existing literature on private tutoring, which claim

the ineffectiveness of formal education and unmet needs of students and parents from formal

education (Kim & Lee, 2010; Chun et al., 2003; Kim, 2004). Lee and Hong (2001) argued that

this is due to the rigidity of the formal school system in Korea caused by the government’s heavy

regulation and strict control. In addition, Lee, Park, and Lee (2009) claimed that parents believe

that succeeding on the CSAT for entrance to a university is not possible without private tutoring.

Therefore, as the effect of high school tutoring on the CSAT becomes larger, it might be

considered as a possible evidence that formal education is ineffective. In the last chapter, based

on these results and upcoming results obtained from the analyses below, I will provide further

implications for policy makers in education.

This chapter investigated the effect of private tutoring on students’ academic

achievement in various ways. In the next chapter, I will explore how private tutoring affects the

learning environment in formal schooling by using the degree of students’ attention to lessons in

schools. This analysis will provide implications of a larger influence of private tutoring not only

on students, but also on formal schooling.

 117

CHAPTER V. THE IMPACT ON FORMAL SCHOOLING

The second part of the analysis focuses on the impact of private tutoring on formal

schooling. In many studies, the quality of formal schooling is often regarded as one of the

important determinants or causes of taking private tutoring (Kim & Lee, 2010; Buchmann, 1999;

Silova & Bray, 2006). Especially, Silova and Bray (2006) argued that the ineffectiveness of the

public school education provision is found as one of the main determinants of private tutoring.

However, other studies perceive private tutoring as influencing the quality of formal schooling.

For example, tutoring can cause a great lack of interest and attention to lessons in formal schools

on the part of students because they have already covered the topics with tutors (Hussein, 1987;

Nanayakkara & Ranaweera, 1994; Sawada & Kobayashi, 1986). This is a serious issue for

teachers and schools because as more students lose interest in school lessons due to private

tutoring, it dilutes the learning environment of the classroom, which directly affects the quality

of schooling and students’ academic performance. Even though there are a few qualitative

studies that deal with this issue, researchers have not paid much attention to this in a quantitative

way. Therefore, it is not statistically proven whether private tutoring negatively or positively

affects the quality of formal schooling. To fill this gap in research, this chapter focuses on the

influence of private tutoring on the quality of formal schooling. This study is designed to answer

the following research question.

Does private tutoring influence students’ behavior in formal schools that affect the

learning environment in schools? In particular, how is participation in private tutoring

associated with students’ attention to lessons in schools?

 118

As a proxy for measuring the learning environment in schools, the degree of students’

attention to lessons is chosen as a dependent variable. As expected, the estimation of the effects

of private tutoring on the learning environment in formal schooling is also subject to an

endogeneity problem. The possible source of endogeneity would be reverse causality. The

reverse causality means that private tutoring could affect the quality of the learning environment

in formal schooling, and this quality could reversely affect the participation of private tutoring.

In other words, if students are not satisfied with formal schooling due to the unproductive

learning environment, they would tend to find supplementary education such as private tutoring,

to satisfy their desire to learn. This reverse causality would create a problem in measuring the

causal relationship between private tutoring and the quality of formal schooling. I tested this

reverse causality by using school fixed effects to compare estimates with or without school fixed

effects. The estimates presented in the section of empirical results are very similar to the

estimates with school fixed effects (results are not reported in the text). This implies that school

characteristics do not play significant roles in estimation, which also implies that reverse

causality is not a serious issue in this study. Having this threat of endogeneity bias, two quasi-

experimental methods are introduced: Propensity Score Matching and Difference-in-Differences.

5.1 Estimation Models

The main purpose of this section is to observe whether there is any effect of private

tutoring on students’ attention to lessons in formal schools. I began the analysis by conducting

the Ordered Logit estimation.

(5.1) ij 0 1 ij k ij,k ij
 119

where ij is the degree of student i’ attention to lessons in grade j, which is a categorical

variable, PT is a private tutoring variable, and ij,k is a vector of covariates that indicate student

i’s family, school, and regional characteristics in grade j. There is a more detailed description

about how the dependent variable and covariates were constructed in the next section of

description of variables. In this equation, 1 is a coefficient of interest that indicates the effect of

taking private tutoring on students’ attention to lessons after controlling for the students’

background characteristics. However, 1 tends to be biased in the linear estimation because the

dependent variable can affect the independent variable, which is the participation in private

tutoring (reverse causality). In addition, as explained in Chapter I, students who take private

tutoring tend to have different unobserved characteristics from those who do not. These

unobserved variables could also affect the dependent variable (attention to lessons in formal

schools). This complication will make the estimate of the tutoring effect biased. In this context,

one of the popular approaches includes Propensity Score Matching (Rosenbaum & Rubin, 1983).

5.1.1 Propensity Score Matching

Propensity Score Matching (PSM) is one of the quasi-experimental methods, which is

used with observational data in order to detect a causal mechanism. In PSM, rather than

controlling for all the covariates in i,k, students in a treatment group (i.e., students who

participate in private tutoring) are matched with students in a control group (i.e., students who do

not participate in private tutoring) using the propensity score, e(X), which is a numerical

summary of i,k. We can estimate e(X) by using logistic or probit regression where the

dependent variable is the treatment (a dichotomous variable that indicates participation in private

tutoring) and the predictors are all confounding covariates. In the process of matching, a simple
 120

one-to-one nearest neighbor matching with replacement is used among various types of matching

(Dehejia & Wahba, 2002). In this method, each treatment unit can be matched to the nearest

comparison unit, even if a comparison unit is matched more than once. Researchers claim that

this method helps in reducing the bias because it minimizes the propensity score distance

between the treatment unit and the matched comparison unit. Using this method of matching, I

constructed a counterfactual to measure the outcome (i.e., the degree of students’ attention to

lessons in formal schools) that students would have obtained, had they not participated in private

tutoring.

(5.2) e(X) = Pr(PTi=1|Xi,k)

This propensity score, e(X), is a predicted probability for each person that receives the

treatment. Using these two groups that are similar in various ways except for the treatment (the

participation in private tutoring), the mean of the dependent variable (i.e., the degree of attention

to lessons in formal schools) for a treatment group is compared to the one for a control group to

observe the treatment effect. This effect is called the average treatment effect on the treated

(ATT). The mathematical representation of difference-in-means in PSM is as follows.

(5.3) τ = E[Y(1)|Z=1,e(X)] – E[Y(0)|Z=0,e(X)] = YZ=1,e(X) – YZ=0,e(X)

where τ is the treatment effect. In addition to this method of difference-in-means, I also

conducted the regression-adjusted matched estimation, which is running a regression of outcome

on the treatment indicator and confounding covariates using weights to construct the sample to
 121

represent matched group. The regression-adjusted matched estimation also provides a sensitivity

check for estimates obtained from the difference-in-means.

Two estimation methods derive , which is the average treatment effect on the treated.

Specifically, measures the effects of private tutoring on the degree of attention to school

lessons of students who participated in private tutoring. By constructing the counterfactual

group by matching, I compared the mean of the degree of attention to lessons of the treated

group with the one of the control group. This effect provides an overall picture on the average

effect of private tutoring on students’ attention to lessons in the absence of explaining the effect

within each ability group. The effect of private tutoring for high-achieving students may be

different from the one for low-achieving students. To uncover the differential effects of private

tutoring by ability group, I observed the effect by each subgroup of students. I constructed three

subgroups of students using students’ academic standing.

It should be noted that correctly calculating standard errors in PSM is a problem for

several reasons. After matching, the observations are no longer independent of each other even

with regular matching. In other words, if there are correlations between matched pairs, standard

errors are subject to be biased. In addition, matching with replacement creates the additional

complication of including some units multiple times. In order to correct for standard errors,

standard errors are calculated by bootstrapping with 1,000 replications.

However, PSM also bears methodological drawbacks that should be stated explicitly.

Several researchers have criticized that this method still has potential problems that are caused

by unobserved variables because only observed characteristics of the sample are used in the

process of matching (Michalopoulos, Bloom & Hill, 2004). In addition, the method is sensitive

to choices of confounding covariates to calculate propensity score, e(X). Even though this
 122

method carries these methodological disadvantages, it can help reduce large biases compared to

Ordered Logit.

5.1.2 Difference-in-Differences Estimation

Another method to observe the effect of private tutoring on students’ attention to lessons

in formal schools is the difference-in-differences (DID) estimation using a natural experiment.

The natural experiment is that the Seoul metropolitan government enacted the regulation in 2008

that prohibits late instruction in hakwon located in Seoul after 10 pm (Kim & Chang, 2010). The

Seoul metropolitan government enacted this regulation to aim at reducing excessive participation

in private tutoring and to secure students’ time for sleeping and resting. According to the

KNSO’s national statistics, the participation rate had decreased by about 2 percent from 2007 to

2008 as shown in Table 3. Therefore, the DID estimation is used to investigate how the

reduction in private tutoring caused by the regulation affects students’ attention to lessons. The

underlying assumption here is that regulating late instruction at hakwon actually reduces the

participation rate in private tutoring. The national statistics showed the actual decrease in the

participation rate from 2007 to 2008. In addition, Kim and Chang (2010) also found that this

governmental law that regulates the operating hours of hakwon contributed to a decrease in

tutoring hours. However, there may be violations of the law by tutors or hakwon that do not

follow the regulation, which is unlikely to be reported in the national statistics. Although this

national experiment has potential problems of illegal practices, it is reasonable to say that the

absolute amount of time that students spend and the participation rate on private tutoring has

decreased after the regulation. However, even though reducing a few hours a week may not have

much of an effect on the degree of attention to lessons, it is meaningful to investigate the effect
 123

of this regulation because it has had quite a large influence on the tutoring market and reducing

students’ participation in tutoring.

Using this regulation as an “exogenous” shock that acts like a treatment, the DID

estimation compares the dependent variable of students who live in the treated city (i.e., Seoul)

with one of the students who live in another metropolitan city similar to Seoul before and after

the regulation was enforced. However, in the KELS data, the sample size for students who live

in another metropolitan city similar to Seoul, such as Gyeonggi or Busan, is very small. Instead

of using a small sample as a control group, I used the representative sample of students who live

in six metropolitan cities16 in Korea. In the KELS, the “pre-regulation” period is denoted as the

first, second, and third year of the survey covering the period 2005, 2006, and 2007, and the

“post-regulation” period is denoted as fourth, fifth, and sixth year of the survey covering the

period 2008, 2009, and 2010. In this analysis, the difference in outcomes between Seoul and

other metropolitan cities in 2007 (year 0) is compared with the one in 2008 (year 1), 2009 (year 2)

and 2010 (year 3) using the equation below.

(5.4) it 0t 1 it k kit it

it is the dependent variable that indicates the degree of student i’s attention to

lessons in formal schools, and it is a binary indicator whether student i lives in Seoul at

time t. kit is a vector of students’ characteristics in family, school, and region, and it is the

idiosyncratic error. The coefficient 1 is the treatment effect. Since the KELS data is individual-

level longitudinal data, I observed the same students over time unlike the cross-sectional data

16
There are 6 other metropolitan cities in Korea: Busan, Daegu, Incheon, Gwangju, Daejeon, and Ulsan
metropolitan cities.
 124

that has several cohorts of students. Using equation 5.4, I subtracted the results in year 0 from

the ones in year 1, year 2, or year 3 in order to obtain 1 as follows.

(5.5) i 0 1 i k ki i

In addition to the average effect, heterogeneous effects are also observed using three

subgroups of students, constructed by academic achievement. These heterogeneous effects

provide a detailed picture of the effect of this regulation by ability group.

5.2 Variables

Dependent variables. As a dependent variable, a variable indicating the degree of

students’ attention to lessons in formal schools (ATTN) is selected as a proxy for measuring the

quality of formal schooling. This variable is reported by a student’s homeroom teacher in each

grade. The actual question is “How much does this student pay attention to lessons in school?”

This variable consists of five categories (1=having attention to a lesson less than 10 minutes;

2=11-20 minutes; 3=21-30 minutes; 4=31-40 minutes; 5=more than 40 minutes). Therefore, this

is a teacher perception of the attention span of a student. I transformed the dependent variables

into z-scored measures to interpret estimates in a standardized unit. However, this teacher-

reported measure of students’ attention to lessons tends to be subjective because it is only a

teacher’s perception of about 40 students in a classroom without clear criteria to measure their

level of attention. Therefore, this measure has limitation on its accuracy in measuring the actual

attention span of a student.

Treatment variables. Same as the first analysis, treatment variables are a set of

dichotomous variables taking the value 1 if a student participates in any form of private tutoring,
 125

0 otherwise. This variable is measured in each grade by parents (G7PT, G8PT, G9PT, G10PT,

G11PT, and G12PT).

Control variables. To control for observed differences between treatment and control

groups, I included various explanatory variables in a vector . The first set of controls include

student and family characteristics: 1) gender of a student taking the value 1 if a student is female,

0 male (GENDER); 2) household monthly income in the unit of 10,000 Korean Won, which is

approximately 10 USD (HINCOME); 3) father’s education level, which consists of seven

categories (DADEDU; 1=less than elementary school, 2=middle school graduate, 3=high school

graduate; 4=two-year college graduate; 5=Bachelor’s degree; 6= Master’s degree; 7=Ph.D

degree); 4) parental academic aspiration for their children (PARASPIRE), which indicates

parents’ expectation on their children’s final educational level, that is a categorical variable with

seven categories as DADEDU; 5) students’ academic motivation (ACAMOTIV), which

indicates students’ plan for final educational level in the future, that also have the same seven

categories as DADEDU and ACAMOTIV. 6) Students’ verbal, English, and math scores in the

KEDI tests during middle school years are also used to control for students’ previous academic

ability (VER_S, ENG_S, and MAT_S). As school characteristics, similar to the first analysis,

there are three variables: 1) type of school that a student attends taking the value 1 if a school is

private, 0 public (SCH_TYPE); 2) high school track taking the value 1 if a high school follows

general track, 0 if a high school follows a technical track (SCH_TRACK); 3) student-teacher

ratio as a proxy for education quality (ST_RATIO) by subtracting the total number of students

by the total number of teaching faculty in a school. The last set of controls contain regional

characteristics; 1) urban residence taking the value 1 if a student lives in urban area, 0 otherwise

(URBAN); 2) district with a school choice scheme taking the value 1 if a student lives in a
 126

district where he or she could choose a school; 0 if a student is randomly assigned to a school

affected by the School Equalization policy (SCH_CHOICE); 3) residence in Seoul taking the

value 1 if a student lives in Seoul, 0 otherwise (SEOUL). Except GENDER and DADEDU, all

of the variables are measured in each grade as shown in Table 45.

5.3 Estimation and Results

5.3.1 Descriptive Statistics

Tables 45 and 46 present descriptive statistics of the control variables and the dependent

variables, respectively. Male students are more likely to take private tutoring than female

students; 54 percent of students in the group of students who participate in private tutoring

(treated group) are male, while 48 percent of students in the group without private tutoring

(control group) are male students. Households of students with private tutoring, on average,

have much higher monthly income than households of students with no private tutoring. The

difference at most is about 1,880,000 Won, which is approximately equivalent to 1,880 USD for

students in grade 11. Fathers of students taking private tutoring have slightly higher educational

levels than fathers of students without private tutoring. As the existing literature claimed,

parents who support their children with private tutoring tend to have a higher academic

aspiration for their children and, as expected, a similar pattern is observed in terms of students’

academic motivation. In terms of academic achievement in verbal, English, and math, students

who take private tutoring, on average, show higher achievement than those who do not take

tutoring. Students with private tutoring score higher by about 8 raw scores in verbal, and 17 raw

scores in English and math compared to students without private tutoring.

 127

In terms of school characteristics, students assisted by private tutoring are more likely to

attend a school that follows a general track, attend private schools, and come from a school that

has a higher student-teacher ratio. Specifically, more than 80 percent of students who take

private tutoring attend schools that follow a general track, while about half of students who do

not take tutoring do so. There are no statistically significant differences in school type during

middle school in grades 7, 8, and 9, but the difference is significant during the high school years.

On average, schools with students who take tutoring have about two students more per teacher

compared to schools with students who do not take tutoring.

Regarding regional characteristics, students who receive private tutoring are more likely

to live in urban areas than students who do not receive tutoring. About half of students in the

treated group report that they live in an urban area, whereas only about two-fifths of students in

the control group report as such. Students in the treated group are less likely to live in a district

with a school choice scheme than students in the control group; only 35 percent of students who

receive private tutoring have the freedom to choose their high school, but about half of students

who do not receive tutoring have it. This implies that if students do not have a choice in schools,

they are more likely to take tutoring as a choice. Lastly, students who take tutoring are more

likely to live in Seoul than those who do not.

Table 46 presents summary statistics of the dependent variables used in the analysis.

Except for grade 7, students with private tutoring show a higher degree of attention to lessons in

formal schools than those without tutoring. Among seventh graders, the difference between

students in the treatment and control groups is not statistically significant. The overall findings

suggest that tutoring may reduce attention to classroom instruction, but we need to survey this

result through the full statistical equations.

 128

As already explained in Table 18 in the first analysis, during middle school years, more

than 70 percent of students reported that they received private tutoring. However, the

participation rate decreases after students enter high school; 68 percent in grade 10, 59 percent in

grade 11, and 35 percent in grade 12 reported that they currently received private tutoring. This

pattern is similar to the national statistics shown in Table 3.

Table 45. Descriptive statistics of control variables

Private Tutoring
Variables Description Difference t-value3
yes1 no2
Student and Family Characteristics
Gender 1=female; 0=male 0.46 0.52 -0.06 -3.78
Household's Income (unit=10,000 KRW, approximately 10 USD)
G7HINCOME Household's monthly income in grade 7 385.50 265.80 119.70 16.33
G8HINCOME Household's monthly income in grade 8 383.16 246.01 137.15 17.91
G9HINCOME Household's monthly income in grade 9 493.61 310.69 182.92 7.87
G10HINCOME Household's monthly income in grade 10 485.23 304.95 180.28 10.09
G11HINCOME Household's monthly income in grade 11 489.57 301.55 188.02 10.67
G12HINCOME Household's monthly income in grade 12 464.36 312.82 151.54 11.91
Father's Education
dadedu Father's education level (categorical 1/7) 3.86 3.24 0.62 16.70
Parent's Academic Aspiration for their children (categorical)
g7paraspire Parents' plan for child's future education level in G7 4.66 4.26 0.40 13.51
g8paraspire Parents' plan for child's future education level in G8 4.62 4.11 0.51 17.74
g9paraspire Parents' plan for child's future education level in G9 4.54 4.00 0.54 17.85
g10paraspire Parents' plan for child's future education level in G10 3.52 3.04 0.48 19.13
g11paraspire Parents' plan for child's future education level in G11 3.45 3.01 0.44 18.63
g12paraspire Parents' plan for child's future education level in G12 3.54 3.08 0.46 18.20
Student's Academic Motivation (categorical)
g7acamotiv plans for education level in the future in grade 7 4.21 3.93 0.28 9.45
g8acamotiv plans for education level in the future in grade 8 4.16 3.82 0.34 12.28
g9acamotiv plans for education level in the future in grade 9 4.24 3.82 0.42 14.42
g10acamotiv plans for education level in the future in grade 10 3.21 2.79 0.42 17.51
g11acamotiv plans for education level in the future in grade 11 3.25 2.80 0.45 19.79
g12acamotiv plans for education level in the future in grade 12 3.35 2.90 0.45 17.64
Verbal Achievement
G7VER_S KEDI test: Verbal score in grade 7 67.96 60.47 7.49 13.52
G8VER_S KEDI test: Verbal score in grade 8 61.79 52.88 8.91 15.56
G9VER_S KEDI test: Verbal score in grade 9 59.07 49.58 9.49 13.89
English Achievement
G7ENG_S KEDI test: English score in grade 7 60.62 45.42 15.20 22.70
G8ENG_S KEDI test: English score in grade 8 59.84 42.71 17.13 23.71
G9ENG_S KEDI test: English score in grade 9 58.19 39.46 18.73 22.32
Math Achievement
G7MAT_S KEDI test: Math score in grade 7 62.65 46.80 15.85 22.51
G8MAT_S KEDI test: Math score in grade 8 54.90 38.21 16.69 23.02
G9MAT_S KEDI test: Math score in grade 9 55.78 36.50 19.28 24.12
 129

Private Tutoring
Variables Description Difference t-value3
yes1 no2
School Characteristics
High School Track (1=general; 0=technical)
g10track school track in grade 10 0.85 0.49 0.36 32.38
g11track school track in grade 11 0.87 0.57 0.30 27.66
g12track school track in grade 12 0.92 0.67 0.25 20.76
School Type (1=private; 0=public)
g7sch_type school type in grade 7 0.20 0.18 0.02 1.56
g8sch_type school type in grade 8 0.20 0.20 0.00 0.15
g9sch_type school type in grade 9 0.20 0.20 0.00 0.38
g10sch_type school type in grade 10 0.49 0.44 0.05 3.36
g11sch_type school type in grade 11 0.50 0.45 0.05 3.82
g12sch_type school type in grade 12 0.51 0.46 0.06 3.90
Student-Teacher Ratio
g7STratio # of students/# of teachers in grade 7 20.74 19.16 1.58 11.44
g8STratio # of students/# of teachers in grade 8 20.75 18.82 1.93 14.22
g9STratio # of students/# of teachers in grade 9 20.50 18.57 1.93 13.90
g10STratio # of students/# of teachers in grade 10 16.40 14.75 1.65 21.43
g11STratio # of students/# of teachers in grade 11 16.40 14.75 1.65 21.43
g12STratio # of students/# of teachers in grade 12 16.87 15.23 1.64 20.50

Regional Characteristics
Urbanicity (1=urban; 0=suburban or rural)
g7urban Urban residency in grade 7 0.49 0.36 0.13 8.53
g8urban Urban residency in grade 8 0.49 0.36 0.13 8.82
g9urban Urban residency in grade 9 0.48 0.37 0.11 7.10
g10urban Urban residency in grade 10 0.49 0.37 0.12 9.35
g11urban Urban residency in grade 11 0.51 0.39 0.12 8.85
g12urban Urban residency in grade 12 0.58 0.41 0.16 11.47
School Choice (1=choice; 0=no choice)
g7choice Residence with a school choice scheme in grade 7 0.35 0.49 -0.14 -9.49
g8choice Residence with a school choice scheme in grade 8 0.35 0.50 -0.14 -9.72
g9choice Residence with a school choice scheme in grade 9 0.37 0.49 -0.13 -8.03
g10choice Residence with a school choice scheme in grade 10 0.35 0.49 -0.14 -11.00
g11choice Residence with a school choice scheme in grade 11 0.33 0.46 -0.12 -9.36
g12choice Residence with a school choice scheme in grade 12 0.28 0.46 -0.18 -13.08
Residence in Seoul (1=Seoul; 0=non-Seoul)
g7seoul Residence in Seoul in grade 7 0.19 0.12 0.07 6.85
g8seoul Residence in Seoul in grade 8 0.19 0.12 0.07 6.10
g9seoul Residence in Seoul in grade 9 0.19 0.13 0.06 4.83
g10seoul Residence in Seoul in grade 10 0.19 0.13 0.06 6.53
g11seoul Residence in Seoul in grade 11 0.21 0.12 0.09 8.40
g12seoul Residence in Seoul in grade 12 0.26 0.12 0.14 13.19
1
N=4,884 (7th); 4,904 (8th); 3,651 (9th); 4,256 (10th); 3,370 (11th); 1,897 (12th)
2
N=1,341 (7th); 1,381 (8th); 1,333 (9th); 1,952 (10th); 2,294 (11th); 3,487 (12th)
3
t-value for [diff= mean(1) - mean(0)]
 130

Table 46. Descriptive statistics of dependent variables

Private
Standard Tutoring
Variables Average Difference t-value3
Deviation 1 2
yes no
Dependent Variables (categorical: 1-5)
G7ATTN Degree of attention to lessons in grade 7 4.15 1.02 4.15 4.16 -0.01 -0.36
G8ATTN Degree of attention to lessons in grade 8 4.09 1.06 4.22 3.74 0.47 14.47
G9ATTN Degree of attention to lessons in grade 9 3.97 1.12 4.09 3.54 0.55 15.15
G10ATTN Degree of attention to lessons in grade 10 3.97 1.01 4.09 3.73 0.36 13.18
G11ATTN Degree of attention to lessons in grade 11 3.48 1.01 3.58 3.31 0.26 9.42
G12ATTN Degree of attention to lessons in grade 12 3.54 0.98 3.70 3.46 0.24 8.59
1
N=4,884 (7th); 4,904 (8th); 3,651 (9th); 4,256 (10th); 3,370 (11th); 1,897 (12th)
2
N=1,341 (7th); 1,381 (8th); 1,333 (9th); 1,952 (10th); 2,294 (11th); 3,487 (12th)
3
t-value for [diff= mean(1) - mean(0)]
 131

5.3.2 Diagnostics for Propensity Score Matching

After matching, there are two concerns to check for in validating subsequent analyses.

The first diagnostic is to check whether there is a sufficient overlap between the treatment and

control groups. It’s also called a common support between the two groups. If these two groups

do not overlap or overlap insufficiently, I may surmise that the people are in the different

specifications of covariates or I may not be able to proceed with further analyses, at worst. By

plotting the histograms for the propensity scores of both groups, I can compare and check the

overlap between the two groups. Figure 4 presents the histograms for each grade. The red

histograms show the distribution of propensity scores for the treatment group, and blue

histograms indicate the control group. According to the histograms in Figure 4, there seems to

be a sufficient overlap between the red and blue histograms in each grade level. However, six

sets of histograms commonly indicate that there are insufficient overlaps in the right end

(students who have high propensity scores). After many trials to seek a common overlap for

those with high propensity scores, I was unsuccessful in obtaining sufficient commonality

between the treated and untreated groups.

The second diagnostic is to examine the balance between the treatment and control

groups using the psbal command in STATA. This command allows checking for balances in

means and standard deviations using both the unmatched and matched sample between the

treatment and control groups. In order to reach a better balance, I transformed several variables;

I added square terms for score variables and several interaction terms, and used the logarithm of

household income. I also removed variables that seemed less important with regard to the

outcome variable. The covariates listed in Table G in Appendix are the final model

specifications that reached the best balance. Compared to the balance using unmatched datasets,
 132

balances in means and standard deviations are improved after matching. The balance diagnostics

are available from Table G.1 to G.6 in Appendix. Based on these two diagnostics, subsequent

analyses using these matched samples are justified.

Figure 4. Common support between the treatment and control groups

5.3.3 Empirical Results

Average Effects

Table 47 is a truncated version of empirical results. Column 1 presents the empirical

results obtained from the Ordered Logit estimation. First, private tutoring in grade 7 shows a

negative association with students’ attention to lessons in formal schools in grade 7. A student

who receives private tutoring in grade 7 paid less attention to lessons in school by 0.08 standard

deviations than a student who did not receive any private tutoring. However, this association is

not statistically significant. Interesting patterns are shown in the complete table of results
 133

available in Table H.1 in Appendix. As presented in column 2, students who live in urban areas,

on average, were reported to pay less attention to lessons in schools compared with students who

live in suburban or rural areas. As expected, motivated students paid more attention to lessons

than less motivated peers. Also, students who lived in a district with a school choice scheme

paid less attention to schools than those who have the freedom to choose their schools.

Table 47. Summary of estimates of the average effect of private tutoring on students'
attention to lessons in formal schools
(1) (2) (3)
N for PSM1
Dependent variable (z-score) Ordered Logit PSM1a PSM2b
ATTENTION in Grade 7 -0.08 0.01 0.01
4,916
(0.080) (0.057) (0.056)

ATTENTION in Grade 8 0.27** 0.17** 0.14**

4,984
(0.083) (0.059) (0.053)

ATTENTION in Grade 9 0.17* 0.05 0.06

3,552
(0.086) (0.071) (0.056)

ATTENTION in Grade 10 0.16* 0.07 0.09

3,618
(0.078) (0.071) (0.051)

ATTENTION in Grade 11 0.15* 0.07 0.10

3,672
(0.070) (0.059) (0.054)

ATTENTION in Grade 12 0.12 0.10 0.09

3,208
(0.073) (0.061) (0.051)
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Complete table for Ordered Logit is provided in Table H’s in Appendix; (2) a Difference-in-means; (3) b Regression-
adjusted matched estimate

Unlike the estimate in grade 7, students with private tutoring in grade 8 showed a higher

degree of attention to lessons by 0.27 standard deviations than students without private tutoring,

and this estimate is statistically significant at the 1 percent level. In terms of the associations

between covariates and the dependent variable in Table H.2 in Appendix, high-achieving

students in verbal, English, and math pay more attention to lessons than low-achieving students,

as expected. Female students pay more attention than male students, and students with parents

who have higher academic aspiration for their children show higher attention to lessons than
 134

students with parents who have less aspiration. Lastly, students in schools that have a higher

student-teacher ratio pay less attention to lessons than those who attend schools with a lower

student-teacher ratio. However, the association between school choice and students’ attention is

opposite to the results in Table H.1. In grade 8, students in an area with a school choice scheme

demonstrate higher attention to lessons than their peers in an area with no school choice scheme.

It is also noticeable that the average effect of private tutoring on students’ attention

decreases as students move to upper grades as shown in column 1 of Table 47. In grades 9, 10,

and 11, students who take private tutoring also show higher attention to lessons by 0.17, 0.16,

and 0.15 compared to students without private tutoring. The magnitudes of these effects are

smaller than the effect in grade 8, and these estimates are statistically significant at the 5 percent

level. Furthermore, there is no statistical difference in the degree of attention to lessons between

the treatment and control groups in grade 12. In summary for the estimates of Ordered Logit,

private tutoring positively affects students’ attention to school lessons, which is likely to affect

the learning environment in formal schooling.

It must be noted that these are the estimates from Ordered Logit that are unadjusted for

the endogeneity of private tutoring. Therefore, I applied PSM. As mentioned in the section of

estimation models, PSM is designed to estimate the average treatment effect on the treated (ATT)

within the common support region shown in Figure 4. In addition, the standard errors are

calculated by bootstrapping with 1,000 replications for several reasons already mentioned.

Column 2 in Table 47 presents the empirical results using difference-in-means in the PSM

method. The results suggest that only taking private tutoring in grade 8 has positive effects on

students’ attention to school lessons. Students with private tutoring in grade 8 show a higher

degree of attention by 0.17 standard deviations than students with no private tutoring. Compared
 135

to the estimate in column 1, the effect size becomes smaller, but it is still statistically significant

at the 1 percent level. However, the rest of estimates in column 2 are not statistically significant;

there is no statistically significant effect of taking private tutoring on students’ attention to

lessons in grades 7, 9, 10, 11, and 12. In order to check whether estimates in difference-in-

means are robust, I also applied the regression-adjusted matched estimation presented in column

3. Although the effect size of private tutoring effect in grade 8 is 0.14 standard deviations, which

is slightly smaller than the one in column 2, the results in column 3 are consistent with the ones

in column 2. In terms of the magnitude of the effect, 0.19 implies about a five percent change in

the degree of attention to school lessons, which is a small change. However, if many students in

a classroom improve their attention by five percent, that would change the overall learning

environment in a visible way. Therefore, it should be considered as a sizable effect. At this

point, it is questionable whether these results remain similar when I estimate the effect of private

tutoring by ability group. Using three subgroups of ability group, differential effects are

investigated.

Heterogeneous Effects

Using the same methodologies, the effect of private tutoring on students’ attention to

lessons in formal schools is observed for three ability groups as presented in Table 48. These

three ability groups were constructed by making three subgroups using the KEDI scores

measured in grade 7, which is a year prior to the regulation. Observing the effect for each group

provides information on whether there are differential effects by ability group. In grade 7, there

are no heterogeneous effects of private tutoring on students’ attention to lessons, which is

consistent with the average effect. In grade 8, the positive and significant effects are
 136

concentrated in the lowest ability group. The OLS estimate for the low ability group indicates

that taking private tutoring increases students’ attention to lessons by 0.27 standard deviations

after controlling for students’ background characteristics. The estimates with the PSM methods

are slightly smaller but consistent with the OLS estimate as shown in column 2 and 3 of Table 48,

that taking private tutoring raises students’ degree of attention to lessons by 0.22 and 0.20

standard deviations, respectively.

Table 48. Heterogeneous effects of private tutoring on students' attention to lessons in

formal schools

Group 1 (low ability) Group 2 (medium ability) Group 3 (high ability)

(1) (2) (3) (4) (5) (6) (7) (8) (9)
Ordered Ordered Ordered
PSM1a PSM2b PSM1a PSM2b PSM1a PSM2b
Outcomes Logit Logit Logit
ATTENTION IN G7 -0.10 0.17 0.16 -0.10 -0.01 -0.03 -0.02 0.00 0.00
(0.117) (0.096) (0.085) (0.140) (0.109) (0.104) (0.184) (0.101) (0.115)

ATTENTION IN G8 0.27* 0.22* 0.20* 0.22 0.06 0.07 0.32 0.07 0.09
(0.121) (0.101) (0.099) (0.139) (0.091) (0.073) (0.208) (0.094) (0.092)

ATTENTION IN G9 0.19 0.36** 0.30** 0.23 0.10 0.11 -0.13 -0.09 -0.08
(0.126) (0.139) (0.113) (0.142) (0.109) (0.099) (0.227) (0.069) (0.081)

ATTENTION IN G10 0.16 -0.01 0.02 0.15 0.28** 0.29** 0.18 0.10 0.12
(0.123) (0.139) (0.093) (0.124) (0.104) (0.083) (0.178) (0.170) (0.113)

ATTENTION IN G11 0.04 0.30* 0.28** 0.01 -0.01 0.00 0.44*** 0.11 0.15*
(0.125) (0.120) (0.108) (0.118) (0.114) (0.083) (0.126) (0.112) (0.074)

ATTENTION IN G12 0.26 0.05 0.02 0.26* 0.12 0.12 -0.03 -0.02 0.00
(0.162) (0.164) (0.131) (0.127) (0.105) (0.081) (0.111) (0.088) (0.066)
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Complete table for Ordered Logit is provided in Table H’s in Appendix. (2) a Difference-in-means; (3) b Regression-adjusted
matched estimate

However, there are no statistically significant differences between the treatment and

control groups for the middle and highest ability groups in grade 8. A similar pattern is shown in

grade 9. Positive and significant effects of private tutoring at the 1 percent level are detected
 137

only in the lowest ability group. Although the OLS estimate is not statistically significant,

estimates from the PSM method show that private tutoring increases attention of low-achieving

students by 0.36 standard deviations based on the results from difference-in-means. The

regression-adjusted matched estimate is 0.30, which is smaller but still statistically significant at

the 1 percent level. The effects for the middle and upper groups are not statistically significant,

similar to the results in grade 8. When I averaged out the estimates for all groups, the average

effect should be insignificant as shown in Table 47. In grade 10, private tutoring is effective in

raising attention of students in the middle group only. An increase of about 0.3 in the degree of

attention is found for students who received private tutoring. In grade 11, the effect is detected

in the lowest ability group, according to the PSM estimates. Even though the regression-adjusted

matched estimate for the highest ability group (0.15) is statistically significant at the 5 percent

level, this result is not robust because the estimate from difference-in-means in column 8 is not

significant. Similar to the average effect in grade 12, there are no heterogeneous effects detected

in grade 12. In summary, the significant effects are investigated in either the lowest or middle

groups, but mostly in the lowest group. Therefore, mostly low-achieving students are positively

affected by private tutoring on their attention to lessons, which would improve the learning

environment in classrooms.

Effect of Regulation on Private Tutoring

As another way to estimate the effect of private tutoring on students’ attention to lessons

in formal schools, I used a natural experiment, which is the regulatory enactment on late private

tutoring in Seoul. This analysis observes the effects of the regulation on students’ attention using

the DID method employing the same dependent and control variables shown in the previous

analyses with PSM. As column 2 of Table 49 shows, in the first year after the regulation was
 138

enacted, the difference in attention to lessons between students in Seoul and students in other

metropolitan cities increased by 0.11 standard deviations, but this difference is not statistically

significant. When this average estimate is disaggregated by three ability groups, the results

suggest that the difference favoring students in Seoul increased by 0.46 for the highest ability

group after one year of the regulation, and it is statistically significant at the 5 percent level. This

can be interpreted that attention to lessons among students in the highest ability group in Seoul

has increased after the regulation. Since this regulation is supposed to reduce students’ time

spent on private tutoring and reduce the participation in private tutoring nationally, this result

also implies that the reduction in time spent on private tutoring increases the degree of attention

of high-achieving students in classroom. This may also be a result of students getting more rest

and sleep, which increases their energy in schools. There are no statistically significant results in

the second and third year of the regulation.

As explained, only high-achieving students seem to be affected by this regulation. This

phenomenon does not seem spurious because high achievers show higher participation in private

tutoring than low achievers in Korea as the descriptive statistics show. Therefore, the group of

high-achieving students is more subject to be affected by this regulation than the rest. As a result,

on average, there is no statistically significant change in students’ attention to lessons between

Seoul and other metropolitan cities, but the decrease in demand for private tutoring due to the

regulation positively affects high-achieving students in terms of their attention to lessons in

classrooms.
 139

Table 49. Difference-in-differences estimation: effect of the regulation on private tutoring

on students' attention to lessons in formal schools
(Year 1)
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Low Middle High

SEOUL -0.02 0.11 0.06 -0.08 0.46*

(0.075) (0.110) (0.205) (0.184) (0.192)
Covariates Yes Yes Yes Yes

Observations 2,500 1,342 342 468 532

Note: Dependent variable: chg_attention_yr1 (=G10ATTN-G9ATTN)

(Year 2)
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Low Middle High

SEOUL -0.03 0.11 0.05 0.01 0.28

-0.079 -0.112 -0.218 -0.192 -0.186
Covariates Yes Yes Yes Yes

Observations 2283 1241 316 415 510

Note: Dependent variable: chg_attention_yr2 (=G11ATTN-G9ATTN)

(Year 3)
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Low Middle High

SEOUL 0.01 -0.02 -0.21 -0.04 0.21

(0.082) (0.116) (0.233) (0.194) (0.195)
Covariates Yes Yes Yes Yes

Observations 2,145 1,152 275 399 478

*** p<0.001, ** p<0.01, * p<0.05
Note: Dependent variable: chg_attention_yr3 (=G12ATTN-G9ATTN)

5.4 Discussion

The purpose of this part of the analysis is to explore the effect of private tutoring on

formal schooling. To observe whether private tutoring indirectly affects formal schooling, I used

students’ attention to lessons in formal schools reported by teachers as a proxy for the quality of

learning environment in formal schooling. Two methods were used to observe this causal

relationship. First, by using PSM in order to take account of endogeneity bias that the Ordered
 140

Logit estimation is prone to, I constructed a counterfactual and investigated the average

treatment effect on the treated. Also, the heterogeneous effects by ability group were examined.

In summary, private tutoring is statistically significant in explaining the quality of the learning

environment in formal schooling explained by students’ attention to lessons. Students who

received private tutoring showed a higher degree of attention to lessons in formal schools

compared to students who do not. But this positive and significant effect is detected in grade 8

only. This result implies that if there are many students in one classroom who are exposed to

tutoring, those students will pay more attention to lessons, influencing the overall learning

environment in a positive way. However, this finding runs counter to the conclusions of the

existing literature. It is argued that tutoring can take away students’ interest and attention to

lessons in formal schools because they have already covered the topics with tutors (Hussein,

1987; Nanayakkara & Ranaweera, 1994; Sawada & Kobayashi, 1986). Furthermore, this

positive effect could be much larger than estimated when the results in heterogeneous effects are

taken into account as below.

As shown in Table 47, significant effect among the PSM estimates is detected in grade 8.

However, when I narrowed down the analyses into three ability groups, positive effects were

detected at other grade levels too. In grades 9, 10, and 11, positive effects of private tutoring on

students’ attention to lessons were shown in either the bottom or middle group. These positive

effects were not identified when observing the average effect. The focal point of these results is

that most of the positive effects are identified for the low ability group. In many cases, low-

achieving students do not pay attention to lessons or disturb other students’ studying, which

undermines the academic learning environment. Therefore, it is often the case that the learning

environment becomes more supportive as there are fewer low-achieving students in a classroom.
 141

Since the empirical results in this chapter suggest that low-achieving students who have a high

probability of diluting the learning environment in a classroom improve their behavior in a

classroom when tutored, it is highly likely that the quality of the overall academic learning

environment in formal schooling will be upgraded. Even though the size of effects may seem

trivial, I believe that the actual effects would be much larger than shown in the estimation.

Because if this small and positive change in students’ attention to school lessons is aggregated by

many students in a classroom, that would largely increase the quality of the learning environment

in a classroom. Furthermore, the improved learning environment may have an effect in reducing

the achievement gap between the low and high ability groups. Analysis in terms of the influence

on achievement gap is going to be conducted in the next part of this study.

Lastly, using the regulation on late private tutoring, I explored the effect of the reduction

in private tutoring on students’ attention to lessons in formal schools. On average, the regulation

did not influence the learning environment in classrooms. However, high-achieving students

seem to be affected by the regulation. The results imply that as the time spent on private tutoring

decreases due to the regulation, high-achieving students who live in Seoul pay more attention to

lessons than before the regulation. Again, this may be due to more resting for students, which

makes students more productive and motivated in schools. Therefore, the difference in the

degree of attention between students in Seoul (under the regulation) and students in other

metropolitan cities (not affected by the regulation) become larger after the regulation, whereas

students in the low and middle groups are not affected by it. Based on these results, it can be

concluded that this regulation is somewhat effective in increasing high-achieving students’

attention to lessons in formal schools, but is not successful in improving the overall quality of the
 142

learning environment in formal schooling because the behaviors of low- and average-achieving

students are not affected.

Observing the heterogeneous effects of private tutoring on students’ attention to lessons

has generated questions on whether private tutoring actually creates heterogeneous effects on

students’ academic achievement that was not addressed in the first part of the study. In the next

chapter, I will conduct further study to observe how private tutoring affects educational

inequality in the Korean society.

 143

CHAPTER VI. THE IMPACT ON EDUCATIONAL INEQUALITY

In the first and second parts of the empirical analyses, I investigated the effects of private

tutoring on students’ academic achievement and formal schooling. In this further analysis, I

investigated the social consequences of private tutoring. As the existing literature has suggested,

private tutoring has potential social consequences such as increased pressure on students

(Tsukada, 1991; Wijetunge, 1994), alteration of social relationships among students (De Silva,

1994; Paiva et al., 1997; Russell, 1997), and the expansion of social inequalities (Bray, 1999;

Dang, 2008; Jelani & Tan, 2012). Among these several social consequences, social inequality in

education is the most controversial topic in Korea as well as other countries. Bray (1999) argued

that private tutoring may create a mechanism that increases social inequality. Since most forms

of private tutoring in Korea are more easily accessible to students from high-income families, it

can create greater inequality in accessing supplementary education such as private tutoring that

widens the achievement gap among income or ability groups. However, the existing literature

that points out this possibility deriving from private tutoring is mostly speculative and anecdotal

without quantitative evidence. Therefore, it has not been established by quantitative evidence

whether and by how much private tutoring exacerbates or improves social inequality in

education. Using the following research questions, I tried to uncover the relationship between

private tutoring and educational inequality in Korea.

Does private tutoring exacerbate educational inequality? Are there heterogeneous

effects of private tutoring on student academic achievement? For whom are private

tutoring effects significant and how large might they be?

 144

To answer this research question, two forces that create an inequality of achievement due

to tutoring need to be considered. One force is a differential amount of private tutoring by each

quantile group. The other force is the inequality in the effectiveness of tutoring by quantile that

this chapter mostly focuses on. It is the product of these two forces for each quantile that

determine whether tutoring causes inequality in academic achievement. To observe the first

force, I observed the percentage of students who participate in tutoring and the median amount of

tutoring in each quantile group using the weekly hours spent on private tutoring. As shown in

Table 50, in the case of private tutoring in middle school, students in the upper quantiles show a

higher participation rate in private tutoring compared to students in the lower quantiles, as

expected. However, the differences in percentages between quantiles are not dramatic. In terms

of median weekly hours spent on private tutoring, however, there is about a six-hour difference

between the bottom and top quantiles. Based on these statistics, it seems that inequality in the

amount of private tutoring by quantile influences the inequality of achievement to some extent.

Table 50. Inequality in the amount of private tutoring in middle school by quantile group

q10 q25 q50 q75 q90

Percentage of participants in PT (%) 48.3 55.5 50.1 61.3 65.1
Median weekly hours for PT (hours) 3 4.6 6 7.3 9

Using hours of private tutoring as a treatment variable, it seems possible to estimate how

much this first force affects inequality in achievement. However, this study is only limited to

observe the second force (effectiveness of private tutoring by quantile) due to data limitation.

The variable of weekly tutoring hours in high school has a large number of missing observations,
 145

which makes difficult to estimate the effects, while data on tutoring hours in middle school is

available.

In order to observe how much the second force affects inequality in academic

achievement, I applied Quantile Regression (QR) with the instrumental variable, which is also

called a Two-Stage Least Absolute Deviation estimator (2SLAD), originally developed by

Amemiya (1982). Since a simple QR uses linear regression at each quantile, which does not

control for endogeneity of private tutoring, I used a QR framework by employing the 2SLAD

method. In addition, I applied Propensity Score Subclassification, which enables to observe the

difference between the matched groups in each subclass or strata. These three methods will help

investigate the heterogeneous effects of private tutoring on students’ academic achievement.

6.1 Estimation Models

Both the OLS and IV methods that were introduced in Chapter IV aimed to estimate the

average causal effect of private tutoring on students’ academic achievement. That is to say, most

of the studies estimate the effects of a change in private tutoring on the achievement of the

average individual in the sample being analyzed. The alternative QR approach goes further. It

concerns itself not only with the efficiency of private tutoring on the average individual, but

allows the researcher to estimate the marginal effect of private tutoring for individuals at

different points in the achievement distribution. This makes it possible to assess the equity

implications resulting from having private tutoring. Rather than estimating the effects of

independent variables at the mean, the quantile estimator is designed to predict the effects of

independent variables at different points (quantiles) in the conditional distribution of the

dependent variables (i.e., student’s achievement score). For this reason, the estimator has
 146

become very attractive to researchers interested in determining the differential effects of

regressors at varying points in the distribution of a dependent variable (Levin, 2001).

Given the threat of endogeneity of private tutoring generated by the selectivity of

students with private tutoring and unobserved characteristics of these students, this dissertation

takes into account this problem in a QR framework by employing 2SLAD. The 2SLAD

procedure is essentially the quantile estimation analog of 2SLS. Instead of using the actual

independent variables that indicate private tutoring participation, the predicted values for private

tutoring participation derived from the following first-stage equations are used.

Cumulative Effect (first stage):

(6.1) ih(a) 0 1 ih(a) 2 ih ih

Single-year Effect (first stage):

(6.2) ij(a) 0 1 ij(a) 2 ij ij

Same as equation 4.3 and 4.4 in the first set of the analyses, ih(a) and ij(a) are

private tutoring variables, which indicate years of private tutoring and participation in private

tutoring, respectively. In equation 6.1, ih(a) is an instrumental variable, which is a proportion of

student i’s costs of private tutoring on subject a or overall in a monthly household income during

education level h. This instrument is the same as the one used in the first part of the empirical

analysis. Several tests for checking the validity of the instrument have previously been

explained in Section 4.5 in Chapter IV. Similarly, ij(a) in equation 6.2 is a proportion of student

i’s costs of private tutoring on subject a or overall in grade j. ih and ij are random error terms

associated with the reduced form equations. From these equations 6.1 and 6.2, the predicted
 147

values of two variables of private tutoring are calculated, and they are inserted in the second

stage as follows.

Cumulative Effect (second stage):

(5.14) ih(a),q 0,q 1,q ih(a) 2,q ih,q ih,q

Single-year Effect (second stage):

(5.15) ij(a),q 0,q 1,q ij(a) 2,q ij,q ij,q

where ^ represents predicted values, and 1,q and 1,q are the main coefficients of interest that

imply the cumulative effect and single-year effect of private tutoring using QR at a given

quantile of interest, , to see whether there are different effects of private tutoring on student

academic achievement, by each quantile. Here, represents the 10th, 25th, 50th, 75th or 90th

percentiles in the distribution of academic achievement.

The last method used to explore the heterogeneous effects is the propensity score

subclassification (Rosenbaum & Rubin, 1984). This is another way of using propensity scores to

make causal inferences. Instead of using the propensity scores to match students, I used

propensity scores to divide the sample into subclasses. Within each subclass, the covariates are

balanced between the treated and control groups. First, I calculated propensity scores, e(X), in

equation 4.7, for all students using confounding covariates, and identified the five quantiles of

the distribution of propensity scores. These quantiles are used as cut-points that determine the

five subclasses. Within each subclass, treatment effect estimates are calculated by observing

difference-in-means between the treated and control groups. Using 2SLAD and propensity score

subclassification, this chapter will explore the heterogeneous effects of private tutoring.
 148

6.2 Variables

Variables used in this analysis are identical to the ones listed in Section 4.3 for the first

analysis regarding the effects of private tutoring on students’ academic achievement. As a

reminder, two types of dependent variables are used: students’ scores of the test administered by

the KEDI in grades 7, 8, and 9 and decile rank on the CSAT in grade 12. Also, there are two

types of dichotomous independent variables, which indicate total years of private tutoring and

private tutoring participation. The variables that indicate years of participation in private

tutoring aim to estimate the cumulative effects of private tutoring. The dichotomous variables

that indicate private tutoring participation are used to estimate the effects of private tutoring for

each grade. Control variables in a vector are also the same as the ones listed in Table 16. The

descriptive statistics for these variables are also provided in Table 16, 17, and 18, which have

been described in Section 4.4.

 149

6.3 Empirical Results

6.3.1 Private Tutoring in Middle School: Short-term Heterogeneous Effects

Cumulative Effects. Table 51, 52, and 53 present the average and heterogeneous effects

of one, two, and three years of private tutoring, respectively, on students’ academic achievement

in grade 9 based on three estimation models.

Table 51. Heterogeneous effects of 1 year of private tutoring in middle school on overall
academic achievement in grade 9
OLS
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

1 year of PT in middle school -0.03 -0.06 -0.04 -0.10* -0.04 0.15

(0.046) (0.056) (0.042) (0.053) (0.070) (0.116)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 778 778 778 778 778 778

R-squared 0.261
2SLAD
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

1 year of PT in middle school 0.16 -0.00 0.08 0.16*** 0.26*** 0.27***

(0.131) (0.028) (0.064) (0.050) (0.043) (0.038)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 521 4,286 4,286 4,286 4,286 4,286

R-squared 0.294
PSM
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

1 year of PT in middle school 0.06 -0.06 0.06 -0.14 0.09 -0.13

(0.079) (0.139) (0.129) (0.160) (0.154) (0.176)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 778 155 156 154 156 155

Pseudo R2 0.029 0.034 0.042 0.042 0.021 0.034
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VEM_Z (average score in grade 9); (2) Complete tables for OLS and
2SLAD are provided in Table I.1 in Appendix.
 150

Table 52. Heterogeneous effects of 2 years of private tutoring in middle school on overall
academic achievement in grade 9
OLS
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

2 years of PT in middle school 0.19*** 0.10* 0.14* 0.18** 0.30*** 0.41***

(0.048) (0.044) (0.059) (0.069) (0.060) (0.063)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 1,084 1,084 1,084 1,084 1,084 1,084

R-squared 0.240
2SLAD
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

2 years of PT in middle school 0.25 -0.00 0.15 0.31*** 0.49*** 0.50***

(0.132) (0.054) (0.108) (0.086) (0.096) (0.081)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 763 4,286 4,286 4,286 4,286 4,286

R-squared 0.257
PSM
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

2 years of PT in middle school 0.12 0.14 0.20 0.10 0.33* 0.01

(0.070) (0.120) (0.127) (0.137) (0.162) (0.217)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 1,084 214 217 216 218 216

Pseudo R2 0.082 0.028 0.008 0.042 0.042 0.054
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VEM_Z (average score in grade 9); (2) Complete tables for OLS and
2SLAD are provided in Table I.2 in Appendix.
 151

Table 53. Heterogeneous effects of 3 years of private tutoring in middle school on overall
academic achievement in grade 9
OLS
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

3 years of PT in middle school 0.38*** 0.21*** 0.31*** 0.37*** 0.53*** 0.57***

(0.045) (0.061) (0.060) (0.045) (0.050) (0.075)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 2,854 2,854 2,854 2,854 2,854 2,854

R-squared 0.294
2SLAD
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

3 years of PT in middle school 0.54*** -0.01 0.30 0.61** 0.98*** 1.00***

(0.133) (0.097) (0.246) (0.201) (0.199) (0.151)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 2,252 4,286 4,286 4,286 4,286 4,286

R-squared 0.267
PSM
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

3 years of PT in middle school 0.29** 0.43*** 0.42*** 0.55*** 0.12 -0.03

(0.109) (0.088) (0.111) (0.162) (0.267) (0.606)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 2,854 572 571 571 571 447

Pseudo R2 0.260 0.101 0.031 0.025 0.081 0.222
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9VEM_Z (average score in grade 9); (2) Complete tables for OLS and 2SLAD
are provided in Table I.3 in Appendix.

The OLS estimates suggest that there are cumulative effects of private tutoring in middle

school on students’ academic achievement in grade 9. While one year of private tutoring does

not show significant heterogeneous effects, two and three years of private tutoring suggest that

there are larger effects for students in the upper quantiles. For example, students in the 75th and

90th quantiles increase their achievement by more than 0.30 and 0.50 standard deviations after

taking two and three years of private tutoring, respectively, while students in the 10th and 25th

quantiles benefit by about 0.10 and 0.30 standard deviations with two and three years of private

tutoring, respectively. In other words, students in the higher percentiles benefit more from
 152

private tutoring than students in the lower percentiles. In the 2SLAD estimation, students in the

10th and 25th percentiles do experience significant effects of private tutoring on their academic

achievement, while estimates for students in the 50th, 75th, and 90th percentiles remain

statistically significant. However, the effect sizes are much larger than the ones from the OLS

estimation. Taking one year of private tutoring for students in the 90th percentile positively

affects their achievement by 0.27 standard deviations, and this estimate is the largest among

other quantiles. In addition, taking two and three years of private tutoring increases the overall

academic achievement by 0.50 and 1 standard deviations for students in the 90th percentiles.

Since high-achieving students benefit more from private tutoring than low-achieving students, as

the results suggest, private tutoring seems to widen the achievement gap between high and low

achievers, according to the 2SLAD estimates. However, the results from propensity score

subclassification suggest that two years of private tutoring is most beneficial for students in the

75th quantiles, while three years of private tutoring increases academic achievement for students

in the lower quantiles only. Therefore, the consistent results in all three methods are that two

years of private tutoring increases academic achievement for students in the higher quantile,

which implies the greater achievement gap. One concern in the PSM estimation is that the

sample size for matched students in each quantile is small in several estimations. Sample sizes

are sometimes less than 200, which limits tests of significance. The small sample size may be

one of the reasons why several estimates are not statistically significant.

Instead of observing the heterogeneous effects of overall private tutoring, I narrowed

down the analysis by academic subject. In terms of years of verbal tutoring, there are no

statistically significant heterogeneous effects on students’ verbal achievement in grade 917.

However, the results of English and math tutoring are different from the ones of verbal tutoring.
17
Results of the heterogeneous effects of years of verbal tutoring are available in Table I.4, I.5 and I.6 in Appendix.
 153

As shown in Table 54, taking one year of English tutoring is statistically significant in explaining

English achievement in grade 9 for students in the higher quantiles. Especially, both the 2SLAD

and PSM estimates show that taking one year of English tutoring increases English scores by

0.46 and 0.35 standard deviations, respectively, for students in the 75th percentile. Even though

students in other quantiles except for the lowest one benefit from having one year of English

tutoring based on the 2SLAD estimates, these results are not consistent in the PSM method.

Table 55 presents the effects of two years of English tutoring. All methods show that there are

greater effects of two years of English tutoring for students in the upper quantiles. Even though

the effect sizes are not similar, it is consistent that English tutoring broadens the achievement gap

between students in low and high quantiles. Three years of English tutoring didn't show

consistent results of heterogeneity18.

18
Results of the heterogeneous effects of three years of English tutoring are available in Table I.9 in Appendix.
 154

Table 54. Heterogeneous effects of 1 year of English tutoring in middle school on English
achievement in grade 9
OLS
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90
1 year of English PT in MS 0.05 0.08 0.01 0.07 0.06 0.04
(0.050) (0.053) (0.054) (0.073) (0.073) (0.122)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 993 993 993 993 993 993

R-squared 0.231
2SLAD
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90
1 year of English PT in MS 0.24 0.01 0.24* 0.35*** 0.46*** 0.33***
(0.156) (0.075) (0.097) (0.072) (0.079) (0.050)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 687 4,196 4,196 4,196 4,196 4,196

R-squared 0.274
PSM
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90
1 year of English PT in MS 0.04 0.15 0.06 -0.34 0.35* -0.20
(0.074) (0.108) (0.133) (0.173) (0.180) (0.195)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 993 198 197 196 200 199

Pseudo R2 0.031 0.051 0.039 0.057 0.036 0.031
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9ENG_Z (English score in grade 9); (2) Complete tables for OLS and
2SLAD are provided in Table I.7 in Appendix.
 155

Table 55. Heterogeneous effects of 2 years of English tutoring in middle school on English
achievement in grade 9
OLS
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

2 years of English PT in MS 0.23*** 0.13*** 0.19*** 0.21*** 0.30*** 0.27**

(0.049) (0.028) (0.049) (0.050) (0.082) (0.081)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 1,377 1,377 1,377 1,377 1,377 1,377

R-squared 0.274
2SLAD
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

2 years of English PT in MS 0.11 0.13 2.18* 3.16*** 4.14*** 2.96***

(0.293) (0.717) (0.934) (0.582) (0.700) (0.450)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 993 4,196 4,196 4,196 4,196 4,196

R-squared 0.270
PSM
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

2 years of English PT in MS 0.33*** 0.16 0.14 0.19 0.27* 0.66*

(0.079) (0.109) (0.144) (0.134) (0.169) (0.297)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 1,377 278 264 275 279 275

Pseudo R2 0.112 0.044 0.021 0.023 0.037 0.062
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9ENG_Z (English score in grade 9); (2) Complete tables for OLS and 2SLAD
are provided in Table I.8 in Appendix.

The estimates on the effects of math tutoring, on average, show that the estimates from

the three methods indicate that there are more positive effects of math tutoring for students in the

upper quantiles. Table 56 explains that three years of math tutoring raised math achievement by

0.74 and 0.61 standard deviations for students in the 75th and 90th percentiles, while it increased

math achievement by 0.14 and 0.29 standard deviations based on the OLS estimates. A similar

pattern is shown in the 2SLAD estimates; three years of math tutoring increased math scores by

more than three-fourth of a standard deviation for students in the upper quantiles, but there were

no statistically significant effects for students in the lower quantiles. The PSM estimates also
 156

indicate the greater effects for students in the 75th percentile compared to the effects for students

in the 10th and 25th percentiles. Thus, three years of math tutoring generates a bigger

achievement gap between the lower and upper quantiles, but one and two years of math tutoring

do not show consistent results in heterogeneity19.

Table 56. Heterogeneous effects of 3 years of math tutoring in middle school on math
achievement in grade 9
OLS
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

3 years of Math PT in MS 0.46*** 0.14** 0.29*** 0.48*** 0.74*** 0.61***

(0.048) (0.052) (0.061) (0.058) (0.088) (0.111)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 3,198 3,198 3,198 3,198 3,198 3,198

R-squared 0.254
2SLAD
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

3 years of Math PT in MS 0.88*** 0.02 0.31 0.90** 1.13*** 0.76***

(0.136) (0.218) (0.280) (0.280) (0.226) (0.204)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 2,531 4,260 4,260 4,260 4,260 4,260

R-squared 0.215
PSM
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

3 years of Math PT in MS 0.31** 0.42*** 0.43** 0.49** 0.61* -0.12

(0.088) (0.089) (0.120) (0.146) (0.305) (0.433)
Covariates ✓ ✓ ✓ ✓ ✓ ✓

Observations 3,198 642 635 640 641 640

Pseudo R2 0.223 0.092 0.031 0.015 0.042 0.066
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: G9MAT_Z (math score in grade 9); (2) Complete tables for OLS and
2SLAD are provided in Table I.12 in Appendix.

19
Results of the heterogeneous effects of one and two years of math tutoring are available in Table I.10 and I.11 in
Appendix.
 157

Single-year Effects. Using a dichotomous variable of private tutoring participation in

each grade, I explored the effect of tutoring in each grade to identify the most effective time of

private tutoring on academic achievement in middle school. In terms of the short-term effects of

private tutoring in grade 7 as presented in Table 57, the OLS estimates indicate that private

tutoring in grade 7 helps increase overall achievement in grade 7 for students in all quantiles, but

the effects are larger for students in the upper quantiles than students in the lower quantiles. As

an example, the positive effect of private tutoring in grade 7 for students in the 90th quantile

(0.27) is larger than the one for students in the 10th percentile (0.10). These positive effects

remain in the next year only for students in the 50th, 75th, and 90th percentiles, but with smaller

effect sizes. Tutoring in grade 7 does not affect achievement in grade 9 as shown in year 2.

Taking account of the 2SLAD approach, the effect sizes of the estimates are much greater than

the ones of the OLS estimates. Except for the bottom quantile, there are more than three-quarters

of a standard deviation increases in overall achievement in grade 7. Tutoring in grade 7 for

students in the 90th percentile increases the average score by 0.93 standard deviations, whereas

the same tutoring for students in the 10th percentile affects scores only by 0.32 standard

deviations. Therefore, the effect for students in the highest quantile is three times bigger than the

one for students in the bottom quantile. These positive and significant effects remain in the next

two years for students in the higher quantiles only, while the effect disappears for the students in

the lower quantiles. It also means that the achievement gap becomes greater as students move to

upper grades. According to the PSM results, even though the differences in the estimates among

quantiles are smaller compared to the IV estimates, the effects are bigger for high achievers than

the ones for low achievers and this heterogeneity continues until year 2.
 158

The effect of private tutoring in grade 8 is also heterogeneous in different quantiles. The

OLS estimates in Table 58 show that the effects for high-achieving students are larger than the

ones for low-achieving students, and those effects are somewhat consistent in grade 9. However,

the 2SLAD estimates indicate that tutoring in grade 8 positively affects overall achievement for

middle quantiles (25th, 50th, and 75th quantiles). Furthermore, the effect for students in the 25th

quantile is larger than the one for the 50th and 75th percentiles, which is counter to the results in

grade 7. The PSM estimates are consistent with the 2SLAD estimates; students in the 25th

percentile benefit from taking private tutoring in grade 8 by 0.21 standard deviations, while

students in the upper quantiles do not receive significant effects. Moreover, both the 2SLAD and

PSM estimates suggest that the heterogeneity disappears in the next year. In terms of private

tutoring in grade 9, there are no statistically significant heterogeneous effects based on all three

methods. The OLS estimates show that there are similar effects for students in each quantile,

and the 2SLAD and PSM methods do not show statistically significant results. Based on these

results, it is concluded that, on average, tutoring in grade 7 is more beneficial for high-achieving

students than low-achieving students, while tutoring in grade 8 seems more beneficial for

students in the lower quantiles.

Table 57. Heterogeneous effects of private tutoring in grade 7 on overall academic achievement for 3 years
OLS
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in G7 0.22*** 0.10** 0.18*** 0.25*** 0.27*** 0.27*** 0.09** 0.02 0.02 0.08* 0.15*** 0.14*** 0.05 0.02 0.01 0.06 0.09 0.11
(0.022) (0.035) (0.035) (0.028) (0.026) (0.032) (0.026) (0.035) (0.027) (0.034) (0.035) (0.038) (0.033) (0.033) (0.045) (0.049) (0.059) (0.069)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in G8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in G9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,895 5,895 5,895 5,895 5,895 5,895 5,380 5,380 5,380 5,380 5,380 5,380 3,990 3,990 3,990 3,990 3,990 3,990
R-squared 0.444 0.364 0.315
2SLAD
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in G7 0.53*** 0.32* 0.77*** 0.75*** 0.82*** 0.93*** 0.33*** 0.1 0.3 0.61*** 0.73*** 0.82*** 0.22* 0.19 0.1 0.23 0.51** 0.52**
(0.066) (0.150) (0.171) (0.140) (0.109) (0.129) (0.076) (0.132) (0.158) (0.183) (0.114) (0.138) (0.094) (0.148) (0.058) (0.149) (0.179) (0.189)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in G8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in G9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,473 5,882 5,882 5,882 5,882 5,882 5,022 5,377 5,377 5,377 5,377 5,377 3,716 3,990 3,990 3,990 3,990 3,990
R-squared 0.419 0.351 0.307
PSM
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in G7 0.21*** 0.22*** 0.13* 0.34*** 0.24** 0.13 0.15** 0.07 0.15* 0.37*** 0.36** 0.15 0.17*** 0.14* 0.22** 0.31*** 0.26* -0.12
(0.040) (0.053) (0.064) (0.067) (0.084) (0.111) (0.038) (0.054) (0.063) (0.076) (0.094) (0.117) (0.043) (0.057) (0.067) (0.075) (0.098) (0.121)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in G8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in G9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,895 1,180 1,178 1,176 1,180 1,181 5,582 1,122 1,128 1,125 1,112 1,095 5,692 1,143 1,143 1,154 1,137 1,115
Pseudo R2 0.105 0.037 0.021 0.019 0.018 0.028 0.105 0.033 0.025 0.023 0.017 0.031 0.103 0.034 0.022 0.018 0.015 0.029
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G7VEM_Z (average score in grade 7) for Year 0; G8VEM_Z (average score in grade 8) for Year 1; G9VEM_Z (average score in grade 9) for Year 2; (2) Complete tables for OLS and
2SLAD are provided in Table J.1 and Table J.2 in Appendix.

159
Table 58. Heterogeneous effects of private tutoring in grade 8 on overall achievement for 2 years
OLS
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in grade 8 0.12*** 0.07*** 0.11*** 0.10*** 0.11*** 0.17*** 0.09** 0.03 0.07* 0.10*** 0.07* 0.16***
(0.020) (0.018) (0.024) (0.022) (0.024) (0.025) (0.028) (0.036) (0.030) (0.023) (0.029) (0.031)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,501 5,501 5,501 5,501 5,501 5,501 4,077 4,077 4,077 4,077 4,077 4,077
R-squared 0.633 0.54
2SLAD
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in grade 8 0.27*** 0.26 0.32** 0.29*** 0.24* 0.14 0.13 0.05 0.07 0.11 0.18 0.21
(0.066) (0.143) (0.099) (0.069) (0.101) (0.079) (0.097) (0.176) (0.146) (0.134) (0.143) (0.207)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,953 5,000 5,000 5,000 5,000 5,000 3,657 3,696 3,696 3,696 3,696 3,696
R-squared 0.623 0.53
PSM
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in grade 8 0.18** 0.08 0.21** 0.17* 0.10 0.05 0.21* 0.06 0.00 0.26* 0.10 -0.06
(0.057) (0.059) (0.062) (0.073) (0.097) (0.241) (0.082) (0.071) (0.084) (0.114) (0.206) (0.542)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,501 1,100 1,103 1,087 1,107 1,093 4,077 814 814 816 818 815
Pseudo R2 0.231 0.039 0.024 0.014 0.008 0..053 0.318 0.097 0.036 0.022 0.019 0.106
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G8VEM_Z (average score in grade 8) for year 0; G9VEM_Z (average score in grade 9) for year 1; (2) Complete
tables for OLS and 2SLAD are provided in Table J.3 and Table J.4 in Appendix.

160
 161

Table 59. Heterogeneous effects of private tutoring in grade 9 on academic achievement in grade 9
OLS
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

PT in grade 9 0.16*** 0.15*** 0.12*** 0.15*** 0.17*** 0.12*

(0.025) (0.023) (0.031) (0.025) (0.026) (0.047)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,044 4,044 4,044 4,044 4,044 4,044
R-squared 0.575
2SLAD
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

PT in grade 9 0.19 -0.05 -0.08 -0.03 0.15 0.37

(0.103) (0.067) (0.067) (0.174) (0.192) (0.222)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 3,490 4,504 4,504 4,504 4,504 4,504
R-squared 0.568
PSM
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

PT in grade 9 0.14* 0.10 0.14 0.33** 0.03 0.21

(0.068) (0.074) (0.078) (0.098) (0.047) (0.322)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,044 809 807 807 810 808
Pseudo R2 0.724 0.079 0.030 0.031 0.026 0.113
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G9VEM_Z (average score in grade 9); (2) Complete tables for
OLS and 2SLAD are provided in Table J.5 and Table J.6 in Appendix.

When these single-year effects are observed by academic subject, there are different

patterns of heterogeneous effects from the ones described above. There are no significant

heterogeneous effects detected in verbal tutoring as shown in Table 60, 61, and 62. As opposed

to verbal tutoring, English and math tutoring show heterogeneity in its effects on English and

math achievement. Table 63 presents the effect of English tutoring in grade 7 on achievement in

all three years of middle school. The OLS estimates indicate greater effects for students in the

upper quantiles in terms of English achievement in grade 7. Even though the sizes of effects
 162

decrease in the next year, students in the 75th and 90th percentiles benefit the most by having

English tutoring compared to students in the lower quantiles. The effect of English tutoring in

grade 7 remains significant only for students in the 75th percentile in year 2, while the effects for

the rest of students disappear. However, the 2SLAD estimates show a different story. As shown

in Table 63, the 2SLAD estimates indicate that English tutoring affects all students in the

achievement distribution to a larger extent than the OLS estimates. Moreover, the effects remain

significant in the next two years for students in all quantiles even though the effect sizes for

high-achieving students are greater than the ones for low-achieving students. Specifically,

English tutoring in grade 7 contributes to increased English scores of students in the 90th

percentile by 1.03 in grade 7, 0.78 in grade 8, and 0.50 in grade 9. For the lower quantiles,

students in the 25th percentile also benefited from English tutoring in grade 7 on English

achievement by 0.74 in grade 7, 0.74 in grade 8, and 0.71 in grade 9. These estimates imply that

benefits of English tutoring taken in grade 7 are distributed reasonably and equally regardless of

students’ academic standing. Therefore, taking English tutoring in grade 7 does not exacerbate

the achievement gap in English to a larger extent. The absence of heterogeneity is also shown

when estimating the effects with the propensity score subclassification, but the overall size of

effects is much smaller than the ones from the 2SLAD estimates. English tutoring in grades 8

and 9 shows greater effects for students in the upper quantiles, according to the OLS and PSM

methods, while the 2SLAD estimates do not show a similar pattern as shown in Table 64 and 65.

Compared to English, math tutoring shows more heterogeneity in its effects on math

achievement. In Table 66, the IV estimation suggests that math tutoring in grade 7 affects math

scores in the same grade by 0.75, 0.80, and 0.76 standard deviations for students in the 25th, 50th,

and 75th percentiles, respectively. So, its effects are evenly distributed among students in the
 163

middle quantiles. However, while students in the 90th percentile benefit from math tutoring by

0.52, students in the 10th percentile do not receive any benefit of having the same math tutoring.

The differences in effects between quantiles become greater in the next year. Math tutoring in

grade 7 has a larger effect on math scores achieved in grade 8 than math achievement in grade 7.

Taking math tutoring in grade 7 has positive effects on math scores in grade 8 by 0.84, 0.97, and

1.12 standard deviations for students in the 50th, 75th, and 90th percentiles, respectively. And

these effects are much larger than the effects observed in grade 7. This ascending pattern of

effect sizes is in contrast to the pattern shown in the analyses for total tutoring and English

tutoring. However, in year 2, most of the effects of math tutoring in grade 7 disappear except for

the effect in the 75th percentile. With smaller effect sizes, this pattern is also revealed in the

OLS and PSM estimates; there are greater effects for students in the upper quantiles and these

effects last in the next year.

Interestingly, math tutoring in grade 8 seems to reduce the achievement gap in math

achievement, according to the 2SLAD and PSM methods. According to Table 67, while math

tutoring is not statistically significant in explaining math achievement for students in the 90th

percentile, it helps increase math scores for students in the lower quantiles. In addition, there is a

slightly larger effect for students in the 25th percentile (0.47) than students in the 75th quantiles

(0.40) in year 1 as well. The PSM estimates also suggest that math tutoring in grade 8 increases

math achievement by 0.14 for students in the 10th percentile, while it does not have significant

effects for students in the 75th and 90th percentiles. This pattern is also shown in the next year.

Thus, taking math tutoring in grade 8 seems to reduce the achievement gap in math achievement.

However, the effects of math tutoring in grade 9 do not show a consistently significant pattern in

terms of heterogeneity.
Table 60. Heterogeneous effects of verbal tutoring in grade 7 on verbal achievement for 3 years
OLS
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Verbal PT in grade 7 -0.01 -0.06 -0.03 -0.02 0.01 0.00 -0.05 -0.02 -0.04 -0.06 -0.07* -0.05 -0.02 -0.01 -0.03 0.03 0.00 -0.06*
(0.022) (0.045) (0.032) (0.021) (0.020) (0.024) (0.025) (0.039) (0.044) (0.034) (0.032) (0.039) (0.031) (0.050) (0.063) (0.034) (0.036) (0.029)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,855 5,855 5,855 5,855 5,855 5,855 5,357 5,357 5,357 5,357 5,357 5,357 4,073 4,073 4,073 4,073 4,073 4,073
R-squared 0.316 0.257 0.198
2SLAD
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Verbal PT in grade 7 0.01 0.04 -0.01 -0.03 0.04 0.01 -0.08 -0.03 -0.06 -0.13* -0.05 0.04 -0.12 -0.02 -0.09 -0.09 0.02 -0.11
(0.056) (0.101) (0.077) (0.088) (0.092) (0.090) (0.071) (0.128) (0.072) (0.063) (0.106) (0.120) (0.087) (0.174) (0.154) (0.113) (0.091) (0.080)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,435 5,825 5,825 5,825 5,825 5,825 5,004 5,346 5,346 5,346 5,346 5,346 3,790 4,058 4,058 4,058 4,058 4,058
R-squared 0.307 0.25 0.192
PSM
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Verbal PT in grade 7 0.02 0.02 0.02 -0.01 0.04 -0.00 -0.01 -0.14 0.01 -0.07 0.01 0.08 -0.02 -0.19 -0.19 0.06 -0.03 0.14
(0.034) (0.082) (0.073) (0.079) (0.080) (0.069) (0.041) (0.100) (0.084) (0.083) (0.091) (0.080) (0.050) (0.118) (0.096) (0.094) (0.095) (0.113)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,855 1,168 1,169 1,170 1,175 1,173 5,357 1,073 1,064 1,076 1,076 1,068 4,073 815 811 814 819 814
Pseudo R2 0.025 0.051 0.016 0.014 0.009 0.012 0.121 0.037 0.003 0.054 0.012 0.015 0.155 0.029 0.012 0.016 0.011 0.013
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G7VER_Z (verbal score in grade 7) for Year 0; G8VER_Z (verbal score in grade 8) for Year 1; G9VER_Z (verbal score in grade 9) for Year 2; (2) Complete tables for OLS and 2SLAD
are provided in Table K.1 and Table K.2 in Appendix.

164
Table 61. Heterogeneous effects of verbal tutoring in grade 8 on verbal achievement for 2 years
OLS
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Verbal PT in grade 8 0.04* 0.00 0.02 0.02 0.04 0.04 0.02 -0.02 -0.05 0.01 0.07* 0.02
(0.021) (0.055) (0.041) (0.027) (0.027) (0.033) (0.028) (0.088) (0.049) (0.035) (0.029) (0.044)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,478 5,478 5,478 5,478 5,478 5,478 4,164 4,164 4,164 4,164 4,164 4,164
R-squared 0.469 0.392
2SLAD
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Verbal PT in grade 8 0.00 -0.15 0.05 -0.03 0.00 -0.03 -0.05 -0.05 0.04 0.14 0.13 -0.02
(0.053) (0.130) (0.073) (0.057) (0.058) (0.094) (0.077) (0.140) (0.104) (0.076) (0.085) (0.081)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,935 4,991 4,991 4,991 4,991 4,991 3,728 3,738 3,738 3,738 3,738 3,738
R-squared 0.462 0.381
PSM
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Verbal PT in grade 8 0.04 -0.01 -0.09 0.11 0.10 0.13 0.03 0.14 0.16 0.04 0.27* -0.15
(0.041) (0.096) (0.079) (0.079) (0.079) (0.091) (0.054) (0.132) (0.104) (0.091) (0.109) (0.107)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,478 1,094 1,091 1,099 1,098 1,096 4,164 833 832 838 831 830
Pseudo R2 0.126 0.022 0.006 0.046 0.002 0.016 0.219 0.031 0.037 0.013 0.036 0.017
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G8VER_Z (verbal score in grade 8) for year 0; G9VER_Z (verbal score in grade 9) for year 1; (2) Complete tables for
OLS and 2SLAD are provided in Table K.3 and Table K.4 in Appendix.

165
Table 62. Heterogeneous effects of verbal tutoring in grade 9 on verbal achievement in grade 9
OLS
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

Verbal PT in grade 9 0.00 0.02 -0.01 -0.01 -0.04 -0.02

(0.027) (0.037) (0.039) (0.035) (0.034) (0.039)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,133 4,133 4,133 4,133 4,133 4,133
R-squared 0.425
2SLAD
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

Verbal PT in grade 9 -0.04 0.05 -0.15 0.00 0.09 -0.07

(0.093) (0.150) (0.080) (0.140) (0.129) (0.096)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 3,556 4,404 4,404 4,404 4,404 4,404
R-squared 0.414
PSM
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

Verbal PT in grade 9 -0.02 -0.14 0.00 0.15 -0.08 -0.08

(0.052) (0.111) (0.098) (0.097) (0.114) (0.116)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,133 834 815 825 825 833
Pseudo R2 0.164 0.012 0.014 0.024 0.017 0.011
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G9VER_Z (verbal score in grade 9); (2) Complete tables
for OLS and 2SLAD are provided in Table K.5 and Table K.6 in Appendix.

166
Table 63. Heterogeneous effects of English tutoring in grade 7 on English achievement for 3 years
OLS

Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

English PT in grade 7 0.25*** 0.13*** 0.21*** 0.27*** 0.33*** 0.33*** 0.12*** 0.02 0.08** 0.12** 0.21*** 0.21*** 0.07* 0.01 0.06 0.06 0.18*** 0.11
(0.024) (0.037) (0.023) (0.036) (0.030) (0.037) (0.029) (0.030) (0.028) (0.041) (0.044) (0.038) (0.033) (0.039) (0.039) (0.055) (0.042) (0.063)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,910 5,910 5,910 5,910 5,910 5,910 5,380 5,380 5,380 5,380 5,380 5,380 4,546 4,546 4,546 4,546 4,546 4,546
R-squared 0.403 0.338 0.314

2SLAD

Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

English PT in grade 7 0.72*** 0.32* 0.74*** 1.07*** 1.20*** 1.03*** 0.61*** 0.23 0.74** 1.02*** 0.86*** 0.78*** 0.64*** 0.42* 0.71*** 0.68*** 0.75*** 0.50***
(0.071) (0.145) (0.173) (0.126) (0.140) (0.115) (0.086) (0.239) (0.259) (0.154) (0.133) (0.183) (0.098) (0.191) (0.149) (0.108) (0.122) (0.066)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,487 5,880 5,880 5,880 5,880 5,880 5,023 5,364 5,364 5,364 5,364 5,364 4,248 4,532 4,532 4,532 4,532 4,532
R-squared 0.362 0.301 0.269

PSM

Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

English PT in grade 7 0.24*** 0.29*** 0.23** 0.24** 0.24* 0.32** 0.11* 0.09 0.14 0.17* 0.14 0.22 0.05 -0.02 0.02 0.16 0.23 0.12
(0.035) (0.057) (0.062) (0.075) (0.091) (0.099) (0.050) (0.073) (0.077) (0.087) (0.108) (0.122) (0.056) (0.085) (0.093) (0.099) (0.131) (0.124)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,910 1,183 1,181 1,179 1,183 1,184 5,380 1,071 1,080 1,077 1,081 1,071 4,546 906 909 911 918 901
Pseudo R2 0.084 0.039 0.010 0.009 0.005 0.014 0.191 0.039 0.035 0.009 0.019 0.017 0.219 0.049 0.023 0.014 0.016 0.024
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G7ENG_Z (English score in grade 7) for Year 0; G8ENG_Z (English score in grade 8) for Year 1; G9ENG_Z (English score in grade 9) for Year 2; (2) Complete tables for OLS and 2SLAD are provided in Table L.1 and Table

167
L.2 in Appendix.
Table 64. Heterogeneous effects of English tutoring in grade 8 on English achievement for 2 years
OLS
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

English PT in grade 8 0.14*** 0.08 0.13*** 0.16*** 0.14*** 0.16*** 0.09** 0.04 0.10** 0.06 0.10*** 0.08*
(0.024) (0.057) (0.025) (0.027) (0.031) (0.045) (0.030) (0.033) (0.032) (0.033) (0.027) (0.033)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,502 5,502 5,502 5,502 5,502 5,502 4,643 4,643 4,643 4,643 4,643 4,643
R-squared 0.561 0.504
2SLAD
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

English PT in grade 8 0.31*** 0.19 0.45*** 0.40*** 0.26* 0.30*** 0.18 -0.11 0.27 0.29** 0.12 0.12
(0.073) (0.108) (0.105) (0.090) (0.104) (0.068) (0.102) (0.206) (0.139) (0.101) (0.092) (0.157)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,955 5,003 5,003 5,003 5,003 5,003 4,188 4,230 4,230 4,230 4,230 4,230
R-squared 0.555 0.499
PSM
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

English PT in grade 8 0.18** 0.09 0.08 0.14 0.22* 0.06 0.20* 0.09 0.07 0.07 0.22 0.17
(0.049) (0.068) (0.080) (0.097) (0.104) (0.161) (0.077) (0.080) (0.093) (0.129) (0.174) (0.230)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,502 1,099 1,104 1,097 1,107 1,095 4,643 930 931 926 931 920
Pseudo R2 0.221 0.045 0.028 0.004 0.012 0.035 0.313 0.098 0.025 0.027 0.019 0.056
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G8VER_Z (verbal score in grade 8) for year 0; G9VER_Z (verbal score in grade 9) for year 1; (2) Complete tables for OLS
and 2SLAD are provided in Table L.3 and Table L.4 in Appendix.

168
Table 65. Heterogeneous effects of English tutoring in grade 9 on English achievement in grade 9
OLS
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 9 0.16*** 0.12** 0.14*** 0.16*** 0.23*** 0.17**

(0.028) (0.037) (0.028) (0.034) (0.029) (0.056)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,609 4,609 4,609 4,609 4,609 4,609
R-squared 0.536
2SLAD
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 9 -0.15 -0.14 -0.34 0.33 2.02 0.37

(0.272) (1.478) (1.406) (2.044) (1.919) (1.349)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,002 4,423 4,423 4,423 4,423 4,423
R-squared 0.524
PSM
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 9 0.22** 0.22** 0.19* 0.17 -0.08 0.62**

(0.068) (0.076) (0.088) (0.104) (0.129) (0.194)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,609 919 922 926 920 914
Pseudo R2 0.265 0.044 0.024 0.008 0.015 0.030
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G9ENG_Z (English score in grade 9); (2) Complete tables for OLS and 2SLAD are

169
provided in Table L.5 and Table L.6 in Appendix.
Table 66. Heterogeneous effects of math tutoring in grade 7 on math achievement for 3 years
OLS
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Math PT in grade 7 0.26*** 0.17*** 0.23*** 0.26*** 0.31*** 0.22*** 0.09** 0.04 0.04 0.10** 0.15*** 0.12* 0.04 -0.01 -0.01 0.03 0.12** 0.05
(0.024) (0.045) (0.034) (0.027) (0.028) (0.042) (0.030) (0.036) (0.037) (0.033) (0.042) (0.049) (0.033) (0.039) (0.042) (0.051) (0.042) (0.061)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,859 5,859 5,859 5,859 5,859 5,859 5,391 5,391 5,391 5,391 5,391 5,391 4,641 4,641 4,641 4,641 4,641 4,641
R-squared 0.348 0.286 0.259
2SLAD
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Math PT in grade 7 0.53*** 0.46 0.75*** 0.80*** 0.76*** 0.52*** 0.36*** 0.16 0.27 0.84*** 0.97** 1.12*** 0.34** 0.38 0.26 0.46 0.61** 0.61
(0.081) (0.312) (0.176) (0.131) (0.162) (0.106) (0.085) (0.220) (0.224) (0.237) (0.304) (0.230) (0.132) (0.260) (0.260) (0.295) (0.213) (0.347)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,445 5,830 5,830 5,830 5,830 5,830 5,035 5,372 5,372 5,372 5,372 5,372 4,336 4,622 4,622 4,622 4,622 4,622
R-squared 0.33 0.275 0.244
PSM
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Math PT in grade 7 0.25*** 0.19* 0.13 0.15 0.25* 0.19 0.16** 0.15* 0.08 0.12 0.15 0.24* 0.07 0.07 0.00 -0.11 0.11 0.23
(0.042) (0.081) (0.091) (0.093) (0.104) (0.097) (0.048) (0.067) (0.081) (0.092) (0.112) (0.122) (0.056) (0.084) (0.087) (0.108) (0.128) (0.145)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,859 809 901 932 945 986 5,391 1,072 1,083 1,080 1,084 1,072 4,641 929 928 926 926 932
Pseudo R2 0.075 0.040 0.014 0.017 0.007 0.012 0.187 0.039 0.045 0.016 0.019 0.019 0.210 0.064 0.022 0.011 0.020 0.038
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G7MAT_Z (math score in grade 7) for Year 0; G8MAT_Z (math score in grade 8) for Year 1; G9MAT_Z (math score in grade 9) for Year 2; (2) Complete tables for OLS and 2SLAD are provided in Table
M.1 and Table M.2 in Appendix.

170
Table 67. Heterogeneous effects of math tutoring in grade 8 on math achievement for 2 years
OLS
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Math PT in grade 8 0.17*** 0.16*** 0.13*** 0.17*** 0.20*** 0.19*** 0.11*** 0.04 0.11** 0.10** 0.13*** 0.15*
(0.026) (0.044) (0.036) (0.039) (0.048) (0.057) (0.032) (0.043) (0.036) (0.040) (0.037) (0.070)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,513 5,513 5,513 5,513 5,513 5,513 4,739 4,739 4,739 4,739 4,739 4,739
R-squared 0.471 0.419
2SLAD
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Math PT in grade 8 0.45*** 0.39** 0.52*** 0.48*** 0.40*** 0.15 0.43*** 0.28 0.47*** 0.48*** 0.40** 0.37
(0.083) (0.120) (0.141) (0.090) (0.106) (0.157) (0.111) (0.189) (0.132) (0.131) (0.151) (0.189)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,964 5,011 5,011 5,011 5,011 5,011 4,275 4,316 4,316 4,316 4,316 4,316
R-squared 0.455 0.405
PSM
Year 0 (Grade 8) Year 1 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

Math PT in grade 8 0.31*** 0.14* 0.06 0.29** 0.19 0.21 0.19* 0.13 0.19* 0.08 0.16 0.14
(0.057) (0.064) (0.082) (0.092) (0.117) (0.192) (0.076) (0.083) (0.087) (0.114) (0.148) (0.274)
Covariates ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Observations 5,513 1,096 1,111 1,100 1,108 1,097 4,739 946 950 947 946 950
Pseudo R2 0.223 0.046 0.014 0.021 0.008 0.047 0.305 0.111 0.021 0.025 0.010 0.059
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G8MAT_Z (math score in grade 8) for year 0; G9MAT_Z (math score in grade 9) for year 1; (2) Complete tables for OLS and 2SLAD are provided in
Table M.3 and Table M.4 in Appendix.

171
Table 68. Heterogeneous effects of math tutoring in grade 9 on math achievement in grade 9
OLS
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

Math PT in grade 9 0.24*** 0.16*** 0.22*** 0.23*** 0.32*** 0.30***

(0.031) (0.035) (0.033) (0.047) (0.059) (0.055)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,697 4,697 4,697 4,697 4,697 4,697
R-squared 0.443
2SLAD
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

Math PT in grade 9 -0.09 -0.17 -0.34 1.83 1.23 0.40

(0.287) (1.302) (1.575) (1.228) (1.336) (1.293)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,073 4,491 4,491 4,491 4,491 4,491
R-squared 0.427
PSM
Year 0 (Grade 9)
VARIABLES Avg q10 q25 q50 q75 q90

Math PT in grade 9 0.30*** 0.31*** 0.17 0.37** 0.06 0.00

(0.067) (0.074) (0.094) (0.097) (0.154) (0.257)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Observations 4,697 940 943 931 936 934
Pseudo R2 0.268 0.057 0.021 0.016 0.018 0.042
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: G9MAT_Z (math score in grade 9); (2) Complete tables for OLS and
2SLAD are provided in Table M.5 and Table M.6 in Appendix.

172
 173

6.3.2 Private Tutoring in Middle School: Long-term Heterogeneous Effects

Using the CSAT scores, I also observed the heterogeneous effects of private tutoring in

middle school on the university entrance exam. Heterogeneous effects of years of private

tutoring are not observed, according to all three estimations20. Similarly, cumulative measures of

verbal, English, and math tutoring do not show any differential effects depending on the

academic standing21.

Single-year analyses also tend to provide evidence on the absence of heterogeneity except

for private tutoring in grade 7, English tutoring in grades 7 and 8, and math tutoring in grade 7.

As explained, no heterogeneous effects are found in the analysis using cumulative measures of

private tutoring. Unlike the cumulative effects, private tutoring in grade 7 shows that it benefits

students in the 75th percentile by 1.01 deciles, with no significant effects for students in other

quantiles, according to PSM (Table 69). However, the OLS and 2SLAD estimates do not

support this result. Private tutoring in grades 8 and 9 does not show any heterogeneous effects22.

Verbal tutoring in grades 7 and 8 also does not heterogeneously affect verbal

achievement on the CSAT23, but verbal tutoring in grade 9 positively affects CSAT verbal scores

for students in the top quantile as presented in Table 70. According to the PSM estimates,

students in the top quantile benefit from verbal tutoring in grade 9 by 0.63 deciles, while other

students do not benefit from it. Also, distinguishing heterogeneity is observed in English

tutoring taken in grades 7 and 8. English tutoring in grade 7 improves English CSAT scores by

20
Results of the heterogeneous effects of one, two and three years of private tutoring are available in Table N.1 –
N.3 in Appendix.
21
Results of the heterogeneous effects of one, two and three years of verbal, English and math tutoring are available
in Table N.4 – N.12 in Appendix.
22
Results of the heterogeneous effects of private tutoring in grade 8 and 9 are available in Table O.2, and O.3 in
Appendix.
23
Results of the heterogeneous effects of verbal tutoring in grade 7 and 8 are available in Table O.4, and O.5 in
Appendix.
 174

0.70, 0.91, and 0.98 deciles for students in the 50th, 75th, and 90th percentiles, respectively, as

shown in Table 71. The PSM estimates also indicate that students in the 75th percentile benefit

from taking English tutoring in grade 7 by 0.56 deciles, while students in the 10th percentile

receive a negative impact on their English CSAT scores by a similar amount. Adversely,

English tutoring in grade 8 has positive and significant effects for students in the bottom quantile

only. Taking English tutoring in grade 8 increases English CSAT scores by 0.67 deciles for

students in the 10th percentile, while other students do not benefit from it, according to the PSM

estimates in Table 72. The OLS and 2SLAD estimates do not suggest any heterogeneous effects

of English tutoring in grade 8. However, English tutoring in grade 9 does not affect English

achievement on the CSAT24. In terms of math tutoring, math tutoring in grade 7 has a positive

and significant effect on CSAT math scores for students in the lower quantiles (Table 73). The

OLS estimates indicate that math tutoring in grade 7 increases math scores by 0.36 deciles for

students in the 10th percentile, and the PSM estimates show an increase in math scores by 0.71

deciles for students in the 25th percentile. However, the 2SLAD estimates are not consistent

with this result. Math tutoring in grade 8 and 9 does not show heterogeneous effects, which is

consistent with the previous results25.

24
Results of the effects of English tutoring in grade 9 are available in Table O.9 in Appendix.
25
Results of the effects of math tutoring by grade 8 and 9 are available in Table O.11, and O.12 in Appendix.
 175

Table 69. Long-term heterogeneous effects of private tutoring in grade 7 on overall achievement on
the CSAT
OLS
VARIABLES Avg q10 q25 q50 q75 q90

PT in grade 7 0.14 0.28 0.17 0.13 0.09 -0.00

(0.087) (0.158) (0.119) (0.125) (0.147) (0.153)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,905 1,905 1,905 1,905 1,905 1,905
R-squared 0.296
2SLAD
VARIABLES Avg q10 q25 q50 q75 q90

PT in grade 7 0.42 0.04 0.19 0.57 0.39 0.20

(0.274) (0.608) (0.142) (0.331) (0.368) (0.360)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,800 1,888 1,888 1,888 1,888 1,888
R-squared 0.287
PSM
VARIABLES Avg q10 q25 q50 q75 q90

PT in grade 7 0.33 -0.04 0.01 0.15 1.01* 0.74

(0.209) (0.267) (0.215) (0.347) (0.397) (0.701)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,905 270 328 405 449 349
Pseudo R2 0.227 0.107 0.029 0.059 0.056 0.300
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: CSAT_avg (average CSAT score); (2) Complete tables for
OLS and 2SLAD are provided in Table O.1 in Appendix.
 176

Table 70. Long-term heterogeneous effects of verbal tutoring in grade 9 on the CSAT verbal
achievement
OLS
VARIABLES Avg q10 q25 q50 q75 q90

Verbal PT in grade 9 -0.07 0.09 0.01 -0.10 -0.05 -0.32*

(0.072) (0.085) (0.080) (0.111) (0.097) (0.139)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,044 2,044 2,044 2,044 2,044 2,044
R-squared 0.252
2SLAD
VARIABLES Avg q10 q25 q50 q75 q90

Verbal PT in grade 9 0.07 -0.19 0.05 -0.12 0.01 0.74

(0.290) (0.608) (0.499) (0.274) (0.573) (0.830)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,812 2,273 2,273 2,273 2,273 2,273
R-squared 0.243
PSM
VARIABLES Avg q10 q25 q50 q75 q90

Verbal PT in grade 9 -0.04 0.16 -0.13 0.03 -0.11 0.63*

(0.129) (0.327) (0.239) (0.234) (0.253) (0.246)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Verbal PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,044 350 394 431 426 443
Pseudo R2 0.201 0.077 0.034 0.041 0.031 0.045
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: CSAT_V (CSAT verbal score); (2) Complete tables for OLS and
2SLAD are provided in Table O.6 in Appendix.
 177

Table 71. Long-term heterogeneous effects of English tutoring in grade 7 on the CSAT English
achievement
OLS
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 7 0.10 0.19 0.01 0.11 0.16 -0.07

(0.085) (0.167) (0.108) (0.107) (0.120) (0.166)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
English PT in high
✓ ✓ ✓ ✓ ✓ ✓
school
Observations 2,302 2,302 2,302 2,302 2,302 2,302
R-squared 0.247
2SLAD
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 7 0.65** 0.32 0.05 0.70*** 0.91** 0.98*

(0.251) (0.336) (0.453) (0.199) (0.302) (0.407)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
English PT in high
✓ ✓ ✓ ✓ ✓ ✓
school
Observations 2,179 2,281 2,281 2,281 2,281 2,281
R-squared 0.229
PSM
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 7 0.20 -0.58* 0.30 -0.03 0.56* -0.33

(0.145) (0.274) (0.253) (0.282) (0.276) (0.277)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
English PT in high
✓ ✓ ✓ ✓ ✓ ✓
school
Observations 2,302 338 410 491 519 536
Pseudo R2 0.174 0.071 0.035 0.022 0.034 0.048
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: CSAT_E (CSAT English score); (2) Complete tables for OLS and
2SLAD are provided in Table O.7 in Appendix.
 178

Table 72. Heterogeneous effects of English tutoring in grade 8 on the CSAT English achievement
OLS
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 8 0.13 0.10 0.17 0.16 0.09 0.03

(0.090) (0.164) (0.129) (0.102) (0.125) (0.170)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
English PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,339 2,339 2,339 2,339 2,339 2,339
R-squared 0.342
2SLAD
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 8 0.39 0.33 0.39 1.04* 0.20 0.60

(0.317) (0.507) (0.437) (0.430) (0.389) (0.467)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
English PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,148 2,161 2,161 2,161 2,161 2,161
R-squared 0.342
PSM
VARIABLES Avg q10 q25 q50 q75 q90

English PT in grade 8 0.15 0.67* -0.02 -0.38 0.37 0.97

(0.194) (0.323) (0.239) (0.403) (0.384) (0.708)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 7 ✓ ✓ ✓ ✓ ✓ ✓
English PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
English PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,339 302 469 477 527 562
Pseudo R2 0.307 0.171 0.046 0.024 0.022 0.055
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: CSAT_E (CSAT English score); (2) Complete tables for OLS and
2SLAD are provided in Table O.8 in Appendix.
 179

Table 73. Heterogeneous effects of math tutoring in grade 7 on the CSAT math achievement
OLS
VARIABLES Avg q10 q25 q50 q75 q90

Math PT in grade 7 0.11 0.36* 0.12 0.06 -0.02 0.14

(0.091) (0.174) (0.153) (0.094) (0.151) (0.197)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,294 2,294 2,294 2,294 2,294 2,294
R-squared 0.238
2SLAD
VARIABLES Avg q10 q25 q50 q75 q90

Math PT in grade 7 -0.13 0.03 -0.26 -0.54 -0.06 2.62**

(0.450) (0.787) (0.578) (0.489) (0.728) (0.987)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,172 2,265 2,265 2,265 2,265 2,265
R-squared 0.236
PSM
VARIABLES Avg q10 q25 q50 q75 q90

Math PT in grade 7 0.08 -0.01 0.71* 0.28 0.55 0.16

(0.175) (0.305) (0.278) (0.269) (0.399) (0.452)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 8 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 9 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in high school ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,294 350 425 463 509 542
Pseudo R2 0.186 0.098 0.050 0.016 0.046 0.053
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: CSAT_M (CSAT Math score); (2) Complete tables for OLS and
2SLAD are provided in Table O.10 in Appendix.
 180

6.3.3 Private Tutoring in High School: Short-term Heterogeneous Effects

Private tutoring in high school seems to improve educational inequality on the CSAT.

For example, while one year of private tutoring in high school is not significant in explaining

heterogeneity26, two years of private tutoring in high school improves overall CSAT results by

0.45, 0.52, and 0.37 deciles for students in the 10th, 25th, and 50th percentiles, respectively,

according to the OLS estimation (Table 74). The 2SLAD estimation also suggests that it has

positive and significant effects only for students in the 25th and 50th percentiles by 0.71 and 0.57

deciles, respectively, without having significant effects for students in the upper quantiles.

However, the PSM estimates are not consistent with this result. Moreover, three years of private

tutoring in high school shows a greater impact on the CSAT for students in the lower quantiles,

which is consistent with all three estimation methods (Table 75). The OLS estimates show that

three years of private tutoring in high school increases overall achievement on the CSAT for

students in all quantiles, but the effects for students in the lower quantiles are greater than the

ones for students in the upper quantiles. It improves CSAT scores for students in the 10th and

25th percentiles by 0.56 and 0.72 deciles, respectively, while it benefits students in the 75th and

90th percentiles by 0.48 and 0.45 deciles each. Similarly, the 2SLAD estimates explain that

students in the 10th and 25th percentiles benefit from three years of tutoring in high school by

0.54 and 0.75 deciles, respectively, while students in the 75th percentile increase their CSAT

scores by 0.46 deciles after taking three years of private tutoring in high school. Moreover, the

PSM estimates also show that only students in the 25th percentile receive higher CSAT scores by

0.75 deciles after three years of tutoring, without having significant effects for students in other

quantiles. Therefore, it is evident that three years of private tutoring improves overall CSAT

26
Results of the effects of private tutoring in high school are available in Table P.1 in Appendix.
 181

achievement only for students in the lower quantiles, which contributed to reducing the

achievement gap. This pattern is opposite to the effects of private tutoring in middle school,

which often contributes to exacerbating the achievement gap as explained in the previous section.

Table 74. Heterogeneous effects of 2 years of private tutoring in high school on the CSAT
OLS
VARIABLES Average q10 q25 q50 q75 q90

2 years of PT in high school 0.29** 0.45** 0.52*** 0.37** 0.09 0.14

(0.101) (0.149) (0.122) (0.117) (0.152) (0.194)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 826 826 826 826 826 826
R-squared 0.407
2SLAD
VARIABLES Average q10 q25 q50 q75 q90

2 years of PT in high school 0.51 0.51 0.71*** 0.57*** 0.44 0.32

(0.304) (0.285) (0.162) (0.127) (0.241) (0.326)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 627 1,571 1,571 1,571 1,571 1,571
R-squared 0.402
PSM
VARIABLES Average q10 q25 q50 q75 q90

2 years of PT in high school 0.26 0.15 0.43 0.44 0.78 -0.46

(0.232) (0.729) (0.306) (0.343) (0.459) (0.584)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 826 83 125 187 207 223
Pseudo R2 0.235 0.224 0.164 0.049 0.062 0.142
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_avg (average CSAT score); (2) Complete tables for OLS and 2SLAD are
provided in Table P.2 in Appendix.
 182

Table 75. Heterogeneous effects of 3 years of private tutoring in high school on the CSAT
OLS
VARIABLES Average q10 q25 q50 q75 q90

3 years of PT in high school 0.59*** 0.56* 0.72*** 0.64*** 0.48*** 0.45**

(0.105) (0.245) (0.166) (0.173) (0.140) (0.175)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 985 985 985 985 985 985
R-squared 0.453
2SLAD
VARIABLES Average q10 q25 q50 q75 q90

3 years of PT in high school 0.51* 0.54* 0.75*** 0.61*** 0.46* 0.34

(0.251) (0.268) (0.184) (0.160) (0.214) (0.329)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 782 1,571 1,571 1,571 1,571 1,571
R-squared 0.442
PSM
VARIABLES Average q10 q25 q50 q75 q90

3 years of PT in high school 0.30 0.33 0.75* 0.21 0.41 1.09

(0.322) (0.938) (0.273) (0.346) (0.367) (1.461)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 985 93 166 241 233 120
Pseudo R2 0.393 0.476 0.158 0.062 0.045 0.395
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_avg (average CSAT score); (2) Complete tables for OLS and 2SLAD are
provided in Table P.3 in Appendix.
 183

When analyzing the cumulative effects by academic subject, distinguished patterns are

shown only in math tutoring. There are no consistent and significant cumulative effects of verbal

and English tutoring based on the three estimation models27. However, three years of math

tutoring in high school shows heterogeneous effects. As shown in Table 76, the results suggest

that students in the 10th and 25th percentiles see greater effects of three years of math tutoring in

high school than students in the 75th and 90th percentiles. CSAT math scores go up by 1.32 and

1.35 deciles for students in the 10th and 25th percentiles, respectively, while scores increase by

0.98 and 0.93 deciles for those in the 75th and 90th percentiles, respectively. Similar patterns are

observed when estimating these effects with the 2SLAD method. In the PSM estimation,

students only in the 25th and 50th percentiles benefit from it by 1.18 and 1.11 deciles,

respectively, while students in other quantiles do not experience significant changes on their

CSAT math scores. Thus, even though one and two years of math tutoring do not show

consistent results in heterogeneity28, three years of math tutoring contributes to improve the math

achievement for students in the lower quantiles, which is beneficial in reducing educational

inequality on the CSAT math achievement.

27
Results of the heterogeneous effects of years of verbal and English tutoring in high school are available in Table
P.4 – P.9 in Appendix.
28
Results for the heterogeneous effects of one and two years of math tutoring in high school are available in Table
P.10 and P.11 in Appendix.
 184

Table 76. Heterogeneous effects of 3 years of math tutoring on the CSAT math achievement

OLS
VARIABLES Average q10 q25 q50 q75 q90

3 years of Math PT in high school 1.18*** 1.32*** 1.35*** 1.21*** 0.98*** 0.93***
(0.121) (0.233) (0.185) (0.176) (0.233) (0.245)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,081 1,081 1,081 1,081 1,081 1,081
R-squared 0.401
2SLAD
VARIABLES Average q10 q25 q50 q75 q90

3 years of Math PT in high school 1.08*** 0.91* 1.32*** 1.26*** 1.10*** 0.85**
(0.246) (0.363) (0.228) (0.269) (0.333) (0.327)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 854 1,899 1,899 1,899 1,899 1,899
R-squared 0.386
PSM
VARIABLES Average q10 q25 q50 q75 q90

3 years of Math PT in high school 0.85* 1.00 1.18* 1.11** 0.52 0.60
(0.318) (0.877) (0.415) (0.316) (0.530) (0.617)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,081 102 175 260 263 224
Pseudo R2 0.383 0.604 0.110 0.046 0.047 0.093
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variable: CSAT_M (CSAT Math score); (2) Complete tables for OLS and 2SLAD are
provided in Table P.12 in Appendix.
 185

The results of single-year effects are consistent with the previous results of the

cumulative effects. The results suggest that there are no significant single-year heterogeneous

effects of overall, verbal, and English tutoring29. However, the significant heterogeneous effects

in all the estimations are observed in math tutoring, especially tutoring in grade 12 (Table 77).

Math tutoring in grade 12 increases CSAT math scores for students in all quantiles, according to

the OLS estimates. However, it is most beneficial for students in the 10th percentile by 0.57

deciles, which is about twice as large as the effect for students in the 90th percentile (0.27

deciles). The 2SLAD estimates also suggest that math tutoring in grade 12 increases math scores

by 0.73 deciles for students in the 25th percentiles, while students in the 75th percentile benefit

from it by 0.61 deciles. The differences in the effects between the lower and upper quantiles are

larger when estimating with the PSM method. Students in the 25th percentile have higher math

scores by 1.04 deciles after private tutoring in grade 12, while it increases scores by 0.40 and

0.59 deciles for students in the 75th and 90th percentiles, respectively. Even though there are no

significant heterogeneous effects of math tutoring in grades 10 and 1130, math tutoring in grade

12 shows significant differential effects on the CSAT math achievement by quantile.

Furthermore, it helps reduce the math achievement gap on the CSAT.

29
Results for the single-year heterogeneous effects of overall, verbal and English tutoring are available in Table Q.1
– Q.9 in Appendix.
30
Results for the single-year effects of math tutoring in grades 10 and 11 are available in Table Q10 and Q.11 in
Appendix.
 186

Table 77. Heterogeneous effects of math tutoring in grade 12 on the CSAT math achievement
OLS
VARIABLES Average q10 q25 q50 q75 q90

Math PT in grade 12 0.43*** 0.57*** 0.49*** 0.39*** 0.37*** 0.27*

(0.072) (0.160) (0.087) (0.085) (0.102) (0.120)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 10 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 11 ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,338 2,338 2,338 2,338 2,338 2,338
R-squared 0.345
2SLAD
VARIABLES Average q10 q25 q50 q75 q90

Math PT in grade 12 0.42** 0.40 0.73* 0.48 0.61* 0.45

(0.157) (0.362) (0.285) (0.252) (0.287) (0.366)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 10 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 11 ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,117 2,117 2,117 2,117 2,117 2,117
R-squared 0.343
PSM
VARIABLES Average q10 q25 q50 q75 q90

Math PT in grade 12 0.37** 0.62 1.04* 0.71** 0.40* 0.59**

(0.113) (0.806) (0.363) (0.243) (0.179) (0.184)
Covariates ✓ ✓ ✓ ✓ ✓ ✓
Math PT in middle school ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 10 ✓ ✓ ✓ ✓ ✓ ✓
Math PT in grade 11 ✓ ✓ ✓ ✓ ✓ ✓
Observations 2,338 264 431 474 547 580
Pseudo R2 0.204 0.042 0.022 0.062 0.017 0.025
*** p<0.001, ** p<0.01, * p<0.05
Notes: (1) Dependent variables: CSAT_M (CSAT Math score); (2) Complete tables for OLS and 2SLAD
are provided in Table Q.12 in Appendix.

6.4 Discussion

In summary, private tutoring differentially affects students’ achievement depending on

students’ academic ability. In terms of total private tutoring taken during middle school years,

on average, its effects on the average achievement in grade 9 for high achievers are larger than

the effects for low achievers. Similar patterns appear in the differential effects of English and

math tutoring in middle school without having significant heterogeneous effects of verbal
 187

tutoring in middle school. Therefore, it is evident that private tutoring taken during middle

school years broadens the achievement gap in middle school between high- and low-achieving

students, which exacerbates educational inequality. These findings correspond with the existing

literature, which claims that private tutoring adversely affects educational inequality (Bray, 1999;

Yang, 2007; Tansel & Bircan, 2005). Moreover, this study also finds interesting results in terms

of different effects by grade level, which previous literatures have not analyzed. The results

suggest that tutoring at an earlier grade level shows more dramatic heterogeneity than tutoring at

a later grade level. In other words, the heterogeneous effects of tutoring in grade 7 are greater

than the heterogeneous effects of tutoring in grade 8. Also, this heterogeneity shown in grade 7

is sustained for more years, while the heterogeneity in grade 8 does not last more than one year.

Therefore, private tutoring at an early grade level in middle school has a more negative impact

on educational inequality than tutoring at later grade levels. This pattern also applies to English

and math tutoring.

Instead of using academic achievement in middle school as dependent variables, I used

the CSAT results to observe the long-term effects of private tutoring in middle school on

performance in the university entrance examination. Similar to the average effects discussed in

Chapter IV, significant heterogeneous effects were detected only for English tutoring, and high-

achieving students benefit more from it compared to low-achieving students. As before, the

heterogeneity is severe in English tutoring taken in grade 7; it broadens the achievement gap in

CSAT English by about 1 decile between students in the lower and upper quantiles (Table 73).

However, private tutoring in high school years affects academic achievement in an

opposite direction. Overall findings indicate that private tutoring in high school contributes to

reduce the achievement gap between low- and high-achieving students as opposed to private
 188

tutoring in middle school. Taking three years of private tutoring increases overall CSAT scores

by 0.75 deciles for students in the 25th percentile, which is larger than the effects for those in the

upper quantiles (Table 75). Especially, math tutoring in high school shows that the effect for low

achievers is larger than the one for high achievers (Table 76 and 77). Overall results explain that

private tutoring in high school does not help improve the achievement of high achievers, whereas

it is more beneficial to low achievers, which contradicts the arguments in the existing literature

(Bray, 1999; Yang, 2007; Tansel & Bircan, 2005).

This positive aspect of private tutoring in high school in terms of educational inequality

can be interpreted in two ways. First, it can be explained by diminishing marginal returns to

education. It means that private tutoring predominantly influences academic achievement for

low-achieving students who have a large room to grow, but the same private tutoring has a

relatively small influence for high achievers who have less room for growth. However, this

argument does not apply to private tutoring in middle school, which has a larger effect on high-

achieving students. Secondly, the characteristics of private tutoring in high school may explain

the positive influence of high school tutoring on educational inequality. As mentioned before,

tutoring in high school is more focused on preparing for the CSAT (Lee, Park & Lee, 2009),

which is different from tutoring in middle school that mimics formal education. Thus, private

tutoring in high school is more like “drill and practice” designed for the CSAT that would be

more necessary for low achievers than high achievers. With respect to this type of private

tutoring, the effect would be maximized if students take tutoring when the examination is

approaching. The results also follow this scenario because private tutoring in later grades

(grades 11 and 12) in high school is more effective than tutoring at an early grade level (grade

10).
 189

This study of heterogeneity in private tutoring draws interesting social implications.

Even though private tutoring in middle school contributes to exacerbate educational inequality

between ability groups, which can also be considered as income groups, private tutoring in high

school plays an opposite role to improve educational inequality by benefiting low achievers to a

larger extent. Therefore, it is hard to generalize the effect of private tutoring on educational

inequality, and more attention should be placed on the characteristics of private tutoring, which

play a different role in inequality in education.

 190

CHAPTER VII: CONCLUSION

Over the last two decades, private tutoring has emerged as an important area in education

as its demand has been growing around the world. The private tutoring sector is often

considered the third emerging education sector in addition to public and private school sectors

(Dang & Rogers, 2008). Especially, private tutoring in Asia is most predominant and expanding

at an alarming rate (Bray & Lykins, 2012). However, empirical studies that explored the causal

impact of private tutoring are lacking. Moreover, the existing evidence of its effectiveness is still

mixed. Using the Korean Educational Longitudinal Survey (KELS), this dissertation aimed at

investigating the causal impact of private tutoring in Korea on three outcomes: students’

academic achievement, the quality of the learning environment in formal schooling, and

educational inequality. In order to uncover the causal relationship between private tutoring and

these three outcomes, several quasi-experimental methods were employed as described in the

following section with a summary of findings.

7.1 Summary of Findings

For the first research question, I explored how private tutoring affects students’ academic

achievement in both the short- and long-term by using Ordinary Least Squares, Instrumental

Variable, and Propensity Score Matching methods. The results suggest that private tutoring in

middle school, on average, has positive short-term effects on students’ academic achievement in

middle school, but the long-term effects of middle school tutoring on the CSAT are minimal. By

subject area, English and math tutoring in middle school are effective in improving academic

achievement in English and math during middle school years, whereas verbal tutoring does not
 191

have a positive influence on verbal performance in middle school. However, in the long-term,

only English tutoring in middle school is helpful to achieve better CSAT English performance.

It is also important to note that taking private tutoring in grade 7 is the most influential in

increasing academic achievement in middle school years.

The effects of private tutoring in high school on overall CSAT performance are mixed.

However, by subject, math tutoring in high school has a significant impact on the CSAT. When

observing the impact by grade level, math tutoring in grades 11 and 12 increases CSAT math

scores by about 0.4 to 0.6 deciles, which are quite large. Overall, private tutoring in middle

school and high school brings positive impact on students’ academic performance in secondary

school. However, the cost-effectiveness is still questionable.

The second research question focuses on observing the impact of private tutoring on

students’ formal schooling using the degree of students’ attention to lessons in formal schools as

a proxy for the quality of formal schooling by employing Ordered Logit, Propensity Score

Matching, and Difference-in-Differences methods. On average, private tutoring shows a positive

influence on student’s attention to lessons only in grade 8, but the magnitude of the influence is

not substantial. However, when analyzing it by ability groups, positive effects are detected

mostly in the low ability group, which means that low achievers pay more attention to lessons in

formal schools if they take tutoring. It also improves the overall learning environment in

classrooms, which, in turn, increases the quality of schooling. However, private tutoring seems

to have a different effect on high achievers. The regulation on the operating hours of private

tutoring institutes demonstrated that high achievers pay more attention to lessons in formal

schools when tutoring hours are cut back. However, this regulation does not influence low
 192

achievers. The overall results suggest that private tutoring does affect the learning environment

in formal schooling in a positive direction.

The last research question aims at finding the causal relationship between private tutoring

and educational inequality by using Quantile Regression, Two-Stage Least Absolute Deviation

estimator, and Propensity Score Subclassification. The overall results indicate that private

tutoring has heterogeneous effects on academic achievement. In the case of private tutoring in

middle school, its effects on high achievers are, on average, bigger than that on low achievers.

This pattern also appears in terms of the effects of English and math tutoring in middle school.

Thus, private tutoring taken in middle school broadens the achievement gap between high and

low achievers, which exacerbates educational inequality. In addition, tutoring at an earlier grade

level shows more dramatic heterogeneity than tutoring at a later grade level, which means that

tutoring taken in grade 7 has the largest influence in widening the achievement gap in middle

school. However, except for English tutoring, most of private tutoring in middle school does not

show significant heterogeneity in terms of its effects on the CSAT performance.

In contrast to the effects of private tutoring in middle school, private tutoring in high

school contributes to reduce educational inequality. On average, low achievers benefit more on

the CSAT from taking private tutoring in high school compared to high achievers. For example,

three years of private tutoring is beneficial only for students at the 25th percentile. Moreover,

three years of math tutoring and math tutoring in grade 12 contribute the most to lessen the

achievement gap on the CSAT. Therefore, private tutoring in high school brings a positive

influence to the society by reducing educational inequality, whereas private tutoring in middle

school worsens educational inequality.

 193

7.2 Policy Responses

7.2.1 History of policies on private tutoring in Korea

In the past decades, the Korean government has put a lot of effort into curbing the

demand for private tutoring by proposing various policies. Most of the approaches from the

1960s to 1980s focused on reforming assessment and selection systems, which is one of the

domains that Bray and Lykins (2012) have addressed. When the demand for private tutoring

emerged as a social problem in the 1960s, the Korean government, as its very first approach,

abolished entrance examinations to middle school in 1969, which was the major driver that

fanned the demand for private tutoring. Since then, students have been assigned to neighborhood

middle schools by lottery. However, the desire for private tutoring did not subside, and parents

of middle school students showed a tremendous demand for private tutoring for high school

admission (Kim & Chang, 2010). To cool off this consistent demand, the Korean government

enacted the High School Equalization Policy in 1974, which abolished the high school entrance

examination and deprived students of the freedom to choose their high schools. However, this

policy was also unsuccessful to alleviate this demand. As the most radical approach, in 1980, the

military government announced the Educational Reform Bill, which placed a total ban on all the

supplementary, for-profit, tutoring activities. Even though all the tutoring activities were

prohibited, there were still illegal tutoring lessons for students and parents with a desire to enter

prestigious universities. After acknowledging that even a total ban does not solve the problem,

the military government started relaxing the ban in stages from 1981 (Ministry of Education,

1989).

Even after year 2000, when the Constitutional Court declared that the prohibition of

private tutoring was an infringement of human rights and unconstitutional (Kim & Chang, 2010),
 194

various direct regulations and plans have been introduced. The Korean government started

quality control of private tutoring institutions in terms of facility, qualifications of tutors, and

fees of private tutoring institutions, etc. In addition, the government introduced a national model

of private tutoring (Han, 2004). As substitutes for private tutoring, the government introduced

the Educational Broadcasting System (EBS). EBS broadcasts tutoring lessons instructed by

school teachers or famous tutors from the private tutoring sector, free of charge. In addition, the

government supported formal schools to strengthen their after-school programs by diversifying

the curriculum based on students’ academic levels and by hiring famous tutors. Recently, the

entrance examination to enter special purpose high schools (SPHSs), which was introduced in

the late 1970s as a response to criticism of the High School Equalization Policy, has been

abolished to cut back on the demand for private tutoring. As described, the Korean government

attempted various approaches in several domains regarding the systems of assessment and

selection, curriculum, technology, and so forth (Bray & Lykins, 2012). Even though the demand

has been slowly decreasing, according to the national statistics, all of these efforts did not

dramatically change household consumption of private tutoring for children. Having with these

historical approaches in mind, I propose several suggestions to policy makers in Korea.

7.2.2 Suggestions to Policy Makers in Education

First, the Korean government should support formal schools to create a consumer-

responsive curriculum. As explained in Chapter III, one of the major reasons students participate

in private tutoring is because formal education fails to meet the educational needs of students and

parents (Chun et al., 2003; Kim, 2004). Due to this reason, various programs have been

introduced to formal schools, but they do not satisfy students and parents for the following
 195

reasons. First, a large part of the curriculum is targeted at average ability students without

diversifying the educational levels. However, the curriculum targeting average students does not

satisfy others who have different levels of ability. Especially, for high-achieving students with a

high probability of taking private tutoring, the curriculum in formal schooling is not challenging

and does not satisfy their educational needs. In addition, low-achieving students face difficulties

in following the curriculum because it is hard to receive additional support in formal schools.

Fortunately, after-school programs that have been strengthened since the mid-2000s provide

substantial support for these low-achieving disadvantaged students, but high achievers still have

to satisfy their educational desire through private tutoring. Therefore, the government should

support schools to create a quality curriculum that meets the needs of students of different levels.

This can be made possible if the government provides more financial support and gives more

flexibility to schools in designing their curriculum. The lack of flexibility under the current

heavy regulation and strict control on the formal education system hampers schools from

designing a consumer-responsive curriculum.

Secondly, the government should pay more attention to better allocate resources in order

to improve the quality of formal education. Quality education entails many aspects of education

such as curriculum, teacher proficiency, facilities, and additional services, etc. However, the

government’s resource allocation hasn’t seemed balanced in recent years, so all these aspects of

education have not been equally emphasized. In addition, resources have been sporadically

allocated and altered depending on one or two influential government authorities without

establishing a long-term goal. For example, recently, most of the resources in education have

been allocated to the national free meal program for all students in elementary and middle

schools, regardless of students’ economic status, as proposed in 2011. Due to this program,
 196

resources that were allocated to the curriculum and facilities had to be reallocated to this free

meal program even though there are plenty of schools in dire need of the money for their

curriculum and facilities. It became a huge problem for many schools in rural areas that still

have old buildings with wooden floors and leaking ceilings without heating systems (Choi, 2012).

Furthermore, budgets for curriculum development and after-school programs have also shrunk

(Kim, 2013). This spontaneous budget planning is somewhat attributed to the tight linkage

between education and politics in Korea. Since educational policies are often used by politicians

to attract votes, a number of policies and initiatives have been proposed and executed without

proper consideration of their consequences on students and schools. Therefore, it is crucial to

allocate educational resources in balance with a long-term plan to improve the overall quality of

formal schooling in order to satisfy students and parents.

As another way to improve the quality of formal schooling to decrease the demand for

private tutoring, the government should create realistic plans for teacher empowerment. “The

quality of a school system rests on the quality of its teachers (McKinsey & Company, 2007).”

As many articles have already introduced, school teachers in Korea are recruited from the top 5

percent of each cohort of college graduates. Even though teacher quality in Korea is ranked top

among OECD countries (McKinsey & Company, 2007), not enough attention has been paid to

teacher empowerment after recruitment. For example, there is a lack of incentives for individual

teachers to improve their performance. In other words, there is no premium system for teachers

who make consistent efforts to improve their pedagogy and adopt new materials and technology

to maximize educational outputs. Those teachers can be recognized through evaluations from

stakeholders in education, but there are mostly bureaucratic evaluations in Korea. Due to this

environment in the teacher market, school teachers become less inclined to change and less
 197

likely to stay attuned to market dynamics in education compared to instructors in the private

tutoring sector who are under fierce competition and high-stakes evaluations. This would create

a large discrepancy between formal schooling and the educational needs of students and parents.

I am not arguing that competition and evaluation are a panacea since there are many side effects

of competition among teachers and high-stakes teacher evaluations as previous research has

argued (Glazerman & Seifullah, 2010). Acknowledging undesirable effects of existing teacher

policies, the government should place more attention on creating a motivating environment to

empower school teachers.

The grading system in secondary schools should be revised in order to reduce the

influence of private tutoring on students’ overall academic performance. The current grading

system is mostly exam-oriented; grade point average (GPA) is usually based on the performance

of two examinations in each semester. Students only have limited opportunities to get evaluated,

and there are no other alternatives to make up for their performance if they do not perform well

on the examinations. This uniform and exam-oriented grading system encourages fierce

competition among students and may drive up the demand for private tutoring in order to excel

in the examinations. As the empirical results suggest in Chapter IV, private tutoring in middle

school positively affects students’ academic achievement during middle school years. Thus, the

current grading system indirectly encourages students to receive additional tutoring since the

influence of private tutoring on their GPA can be substantial. To solve this issue, schools must

introduce various criteria or tools to evaluate students. For example, schools may reduce the

weight of exam results on the GPA and introduce other methods for evaluation such as written

assignments, quizzes, and class participation, etc. By mixing objective and subjective methods

of evaluation, teachers should evaluate students by considering various aspects instead of solely
 198

using test scores. The Korean government has started showing concerns regarding the grading

system since 2011, and the grading system has changed from relative evaluation to absolute

evaluation by giving schools more flexibility to evaluate students. However, it is important to

establish specific grading rubrics for absolute evaluation in order to prevent teachers and schools

from misusing the evaluation system. Therefore, the demand for private tutoring would decrease

if the weight of tests on the GPA is reduced by introducing other methods in the grading system.

Last but not least, parents should be actively engaged in formal education. Schools

should urge parents to participate in various school activities and offer them many opportunities

for communication. Through these opportunities and activities, parents would be fully aware of

what is happening to students in schools. Schools should also be aware of the needs of students

and parents by communicating with each other in order to increase the quality of formal

schooling and create a consumer-responsive curriculum. Instead of having students and parents

fulfill their educational needs in the private tutoring sector, schools should play a major role in

satisfying parents and students by incorporating their educational needs in the formal curriculum.

This is possible when there is a steady communication channel between parents and schools.

As a result, it is crucial to improve the quality of formal education by emphasizing a

consumer-responsive curriculum, balanced resource allocation, teacher empowerment, improved

grading system, and parent engagement. If students and parents are satisfied with formal

schooling, parents will naturally reduce their demand for private tutoring for their children. All

the future policies related to private tutoring should consider these elements of education.

7.3 Limitations of the Study and Directions for Future Research

Several limitations result from the availability and structure of variables. First of all, the

achievement scores during high school years are not available in the KELS data, so it was
 199

impossible to observe the effect of private tutoring on academic achievement during high school

years. The only available achievement data in high school is the CSAT results, which enabled

me to explore the effect of private tutoring on the university entrance examination. However, the

CSAT data is measured by decile rank, which is not as precise as percentile rank. This crude

measure of the CSAT makes it difficult to accurately calculate the estimates of private tutoring

effects on the CSAT. Secondly, this study is not able to explore the effect of private tutoring on

college attendance, which is a better measure to investigate both academic performance and

educational inequality caused by private tutoring. Variables that indicate whether students attend

colleges and what type of colleges they attend are available in the 7th wave of the KELS data,

which are not publicly available yet. In future research, it will be crucial to use more accurately

measured variables in order to reduce biases that result from measurement error. In addition,

more diverse measures in terms of students’ academic achievement will be necessary to explore

various educational outcomes of students.

Besides the limitations related to availability and precision of variables, the biggest

limitation of this study is that it fails to render implications on the cost-effectiveness of private

tutoring, which stakeholders in education would find the most curious and useful. This study

was able to explore the effectiveness of private tutoring as presented in Chapter IV. However, it

is still under the veil whether taking private tutoring is cost-effective after taking costs of private

tutoring into account. As a mini study of cost-effectiveness of private tutoring, I compared the

cost-effective ratios of four types of private tutoring as follows.

 200

Table 78. Cost-effectiveness ratios of four types of private tutoring

Effects on average
Annual Costs
score in grade 9 C/E ratio
(monthly cost x 12)
(PSM estimates)
Three years of private tutoring 665.28 0.31 2146.06

Private tutoring in grade 7 159.96 0.12 1333.00

Private tutoring in grade 8 202.68 0.21 965.14

Private tutoring in grade 9 302.64 0.22 1375.64

Since information on the monthly cost of each type of private tutoring is available in the KELS

data, I multiplied it by 12 to calculate the annual costs of each type of private tutoring by

assuming that students take tutoring year-round. In terms of the effects, I used the estimates for

the effect of each type on academic achievement in grade 9. According to the cost-effectiveness

ratios in Table 78, private tutoring in grade 8 is the most cost-effective to raise academic

achievement in grade 9, followed by private tutoring in grade 7 and 9. Three years of private

tutoring is the least cost-effective, which means that it requires the largest amount of cost to see

the same effect. Future research should include a further study on the cost-effectiveness of

private tutoring, which will draw more practical inferences on utilizing private tutoring for

students in secondary schools.

 201

REFERENCE

Angrist, J. D., Imbens, G W. & Rubin, D. B. (1996). Identification of Causal Effects Using
Instrumental Variables. Journal of the American Statistical Association, 91, 444-472.

Assad, R. & El-Badawy, A. (2004). Private and group tutoring in Egypt: When is the gender
inequality. Paper presented at the workshop on Gender, Work, and Family in the Middle
East and North Africa.

Baker, D. P., Akiba, M., LeTendre, G. K., & Wiseman, A. W. (2001). Worldwide Shadow
Education: Outside-School Learning, Institutional Quality of Schooling, and Cross-
National Mathematics Achievement. Educational Evaluation and Policy Analysis, 23(1),
1-17.

Baker, D. P. & LeTendre, G. K. (2005). National differences, global similarities: World culture
and the future of schooling. Stanford CA: Stanford University Press.

Ban, S., Jung, S., & Yang, S. (2005). Analyzing the effect of private tutoring on academic
achievement. The first Korean Education and Employment Panel conference. Seoul, Korea.

Banerjee, A.V., Cole, S., Duflo, E. & Linden, L. (2007). Remedying Education: Evidence from
Two Randomized Experiments in India. Quarterly Journal of Economics, 122(3), 1235-
1264.

Baumol, W. J. (1996). Children of performing arts, the economic dilemma: The climbing costs
of health care and education. Journal of Cultural Economics, 20(3), 183-206.

Becker, G. (1962). Investment in Human Capital: A Theoretical Analysis. Journal of Political

Economy, 70(5), 9-49.

Bennell, P. (1998). Rates of Return to Education in Asia: A Review of the Evidence. Education
Economics, 6(2), 107-120.

Biswal, B. P. (1999). Private Tutoring and Public Corruption: A Cost-effective Education

System for Developing Countries. The Developing Economics, 37(2), 222-240.

Bloom, B. S. (1984). The 2 Sigma Problem: The Search for Methods of Group Instruction as
Effective as One-to-One Tutoring” Educational Researcher, 13(6), 4-16.

Bray, M. (1998). Financing Education in Developing Asia: Themes, Tensions, and Policies.
International Journal of Education Research, 29, 627-642.
 202

Bray, M. (1999). The shadow education system: Private tutoring and its implications for planners.
Fundamentals of Educational Planning, No.61. Paris: UNESCO International Institute for
Educational Planning (IIEP).

Bray, M. (2006). Private Supplementary Tutoring: Comparative Perspectives on Patterns and

Implications. Compare, 36(4), 515-530. Paris: UNESCO International Institute for
Educational Planning (IIEP).

Bray, M. & Kwok, P. (2003). Demand for Private Supplementary Tutoring: Conceptual
Considerations, and Socio-economic Patterns in Hong Kong. Economics of Education
Review, 22(6), 611-620.

Bray, M. & Lykins, C. (2012). Shadow Education: Private Supplementary Tutoring and Its
Implications for Policy Makers in Asia. Asian Development Bank.

Briggs, D. C. (2001). The Effect of Admission Test Preparation: Evidence from NELS: 88.
Chance, 14(1), 10-18.

Buchmann, C. (1999). The state and schooling in Kenya: Historical development and current
challenges. Africa Today, 46(1), 95-116. doi:10.1353/at.2003.0072
---. (2002). Getting Ahead in Kenya: Social Capital, Shadow Education, and Achievement.
Research in the Sociology of Education, 13, 133-159.

Burde, D. & Linden, L. (2010). The effect of village-based schools: evidence from a RCT in
Afghanistan. NYU Steinhardt.

Caillods, F., Gottelmann-Duret, G., Radi, M., & Hddigui, E. M. (1998). La formation
scientifique au Maroc: conditions et options de politique. Paris: UNCESCO/IIEP.

Carnoy, M. (1997). Recent Research on Market Returns to Education. International Journal of

Educational Research, 27(6), 483-490.

Cheo, R. & Quah, E. (2005). Mothers, Maids and Tutors: An Empirical Evaluation of Their
Effect on Children’s Academic Grades in Singapore. Education Economics, 13(3), 269-
285.

Choi, H. (2007). Does private tutoring help advancement to higher education? (Working Paper
2007-1). Seoul: Korea Labor Institute.

Choi, S. H. (2012, November 12). Catastrophe that expansion of free meal policy will bring. The
Dong-A Ilbo. Retrieved from http://news.donga.com/3/all/20121113/50806408/1

Choi, S., Kim Y., Yoo H., Kim H., & Lee, H. (2003). Analysis of Private Tutoring Status and It’s
Expense Scale. KEDI Research Report CR 2003-19. Seoul: Korean Educational
Development Institute. (in Korean)
 203

Chun, S., et al. (2003). A study of attitudes influencing the level of satisfaction with public
education. The collection of treatises of management education. 30.

Dang, H. (2007a). The Determinants and Impact of Private Tutoring Classes in Vietnam.
(Unpublished doctoral dissertation). University of Minnesota, Minnesota.
---. (2007b). The Determinants and Impact of Private Tutoring Classes in Vietnam.
Economics of Education Review, 26(6), 684-699.

Dang, H. (2008). Private tutoring in Vietnam: An investigation of its causes and impacts with
policy implications. Saarbrucken: VDM Verlag Dr. Muller.

Dang, H. & Rogers, F. H. (2008). How to Interpret the Growing Phenomenon of Private
Tutoring: Human Capital Deepening, Inequality Increasing, or Waste of Resources?
(World Bank Policy Research Working Paper 4530). Retrieved from
http://ssrn.com/abstract=1098628

Davies, S. (2004). School Choice by Default? Understanding the Demand for Private Tutoring in
Canada. American Journal of Education, 110, 233-255.

Dawson, W. (2010). Private Tutoring and Mass Schooling in East Asia: Reflections of Inequality
in Japan, South Korea, and Cambodia. Asia Pacific Education Review, 11(1), 14-24.

De Silva, W. A. (1994). Extra-school tutoring in the Asian context with special reference to Sri
Lanka. Maharagama: Department of Educational Research, National Institute of Education.

Dehejia, R. H. &Wahba, Sadek. (2002). Propensity Score-Matching Methods for

Nonexperimental Causal Studies. The Review of Economics and Statistics, 84(1), 151-161.

Dore, R. (1976). The diploma disease: Education, qualification and development. London:
George Allen and Unwin.

Dore, R. (1997). Reflections on the Diploma Disease Twenty Years Later. Assessment in
Education: Principles, Policy & Practice, 4(1), 189-20

Filmer, D. & Pritchett, L. (2001). Estimating Wealth Effects Without Expenditure Data—Or
Tears: An Application to Educational Enrollments in States of India. Demography, 38(1),
115-132.

Foondun, R. A. (2002). The Issue of Private Tuition: An Analysis of the Practice in Mauritius
and Selected South-East Asian Countries. International Review of Education, 48(6), 485-
515.

Glazerman, S. & Seifullah, A. (2010). An Evaluation of the Teacher Advancement Program

(TAP) in Chicago: Year Two Impact Report. Princeton: Mathematica Policy Research.
Retrieved from http://www.eric.ed.gov/PDFS/ED510712.pdf
 204

Glewwe, P. & Kremer, M. (2006). School, teachers, and education outcomes in developing
countries. In E. A. Hanushek, & F. Welch (Eds.), Handbook of the economics of education.
Amsterdam: North Holland.

Glewwe, P. & Jayachandran, S. (2006). Incentives to Teach Badly? After-School Tutoring in

Developing Countries. Mimeo, University of Minnesota.

Gordon, E. E., & Gordon, E. H. (1990). Centuries of tutoring— a history of alternative education
in America and Western Europe. MD: University Press of America.

Greene, W. H. (2003). Econometric Analysis. New Jersey: Prentice Hall, 7th Edition.

Ha, T. T., & Harpham, T. (2005). Primary education in Vietnam: Extra classes and outcomes.
International Education Journal, 6(5), 626–634.

Han, J. (2004). National private tutoring. Seoul, Korea: Hakjisa.

Han, D., Sung, B., & Gil, I. (2001). A study on the comparison between the effects of private
tutoring versus in-school education on academic achievement of high school students. The
Korean Journal of Sociology of Education, 11(1): 33-54.

Hanushek, E. (1986). The Economics of Schooling: Production and Efficiency in Public Schools.
Journal of Economic Literature, 24, 1141-1171.

Heckman, J. (1978). Dummy endogenous variables in a simultaneous equations system.

Econometrica, 46, 931-961.

Heckman, J. (1979). Sample selection bias as a specification error. Econometrica, 47,

153-161.

Hussein, M. G. A. (1987). Private Tutoring: a Hidden Educational Problem. Educational Studies

in Mathematics, 18(1), 91-96.

Hargreaves, E. (1997). The diploma disease in Egypt: learning, teaching and the monster of
secondary leaving certificate. Assessment in Education: Principles, Policy and Practice,
4(1), 161-176.

Hyun, J., Lee, J., & Lee, H. (2002). A Study on the Education fever of Korean parents. Seoul:
Korean Educational Development Institutes.

Jacob, B. A. & Lefgren, L. (2004). Remedial Education And Student Achievement: A

Regression-Discontinuity Analysis. Review of Economics and Statistics, 86(1), 226-244.

Jelani, J. & Tan, A. K. G. (2012). Determinants of Participation and Expenditure Patterns of

Private Tuition Received by Primary School Students in Penang, Malaysia: An
Exploratory Study. Asia Pacific Journal of Education, 32(1), 19–35.
 205

Kang, C. (2007). Does money matter? the effect of private educational expenditures on academic
performance: Evidence from exogenous variation in birth order. (Working Paper No.
0704). National University of Singapore. Retrieved from
http://www.fas.nus.edu.sg/ecs/pub/wp/wp0704.pdf

Katsillis, J. and Rubinson, R. (1990). Cultural Capital, Student Achievement, and Educational
Reproduction: The Case of Greece. American Sociological Review, 55(2), 270-279.

Kim, C. (2011, April 19). Jaesusang is about 21%-24% of students who take the university
entrance exam each year. Korea JoongAng Daily. Retrieved from
http://article.joinsmsn.com/news/article/article.asp?total_id=5366627&cloc=olink|article|d
efault

Kim, M. (2003). Private institute education: Competition and anxiety of the South Korean
middle class. The International Conference on Education Research. Seoul, Korea.

Kim, H. (2004). Analyzing the structure of variables affecting on private tutoring expense. The
Journal of Educational Administration, 22(1), 27-45.

Kim, H. I. (2013, January, 11). Reduction in educational budgets for low-income students after
expanding free meal plan and childcare programs. The Chosun Ilbo. Retrieved from
http://news.chosun.com/site/data/html_dir/2013/01/11/2013011100183.html

Kim, J. (2007b). A game theoretical approach to private tutoring in South Korea. (Occasional
Paper No. 144). National Center for the Study of Privatization in Education. New York:
Teachers College.

Kim, J. (2007a). The determinants of demand for private tutoring in South Korea. (Occasional
Paper No. 143). National Center for the Study of Privatization in Education. New York:
Teachers College.

Kim, J. & Chang, J. (2010). Do Governmental Regulations for Cram Schools Decrease the
Number of Hours Students Spend on Private Tutoring? KEDI Journal of Educational
Policy, 7(1), 3–21.

Kim, K. & Park, D. (2012). Impacts of Urban Economic Factors on Private Tutoring Industry.
Asia Pacific Education Review, 13(2), 273–280.

Kim, M., Kang, Y., Park, S., Lee, H., & Hwang, Y. (2006). Scales of the entrance examination
industry and its development: A study of the entrance examination industry and college
admission policy. Seoul: Korean Educational Development Institute.

Kim, S. & Lee, J. (2001). Demand for Education and Developmental State: Private Tutoring in
South Korea. Social Science Research Network Electronic Paper Collection. Retrieved
from http://papers.ssrn.com/paper.taf?abstract_id=268284
 206

Kim, S. and Lee, J. (2010). Private Tutoring and Demand for Education in South Korea.
Economic Development and Cultural Change, 58(2), 259-296.

Kim, T. (2005). Shadow Education: School Quality and Demand for Tutoring in Korea. Kyoto
University 21COE Discussion Paper No. 055.

Kim, T., Lee, J., & Lee, Y. (2003). Empirical Analysis of High School Equalization Policy on
the Effect of Student Achievement. CEPRI Research Report. Seoul: CEPRI at the KDI
School of Public Policy and Management.

Kim, K. O., Kim, D., Suh, Y. J., & Rhee, C. Y. (2003). The Change of College Entrance
Examination Policy: Who is admitted into Seoul National University? Korean Social
Science Review, 25(3), 3-187.

Kim, Y., Kang, S., Ryu, H., & Namgung, J. (2003). An analytical study on the qualitative level
and actual condition of school education. Seoul: Korean Educational Development
Institute.

Kim, Y. B. & Kim, M. S. (2002). Analysis of Cramming Institutions’ Education Status. (KEDI
Research Report RR 2002-1). Seoul: Korean Educational Development Institute.

Kong, E., Han, Y., Jang, S. M., & Huh, S. Y. (2001). A study of the reality of private tutoring
expense. Seoul: Korean Educational Development Institutes.

Korea National Statistics Office. (2008). 2007 Survey Report on Private Tutoring Costs and
Scale in Korea. (in Korean)

Korea National Statistics Office. http://kostat.go.kr/

Kramer, W. & Werner, P. (1998). Parental and paid tutoring. Germany: Deutscher-Institus
Verlag.

Kulpoo, D. (1998). The quality of education: some policy suggestions based on a survey of
schools in Mauritius. SACMEQ Policy Research: Report No. 1, Paris: UNESCO/IIEP.

Kwak, B. (2004). Struggle against Private Lessons in Korean Education Context. Paper
presented at the 28th annual conference of the Pacific Circle Consortium. Hong Kong
Institute of Education, 21-23 April.

Lavy, V. & Schlosser, A. (2005). Targeted Remedial Education for Underperforming Teenagers:
Costs and Benefits. Journal of Labor Economics, 23(4), 839-874.

Lee, C. J. (2005). Perspective: Korean Education Fever and Private Tutoring. KEDI Journal of
Educational Policy, 29(1). Seoul: KEDI.
 207

Lee, C., Park, H., & Lee, H. (2009). Shadow Education Systems. In Gary Sykes, Barbara
Schneider, and David N. Plank (Eds.), Handbook of Education Policy Research. New
York: Routledge for the American Educational Research Association, pp.901–919.

Lee, H. (2001). Educational effect of private tutoring. Paper presented at the Korea Education
Forum-3rd Educational Problem Seminar. Seoul: Korean Educational Development
Institute.

Lee, J. (2007). Two Worlds of Private Tutoring: The Prevalence and Causes of After-school
Mathematics Tutoring in Korea and the United States. Teachers College Record, 109(5),
1207-1234.

Lee, J., Kim, Y., & Yoon, C. (2004). The Effects of Pre-class Tutoring on Student Achievement,
Challenges and Implications for Public Education in Korea. KEDI Journal of Educational
Policy, 1(1), 25-42.

Lee, J., & Hong, S. (2001). Public Education vs. Private Education: Choice and Equality of
Korean Education. Retrieved from
http://cepri.kdischool.ac.kr/kor/data/research/2001_1.hwp

Lee, K. (2003). The best of intentions: Meritocratic selection to higher education and the
development of shadow education in Korea. (Doctoral dissertation). Pennsylvania State
University.

Lee, S. C. (2007). The Timing of Signaling: To Study in High School or in College?

International Economic Review, 48(3), 785-807.

Lee, S. (2006). Prestige- Oriented View of College Entrance and Shadow Education in South
Korea: Factors Influencing Parent Expenditures on Private Tutoring. (Doctoral
dissertation). Pennsylvania State University.

Levin, J. (2001). For whom the reductions count: A quantile regression analysis of class size and
peer effects on scholastic achievement. Empirical Economics, 26, 221-246.

Lipscomb, S. (2007). Secondary School Extracurricular Involvement and Academic

Achievement: A Fixed-Effects Approach. Economics of Education Review, 26(4), 463-
472.

Lucas, R. E. (1988). On the mechanics of economic development. Journal of Monetary

Economics, 22(1), 3-42.

Mankiw, N. Gregory, D. R. & Weil, D. N. (1992). A Contribution to the Empirics of Economic

Growth. The Quarterly Journal of Economics, 107(2), 407-437.

McKinsey & Company. (2007). How the world’s best-performing school systems come out on
top. Retrieved from
 208

http://www.mckinsey.com/locations/UK_Ireland/~/media/Reports/UKI/Education_report.
ashx

Ministry of Education. (1989). Palliative measures to ban the private tutoring. Seoul.

Ministry of Education, Culture, Sports, Science and Technology-Japan (MEXT). (1994). Survey
on extra-school education. Tokyo.

Mischo, C. & Haag, L. (2002). Expansion and Effectiveness of Private Tutoring. European
Journal of Psychology of Education, 17(3), 263-273.

Mitchell, R. (1968). Pupil, parent and school: A Hong Kong study. Taipei: Orient Cultural
Service.

Murnane, R. J. & Willett, J. B. (2010). Methods matter: Improving causal inference in

educational and social science research. Oxford University Press, USA.

Nam, K. (2007). Time-series analysis of the scale of private supplementary tutoring. Economic
study of educational finance, 16(1), 57–79. (in Korean)

Nanayakkara, G. L. S. & Ranaweera, M. (1994). Impact of private tuition and the educational
challeges of the 21st century. Economic Review [The People’s Bank, Colombo], 20(2,3),
11-17.

Ono, H. (2007). Does Examination Hell Pay Off? A Cost-Benefit Analysis of “Ronin’ and
College Education in Japan. Economics of Education Review, 26(3), 271-284.

Paik, I. (1999). Study on the determinants of demand for private tutoring. European Journal of
Psychology of Education, 17, 263-273.

Park, D., Park, C., & Kim, S. (2001). The effects of school and student level background
variables on math and science achievements in middle schools. The Korean Journal of
Educational Evaluation, 14(1), 127-149.

Pindyck, R. & Rubinfeld, D. (2000). Microeconomics. New York: Macmillan.

Polydorides, G. (1986). The determinants of educational achievement at the end of secondary

schooling: the case of Greece. Paper presented at the annual meeting of the American
Educational Research Association, San Francisco, 16-20 April.

Psachoaropolous, G. (1994). Returns to Investment in Education: A Global Update. World

Development, 22(9), 1325-1343.

Psacharopoulous, G., & Papakonstantinou, G. (2005). The real university cost in a “free” higher
education country. Economics of Education Review, 24(1), 103–108.
 209

Roesgaard, M. H. (2006). Japanese education and the cram school business: Functions,
challenges and perspectives of the Juku. Copenhagen: NIAS Press.

Rosenbaum, P. R. & Rubin, D. B. (1984). Reducing bias in observational studies using

subclassification on the propensity score. Journal of the American Statistical Association,
79(387), 516-524.

Russell, N. U. (1997). Lessons from Japanese Cram Schools. In Cummings, W.K. and Altbach,
P. (Eds). The Challenge of Eastern-Asian Education: Lessons for America. Albany: State
University of New York Press. Retrieved from
http://books.google.com/books?hl=en&lr=&id=qjS2jI27_6AC&oi=fnd&pg=PA153&dq=
Lessons+from+Japanese+Cram+Schools&ots=eQlCSQF8Wi&sig=ahDN3Euc6310z-
GFGDD513mt7G0#v=onepage&q=Lessons%20from%20Japanese%20Cram%20Schools
&f=false.

Rohlen, T. & LeTendre, G. (1996). Teaching and learning in Japan. New York: Cambridge
University Press.

Russell, N. U. (2002). The role of the private sector in determining national standards: How
juku undermine Japanese educational authority. In. G. Decoker (Ed.), National standards
and school reform in Japan and the United States. New York: Teachers College Press.

Sang, K. (2005). The Effects of Private Tutoring on High School Student's Achievement in
Mathematics. (Doctoral dissertation). Seoul, Korea: Seoul National University.

Sawada, T. & Kobayashi, S. (1986). An analysis of the Effect of Arithmetic and Mathematics
education at Juku. Translated with an afterword by Horvath, P.; Compendium 12, Tokyo:
National Institute for Educational Research.

Schultz, T. W. (1961). Investment in Human Capital. The American Economic Review, 51(1), 1-
17.

Silova, I. & Bray, M. (2006). The hidden marketplace: Private tutoring in former socialist
countries. In Silova, Iveta, Virginija Budiene and Mark Bray. (Eds). Education in a
Hidden Marketplace: Monitoring of Private Tutoring. New York: Open Society Institute.

Stevenson, D. L. & Baker, D. P. (1992). Shadow education and allocation in formal schooling:
transition to university in Japan. American Journal of sociology, 97(6), 1639-57.

Suryadarma, D., Suryahadi, A., Sumarto, S. & Rogers, F. H. (2006). Improving Student
Performance in Public Primary Schools in Developing Countries: Evidence from
Indonesia. Education Economics, 14(4), 401-429.

Sweeting, A. (1983). Hong Kong. In R. M. Thomas & T. W. Postlethwaite (Eds.), Schooling in

East Asia: Forces of change (pp. 272-297). New York: Pergamon.
 210

Tansel, A. & Bircan, F. (2005). Effect of private tutoring on university entrance examination
performance in Turkey. IZA Discussion Paper No. 1609. Retrieved from
http://ssrn.com/abstract=721925

Tansel, A. & Bircan, F. (2006). Demand for Education in Turkey: A Tobit Analysis of Private
Tutoring Expenditures. Economics of Education Review, 25(3), 303-313.

Tsukada, M. (1991). Yobiko life: a study of the legitimation process of social stratification in
Japan. Berkeley: Institute of East Asian Studies, University of California.

Turner, R. (1960). Sponsored and Contest Mobility in the School System. American Sociological
Review, 25(6), 855-867.

UNICEF. (1994). Situation Analysis of Women and Children in Mauritius. Port Louis: UNICEF.

Wijetunge, S. (1994). Some psychological aspects of the private tuition system. Economic
Review [The People’s Bank, Colombo], 20(2,3), 15-17.

Won, J. (2001). A Study on Social and Economic Effect of Private Education. (Doctoral
dissertation). Cohen University & Theological Seminary.

Yang, J. & Kim, K. (2003). Effects of middle school organization on academic achievement in
Korea: An HLM analysis of TIMSS_R. The Korean Journal of Sociology of Education,
13(2), 165-184.

Yoo, H., Chung, J, & Kim, S. (2009). Effect of Policy of Private Tutoring on Changes in Private
Tutoring Market in South Korea, Taiwan, and Japan. Korea Research Institute for
Vocational Education and Training.

Yun, C. (1997). A Study on the Reality of Private Tutoring. Seoul: Institute of Education Study
of the Seoul National University.
 211

APPENDIX
 212

Table A. Monthly salary schedule for K-12 teachers in 2011

Salary Salary
Salary Salary
Class Class
1 1,243,700 21 2,370,400
2 1,281,500 22 2,457,900
3 1,319,700 23 2,544,800
4 1,357,800 24 2,631,800
5 1,396,200 25 2,718,700
6 1,434,500 26 2,805,900
7 1,472,200 27 2,896,800
8 1,510,100 28 2,987,800
9 1,548,600 29 3,082,600
10 1,590,300 30 3,178,000
11 1,631,400 31 3,272,900
12 1,673,300 32 3,367,700
13 1,749,600 33 3,464,000
14 1,826,100 34 3,560,000
15 1,902,500 35 3,656,300
16 1,979,000 36 3,751,900
17 2,054,800 37 3,835,400
18 2,134,000 38 3,918,900
19 2,213,000 39 4,002,500
20 2,291,700 40 4,085,600
Unit: 1 Korean Won
Source: 2011 Teacher Salary schedule (K-12 school teachers), The Korean Teachers and Education Workers’
Union. http://www.eduhope.net/ (in Korean)
Note: 1,000 Korean won equal approximately 1 USD
 213

Table B.1. Treatment variables used in the section 5.1

Type of
Variable Name Description
variable
Measures for cumulative tutoring
PT1yr_ms dichotomous 1 year of any private tutoring in middle school
PT2yr_ms dichotomous 2 years of any private tutoring in middle school
PT3yr_ms dichotomous 3 years of any private tutoring in middle school
PT1yr_hs dichotomous 1 year of any private tutoring in high school
PT2yr_hs dichotomous 2 years of any private tutoring in high school
PT3yr_hs dichotomous 3 years of any private tutoring in high school
VERPT1yr_ms dichotomous 1 year of Verbal private tutoring in middle school
VERPT2yr_ms dichotomous 2 years of Verbal private tutoring in middle school
VERPT3yr_ms dichotomous 3 years of Verbal private tutoring in middle school
VERPT1yr_hs dichotomous 1 year of Verbal private tutoring in high school
VERPT2yr_hs dichotomous 2 years of Verbal private tutoring in high school
VERPT3yr_hs dichotomous 3 years of Verbal private tutoring in high school
ENGPT1yr_ms dichotomous 1 year of English private tutoring in middle school
ENGPT2yr_ms dichotomous 2 years of English private tutoring in middle school
ENGPT3yr_ms dichotomous 3 years of English private tutoring in middle school
ENGPT1yr_hs dichotomous 1 year of English private tutoring in high school
ENGPT2yr_hs dichotomous 2 years of English private tutoring in high school
ENGPT3yr_hs dichotomous 3 years of English private tutoring in high school
MATHPT1yr_ms dichotomous 1 year of Math private tutoring in middle school
MATHPT2yr_ms dichotomous 2 years of Math private tutoring in middle school
MATHPT3yr_ms dichotomous 3 years of Math private tutoring in middle school
MATHPT1yr_hs dichotomous 1 year of Math private tutoring in high school
MATHPT2yr_hs dichotomous 2 years of Math private tutoring in high school
MATHPT3yr_hs dichotomous 3 years of Math private tutoring in high school
Measure for Single-year tutoring
PT_G7 dichotomous Participation in any private tutoring in Grade 7
PT_G8 dichotomous Participation in any private tutoring in Grade 8
PT_G9 dichotomous Participation in any private tutoring in Grade 9
PT_G10 dichotomous Participation in any private tutoring in Grade 10
PT_G11 dichotomous Participation in any private tutoring in Grade 11
PT_G12 dichotomous Participation in any private tutoring in Grade 12
VERBALPT_G7 dichotomous Participation in Verbal private tutoring in grade 7
VERBALPT_G8 dichotomous Participation in Verbal private tutoring in grade 8
VERBALPT_G9 dichotomous Participation in Verbal private tutoring in grade 9
VERBALPT_G10 dichotomous Participation in Verbal private tutoring in grade 10
VERBALPT_G11 dichotomous Participation in Verbal private tutoring in grade 11
VERBALPT_G12 dichotomous Participation in Verbal private tutoring in grade 12
ENGPT_G7 dichotomous Participation in English private tutoring in grade 7
ENGPT_G8 dichotomous Participation in English private tutoring in grade 8
ENGPT_G9 dichotomous Participation in English private tutoring in grade 9
ENGPT_G10 dichotomous Participation in English private tutoring in grade 10
ENGPT_G11 dichotomous Participation in English private tutoring in grade 11
 214

ENGPT_G12 dichotomous Participation in English private tutoring in grade 12

MATHPT_G7 dichotomous Participation in Math private tutoring in grade 7
MATHPT_G8 dichotomous Participation in Math private tutoring in grade 8
MATHPT_G9 dichotomous Participation in Math private tutoring in grade 9
MATHPT_G10 dichotomous Participation in Math private tutoring in grade 10
MATHPT_G11 dichotomous Participation in Math private tutoring in grade 11
MATHPT_G12 dichotomous Participation in Math private tutoring in grade 12

Table B.2. Dependent variables used in the section 5.1

Variable Name Description Test Unit

G7VEM_Z Average score of Verbal, English and Math in grade 7 KEDI z-score
G8VEM_Z Average score of Verbal, English and Math in grade 8 KEDI z-score
G9VEM_Z Average score of Verbal, English and Math in grade 9 KEDI z-score
G7VER_Z Verbal score in grade 7 KEDI z-score
G8VER_Z Verbal score in grade 8 KEDI z-score
G9VER_Z Verbal score in grade 9 KEDI z-score
G7ENG_Z English score in grade 7 KEDI z-score
G8ENG_Z English score in grade 8 KEDI z-score
G9ENG_Z English score in grade 9 KEDI z-score
G7MAT_Z Math score in grade 7 KEDI z-score
G8MAT_Z Math score in grade 8 KEDI z-score
G9MAT_Z Math score in grade 9 KEDI z-score
CSAT_avg Average CSAT score CSAT decile rank
CSAT_V Verbal score CSAT decile rank
CSAT_E English score CSAT decile rank
CSAT_M Math score CSAT decile rank

Table B.3. Control variables used in Chapter IV

Variable Name Type of variable Description
GENDER dichotomous 1=female; 0=male
SES continuous socio-economic status
whether a student spends more than 10 hours per week
SELF-STUDY dichotomous
for independent study (1=yes; 0=no)
SELF-ESTEEM continuous level of self-esteem
PRESCORE continuous previous achievement score
SCHOOL TYPE dichotomous school type (1=private; 0=public)
SCHOOL TRACK dichotomous school track (1=general; 0=technical)
STU-TEA RATIO continuous student-teacher ratio
URBAN dichotomous urban residence (1=yes; 0=no)
SCHOOL CHOICE dichotomous residence with a school choice scheme (1=yes; 0=no)
 215

Table C.1. Effects of one year of private Table C.2. Effects of two years of private
tutoring on academic achievement in grade 9 tutoring on academic achievement in grade 9
(1) (2) (1) (2)
VARIABLES OLS IV VARIABLES OLS IV

1 year of PT in middle 2 years of PT in middle

-0.03 0.16 0.19*** 0.25
school school
(0.046) (0.131) (0.048) (0.132)
URBAN -0.01 -0.06 URBAN -0.00 -0.02
(0.072) (0.089) (0.068) (0.081)
GENDER (1=female) 0.25*** 0.26*** GENDER (1=female) 0.27*** 0.28***
(0.046) (0.055) (0.043) (0.051)
SES 0.17*** 0.18*** SES 0.08** 0.11**
(0.036) (0.048) (0.032) (0.042)
SCHOOL TYPE 0.07 0.07 SCHOOL TYPE 0.10 0.08
(0.059) (0.069) (0.055) (0.063)
STU-TEA RATIO -0.00 -0.00 STU-TEA RATIO 0.00 0.00
(0.005) (0.006) (0.004) (0.005)
SCHOOL CHOCIE 0.01 -0.04 SCHOOL CHOCIE 0.05 0.05
(0.069) (0.087) (0.067) (0.081)
PRESCORE 0.15*** 0.16*** PRESCORE 0.15*** 0.16***
(0.014) (0.016) (0.012) (0.015)
SELF-STUDY 0.15 0.12 SELF-STUDY 0.20* 0.15
(0.102) (0.121) (0.086) (0.100)
SELF-ESTEEM 0.10* 0.13** SELF-ESTEEM 0.11** 0.12*
(0.042) (0.050) (0.041) (0.050)
Constant -1.25*** -1.19*** Constant -1.43*** -1.47***
(0.187) (0.254) (0.179) (0.231)

Observations 778 521 Observations 1,084 763

R-squared 0.261 0.294 R-squared 0.240 0.257
Standard errors in parentheses Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05 *** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in Dependent variable: G9VEM_Z (average score in
grade 9) grade 9)
 216

Table C.3. Effects of three years of private Table C.4. Effects of one year of verbal
tutoring on academic achievement in grade 9 tutoring on verbal achievement in grade 9
(1) (2) (1) (2)
VARIABLES OLS IV VARIABLES OLS IV

3 years of PT in middle 1 year of Verbal PT in middle

0.38*** 0.54*** -0.17*** -0.07
school school
(0.045) (0.133) (0.043) (0.121)
URBAN -0.07 -0.10* URBAN -0.01 -0.08
(0.039) (0.045) (0.064) (0.074)
GENDER (1=female) 0.28*** 0.27*** GENDER (1=female) 0.49*** 0.49***
(0.027) (0.031) (0.043) (0.049)
SES 0.15*** 0.15*** SES 0.17*** 0.17***
(0.018) (0.023) (0.027) (0.032)
SCHOOL TYPE 0.08* 0.10* SCHOOL TYPE 0.18*** 0.20***
(0.034) (0.039) (0.052) (0.059)
STU-TEA RATIO 0.01* 0.01* STU-TEA RATIO 0.00 0.00
(0.004) (0.004) (0.005) (0.005)
SCHOOL CHOCIE -0.04 -0.04 SCHOOL CHOCIE -0.00 -0.06
(0.042) (0.049) (0.067) (0.078)
PRESCORE 0.16*** 0.15*** PRESCORE 0.17*** 0.17***
(0.008) (0.010) (0.012) (0.014)
SELF-STUDY 0.04 0.05 SELF-STUDY 0.17* 0.16
(0.039) (0.044) (0.071) (0.080)
SELF-ESTEEM 0.13*** 0.13*** SELF-ESTEEM 0.06 0.10*
(0.025) (0.029) (0.038) (0.044)
Constant -1.42*** -1.52*** Constant -1.79*** -1.66***
(0.124) (0.166) (0.188) (0.224)

Observations 2,854 2,252 Observations 1,768 1,304

R-squared 0.294 0.267 R-squared 0.249 0.259
Standard errors in parentheses Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05 *** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in grade Dependent variable: G9VER_Z (verbal score in grade 9)
9)
 217

Table C.5. Effects of two years of verbal Table C.6. Effects of three years of verbal
tutoring on verbal achievement in grade 9 tutoring on verbal achievement in grade 9
(1) (2) (1) (2)
VARIABLES OLS IV VARIABLES OLS IV

2 years of Verbal PT in 3 years of Verbal PT in

-0.05 -0.11 0.02 -0.12
middle school middle school
(0.042) (0.087) (0.040) (0.083)
URBAN -0.06 -0.06 URBAN -0.05 -0.10
(0.060) (0.069) (0.055) (0.062)
GENDER (1=female) 0.48*** 0.47*** GENDER (1=female) 0.47*** 0.44***
(0.041) (0.049) (0.038) (0.044)
SES 0.13*** 0.16*** SES 0.06* 0.09**
(0.027) (0.031) (0.025) (0.029)
SCHOOL TYPE 0.11* 0.13* SCHOOL TYPE 0.03 0.03
(0.051) (0.059) (0.047) (0.052)
STU-TEA RATIO 0.01 0.00 STU-TEA RATIO 0.01 0.01
(0.005) (0.005) (0.005) (0.005)
SCHOOL CHOCIE -0.05 -0.06 SCHOOL CHOCIE -0.06 -0.11
(0.062) (0.072) (0.058) (0.066)
PRESCORE 0.16*** 0.14*** PRESCORE 0.16*** 0.16***
(0.012) (0.014) (0.011) (0.013)
SELF-STUDY 0.11 0.11 SELF-STUDY -0.00 0.02
(0.064) (0.073) (0.058) (0.064)
SELF-ESTEEM 0.12*** 0.16*** SELF-ESTEEM 0.14*** 0.14***
(0.037) (0.043) (0.034) (0.039)
Constant -1.61*** -1.43*** Constant -1.66*** -1.42***
(0.181) (0.216) (0.172) (0.200)

Observations 1,900 1,394 Observations 2,207 1,714

R-squared 0.215 0.207 R-squared 0.203 0.194
Standard errors in parentheses Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05 *** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VER_Z (verbal score in grade Dependent variable: G9VER_Z (verbal score in grade
9) 9)
 218

Table C.7. Effects of one year of English Table C.8. Effects of two year of English
tutoring on English achievement in grade 9 tutoring on English achievement in grade 9
(1) (2) (1) (2)
VARIABLES OLS IV VARIABLES OLS IV

1 year of English PT in 2 years of English PT in

0.05 0.24 0.23*** 0.11
middle school middle school
(0.050) (0.156) (0.049) (0.293)
URBAN 0.12 0.08 URBAN 0.06 0.10
(0.075) (0.093) (0.065) (0.080)
GENDER (1=female) 0.34*** 0.34*** GENDER (1=female) 0.31*** 0.30***
(0.049) (0.059) (0.044) (0.053)
SES 0.22*** 0.27*** SES 0.19*** 0.23***
(0.037) (0.048) (0.032) (0.053)
SCHOOL TYPE 0.14* 0.07 SCHOOL TYPE 0.11* 0.08
(0.062) (0.072) (0.055) (0.065)
STU-TEA RATIO -0.00 -0.00 STU-TEA RATIO 0.00 0.01
(0.005) (0.006) (0.005) (0.006)
SCHOOL CHOCIE 0.13 0.07 SCHOOL CHOCIE 0.04 0.09
(0.074) (0.093) (0.067) (0.080)
PRESCORE 0.14*** 0.16*** PRESCORE 0.14*** 0.15***
(0.015) (0.017) (0.013) (0.018)
SELF-STUDY 0.03 -0.08 SELF-STUDY 0.22** 0.11
(0.103) (0.120) (0.082) (0.097)
SELF-ESTEEM 0.13** 0.16** SELF-ESTEEM 0.16*** 0.19***
(0.044) (0.051) (0.040) (0.051)
Constant -1.82*** -1.83*** Constant -1.79*** -1.83***
(0.215) (0.280) (0.192) (0.280)

Observations 993 687 Observations 1,377 993

R-squared 0.231 0.274 R-squared 0.274 0.270
Standard errors in parentheses Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05 *** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in Dependent variable: G9ENG_Z (English score in
grade 9) grade 9)
 219

Table C.9 Effects of three year of English Table C.10. Effects of one year of math
tutoring on English achievement in grade 9 tutoring on math achievement in grade 9
(1) (2) (1) (2)
VARIABLES OLS IV VARIABLES OLS IV

3 years of English PT in 1 year of Math PT in

0.41*** 1.06*** 0.05 0.21
middle school middle school
(0.046) (0.137) (0.048) (0.128)
URBAN -0.07 -0.10* URBAN -0.18* -0.19*
(0.043) (0.051) (0.074) (0.089)
GENDER (1=female) 0.35*** 0.37*** GENDER (1=female) -0.02 -0.05
(0.030) (0.035) (0.048) (0.056)
SES 0.21*** 0.17*** SES 0.16*** 0.14**
(0.020) (0.027) (0.035) (0.042)
SCHOOL TYPE 0.07 0.08 SCHOOL TYPE 0.07 0.10
(0.038) (0.044) (0.061) (0.070)
STU-TEA RATIO 0.00 0.00 STU-TEA RATIO 0.00 -0.00
(0.004) (0.005) (0.005) (0.006)
SCHOOL CHOCIE -0.11* -0.10 SCHOOL CHOCIE -0.02 -0.03
(0.047) (0.055) (0.072) (0.088)
PRESCORE 0.16*** 0.14*** PRESCORE 0.18*** 0.18***
(0.009) (0.011) (0.014) (0.016)
SELF-STUDY -0.00 -0.05 SELF-STUDY 0.17 0.20
(0.043) (0.049) (0.095) (0.111)
SELF-ESTEEM 0.25*** 0.21*** SELF-ESTEEM 0.07 0.12*
(0.027) (0.032) (0.044) (0.051)
Constant -1.58*** -2.11*** Constant -1.28*** -1.24***
(0.141) (0.186) (0.206) (0.261)

Observations 3,142 2,492 Observations 992 693

R-squared 0.310 0.252 R-squared 0.242 0.272
Standard errors in parentheses Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05 *** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in grade Dependent variable: G9MAT_Z (math score in
9) grade 9)
 220

Table C.11. Effects of two year of math Table C.12 Effects of three year of math
tutoring on math achievement in grade 9 tutoring on math achievement in grade 9
(1) (2) (1) (2)
VARIABLES OLS IV VARIABLES OLS IV

2 years of Math PT in 3 years of Math PT in

0.29*** 0.01 0.46*** 0.88***
middle school middle school
(0.049) (0.307) (0.048) (0.136)
URBAN -0.19** -0.24** URBAN -0.14** -0.17***
(0.067) (0.081) (0.044) (0.051)
GENDER (1=female) 0.00 -0.02 GENDER (1=female) 0.01 0.01
(0.044) (0.052) (0.031) (0.036)
SES 0.15*** 0.21*** SES 0.17*** 0.17***
(0.031) (0.051) (0.021) (0.026)
SCHOOL TYPE 0.15** 0.18** SCHOOL TYPE 0.15*** 0.14**
(0.055) (0.066) (0.039) (0.045)
STU-TEA RATIO 0.00 0.01 STU-TEA RATIO 0.01 0.00
(0.005) (0.006) (0.004) (0.005)
SCHOOL CHOCIE -0.05 -0.10 SCHOOL CHOCIE -0.06 -0.03
(0.067) (0.080) (0.048) (0.056)
PRESCORE 0.16*** 0.18*** PRESCORE 0.18*** 0.16***
(0.013) (0.017) (0.010) (0.011)
SELF-STUDY 0.08 0.09 SELF-STUDY 0.04 0.03
(0.080) (0.094) (0.044) (0.049)
SELF-ESTEEM 0.13** 0.15** SELF-ESTEEM 0.11*** 0.09**
(0.040) (0.051) (0.028) (0.032)
Constant -1.22*** -1.00*** Constant -1.34*** -1.58***
(0.193) (0.283) (0.146) (0.188)

Observations 1,423 1,031 Observations 3,198 2,531

R-squared 0.244 0.245 R-squared 0.254 0.215
Standard errors in parentheses Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05 *** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9MAT_Z (math score in Dependent variable: G9MAT_Z (math score in grade
grade 9) 9)
 221

Table C.13. Effects of private tutoring in grade 7 on academic achievement for 3 years
Year 0 Year 1 Year 2
(1) (2) (3) (4) (5) (6)
VARIABLES OLS IV OLS IV OLS IV

PT in G7 0.22*** 0.53*** 0.09** 0.33*** 0.05 0.22*

(0.022) (0.066) (0.026) (0.076) (0.033) (0.094)
URBANICITY 0.07** 0.04 0.01 -0.01 -0.05 -0.06
(0.025) (0.026) (0.028) (0.029) (0.034) (0.035)
GENDER (1=female) 0.25*** 0.26*** 0.29*** 0.29*** 0.28*** 0.28***
(0.017) (0.018) (0.019) (0.020) (0.023) (0.024)
SES 0.19*** 0.16*** 0.16*** 0.16*** 0.15*** 0.14***
(0.011) (0.013) (0.013) (0.014) (0.016) (0.017)
SCHOOL TYPE (1=private) 0.06** 0.04 0.06* 0.05 0.09** 0.07*
(0.021) (0.022) (0.024) (0.025) (0.029) (0.030)
STU-TEA RATIO 0.01*** 0.01*** 0.01*** 0.01*** 0.01* 0.01*
(0.002) (0.002) (0.002) (0.002) (0.003) (0.003)
SCHOOL CHOICE -0.07** -0.07** -0.09** -0.10*** -0.02 -0.02
(0.026) (0.028) (0.029) (0.031) (0.036) (0.037)
PRESCORE (G6) 0.20*** 0.19*** 0.17*** 0.16*** 0.16*** 0.16***
(0.005) (0.006) (0.006) (0.006) (0.007) (0.007)
STUDY HOUR 0.10*** 0.09*** 0.09** 0.08** 0.07 0.05
(1= >10hrs per week) (0.025) (0.026) (0.028) (0.029) (0.035) (0.036)
SELF ESTEEM 0.19*** 0.19*** 0.18*** 0.18*** 0.12*** 0.12***
(0.015) (0.016) (0.017) (0.018) (0.021) (0.022)
PT in G8 0.24*** 0.15*** 0.18*** 0.12**
(0.027) (0.040) (0.035) (0.045)
PT in G9 0.23*** 0.20***
(0.033) (0.038)
Constant -1.92*** -2.06*** -1.77*** -1.83*** -1.78*** -1.85***
(0.078) (0.090) (0.088) (0.096) (0.106) (0.115)

Observations 5,895 5,473 5,380 5,022 3,990 3,716

R-squared 0.444 0.419 0.364 0.351 0.315 0.307
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7VEM_Z (average score in grade 7) for (1) and (2); G8VEM_Z (average score in grade
8) for (3) and (4); G9VEM_Z (average score in grade 9) for (5) and (6)
 222

Table C.14. Effects of private tutoring in grade 8 on academic achievement for 2 years
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4)
VARIABLES OLS IV OLS IV

PT in G8 0.12*** 0.27*** 0.09** 0.13

(0.020) (0.066) (0.028) (0.097)
URBANICITY -0.04 -0.05* -0.09** -0.09**
(0.021) (0.022) (0.028) (0.029)
GENDER (1=female) 0.10*** 0.11*** 0.09*** 0.09***
(0.014) (0.015) (0.019) (0.020)
SES 0.04*** 0.04*** 0.05*** 0.06***
(0.010) (0.011) (0.013) (0.014)
SCHOOL TYPE (1=private) 0.03 0.03 0.06* 0.07**
(0.018) (0.019) (0.023) (0.025)
STU-TEA RATIO 0.00 0.00 -0.00 -0.00
(0.002) (0.002) (0.002) (0.002)
SCHOOL CHOICE -0.05* -0.05* 0.01 0.01
(0.022) (0.023) (0.029) (0.031)
PRESCORE (G7) 0.73*** 0.72*** 0.68*** 0.68***
(0.010) (0.011) (0.013) (0.014)
STUDY HOUR 0.03 0.01 0.01 -0.01
(1= >10hrs per week) (0.021) (0.023) (0.029) (0.030)
SELF ESTEEM 0.07*** 0.06*** 0.03 0.03
(0.013) (0.013) (0.017) (0.018)
PT in G7 -0.03 -0.08** -0.06* -0.08*
(0.020) (0.031) (0.027) (0.039)
PT in G9 0.16*** 0.15***
(0.027) (0.040)
Constant -0.18** -0.25*** -0.25** -0.24**
(0.062) (0.073) (0.082) (0.092)

Observations 5,501 4,953 4,077 3,657

R-squared 0.633 0.623 0.540 0.530
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8VEM_Z (average score in grade 8) for (1) and (2); G9VEM_Z (average score in
grade 9) for (3) and (4)
 223

Table C.15. Effects of private tutoring in grade 9 on academic achievement in grade 9

Year 0 (Grade 9)
(1) (2)
VARIABLES OLS IV

PT in G9 0.16*** 0.19*
(0.023) (0.090)
URBANICITY -0.07** -0.07**
(0.025) (0.028)
GENDER (1=female) 0.06** 0.06**
(0.018) (0.019)
SES 0.05*** 0.06***
(0.012) (0.014)
SCHOOL TYPE (1=private) 0.06** 0.07**
(0.021) (0.023)
STU-TEA RATIO -0.00 0.00
(0.002) (0.002)
SCHOOL CHOICE 0.03 0.02
(0.027) (0.029)
PRESCORE (G8) 0.70*** 0.69***
(0.012) (0.013)
STUDY HOURS (1= >10hrs per week) 0.00 0.00
(0.027) (0.029)
SELF ESTEEM 0.05*** 0.05**
(0.015) (0.017)
PT in G8 0.05* 0.01
(0.024) (0.046)
Constant -0.23** -0.25**
(0.074) (0.086)

Observations 4,378 3,761

R-squared 0.585 0.579
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in grade 9)
 224

Table D.1. Effects of verbal tutoring in grade 7 on verbal achievement for 3 years
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES OLS IV OLS IV OLS IV

VERBAL PT in G7 -0.01 0.01 -0.05 -0.08 -0.02 -0.12

(0.022) (0.056) (0.025) (0.071) (0.031) (0.087)
URBANICITY 0.05 0.05 0.02 0.01 -0.04 -0.04
(0.031) (0.032) (0.034) (0.035) (0.041) (0.042)
GENDER (1=female) 0.45*** 0.45*** 0.47*** 0.47*** 0.47*** 0.47***
(0.021) (0.022) (0.023) (0.024) (0.028) (0.029)
SES 0.14*** 0.13*** 0.10*** 0.11*** 0.11*** 0.12***
(0.014) (0.015) (0.015) (0.016) (0.018) (0.019)
SCHOOL TYPE 0.04 0.02 0.05 0.04 0.08* 0.08*
(1=private) (0.026) (0.027) (0.029) (0.030) (0.035) (0.036)
STU/TEA RATIO 0.02*** 0.01*** 0.02*** 0.02*** 0.01** 0.01**
(0.003) (0.003) (0.003) (0.003) (0.003) (0.004)
SCHOOL CHOICE -0.05 -0.05 -0.08* -0.09* -0.03 -0.03
(0.033) (0.034) (0.036) (0.037) (0.043) (0.045)
PRESCORE (G6) 0.20*** 0.20*** 0.17*** 0.17*** 0.15*** 0.15***
(0.006) (0.007) (0.007) (0.007) (0.008) (0.009)
STUDY HOUR 0.06* 0.07* 0.05 0.04 0.08 0.07
(1= >10hrs per week) (0.032) (0.032) (0.035) (0.036) (0.043) (0.044)
SELF ESTEEM 0.19*** 0.19*** 0.17*** 0.17*** 0.10*** 0.11***
(0.019) (0.020) (0.021) (0.021) (0.025) (0.026)
VERBAL PT in G8 0.09*** 0.09** 0.06 0.07
(0.025) (0.035) (0.033) (0.043)
VERBAL PT in G9 0.01 0.03
(0.032) (0.035)
Constant -2.12*** -2.07*** -1.96*** -1.89*** -1.77*** -1.72***
(0.096) (0.102) (0.105) (0.110) (0.126) (0.133)

Observations 5,855 5,435 5,357 5,004 4,073 3,790

R-squared 0.316 0.307 0.257 0.250 0.198 0.192
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7VER_Z (verbal score in grade 7) (1) and (2); G8VER_Z (verbal score in grade
8) for (3) and (4); G9VER_Z (verbal score in grade 9) for (5) and (6)
 225

Table D.2. Effects of verbal tutoring in grade 8 on verbal achievement for 2 years
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4)
VARIABLES OLS IV OLS IV

VERBAL PT in G8 0.04* -0.00 0.00 -0.11

(0.021) (0.053) (0.028) (0.075)
URBANICITY -0.04 -0.04 -0.08* -0.08*
(0.028) (0.029) (0.034) (0.036)
GENDER (1=female) 0.29*** 0.30*** 0.26*** 0.25***
(0.020) (0.021) (0.024) (0.026)
SES -0.04** -0.04** -0.01 0.00
(0.013) (0.014) (0.016) (0.017)
SCHOOL TYPE (1=private) 0.02 0.02 0.05 0.06
(0.024) (0.025) (0.029) (0.031)
STU-TEA RATIO 0.01** 0.01** 0.00 0.00
(0.002) (0.002) (0.003) (0.003)
SCHOOL CHOICE -0.03 -0.04 0.02 0.00
(0.030) (0.031) (0.036) (0.038)
PRESCORE (G7) 0.73*** 0.72*** 0.71*** 0.71***
(0.013) (0.014) (0.015) (0.016)
STUDY HOURS -0.02 -0.03 0.00 -0.01
(1= >10hrs per week) (0.029) (0.030) (0.036) (0.038)
SELF ESTEEM 0.05** 0.04* 0.01 0.01
(0.017) (0.018) (0.021) (0.022)
VERBAL PT in G7 -0.06** -0.04 -0.02 0.02
(0.021) (0.029) (0.026) (0.035)
VERBAL PT in G9 -0.00 0.02
(0.027) (0.036)

Constant -0.49*** -0.49*** -0.38*** -0.32**

(0.082) (0.089) (0.100) (0.107)

Observations 5,478 4,935 4,133 3,701

R-squared 0.469 0.462 0.425 0.420
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8VER_Z (verbal score in grade 8) for (1) and (2); G9VER_Z (verbal
score in grade 9) for (3) and (4)
 226

Table D.3. Effects of verbal tutoring in grade 9 on verbal achievement in grade 9

Year 0 (Grade 9)
(1) (2)
VARIABLES OLS IV

VERBAL PT in G9 -0.00 -0.04

(0.027) (0.093)
URBANICITY -0.08* -0.09*
(0.034) (0.037)
GENDER (1=female) 0.26*** 0.27***
(0.024) (0.027)
SES -0.01 -0.01
(0.016) (0.017)
SCHOOL TYPE (1=private) 0.05 0.07*
(0.029) (0.031)
STU-TEA RATIO 0.00 0.00
(0.003) (0.003)
SCHOOL CHOICE 0.02 0.00
(0.036) (0.039)
PRESCORE (G8) 0.71*** 0.69***
(0.015) (0.017)
STUDY HOURS (1= >10hrs per week) 0.00 0.01
(0.036) (0.039)
SELF-ESTEEM 0.01 0.01
(0.021) (0.023)
VERBAL PT in G7 -0.02 -0.01
(0.026) (0.031)
VERBAL PT in G8 0.00 -0.01
(0.028) (0.043)
Constant -0.38*** -0.40***
(0.100) (0.112)

Observations 4,133 3,556

R-squared 0.425 0.414
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VER_Z (verbal score in grade 9)
 227

Table D.4. Effects of English tutoring in grade 7 on English achievement for 3 years
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES OLS IV OLS IV OLS IV

ENGLISH PT in G7 0.25*** 0.72*** 0.12*** 0.61*** 0.07* 0.64***

(0.024) (0.071) (0.029) (0.086) (0.033) (0.098)
URBANICITY 0.10*** 0.05 0.05 0.01 0.00 -0.03
(0.029) (0.032) (0.032) (0.034) (0.036) (0.039)
GENDER (1=female) 0.28*** 0.30*** 0.36*** 0.37*** 0.34*** 0.35***
(0.020) (0.022) (0.022) (0.024) (0.025) (0.027)
SES 0.28*** 0.25*** 0.25*** 0.23*** 0.21*** 0.19***
(0.013) (0.016) (0.015) (0.017) (0.017) (0.019)
SCHOOL TYPE (1=private) 0.08** 0.06* 0.06* 0.04 0.09** 0.06
(0.025) (0.027) (0.028) (0.029) (0.031) (0.033)
STU-TEA RATIO 0.01*** 0.01*** 0.01*** 0.01** 0.01 0.01*
(0.002) (0.003) (0.003) (0.003) (0.003) (0.003)
SCHOOL CHOICE -0.11*** -0.12*** -0.08* -0.10** -0.05 -0.05
(0.031) (0.033) (0.034) (0.036) (0.038) (0.041)
PRESCORE (G6) 0.18*** 0.17*** 0.15*** 0.14*** 0.16*** 0.15***
(0.006) (0.007) (0.007) (0.007) (0.008) (0.008)
STUDY HOURS 0.13*** 0.10** 0.11** 0.09* 0.03 -0.01
(1= >10hrs per week) (0.030) (0.032) (0.033) (0.035) (0.037) (0.040)
SELF ESTEEM 0.23*** 0.21*** 0.23*** 0.22*** 0.21*** 0.20***
(0.018) (0.020) (0.020) (0.021) (0.022) (0.024)
ENGLISH PT in G8 0.25*** 0.05 0.18*** -0.02
(0.029) (0.045) (0.036) (0.050)
ENGLISH PT in G9 0.23*** 0.13**
(0.035) (0.041)
Constant -1.82*** -2.02*** -1.77*** -1.90*** -1.81*** -1.99***
(0.092) (0.105) (0.102) (0.112) (0.114) (0.126)

Observations 5,910 5,487 5,380 5,023 4,546 4,248

R-squared 0.403 0.362 0.338 0.301 0.314 0.269
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7ENG_Z (English score in grade 7) for (1) and (2); G8ENG_Z (English score in grade 8)
for (3) and (4); G9ENG_Z (English score in grade 9) for (5) and (6)
 228

Table D.5. Effects of English tutoring in grade 8 on English achievement for 2 years
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4)
VARIABLES OLS IV OLS IV

ENGLISH PT in G8 0.14*** 0.31*** 0.09** 0.18

(0.024) (0.073) (0.030) (0.102)
URBANICITY 0.00 -0.01 -0.04 -0.04
(0.026) (0.027) (0.030) (0.032)
GENDER (1=female) 0.17*** 0.17*** 0.15*** 0.16***
(0.018) (0.019) (0.021) (0.022)
SES 0.11*** 0.12*** 0.10*** 0.10***
(0.012) (0.014) (0.015) (0.016)
SCHOOL TYPE (1=private) 0.03 0.04 0.06* 0.07**
(0.022) (0.023) (0.026) (0.027)
STU-TEA RATIO -0.00 -0.00 -0.00 -0.00
(0.002) (0.002) (0.003) (0.003)
SCHOOL CHOICE -0.03 -0.04 -0.01 -0.00
(0.027) (0.029) (0.032) (0.034)
PRESCORE (G7) 0.74*** 0.73*** 0.70*** 0.70***
(0.012) (0.014) (0.014) (0.016)
STUDY HOURS 0.05 0.03 -0.02 -0.05
(1= >10hrs per week) (0.027) (0.028) (0.031) (0.033)
SELF ESTEEM 0.09*** 0.09*** 0.11*** 0.10***
(0.016) (0.017) (0.018) (0.019)
ENGLISH PT in G7 0.00 -0.06 -0.03 -0.07
(0.023) (0.035) (0.028) (0.040)
ENGLISH PT in G9 0.17*** 0.14**
(0.029) (0.045)
Constant -0.30*** -0.37*** -0.29** -0.33**
(0.077) (0.090) (0.091) (0.102)

Observations 5,502 4,955 4,643 4,188

R-squared 0.561 0.555 0.504 0.499
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables:G8ENG_Z (English score in grade 8) for (1) and (2); G9ENG_Z (English
score in grade 9) for (3) and (4)
 229

Table D.6. Effects of English tutoring in grade 9 on English achievement in grade 9

Year 0 (Grade 9)
(1) (2)
VARIABLES OLS IV

ENGLISH PT in G9 0.16*** -0.15

(0.028) (0.272)
URBANICITY -0.02 -0.00
(0.029) (0.032)
GENDER (1=female) 0.12*** 0.12***
(0.021) (0.024)
SES 0.11*** 0.12***
(0.014) (0.018)
SCHOOL TYPE (1=private) 0.04 0.05
(0.025) (0.027)
STU/TEA RATIO -0.00 -0.00
(0.002) (0.003)
SCHOOL CHOICE 0.01 0.02
(0.031) (0.035)
PRESCORE (G8) 0.73*** 0.74***
(0.014) (0.018)
STUDY HOURS -0.03 -0.04
(1= >10hrs per week) (0.030) (0.033)
SELF-ESTEEM 0.13*** 0.13***
(0.018) (0.020)
ENGLISH PT in G7 0.01 0.07
(0.027) (0.050)
ENGLISH PT in G8 0.06* 0.14
(0.029) (0.100)
Constant -0.27** -0.22
(0.088) (0.119)

Observations 4,609 4,002

R-squared 0.536 0.524
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in grade 9)
 230

Table D.7. Effects of math tutoring in grade 7 on math achievement for 3 years
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES OLS IV OLS IV OLS IV

MATH PT in G7 0.26*** 0.53*** 0.09** 0.36*** 0.04 0.34**

(0.024) (0.081) (0.030) (0.085) (0.033) (0.132)
URBANICITY 0.05 0.03 -0.02 -0.06 -0.14*** -0.17***
(0.031) (0.032) (0.034) (0.035) (0.037) (0.039)
GENDER (1=female) 0.02 0.02 0.04 0.05* 0.01 0.01
(0.021) (0.022) (0.023) (0.024) (0.026) (0.027)
SES 0.16*** 0.14*** 0.16*** 0.16*** 0.16*** 0.15***
(0.014) (0.016) (0.016) (0.017) (0.017) (0.019)
SCHOOL TYPE (1=private) 0.05 0.04 0.07* 0.06 0.15*** 0.13***
(0.026) (0.028) (0.029) (0.030) (0.032) (0.033)
STU/TEA RATIO 0.01*** 0.01*** 0.01*** 0.01** 0.01* 0.01
(0.003) (0.003) (0.003) (0.003) (0.003) (0.003)
SCHOOL CHOICE -0.06* -0.07* -0.11** -0.12*** -0.04 -0.05
(0.032) (0.034) (0.036) (0.037) (0.039) (0.041)
PRESCORE (G6) 0.22*** 0.21*** 0.18*** 0.18*** 0.17*** 0.17***
(0.006) (0.007) (0.007) (0.007) (0.008) (0.008)
STUDY HOURS 0.11*** 0.09** 0.11*** 0.11** 0.06 0.06
(1= >10hrs per week) (0.031) (0.033) (0.035) (0.036) (0.038) (0.039)
SELF ESTEEM 0.17*** 0.16*** 0.15*** 0.15*** 0.10*** 0.10***
(0.019) (0.020) (0.021) (0.022) (0.023) (0.024)
MATH PT in G8 0.28*** 0.16*** 0.20*** 0.09
(0.031) (0.046) (0.037) (0.058)
MATH PT in G9 0.30*** 0.25***
(0.036) (0.044)
Constant -1.72*** -1.83*** -1.44*** -1.52*** -1.41*** -1.50***
(0.096) (0.108) (0.106) (0.114) (0.117) (0.128)

Observations 5,859 5,445 5,391 5,035 4,641 4,336

R-squared 0.348 0.330 0.286 0.275 0.259 0.244
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7MAT_Z (math score in grade 7) for (1) and (2); G8MAT_Z (math score in grade 8) for
(3) and (4); G9MAT_Z (math score in grade 9) for (5) and (6)
 231

Table D.8. Effects of math tutoring in grade 8 on math achievement for 2 years
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4)
VARIABLES OLS IV OLS IV

MATH PT in G8 0.17*** 0.45*** 0.11*** 0.43***

(0.026) (0.083) (0.032) (0.111)
URBANICITY -0.08** -0.09** -0.18*** -0.17***
(0.029) (0.030) (0.032) (0.034)
GENDER (1=female) -0.15*** -0.14*** -0.18*** -0.18***
(0.020) (0.021) (0.023) (0.024)
SES 0.04** 0.03* 0.05*** 0.05**
(0.014) (0.015) (0.015) (0.017)
SCHOOL TYPE (1=private) 0.04 0.04 0.12*** 0.13***
(0.025) (0.026) (0.028) (0.030)
STU-TEA RATIO 0.00 -0.00 -0.00 -0.00
(0.002) (0.003) (0.003) (0.003)
SCHOOL CHOICE -0.08* -0.07* -0.01 0.01
(0.030) (0.033) (0.034) (0.037)
PRESCORE (G7) 0.73*** 0.71*** 0.67*** 0.67***
(0.014) (0.015) (0.015) (0.017)
STUDY HOURS 0.05 0.03 -0.00 -0.03
(1= >10hrs per week) (0.029) (0.031) (0.033) (0.035)
SELF ESTEEM 0.06** 0.05* 0.02 0.01
(0.017) (0.019) (0.020) (0.021)
MATH PT in G7 0.00 -0.09* -0.03 -0.13**
(0.025) (0.040) (0.029) (0.044)
MATH PT in G9 0.25*** 0.14**
(0.032) (0.048)
Constant 0.23** 0.09 0.14 0.04
(0.085) (0.099) (0.097) (0.108)

Observations 5,513 4,964 4,739 4,275

R-squared 0.471 0.455 0.419 0.405
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8MAT_Z (math score in grade 8) for (1) and (2); G9MAT_Z (math
score in grade 9) for (3) and (4)
 232

Table D.9. Effects of math tutoring in grade 9 on math achievement in grade 9

Year 0 (Grade 9)
(1) (2)
VARIABLES OLS IV

MATH PT in G9 0.24*** -0.09

(0.031) (0.287)
URBANICITY -0.15*** -0.15***
(0.032) (0.035)
GENDER (1=female) -0.21*** -0.22***
(0.022) (0.026)
SES 0.07*** 0.09***
(0.015) (0.020)
SCHOOL TYPE (1=private) 0.11*** 0.13***
(0.027) (0.030)
STU-TEA RATIO -0.00 0.00
(0.003) (0.004)
SCHOOL CHOICE 0.01 0.01
(0.034) (0.037)
PRESCORE (G8) 0.68*** 0.70***
(0.015) (0.019)
STUDY HOURS -0.00 -0.00
(1= >10hrs per week) (0.033) (0.036)
SELF ESTEEM 0.04* 0.05*
(0.019) (0.023)
MATH PT in G7 0.01 0.05
(0.029) (0.053)
MATH PT in G8 0.08** 0.19
(0.031) (0.103)
Constant 0.11 0.22
(0.095) (0.136)

Observations 4,697 4,073

R-squared 0.443 0.427
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9MAT_Z (math score in grade 9)
 233

Table E.1. Long-term cumulative effects of private tutoring in middle school on the CSAT
achievement
1 year of PT in MS 2 years of PT in MS 3 years of PT in MS

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.03 0.34 0.13 0.36 0.04 0.48

(0.180) (0.482) (0.166) (0.410) (0.134) (0.456)
URBANICITY -0.63* -0.91** -0.07 -0.12 -0.03 -0.15
(0.272) (0.308) (0.206) (0.232) (0.092) (0.103)
GENDER (1=female) -0.05 -0.24 0.13 0.09 0.05 0.03
(0.178) (0.207) (0.138) (0.162) (0.064) (0.072)
SES 0.02 -0.17 0.05 0.06 0.31*** 0.25***
(0.153) (0.191) (0.104) (0.117) (0.043) (0.051)
SCHOOL TYPE (1=private) -0.16 -0.13 0.31 0.17 0.05 -0.04
(0.257) (0.287) (0.175) (0.197) (0.081) (0.091)
STU-TEA RATIO 0.03 0.03 -0.00 -0.00 0.00 0.01
(0.018) (0.020) (0.014) (0.016) (0.008) (0.009)
SCHOOL CHOICE -0.34 -0.53 -0.14 -0.21 -0.04 -0.12
(0.280) (0.315) (0.213) (0.238) (0.100) (0.110)
PRESCORE (G6) 0.27*** 0.32*** 0.24*** 0.28*** 0.26*** 0.26***
(0.054) (0.061) (0.039) (0.046) (0.020) (0.022)
STUDY HOURS (1= >10hrs /week) -0.24 -0.26 0.17 0.11 0.15 0.15
(0.332) (0.381) (0.231) (0.259) (0.087) (0.095)
SELF ESTEEM 0.23 0.07 0.24 0.11 0.31*** 0.28***
(0.155) (0.177) (0.130) (0.152) (0.057) (0.065)
PT in G10 0.10 0.16 0.03 0.09 0.35*** 0.32**
(0.186) (0.221) (0.155) (0.191) (0.097) (0.122)
PT in G11 0.35 0.23 0.21 -0.01 0.33*** 0.32**
(0.206) (0.242) (0.158) (0.189) (0.087) (0.101)
PT in G12 0.54* 0.60* 0.36* 0.45* 0.29*** 0.34***
(0.253) (0.298) (0.175) (0.199) (0.072) (0.079)
Constant 3.46*** 3.68*** 3.40*** 3.36*** 3.24*** 2.95***
(0.801) (0.951) (0.598) (0.712) (0.316) (0.455)

Observations 246 181 421 325 1,452 1,190

R-squared 0.275 0.294 0.181 0.180 0.318 0.310
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)
 234

Table E.2. Long-term single-year effects of private tutoring in middle school on the CSAT
PT in G7 PT in G8 PT in G9

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.14 0.42 0.09 -0.18 -0.07 0.31

(0.087) (0.274) (0.086) (0.293) (0.081) (0.385)
URBANICITY -0.00 -0.03 -0.03 -0.05 -0.04 -0.09
(0.084) (0.087) (0.075) (0.078) (0.075) (0.081)
GENDER (1=female) 0.07 0.06 -0.11* -0.13* -0.17** -0.17**
(0.058) (0.060) (0.053) (0.055) (0.052) (0.059)
SES 0.28*** 0.25*** 0.18*** 0.14*** 0.18*** 0.17***
(0.040) (0.043) (0.036) (0.038) (0.036) (0.039)
SCHOOL TYPE (1=private) 0.13 0.10 0.13* 0.10 0.13 0.10
(0.074) (0.077) (0.067) (0.069) (0.065) (0.070)
STU-TEA RATIO 0.00 0.01 -0.01 -0.01 -0.00 -0.00
(0.007) (0.007) (0.007) (0.007) (0.006) (0.007)
SCHOOL CHOICE -0.04 -0.05 0.04 0.03 0.02 -0.04
(0.090) (0.093) (0.081) (0.084) (0.080) (0.087)
PRESCORE (G6) 0.25*** 0.25*** 0.95*** 0.97*** 0.84*** 0.82***
(0.018) (0.019) (0.039) (0.041) (0.036) (0.038)
STUDY HOURS (1= >10hrs /week) 0.13 0.14 0.08 0.09 0.09 0.10
(0.082) (0.084) (0.074) (0.077) (0.073) (0.078)
SELF ESTEEM 0.28*** 0.27*** 0.21*** 0.20*** 0.26*** 0.22***
(0.052) (0.054) (0.046) (0.048) (0.045) (0.051)
PT in G7 - - -0.06 0.02 - -
- - (0.080) (0.110) - -
PT in G8 0.17 0.14 - - 0.02 -0.17
(0.094) (0.120) - - (0.081) (0.178)
PT in G9 -0.11 -0.22* -0.12 -0.04 - -
(0.093) (0.104) (0.085) (0.132) - -
PT in G10 0.19* 0.19* 0.08 0.09 0.16* 0.16
(0.080) (0.084) (0.073) (0.077) (0.071) (0.085)
PT in G11 0.31*** 0.30*** 0.30*** 0.30*** 0.31*** 0.27**
(0.075) (0.078) (0.068) (0.070) (0.066) (0.083)
PT in G12 0.32*** 0.32*** 0.19** 0.21*** 0.22*** 0.25***
(0.065) (0.068) (0.060) (0.062) (0.059) (0.063)
Constant 3.14*** 3.01*** 5.24*** 5.36*** 5.25*** 5.22***
(0.276) (0.306) (0.236) (0.254) (0.229) (0.270)

Observations 1,905 1,800 1,934 1,770 2,021 1,785

R-squared 0.296 0.287 0.405 0.403 0.390 0.378
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)
 235

Table E.3. Long-term cumulative effects of verbal tutoring in middle school on the CSAT verbal
achievement
1 year of Verbal PT 2 years of Verbal PT 3 years of Verbal PT

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest -0.17 0.52* -0.21* -0.28 -0.26** -0.47*

(0.106) (0.256) (0.103) (0.222) (0.098) (0.192)
URBANICITY -0.22 -0.28 -0.14 -0.14 -0.13 -0.14
(0.158) (0.180) (0.144) (0.160) (0.127) (0.140)
GENDER (1=female) 0.20 0.21 0.40*** 0.39*** 0.18* 0.13
(0.107) (0.121) (0.100) (0.115) (0.088) (0.099)
SES 0.29*** 0.27*** 0.26*** 0.24** 0.19** 0.19**
(0.068) (0.079) (0.067) (0.075) (0.060) (0.066)
SCHOOL TYPE (1=private) 0.01 -0.17 0.03 0.01 -0.02 -0.15
(0.133) (0.154) (0.128) (0.144) (0.114) (0.126)
STU-TEA RATIO 0.02 0.01 0.01 0.01 0.02* 0.02*
(0.012) (0.014) (0.011) (0.013) (0.011) (0.012)
SCHOOL CHOICE 0.00 -0.08 -0.14 -0.14 -0.08 -0.04
(0.174) (0.198) (0.154) (0.171) (0.139) (0.151)
PRESCORE (G6) 0.31*** 0.31*** 0.31*** 0.31*** 0.29*** 0.30***
(0.032) (0.037) (0.029) (0.033) (0.027) (0.030)
STUDY HOURS (1= >10hrs /week) -0.06 -0.03 -0.06 -0.02 -0.02 0.05
(0.158) (0.176) (0.141) (0.156) (0.127) (0.137)
SELF ESTEEM 0.33*** 0.28** 0.31*** 0.25* 0.43*** 0.38***
(0.094) (0.108) (0.089) (0.101) (0.080) (0.089)
VERBAL PT in G10 -0.10 -0.18 -0.01 0.07 0.10 0.15
(0.115) (0.136) (0.106) (0.126) (0.095) (0.113)
VERBAL PT in G11 0.13 0.05 -0.02 -0.11 0.04 0.02
(0.143) (0.163) (0.129) (0.145) (0.113) (0.124)
VERBAL PT in G12 0.57** 0.53** 0.40* 0.52** 0.32* 0.36*
(0.180) (0.200) (0.167) (0.181) (0.141) (0.150)
Constant 3.34*** 3.35*** 3.38*** 3.38*** 3.52*** 3.55***
(0.492) (0.568) (0.449) (0.514) (0.416) (0.467)

Observations 791 635 925 722 1,114 909

R-squared 0.254 0.204 0.222 0.214 0.212 0.210
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)
 236

Table E.4. Long-term single-year effects of verbal tutoring in middle school on the CSAT verbal
achievement
VERBAL PT in G7 VERBAL PT in G8 VERBAL PT in G9

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest -0.04 -0.08 -0.04 -0.33 -0.07 0.07

(0.071) (0.188) (0.071) (0.176) (0.072) (0.290)
URBANICITY -0.01 -0.01 -0.05 -0.04 -0.03 -0.05
(0.094) (0.097) (0.088) (0.090) (0.089) (0.097)
GENDER (1=female) 0.25*** 0.24*** 0.07 0.06 0.03 0.04
(0.065) (0.067) (0.061) (0.063) (0.063) (0.072)
SES 0.23*** 0.21*** 0.11** 0.07 0.12** 0.12*
(0.044) (0.045) (0.041) (0.043) (0.043) (0.046)
SCHOOL TYPE (1=private) 0.09 0.08 0.10 0.07 0.11 0.08
(0.081) (0.084) (0.076) (0.078) (0.078) (0.083)
STU-TEA RATIO 0.01 0.01 -0.00 -0.00 0.00 0.00
(0.008) (0.008) (0.007) (0.008) (0.008) (0.008)
SCHOOL CHOICE 0.05 0.05 0.14 0.15 0.13 0.11
(0.101) (0.104) (0.094) (0.097) (0.096) (0.103)
PRESCORE (G6) 0.29*** 0.29*** 0.95*** 0.96*** 0.83*** 0.82***
(0.020) (0.021) (0.043) (0.045) (0.042) (0.045)
STUDY HOURS (1= >10hrs /week) 0.03 0.04 -0.02 -0.00 -0.01 0.02
(0.092) (0.094) (0.086) (0.089) (0.088) (0.093)
SELF ESTEEM 0.30*** 0.30*** 0.24*** 0.25*** 0.33*** 0.29***
(0.058) (0.060) (0.053) (0.055) (0.054) (0.059)
VERBAL PT in G7 - - -0.12 -0.03 -0.09 -0.12
- - (0.067) (0.084) (0.069) (0.078)
VERBAL PT in G8 -0.02 -0.03 - - -0.06 -0.10
(0.076) (0.095) - - (0.073) (0.133)
VERBAL PT in G9 -0.13 -0.12 -0.08 0.03 - -
(0.075) (0.081) (0.071) (0.092) - -
VERBAL PT in G10 0.03 0.03 -0.02 -0.02 -0.02 -0.02
(0.068) (0.071) (0.064) (0.066) (0.065) (0.081)
VERBAL PT in G11 0.08 0.07 0.10 0.08 0.10 0.09
(0.081) (0.083) (0.075) (0.077) (0.077) (0.082)
VERBAL PT in G12 0.33*** 0.35*** 0.20* 0.23* 0.20* 0.20*
(0.101) (0.102) (0.095) (0.096) (0.097) (0.103)
Constant 3.32*** 3.32*** 5.39*** 5.47*** 5.40*** 5.36***
(0.301) (0.313) (0.263) (0.272) (0.268) (0.298)

Observations 2,032 1,921 2,064 1,889 2,044 1,812

R-squared 0.198 0.194 0.282 0.278 0.252 0.243
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)
 237

Table E.5. Long-term cumulative effects of English tutoring in middle school on the CSAT
English achievement
1 year of English PT 2 years of English PT 3 years of English PT

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest -0.18 -0.02 -0.13 0.54 0.09 1.23**

(0.180) (0.521) (0.163) (0.408) (0.132) (0.444)
URBANICITY -0.44 -0.62* 0.03 -0.19 0.02 -0.06
(0.273) (0.305) (0.189) (0.220) (0.103) (0.115)
GENDER (1=female) 0.38* 0.17 0.14 0.18 0.21** 0.20*
(0.178) (0.203) (0.132) (0.154) (0.071) (0.081)
SES 0.15 0.03 0.30** 0.23* 0.41*** 0.35***
(0.139) (0.167) (0.096) (0.112) (0.049) (0.058)
SCHOOL TYPE (1=private) 0.38 0.21 0.16 0.03 0.07 0.01
(0.234) (0.262) (0.173) (0.200) (0.090) (0.102)
STU/TEA RATIO 0.02 0.02 -0.00 -0.01 0.01 0.01
(0.018) (0.021) (0.015) (0.017) (0.009) (0.010)
SCHOOL CHOICE -0.29 -0.29 -0.34 -0.49* -0.11 -0.10
(0.284) (0.317) (0.201) (0.233) (0.111) (0.123)
PRESCORE (G6) 0.25*** 0.30*** 0.27*** 0.28*** 0.28*** 0.28***
(0.052) (0.060) (0.039) (0.047) (0.023) (0.026)
STUDY HOURS (1= >10hrs /week) -0.13 0.02 0.19 0.05 0.12 0.12
(0.323) (0.378) (0.220) (0.261) (0.097) (0.108)
SELF-ESTEEM 0.47** 0.32 0.30* 0.22 0.40*** 0.31***
(0.156) (0.179) (0.119) (0.143) (0.064) (0.074)
ENGLISH PT in G10 0.04 0.09 0.05 0.01 0.13 0.02
(0.188) (0.221) (0.142) (0.174) (0.087) (0.107)
ENGLISH PT in G11 0.44* 0.42 0.42** 0.34 0.14 0.11
(0.216) (0.266) (0.154) (0.184) (0.082) (0.096)
ENGLISH PT in G12 0.35 0.15 0.15 0.12 0.23** 0.21*
(0.289) (0.338) (0.183) (0.213) (0.082) (0.092)
Constant 3.16*** 3.14** 3.69*** 3.58*** 3.05*** 2.22***
(0.819) (0.959) (0.587) (0.713) (0.347) (0.486)

Observations 343 260 595 462 1,678 1,375

R-squared 0.203 0.203 0.204 0.173 0.257 0.225
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)
 238

Table E.6. Long-term single-year effects of English tutoring in middle school on the CSAT
English achievement
ENGLISH PT in G7 ENGLISH PT in G8 ENGLISH PT in G9

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.10 0.65** 0.13 0.39 -0.05 0.51

(0.085) (0.251) (0.090) (0.317) (0.092) (0.389)
URBANICITY 0.03 -0.01 -0.02 -0.05 0.01 -0.02
(0.090) (0.094) (0.083) (0.086) (0.083) (0.089)
GENDER (1=female) 0.24*** 0.23*** 0.03 0.04 -0.05 -0.05
(0.062) (0.065) (0.058) (0.060) (0.059) (0.064)
SES 0.39*** 0.37*** 0.25*** 0.20*** 0.26*** 0.25***
(0.043) (0.046) (0.040) (0.043) (0.041) (0.044)
SCHOOL TYPE (1=private) 0.15 0.12 0.14* 0.13 0.15* 0.14
(0.079) (0.083) (0.073) (0.076) (0.073) (0.079)
STU/TEA RATIO 0.01 0.01 -0.00 -0.00 -0.00 -0.00
(0.008) (0.008) (0.007) (0.008) (0.007) (0.008)
SCHOOL CHOICE -0.16 -0.17 -0.06 -0.08 -0.06 -0.10
(0.097) (0.101) (0.089) (0.093) (0.090) (0.096)
PRESCORE (G6) 0.26*** 0.26*** 0.99*** 1.03*** 0.92*** 0.91***
(0.020) (0.020) (0.042) (0.045) (0.040) (0.043)
STUDY HOURS (1= >10hrs /week) 0.11 0.12 0.06 0.07 0.08 0.09
(0.088) (0.091) (0.081) (0.084) (0.081) (0.087)
SELF-ESTEEM 0.37*** 0.35*** 0.27*** 0.24*** 0.33*** 0.30***
(0.056) (0.058) (0.051) (0.053) (0.051) (0.056)
ENGLISH PT in G7 - - -0.08 -0.09 -0.03 -0.09
- - (0.079) (0.107) (0.079) (0.092)
ENGLISH PT in G8 0.22* 0.05 - - 0.12 -0.10
(0.097) (0.124) - - (0.090) (0.167)
ENGLISH PT in G9 -0.04 -0.14 -0.00 -0.11 - -
(0.099) (0.109) (0.091) (0.148) - -
ENGLISH PT in G10 0.08 0.06 0.04 0.03 0.07 0.06
(0.073) (0.077) (0.067) (0.071) (0.068) (0.081)
ENGLISH PT in G11 0.24*** 0.24** 0.21** 0.18* 0.22*** 0.21**
(0.071) (0.074) (0.065) (0.069) (0.066) (0.073)
ENGLISH PT in G12 0.20** 0.20** 0.15* 0.17* 0.17* 0.16*
(0.075) (0.077) (0.069) (0.072) (0.069) (0.074)
Constant 3.08*** 2.88*** 5.13*** 5.05*** 5.18*** 5.07***
(0.299) (0.323) (0.258) (0.287) (0.259) (0.303)

Observations 2,302 2,179 2,339 2,148 2,319 2,062

R-squared 0.247 0.229 0.342 0.342 0.337 0.326
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)
 239

Table E.7. Long-term cumulative effects of math tutoring in middle school on the CSAT math
achievement
1 year of Math PT 2 years of Math PT 3 years of Math PT

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest -0.04 -0.34 -0.17 0.13 -0.15 0.56

(0.192) (0.475) (0.172) (0.442) (0.143) (0.461)
URBANICITY -0.23 -0.47 -0.02 -0.17 -0.07 -0.19
(0.291) (0.322) (0.209) (0.238) (0.109) (0.120)
GENDER (1=female) -0.52** -0.74*** -0.25 -0.36* -0.20** -0.21*
(0.192) (0.216) (0.142) (0.161) (0.076) (0.086)
SES 0.19 0.14 0.13 0.14 0.23*** 0.19**
(0.140) (0.151) (0.101) (0.112) (0.052) (0.060)
SCHOOL TYPE (1=private) -0.11 -0.09 0.15 0.05 -0.05 -0.13
(0.265) (0.290) (0.182) (0.208) (0.097) (0.110)
STU-TEA RATIO 0.02 0.02 0.02 0.02 0.01 0.01
(0.020) (0.022) (0.015) (0.018) (0.010) (0.012)
SCHOOL CHOICE 0.07 -0.17 0.00 -0.07 -0.15 -0.26*
(0.306) (0.335) (0.214) (0.243) (0.119) (0.130)
PRESCORE (G6) 0.33*** 0.35*** 0.25*** 0.25*** 0.24*** 0.24***
(0.058) (0.063) (0.043) (0.049) (0.024) (0.027)
STUDY HOURS (1= >10hrs /week) 0.06 0.15 0.11 0.03 0.21* 0.18
(0.316) (0.359) (0.226) (0.259) (0.103) (0.113)
SELF ESTEEM 0.03 -0.12 0.33** 0.23 0.23*** 0.19*
(0.173) (0.194) (0.127) (0.147) (0.069) (0.077)
MATH PT in G10 0.13 0.33 0.30 0.35 0.42*** 0.38**
(0.206) (0.239) (0.159) (0.197) (0.107) (0.133)
MATH PT in G11 0.66** 0.69** 0.54** 0.41* 0.59*** 0.49***
(0.230) (0.255) (0.166) (0.194) (0.099) (0.116)
MATH PT in G12 0.39 0.06 0.84*** 0.89*** 0.55*** 0.55***
(0.291) (0.325) (0.189) (0.214) (0.087) (0.096)
Constant 3.06*** 3.89*** 3.03*** 3.30*** 3.56*** 3.31***
(0.858) (0.974) (0.622) (0.730) (0.374) (0.491)

Observations 344 266 594 466 1,668 1,369

R-squared 0.235 0.258 0.243 0.230 0.256 0.242
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)
 240

Table E.8. Long-term single-year effects of math tutoring in middle school on the CSAT math
achievement
MATH PT in G7 MATH PT in G8 MATH PT in G9

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.11 -0.13 -0.07 0.01 -0.14 0.44

(0.091) (0.450) (0.098) (0.362) (0.102) (0.437)
URBANICITY -0.05 -0.07 -0.09 -0.12 -0.06 -0.12
(0.097) (0.102) (0.091) (0.094) (0.092) (0.098)
GENDER (1=female) -0.21** -0.25*** -0.39*** -0.41*** -0.46*** -0.47***
(0.067) (0.071) (0.065) (0.067) (0.066) (0.072)
SES 0.21*** 0.21*** 0.10* 0.08 0.10* 0.09
(0.046) (0.051) (0.045) (0.047) (0.045) (0.048)
SCHOOL TYPE (1=private) 0.05 0.07 0.09 0.06 0.08 0.05
(0.085) (0.090) (0.081) (0.084) (0.081) (0.086)
STU-TEA RATIO 0.00 0.01 -0.00 -0.00 -0.00 -0.00
(0.008) (0.009) (0.008) (0.008) (0.008) (0.009)
SCHOOL CHOICE -0.09 -0.13 -0.04 -0.06 -0.04 -0.14
(0.104) (0.108) (0.099) (0.102) (0.099) (0.106)
PRESCORE (G6) 0.23*** 0.22*** 0.87*** 0.90*** 0.81*** 0.79***
(0.021) (0.022) (0.048) (0.050) (0.045) (0.048)
STUDY HOURS (1= >10hrs /week) 0.14 0.16 0.06 0.03 0.08 0.09
(0.094) (0.097) (0.089) (0.093) (0.090) (0.097)
SELF ESTEEM 0.24*** 0.23*** 0.15** 0.14* 0.21*** 0.16**
(0.060) (0.062) (0.056) (0.058) (0.056) (0.062)
MATH PT in G7 - - -0.03 -0.06 0.01 -0.09
(0.087) (0.126) (0.087) (0.103)
MATH PT in G8 -0.04 0.04 - - -0.10 -0.25
(0.104) (0.176) - - (0.099) (0.175)
MATH PT in G9 -0.16 -0.17 -0.12 -0.19 - -
(0.107) (0.131) (0.102) (0.159) - -
MATH PT in G10 0.30*** 0.33*** 0.18* 0.18* 0.20* 0.17
(0.088) (0.092) (0.084) (0.089) (0.085) (0.105)
MATH PT in G11 0.62*** 0.65*** 0.63*** 0.61*** 0.65*** 0.61***
(0.083) (0.088) (0.079) (0.083) (0.079) (0.090)
MATH PT in G12 0.55*** 0.58*** 0.38*** 0.38*** 0.38*** 0.37***
(0.078) (0.082) (0.075) (0.078) (0.075) (0.080)
Constant 3.57*** 3.74*** 5.44*** 5.54*** 5.45*** 5.47***
(0.320) (0.354) (0.284) (0.309) (0.285) (0.332)

Observations 2,294 2,172 2,330 2,142 2,310 2,055

R-squared 0.238 0.236 0.298 0.297 0.296 0.284
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)
 241

Table F.1. Cumulative effects of private tutoring in high school on the CSAT achievement
1 year of PT in HS 2 years of PT in HS 3 years of PT in HS

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.06 0.36 0.29** 0.51 0.59*** 0.51*

(0.104) (0.376) (0.101) (0.304) (0.105) (0.251)
URBANICITY -0.12 -0.27 -0.22 -0.20 -0.08 -0.11
(0.159) (0.189) (0.119) (0.137) (0.108) (0.120)
GENDER (1=female) -0.15 -0.23 -0.12 -0.08 -0.04 -0.05
(0.100) (0.122) (0.083) (0.097) (0.072) (0.082)
SES 0.11 0.09 0.10 0.08 0.13** 0.13*
(0.080) (0.100) (0.054) (0.064) (0.048) (0.058)
SCHOOL TYPE (1=private) 0.03 0.09 0.05 0.07 0.21** 0.23**
(0.105) (0.124) (0.085) (0.098) (0.074) (0.082)
SCHOOL TRACK 0.34* 0.46* 0.54*** 0.66*** 0.52*** 0.75***
(0.141) (0.181) (0.134) (0.170) (0.132) (0.165)
STU-TEA RATIO -0.01 -0.02 -0.03 -0.05* -0.05** -0.04*
(0.020) (0.024) (0.018) (0.021) (0.016) (0.019)
SCHOOL CHOICE -0.06 -0.23 -0.29* -0.26 -0.11 -0.17
(0.155) (0.185) (0.123) (0.148) (0.115) (0.129)
PRESCORE (G9) 0.82*** 0.80*** 0.74*** 0.70*** 0.73*** 0.68***
(0.071) (0.085) (0.056) (0.065) (0.051) (0.057)
STUDY HOURS (1= >10hrs /week) 0.46*** 0.35* 0.40*** 0.42*** 0.26** 0.16
(0.121) (0.144) (0.093) (0.109) (0.081) (0.090)
SELF ESTEEM 0.28*** 0.34** 0.31*** 0.35*** 0.43*** 0.48***
(0.084) (0.104) (0.067) (0.080) (0.061) (0.068)
PT in middle school -0.06 -0.10 0.02 0.03 -0.03 0.03
(0.047) (0.065) (0.047) (0.071) (0.050) (0.071)
Constant 5.09*** 5.56*** 5.51*** 5.44*** 5.44*** 5.24***
(0.449) (0.561) (0.386) (0.459) (0.356) (0.407)

Observations 639 465 826 627 985 782

R-squared 0.327 0.319 0.407 0.402 0.453 0.442
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)
 242

Table F.2. Single-year effects of private tutoring in high school on the CSAT achievement
PT in G10 PT in G11 PT in G12

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.07 0.20 0.25*** 0.36 0.26*** 0.16

(0.073) (0.335) (0.067) (0.283) (0.059) (0.145)
URBANICITY -0.06 -0.09 -0.06 -0.06 -0.06 -0.06
(0.076) (0.082) (0.076) (0.078) (0.076) (0.080)
GENDER (1=female) -0.07 -0.10 -0.07 -0.08 -0.07 -0.06
(0.052) (0.060) (0.052) (0.054) (0.052) (0.055)
SES 0.13*** 0.13*** 0.13*** 0.12*** 0.13*** 0.13***
(0.035) (0.039) (0.035) (0.037) (0.035) (0.037)
SCHOOL TYPE (1=private) 0.17** 0.17** 0.17** 0.18*** 0.17** 0.15**
(0.053) (0.058) (0.053) (0.055) (0.053) (0.056)
SCHOOL TRACK 0.56*** 0.59*** 0.56*** 0.56*** 0.56*** 0.61***
(0.094) (0.121) (0.094) (0.099) (0.094) (0.104)
STU-TEA RATIO -0.03** -0.04** -0.03** -0.03** -0.03** -0.03**
(0.012) (0.013) (0.012) (0.012) (0.012) (0.012)
SCHOOL CHOICE -0.09 -0.14 -0.09 -0.10 -0.09 -0.12
(0.080) (0.088) (0.080) (0.082) (0.080) (0.084)
PRESCORE (G9) 0.74*** 0.72*** 0.74*** 0.72*** 0.74*** 0.73***
(0.036) (0.039) (0.036) (0.038) (0.036) (0.037)
STUDY HOURS (1= >10hrs /week) 0.34*** 0.33*** 0.34*** 0.31*** 0.34*** 0.34***
(0.057) (0.062) (0.057) (0.059) (0.057) (0.060)
SELF ESTEEM 0.35*** 0.34*** 0.35*** 0.37*** 0.35*** 0.38***
(0.044) (0.048) (0.044) (0.045) (0.044) (0.046)
PT in middle schoo -0.01 -0.02 -0.01 -0.03 -0.01 -0.01
(0.033) (0.044) (0.033) (0.041) (0.033) (0.036)
PT in grade 10 0.07 0.04 0.07 0.06
(0.073) (0.107) (0.073) (0.078)
PT in grade 11 0.25*** 0.22* 0.25*** 0.31***
(0.067) (0.102) (0.067) (0.086)
PT in grade 12 0.26*** 0.26*** 0.26*** 0.23*
(0.059) (0.067) (0.059) (0.095)
Constant 5.14*** 5.28*** 5.14*** 5.16*** 5.14*** 5.16***
(0.257) (0.293) (0.257) (0.267) (0.257) (0.272)

Observations 1,940 1,680 1,940 1,827 1,940 1,752

R-squared 0.421 0.407 0.421 0.413 0.421 0.422
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)
 243

Table F.3. Cumulative effects of verbal tutoring in high school on the CSAT verbal achievement
2 years of Verbal PT in 3 years of Verbal PT in
1 year of Verbal PT in HS
HS HS
(1) (2) (3) (4) (5) (6)
VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.01 0.54 0.24** 0.34 0.23 0.23

(0.071) (0.464) (0.093) (0.240) (0.143) (0.258)
URBANICITY -0.03 -0.09 -0.17 -0.12 -0.16 -0.09
(0.101) (0.124) (0.121) (0.139) (0.136) (0.155)
GENDER (1=female) 0.12 0.03 0.14 0.11 0.17 0.15
(0.072) (0.090) (0.081) (0.093) (0.092) (0.107)
SES 0.08 0.04 0.10 0.09 0.07 0.08
(0.047) (0.060) (0.053) (0.066) (0.061) (0.073)
SCHOOL TYPE (1=private) 0.08 0.03 -0.01 -0.04 -0.08 -0.11
(0.072) (0.084) (0.083) (0.095) (0.093) (0.107)
SCHOOL TRACK 0.58*** 0.68*** 0.64*** 0.81*** 0.66*** 0.82***
(0.119) (0.151) (0.138) (0.168) (0.149) (0.187)
STU-TEA RATIO -0.02 -0.02 -0.01 -0.02 -0.03 -0.02
(0.016) (0.020) (0.018) (0.021) (0.020) (0.024)
SCHOOL CHOICE -0.01 0.03 -0.08 -0.04 -0.19 -0.13
(0.105) (0.126) (0.123) (0.137) (0.137) (0.154)
PRESCORE (G9) 0.77*** 0.76*** 0.66*** 0.66*** 0.72*** 0.66***
(0.048) (0.057) (0.055) (0.063) (0.063) (0.074)
STUDY HOURS (1= >10hrs /week) 0.11 0.11 0.20* 0.20 0.14 0.17
(0.078) (0.092) (0.089) (0.102) (0.103) (0.118)
SELF ESTEEM 0.31*** 0.34*** 0.42*** 0.45*** 0.41*** 0.43***
(0.060) (0.071) (0.065) (0.075) (0.074) (0.087)
Verbal PT in middle school -0.08* -0.13* -0.10** -0.12* -0.07 -0.06
(0.031) (0.052) (0.036) (0.046) (0.039) (0.047)
Constant 5.26*** 5.07*** 5.27*** 5.15*** 5.56*** 5.22***
(0.333) (0.448) (0.381) (0.444) (0.422) (0.498)

Observations 1,596 1,227 1,208 938 988 760

R-squared 0.256 0.230 0.273 0.275 0.290 0.270
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)
 244

Table F.4. Single-year effects of verbal tutoring in high school on the CSAT verbal achievement
Verbal PT in G10 Verbal PT in G11 Verbal PT in G12

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.00 0.11 0.13 0.05 0.21* 0.09

(0.064) (0.291) (0.075) (0.279) (0.094) (0.260)
URBANICITY -0.04 -0.05 -0.04 -0.02 -0.04 -0.02
(0.090) (0.102) (0.090) (0.094) (0.090) (0.094)
GENDER (1=female) 0.10 0.04 0.10 0.09 0.10 0.09
(0.062) (0.069) (0.062) (0.064) (0.062) (0.065)
SES 0.08* 0.08 0.08* 0.08 0.08* 0.07
(0.040) (0.047) (0.040) (0.043) (0.040) (0.043)
SCHOOL TYPE (1=private) 0.09 0.07 0.09 0.11 0.09 0.08
(0.063) (0.068) (0.063) (0.065) (0.063) (0.066)
SCHOOL TRACK 0.66*** 0.72*** 0.66*** 0.68*** 0.66*** 0.71***
(0.110) (0.126) (0.110) (0.114) (0.110) (0.121)
STU-TEA RATIO -0.03* -0.03 -0.03* -0.03 -0.03* -0.03*
(0.014) (0.015) (0.014) (0.014) (0.014) (0.014)
SCHOOL CHOICE 0.02 0.01 0.02 0.03 0.02 0.02
(0.094) (0.102) (0.094) (0.098) (0.094) (0.099)
PRESCORE (G9) 0.72*** 0.72*** 0.72*** 0.71*** 0.72*** 0.72***
(0.041) (0.045) (0.041) (0.043) (0.041) (0.044)
STUDY HOURS (1= >10hrs /week) 0.17* 0.18* 0.17* 0.16* 0.17* 0.18*
(0.068) (0.073) (0.068) (0.069) (0.068) (0.072)
SELF ESTEEM 0.34*** 0.32*** 0.34*** 0.35*** 0.34*** 0.38***
(0.051) (0.057) (0.051) (0.053) (0.051) (0.054)
Verbal PT in middle school -0.08** -0.11** -0.08** -0.07* -0.08** -0.08**
(0.028) (0.038) (0.028) (0.029) (0.028) (0.030)
Verbal PT in grade 10 0.00 0.00 0.00 -0.00
(0.064) (0.080) (0.064) (0.068)
Verbal PT in grade 11 0.13 0.11 0.13 0.15
(0.075) (0.102) (0.075) (0.094)
Verbal PT in grade 12 0.21* 0.23* 0.21* 0.23
(0.094) (0.102) (0.094) (0.130)
Constant 5.27*** 5.31*** 5.27*** 5.25*** 5.27*** 5.31***
(0.296) (0.329) (0.296) (0.313) (0.296) (0.311)

Observations 2,070 1,790 2,070 1,952 2,070 1,870

R-squared 0.269 0.261 0.269 0.259 0.269 0.272
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)
 245

Table F.5. Cumulative effects of English tutoring in high school on the CSAT English
achievement
1 year of English PT 2 years of English PT 3 years of English PT

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest -0.12 0.41 0.22* 0.18 0.36*** 0.29

(0.090) (0.349) (0.090) (0.221) (0.100) (0.164)
URBANICITY 0.04 -0.03 -0.21 -0.20 -0.16 -0.16
(0.128) (0.150) (0.118) (0.134) (0.133) (0.146)
GENDER (1=female) 0.16 0.14 0.02 0.02 0.10 0.09
(0.088) (0.104) (0.081) (0.093) (0.088) (0.100)
SES 0.26*** 0.24** 0.18*** 0.17** 0.25*** 0.23**
(0.066) (0.082) (0.052) (0.063) (0.064) (0.077)
SCHOOL TYPE (1=private) 0.09 0.21 0.11 0.17 0.09 0.16
(0.091) (0.106) (0.083) (0.095) (0.092) (0.103)
SCHOOL TRACK 0.57*** 0.67*** 0.62*** 0.76*** 0.54*** 0.66***
(0.139) (0.182) (0.140) (0.166) (0.146) (0.174)
STU-TEA RATIO -0.02 -0.04 -0.04 -0.05* -0.04* -0.04
(0.018) (0.022) (0.018) (0.022) (0.019) (0.022)
SCHOOL CHOICE -0.13 -0.22 -0.31* -0.29* -0.20 -0.25
(0.131) (0.154) (0.121) (0.137) (0.135) (0.149)
PRESCORE (G9) 0.79*** 0.80*** 0.79*** 0.79*** 0.80*** 0.77***
(0.059) (0.072) (0.057) (0.064) (0.061) (0.068)
STUDY HOURS (1= >10hrs /week) 0.41*** 0.28* 0.42*** 0.36*** 0.18 0.03
(0.100) (0.119) (0.090) (0.105) (0.099) (0.111)
SELF ESTEEM 0.44*** 0.46*** 0.40*** 0.41*** 0.51*** 0.54***
(0.074) (0.089) (0.066) (0.076) (0.074) (0.083)
English PT in middle school 0.02 -0.02 0.01 0.05 0.03 0.09
(0.043) (0.059) (0.045) (0.060) (0.048) (0.057)
Constant 4.78*** 4.88*** 5.48*** 5.45*** 5.32*** 5.29***
(0.402) (0.491) (0.394) (0.461) (0.416) (0.481)

Observations 1,157 878 1,228 956 963 757

R-squared 0.327 0.298 0.363 0.352 0.399 0.377
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)
 246

Table F.6. Single-year effects of English tutoring in high school on the CSAT English
achievement
English PT in G10 English PT in G11 English PT in G12

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.03 0.26 0.18** 0.33 0.19** 0.14

(0.067) (0.267) (0.065) (0.212) (0.068) (0.144)
URBANICITY -0.02 -0.05 -0.02 -0.02 -0.02 -0.03
(0.083) (0.091) (0.083) (0.086) (0.083) (0.088)
GENDER (1=female) 0.07 0.05 0.07 0.07 0.07 0.07
(0.058) (0.063) (0.058) (0.060) (0.058) (0.061)
SES 0.24*** 0.25*** 0.24*** 0.21*** 0.24*** 0.23***
(0.039) (0.044) (0.039) (0.041) (0.039) (0.042)
SCHOOL TYPE (1=private) 0.14* 0.19** 0.14* 0.16** 0.14* 0.14*
(0.059) (0.064) (0.059) (0.061) (0.059) (0.062)
SCHOOL TRACK 0.63*** 0.67*** 0.63*** 0.64*** 0.63*** 0.69***
(0.106) (0.122) (0.106) (0.110) (0.106) (0.116)
STU-TEA RATIO -0.03* -0.04** -0.03* -0.03* -0.03* -0.03*
(0.013) (0.015) (0.013) (0.014) (0.013) (0.014)
SCHOOL CHOICE -0.15 -0.18 -0.15 -0.14 -0.15 -0.19*
(0.088) (0.095) (0.088) (0.091) (0.088) (0.093)
PRESCORE (G9) 0.73*** 0.72*** 0.73*** 0.71*** 0.73*** 0.72***
(0.039) (0.042) (0.039) (0.041) (0.039) (0.041)
STUDY HOURS (1= >10hrs /week) 0.38*** 0.36*** 0.38*** 0.34*** 0.38*** 0.38***
(0.063) (0.068) (0.063) (0.065) (0.063) (0.066)
SELF ESTEEM 0.43*** 0.42*** 0.43*** 0.45*** 0.43*** 0.45***
(0.048) (0.053) (0.048) (0.050) (0.048) (0.051)
English PT in middle school 0.02 0.01 0.02 0.01 0.02 0.02
(0.034) (0.043) (0.034) (0.037) (0.034) (0.037)
English PT in grade 10 0.03 0.01 0.03 0.03
(0.067) (0.086) (0.067) (0.071)
English PT in grade 11 0.18** 0.13 0.18** 0.20*
(0.065) (0.093) (0.065) (0.077)
English PT in grade 12 0.19** 0.17* 0.19** 0.14
(0.068) (0.075) (0.068) (0.094)
Constant 4.95*** 4.99*** 4.95*** 4.91*** 4.95*** 4.96***
(0.284) (0.314) (0.284) (0.296) (0.284) (0.300)

Observations 2,344 2,032 2,344 2,212 2,344 2,114

R-squared 0.348 0.336 0.348 0.335 0.348 0.349
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)
 247

Table F.7. Cumulative effects of math tutoring in high school on the CSAT math achievement
1 year of Math PT in HS 2 years of Math PT in HS 3 years of Math PT in HS

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.20 0.56 0.73*** 1.01*** 1.18*** 1.08***

(0.111) (0.374) (0.116) (0.283) (0.121) (0.246)
URBANICITY -0.05 -0.15 -0.21 -0.24 -0.03 -0.03
(0.162) (0.190) (0.140) (0.158) (0.133) (0.148)
GENDER (1=female) -0.41*** -0.50*** -0.36*** -0.31** -0.30*** -0.36***
(0.107) (0.128) (0.097) (0.111) (0.090) (0.102)
SES 0.00 -0.06 -0.00 -0.02 0.06 0.08
(0.079) (0.099) (0.065) (0.078) (0.061) (0.075)
SCHOOL TYPE (1=private) 0.17 0.19 0.11 0.11 0.06 0.04
(0.110) (0.128) (0.099) (0.113) (0.092) (0.103)
SCHOOL TRACK 0.25 0.28 0.44** 0.50* 0.39* 0.52**
(0.156) (0.200) (0.163) (0.206) (0.157) (0.193)
STU-TEA RATIO -0.01 -0.02 -0.05* -0.06* -0.03 -0.02
(0.022) (0.026) (0.021) (0.025) (0.021) (0.024)
SCHOOL CHOICE 0.01 -0.11 -0.27 -0.36* -0.10 -0.20
(0.162) (0.192) (0.143) (0.165) (0.141) (0.160)
PRESCORE (G9) 0.82*** 0.78*** 0.78*** 0.70*** 0.72*** 0.66***
(0.072) (0.084) (0.066) (0.075) (0.065) (0.074)
STUDY HOURS (1= >10hrs /week) 0.72*** 0.68*** 0.58*** 0.50*** 0.46*** 0.33**
(0.125) (0.147) (0.106) (0.125) (0.100) (0.114)
SELF ESTEEM 0.29** 0.37*** 0.25** 0.27** 0.32*** 0.37***
(0.088) (0.105) (0.078) (0.091) (0.075) (0.083)
Math PT in middle school -0.08 -0.14* -0.05 -0.06 -0.12* -0.09
(0.050) (0.067) (0.053) (0.072) (0.055) (0.070)
Constant 4.97*** 5.36*** 5.81*** 5.98*** 5.40*** 5.42***
(0.483) (0.597) (0.443) (0.517) (0.440) (0.500)

Observations 910 680 1,083 832 1,081 854

R-squared 0.259 0.247 0.324 0.308 0.401 0.386
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)
 248

Table F.8. Single-year effects of math tutoring in high school on the CSAT math achievement
Math PT in G10 Math PT in G11 Math PT in G12

(1) (2) (3) (4) (5) (6)

VARIABLES OLS IV OLS IV OLS IV

Coefficient of interest 0.14 0.14 0.55*** 0.63** 0.43*** 0.42**

(0.082) (0.322) (0.076) (0.241) (0.072) (0.157)
URBANICITY -0.08 -0.10 -0.08 -0.09 -0.08 -0.07
(0.089) (0.096) (0.089) (0.092) (0.089) (0.095)
GENDER (1=female) -0.35*** -0.39*** -0.35*** -0.38*** -0.35*** -0.34***
(0.063) (0.069) (0.063) (0.065) (0.063) (0.066)
SES 0.04 0.04 0.04 0.04 0.04 0.04
(0.042) (0.047) (0.042) (0.044) (0.042) (0.045)
SCHOOL TYPE (1=private) 0.17** 0.16* 0.17** 0.18** 0.17** 0.19**
(0.064) (0.068) (0.064) (0.065) (0.064) (0.067)
SCHOOL TRACK 0.41*** 0.41** 0.41*** 0.39** 0.41*** 0.45***
(0.116) (0.139) (0.116) (0.121) (0.116) (0.127)
STU-TEA RATIO -0.04* -0.04** -0.04* -0.04* -0.04* -0.04*
(0.014) (0.016) (0.014) (0.015) (0.014) (0.015)
SCHOOL CHOICE -0.13 -0.21* -0.13 -0.17 -0.13 -0.16
(0.094) (0.101) (0.094) (0.098) (0.094) (0.100)
PRESCORE (G9) 0.75*** 0.71*** 0.75*** 0.75*** 0.75*** 0.75***
(0.043) (0.046) (0.043) (0.045) (0.043) (0.045)
STUDY HOURS (1= >10hrs /week) 0.50*** 0.49*** 0.50*** 0.48*** 0.50*** 0.47***
(0.068) (0.074) (0.068) (0.071) (0.068) (0.072)
SELF ESTEEM 0.26*** 0.26*** 0.26*** 0.28*** 0.26*** 0.29***
(0.052) (0.056) (0.052) (0.054) (0.052) (0.055)
Math PT in middle school -0.05 -0.05 -0.05 -0.07 -0.05 -0.06
(0.037) (0.047) (0.037) (0.042) (0.037) (0.040)
Math PT in grade 10 0.14 0.12 0.14 0.11
(0.082) (0.103) (0.082) (0.088)
Math PT in grade 11 0.55*** 0.58*** 0.55*** 0.60***
(0.076) (0.105) (0.076) (0.094)
Math PT in grade 12 0.43*** 0.45*** 0.43*** 0.40***
(0.072) (0.080) (0.072) (0.097)
Constant 5.40*** 5.67*** 5.40*** 5.56*** 5.40*** 5.43***
(0.305) (0.336) (0.305) (0.319) (0.305) (0.323)

Observations 2,338 2,033 2,338 2,209 2,338 2,117

R-squared 0.345 0.334 0.345 0.341 0.345 0.343
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)
 249

Table G.1. Balance diagnostics using the sample in grade 7

Mean SD
Variable Sample Treated Control Treated Control
PRESCORE_squared Unmatched 42.639 32.800 21.6 20.9
Matched 42.639 43.290 21.6 22.6

SEOUL_G7 Unmatched 0.201 0.118 0.4 0.3

Matched 0.201 0.198 0.4 0.4

URBANICITY_G7 Unmatched 0.503 0.364 0.5 0.5

Matched 0.503 0.505 0.5 0.5

GENDER Unmatched 1.474 1.533 0.5 0.5

Matched 1.474 1.459 0.5 0.5

HOUSEHOLD INCOME_G7 (log) Unmatched 5.826 5.441 0.5 0.6

Matched 5.826 5.799 0.5 0.6

DADEDU Unmatched 3.877 3.283 1.2 1.2

Matched 3.877 3.920 1.2 1.3

PARENTAL ASPIRATION_G7 Unmatched 4.678 4.297 0.9 0.9

Matched 4.678 4.717 0.9 1.0

SCHOOL TYPE_G7 Unmatched 0.207 0.191 0.4 0.4

Matched 0.207 0.186 0.4 0.4

STU-TEA RATIO Unmatched 20.901 19.448 4.0 5.3

Matched 20.901 20.960 4.0 4.2

SCHOOL CHOICE_G7 Unmatched 1.338 1.477 0.5 0.5

Matched 1.338 1.339 0.5 0.5
 250

Table G.2. Balance diagnostics using the sample in grade 8

Mean SD
Variable Sample Treated Control Treated Control
VERBAL SCORE_G7 Unmatched 68.515 59.964 17.5 18.5
Matched 68.515 68.821 17.5 17.1

ENGLISH SCORE_G7 Unmatched 60.872 45.397 22.2 18.9

Matched 60.872 59.616 22.2 21.6

MATH SCORE_G7 Unmatched 63.026 47.020 22.7 21.9

Matched 63.026 63.321 22.7 22.4

URBANICITY_G8 Unmatched 0.500 0.358 0.5 0.5

Matched 0.500 0.494 0.5 0.5

GENDER Unmatched 1.480 1.517 0.5 0.5

Matched 1.480 1.432 0.5 0.5

HOUSEHOLD INCOME_G7 Unmatched 5.805 5.416 0.5 0.6

Matched 5.805 5.812 0.5 0.6

DADEDU Unmatched 3.848 3.200 1.2 1.1

Matched 3.848 3.834 1.2 1.2

ACADEMIC
MOTIVATION_G8 Unmatched 4.180 3.854 0.8 0.8
Matched 4.180 4.170 0.8 0.9

PARENTAL ASPIRATION_G8 Unmatched 4.635 4.141 0.9 0.9

Matched 4.635 4.679 0.9 1.0

STU-TEA RATIO_G8 Unmatched 20.890 19.126 4.0 5.3

Matched 20.890 20.688 4.0 4.4

SCHOOL CHOICE_G8 Unmatched 1.343 1.494 0.5 0.5

Matched 1.343 1.354 0.5 0.5
 251

Table G.3. Balance diagnostics using the sample in grade 9

Mean SD
Variable Sample Treated Control Treated Control
VERBAL SCORE_G8 Unmatched 62.270 53.399 18.7 18.4
Matched 62.270 62.430 18.7 19.1

ENGLISH SCORE_G8 Unmatched 60.099 43.988 24.4 20.5

Matched 60.099 58.980 24.4 24.3

MATH SCORE_G8 Unmatched 55.477 39.517 24.7 20.2

Matched 55.477 55.218 24.7 23.0

SEOUL_G9 Unmatched 0.195 0.129 0.4 0.3

Matched 0.195 0.153 0.4 0.4

URBANICITY_G9 Unmatched 0.484 0.361 0.5 0.5

Matched 0.484 0.451 0.5 0.5

GENDER Unmatched 1.469 1.509 0.5 0.5

Matched 1.469 1.468 0.5 0.5

HOUSEHOLD INCOME_G9 Unmatched 5.940 5.470 0.6 0.7

Matched 5.940 5.969 0.6 0.7

DADEDU Unmatched 3.825 3.181 1.1 1.1

Matched 3.825 3.872 1.1 1.3

PARENTAL ASPIRATION_G9 Unmatched 4.557 4.035 0.9 0.9

Matched 4.557 4.551 0.9 1.0

SCHOOL TYPE_G9 Unmatched 0.205 0.208 0.4 0.4

Matched 0.205 0.214 0.4 0.4

STU-TEA RATIO_G9 Unmatched 20.651 18.677 3.8 5.1

Matched 20.651 20.450 3.8 3.9

SCHOOL CHOICE_G9 Unmatched 1.361 1.498 0.5 0.5

Matched 1.361 1.353 0.5 0.5
 252

Table G.4. Balance diagnostics using the sample in grade 10

Mean SD
Variable Sample Treated Control Treated Control
VERBAL SCORE_G9 Unmatched 61.771 51.969 20.0 20.1
Matched 61.771 61.809 20.0 20.1

ENGLISH SCORE_G9 Unmatched 61.403 43.363 25.9 22.6

Matched 61.403 61.329 25.9 26.0

MATH SCORE_G9 Unmatched 57.689 40.908 25.6 22.6

Matched 57.689 58.109 25.6 25.6

SEOUL_G10 Unmatched 0.193 0.137 0.4 0.3

Matched 0.193 0.187 0.4 0.4

URBANICITY_G10 Unmatched 0.502 0.379 0.5 0.5

Matched 0.502 0.513 0.5 0.5

GENDER Unmatched 1.503 1.459 0.5 0.5

Matched 1.503 1.499 0.5 0.5

HOUSEHOLD INCOME_G10 Unmatched 5.939 5.523 0.6 0.6

Matched 5.939 5.868 0.6 0.6

DADEDU Unmatched 3.897 3.223 1.1 1.0

Matched 3.897 3.992 1.1 1.2

ACADEMIC MOTIVATION_G10 Unmatched 3.234 2.839 0.8 0.8

Matched 3.234 3.302 0.8 0.9

PARENTAL ASPIRATION_G10 Unmatched 3.533 3.072 0.9 0.8

Matched 3.533 3.514 0.9 0.9

SCHOOL TYPE_G10 Unmatched 0.484 0.450 0.5 0.5

Matched 0.484 0.449 0.5 0.5

SCHOOL TRACK_G10 Unmatched 0.874 0.526 0.3 0.5

Matched 0.874 0.886 0.3 0.3

STU-TEA RATIO_G10 Unmatched 16.514 14.877 2.5 3.2

Matched 16.514 16.349 2.5 2.9

SCHOOL CHOICE_G10 Unmatched 1.341 1.484 0.5 0.5

Matched 1.341 1.339 0.5 0.5
 253

Table G.5. Balance diagnostics using the sample in grade 11

Mean SD
Variable Sample Treated Control Treated Control

VERBAL SCORE_G9 Unmatched 62.441 53.706 19.9 19.9

Matched 62.441 62.657 19.9 18.6

ENGLISH SCORE_G9 Unmatched 62.956 46.411 25.7 24.4

Matched 62.956 63.48 25.7 25.9

MATH SCORE_G9 Unmatched 59.75 43.875 25.5 23.7

Matched 59.75 60.495 25.5 25.4

SEOUL_G11 Unmatched 0.21863 0.11842 0.4 0.3

Matched 0.21863 0.20352 0.4 0.4

URBANICITY_G11 Unmatched 0.517 0.40235 0.5 0.5

Matched 0.517 0.50189 0.5 0.5

GENDER Unmatched 1.4893 1.4931 0.5 0.5

Matched 1.4893 1.4293 0.5 0.5

HOUSEHOLD INCOME_G11 Unmatched 5.9098 5.4449 0.7 0.7

Matched 5.9098 5.9435 0.7 0.8

DADEDU Unmatched 3.9261 3.3255 1.2 1.0

Matched 3.9261 3.8196 1.2 1.1

ACADEMIC MOTIVATION_G11 Unmatched 3.2673 2.8324 0.7 0.8

Matched 3.2673 3.2497 0.7 0.8

GENDER*MOTIVATION Unmatched 4.8586 4.2306 2.0 1.8

Matched 4.8586 4.6425 2.0 2.0

PARENTAL ASPIRATION_G11 Unmatched 3.4461 3.0332 0.9 0.8

Matched 3.4461 3.4473 0.9 0.9

SCHOOL TYPE_G11 Unmatched 0.50105 0.45776 0.5 0.5

Matched 0.50105 0.51951 0.5 0.5

SCHOOL TRACK_G11 Unmatched 0.89005 0.58241 0.3 0.5

Matched 0.89005 0.8888 0.3 0.3

STU-TEA RATIO_G11 Unmatched 16.945 15.318 2.6 3.1

Matched 16.945 16.797 2.6 2.8

SCHOOL CHOICE_G11 Unmatched 1.3273 1.4404 0.5 0.5

Matched 1.3273 1.3525 0.5 0.5
 254

Table G.6. Balance diagnostics using the sample in grade 12

Mean SD
Variable Sample Treated Control Treated Control

VERBAL SCORE_G9 Unmatched 64.243 57.291 19.1 20.2

Matched 64.243 64.422 19.1 20.2

ENGLISH SCORE_G9 Unmatched 66.993 51.782 24.9 25.5

Matched 66.993 67.384 24.9 25.6

MATH SCORE_G9 Unmatched 62.818 49.384 24.7 25.5

Matched 62.818 63.569 24.7 26.2

SEOUL_G12 Unmatched 0.24758 0.11873 0.4 0.3

Matched 0.24758 0.21859 0.4 0.4

URBANICITY_G12 Unmatched 0.58067 0.4312 0.5 0.5

Matched 0.58067 0.57844 0.5 0.5

GENDER Unmatched 1.51 1.4824 0.5 0.5

Matched 1.51 1.4952 0.5 0.5

HOUSEHOLD INCOME_G12 Unmatched 5.9986 5.5716 0.6 0.6

Matched 5.9986 5.988 0.6 0.7

DADEDU Unmatched 4.0825 3.5147 1.1 1.1

Matched 4.0825 4.1524 1.1 1.1

ACADEMIC MOTIVATION_G12 Unmatched 3.3673 2.9553 0.7 0.9

Matched 3.3673 3.4112 0.7 0.9

PARENTAL ASPIRATION_G12 Unmatched 3.5546 3.1207 0.8 0.9

Matched 3.5546 3.5933 0.8 0.9

SCHOOL TYPE_G12 Unmatched 0.51673 0.46349 0.5 0.5

Matched 0.51673 0.52119 0.5 0.5

SCHOOL TRACK_G12 Unmatched 0.9316 0.7228 0.3 0.4

Matched 0.9316 0.94052 0.3 0.2

STU-TEA RATIO_G12 Unmatched 17.118 15.691 2.4 3.1

Matched 17.118 17.159 2.4 2.3

SCHOOL CHOICE_G12 Unmatched 1.2669 1.4406 0.4 0.5

Matched 1.2669 1.2743 0.4 0.4
 255

Table H.1. Ordered Logit estimation of the effect of private tutoring on students’ attention to
lessons in formal schools in grade 7
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Q1 Q2 Q3

PT_G7 -0.03 -0.08 -0.10 -0.10 -0.02

(0.060) (0.080) (0.117) (0.140) (0.184)
PRESCORE (6th) 0.00 -0.04 0.02 0.28*
(0.018) (0.051) (0.102) (0.119)
SEOUL 0.05 -0.08 0.08 0.14
(0.089) (0.151) (0.148) (0.173)
URBANICITY -0.22* -0.25 -0.07 -0.34
(0.095) (0.165) (0.156) (0.177)
GENDER -0.01 -0.01 0.01 0.00
(0.061) (0.101) (0.102) (0.116)
HHINCOME 0.00 0.00 -0.00 0.00
(0.000) (0.000) (0.000) (0.000)
DADEDU 0.02 0.02 0.03 0.00
(0.030) (0.051) (0.049) (0.055)
ACADEMIC MOTIVATION 0.09* 0.16* 0.12 0.01
(0.040) (0.067) (0.070) (0.071)
PARENTAL ASIPRATION -0.02 0.04 -0.04 -0.04
(0.035) (0.063) (0.057) (0.063)
SCHOOL TYPE (1=private) -0.12 0.07 -0.31* -0.06
(0.075) (0.126) (0.127) (0.142)
STU-TEA RATIO -0.01 -0.03* -0.00 -0.00
(0.008) (0.012) (0.014) (0.016)
SCHOOL CHOICE -0.20* -0.40* 0.02 -0.18
(0.095) (0.162) (0.157) (0.183)
cut1

Constant -4.20*** -4.39*** -4.52*** -3.84*** -2.32

(0.117) (0.350) (0.595) (0.858) (1.200)
cut2

Constant -2.28*** -2.49*** -2.65*** -1.95* -0.34

(0.065) (0.331) (0.566) (0.834) (1.178)
cut3

Constant -1.27*** -1.53*** -1.64** -0.99 0.56

(0.057) (0.328) (0.563) (0.831) (1.177)
cut4

Constant 0.04 -0.22 -0.39 0.36 1.92

(0.054) (0.328) (0.562) (0.831) (1.179)

Observations 5,810 3,947 1,433 1,428 1,086

Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G7ATTN (degree of attention to lessons in grade 7)
 256

Table H.2. Ordered Logit estimation of the effect of private tutoring on students’ attention to
lessons in formal schools in grade 8
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Q1 Q2 Q3

PT_G8 0.81*** 0.27** 0.27* 0.22 0.32

(0.058) (0.083) (0.121) (0.139) (0.208)
VERBAL SCORE_G7 0.02*** 0.03*** 0.03*** 0.03***
(0.002) (0.004) (0.005) (0.007)
ENGLISH SCORE_G7 0.02*** 0.02*** 0.02*** 0.02***
(0.002) (0.005) (0.004) (0.005)
MATH SCORE_G7 0.02*** 0.02*** 0.02*** 0.03***
(0.002) (0.004) (0.004) (0.005)
SEOUL -0.03 -0.17 -0.05 0.04
(0.097) (0.183) (0.153) (0.177)
URBANICITY 0.14 0.33 0.09 0.00
(0.100) (0.189) (0.159) (0.179)
GENDER (1=female) 0.40*** 0.34** 0.44*** 0.42**
(0.067) (0.113) (0.109) (0.129)
HHINCOME -0.00* -0.00 -0.00* 0.00
(0.000) (0.000) (0.000) (0.000)
DADEDU -0.03 -0.06 -0.00 -0.06
(0.032) (0.057) (0.050) (0.061)
ACADEMIC MOTIVATION 0.19*** 0.14* 0.22** 0.19*
(0.045) (0.070) (0.080) (0.086)
PARENTAL ASIPRATION 0.14*** 0.07 0.11 0.22**
(0.039) (0.068) (0.066) (0.070)
SCHOOL TYPE (1=private) 0.01 -0.18 0.04 0.17
(0.078) (0.134) (0.127) (0.156)
STU-TEA RATIO -0.03** 0.00 -0.05** -0.07**
(0.008) (0.012) (0.014) (0.022)
SCHOOL CHOICE 0.26** 0.30 0.25 0.17
(0.097) (0.169) (0.161) (0.188)
cut1

Constant -3.16*** 0.87* 0.94 0.85 1.73

(0.094) (0.360) (0.585) (0.848) (1.148)
cut2

Constant -1.61*** 2.53*** 2.59*** 2.62** 3.32**

(0.059) (0.348) (0.580) (0.811) (1.078)
cut3

Constant -0.57*** 3.78*** 3.71*** 4.06*** 4.77***

(0.052) (0.349) (0.585) (0.807) (1.065)
cut4

Constant 0.76*** 5.45*** 5.32*** 5.79*** 6.49***

(0.053) (0.355) (0.595) (0.816) (1.069)

Observations 5,954 4,044 1,161 1,393 1,490

Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G8ATTN (degree of attention to lessons in grade 8)
 257

Table H.3. Ordered Logit estimation of the effect of private tutoring on students’ attention to
lessons in formal schools in grade 9
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Q1 Q2 Q3

PT_G9 0.84*** 0.17* 0.19 0.23 -0.13

(0.061) (0.086) (0.126) (0.142) (0.227)
VERBAL SCORE_G8 0.02*** 0.03*** 0.02*** 0.02**
(0.003) (0.005) (0.005) (0.007)
ENGLISH SCORE_G8 0.01*** 0.02*** 0.01 0.01
(0.002) (0.006) (0.005) (0.006)
MATH SCORE_G8 0.01*** 0.02*** 0.01** 0.02***
(0.002) (0.005) (0.004) (0.005)
SEOUL 0.14 -0.01 0.26 0.15
(0.108) (0.195) (0.182) (0.190)
URBANICITY -0.25* -0.37 -0.25 -0.21
(0.113) (0.209) (0.186) (0.197)
GENDER (1=female) 0.49*** 0.28* 0.54*** 0.59***
(0.074) (0.126) (0.122) (0.139)
HHINCOME -0.00 0.00 0.00 -0.00
(0.000) (0.000) (0.000) (0.000)
DADEDU -0.08* -0.02 -0.09 -0.10
(0.034) (0.061) (0.057) (0.062)
ACADEMIC MOTIVATION 0.28*** 0.22** 0.36*** 0.31***
(0.047) (0.071) (0.093) (0.087)
PARENTAL ASIPRATION 0.21*** 0.21** 0.19* 0.24**
(0.043) (0.073) (0.075) (0.077)
SCHOOL TYPE (1=private) -0.09 -0.08 -0.13 -0.10
(0.085) (0.149) (0.140) (0.159)
STU-TEA RATIO -0.00 0.00 -0.01 -0.02
(0.009) (0.014) (0.015) (0.021)
SCHOOL CHOICE 0.10 -0.01 -0.06 0.41*
(0.106) (0.176) (0.178) (0.207)
cut1

Constant -2.76*** 1.12** 1.46* 0.22 1.01

(0.088) (0.371) (0.606) (0.854) (1.017)
cut2

Constant -1.30*** 2.73*** 3.03*** 2.16** 2.47**

(0.059) (0.363) (0.606) (0.822) (0.954)
cut3

Constant -0.28*** 3.89*** 4.12*** 3.38*** 3.92***

(0.054) (0.366) (0.613) (0.821) (0.942)
cut4

Constant 1.02*** 5.56*** 5.72*** 5.03*** 5.77***

(0.056) (0.372) (0.625) (0.829) (0.949)

Observations 4,721 3,114 972 1,075 1,067

Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ATTN (degree of attention to lessons in grade 9)
 258

Table H.4. Ordered Logit estimation of the effect of private tutoring on students’ attention to
lessons in formal schools in grade 10
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Q1 Q2 Q3

PT_G10 0.65*** 0.16* 0.16 0.15 0.18

(0.051) (0.078) (0.123) (0.124) (0.178)
VERBAL SCORE_G9 0.02*** 0.02*** 0.01* 0.03***
(0.002) (0.004) (0.004) (0.006)
ENGLISH SCORE_G9 0.01*** 0.01* 0.01 0.02***
(0.002) (0.005) (0.003) (0.005)
MATH SCORE_G9 0.01*** 0.01 0.01** 0.01***
(0.002) (0.005) (0.003) (0.004)
SEOUL 0.09 -0.25 0.14 0.48**
(0.097) (0.178) (0.158) (0.179)
URBANICITY -0.23* -0.49* -0.14 -0.15
(0.101) (0.192) (0.171) (0.171)
GENDER (1=female) 0.35*** 0.25* 0.46*** 0.30*
(0.067) (0.119) (0.112) (0.123)
HHINCOME -0.00 -0.00 -0.00 -0.00
(0.000) (0.000) (0.000) (0.000)
DADEDU 0.01 0.04 -0.01 -0.01
(0.031) (0.059) (0.052) (0.053)
ACADEMIC MOTIVATION 0.27*** 0.23** 0.27** 0.25**
(0.048) (0.079) (0.084) (0.087)
PARENTAL ASIPRATION 0.09* 0.14 0.15* -0.01
(0.041) (0.078) (0.068) (0.070)
SCHOOL TYPE (1=private) 0.16* 0.31* -0.02 0.22
(0.067) (0.123) (0.111) (0.119)
SCHOOL TRACK (1=academic) -0.36*** -0.38** -0.31 -0.21
(0.096) (0.139) (0.158) (0.278)
STU-TEA RATIO -0.01 0.00 -0.00 -0.02
(0.014) (0.023) (0.025) (0.028)
SCHOOL CHOICE 0.05 -0.21 0.29 -0.01
(0.098) (0.177) (0.165) (0.175)
cut1

Constant -3.61*** -0.88* -1.05 -1.53* 0.54

(0.101) (0.344) (0.591) (0.736) (0.936)
cut2

Constant -1.88*** 0.88** 0.60 0.74 2.24**

(0.054) (0.325) (0.580) (0.666) (0.845)
cut3

Constant -0.50*** 2.34*** 1.91** 2.41*** 3.96***

(0.043) (0.325) (0.581) (0.663) (0.830)
cut4

Constant 1.00*** 4.11*** 3.51*** 4.22*** 5.89***

(0.045) (0.330) (0.588) (0.672) (0.837)

Observations 6,096 3,635 1,068 1,265 1,302

Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G10ATTN (degree of attention to lessons in grade 10)
 259

Table H.5. Ordered Logit estimation of the effect of private tutoring on students’ attention to
lessons in formal schools in grade 11
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Q1 Q2 Q3

PT_G11 0.46*** 0.15* 0.04 0.01 0.44***

(0.051) (0.070) (0.125) (0.118) (0.126)
VERBAL SCORE_G9 0.02*** 0.01* 0.01*** 0.02***
(0.002) (0.004) (0.004) (0.005)
ENGLISH SCORE_G9 0.01*** 0.00 0.00 0.02***
(0.002) (0.005) (0.003) (0.005)
MATH SCORE_G9 0.00** -0.00 -0.00 0.01**
(0.002) (0.005) (0.003) (0.004)
SEOUL 0.06 -0.05 0.16 0.17
(0.092) (0.174) (0.158) (0.154)
URBANICITY -0.05 0.04 -0.07 -0.06
(0.095) (0.191) (0.167) (0.150)
GENDER (1=female) 0.43*** 0.44*** 0.50*** 0.34**
(0.065) (0.121) (0.113) (0.109)
HHINCOME -0.00 0.00 0.00 -0.00
(0.000) (0.000) (0.000) (0.000)
DADEDU -0.08** -0.07 -0.05 -0.15**
(0.030) (0.059) (0.053) (0.048)
ACADEMIC MOTIVATION 0.31*** 0.24** 0.39*** 0.28***
(0.046) (0.079) (0.091) (0.076)
PARENTAL ASIPRATION 0.18*** 0.21* 0.13 0.21**
(0.042) (0.086) (0.073) (0.066)
SCHOOL TYPE (1=private) 0.07 0.05 0.07 0.03
(0.064) (0.122) (0.110) (0.106)
SCHOOL TRACK (1=academic) -0.45*** -0.60*** -0.32* 0.02
(0.095) (0.148) (0.161) (0.236)
STU-TEA RATIO -0.00 0.05* -0.02 -0.04
(0.013) (0.022) (0.025) (0.023)
SCHOOL CHOICE 0.05 0.15 0.15 -0.11
(0.093) (0.176) (0.164) (0.152)
cut1

Constant -3.10*** -0.41 -0.20 -0.91 0.21

(0.080) (0.313) (0.567) (0.695) (0.759)
cut2

Constant -1.34*** 1.54*** 1.70** 1.12 2.32**

(0.046) (0.302) (0.560) (0.675) (0.706)
cut3

Constant 0.20*** 3.31*** 3.39*** 2.87*** 4.26***

(0.041) (0.306) (0.568) (0.679) (0.705)
cut4

Constant 2.00*** 5.36*** 5.54*** 4.96*** 6.31***

(0.050) (0.313) (0.585) (0.689) (0.718)

Observations 5,415 3,682 1,069 1,233 1,380

Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G11ATTN (degree of attention to lessons in grade 11)
 260

Table H.6. Ordered Logit estimation of the effect of private tutoring on students’ attention to
lessons in formal schools in grade 12
Average Effect Heterogeneous Effect
(1) (2) (3) (4) (5)
VARIABLES Empty Full Q1 Q2 Q3

PT_G12 0.44*** 0.12 0.26 0.26* -0.03

(0.053) (0.073) (0.162) (0.127) (0.111)
VERBAL SCORE_G9 0.01*** 0.00 0.01 0.01**
(0.002) (0.005) (0.004) (0.005)
ENGLISH SCORE_G9 0.01*** 0.01 0.00 0.02***
(0.002) (0.006) (0.004) (0.005)
MATH SCORE_G9 0.01*** 0.00 -0.00 0.01**
(0.002) (0.005) (0.004) (0.004)
SEOUL 0.04 0.11 -0.29 0.36*
(0.100) (0.192) (0.170) (0.167)
URBANICITY 0.06 0.21 0.30 -0.14
(0.105) (0.224) (0.183) (0.163)
GENDER (1=female) 0.26*** 0.20 0.23 0.33**
(0.070) (0.134) (0.119) (0.115)
HHINCOME 0.00 0.00 0.00 0.00
(0.000) (0.000) (0.000) (0.000)
DADEDU -0.08* -0.09 -0.13* -0.05
(0.031) (0.063) (0.056) (0.049)
ACADEMIC MOTIVATION 0.32*** 0.18* 0.43*** 0.34***
(0.050) (0.093) (0.090) (0.079)
PARENTAL ASIPRATION 0.16*** 0.07 0.16 0.17*
(0.046) (0.090) (0.083) (0.072)
SCHOOL TYPE (1=private) 0.22** 0.03 0.37** 0.20
(0.069) (0.135) (0.118) (0.111)
SCHOOL TRACK (1=academic) -0.51*** -0.70*** -0.41* 0.23
(0.104) (0.158) (0.170) (0.299)
STU-TEA RATIO -0.01 0.03 -0.02 -0.03
(0.014) (0.025) (0.025) (0.024)
SCHOOL CHOICE 0.22* 0.46* 0.47** -0.06
(0.105) (0.213) (0.181) (0.167)
cut1

Constant -3.64*** -1.18*** -1.22 -1.62* -0.08

(0.096) (0.350) (0.680) (0.730) (0.818)
cut2

Constant -1.67*** 0.86** 0.73 0.50 2.07**

(0.043) (0.331) (0.665) (0.700) (0.751)
cut3

Constant -0.01 2.77*** 2.47*** 2.54*** 4.15***

(0.034) (0.333) (0.671) (0.704) (0.749)
cut4

Constant 1.79*** 4.75*** 4.47*** 4.63*** 6.13***

(0.043) (0.340) (0.683) (0.713) (0.760)

Observations 5,135 3,268 878 1,136 1,254

Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G12ATTN (degree of attention to lessons in grade 12)
Table I.1. Heterogeneous effects of one year of private tutoring in middle school on overall academic achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of PT -0.03 -0.06 -0.04 -0.10* -0.04 0.15 0.16 -0.00 0.08 0.16*** 0.26*** 0.27***
(0.046) (0.056) (0.042) (0.053) (0.070) (0.116) (0.131) (0.028) (0.064) (0.050) (0.043) (0.038)
URBAN -0.01 -0.01 0.09 0.07 -0.04 -0.07 -0.06 0.03 -0.04 -0.10** -0.08* -0.08
(0.072) (0.075) (0.060) (0.153) (0.121) (0.188) (0.089) (0.054) (0.067) (0.033) (0.040) (0.047)
GENDER (1=female) 0.25*** 0.12 0.24*** 0.29*** 0.29*** 0.15* 0.26*** 0.23*** 0.34*** 0.32*** 0.19*** 0.07*
(0.046) (0.063) (0.059) (0.076) (0.071) (0.071) (0.055) (0.021) (0.036) (0.032) (0.026) (0.034)
SES 0.17*** 0.04 0.10*** 0.18*** 0.19*** 0.18*** 0.18*** 0.12*** 0.21*** 0.22*** 0.20*** 0.15***
(0.036) (0.048) (0.028) (0.047) (0.048) (0.054) (0.048) (0.029) (0.029) (0.022) (0.018) (0.018)
SCHOOL TYPE 0.07 0.10 0.12* 0.02 0.05 0.09 0.07 0.10* 0.13* 0.12*** 0.10*** 0.10**
(0.059) (0.072) (0.060) (0.057) (0.085) (0.179) (0.069) (0.041) (0.052) (0.028) (0.028) (0.036)
STU-TEA RATIO -0.00 -0.00 -0.00 -0.01 -0.00 0.01 -0.00 0.01 0.01 0.01 0.01*** 0.01**
(0.005) (0.005) (0.005) (0.006) (0.008) (0.012) (0.006) (0.003) (0.003) (0.004) (0.003) (0.004)
SCHOOL CHOICE 0.01 -0.01 0.10 0.09 -0.08 -0.00 -0.04 -0.02 -0.04 -0.09* -0.03 -0.00
(0.069) (0.053) (0.058) (0.149) (0.101) (0.220) (0.087) (0.041) (0.063) (0.039) (0.035) (0.041)
PRESCORE 0.15*** 0.07*** 0.11*** 0.16*** 0.19*** 0.21*** 0.16*** 0.10*** 0.16*** 0.20*** 0.19*** 0.14***
(0.014) (0.016) (0.011) (0.019) (0.018) (0.025) (0.016) (0.015) (0.009) (0.006) (0.006) (0.010)
SELF-STUDY 0.15 0.12 0.28 0.18 0.09 0.02 0.12 -0.01 0.06 0.08 0.10* 0.06
(0.102) (0.179) (0.197) (0.118) (0.153) (0.166) (0.121) (0.058) (0.068) (0.057) (0.039) (0.037)
SELF-ESTEEM 0.10* 0.03 0.06 0.08 0.15* 0.01 0.13** 0.08* 0.19*** 0.20*** 0.14*** 0.14***
(0.042) (0.048) (0.035) (0.052) (0.065) (0.094) (0.050) (0.035) (0.034) (0.020) (0.018) (0.021)
Constant -1.25*** -1.57*** -1.68*** -1.40*** -0.86** -0.87 -1.19*** -1.78*** -1.75*** -1.44*** -1.06*** -0.35**
(0.187) (0.126) (0.168) (0.328) (0.291) (0.655) (0.254) (0.118) (0.151) (0.135) (0.104) (0.133)

Observations 778 778 778 778 778 778 521 4,286 4,286 4,286 4,286 4,286
R-squared 0.261 0.294
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in grade 9)

261
Table I.2. Heterogeneous effects of two years of private tutoring in middle school on overall academic achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of PT 0.19*** 0.10* 0.14* 0.18** 0.30*** 0.41*** 0.25 -0.00 0.15 0.31*** 0.49*** 0.50***
(0.048) (0.044) (0.059) (0.069) (0.060) (0.063) (0.132) (0.054) (0.108) (0.086) (0.096) (0.081)
URBAN -0.00 0.00 0.16 0.02 -0.02 -0.09 -0.02 0.03 -0.04 -0.10** -0.08 -0.08*
(0.068) (0.072) (0.089) (0.109) (0.111) (0.146) (0.081) (0.061) (0.050) (0.032) (0.044) (0.034)
GENDER (1=female) 0.27*** 0.18*** 0.28*** 0.31*** 0.28*** 0.16 0.28*** 0.23*** 0.34*** 0.32*** 0.20*** 0.08**
(0.043) (0.044) (0.070) (0.062) (0.060) (0.091) (0.051) (0.041) (0.022) (0.029) (0.031) (0.031)
SES 0.08** -0.05 0.02 0.14* 0.11*** 0.13** 0.11** 0.12*** 0.20*** 0.20*** 0.16*** 0.12***
(0.032) (0.042) (0.046) (0.060) (0.031) (0.040) (0.042) (0.024) (0.022) (0.022) (0.025) (0.017)
SCHOOL TYPE 0.10 0.12* 0.09 0.04 -0.02 0.12 0.08 0.10* 0.13*** 0.11*** 0.08** 0.07*
(0.055) (0.053) (0.082) (0.070) (0.085) (0.136) (0.063) (0.042) (0.030) (0.029) (0.025) (0.033)
STU-TEA RATIO 0.00 0.00 -0.00 -0.00 -0.00 0.00 0.00 0.01 0.01 0.01 0.01* 0.01*
(0.004) (0.005) (0.007) (0.008) (0.008) (0.008) (0.005) (0.003) (0.004) (0.004) (0.004) (0.004)
SCHOOL CHOICE 0.05 -0.03 0.08 0.06 0.07 0.03 0.05 -0.02 -0.04 -0.10* -0.04 -0.01
(0.067) (0.066) (0.090) (0.134) (0.110) (0.142) (0.081) (0.039) (0.049) (0.044) (0.036) (0.039)
PRESCORE 0.15*** 0.04** 0.10*** 0.18*** 0.20*** 0.17*** 0.16*** 0.10*** 0.15*** 0.20*** 0.18*** 0.13***
(0.012) (0.013) (0.017) (0.016) (0.012) (0.017) (0.015) (0.015) (0.013) (0.007) (0.008) (0.009)
SELF-STUDY 0.20* 0.08 0.13 0.29** 0.30** 0.11 0.15 -0.01 0.05 0.08* 0.09* 0.05
(0.086) (0.139) (0.145) (0.093) (0.094) (0.133) (0.100) (0.055) (0.042) (0.033) (0.035) (0.030)
SELF-ESTEEM 0.11** 0.03 0.02 0.12 0.15** 0.14 0.12* 0.08* 0.18*** 0.19*** 0.13*** 0.13***
(0.041) (0.053) (0.093) (0.064) (0.057) (0.085) (0.050) (0.036) (0.029) (0.019) (0.025) (0.026)
Constant -1.43*** -1.61*** -1.80*** -1.53*** -1.22*** -0.73* -1.47*** -1.78*** -1.78*** -1.49*** -1.14*** -0.43*
(0.179) (0.191) (0.254) (0.317) (0.250) (0.325) (0.231) (0.130) (0.128) (0.147) (0.152) (0.183)

Observations 1,084 1,084 1,084 1,084 1,084 1,084 763 4,286 4,286 4,286 4,286 4,286
R-squared 0.240 0.257
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in grade 9)

262
Table I.3. Heterogeneous effects of three years of private tutoring in middle school on overall academic achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of PT 0.38*** 0.21*** 0.31*** 0.37*** 0.53*** 0.57*** 0.54*** -0.01 0.30 0.61** 0.98*** 1.00***
(0.045) (0.061) (0.060) (0.045) (0.050) (0.075) (0.133) (0.097) (0.246) (0.201) (0.199) (0.151)
URBAN -0.07 -0.08 -0.09 -0.11* 0.00 0.01 -0.10* 0.03 -0.04 -0.10** -0.08** -0.07*
(0.039) (0.055) (0.068) (0.051) (0.056) (0.072) (0.045) (0.040) (0.052) (0.030) (0.026) (0.037)
GENDER (1=female) 0.28*** 0.30*** 0.40*** 0.33*** 0.19*** 0.11* 0.27*** 0.23*** 0.35*** 0.33*** 0.21*** 0.09***
(0.027) (0.042) (0.043) (0.038) (0.041) (0.051) (0.031) (0.033) (0.035) (0.018) (0.028) (0.025)
SES 0.15*** 0.09** 0.13** 0.17*** 0.17*** 0.13*** 0.15*** 0.12*** 0.19*** 0.20*** 0.15*** 0.10***
(0.018) (0.028) (0.048) (0.023) (0.020) (0.031) (0.023) (0.024) (0.036) (0.029) (0.023) (0.017)
SCHOOL TYPE 0.08* 0.04 0.06 0.11* 0.04 0.08 0.10* 0.10* 0.13* 0.11*** 0.08** 0.08*
(0.034) (0.053) (0.058) (0.044) (0.021) (0.048) (0.039) (0.042) (0.051) (0.032) (0.030) (0.038)
STU-TEA RATIO 0.01* 0.01 0.01** 0.00 0.01 0.00 0.01* 0.01 0.00 0.00 0.00 0.00
(0.004) (0.004) (0.004) (0.005) (0.003) (0.004) (0.004) (0.004) (0.003) (0.003) (0.003) (0.005)
SCHOOL CHOICE -0.04 -0.11 -0.06 -0.08 0.03 0.01 -0.04 -0.02 -0.03 -0.09* -0.02 0.01
(0.042) (0.066) (0.085) (0.052) (0.052) (0.080) (0.049) (0.043) (0.049) (0.039) (0.033) (0.042)
PRESCORE 0.16*** 0.12*** 0.18*** 0.20*** 0.17*** 0.12*** 0.15*** 0.10*** 0.15*** 0.19*** 0.17*** 0.12***
(0.008) (0.009) (0.013) (0.009) (0.008) (0.009) (0.010) (0.014) (0.013) (0.009) (0.012) (0.012)
SELF-STUDY 0.04 -0.07 0.07 0.10 0.06 0.10* 0.05 -0.01 0.05 0.07 0.07* 0.04
(0.039) (0.110) (0.106) (0.061) (0.048) (0.046) (0.044) (0.072) (0.068) (0.050) (0.035) (0.032)
SELF-ESTEEM 0.13*** 0.08 0.13*** 0.14*** 0.12*** 0.12*** 0.13*** 0.08*** 0.18*** 0.18*** 0.11*** 0.11***
(0.025) (0.048) (0.035) (0.027) (0.027) (0.026) (0.029) (0.023) (0.033) (0.033) (0.027) (0.030)
Constant -1.42*** -1.87*** -2.07*** -1.46*** -1.11*** -0.37* -1.52*** -1.78*** -1.86*** -1.66*** -1.41*** -0.71***
(0.124) (0.172) (0.214) (0.158) (0.102) (0.183) (0.166) (0.128) (0.186) (0.204) (0.157) (0.161)

Observations 2,854 2,854 2,854 2,854 2,854 2,854 2,252 4,286 4,286 4,286 4,286 4,286
R-squared 0.294 0.267
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in grade 9)

263
Table I.4. Heterogeneous effects of one year of verbal tutoring in middle school on verbal achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of Verbal PT -0.17*** -0.17 -0.24** -0.17** -0.15** -0.13** -0.07 -0.05 -0.08 -0.08 -0.06 -0.07
(0.043) (0.088) (0.082) (0.066) (0.049) (0.045) (0.121) (0.052) (0.066) (0.044) (0.049) (0.039)
URBAN -0.01 0.14 0.06 -0.11 -0.02 -0.05 -0.08 0.02 -0.06 -0.07 -0.04 -0.01
(0.064) (0.095) (0.074) (0.084) (0.061) (0.060) (0.074) (0.066) (0.069) (0.043) (0.029) (0.034)
GENDER (1=female) 0.49*** 0.41*** 0.64*** 0.62*** 0.38*** 0.20*** 0.49*** 0.44*** 0.69*** 0.48*** 0.32*** 0.19***
(0.043) (0.058) (0.059) (0.065) (0.050) (0.048) (0.049) (0.045) (0.050) (0.025) (0.027) (0.021)
SES 0.17*** 0.03 0.21*** 0.21*** 0.15*** 0.11*** 0.17*** 0.08 0.18*** 0.15*** 0.13*** 0.10***
(0.027) (0.046) (0.031) (0.029) (0.026) (0.022) (0.032) (0.041) (0.027) (0.023) (0.020) (0.017)
SCHOOL TYPE 0.18*** 0.16 0.31*** 0.26*** 0.10* 0.04 0.20*** 0.06 0.16** 0.14*** 0.06* 0.03
(0.052) (0.094) (0.088) (0.057) (0.040) (0.048) (0.059) (0.061) (0.048) (0.035) (0.031) (0.024)
STU-TEA RATIO 0.00 -0.00 0.01 0.00 0.00 0.00 0.00 0.01 0.01* 0.01* 0.01 0.01*
(0.005) (0.007) (0.008) (0.007) (0.006) (0.004) (0.005) (0.004) (0.005) (0.004) (0.004) (0.003)
SCHOOL CHOICE -0.00 0.03 -0.01 -0.10 -0.04 -0.03 -0.06 -0.08 -0.10 -0.09 0.02 0.07*
(0.067) (0.107) (0.080) (0.109) (0.068) (0.062) (0.078) (0.052) (0.071) (0.052) (0.041) (0.029)
PRESCORE 0.17*** 0.09*** 0.19*** 0.21*** 0.17*** 0.15*** 0.17*** 0.07*** 0.18*** 0.19*** 0.15*** 0.12***
(0.012) (0.024) (0.024) (0.020) (0.016) (0.019) (0.014) (0.013) (0.013) (0.012) (0.010) (0.010)
SELF-STUDY 0.17* 0.23* 0.14 0.18*** 0.16** 0.18** 0.16 0.06 0.01 0.08 0.12** 0.08*
(0.071) (0.115) (0.151) (0.044) (0.057) (0.069) (0.080) (0.070) (0.046) (0.050) (0.039) (0.034)
SELF-ESTEEM 0.06 -0.03 0.10 0.11* 0.05 0.02 0.10* 0.14*** 0.19*** 0.18*** 0.10*** 0.05
(0.038) (0.051) (0.057) (0.050) (0.045) (0.054) (0.044) (0.041) (0.041) (0.033) (0.023) (0.030)
Constant -1.79*** -2.46*** -2.75*** -1.90*** -0.84** -0.04 -1.66*** -2.34*** -2.62*** -1.69*** -0.75*** -0.05
(0.188) (0.314) (0.192) (0.270) (0.264) (0.212) (0.224) (0.133) (0.196) (0.175) (0.102) (0.081)

Observations 1,768 1,768 1,768 1,768 1,768 1,768 1,304 4,187 4,187 4,187 4,187 4,187
R-squared 0.249 0.259
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VER_Z (verbal score in grade 9)

264
Table I.5. Heterogeneous effects of two years of verbal tutoring in middle school on verbal achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of Verbal PT -0.05 -0.02 -0.14* -0.07 -0.07 -0.05 -0.11 -0.05 -0.08 -0.08 -0.06 -0.07
(0.042) (0.060) (0.065) (0.054) (0.045) (0.050) (0.087) (0.054) (0.069) (0.062) (0.045) (0.049)
URBAN -0.06 -0.01 -0.04 -0.09 -0.04 -0.03 -0.06 0.02 -0.07 -0.07 -0.05 -0.01
(0.060) (0.080) (0.105) (0.109) (0.071) (0.066) (0.069) (0.092) (0.080) (0.043) (0.044) (0.037)
GENDER (1=female) 0.48*** 0.41*** 0.69*** 0.52*** 0.40*** 0.18*** 0.47*** 0.44*** 0.69*** 0.47*** 0.32*** 0.19***
(0.041) (0.052) (0.059) (0.055) (0.046) (0.045) (0.049) (0.051) (0.052) (0.032) (0.027) (0.024)
SES 0.13*** 0.02 0.14*** 0.17*** 0.12*** 0.09*** 0.16*** 0.08* 0.18*** 0.15*** 0.13*** 0.10***
(0.027) (0.054) (0.041) (0.028) (0.025) (0.024) (0.031) (0.034) (0.029) (0.021) (0.015) (0.015)
SCHOOL TYPE 0.11* 0.02 0.18** 0.16* 0.08 0.07 0.13* 0.06 0.16* 0.14*** 0.07 0.04
(0.051) (0.069) (0.060) (0.066) (0.052) (0.051) (0.059) (0.068) (0.073) (0.036) (0.046) (0.037)
STU-TEA RATIO 0.01 0.01 0.01 0.01 0.01 -0.00 0.00 0.01* 0.01 0.01* 0.01* 0.01*
(0.005) (0.006) (0.006) (0.005) (0.005) (0.004) (0.005) (0.004) (0.007) (0.004) (0.003) (0.003)
SCHOOL CHOICE -0.05 -0.10 -0.06 -0.10 -0.02 -0.03 -0.06 -0.08 -0.10 -0.09* 0.02 0.07
(0.062) (0.081) (0.106) (0.104) (0.069) (0.060) (0.072) (0.070) (0.070) (0.040) (0.048) (0.037)
PRESCORE 0.16*** 0.08** 0.19*** 0.20*** 0.16*** 0.13*** 0.14*** 0.07*** 0.18*** 0.19*** 0.15*** 0.12***
(0.012) (0.024) (0.027) (0.015) (0.016) (0.015) (0.014) (0.020) (0.018) (0.009) (0.012) (0.011)
SELF-STUDY 0.11 0.16 0.13 0.14* 0.14 0.08 0.11 0.06 0.02 0.08 0.12** 0.09*
(0.064) (0.116) (0.184) (0.070) (0.081) (0.058) (0.073) (0.054) (0.074) (0.067) (0.047) (0.035)
SELF-ESTEEM 0.12*** 0.09 0.15* 0.14** 0.07* 0.06 0.16*** 0.14*** 0.19*** 0.18*** 0.10*** 0.05*
(0.037) (0.056) (0.073) (0.048) (0.031) (0.044) (0.043) (0.031) (0.042) (0.023) (0.025) (0.025)
Constant -1.61*** -2.31*** -2.66*** -1.71*** -0.85** 0.17 -1.43*** -2.34*** -2.62*** -1.69*** -0.75*** -0.05
(0.181) (0.260) (0.283) (0.239) (0.285) (0.160) (0.216) (0.228) (0.213) (0.119) (0.148) (0.113)

Observations 1,900 1,900 1,900 1,900 1,900 1,900 1,394 4,187 4,187 4,187 4,187 4,187
R-squared 0.215 0.207
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VER_Z (verbal score in grade 9)

265
Table I.6. Heterogeneous effects of three years of verbal tutoring in middle school on verbal achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of Verbal PT 0.02 0.12 -0.02 0.02 -0.04 -0.05 -0.12 -0.07 -0.12 -0.12 -0.10 -0.11
(0.040) (0.071) (0.087) (0.065) (0.042) (0.033) (0.083) (0.069) (0.104) (0.062) (0.051) (0.060)
URBAN -0.05 0.07 0.00 -0.12 -0.04 -0.06 -0.10 0.02 -0.06 -0.07 -0.05 -0.01
(0.055) (0.131) (0.089) (0.086) (0.086) (0.046) (0.062) (0.087) (0.093) (0.048) (0.040) (0.032)
GENDER (1=female) 0.47*** 0.53*** 0.71*** 0.51*** 0.33*** 0.19*** 0.44*** 0.44*** 0.68*** 0.47*** 0.31*** 0.18***
(0.038) (0.085) (0.058) (0.036) (0.034) (0.042) (0.044) (0.059) (0.040) (0.037) (0.038) (0.030)
SES 0.06* 0.01 0.05 0.07* 0.09*** 0.05 0.09** 0.08*** 0.19*** 0.15*** 0.13*** 0.10***
(0.025) (0.035) (0.050) (0.031) (0.026) (0.028) (0.029) (0.023) (0.031) (0.020) (0.025) (0.018)
SCHOOL TYPE 0.03 0.00 0.05 0.12* 0.06 0.02 0.03 0.06 0.16*** 0.14*** 0.07 0.04
(0.047) (0.080) (0.095) (0.055) (0.048) (0.040) (0.052) (0.060) (0.049) (0.030) (0.036) (0.024)
STU-TEA RATIO 0.01 -0.00 0.01 0.01 0.01 0.00 0.01 0.01* 0.01 0.01* 0.01 0.01
(0.005) (0.010) (0.009) (0.008) (0.005) (0.004) (0.005) (0.004) (0.007) (0.004) (0.004) (0.004)
SCHOOL CHOICE -0.06 -0.04 -0.01 -0.14 -0.00 -0.04 -0.11 -0.08 -0.10 -0.09 0.01 0.06*
(0.058) (0.138) (0.104) (0.104) (0.079) (0.049) (0.066) (0.087) (0.087) (0.046) (0.032) (0.025)
PRESCORE 0.16*** 0.07*** 0.19*** 0.20*** 0.16*** 0.13*** 0.16*** 0.07*** 0.18*** 0.19*** 0.15*** 0.12***
(0.011) (0.022) (0.024) (0.015) (0.012) (0.010) (0.013) (0.015) (0.014) (0.010) (0.009) (0.008)
SELF-STUDY -0.00 -0.04 -0.13 0.04 0.04 0.02 0.02 0.06 0.02 0.08 0.12** 0.08**
(0.058) (0.069) (0.086) (0.088) (0.065) (0.055) (0.064) (0.078) (0.067) (0.064) (0.043) (0.030)
SELF-ESTEEM 0.14*** 0.17** 0.24*** 0.15** 0.08* 0.07* 0.14*** 0.15*** 0.19*** 0.18*** 0.10*** 0.05*
(0.034) (0.058) (0.065) (0.051) (0.031) (0.029) (0.039) (0.035) (0.047) (0.034) (0.027) (0.024)
Constant -1.66*** -2.43*** -2.89*** -1.75*** -0.78*** 0.14 -1.42*** -2.33*** -2.61*** -1.68*** -0.74*** -0.04
(0.172) (0.381) (0.291) (0.271) (0.154) (0.161) (0.200) (0.229) (0.206) (0.128) (0.129) (0.123)

Observations 2,207 2,207 2,207 2,207 2,207 2,207 1,714 4,187 4,187 4,187 4,187 4,187
R-squared 0.203 0.194
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VER_Z (verbal score in grade 9)

266
Table I.7. Heterogeneous effects of one year of English tutoring in middle school on English achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of English PT 0.05 0.08 0.01 0.07 0.06 0.04 0.24 0.01 0.24* 0.35*** 0.46*** 0.33***
(0.050) (0.053) (0.054) (0.073) (0.073) (0.122) (0.156) (0.075) (0.097) (0.072) (0.079) (0.050)
URBAN 0.12 0.02 0.03 0.14 0.17 0.24 0.08 0.05 -0.02 -0.01 -0.07 -0.09*
(0.075) (0.056) (0.074) (0.104) (0.088) (0.130) (0.093) (0.055) (0.056) (0.051) (0.050) (0.038)
GENDER (1=female) 0.34*** 0.16*** 0.21*** 0.39*** 0.36*** 0.44*** 0.34*** 0.19*** 0.44*** 0.37*** 0.25*** 0.12***
(0.049) (0.031) (0.038) (0.062) (0.068) (0.096) (0.059) (0.036) (0.043) (0.038) (0.040) (0.024)
SES 0.22*** 0.03 0.12** 0.19* 0.35*** 0.36*** 0.27*** 0.16*** 0.26*** 0.28*** 0.24*** 0.17***
(0.037) (0.029) (0.039) (0.073) (0.060) (0.064) (0.048) (0.028) (0.027) (0.019) (0.018) (0.023)
SCHOOL TYPE 0.14* 0.02 0.09 0.20 0.12 0.22 0.07 0.06 0.12* 0.11** 0.08* 0.11*
(0.062) (0.055) (0.047) (0.114) (0.121) (0.138) (0.072) (0.033) (0.050) (0.039) (0.031) (0.047)
STU-TEA RATIO -0.00 -0.00 -0.01 -0.01 -0.01 -0.01 -0.00 0.01 0.01 0.01 0.01 0.01*
(0.005) (0.003) (0.004) (0.007) (0.006) (0.007) (0.006) (0.004) (0.005) (0.005) (0.004) (0.005)
SCHOOL CHOICE 0.13 -0.01 0.05 0.18* 0.08 0.05 0.07 0.00 -0.06 -0.05 -0.12 -0.07
(0.074) (0.066) (0.060) (0.084) (0.089) (0.135) (0.093) (0.062) (0.045) (0.060) (0.070) (0.043)
PRESCORE 0.14*** 0.04** 0.08*** 0.15*** 0.18*** 0.18*** 0.16*** 0.08*** 0.15*** 0.21*** 0.17*** 0.11***
(0.015) (0.016) (0.018) (0.026) (0.024) (0.023) (0.017) (0.010) (0.010) (0.012) (0.011) (0.008)
SELF-STUDY 0.03 -0.02 0.07 0.02 -0.03 0.09 -0.08 -0.00 0.03 0.05 -0.04 -0.03
(0.103) (0.099) (0.095) (0.147) (0.184) (0.199) (0.120) (0.084) (0.100) (0.045) (0.033) (0.037)
SELF-ESTEEM 0.13** 0.07 0.10* 0.15* 0.14* 0.13 0.16** 0.13*** 0.24*** 0.25*** 0.23*** 0.16***
(0.044) (0.036) (0.043) (0.067) (0.068) (0.085) (0.051) (0.034) (0.045) (0.036) (0.033) (0.029)
Constant -1.82*** -1.75*** -1.77*** -2.10*** -1.31*** -0.86* -1.83*** -2.02*** -2.38*** -2.09*** -1.06*** -0.09
(0.215) (0.170) (0.181) (0.309) (0.245) (0.368) (0.280) (0.214) (0.160) (0.157) (0.196) (0.151)

Observations 993 993 993 993 993 993 687 4,196 4,196 4,196 4,196 4,196
R-squared 0.231 0.274
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in grade 9)

267
Table I.8. Heterogeneous effects of two years of English tutoring in middle school on English achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of English PT 0.23*** 0.13*** 0.19*** 0.21*** 0.30*** 0.27** 0.11 0.13 2.18* 3.16*** 4.14*** 2.96***
(0.049) (0.028) (0.049) (0.050) (0.082) (0.081) (0.293) (0.717) (0.934) (0.582) (0.700) (0.450)
URBAN 0.06 0.00 0.05 0.07 0.06 0.15 0.10 0.05 -0.13 -0.18** -0.29*** -0.24***
(0.065) (0.040) (0.055) (0.096) (0.079) (0.076) (0.080) (0.074) (0.073) (0.055) (0.053) (0.049)
GENDER (1=female) 0.31*** 0.17*** 0.30*** 0.44*** 0.28*** 0.18* 0.30*** 0.19*** 0.53*** 0.50*** 0.42*** 0.24***
(0.044) (0.051) (0.073) (0.064) (0.077) (0.075) (0.053) (0.055) (0.058) (0.061) (0.050) (0.041)
SES 0.19*** 0.06 0.15*** 0.20*** 0.22*** 0.15*** 0.23*** 0.14 -0.00 -0.10 -0.26** -0.18**
(0.032) (0.054) (0.031) (0.051) (0.045) (0.037) (0.053) (0.102) (0.133) (0.092) (0.092) (0.065)
SCHOOL TYPE 0.11* 0.04 0.12* 0.06 0.14 0.27** 0.08 0.06 0.10* 0.08* 0.04 0.09*
(0.055) (0.046) (0.056) (0.063) (0.095) (0.089) (0.065) (0.033) (0.042) (0.035) (0.041) (0.035)
STU-TEA RATIO 0.00 0.00 0.00 0.01 0.00 -0.01 0.01 0.01 -0.00 -0.01 -0.01* -0.00
(0.005) (0.003) (0.005) (0.005) (0.008) (0.008) (0.006) (0.005) (0.005) (0.005) (0.006) (0.005)
SCHOOL CHOICE 0.04 -0.01 0.04 0.10 0.02 -0.05 0.09 0.00 -0.05 -0.03 -0.09** -0.05
(0.067) (0.040) (0.067) (0.070) (0.101) (0.087) (0.080) (0.049) (0.043) (0.045) (0.030) (0.042)
PRESCORE 0.14*** 0.05** 0.11*** 0.16*** 0.19*** 0.14*** 0.15*** 0.07* 0.09* 0.11*** 0.05 0.02
(0.013) (0.016) (0.018) (0.016) (0.025) (0.018) (0.018) (0.029) (0.036) (0.027) (0.028) (0.018)
SELF-STUDY 0.22** 0.12 0.36 0.37*** 0.14 0.14 0.11 -0.00 0.03 0.04 -0.05 -0.04
(0.082) (0.173) (0.200) (0.082) (0.129) (0.102) (0.097) (0.057) (0.083) (0.045) (0.037) (0.043)
SELF-ESTEEM 0.16*** 0.10* 0.08 0.17* 0.20** 0.20* 0.19*** 0.12** 0.11 0.06 -0.01 -0.01
(0.040) (0.048) (0.059) (0.079) (0.064) (0.086) (0.051) (0.042) (0.080) (0.048) (0.045) (0.037)
Constant -1.79*** -1.86*** -2.19*** -2.29*** -1.39*** -0.18 -1.83*** -2.08*** -3.31*** -3.44*** -2.82*** -1.35***
(0.192) (0.153) (0.236) (0.201) (0.325) (0.244) (0.280) (0.377) (0.469) (0.333) (0.341) (0.209)

Observations 1,377 1,377 1,377 1,377 1,377 1,377 993 4,196 4,196 4,196 4,196 4,196
R-squared 0.274 0.270
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in grade 9)

268
Table I.9. Heterogeneous effects of three years of English tutoring in middle school on English achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of English PT 0.41*** 0.20*** 0.29*** 0.49*** 0.65*** 0.52*** 1.06*** 0.04 0.64* 0.93*** 1.22*** 0.87***
(0.046) (0.046) (0.056) (0.060) (0.046) (0.055) (0.137) (0.216) (0.276) (0.247) (0.207) (0.164)
URBAN -0.07 0.02 -0.06 -0.08 -0.07* -0.09** -0.10* 0.05 -0.01 -0.00 -0.07 -0.08*
(0.043) (0.097) (0.067) (0.066) (0.034) (0.027) (0.051) (0.060) (0.068) (0.054) (0.035) (0.035)
GENDER (1=female) 0.35*** 0.26*** 0.52*** 0.40*** 0.22*** 0.14*** 0.37*** 0.19*** 0.46*** 0.39*** 0.28*** 0.14***
(0.030) (0.067) (0.043) (0.036) (0.039) (0.030) (0.035) (0.046) (0.036) (0.037) (0.039) (0.029)
SES 0.21*** 0.13*** 0.24*** 0.24*** 0.22*** 0.14*** 0.17*** 0.16*** 0.23*** 0.23*** 0.18*** 0.13***
(0.020) (0.033) (0.042) (0.033) (0.019) (0.014) (0.027) (0.030) (0.040) (0.028) (0.026) (0.024)
SCHOOL TYPE 0.07 0.03 0.09 0.09 0.06 0.05 0.08 0.06 0.11** 0.09* 0.06* 0.10**
(0.038) (0.044) (0.053) (0.071) (0.043) (0.039) (0.044) (0.034) (0.036) (0.042) (0.030) (0.038)
STU-TEA RATIO 0.00 -0.00 0.00 0.01 0.00 0.01 0.00 0.01 0.00 0.00 0.00 0.01
(0.004) (0.006) (0.005) (0.005) (0.002) (0.004) (0.005) (0.005) (0.004) (0.004) (0.004) (0.004)
SCHOOL CHOICE -0.11* -0.07 -0.15* -0.08 -0.11** -0.09** -0.10 0.00 -0.05 -0.03 -0.10* -0.06
(0.047) (0.095) (0.072) (0.070) (0.035) (0.029) (0.055) (0.066) (0.062) (0.060) (0.047) (0.043)
PRESCORE 0.16*** 0.09*** 0.17*** 0.20*** 0.15*** 0.09*** 0.14*** 0.08*** 0.14*** 0.19*** 0.15*** 0.10***
(0.009) (0.016) (0.017) (0.019) (0.013) (0.011) (0.011) (0.013) (0.014) (0.012) (0.010) (0.011)
SELF-STUDY -0.00 -0.08 0.00 -0.01 -0.04 -0.03 -0.05 -0.00 0.03 0.04 -0.05 -0.04
(0.043) (0.062) (0.083) (0.047) (0.040) (0.040) (0.049) (0.077) (0.075) (0.056) (0.032) (0.040)
SELF-ESTEEM 0.25*** 0.15*** 0.31*** 0.28*** 0.18*** 0.15*** 0.21*** 0.13*** 0.22*** 0.21*** 0.19*** 0.13***
(0.027) (0.031) (0.046) (0.043) (0.033) (0.029) (0.032) (0.028) (0.035) (0.051) (0.049) (0.028)
Constant -1.58*** -2.03*** -2.40*** -1.97*** -0.93*** 0.09 -2.11*** -2.04*** -2.61*** -2.42*** -1.49*** -0.39*
(0.141) (0.279) (0.232) (0.210) (0.096) (0.111) (0.186) (0.281) (0.300) (0.235) (0.190) (0.161)

Observations 3,142 3,142 3,142 3,142 3,142 3,142 2,492 4,196 4,196 4,196 4,196 4,196
R-squared 0.310 0.252
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in grade 9)

269
Table I.10. Heterogeneous effects of one year of math tutoring in middle school on math achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of Math PT 0.05 -0.04 0.01 0.03 0.11 0.29*** 0.21 0.01 0.09 0.26*** 0.33*** 0.22***
(0.048) (0.036) (0.042) (0.049) (0.069) (0.070) (0.128) (0.055) (0.073) (0.069) (0.076) (0.062)
URBAN -0.18* -0.11* -0.00 -0.14 -0.36** -0.20 -0.19* 0.03 -0.08 -0.24*** -0.11** -0.03
(0.074) (0.045) (0.109) (0.093) (0.128) (0.208) (0.089) (0.031) (0.055) (0.035) (0.035) (0.052)
GENDER (1=female) -0.02 0.02 0.04 -0.01 0.02 -0.08 -0.05 0.04 0.04 0.02 -0.10** -0.13***
(0.048) (0.037) (0.047) (0.063) (0.072) (0.077) (0.056) (0.039) (0.040) (0.038) (0.034) (0.039)
SES 0.16*** 0.04 0.05 0.20*** 0.18*** 0.18*** 0.14** 0.09** 0.21*** 0.25*** 0.22*** 0.16***
(0.035) (0.026) (0.041) (0.043) (0.046) (0.053) (0.042) (0.030) (0.030) (0.026) (0.028) (0.028)
SCHOOL TYPE 0.07 -0.00 0.00 0.15* 0.04 -0.02 0.10 0.10 0.17** 0.18*** 0.15*** 0.14**
(0.061) (0.045) (0.077) (0.068) (0.139) (0.101) (0.070) (0.055) (0.057) (0.050) (0.031) (0.047)
STU-TEA RATIO 0.00 0.01 0.00 0.00 -0.00 0.00 -0.00 0.01** 0.00 0.01 0.01* 0.01
(0.005) (0.004) (0.005) (0.005) (0.007) (0.010) (0.006) (0.003) (0.004) (0.005) (0.006) (0.005)
SCHOOL CHOICE -0.02 0.08 0.12 -0.03 -0.17 -0.12 -0.03 0.06 -0.04 -0.12* 0.01 0.11*
(0.072) (0.048) (0.070) (0.075) (0.134) (0.182) (0.088) (0.044) (0.061) (0.057) (0.033) (0.051)
PRESCORE 0.18*** 0.09*** 0.12*** 0.14*** 0.24*** 0.29*** 0.18*** 0.08*** 0.15*** 0.22*** 0.22*** 0.16***
(0.014) (0.017) (0.012) (0.016) (0.017) (0.024) (0.016) (0.008) (0.009) (0.013) (0.013) (0.015)
SELF-STUDY 0.17 -0.12 0.28 0.21** 0.21 -0.02 0.20 -0.02 0.09 0.21** 0.09 0.07
(0.095) (0.078) (0.147) (0.076) (0.182) (0.137) (0.111) (0.065) (0.079) (0.066) (0.050) (0.055)
SELF-ESTEEM 0.07 0.07 0.05 0.09* 0.04 0.04 0.12* 0.03 0.12* 0.16*** 0.12* 0.14**
(0.044) (0.051) (0.074) (0.045) (0.091) (0.066) (0.051) (0.038) (0.051) (0.047) (0.051) (0.045)
Constant -1.28*** -2.00*** -1.94*** -1.26*** -0.88* -0.69 -1.24*** -1.96*** -1.67*** -1.50*** -1.11*** -0.08
(0.206) (0.162) (0.210) (0.249) (0.403) (0.481) (0.261) (0.121) (0.161) (0.211) (0.173) (0.162)

Observations 992 992 992 992 992 992 693 4,260 4,260 4,260 4,260 4,260
R-squared 0.242 0.272
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9MAT_Z (math score in grade 9)

270
Table I.11. Heterogeneous effects of two years of math tutoring in middle school on math achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of Math PT 0.29*** 0.08* 0.14* 0.29*** 0.46*** 0.50*** 0.01 0.08 1.16 3.37*** 4.22*** 2.83***
(0.049) (0.039) (0.061) (0.055) (0.074) (0.089) (0.307) (0.746) (0.931) (0.912) (0.866) (0.790)
URBAN -0.19** -0.00 -0.02 -0.14 -0.31** -0.24* -0.24** 0.02 -0.14 -0.40*** -0.32*** -0.17*
(0.067) (0.050) (0.077) (0.133) (0.113) (0.119) (0.081) (0.059) (0.079) (0.055) (0.061) (0.069)
GENDER (1=female) 0.00 0.05 -0.00 -0.01 0.05 -0.07 -0.02 0.05 0.07 0.12* 0.02 -0.05
(0.044) (0.058) (0.062) (0.063) (0.072) (0.080) (0.052) (0.059) (0.070) (0.049) (0.043) (0.044)
SES 0.15*** 0.05 0.06 0.16** 0.20*** 0.22*** 0.21*** 0.08 0.08 -0.12 -0.24* -0.14
(0.031) (0.034) (0.046) (0.052) (0.054) (0.063) (0.051) (0.092) (0.114) (0.110) (0.098) (0.086)
SCHOOL TYPE 0.15** 0.07 0.11* 0.11 0.20 0.15 0.18** 0.10 0.12 0.04 -0.03 0.02
(0.055) (0.048) (0.048) (0.097) (0.104) (0.111) (0.066) (0.055) (0.069) (0.058) (0.050) (0.060)
STU-TEA RATIO 0.00 0.00 0.00 0.01 -0.00 -0.00 0.01 0.01 -0.00 -0.01 -0.01 -0.01
(0.005) (0.004) (0.006) (0.005) (0.010) (0.010) (0.006) (0.005) (0.007) (0.005) (0.007) (0.007)
SCHOOL CHOICE -0.05 0.06 -0.01 -0.00 -0.07 -0.09 -0.10 0.06 -0.03 -0.08 0.06 0.15**
(0.067) (0.061) (0.086) (0.127) (0.123) (0.114) (0.080) (0.040) (0.059) (0.072) (0.048) (0.055)
PRESCORE 0.16*** 0.07*** 0.11*** 0.17*** 0.20*** 0.21*** 0.18*** 0.08*** 0.12*** 0.14*** 0.13*** 0.09***
(0.013) (0.020) (0.019) (0.024) (0.019) (0.020) (0.017) (0.021) (0.023) (0.029) (0.029) (0.018)
SELF-STUDY 0.08 -0.09 0.00 0.22 0.14 0.00 0.09 -0.02 0.12 0.30*** 0.20** 0.14**
(0.080) (0.082) (0.180) (0.128) (0.079) (0.118) (0.094) (0.056) (0.076) (0.051) (0.064) (0.049)
SELF-ESTEEM 0.13** 0.05 0.13* 0.17** 0.18* 0.10 0.15** 0.02 0.06 -0.03 -0.11 -0.01
(0.040) (0.047) (0.062) (0.057) (0.075) (0.072) (0.051) (0.059) (0.067) (0.061) (0.058) (0.050)
Constant -1.22*** -1.90*** -1.59*** -1.48*** -0.83* -0.21 -1.00*** -2.00*** -2.20*** -3.03*** -3.02*** -1.36*
(0.193) (0.162) (0.251) (0.351) (0.378) (0.248) (0.283) (0.426) (0.549) (0.547) (0.446) (0.532)

Observations 1,423 1,423 1,423 1,423 1,423 1,423 1,031 4,260 4,260 4,260 4,260 4,260
R-squared 0.244 0.245
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9MAT_Z (math score in grade 9)

271
Table I.12. Heterogeneous effects of three years of math tutoring in middle school on math achievement in grade 9
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of Math PT 0.46*** 0.14** 0.29*** 0.48*** 0.74*** 0.61*** 0.88*** 0.02 0.31 0.90** 1.13*** 0.76***
(0.048) (0.052) (0.061) (0.058) (0.088) (0.111) (0.136) (0.218) (0.280) (0.280) (0.226) (0.204)
URBAN -0.14** -0.05 -0.15** -0.21*** -0.14* -0.03 -0.17*** 0.02 -0.08 -0.24*** -0.11** -0.03
(0.044) (0.041) (0.052) (0.052) (0.055) (0.044) (0.051) (0.042) (0.049) (0.063) (0.042) (0.039)
GENDER (1=female) 0.01 0.07 0.07 0.00 -0.06 -0.12*** 0.01 0.04 0.05 0.04 -0.07* -0.11***
(0.031) (0.046) (0.038) (0.035) (0.031) (0.031) (0.036) (0.032) (0.042) (0.040) (0.033) (0.023)
SES 0.17*** 0.13*** 0.15*** 0.21*** 0.18*** 0.10*** 0.17*** 0.09** 0.20*** 0.21*** 0.17*** 0.13***
(0.021) (0.037) (0.035) (0.026) (0.021) (0.019) (0.026) (0.030) (0.038) (0.038) (0.030) (0.030)
SCHOOL TYPE 0.15*** 0.11 0.10 0.16* 0.12* 0.10* 0.14** 0.10** 0.17*** 0.18*** 0.14** 0.14*
(0.039) (0.088) (0.071) (0.072) (0.048) (0.039) (0.045) (0.040) (0.036) (0.032) (0.049) (0.064)
STU-TEA RATIO 0.01 0.00 0.01 0.01 0.00 0.01 0.00 0.01* 0.00 0.00 0.01 0.00
(0.004) (0.004) (0.005) (0.004) (0.005) (0.006) (0.005) (0.004) (0.006) (0.004) (0.005) (0.004)
SCHOOL CHOICE -0.06 -0.04 -0.07 -0.07 -0.07 0.08 -0.03 0.06 -0.04 -0.11 0.03 0.12
(0.048) (0.048) (0.051) (0.071) (0.073) (0.052) (0.056) (0.046) (0.056) (0.060) (0.052) (0.071)
PRESCORE 0.18*** 0.11*** 0.16*** 0.22*** 0.18*** 0.14*** 0.16*** 0.08*** 0.14*** 0.20*** 0.20*** 0.14***
(0.010) (0.012) (0.013) (0.011) (0.013) (0.011) (0.011) (0.012) (0.014) (0.014) (0.015) (0.012)
SELF-STUDY 0.04 -0.08 0.07 0.14* 0.04 0.03 0.03 -0.02 0.08 0.18** 0.05 0.04
(0.044) (0.082) (0.129) (0.066) (0.035) (0.046) (0.049) (0.067) (0.086) (0.058) (0.044) (0.040)
SELF-ESTEEM 0.11*** 0.00 0.12* 0.13** 0.15*** 0.14*** 0.09** 0.03 0.11* 0.13*** 0.08* 0.12***
(0.028) (0.049) (0.056) (0.047) (0.040) (0.039) (0.032) (0.033) (0.046) (0.038) (0.037) (0.033)
Constant -1.34*** -1.93*** -1.89*** -1.58*** -0.78*** -0.19 -1.58*** -1.97*** -1.77*** -1.81*** -1.49*** -0.33
(0.146) (0.126) (0.191) (0.152) (0.203) (0.217) (0.188) (0.163) (0.235) (0.307) (0.221) (0.239)

Observations 3,198 3,198 3,198 3,198 3,198 3,198 2,531 4,260 4,260 4,260 4,260 4,260
R-squared 0.254 0.215
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9MAT_Z (math score in grade 9)

272
Table J.1. Heterogeneous effects of private tutoring in grade 7 on academic achievement for 3 years using the OLS estimation

Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in G7 0.22*** 0.10** 0.18*** 0.25*** 0.27*** 0.27*** 0.09** 0.02 0.02 0.08* 0.15*** 0.14*** 0.05 0.02 0.01 0.06 0.09 0.11
(0.022) (0.035) (0.035) (0.028) (0.026) (0.032) (0.026) (0.035) (0.027) (0.034) (0.035) (0.038) (0.033) (0.033) (0.045) (0.049) (0.059) (0.069)
URBANICITY 0.07** 0.08 0.08 0.04 0.04 0.07* 0.01 0.06 0.04 0.04 -0.02 -0.01 -0.05 -0.03 -0.04 -0.10 -0.03 -0.05
(0.025) (0.043) (0.046) (0.025) (0.029) (0.030) (0.028) (0.046) (0.040) (0.046) (0.046) (0.051) (0.034) (0.059) (0.054) (0.054) (0.043) (0.044)
GENDER (1=female) 0.25*** 0.34*** 0.32*** 0.27*** 0.19*** 0.13*** 0.29*** 0.39*** 0.38*** 0.33*** 0.21*** 0.13*** 0.28*** 0.26*** 0.38*** 0.33*** 0.20*** 0.11***
(0.017) (0.026) (0.020) (0.022) (0.017) (0.023) (0.019) (0.029) (0.030) (0.027) (0.019) (0.021) (0.023) (0.027) (0.034) (0.021) (0.034) (0.028)
SES 0.19*** 0.18*** 0.20*** 0.20*** 0.18*** 0.14*** 0.16*** 0.11*** 0.18*** 0.20*** 0.20*** 0.14*** 0.15*** 0.08** 0.14*** 0.18*** 0.16*** 0.15***
(0.011) (0.019) (0.019) (0.014) (0.013) (0.019) (0.013) (0.026) (0.028) (0.015) (0.011) (0.015) (0.016) (0.026) (0.027) (0.020) (0.019) (0.021)
SCHOOL TYPE (1=private) 0.06** 0.08* 0.04 0.08*** 0.07*** 0.04 0.06* 0.08 0.07 0.04 0.02 0.01 0.09** 0.08 0.07 0.11** 0.04 0.07
(0.021) (0.031) (0.039) (0.021) (0.020) (0.039) (0.024) (0.052) (0.044) (0.031) (0.038) (0.053) (0.029) (0.047) (0.057) (0.041) (0.026) (0.048)
STU-TEA RATIO 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01** 0.01*** 0.01** 0.01** 0.01** 0.01*** 0.01* 0.01* 0.01* 0.01 0.01 0.01 0.00
(0.002) (0.003) (0.003) (0.003) (0.003) (0.004) (0.002) (0.002) (0.003) (0.004) (0.003) (0.004) (0.003) (0.003) (0.004) (0.004) (0.004) (0.004)
SCHOOL CHOICE -0.07** -0.05 -0.09* -0.10** -0.10** -0.01 -0.09** -0.05 -0.07 -0.07 -0.12* -0.11* -0.02 -0.06 -0.00 -0.05 0.01 -0.03
(0.026) (0.041) (0.043) (0.030) (0.032) (0.044) (0.029) (0.056) (0.052) (0.062) (0.054) (0.047) (0.036) (0.056) (0.048) (0.053) (0.039) (0.044)
PRESCORE (G6) 0.20*** 0.18*** 0.21*** 0.22*** 0.20*** 0.16*** 0.17*** 0.11*** 0.17*** 0.20*** 0.19*** 0.16*** 0.16*** 0.09*** 0.15*** 0.19*** 0.18*** 0.14***
(0.005) (0.007) (0.008) (0.007) (0.007) (0.009) (0.006) (0.011) (0.009) (0.009) (0.010) (0.010) (0.007) (0.010) (0.010) (0.011) (0.011) (0.012)
STUDY HOUR (1=
>10hrs/week) 0.10*** 0.15*** 0.11** 0.11*** 0.08*** 0.06 0.09** 0.11 0.15** 0.08** 0.06* 0.09*** 0.07 -0.04 0.10 0.14* 0.09 0.10*
(0.025) (0.036) (0.037) (0.021) (0.019) (0.029) (0.028) (0.063) (0.046) (0.028) (0.027) (0.026) (0.035) (0.080) (0.054) (0.058) (0.046) (0.044)
SELF ESTEEM 0.19*** 0.18*** 0.21*** 0.21*** 0.18*** 0.19*** 0.18*** 0.14*** 0.17*** 0.18*** 0.19*** 0.17*** 0.12*** 0.05* 0.10*** 0.12*** 0.13*** 0.11**
(0.015) (0.026) (0.030) (0.018) (0.015) (0.018) (0.017) (0.024) (0.031) (0.025) (0.023) (0.031) (0.021) (0.024) (0.024) (0.027) (0.031) (0.040)
PT in G8 0.24*** 0.18*** 0.22*** 0.23*** 0.25*** 0.28*** 0.18*** 0.08* 0.17*** 0.18*** 0.20*** 0.22***
(0.027) (0.034) (0.038) (0.030) (0.039) (0.042) (0.035) (0.035) (0.045) (0.052) (0.056) (0.053)
PT in G9 0.23*** 0.21*** 0.20*** 0.25*** 0.28*** 0.22***
(0.033) (0.025) (0.030) (0.048) (0.056) (0.053)
- - - - - - - - - - - - - - - - - -
Constant 1.92*** 2.77*** 2.45*** 1.99*** 1.38*** 0.80*** 1.77*** 2.39*** 2.34*** 2.01*** 1.28*** 0.63*** 1.78*** 2.16*** 2.37*** 1.95*** 1.35*** 0.49***
(0.078) (0.127) (0.118) (0.086) (0.100) (0.137) (0.088) (0.138) (0.138) (0.200) (0.135) (0.135) (0.106) (0.128) (0.126) (0.169) (0.165) (0.141)

Observations 5,895 5,895 5,895 5,895 5,895 5,895 5,380 5,380 5,380 5,380 5,380 5,380 3,990 3,990 3,990 3,990 3,990 3,990
R-squared 0.444 0.364 0.315
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7VEM_Z (average score in grade 7) for (1)-(6); G8VEM_Z (average score in grade 8) for (7)-(12); G9VEM_Z (average score in grade 9) for (13)-(18)

273
Table J.2. Heterogeneous effects of private tutoring in grade 7 on overall achievement for 3 years using the 2SLAD estimation
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in G7 (predicted) 0.53*** 0.32* 0.77*** 0.75*** 0.82*** 0.93*** 0.33*** 0.10 0.30 0.61*** 0.73*** 0.82*** 0.22* 0.19 0.10 0.23 0.51** 0.52**
(0.066) (0.150) (0.171) (0.140) (0.109) (0.129) (0.076) (0.132) (0.158) (0.183) (0.114) (0.138) (0.094) (0.148) (0.058) (0.149) (0.179) (0.189)
URBANICITY 0.04 0.06 0.04 -0.02 0.00 0.05* -0.01 0.05 -0.02 -0.04 -0.05 -0.06 -0.06 -0.04 -0.04 -0.09** -0.05 -0.07
(0.026) (0.062) (0.037) (0.032) (0.027) (0.027) (0.029) (0.070) (0.049) (0.048) (0.032) (0.037) (0.035) (0.046) (0.056) (0.035) (0.038) (0.049)
GENDER (1=female) 0.26*** 0.35*** 0.32*** 0.27*** 0.20*** 0.13*** 0.29*** 0.39*** 0.38*** 0.33*** 0.23*** 0.14*** 0.28*** 0.27*** 0.36*** 0.32*** 0.22*** 0.11**
(0.018) (0.033) (0.033) (0.026) (0.025) (0.029) (0.020) (0.026) (0.025) (0.025) (0.022) (0.020) (0.024) (0.030) (0.029) (0.027) (0.026) (0.039)
SES 0.16*** 0.18*** 0.16*** 0.16*** 0.14*** 0.08*** 0.16*** 0.11*** 0.19*** 0.19*** 0.17*** 0.10*** 0.14*** 0.07** 0.14*** 0.17*** 0.14*** 0.12***
(0.013) (0.028) (0.022) (0.014) (0.013) (0.018) (0.014) (0.029) (0.027) (0.018) (0.016) (0.024) (0.017) (0.026) (0.029) (0.024) (0.024) (0.021)
SCHOOL TYPE (1=private) 0.04 0.06* 0.03 0.02 0.05 0.03 0.05 0.08** 0.06 0.03 0.01 -0.02 0.07* 0.07* 0.06 0.12** 0.06 0.07
(0.022) (0.024) (0.022) (0.023) (0.024) (0.027) (0.025) (0.024) (0.043) (0.024) (0.026) (0.046) (0.030) (0.035) (0.060) (0.039) (0.042) (0.047)
STU-TEA RATIO 0.01*** 0.01** 0.01** 0.01*** 0.01*** 0.01*** 0.01*** 0.00 0.01 0.01** 0.01*** 0.01* 0.01* 0.00 0.01* 0.00 0.01 0.00
(0.002) (0.004) (0.003) (0.002) (0.003) (0.004) (0.002) (0.003) (0.004) (0.003) (0.002) (0.003) (0.003) (0.003) (0.002) (0.004) (0.003) (0.004)
SCHOOL CHOICE -0.07** -0.05 -0.07 -0.12** -0.10** 0.03 -0.10*** -0.08 -0.12* -0.13** -0.11*** -0.11** -0.02 -0.04 -0.01 -0.05 0.01 -0.01
(0.028) (0.064) (0.043) (0.040) (0.034) (0.034) (0.031) (0.058) (0.050) (0.044) (0.029) (0.035) (0.037) (0.037) (0.044) (0.056) (0.036) (0.053)
PRESCORE (G6) 0.19*** 0.17*** 0.19*** 0.21*** 0.19*** 0.15*** 0.16*** 0.10*** 0.16*** 0.19*** 0.18*** 0.16*** 0.16*** 0.09*** 0.14*** 0.19*** 0.17*** 0.14***
(0.006) (0.012) (0.012) (0.011) (0.008) (0.009) (0.006) (0.009) (0.008) (0.008) (0.007) (0.008) (0.007) (0.010) (0.012) (0.008) (0.008) (0.010)
STUDY HOUR (1=>10hrs/week) 0.09*** 0.15*** 0.11*** 0.09*** 0.04 0.02 0.08** 0.11*** 0.12* 0.05 0.05 0.06* 0.05 -0.06 0.08 0.13** 0.08* 0.09
(0.026) (0.044) (0.028) (0.026) (0.032) (0.036) (0.029) (0.032) (0.052) (0.050) (0.041) (0.028) (0.036) (0.088) (0.074) (0.049) (0.038) (0.064)
SELF ESTEEM 0.19*** 0.18*** 0.19*** 0.21*** 0.15*** 0.14*** 0.18*** 0.15*** 0.19*** 0.17*** 0.16*** 0.14*** 0.12*** 0.05 0.13** 0.15*** 0.14*** 0.11***
(0.016) (0.031) (0.032) (0.024) (0.022) (0.024) (0.018) (0.024) (0.037) (0.029) (0.023) (0.024) (0.022) (0.042) (0.045) (0.035) (0.024) (0.028)
PT_G8 0.15*** 0.16** 0.11 0.00 0.02 0.01 0.12** 0.02 0.13** 0.12* 0.03 0.07
(0.040) (0.058) (0.079) (0.073) (0.071) (0.079) (0.045) (0.052) (0.046) (0.053) (0.047) (0.061)
PT_G9 0.20*** 0.16*** 0.17*** 0.22*** 0.21*** 0.18**
(0.038) (0.043) (0.045) (0.056) (0.041) (0.064)
- - - - - - - - - - - - - - - -
Constant 2.06*** 2.83*** 2.71*** 2.21*** 1.68*** 1.29*** -1.83*** 2.31*** 2.26*** 2.00*** -1.53*** 0.91*** 1.85*** 2.19*** 2.28*** 1.97*** 1.52*** 0.71***
(0.090) (0.180) (0.157) (0.150) (0.118) (0.127) (0.096) (0.130) (0.134) (0.111) (0.065) (0.096) (0.115) (0.103) (0.087) (0.177) (0.111) (0.153)

Observations 5,473 5,882 5,882 5,882 5,882 5,882 5,022 5,377 5,377 5,377 5,377 5,377 3,716 3,990 3,990 3,990 3,990 3,990
R-squared 0.419 0.351 0.307
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7VEM_Z (average score in grade 7) for (1)-(6); G8VEM_Z (average score in grade 8) for (7)-(12); G9VEM_Z (average score in grade 9) for (13)-(18)

274
Table J.3. Heterogeneous effects of private tutoring in grade 8 on overall achievement for 2 years using the OLS estimation
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in G8 0.12*** 0.07*** 0.11*** 0.10*** 0.11*** 0.17*** 0.09** 0.03 0.07* 0.10*** 0.07* 0.16***
(0.020) (0.018) (0.024) (0.022) (0.024) (0.025) (0.028) (0.036) (0.030) (0.023) (0.029) (0.031)
URBANICITY -0.04 -0.06 -0.05 -0.02 -0.04 -0.07* -0.09** -0.07 -0.11*** -0.06* -0.05 -0.07*
(0.021) (0.051) (0.024) (0.028) (0.031) (0.026) (0.028) (0.058) (0.033) (0.028) (0.028) (0.027)
GENDER (1=female) 0.10*** 0.21*** 0.16*** 0.08*** 0.03* 0.03* 0.09*** 0.15*** 0.14*** 0.08*** 0.04* -0.01
(0.014) (0.023) (0.020) (0.017) (0.015) (0.015) (0.019) (0.039) (0.029) (0.020) (0.018) (0.015)
SES 0.04*** 0.01 0.04** 0.05*** 0.05*** 0.04*** 0.05*** 0.04 0.06** 0.04** 0.05* 0.04*
(0.010) (0.022) (0.016) (0.012) (0.011) (0.009) (0.013) (0.029) (0.023) (0.013) (0.019) (0.018)
SCHOOL TYPE (1=private) 0.03 0.06 0.02 0.00 0.02 0.03 0.06* 0.08 0.06* 0.03 0.06** 0.05
(0.018) (0.036) (0.019) (0.020) (0.022) (0.023) (0.023) (0.049) (0.029) (0.034) (0.020) (0.035)
STU-TEA RATIO 0.00 0.00 0.00 0.00 0.00 0.00 -0.00 -0.00 -0.00 -0.00 -0.00 -0.00
(0.002) (0.004) (0.002) (0.002) (0.002) (0.003) (0.002) (0.005) (0.003) (0.002) (0.002) (0.003)
SCHOOL CHOICE -0.05* -0.06 -0.04 -0.01 -0.05 -0.09*** 0.01 -0.04 -0.01 0.05 0.05 0.02
(0.022) (0.057) (0.030) (0.031) (0.030) (0.027) (0.029) (0.040) (0.035) (0.026) (0.034) (0.034)
PRESCORE (G7) 0.73*** 0.67*** 0.78*** 0.80*** 0.77*** 0.66*** 0.68*** 0.47*** 0.72*** 0.80*** 0.75*** 0.62***
(0.010) (0.031) (0.021) (0.012) (0.014) (0.019) (0.013) (0.044) (0.014) (0.012) (0.019) (0.022)
STUDY HOUR (1= >10hrs/week) 0.03 0.04 0.05* 0.02 0.02 0.02 0.01 -0.07 0.03 0.06 0.06* 0.05
(0.021) (0.058) (0.022) (0.026) (0.033) (0.031) (0.029) (0.063) (0.055) (0.039) (0.023) (0.031)
SELF ESTEEM 0.07*** 0.09*** 0.06*** 0.05*** 0.05** 0.06*** 0.03 -0.01 0.01 0.01 0.03 0.04**
(0.013) (0.026) (0.016) (0.012) (0.017) (0.016) (0.017) (0.033) (0.021) (0.019) (0.021) (0.016)
PT in G7 -0.03 -0.05 -0.04* -0.02 -0.01 0.00 -0.06* -0.04 -0.05 -0.00 -0.07** -0.10
(0.020) (0.037) (0.019) (0.027) (0.036) (0.028) (0.027) (0.047) (0.031) (0.027) (0.024) (0.052)
PT in G9 0.16*** 0.18*** 0.13** 0.14*** 0.18*** 0.15***
(0.027) (0.046) (0.042) (0.028) (0.043) (0.037)
- - -
Constant -0.18** 0.89*** 0.55*** -0.15* 0.27*** 0.59*** -0.25** 1.07*** -0.59*** -0.22** 0.12 0.59***
(0.062) (0.127) (0.080) (0.073) (0.082) (0.102) (0.082) (0.159) (0.125) (0.071) (0.081) (0.113)

Observations 5,501 5,501 5,501 5,501 5,501 5,501 4,077 4,077 4,077 4,077 4,077 4,077
R-squared 0.633 0.540
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8VEM_Z (average score in grade 8) for (1)-(6); G9VEM_Z (average score in grade 9) for (7)-(12)

275
Table J.4. Heterogeneous effects of private tutoring in grade 8 on overall achievement for 2 years using the 2SLAD estimation
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

PT in G8 (predicted) 0.27*** 0.26 0.32** 0.29*** 0.24* 0.14 0.13 0.05 0.07 0.11 0.18 0.21
(0.066) (0.143) (0.099) (0.069) (0.101) (0.079) (0.097) (0.176) (0.146) (0.134) (0.143) (0.207)
URBANICITY -0.05* -0.06 -0.06* -0.03 -0.06* -0.04 -0.09** -0.07* -0.10* -0.07* -0.05 -0.06
(0.022) (0.036) (0.025) (0.022) (0.024) (0.033) (0.029) (0.034) (0.042) (0.027) (0.040) (0.035)
GENDER (1=female) 0.11*** 0.23*** 0.17*** 0.08*** 0.04* 0.02 0.09*** 0.15*** 0.15*** 0.07*** 0.04* -0.00
(0.015) (0.028) (0.018) (0.017) (0.018) (0.021) (0.020) (0.036) (0.030) (0.016) (0.017) (0.028)
SES 0.04*** 0.02 0.05** 0.05*** 0.04* 0.04*** 0.06*** 0.06 0.08*** 0.05*** 0.05** 0.04*
(0.011) (0.025) (0.016) (0.011) (0.016) (0.012) (0.014) (0.035) (0.020) (0.016) (0.016) (0.019)
SCHOOL TYPE (1=private) 0.03 0.03 0.03 0.02 0.01 0.02 0.07** 0.10 0.07 0.06 0.06* 0.04
(0.019) (0.041) (0.026) (0.021) (0.017) (0.023) (0.025) (0.061) (0.043) (0.044) (0.024) (0.039)
STU-TEA RATIO 0.00 0.00 0.00 0.00 0.00 0.00 -0.00 -0.00 -0.00 -0.00 -0.00 -0.00
(0.002) (0.004) (0.003) (0.002) (0.002) (0.002) (0.002) (0.004) (0.003) (0.003) (0.003) (0.003)
SCHOOL CHOICE -0.05* -0.06 -0.03 -0.01 -0.07** -0.06 0.01 -0.06 -0.02 0.05 0.05 0.04
(0.023) (0.039) (0.026) (0.027) (0.025) (0.033) (0.031) (0.046) (0.046) (0.041) (0.041) (0.031)
PRESCORE (G7) 0.72*** 0.66*** 0.77*** 0.79*** 0.76*** 0.66*** 0.68*** 0.48*** 0.72*** 0.80*** 0.74*** 0.61***
(0.011) (0.025) (0.018) (0.011) (0.013) (0.018) (0.014) (0.034) (0.024) (0.014) (0.013) (0.024)
STUDY HOUR (1= >10hrs/week) 0.01 0.03 0.04 0.01 0.02 0.01 -0.01 -0.10 -0.01 0.04 0.06 0.06
(0.023) (0.050) (0.027) (0.022) (0.026) (0.027) (0.030) (0.061) (0.037) (0.029) (0.033) (0.048)
SELF ESTEEM 0.06*** 0.07** 0.05*** 0.04*** 0.05** 0.06*** 0.03 -0.00 0.01 0.01 0.03 0.04
(0.013) (0.025) (0.011) (0.011) (0.018) (0.014) (0.018) (0.037) (0.026) (0.017) (0.021) (0.031)
PT in G7 -0.08** -0.11* -0.12*** -0.11*** -0.04 0.02 -0.08* -0.07 -0.06 -0.02 -0.09 -0.12
(0.031) (0.055) (0.032) (0.029) (0.042) (0.035) (0.039) (0.067) (0.054) (0.062) (0.057) (0.086)
PT in G9 0.15*** 0.17* 0.12 0.12 0.14* 0.15
(0.040) (0.075) (0.074) (0.066) (0.068) (0.092)
Constant -0.25*** -1.02*** -0.66*** -0.24* 0.22 0.54*** -0.24** -0.99*** -0.59*** -0.20* 0.08 0.50***
(0.073) (0.164) (0.119) (0.095) (0.111) (0.138) (0.092) (0.133) (0.102) (0.083) (0.086) (0.116)

Observations 4,953 5,000 5,000 5,000 5,000 5,000 3,657 3,696 3,696 3,696 3,696 3,696
R-squared 0.623 0.530
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8VEM_Z (average score in grade 8) for (1)-(6); G9VEM_Z (average score in grade 9) for (7)-(12)

276
 277

Table J.5. Heterogeneous effects of private tutoring in grade 9 on overall achievement in grade 9
using the OLS estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Avg q10 q25 q50 q75 q90

PT in G9 0.16*** 0.13*** 0.11*** 0.16*** 0.17*** 0.13***

(0.023) (0.031) (0.027) (0.026) (0.023) (0.034)
URBANICITY -0.07** -0.13** -0.06 -0.06* -0.02 -0.03
(0.025) (0.047) (0.034) (0.025) (0.023) (0.025)
GENDER (1=female) 0.06** 0.08** 0.08** 0.03 0.02 -0.01
(0.018) (0.031) (0.025) (0.015) (0.019) (0.018)
SES 0.05*** 0.07* 0.04* 0.03* 0.05*** 0.06***
(0.012) (0.029) (0.016) (0.010) (0.007) (0.013)
SCHOOL TYPE (1=private) 0.06** 0.11*** 0.05 0.04 0.05* 0.08**
(0.021) (0.027) (0.031) (0.019) (0.024) (0.028)
STU-TEA RATIO -0.00 0.00 0.00 -0.00 0.00 -0.00
(0.002) (0.003) (0.003) (0.002) (0.002) (0.003)
SCHOOL CHOICE 0.03 -0.05 0.04 0.03 0.09*** 0.07*
(0.027) (0.049) (0.037) (0.020) (0.028) (0.027)
PRESCORE (G8) 0.70*** 0.56*** 0.78*** 0.81*** 0.73*** 0.62***
(0.012) (0.028) (0.014) (0.010) (0.011) (0.014)
STUDY HOUR (1= >10hrs/week) 0.00 -0.04 0.03 0.02 0.02 0.04
(0.027) (0.069) (0.039) (0.029) (0.038) (0.023)
SELF ESTEEM 0.05*** 0.02 0.04* 0.04*** 0.06*** 0.07***
(0.015) (0.026) (0.019) (0.012) (0.018) (0.018)
PT in G8 0.05* 0.05 0.03 0.03 0.06** 0.09*
(0.024) (0.028) (0.030) (0.030) (0.020) (0.043)
Constant -0.23** -0.90*** -0.60*** -0.14* -0.01 0.50***
(0.074) (0.124) (0.092) (0.063) (0.074) (0.092)

Observations 4,378 4,378 4,378 4,378 4,378 4,378

R-squared 0.585
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in grade 9)
 278

Table J.6. Heterogeneous effects of private tutoring in grade 9 on overall achievement in grade 9
using the 2SLAD estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Avg q10 q25 q50 q75 q90

PT in G9 (predicted) 0.19* -0.04 -0.07 -0.01 0.17 0.34

(0.090) (0.170) (0.153) (0.172) (0.179) (0.176)
URBANICITY -0.07** -0.14** -0.08** -0.04 -0.00 -0.02
(0.028) (0.048) (0.029) (0.023) (0.024) (0.019)
GENDER (1=female) 0.06** 0.06 0.08*** 0.05** 0.02 -0.02
(0.019) (0.034) (0.022) (0.019) (0.015) (0.021)
SES 0.06*** 0.09*** 0.06*** 0.05** 0.04** 0.04**
(0.014) (0.020) (0.016) (0.016) (0.014) (0.017)
SCHOOL TYPE (1=private) 0.07** 0.10** 0.08*** 0.06** 0.07** 0.10***
(0.023) (0.032) (0.022) (0.021) (0.025) (0.028)
STU-TEA RATIO 0.00 0.00 0.00 -0.00 -0.00 -0.00
(0.002) (0.003) (0.003) (0.002) (0.002) (0.003)
SCHOOL CHOICE 0.02 -0.09* -0.00 0.02 0.09*** 0.07*
(0.029) (0.040) (0.026) (0.018) (0.026) (0.032)
PRESCORE (G8) 0.69*** 0.61*** 0.80*** 0.82*** 0.75*** 0.61***
(0.013) (0.034) (0.014) (0.011) (0.015) (0.020)
STUDY HOUR (1= >10hrs/week) 0.00 -0.04 0.00 -0.00 -0.01 0.03
(0.029) (0.059) (0.046) (0.024) (0.031) (0.031)
SELF ESTEEM 0.05** 0.03 0.06*** 0.05** 0.07*** 0.07**
(0.017) (0.024) (0.019) (0.017) (0.016) (0.025)
PT in G8 0.01 0.04 0.07 0.07 0.04 -0.03
(0.046) (0.073) (0.065) (0.071) (0.085) (0.089)
Constant -0.25** -0.65*** -0.45*** -0.08 0.03 0.37***
(0.086) (0.138) (0.082) (0.070) (0.080) (0.102)

Observations 3,761 4,839 4,839 4,839 4,839 4,839

R-squared 0.579
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VEM_Z (average score in grade 9)
Table K.1. Heterogeneous effects of verbal tutoring in grade 7 on verbal achievement for 3 years using the OLS estimation
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

VERBAL PT in G7 -0.01 -0.06 -0.03 -0.02 0.01 -0.00 -0.05 -0.02 -0.04 -0.06 -0.07* -0.05 -0.02 -0.01 -0.03 0.03 -0.00 -0.06*
(0.022) (0.045) (0.032) (0.021) (0.020) (0.024) (0.025) (0.039) (0.044) (0.034) (0.032) (0.039) (0.031) (0.050) (0.063) (0.034) (0.036) (0.029)
URBANICITY 0.05 0.17 0.03 0.03 0.05 0.03 0.02 0.10 0.04 -0.01 -0.01 0.00 -0.04 0.03 -0.02 -0.09 -0.05 -0.04
(0.031) (0.099) (0.032) (0.038) (0.035) (0.043) (0.034) (0.072) (0.047) (0.034) (0.037) (0.018) (0.041) (0.075) (0.092) (0.066) (0.040) (0.033)
GENDER (1=female) 0.45*** 0.70*** 0.54*** 0.44*** 0.31*** 0.23*** 0.47*** 0.65*** 0.64*** 0.49*** 0.36*** 0.29*** 0.47*** 0.44*** 0.71*** 0.56*** 0.34*** 0.21***
(0.021) (0.045) (0.026) (0.027) (0.024) (0.016) (0.023) (0.041) (0.034) (0.026) (0.023) (0.025) (0.028) (0.042) (0.059) (0.047) (0.030) (0.028)
SES 0.14*** 0.16*** 0.15*** 0.15*** 0.12*** 0.08*** 0.10*** 0.10*** 0.10** 0.14*** 0.11*** 0.07** 0.11*** 0.05 0.12** 0.15*** 0.11*** 0.08***
(0.014) (0.032) (0.020) (0.021) (0.017) (0.019) (0.015) (0.026) (0.029) (0.022) (0.019) (0.021) (0.018) (0.029) (0.045) (0.026) (0.018) (0.020)
SCHOOL TYPE (1=private) 0.04 0.04 0.05 0.02 0.03 0.03 0.05 0.09 0.11** 0.03 0.02 -0.03 0.08* 0.05 0.10 0.12** 0.07* 0.03
(0.026) (0.062) (0.042) (0.038) (0.027) (0.033) (0.029) (0.051) (0.039) (0.027) (0.033) (0.033) (0.035) (0.058) (0.073) (0.038) (0.033) (0.038)
STU-TEA RATIO 0.02*** 0.02*** 0.02*** 0.02*** 0.01*** 0.01 0.02*** 0.02** 0.02*** 0.02*** 0.01*** 0.01* 0.01** 0.00 0.01* 0.01** 0.01** 0.00
(0.003) (0.007) (0.006) (0.005) (0.003) (0.003) (0.003) (0.007) (0.005) (0.003) (0.003) (0.004) (0.003) (0.004) (0.006) (0.003) (0.003) (0.002)
SCHOOL CHOICE -0.05 0.08 -0.06 -0.07 -0.02 -0.03 -0.08* -0.02 -0.09 -0.08* -0.08 -0.08** -0.03 -0.05 0.00 -0.07 -0.01 -0.01
(0.033) (0.113) (0.050) (0.043) (0.040) (0.047) (0.036) (0.076) (0.052) (0.041) (0.044) (0.031) (0.043) (0.065) (0.076) (0.056) (0.043) (0.043)
PRESCORE (G6) 0.20*** 0.21*** 0.24*** 0.22*** 0.18*** 0.13*** 0.17*** 0.15*** 0.20*** 0.20*** 0.18*** 0.14*** 0.15*** 0.07*** 0.19*** 0.19*** 0.15*** 0.13***
(0.006) (0.013) (0.013) (0.010) (0.007) (0.006) (0.007) (0.020) (0.013) (0.008) (0.008) (0.011) (0.008) (0.010) (0.014) (0.009) (0.011) (0.009)
STUDY HOUR 0.06* 0.18** 0.07 0.05 -0.01 -0.02 0.05 0.08 0.08 0.02 0.03 -0.03 0.08 0.07 0.02 0.09 0.08** 0.05
(1= >10hrs/week) (0.032) (0.067) (0.064) (0.028) (0.028) (0.046) (0.035) (0.061) (0.054) (0.031) (0.038) (0.041) (0.043) (0.043) (0.083) (0.054) (0.030) (0.031)
SELF ESTEEM 0.19*** 0.18*** 0.21*** 0.18*** 0.17*** 0.17*** 0.17*** 0.20*** 0.16*** 0.16*** 0.17*** 0.16*** 0.10*** 0.06 0.11* 0.13*** 0.10*** 0.05
(0.019) (0.054) (0.042) (0.027) (0.021) (0.032) (0.021) (0.046) (0.034) (0.025) (0.023) (0.019) (0.025) (0.042) (0.047) (0.034) (0.023) (0.033)
VERBAL PT in G8 0.09*** 0.04 0.12* 0.11** 0.06 0.01 0.06 0.06 0.01 0.03 0.04 0.12**
(0.025) (0.054) (0.050) (0.041) (0.040) (0.037) (0.033) (0.054) (0.065) (0.040) (0.033) (0.038)
VERBAL PT in G9 0.01 0.05 0.07 0.01 -0.04 -0.09*
(0.032) (0.049) (0.069) (0.049) (0.040) (0.038)
- - - - - - - - - - - - - - -
Constant 2.12*** 4.00*** 3.13*** 2.08*** 1.15*** -0.21 1.96*** 3.46*** 3.06*** 2.09*** 1.10*** -0.21 1.77*** 2.44*** 3.07*** 1.92*** 0.84*** -0.01
(0.096) (0.286) (0.166) (0.104) (0.100) (0.138) (0.105) (0.277) (0.158) (0.134) (0.131) (0.125) (0.126) (0.198) (0.244) (0.156) (0.125) (0.128)

Observations 5,855 5,855 5,855 5,855 5,855 5,855 5,357 5,357 5,357 5,357 5,357 5,357 4,073 4,073 4,073 4,073 4,073 4,073
R-squared 0.316 0.257 0.198
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7VER_Z (verbal score in grade 7) for (1)-(6); G8VER_Z (verbal score in grade 8) for (7)-(12); G9VER_Z (verbal score in grade 9) for (13)-(18)

279
Table K.2. Heterogeneous effects of verbal tutoring in grade 7 on verbal achievement for 3 years using the 2SLAD estimation
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90 Avg q10 q25 q50 q75 q90

VERBAL PT in G7 (predicted) 0.01 0.04 -0.01 -0.03 0.04 0.01 -0.08 -0.03 -0.06 -0.13* -0.05 0.04 -0.12 -0.02 -0.09 -0.09 0.02 -0.11
(0.056) (0.101) (0.077) (0.088) (0.092) (0.090) (0.071) (0.128) (0.072) (0.063) (0.106) (0.120) (0.087) (0.174) (0.154) (0.113) (0.091) (0.080)
URBANICITY 0.05 0.14 0.02 0.03 0.04 0.01 0.01 0.04 0.01 -0.02 -0.02 -0.01 -0.04 0.04 -0.03 -0.09 -0.03 -0.03
(0.032) (0.090) (0.046) (0.038) (0.032) (0.033) (0.035) (0.067) (0.065) (0.034) (0.039) (0.032) (0.042) (0.090) (0.114) (0.053) (0.038) (0.043)
GENDER (1=female) 0.45*** 0.65*** 0.52*** 0.44*** 0.31*** 0.23*** 0.47*** 0.64*** 0.63*** 0.48*** 0.34*** 0.28*** 0.47*** 0.42*** 0.70*** 0.54*** 0.33*** 0.21***
(0.022) (0.048) (0.024) (0.023) (0.020) (0.029) (0.024) (0.052) (0.034) (0.031) (0.030) (0.034) (0.029) (0.052) (0.053) (0.037) (0.034) (0.033)
SES 0.13*** 0.17*** 0.15*** 0.15*** 0.11*** 0.07*** 0.11*** 0.14*** 0.11*** 0.16*** 0.10*** 0.07*** 0.12*** 0.07* 0.15*** 0.16*** 0.11*** 0.08***
(0.015) (0.040) (0.030) (0.020) (0.015) (0.019) (0.016) (0.028) (0.028) (0.019) (0.020) (0.014) (0.019) (0.032) (0.030) (0.026) (0.021) (0.016)
SCHOOL TYPE (1=private) 0.02 0.03 0.02 0.01 0.00 0.01 0.04 0.09 0.12*** 0.02 0.02 -0.03 0.08* 0.05 0.13* 0.15*** 0.07* 0.03
(0.027) (0.063) (0.051) (0.038) (0.027) (0.025) (0.030) (0.059) (0.035) (0.029) (0.029) (0.026) (0.036) (0.049) (0.060) (0.044) (0.036) (0.030)
STU/TEA RATIO 0.01*** 0.02*** 0.02*** 0.02*** 0.01*** 0.01 0.02*** 0.01** 0.02*** 0.02*** 0.01*** 0.01** 0.01** 0.00 0.01 0.01 0.01** 0.01
(0.003) (0.004) (0.003) (0.003) (0.003) (0.003) (0.003) (0.005) (0.004) (0.003) (0.003) (0.003) (0.004) (0.005) (0.006) (0.004) (0.003) (0.004)
SCHOOL CHOICE -0.05 0.03 -0.06 -0.08 -0.02 -0.05 -0.09* -0.09 -0.14** -0.10* -0.08 -0.10** -0.03 -0.03 -0.03 -0.07 0.02 0.03
(0.034) (0.090) (0.051) (0.049) (0.034) (0.032) (0.037) (0.069) (0.054) (0.040) (0.047) (0.037) (0.045) (0.069) (0.099) (0.052) (0.057) (0.049)
PRESCORE (G6) 0.20*** 0.20*** 0.24*** 0.23*** 0.18*** 0.14*** 0.17*** 0.13*** 0.19*** 0.21*** 0.18*** 0.14*** 0.15*** 0.08*** 0.18*** 0.19*** 0.15*** 0.13***
(0.007) (0.012) (0.007) (0.008) (0.006) (0.008) (0.007) (0.021) (0.017) (0.009) (0.011) (0.008) (0.009) (0.011) (0.016) (0.012) (0.011) (0.010)
STUDY HOUR 0.07* 0.14* 0.09* 0.06 0.01 0.01 0.04 0.05 0.05 0.02 0.01 -0.01 0.07 0.07 0.02 0.09 0.08* 0.09**
(1= >10hrs/week) (0.032) (0.069) (0.047) (0.037) (0.032) (0.035) (0.036) (0.065) (0.051) (0.052) (0.039) (0.042) (0.044) (0.065) (0.087) (0.063) (0.033) (0.027)
SELF ESTEEM 0.19*** 0.20*** 0.20*** 0.19*** 0.16*** 0.14*** 0.17*** 0.22*** 0.17*** 0.15*** 0.16*** 0.13*** 0.11*** 0.07 0.16*** 0.16*** 0.11*** 0.05*
(0.020) (0.035) (0.024) (0.027) (0.022) (0.030) (0.021) (0.033) (0.034) (0.030) (0.020) (0.030) (0.026) (0.041) (0.046) (0.038) (0.025) (0.024)
VERBAL PT in G8 0.09** 0.06 0.14** 0.12*** 0.05 -0.03 0.07 0.08 0.02 0.06 0.02 0.11*
(0.035) (0.092) (0.051) (0.032) (0.044) (0.062) (0.043) (0.065) (0.064) (0.039) (0.033) (0.044)
VERBAL PT in G9 0.03 0.05 0.06 0.04 -0.03 -0.06
(0.035) (0.052) (0.063) (0.041) (0.037) (0.041)
- - - - - - - - - - - - - - -
Constant 2.07*** 3.77*** 3.01*** 2.11*** 1.19*** -0.18* 1.89*** 3.09*** 2.91*** 2.09*** 1.07*** -0.19 1.72*** 2.45*** 2.81*** 1.81*** 0.91*** -0.12
(0.102) (0.197) (0.127) (0.094) (0.100) (0.085) (0.110) (0.203) (0.155) (0.124) (0.159) (0.121) (0.133) (0.220) (0.257) (0.148) (0.153) (0.140)

Observations 5,435 5,825 5,825 5,825 5,825 5,825 5,004 5,346 5,346 5,346 5,346 5,346 3,790 4,058 4,058 4,058 4,058 4,058
R-squared 0.307 0.250 0.192
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7VER_Z (verbal score in grade 7) for (1)-(6); G8VER_Z (verbal score in grade 8) for (7)-(12); G9VER_Z (verbal score in grade 9) for (13)-(18)

280
Table K.3. Heterogeneous effects of verbal tutoring in grade 8 on verbal achievement for 2 years using the OLS estimation
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G8 0.04* 0.00 0.02 0.02 0.04 0.04 0.02 -0.02 -0.05 0.01 0.07* 0.02
(0.021) (0.055) (0.041) (0.027) (0.027) (0.033) (0.028) (0.088) (0.049) (0.035) (0.029) (0.044)
URBANICITY -0.04 0.01 -0.04 -0.06 -0.03 -0.04 -0.09* -0.06 -0.03 -0.10* -0.07 -0.05
(0.028) (0.069) (0.043) (0.042) (0.041) (0.039) (0.035) (0.060) (0.044) (0.048) (0.048) (0.038)
GENDER (1=female) 0.29*** 0.42*** 0.36*** 0.29*** 0.22*** 0.18*** 0.30*** 0.36*** 0.39*** 0.31*** 0.21*** 0.09***
(0.020) (0.058) (0.052) (0.037) (0.028) (0.027) (0.025) (0.070) (0.051) (0.038) (0.027) (0.026)
SES -0.04** -0.04 -0.05 -0.03 -0.02 -0.02 -0.02 -0.02 -0.01 -0.03 -0.00 0.00
(0.013) (0.032) (0.028) (0.020) (0.020) (0.020) (0.016) (0.041) (0.025) (0.020) (0.013) (0.012)
SCHOOL TYPE (1=private) 0.02 0.02 0.06 -0.01 -0.00 -0.00 0.06* 0.05 0.13*** 0.07 0.03 -0.00
(0.024) (0.052) (0.033) (0.024) (0.028) (0.043) (0.030) (0.041) (0.038) (0.036) (0.031) (0.043)
STU-TEA RATIO 0.01** 0.01 0.01* 0.01** 0.00 -0.00 0.00 -0.00 0.00 0.00 0.00 -0.00
(0.002) (0.006) (0.004) (0.003) (0.002) (0.003) (0.003) (0.006) (0.006) (0.003) (0.002) (0.002)
SCHOOL CHOICE -0.03 -0.00 -0.01 -0.02 -0.03 -0.05 0.01 -0.03 0.06 -0.00 0.09 0.07*
(0.030) (0.076) (0.041) (0.039) (0.041) (0.048) (0.037) (0.050) (0.049) (0.044) (0.046) (0.030)
PRESCORE (G7) 0.73*** 0.76*** 0.83*** 0.79*** 0.68*** 0.59*** 0.68*** 0.57*** 0.82*** 0.80*** 0.64*** 0.51***
(0.013) (0.027) (0.019) (0.015) (0.016) (0.024) (0.016) (0.044) (0.017) (0.019) (0.019) (0.021)
STUDY HOUR (1= >10hrs/week) -0.02 0.00 -0.01 -0.02 -0.02 -0.04 0.02 0.07 0.00 0.02 0.04 0.05
(0.029) (0.056) (0.047) (0.030) (0.023) (0.038) (0.037) (0.078) (0.043) (0.046) (0.025) (0.025)
SELF ESTEEM 0.05** 0.08 0.05 0.04* 0.05** 0.04 -0.01 -0.08 -0.05* -0.03 0.01 0.01
(0.017) (0.039) (0.028) (0.018) (0.015) (0.024) (0.021) (0.059) (0.025) (0.033) (0.022) (0.028)
VERBAL PT in G7 -0.06** -0.05 -0.08* -0.07** -0.06* -0.06* -0.05 -0.08 0.02 -0.01 -0.07* -0.03
(0.021) (0.052) (0.032) (0.022) (0.027) (0.023) (0.027) (0.048) (0.042) (0.035) (0.030) (0.042)
VERBAL PT in G9 0.00 0.06 0.00 -0.03 -0.02 -0.03
(0.027) (0.045) (0.054) (0.029) (0.022) (0.034)
Constant -0.49*** -1.69*** -1.07*** -0.45*** 0.21* 0.72*** -0.43*** -1.44*** -1.11*** -0.37** 0.11 0.75***
(0.082) (0.242) (0.152) (0.097) (0.102) (0.129) (0.102) (0.218) (0.170) (0.136) (0.103) (0.097)

Observations 5,478 5,478 5,478 5,478 5,478 5,478 4,164 4,164 4,164 4,164 4,164 4,164
R-squared 0.469 0.392
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8VER_Z (verbal score in grade 8) for (1)-(6); G9VER_Z (verbal score in grade 9) for (7)-(12)

281
Table K.4. Heterogeneous effects of verbal tutoring in grade 8 on verbal achievement for 2 years using the 2SLAD estimation
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G8 (predicted) -0.00 -0.15 0.05 -0.03 0.00 -0.03 -0.05 -0.05 0.04 0.14 0.13 -0.02
(0.053) (0.130) (0.073) (0.057) (0.058) (0.094) (0.077) (0.140) (0.104) (0.076) (0.085) (0.081)
URBANICITY -0.04 0.01 -0.05 -0.07 -0.02 -0.03 -0.10* -0.08 -0.06 -0.11*** -0.08* -0.07
(0.029) (0.057) (0.046) (0.042) (0.025) (0.035) (0.037) (0.055) (0.038) (0.030) (0.043) (0.047)
GENDER (1=female) 0.30*** 0.43*** 0.38*** 0.29*** 0.22*** 0.18*** 0.29*** 0.36*** 0.41*** 0.31*** 0.22*** 0.07*
(0.021) (0.043) (0.038) (0.030) (0.030) (0.038) (0.026) (0.065) (0.041) (0.040) (0.035) (0.031)
SES -0.04** -0.02 -0.06** -0.03* -0.02 -0.02 -0.01 -0.01 -0.02 -0.02 -0.00 0.01
(0.014) (0.027) (0.019) (0.016) (0.017) (0.020) (0.018) (0.041) (0.026) (0.023) (0.016) (0.020)
SCHOOL TYPE (1=private) 0.02 0.03 0.06 -0.03 -0.02 -0.00 0.07* 0.04 0.16*** 0.08* 0.04 -0.00
(0.025) (0.065) (0.035) (0.039) (0.028) (0.039) (0.032) (0.067) (0.039) (0.036) (0.029) (0.043)
STU-TEA RATIO 0.01** 0.01 0.01** 0.01** 0.01 0.00 0.00 -0.00 0.00 0.00 -0.00 -0.00
(0.002) (0.004) (0.003) (0.003) (0.003) (0.004) (0.003) (0.005) (0.005) (0.004) (0.003) (0.004)
SCHOOL CHOICE -0.04 -0.01 -0.03 -0.03 -0.01 -0.04 -0.01 -0.07 0.01 -0.02 0.06 0.05
(0.031) (0.050) (0.048) (0.045) (0.026) (0.035) (0.040) (0.054) (0.039) (0.033) (0.048) (0.053)
PRESCORE (G7) 0.72*** 0.76*** 0.83*** 0.78*** 0.68*** 0.58*** 0.67*** 0.55*** 0.80*** 0.79*** 0.64*** 0.51***
(0.014) (0.032) (0.022) (0.016) (0.021) (0.020) (0.017) (0.056) (0.022) (0.024) (0.033) (0.023)
STUDY HOUR (1= >10hrs/week) -0.03 0.02 -0.02 -0.03 -0.03 -0.04 -0.00 0.04 -0.00 0.01 0.05 0.07*
(0.030) (0.073) (0.039) (0.029) (0.029) (0.026) (0.039) (0.070) (0.045) (0.044) (0.026) (0.027)
SELF-ESTEEM 0.04* 0.07 0.02 0.03 0.05* 0.03 -0.01 -0.06 -0.06 -0.03 0.00 -0.01
(0.018) (0.040) (0.023) (0.027) (0.023) (0.025) (0.023) (0.053) (0.047) (0.030) (0.023) (0.030)
VERBAL PT in G7 -0.04 0.01 -0.08 -0.04 -0.04 -0.02 -0.01 -0.08 0.01 -0.03 -0.07* -0.01
(0.029) (0.071) (0.046) (0.031) (0.032) (0.052) (0.037) (0.073) (0.043) (0.052) (0.036) (0.039)
VERBAL PT in G9 0.01 0.06 -0.04 -0.07* -0.04 -0.01
(0.037) (0.058) (0.052) (0.037) (0.037) (0.043)
Constant -0.49*** -1.64*** -1.08*** -0.45** 0.12 0.65*** -0.38*** -1.42*** -1.10*** -0.36** 0.22 0.87***
(0.089) (0.158) (0.133) (0.160) (0.101) (0.138) (0.110) (0.175) (0.157) (0.123) (0.144) (0.159)

Observations 4,935 4,991 4,991 4,991 4,991 4,991 3,728 3,738 3,738 3,738 3,738 3,738
R-squared 0.462 0.381
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8VER_Z (verbal score in grade 8) for (1)-(6); G9VER_Z (verbal score in grade 9) for (7)-(12)

282
 283

Table K.5. Heterogeneous effects of verbal tutoring in grade 9 on verbal achievement in grade 9
using the OLS estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

VERBAL PT in G9 -0.00 0.02 -0.01 -0.01 -0.04 -0.02

(0.027) (0.037) (0.039) (0.035) (0.034) (0.039)
URBANICITY -0.08* -0.10 -0.05 -0.05 -0.08* -0.03
(0.034) (0.061) (0.046) (0.035) (0.034) (0.043)
GENDER (1=female) 0.26*** 0.30*** 0.35*** 0.28*** 0.20*** 0.12***
(0.024) (0.047) (0.037) (0.032) (0.025) (0.032)
SES -0.01 0.02 -0.00 -0.02 -0.01 -0.01
(0.016) (0.033) (0.021) (0.014) (0.017) (0.017)
SCHOOL TYPE (1=private) 0.05 0.00 0.10** 0.06 0.01 -0.01
(0.029) (0.060) (0.036) (0.031) (0.033) (0.040)
STU-TEA RATIO 0.00 -0.00 -0.00 0.00 -0.00 -0.00
(0.003) (0.004) (0.004) (0.004) (0.004) (0.003)
SCHOOL CHOICE 0.02 -0.01 0.04 0.06 0.03 0.09
(0.036) (0.043) (0.049) (0.050) (0.037) (0.050)
PRESCORE (G8) 0.71*** 0.66*** 0.84*** 0.81*** 0.66*** 0.49***
(0.015) (0.039) (0.016) (0.014) (0.020) (0.024)
STUDY HOUR (1= >10hrs/week) 0.00 0.03 -0.05 -0.01 0.03 0.07*
(0.036) (0.072) (0.056) (0.039) (0.037) (0.033)
SELF ESTEEM 0.01 0.00 0.00 0.00 -0.00 0.05
(0.021) (0.052) (0.024) (0.027) (0.018) (0.030)
VERBAL PT in G7 -0.02 -0.01 0.03 0.00 -0.04 -0.07*
(0.026) (0.049) (0.040) (0.031) (0.033) (0.033)
VERBAL PT in G8 0.00 -0.03 -0.02 -0.02 0.04 0.07*
(0.028) (0.043) (0.042) (0.042) (0.038) (0.031)
Constant -0.38*** -1.27*** -0.98*** -0.50*** 0.28* 0.66***
(0.100) (0.152) (0.148) (0.143) (0.116) (0.106)

Observations 4,133 4,133 4,133 4,133 4,133 4,133

R-squared 0.425
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VER_Z (verbal score in grade 9)
 284

Table K.6. Heterogeneous effects of verbal tutoring in grade 9 on verbal achievement in grade 9
using the 2SLAD estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

VERBAL PT in G9 (predicted) -0.04 0.05 -0.15 0.00 0.09 -0.07

(0.093) (0.150) (0.080) (0.140) (0.129) (0.096)
URBANICITY -0.09* -0.09 -0.07 -0.06* -0.06 -0.03
(0.037) (0.060) (0.043) (0.032) (0.032) (0.034)
GENDER (1=female) 0.27*** 0.30*** 0.35*** 0.29*** 0.23*** 0.14***
(0.027) (0.054) (0.034) (0.027) (0.028) (0.033)
SES -0.01 0.02 0.00 -0.01 -0.02 0.00
(0.017) (0.027) (0.024) (0.018) (0.012) (0.025)
SCHOOL TYPE (1=private) 0.07* 0.07 0.11* 0.06* 0.04 0.02
(0.031) (0.076) (0.049) (0.028) (0.024) (0.029)
STU-TEA RATIO 0.00 -0.00 0.00 0.00 0.00 0.00
(0.003) (0.007) (0.004) (0.003) (0.003) (0.004)
SCHOOL CHOICE 0.00 -0.08 -0.00 0.02 0.05 0.08*
(0.039) (0.062) (0.044) (0.045) (0.042) (0.038)
PRESCORE (G8) 0.69*** 0.68*** 0.82*** 0.78*** 0.65*** 0.47***
(0.017) (0.030) (0.017) (0.015) (0.018) (0.028)
STUDY HOUR (1= >10hrs/week) 0.01 -0.05 -0.05 -0.02 0.02 0.03
(0.039) (0.082) (0.041) (0.042) (0.046) (0.047)
SELF ESTEEM 0.01 0.00 0.00 0.01 0.03 0.05
(0.023) (0.046) (0.034) (0.026) (0.022) (0.026)
VERBAL PT in G7 -0.01 -0.03 0.03 0.01 -0.05 -0.09**
(0.031) (0.055) (0.050) (0.032) (0.026) (0.034)
VERBAL PT in G8 -0.01 -0.08 -0.02 -0.06 -0.02 0.06
(0.043) (0.073) (0.044) (0.058) (0.053) (0.040)
Constant -0.40*** -1.15*** -0.89*** -0.42*** 0.12 0.62***
(0.112) (0.183) (0.089) (0.100) (0.076) (0.105)

Observations 3,556 4,404 4,404 4,404 4,404 4,404

R-squared 0.414
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9VER_Z (verbal score in grade 9)
Table L.1. Heterogeneous effects of English tutoring in grade 7 on English achievement for 3 years using the OLS estimation
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G7 0.25*** 0.13*** 0.21*** 0.27*** 0.33*** 0.33*** 0.12*** 0.02 0.08** 0.12** 0.21*** 0.21*** 0.07* 0.01 0.06 0.06 0.18*** 0.11
(0.024) (0.037) (0.023) (0.036) (0.030) (0.037) (0.029) (0.030) (0.028) (0.041) (0.044) (0.038) (0.033) (0.039) (0.039) (0.055) (0.042) (0.063)
URBANICITY 0.10*** 0.12* 0.08* 0.05 0.10** 0.10* 0.05 0.11* 0.07 0.05 0.03 0.01 0.00 0.03 0.02 0.04 -0.03 -0.06
(0.029) (0.055) (0.037) (0.040) (0.037) (0.047) (0.032) (0.052) (0.041) (0.053) (0.029) (0.029) (0.036) (0.054) (0.052) (0.045) (0.049) (0.045)
GENDER (1=female) 0.28*** 0.34*** 0.33*** 0.29*** 0.23*** 0.17*** 0.36*** 0.40*** 0.45*** 0.37*** 0.28*** 0.20*** 0.34*** 0.21*** 0.45*** 0.41*** 0.25*** 0.14***
(0.020) (0.031) (0.028) (0.025) (0.024) (0.031) (0.022) (0.037) (0.033) (0.033) (0.033) (0.029) (0.025) (0.037) (0.032) (0.037) (0.034) (0.030)
SES 0.28*** 0.20*** 0.27*** 0.32*** 0.28*** 0.25*** 0.25*** 0.17*** 0.28*** 0.32*** 0.24*** 0.20*** 0.21*** 0.12*** 0.22*** 0.22*** 0.23*** 0.16***
(0.013) (0.026) (0.020) (0.018) (0.017) (0.020) (0.015) (0.024) (0.031) (0.019) (0.023) (0.020) (0.017) (0.022) (0.034) (0.032) (0.026) (0.012)
SCHOOL TYPE (1=private) 0.08** 0.04 0.02 0.10** 0.10** 0.08 0.06* 0.01 0.07 0.04 0.02 0.05 0.09** 0.05 0.08 0.10* 0.06 0.08
(0.025) (0.040) (0.036) (0.032) (0.037) (0.041) (0.028) (0.041) (0.047) (0.036) (0.037) (0.035) (0.031) (0.038) (0.044) (0.045) (0.050) (0.050)
STU-TEA RATIO 0.01*** 0.01* 0.01** 0.01*** 0.01*** 0.01** 0.01*** 0.00 0.01* 0.01* 0.01** 0.01* 0.01 0.01 0.01* 0.00 0.00 0.01*
(0.002) (0.004) (0.003) (0.003) (0.004) (0.005) (0.003) (0.003) (0.004) (0.004) (0.003) (0.002) (0.003) (0.004) (0.003) (0.004) (0.004) (0.003)
SCHOOL CHOICE -0.11*** -0.05 -0.13** -0.15** -0.14** -0.06 -0.08* 0.03 -0.04 -0.10* -0.09* -0.07 -0.05 -0.01 -0.02 -0.01 -0.10 -0.09
(0.031) (0.060) (0.040) (0.048) (0.047) (0.048) (0.034) (0.043) (0.049) (0.043) (0.037) (0.045) (0.038) (0.054) (0.051) (0.053) (0.057) (0.052)
PRESCORE (G6) 0.18*** 0.15*** 0.19*** 0.20*** 0.20*** 0.15*** 0.15*** 0.08*** 0.14*** 0.18*** 0.17*** 0.13*** 0.16*** 0.07*** 0.15*** 0.20*** 0.16*** 0.11***
(0.006) (0.013) (0.008) (0.008) (0.010) (0.008) (0.007) (0.010) (0.009) (0.010) (0.009) (0.010) (0.008) (0.008) (0.009) (0.014) (0.010) (0.010)
STUDY HOUR 0.13*** 0.17*** 0.14*** 0.10* 0.09** 0.07 0.11** 0.13* 0.20** 0.12* 0.08 0.03 0.03 -0.05 0.08 0.06 -0.02 -0.02
(1= >10hrs/week) (0.030) (0.041) (0.031) (0.043) (0.031) (0.042) (0.033) (0.054) (0.072) (0.045) (0.044) (0.039) (0.037) (0.063) (0.076) (0.036) (0.045) (0.035)
SELF-ESTEEM 0.23*** 0.11** 0.26*** 0.28*** 0.25*** 0.21*** 0.23*** 0.19*** 0.25*** 0.24*** 0.23*** 0.16*** 0.21*** 0.12** 0.21*** 0.24*** 0.19*** 0.16***
(0.018) (0.036) (0.033) (0.030) (0.023) (0.029) (0.020) (0.026) (0.028) (0.034) (0.031) (0.027) (0.022) (0.040) (0.049) (0.039) (0.032) (0.029)
ENGLISH PT in G8 0.25*** 0.15** 0.21*** 0.26*** 0.32*** 0.22*** 0.18*** 0.11*** 0.19*** 0.21** 0.20** 0.12
(0.029) (0.046) (0.030) (0.042) (0.063) (0.055) (0.036) (0.033) (0.048) (0.068) (0.060) (0.068)
ENGLISH PT in G9 0.23*** 0.10** 0.15** 0.31*** 0.29*** 0.23**
(0.035) (0.036) (0.045) (0.053) (0.059) (0.070)
- - - - - - - - - - - - - - -
Constant -1.82*** 2.71*** 2.33*** 1.84*** 1.32*** 0.61*** 1.77*** 2.37*** 2.40*** 1.94*** 1.32*** -0.33* 1.81*** 2.11*** 2.59*** 2.25*** 1.12*** -0.04
(0.092) (0.206) (0.100) (0.136) (0.138) (0.145) (0.102) (0.118) (0.152) (0.116) (0.151) (0.144) (0.114) (0.166) (0.132) (0.139) (0.150) (0.129)

Observations 5,910 5,910 5,910 5,910 5,910 5,910 5,380 5,380 5,380 5,380 5,380 5,380 4,546 4,546 4,546 4,546 4,546 4,546
R-squared 0.403 0.338 0.314
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7ENG_Z (English score in grade 7) for (1)-(6); G8ENG_Z (English score in grade 8) for (7)-(12); G9ENG_Z (English score in grade 9) for (13)-(18)

285
Table L.2. Heterogeneous effects of English tutoring in grade 7 on English achievement for 3 years using the 2SLAD estimation
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES IV q10 q25 q50 q75 q90 IV q10 q25 q50 q75 q90 IV q10 q25 q50 q75 q90

ENGLISH PT in G7 (predicted) 0.72*** 0.32* 0.74*** 1.07*** 1.20*** 1.03*** 0.61*** 0.23 0.74** 1.02*** 0.86*** 0.78*** 0.64*** 0.42* 0.71*** 0.68*** 0.75*** 0.50***
(0.071) (0.145) (0.173) (0.126) (0.140) (0.115) (0.086) (0.239) (0.259) (0.154) (0.133) (0.183) (0.098) (0.191) (0.149) (0.108) (0.122) (0.066)
URBANICITY 0.05 0.06 -0.00 -0.01 0.01 0.02 0.01 0.07 0.01 -0.02 -0.00 0.00 -0.03 0.03 0.00 -0.01 -0.08* -0.07
(0.032) (0.052) (0.048) (0.035) (0.040) (0.033) (0.034) (0.073) (0.050) (0.048) (0.035) (0.048) (0.039) (0.044) (0.054) (0.037) (0.033) (0.039)
GENDER (1=female) 0.30*** 0.33*** 0.34*** 0.33*** 0.26*** 0.18*** 0.37*** 0.40*** 0.46*** 0.38*** 0.31*** 0.20*** 0.35*** 0.22*** 0.45*** 0.42*** 0.27*** 0.13***
(0.022) (0.028) (0.024) (0.023) (0.025) (0.039) (0.024) (0.041) (0.032) (0.032) (0.034) (0.033) (0.027) (0.032) (0.026) (0.041) (0.036) (0.028)
SES 0.25*** 0.19*** 0.24*** 0.25*** 0.21*** 0.17*** 0.23*** 0.17*** 0.28*** 0.28*** 0.20*** 0.16*** 0.19*** 0.11*** 0.20*** 0.21*** 0.20*** 0.14***
(0.016) (0.028) (0.018) (0.020) (0.019) (0.023) (0.017) (0.023) (0.031) (0.019) (0.013) (0.025) (0.019) (0.023) (0.031) (0.027) (0.025) (0.015)
SCHOOL TYPE (1=private) 0.06* 0.02 -0.01 0.08* 0.08 0.03 0.04 0.03 0.03 0.03 0.02 0.03 0.06 0.02 0.06 0.08 0.05 0.06
(0.027) (0.043) (0.029) (0.035) (0.052) (0.036) (0.029) (0.039) (0.040) (0.035) (0.034) (0.038) (0.033) (0.042) (0.055) (0.044) (0.034) (0.046)
STU-TEA RATIO 0.01*** 0.01* 0.00 0.01** 0.01*** 0.01** 0.01** 0.00 0.00 0.01** 0.01*** 0.01* 0.01* 0.01 0.01 0.00 0.01 0.01*
(0.003) (0.003) (0.004) (0.003) (0.003) (0.004) (0.003) (0.005) (0.004) (0.004) (0.003) (0.004) (0.003) (0.003) (0.004) (0.004) (0.005) (0.003)
- -
SCHOOL CHOICE 0.12*** -0.05 -0.13* 0.17*** -0.14** -0.09* -0.10** -0.01 -0.10 -0.12* -0.10* -0.04 -0.05 -0.01 -0.04 -0.04 -0.11* -0.07
(0.033) (0.064) (0.063) (0.042) (0.044) (0.039) (0.036) (0.053) (0.058) (0.056) (0.046) (0.047) (0.041) (0.048) (0.063) (0.044) (0.042) (0.039)
PRESCORE (G6) 0.17*** 0.14*** 0.16*** 0.18*** 0.17*** 0.14*** 0.14*** 0.07*** 0.12*** 0.17*** 0.17*** 0.12*** 0.15*** 0.07*** 0.13*** 0.19*** 0.16*** 0.11***
(0.007) (0.011) (0.008) (0.010) (0.013) (0.011) (0.007) (0.010) (0.011) (0.010) (0.012) (0.011) (0.008) (0.007) (0.008) (0.010) (0.014) (0.013)
STUDY HOUR 0.10** 0.15** 0.14** 0.09** 0.05 0.03 0.09* 0.11 0.16** 0.08 0.04 -0.01 -0.01 -0.06 0.02 0.04 -0.04 -0.04
(1= >10hrs/week) (0.032) (0.052) (0.051) (0.031) (0.037) (0.029) (0.035) (0.065) (0.052) (0.041) (0.030) (0.025) (0.040) (0.061) (0.053) (0.042) (0.047) (0.024)
SELF ESTEEM 0.21*** 0.11** 0.23*** 0.24*** 0.22*** 0.17*** 0.22*** 0.19*** 0.22*** 0.22*** 0.20*** 0.14*** 0.20*** 0.11*** 0.21*** 0.22*** 0.18*** 0.15***
(0.020) (0.035) (0.027) (0.025) (0.030) (0.024) (0.021) (0.037) (0.033) (0.034) (0.022) (0.017) (0.024) (0.025) (0.025) (0.033) (0.038) (0.024)
ENGLISH PT in G8 0.05 0.06 -0.06 -0.12 0.04 0.03 -0.02 -0.03 -0.07 -0.06 -0.03 -0.03
(0.045) (0.096) (0.104) (0.081) (0.065) (0.092) (0.050) (0.066) (0.054) (0.068) (0.064) (0.051)
ENGLISH PT in G9 0.13** 0.04 0.05 0.20** 0.19*** 0.20***
(0.041) (0.052) (0.048) (0.061) (0.058) (0.033)
- - - - - - - - - - - - - - - - -
Constant 2.02*** 2.69*** 2.45*** 2.25*** 1.80*** 0.94*** 1.90*** 2.33*** 2.40*** 2.20*** 1.61*** 0.67*** 1.99*** 2.22*** 2.63*** 2.32*** 1.33*** -0.20
(0.105) (0.155) (0.198) (0.124) (0.183) (0.143) (0.112) (0.159) (0.200) (0.162) (0.114) (0.193) (0.126) (0.163) (0.140) (0.171) (0.186) (0.143)

Observations 5,487 5,880 5,880 5,880 5,880 5,880 5,023 5,364 5,364 5,364 5,364 5,364 4,248 4,532 4,532 4,532 4,532 4,532
R-squared 0.362 0.301 0.269
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7ENG_Z (English score in grade 7) for (1)-(6); G8ENG_Z (English score in grade 8) for (7)-(12); G9ENG_Z (English score in grade 9) for (13)-(18)

286
Table L.3. Heterogeneous effects of English tutoring in grade 8 on English achievement for 2 years using the OLS esitmaiton
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G8 0.14*** 0.08 0.13*** 0.16*** 0.14*** 0.16*** 0.09** 0.04 0.10** 0.06 0.10*** 0.08*
(0.024) (0.057) (0.025) (0.027) (0.031) (0.045) (0.030) (0.033) (0.032) (0.033) (0.027) (0.033)
URBANICITY 0.00 0.04 0.02 -0.01 0.02 -0.02 -0.04 -0.02 -0.06 -0.04 -0.03 -0.05
(0.026) (0.047) (0.028) (0.025) (0.025) (0.031) (0.030) (0.058) (0.053) (0.033) (0.047) (0.051)
GENDER (1=female) 0.17*** 0.26*** 0.21*** 0.15*** 0.11*** 0.07** 0.15*** 0.19*** 0.25*** 0.13*** 0.08** 0.05
(0.018) (0.035) (0.017) (0.017) (0.018) (0.022) (0.021) (0.049) (0.037) (0.026) (0.026) (0.033)
SES 0.11*** 0.11** 0.13*** 0.12*** 0.13*** 0.10*** 0.10*** 0.09* 0.13*** 0.09*** 0.08*** 0.06**
(0.012) (0.032) (0.017) (0.014) (0.015) (0.016) (0.015) (0.042) (0.037) (0.019) (0.016) (0.018)
SCHOOL TYPE (1=private) 0.03 -0.02 0.03 -0.01 0.03 0.03 0.06* 0.09 0.08 0.07 0.05 0.04
(0.022) (0.051) (0.028) (0.032) (0.023) (0.041) (0.026) (0.052) (0.057) (0.039) (0.036) (0.049)
STU-TEA RATIO -0.00 -0.00 -0.00 -0.00 0.00 -0.00 -0.00 -0.00 -0.01 -0.00* -0.00 -0.01
(0.002) (0.003) (0.002) (0.001) (0.003) (0.003) (0.003) (0.003) (0.003) (0.002) (0.002) (0.003)
SCHOOL CHOICE -0.03 0.01 -0.02 -0.02 -0.02 -0.05 -0.01 0.02 0.00 -0.01 -0.02 -0.04
(0.027) (0.050) (0.033) (0.029) (0.019) (0.035) (0.032) (0.066) (0.062) (0.040) (0.038) (0.045)
PRESCORE (G7) 0.74*** 0.62*** 0.80*** 0.83*** 0.76*** 0.62*** 0.70*** 0.43*** 0.78*** 0.82*** 0.75*** 0.59***
(0.012) (0.030) (0.015) (0.012) (0.017) (0.023) (0.014) (0.030) (0.025) (0.012) (0.014) (0.026)
STUDY HOUR (1= >10hrs/week) 0.05 0.07 0.09* 0.04 0.03 -0.01 -0.02 -0.10 0.02 0.00 0.00 -0.01
(0.027) (0.064) (0.037) (0.032) (0.027) (0.033) (0.031) (0.059) (0.047) (0.032) (0.036) (0.034)
SELF ESTEEM 0.09*** 0.09** 0.09*** 0.09*** 0.08*** 0.07*** 0.11*** 0.09* 0.10*** 0.11*** 0.08*** 0.10***
(0.016) (0.029) (0.024) (0.018) (0.021) (0.019) (0.018) (0.041) (0.029) (0.018) (0.016) (0.026)
ENGLISH PT in G7 0.00 -0.04 0.00 0.01 0.01 0.07 -0.03 -0.01 -0.08 -0.03 0.00 0.00
(0.023) (0.034) (0.027) (0.026) (0.024) (0.038) (0.028) (0.042) (0.044) (0.034) (0.030) (0.050)
ENGLISH PT in G9 0.17*** 0.13** 0.15*** 0.21*** 0.15** 0.16***
(0.029) (0.041) (0.038) (0.034) (0.046) (0.044)
Constant -0.30*** -1.18*** -0.74*** -0.28*** 0.15 0.69*** -0.29** -1.35*** -0.80*** -0.20* 0.23* 0.79***
(0.077) (0.131) (0.091) (0.072) (0.093) (0.125) (0.091) (0.195) (0.169) (0.100) (0.097) (0.140)

Observations 5,502 5,502 5,502 5,502 5,502 5,502 4,643 4,643 4,643 4,643 4,643 4,643
R-squared 0.561 0.504
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8ENG_Z (English score in grade 8) for (1)-(6); G9ENG_Z (English score in grade 9) for (7)-(12)

287
Table L.4. Heterogeneous effects of English tutoring in grade 8 on English achievement for 2 years using the 2SLAD estimation
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G8 (predicted) 0.31*** 0.19 0.45*** 0.40*** 0.26* 0.30*** 0.18 -0.11 0.27 0.29** 0.12 0.12
(0.073) (0.108) (0.105) (0.090) (0.104) (0.068) (0.102) (0.206) (0.139) (0.101) (0.092) (0.157)
URBANICITY -0.01 0.03 -0.01 -0.03 0.01 -0.00 -0.04 0.01 -0.03 -0.03 -0.05 -0.04
(0.027) (0.062) (0.035) (0.026) (0.026) (0.027) (0.032) (0.079) (0.055) (0.032) (0.039) (0.066)
GENDER (1=female) 0.17*** 0.25*** 0.22*** 0.17*** 0.11*** 0.09** 0.16*** 0.21*** 0.27*** 0.14*** 0.07** 0.04
(0.019) (0.037) (0.024) (0.017) (0.024) (0.028) (0.022) (0.043) (0.030) (0.021) (0.023) (0.040)
SES 0.12*** 0.12** 0.12*** 0.11*** 0.13*** 0.09*** 0.10*** 0.10*** 0.14*** 0.09*** 0.08*** 0.06**
(0.014) (0.041) (0.018) (0.014) (0.017) (0.012) (0.016) (0.031) (0.021) (0.014) (0.016) (0.021)
SCHOOL TYPE (1=private) 0.04 0.00 0.06** 0.02 0.01 0.07* 0.07** 0.12* 0.09* 0.07* 0.05* 0.04
(0.023) (0.050) (0.022) (0.023) (0.024) (0.036) (0.027) (0.056) (0.038) (0.030) (0.024) (0.028)
STU-TEA RATIO -0.00 -0.00 0.00 -0.00 0.00 -0.00 -0.00 -0.00 -0.01 -0.00 -0.00 -0.01
(0.002) (0.004) (0.003) (0.002) (0.002) (0.003) (0.003) (0.005) (0.004) (0.002) (0.004) (0.004)
SCHOOL CHOICE -0.04 0.00 -0.02 -0.02 -0.03 -0.03 -0.00 0.03 0.01 0.00 -0.02 -0.02
(0.029) (0.050) (0.032) (0.029) (0.033) (0.035) (0.034) (0.067) (0.041) (0.038) (0.038) (0.074)
PRESCORE (G7) 0.73*** 0.63*** 0.79*** 0.82*** 0.75*** 0.60*** 0.70*** 0.46*** 0.78*** 0.82*** 0.75*** 0.58***
(0.014) (0.032) (0.015) (0.015) (0.016) (0.025) (0.016) (0.026) (0.022) (0.017) (0.018) (0.042)
STUDY HOUR (1= >10hrs/week) 0.03 0.07 0.05 0.01 0.03 0.00 -0.05 -0.12 -0.01 -0.01 0.00 -0.02
(0.028) (0.082) (0.046) (0.027) (0.026) (0.023) (0.033) (0.079) (0.052) (0.031) (0.043) (0.037)
SELF ESTEEM 0.09*** 0.08** 0.08** 0.08*** 0.07*** 0.06** 0.10*** 0.09*** 0.09** 0.09*** 0.07*** 0.08**
(0.017) (0.030) (0.027) (0.015) (0.021) (0.020) (0.019) (0.022) (0.030) (0.018) (0.017) (0.026)
ENGLISH PT in G7 -0.06 -0.10* -0.12* -0.08* -0.04 0.00 -0.07 0.02 -0.13** -0.10* -0.01 0.01
(0.035) (0.047) (0.047) (0.041) (0.052) (0.040) (0.040) (0.071) (0.050) (0.040) (0.040) (0.048)
ENGLISH PT in G9 0.14** 0.18 0.09 0.12* 0.14** 0.13*
(0.045) (0.094) (0.074) (0.052) (0.045) (0.059)
Constant -0.37*** -1.20*** -0.95*** -0.39*** 0.14 0.58*** -0.33** -1.35*** -0.88*** -0.31** 0.21** 0.79***
(0.090) (0.137) (0.094) (0.106) (0.096) (0.091) (0.102) (0.207) (0.117) (0.112) (0.065) (0.186)

Observations 4,955 5,003 5,003 5,003 5,003 5,003 4,188 4,230 4,230 4,230 4,230 4,230
R-squared 0.555 0.499
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8ENG_Z (English score in grade 8) for (1)-(6); G9ENG_Z (English score in grade 9) for (7)-(12)

288
 289

Table L.5. Heterogeneous effects of English tutoring in grade 9 on English achievement in grade
9 using the OLS estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

ENGLISH PT in G9 0.16*** 0.12** 0.14*** 0.16*** 0.23*** 0.17**

(0.028) (0.037) (0.028) (0.034) (0.029) (0.056)
URBANICITY -0.02 -0.03 -0.03 -0.04 -0.04* 0.01
(0.029) (0.056) (0.045) (0.026) (0.020) (0.040)
GENDER (1=female) 0.12*** 0.17*** 0.14*** 0.10*** 0.08*** 0.02
(0.021) (0.033) (0.027) (0.021) (0.020) (0.027)
SES 0.11*** 0.09*** 0.11*** 0.08*** 0.09*** 0.09***
(0.014) (0.022) (0.020) (0.013) (0.017) (0.019)
SCHOOL TYPE (1=private) 0.04 -0.01 0.03 0.03 0.06* 0.04
(0.025) (0.041) (0.042) (0.031) (0.031) (0.027)
STU-TEA RATIO -0.00 -0.00 -0.00 -0.00 -0.01** -0.00
(0.002) (0.004) (0.002) (0.002) (0.002) (0.003)
SCHOOL CHOICE 0.01 0.06 0.02 -0.03 -0.03 0.04
(0.031) (0.062) (0.059) (0.026) (0.032) (0.049)
PRESCORE (G8) 0.73*** 0.53*** 0.85*** 0.84*** 0.76*** 0.56***
(0.014) (0.020) (0.017) (0.013) (0.017) (0.030)
STUDY HOUR (1= >10hrs/week) -0.03 -0.06 -0.00 0.02 -0.02 -0.06
(0.030) (0.053) (0.064) (0.031) (0.042) (0.035)
SELF ESTEEM 0.13*** 0.09** 0.12*** 0.11*** 0.10*** 0.10***
(0.018) (0.029) (0.024) (0.014) (0.017) (0.023)
ENGLISH PT in G7 0.01 0.05 -0.01 0.02 0.03 0.06
(0.027) (0.042) (0.035) (0.033) (0.022) (0.042)
ENGLISH PT in G8 0.06* 0.04 0.05 0.04 0.02 0.06
(0.029) (0.034) (0.036) (0.033) (0.033) (0.041)
Constant -0.27** -1.38*** -0.69*** -0.15** 0.28*** 0.64***
(0.088) (0.145) (0.129) (0.051) (0.085) (0.121)

Observations 4,609 4,609 4,609 4,609 4,609 4,609

R-squared 0.536
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in grade 9)
 290

Table L.6. Heterogeneous effects of English tutoring in grade 9 on English achievement in grade
9 using the 2SLAD estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

ENGLISH PT in G9 (predicted) -0.15 -0.14 -0.34 0.33 2.02 0.37

(0.272) (1.478) (1.406) (2.044) (1.919) (1.349)
URBANICITY -0.00 -0.04 -0.01 -0.04 -0.05 0.04
(0.032) (0.058) (0.049) (0.048) (0.042) (0.046)
GENDER (1=female) 0.12*** 0.17*** 0.14** 0.10 0.14* 0.02
(0.024) (0.044) (0.053) (0.076) (0.061) (0.050)
SES 0.12*** 0.08 0.13 0.07 0.01 0.06
(0.018) (0.069) (0.065) (0.079) (0.071) (0.054)
SCHOOL TYPE (1=private) 0.05 0.01 0.03 0.05* 0.10** 0.08*
(0.027) (0.062) (0.046) (0.024) (0.032) (0.035)
STU-TEA RATIO -0.00 0.00 0.00 -0.00 -0.02 -0.00
(0.003) (0.010) (0.010) (0.016) (0.015) (0.011)
SCHOOL CHOICE 0.02 0.00 0.01 -0.03 -0.05 0.05
(0.035) (0.076) (0.053) (0.056) (0.049) (0.060)
PRESCORE (G8) 0.74*** 0.55*** 0.87*** 0.84*** 0.69*** 0.57***
(0.018) (0.075) (0.057) (0.079) (0.076) (0.062)
STUDY HOUR (1= >10hrs/week) -0.04 -0.06 -0.02 0.01 -0.07* -0.04
(0.033) (0.053) (0.046) (0.030) (0.033) (0.031)
SELF ESTEEM 0.13*** 0.09 0.15*** 0.11 0.05 0.11*
(0.020) (0.052) (0.045) (0.061) (0.053) (0.047)
ENGLISH PT in G7 0.07 0.16 0.11 -0.01 -0.27 -0.01
(0.050) (0.214) (0.221) (0.309) (0.287) (0.208)
ENGLISH PT in G8 0.14 0.09 0.16 -0.03 -0.62 -0.03
(0.100) (0.516) (0.485) (0.718) (0.682) (0.484)
Constant -0.22 -1.24** -0.58 -0.22 -0.25 0.52
(0.119) (0.421) (0.396) (0.534) (0.493) (0.335)

Observations 4,002 4,423 4,423 4,423 4,423 4,423

R-squared 0.524
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9ENG_Z (English score in grade 9)
Table M.1. Heterogeneous effects of math tutoring in grade 7 on math achievement for 3 years using the OLS estimation
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G7 0.26*** 0.17*** 0.23*** 0.26*** 0.31*** 0.22*** 0.09** 0.04 0.04 0.10** 0.15*** 0.12* 0.04 -0.01 -0.01 0.03 0.12** 0.05
(0.024) (0.045) (0.034) (0.027) (0.028) (0.042) (0.030) (0.036) (0.037) (0.033) (0.042) (0.049) (0.033) (0.039) (0.042) (0.051) (0.042) (0.061)
URBANICITY 0.05 0.02 0.03 0.05 0.05 0.10*** -0.02 0.00 0.02 -0.01 -0.05 -0.07 -0.14*** -0.04 -0.10 -0.21** -0.16*** -0.04
(0.031) (0.030) (0.031) (0.043) (0.038) (0.029) (0.034) (0.035) (0.044) (0.048) (0.044) (0.057) (0.037) (0.037) (0.078) (0.069) (0.040) (0.052)
GENDER (1=female) 0.02 0.07 0.06* 0.01 -0.05** -0.02 0.04 0.10*** 0.13*** 0.08* -0.02 -0.08 0.01 0.08* 0.05 0.02 -0.04 -0.12***
(0.021) (0.043) (0.031) (0.018) (0.018) (0.022) (0.023) (0.026) (0.036) (0.035) (0.048) (0.047) (0.026) (0.038) (0.039) (0.030) (0.026) (0.033)
SES 0.16*** 0.16*** 0.19*** 0.20*** 0.15*** 0.10*** 0.16*** 0.07*** 0.13*** 0.21*** 0.20*** 0.15*** 0.16*** 0.09*** 0.14*** 0.19*** 0.17*** 0.11***
(0.014) (0.027) (0.020) (0.020) (0.017) (0.018) (0.016) (0.019) (0.015) (0.021) (0.020) (0.023) (0.017) (0.025) (0.028) (0.025) (0.025) (0.020)
SCHOOL TYPE (1=private) 0.05 0.04 0.04* 0.06* 0.06 0.01 0.07* 0.07 0.10* 0.06 0.03 0.03 0.15*** 0.12*** 0.13*** 0.15*** 0.13*** 0.10***
(0.026) (0.048) (0.021) (0.031) (0.036) (0.040) (0.029) (0.049) (0.039) (0.036) (0.044) (0.039) (0.032) (0.027) (0.039) (0.034) (0.038) (0.027)
STU-TEA RATIO 0.01*** 0.01* 0.01*** 0.01*** 0.02*** 0.02*** 0.01*** 0.00 0.01* 0.01 0.02*** 0.02** 0.01* 0.01 0.00 0.01** 0.01 0.01
(0.003) (0.004) (0.002) (0.003) (0.003) (0.003) (0.003) (0.003) (0.003) (0.003) (0.004) (0.005) (0.003) (0.003) (0.005) (0.003) (0.003) (0.004)
SCHOOL CHOICE -0.06* -0.12** -0.05 -0.09* -0.08 0.03 -0.11** -0.08** -0.06 -0.10 -0.15** -0.14** -0.04 -0.01 -0.04 -0.08 -0.05 0.09
(0.032) (0.044) (0.031) (0.039) (0.044) (0.042) (0.036) (0.030) (0.046) (0.057) (0.055) (0.051) (0.039) (0.030) (0.074) (0.068) (0.048) (0.049)
PRESCORE (G6) 0.22*** 0.17*** 0.23*** 0.26*** 0.23*** 0.17*** 0.18*** 0.10*** 0.17*** 0.21*** 0.21*** 0.17*** 0.17*** 0.09*** 0.14*** 0.20*** 0.20*** 0.14***
(0.006) (0.015) (0.009) (0.009) (0.007) (0.009) (0.007) (0.010) (0.010) (0.012) (0.007) (0.015) (0.008) (0.009) (0.007) (0.009) (0.010) (0.008)
STUDY HOUR 0.11*** 0.19** 0.17** 0.11* 0.07* 0.01 0.11*** 0.16* 0.18** 0.12* 0.11* 0.04 0.06 -0.04 0.07 0.17*** 0.05 0.03
(1= >10hrs/week) (0.031) (0.067) (0.064) (0.045) (0.031) (0.036) (0.035) (0.066) (0.055) (0.051) (0.049) (0.063) (0.038) (0.049) (0.050) (0.020) (0.035) (0.044)
SELF ESTEEM 0.17*** 0.16*** 0.18*** 0.18*** 0.15*** 0.15*** 0.15*** 0.08** 0.12*** 0.17*** 0.17*** 0.18*** 0.10*** 0.01 0.09* 0.12*** 0.14*** 0.15***
(0.019) (0.028) (0.031) (0.024) (0.018) (0.018) (0.021) (0.030) (0.031) (0.034) (0.024) (0.024) (0.023) (0.042) (0.043) (0.036) (0.040) (0.037)
MATH PT in G8 0.28*** 0.17*** 0.25*** 0.32*** 0.33*** 0.34*** 0.20*** 0.11** 0.13* 0.23*** 0.27*** 0.20**
(0.031) (0.033) (0.040) (0.048) (0.049) (0.067) (0.037) (0.040) (0.052) (0.044) (0.066) (0.063)
MATH PT in G9 0.30*** 0.10** 0.27*** 0.34*** 0.39*** 0.36***
(0.036) (0.036) (0.048) (0.061) (0.058) (0.068)
- - - - - - - - - - - - - -
Constant 1.72*** 2.35*** 2.39*** 1.87*** 1.20*** 0.55*** 1.44*** 1.93*** 2.13*** 1.66*** 1.07*** -0.20 -1.41*** 2.03*** 1.84*** 1.66*** -1.05*** -0.21
(0.096) (0.155) (0.125) (0.156) (0.128) (0.139) (0.106) (0.099) (0.117) (0.117) (0.131) (0.145) (0.117) (0.103) (0.186) (0.133) (0.107) (0.144)

Observations 5,859 5,859 5,859 5,859 5,859 5,859 5,391 5,391 5,391 5,391 5,391 5,391 4,641 4,641 4,641 4,641 4,641 4,641
R-squared 0.348 0.286 0.259
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7MAT_Z (math score in grade 7) for (1)-(6); G8MAT_Z (math score in grade 8) for (7)-(12); G9MAT_Z (math score in grade 9) for (13)-(18)

291
Table M.2. Heterogeneous effects of math tutoring in grade 7 on math achievement for 3 years using the 2SLAD estimation
Year 0 (Grade 7) Year 1 (Grade 8) Year 2 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18)
VARIABLES IV q10 q25 q50 q75 q90 IV q10 q25 q50 q75 q90 IV q10 q25 q50 q75 q90

MATH PT in G7 (predicted) 0.53*** 0.46 0.75*** 0.80*** 0.76*** 0.52*** 0.36*** 0.16 0.27 0.84*** 0.97** 1.12*** 0.34** 0.38 0.26 0.46 0.61** 0.61
(0.081) (0.312) (0.176) (0.131) (0.162) (0.106) (0.085) (0.220) (0.224) (0.237) (0.304) (0.230) (0.132) (0.260) (0.260) (0.295) (0.213) (0.347)
- - - -
URBANICITY 0.03 -0.03 -0.01 -0.04 0.03 0.06 -0.06 -0.02 -0.05 -0.07 0.11*** -0.10** 0.17*** -0.06 -0.14 0.23*** 0.17*** -0.06
(0.032) (0.044) (0.057) (0.052) (0.047) (0.043) (0.035) (0.046) (0.045) (0.049) (0.031) (0.034) (0.039) (0.045) (0.081) (0.051) (0.042) (0.045)
-
GENDER (1=female) 0.02 0.09* 0.07 0.02 -0.02 -0.04 0.05* 0.11*** 0.13*** 0.08* -0.00 -0.06* 0.01 0.08 0.05 0.03 -0.04 0.12***
(0.022) (0.044) (0.041) (0.038) (0.035) (0.024) (0.024) (0.026) (0.030) (0.032) (0.034) (0.031) (0.027) (0.044) (0.041) (0.038) (0.033) (0.030)
SES 0.14*** 0.16*** 0.16*** 0.15*** 0.12*** 0.07*** 0.16*** 0.07** 0.13*** 0.19*** 0.18*** 0.10*** 0.15*** 0.07* 0.14*** 0.19*** 0.16*** 0.10**
(0.016) (0.039) (0.028) (0.019) (0.022) (0.014) (0.017) (0.026) (0.017) (0.025) (0.023) (0.025) (0.019) (0.030) (0.031) (0.033) (0.027) (0.031)
SCHOOL TYPE (1=private) 0.04 0.03 0.02 0.05* 0.02 -0.01 0.06 0.08* 0.08* 0.02 -0.02 -0.00 0.13*** 0.10** 0.14** 0.15** 0.14*** 0.10*
(0.028) (0.055) (0.038) (0.022) (0.048) (0.041) (0.030) (0.034) (0.031) (0.035) (0.051) (0.049) (0.033) (0.036) (0.044) (0.049) (0.028) (0.047)
STU-TEA RATIO 0.01*** 0.00 0.01* 0.01** 0.01*** 0.02*** 0.01** 0.00 0.01 0.00 0.01** 0.01*** 0.01 0.00 0.00 0.01 0.00 0.00
(0.003) (0.004) (0.004) (0.003) (0.003) (0.002) (0.003) (0.003) (0.003) (0.004) (0.004) (0.004) (0.003) (0.004) (0.005) (0.005) (0.004) (0.004)
- - -
SCHOOL CHOICE -0.07* 0.14*** -0.03 -0.11* -0.05 0.03 0.12*** -0.08 -0.09** -0.13* 0.15*** -0.11* -0.05 -0.01 -0.05 -0.06 -0.05 0.11**
(0.034) (0.041) (0.057) (0.052) (0.043) (0.042) (0.037) (0.050) (0.031) (0.053) (0.044) (0.048) (0.041) (0.051) (0.075) (0.044) (0.041) (0.040)
PRESCORE (G6) 0.21*** 0.17*** 0.22*** 0.25*** 0.22*** 0.17*** 0.18*** 0.09*** 0.16*** 0.21*** 0.20*** 0.18*** 0.17*** 0.09*** 0.13*** 0.19*** 0.20*** 0.15***
(0.007) (0.011) (0.007) (0.006) (0.006) (0.008) (0.007) (0.011) (0.010) (0.014) (0.012) (0.011) (0.008) (0.011) (0.012) (0.017) (0.013) (0.010)
STUDY HOUR 0.09** 0.17** 0.12* 0.11*** 0.05 0.02 0.11** 0.15* 0.18** 0.11 0.08 0.05 0.06 -0.05 0.03 0.15*** 0.05 0.03
(1= >10hrs/week) (0.033) (0.064) (0.054) (0.027) (0.042) (0.045) (0.036) (0.066) (0.057) (0.070) (0.057) (0.048) (0.039) (0.062) (0.091) (0.045) (0.030) (0.038)
SELF ESTEEM 0.16*** 0.17*** 0.18*** 0.15*** 0.13*** 0.11*** 0.15*** 0.08** 0.12*** 0.17*** 0.14*** 0.11*** 0.10*** 0.01 0.09* 0.16*** 0.12*** 0.13***
(0.020) (0.036) (0.034) (0.027) (0.027) (0.023) (0.022) (0.026) (0.032) (0.035) (0.029) (0.033) (0.024) (0.033) (0.044) (0.046) (0.029) (0.026)
MATH PT in G8 0.16*** 0.13 0.15 -0.02 -0.02 -0.09 0.09 -0.03 0.01 0.08 0.10 0.05
(0.046) (0.089) (0.096) (0.107) (0.113) (0.102) (0.058) (0.090) (0.097) (0.096) (0.094) (0.138)
MATH PT in G9 0.25*** 0.03 0.21*** 0.22*** 0.29*** 0.27**
(0.044) (0.055) (0.045) (0.062) (0.066) (0.091)
- - - - - - - - - - - - - - - - -
Constant 1.83*** 2.39*** 2.63*** 2.04*** 1.42*** 0.69*** 1.52*** 1.95*** 2.05*** 1.72*** 1.23*** 0.67*** 1.50*** 2.06*** 1.83*** 1.74*** 1.18*** -0.39*
(0.108) (0.185) (0.159) (0.187) (0.160) (0.095) (0.114) (0.135) (0.133) (0.166) (0.137) (0.193) (0.128) (0.187) (0.222) (0.182) (0.170) (0.154)

Observations 5,445 5,830 5,830 5,830 5,830 5,830 5,035 5,372 5,372 5,372 5,372 5,372 4,336 4,622 4,622 4,622 4,622 4,622
R-squared 0.330 0.275 0.244
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G7MAT_Z (math score in grade 7) for (1)-(6); G8MAT_Z (math score in grade 8) for (7)-(12); G9MAT_Z (math score in grade 9) for (13)-(18)

292
Table M.3. Heterogeneous effects of math tutoring in grade 8 on math achievement for 2 years using the OLS estimation
Year 0 (Grade 8) Year 1 (Grade 9)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G8 0.17*** 0.16*** 0.13*** 0.17*** 0.20*** 0.19*** 0.11*** 0.04 0.11** 0.10** 0.13*** 0.15*
(0.026) (0.044) (0.036) (0.039) (0.048) (0.057) (0.032) (0.043) (0.036) (0.040) (0.037) (0.070)
URBANICITY -0.08** -0.04 -0.04 -0.07* -0.12*** -0.10* -0.18*** -0.12* -0.18*** -0.18*** -0.14*** -0.08
(0.029) (0.044) (0.039) (0.036) (0.034) (0.039) (0.032) (0.051) (0.053) (0.046) (0.034) (0.044)
GENDER (1=female) -0.15*** -0.08 -0.13*** -0.15*** -0.21*** -0.19*** -0.18*** -0.06 -0.14*** -0.22*** -0.21*** -0.23***
(0.020) (0.044) (0.034) (0.025) (0.029) (0.029) (0.023) (0.041) (0.026) (0.023) (0.019) (0.036)
SES 0.04** 0.01 0.03 0.07** 0.05** 0.03 0.05*** 0.04* 0.03 0.07** 0.04 0.05**
(0.014) (0.022) (0.020) (0.022) (0.016) (0.015) (0.015) (0.020) (0.024) (0.022) (0.020) (0.018)
SCHOOL TYPE (1=private) 0.04 0.02 0.02 0.04* 0.04 0.04 0.12*** 0.09* 0.10* 0.11* 0.14*** 0.09**
(0.025) (0.039) (0.036) (0.021) (0.028) (0.037) (0.028) (0.041) (0.043) (0.042) (0.032) (0.033)
STU-TEA RATIO 0.00 -0.00 -0.00 0.00 0.00 0.00 -0.00 -0.00 0.00 -0.00 -0.00 -0.00
(0.002) (0.003) (0.003) (0.002) (0.003) (0.004) (0.003) (0.004) (0.004) (0.003) (0.005) (0.005)
SCHOOL CHOICE -0.08* -0.02 -0.04 -0.05 -0.11*** -0.07* -0.01 -0.03 0.01 -0.01 0.02 0.07
(0.030) (0.043) (0.043) (0.030) (0.033) (0.031) (0.034) (0.057) (0.049) (0.056) (0.041) (0.053)
PRESCORE (G7) 0.73*** 0.54*** 0.74*** 0.80*** 0.81*** 0.69*** 0.67*** 0.44*** 0.67*** 0.80*** 0.76*** 0.59***
(0.014) (0.035) (0.027) (0.013) (0.018) (0.019) (0.015) (0.029) (0.027) (0.019) (0.018) (0.026)
STUDY HOUR (1= >10hrs/week) 0.05 0.11* 0.08* 0.04 -0.02 0.06 -0.00 -0.07 -0.02 -0.00 0.05 0.03
(0.029) (0.047) (0.037) (0.037) (0.033) (0.035) (0.033) (0.057) (0.056) (0.046) (0.035) (0.030)
SELF ESTEEM 0.06** 0.00 0.03 0.06** 0.07*** 0.06* 0.02 -0.03 0.01 0.01 0.03 0.06*
(0.017) (0.032) (0.030) (0.020) (0.021) (0.025) (0.020) (0.031) (0.022) (0.027) (0.023) (0.028)
MATH PT in G7 0.00 -0.05 -0.01 0.01 0.01 0.01 -0.03 -0.03 -0.04 -0.02 -0.02 -0.03
(0.025) (0.040) (0.040) (0.044) (0.047) (0.042) (0.029) (0.037) (0.029) (0.036) (0.031) (0.051)
MATH PT in G9 0.25*** 0.13*** 0.22*** 0.28*** 0.33*** 0.29***
(0.032) (0.029) (0.047) (0.029) (0.037) (0.047)
Constant 0.23** -0.86*** -0.27** 0.18* 0.84*** 1.09*** 0.14 -0.87*** -0.48** 0.22 0.61*** 1.02***
(0.085) (0.139) (0.099) (0.074) (0.120) (0.129) (0.097) (0.149) (0.157) (0.138) (0.136) (0.162)

Observations 5,513 5,513 5,513 5,513 5,513 5,513 4,739 4,739 4,739 4,739 4,739 4,739
R-squared 0.471 0.419
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8MAT_Z (math score in grade 8) for (1)-(6); G9MAT_Z (math score in grade 9) for (7)-(12)

293
Table M.4. Heterogeneous effects of math tutoring in grade 8 on math achievement for 2 years using the 2SLAD estimation
Year 0 (Grade 8) Year 1 (Grade 9)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G8 (predicted) 0.45*** 0.39** 0.52*** 0.48*** 0.40*** 0.15 0.43*** 0.28 0.47*** 0.48*** 0.40** 0.37
(0.083) (0.120) (0.141) (0.090) (0.106) (0.157) (0.111) (0.189) (0.132) (0.131) (0.151) (0.189)
URBANICITY -0.09** -0.01 -0.04 -0.10* -0.10** -0.12*** -0.17*** -0.12** -0.20*** -0.18*** -0.14*** -0.09*
(0.030) (0.043) (0.044) (0.043) (0.039) (0.034) (0.034) (0.040) (0.033) (0.031) (0.033) (0.037)
GENDER (1=female) -0.14*** -0.03 -0.12*** -0.14*** -0.20*** -0.19*** -0.18*** -0.05 -0.14*** -0.22*** -0.21*** -0.23***
(0.021) (0.029) (0.030) (0.032) (0.027) (0.022) (0.024) (0.031) (0.030) (0.031) (0.027) (0.029)
SES 0.03* 0.01 0.03 0.07** 0.03 0.03 0.05** 0.04 0.05 0.07*** 0.04* 0.04*
(0.015) (0.032) (0.031) (0.024) (0.020) (0.031) (0.017) (0.030) (0.030) (0.019) (0.015) (0.021)
SCHOOL TYPE (1=private) 0.04 0.02 0.02 0.05 0.01 0.04 0.13*** 0.10* 0.13** 0.13** 0.14*** 0.12**
(0.026) (0.051) (0.037) (0.034) (0.028) (0.043) (0.030) (0.050) (0.042) (0.042) (0.023) (0.037)
STU-TEA RATIO -0.00 -0.00 -0.00 0.00 0.00 0.00 -0.00 0.00 0.00 -0.00 -0.00 -0.01
(0.003) (0.004) (0.003) (0.003) (0.004) (0.004) (0.003) (0.004) (0.003) (0.004) (0.003) (0.005)
SCHOOL CHOICE -0.07* 0.02 -0.01 -0.07 -0.07 -0.08 0.01 -0.00 0.01 0.03 0.03 0.08*
(0.033) (0.045) (0.029) (0.043) (0.047) (0.049) (0.037) (0.039) (0.047) (0.042) (0.030) (0.038)
PRESCORE (G7) 0.71*** 0.53*** 0.72*** 0.79*** 0.80*** 0.69*** 0.67*** 0.44*** 0.66*** 0.79*** 0.76*** 0.57***
(0.015) (0.033) (0.016) (0.019) (0.014) (0.027) (0.017) (0.042) (0.029) (0.020) (0.016) (0.022)
STUDY HOUR (1= >10hrs/week) 0.03 0.10 0.07 0.02 -0.03 0.07 -0.03 -0.08 -0.06 -0.03 0.04 0.04
(0.031) (0.056) (0.052) (0.037) (0.037) (0.047) (0.035) (0.066) (0.056) (0.058) (0.046) (0.038)
SELF ESTEEM 0.05* 0.01 0.01 0.04 0.06* 0.05* 0.01 -0.03 0.01 0.01 0.02 0.06*
(0.019) (0.035) (0.030) (0.020) (0.026) (0.025) (0.021) (0.033) (0.036) (0.030) (0.025) (0.026)
MATH PT in G7 -0.09* -0.13* -0.15* -0.09 -0.05 0.02 -0.13** -0.12 -0.15** -0.13* -0.13** -0.10
(0.040) (0.064) (0.069) (0.047) (0.061) (0.073) (0.044) (0.073) (0.058) (0.053) (0.049) (0.073)
MATH PT in G9 0.14** 0.01 0.08 0.15* 0.23*** 0.19*
(0.048) (0.072) (0.067) (0.063) (0.056) (0.093)
Constant 0.09 -1.11*** -0.53*** 0.04 0.66*** 1.16*** 0.04 -0.98*** -0.56*** 0.04 0.53*** 1.02***
(0.099) (0.107) (0.099) (0.070) (0.132) (0.143) (0.108) (0.142) (0.132) (0.156) (0.088) (0.131)

Observations 4,964 5,011 5,011 5,011 5,011 5,011 4,275 4,316 4,316 4,316 4,316 4,316
R-squared 0.455 0.405
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variables: G8MAT_Z (math score in grade 8) for (1)-(6); G9MAT_Z (math score in grade 9) for (7)-(12)

294
 295

Table M.5. Heterogeneous effects of math tutoring in grade 9 on math achievement in grade 9
using the OLS estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

MATH PT in G9 0.24*** 0.16*** 0.22*** 0.23*** 0.32*** 0.30***

(0.031) (0.035) (0.033) (0.047) (0.059) (0.055)
URBANICITY -0.15*** -0.19* -0.18*** -0.15*** -0.10** -0.08
(0.032) (0.076) (0.043) (0.031) (0.032) (0.043)
GENDER (1=female) -0.21*** -0.07* -0.18*** -0.27*** -0.25*** -0.25***
(0.022) (0.035) (0.029) (0.028) (0.029) (0.039)
SES 0.07*** 0.08*** 0.06*** 0.04* 0.04** 0.05
(0.015) (0.020) (0.015) (0.020) (0.015) (0.027)
SCHOOL TYPE (1=private) 0.11*** 0.10** 0.09* 0.09* 0.13*** 0.12**
(0.027) (0.036) (0.041) (0.042) (0.030) (0.042)
STU-TEA RATIO -0.00 -0.00 -0.00 0.00 0.00 -0.00
(0.003) (0.005) (0.003) (0.004) (0.004) (0.006)
SCHOOL CHOICE 0.01 -0.08 -0.01 0.01 0.05 0.06
(0.034) (0.074) (0.031) (0.030) (0.040) (0.043)
PRESCORE (G8) 0.68*** 0.51*** 0.70*** 0.82*** 0.75*** 0.58***
(0.015) (0.023) (0.020) (0.019) (0.018) (0.019)
STUDY HOUR (1= >10hrs/week) -0.00 -0.14* 0.02 0.02 0.03 0.03
(0.033) (0.071) (0.065) (0.055) (0.045) (0.048)
SELF ESTEEM 0.04* -0.00 0.04 0.05* 0.06* 0.05
(0.019) (0.036) (0.023) (0.025) (0.027) (0.038)
MATH PT in G7 0.01 -0.00 -0.02 0.04 0.03 -0.05
(0.029) (0.052) (0.039) (0.043) (0.038) (0.049)
MATH PT in G8 0.08** 0.04 0.06* 0.09** 0.11** 0.12
(0.031) (0.025) (0.028) (0.032) (0.041) (0.072)
Constant 0.11 -0.74*** -0.32** 0.13 0.42*** 1.06***
(0.095) (0.155) (0.103) (0.107) (0.102) (0.154)

Observations 4,697 4,697 4,697 4,697 4,697 4,697

R-squared 0.443
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9MAT_Z (math score in grade 9)
 296

Table M.6. Heterogeneous effects of math tutoring in grade 9 on math achievement in grade 9
using the 2SLAD estimation
Year 0 (Grade 9)
(1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90

MATH PT in G9 (predicted) -0.09 -0.17 -0.34 1.83 1.23 0.40

(0.287) (1.302) (1.575) (1.228) (1.336) (1.293)
URBANICITY -0.15*** -0.16** -0.14*** -0.11** -0.10** -0.05
(0.035) (0.059) (0.033) (0.037) (0.031) (0.033)
GENDER (1=female) -0.22*** -0.08 -0.19*** -0.20*** -0.22*** -0.26***
(0.026) (0.063) (0.056) (0.048) (0.040) (0.066)
SES 0.09*** 0.08 0.09 0.00 0.02 0.04
(0.020) (0.061) (0.060) (0.055) (0.053) (0.063)
SCHOOL TYPE (1=private) 0.13*** 0.09* 0.13*** 0.09* 0.14*** 0.14**
(0.030) (0.042) (0.036) (0.045) (0.032) (0.041)
STU-TEA RATIO 0.00 0.00 0.00 -0.01 -0.00 -0.00
(0.004) (0.009) (0.010) (0.009) (0.010) (0.007)
SCHOOL CHOICE 0.01 -0.05 0.02 0.05 0.05 0.08
(0.037) (0.047) (0.033) (0.044) (0.052) (0.056)
PRESCORE (G8) 0.70*** 0.51*** 0.74*** 0.78*** 0.73*** 0.57***
(0.019) (0.056) (0.063) (0.050) (0.055) (0.059)
STUDY HOUR (1= >10hrs/week) -0.00 -0.08 0.02 0.02 0.02 0.03
(0.036) (0.071) (0.052) (0.048) (0.036) (0.049)
SELF ESTEEM 0.05* -0.00 0.04 -0.02 0.01 0.04
(0.023) (0.055) (0.061) (0.057) (0.060) (0.056)
MATH PT in G7 0.05 0.06 0.06 -0.22 -0.12 -0.02
(0.053) (0.194) (0.250) (0.170) (0.194) (0.218)
MATH PT in G8 0.19 0.13 0.22 -0.48 -0.23 0.01
(0.103) (0.455) (0.536) (0.433) (0.466) (0.417)
Constant 0.22 -0.75 -0.21 -0.38 0.17 0.95
(0.136) (0.420) (0.484) (0.370) (0.396) (0.561)

Observations 4,073 4,491 4,491 4,491 4,491 4,491

R-squared 0.427
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: G9MAT_Z (math score in grade 9)
Table N.1. Long-term heterogeneous effects of one year private tutoring in middle school on the CSAT
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 IV q10 q25 q50 q75 q90

1 year of PT in middle school 0.03 -0.08 -0.11 -0.03 0.07 0.22 0.34 0.07 0.02 0.00 0.12 0.16
(0.180) (0.407) (0.169) (0.177) (0.270) (0.307) (0.482) (0.131) (0.078) (0.087) (0.085) (0.128)
URBAN -0.63* -0.68 -0.50 -0.70** -0.78* -0.31 -0.91** -0.07 -0.11 -0.19 -0.09 0.06
(0.272) (0.555) (0.355) (0.262) (0.374) (0.367) (0.308) (0.145) (0.105) (0.109) (0.119) (0.115)
GENDER (1=female) -0.05 0.02 -0.15 -0.03 -0.01 0.30 -0.24 0.19* 0.11 -0.04 -0.03 -0.13
(0.178) (0.492) (0.270) (0.181) (0.277) (0.311) (0.207) (0.091) (0.062) (0.056) (0.071) (0.112)
SES 0.02 -0.37 -0.13 0.31 -0.01 -0.07 -0.17 0.17** 0.24** 0.32*** 0.33*** 0.30***
(0.153) (0.418) (0.288) (0.251) (0.151) (0.248) (0.191) (0.066) (0.072) (0.051) (0.042) (0.066)
SCHOOL TYPE -0.16 -0.08 -0.45 0.07 0.10 -0.36 -0.13 0.08 0.06 0.01 0.04 0.10
(0.257) (0.517) (0.357) (0.316) (0.294) (0.304) (0.287) (0.120) (0.073) (0.103) (0.123) (0.141)
STU-TEA RATIO 0.03 0.02 0.04 0.02 0.03 0.01 0.03 0.02 0.02** 0.00 0.00 0.00
(0.018) (0.038) (0.029) (0.021) (0.026) (0.043) (0.020) (0.012) (0.007) (0.008) (0.009) (0.008)
SCHOOL CHOICE -0.34 -0.80 -0.50 -0.40 -0.37 -0.05 -0.53 -0.07 -0.08 -0.10 -0.10 -0.07
(0.280) (0.520) (0.382) (0.264) (0.369) (0.433) (0.315) (0.168) (0.101) (0.097) (0.102) (0.138)
PRESCORE 0.27*** 0.22 0.34*** 0.30*** 0.30** 0.34*** 0.32*** 0.27*** 0.27*** 0.28*** 0.29*** 0.24***
(0.054) (0.152) (0.081) (0.067) (0.107) (0.096) (0.061) (0.042) (0.029) (0.031) (0.021) (0.028)
SELF-STUDY -0.24 -0.82 0.08 -0.14 -0.43 -1.00* -0.26 0.20 0.01 0.03 0.27* 0.09
(0.332) (0.662) (0.303) (0.243) (0.363) (0.487) (0.381) (0.113) (0.097) (0.158) (0.124) (0.100)
SELF-ESTEEM 0.23 0.62 0.15 0.20 0.29 0.21 0.07 0.26*** 0.24*** 0.28*** 0.19** 0.25**
(0.155) (0.360) (0.178) (0.175) (0.214) (0.307) (0.177) (0.078) (0.070) (0.073) (0.066) (0.078)
PT in grade 10 0.10 0.22 0.27 0.10 0.01 -0.15 0.16 0.67*** 0.36*** 0.36*** 0.24* 0.01
(0.186) (0.377) (0.223) (0.175) (0.322) (0.364) (0.221) (0.126) (0.084) (0.076) (0.109) (0.148)
PT in grade 11 0.35 0.31 0.16 0.30 0.36 0.15 0.23 0.45*** 0.34*** 0.21** 0.28*** 0.20*
(0.206) (0.389) (0.273) (0.286) (0.360) (0.427) (0.242) (0.129) (0.066) (0.081) (0.075) (0.079)
PT in grade 12 0.54* 0.84* 0.42 0.68** 0.72* 0.66 0.60* 0.28 0.29*** 0.34*** 0.31*** 0.43***
(0.253) (0.414) (0.252) (0.240) (0.281) (0.422) (0.298) (0.148) (0.076) (0.101) (0.075) (0.115)
Constant 3.46*** 2.68* 2.53* 3.69*** 4.06*** 3.94*** 3.68*** 0.79 2.01*** 3.37*** 3.92*** 5.25***
(0.801) (1.317) (1.096) (0.724) (1.142) (1.078) (0.951) (0.408) (0.332) (0.309) (0.382) (0.365)

Observations 246 246 246 246 246 246 181 2,128 2,128 2,128 2,128 2,128
R-squared 0.275 0.294
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05

297
Dependent variable: CSAT_avg (average score on the CSAT)
Table N.2. Long-term heterogeneous effects of two years private tutoring in middle school on the CSAT
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 IV q10 q25 q50 q75 q90

2 years of PT in middle school 0.13 0.36 0.09 0.10 0.15 -0.05 0.36 0.12 0.04 0.01 0.21 0.27
(0.166) (0.440) (0.184) (0.215) (0.331) (0.276) (0.410) (0.178) (0.147) (0.124) (0.177) (0.321)
URBAN -0.07 0.15 -0.10 -0.29 -0.23 0.21 -0.12 -0.07 -0.11 -0.19* -0.08 0.07
(0.206) (0.525) (0.214) (0.215) (0.267) (0.300) (0.232) (0.176) (0.083) (0.095) (0.124) (0.166)
GENDER (1=female) 0.13 0.36 0.25 0.04 0.12 -0.01 0.09 0.19 0.11 -0.04 -0.02 -0.13
(0.138) (0.306) (0.236) (0.188) (0.174) (0.272) (0.162) (0.115) (0.083) (0.070) (0.081) (0.093)
SES 0.05 -0.09 0.09 0.16 0.11 -0.05 0.06 0.16 0.23*** 0.31*** 0.32*** 0.28***
(0.104) (0.348) (0.173) (0.158) (0.130) (0.147) (0.117) (0.088) (0.066) (0.061) (0.047) (0.076)
SCHOOL TYPE 0.31 -0.09 0.10 0.17 0.45 0.77 0.17 0.07 0.06 0.01 0.02 0.08
(0.175) (0.440) (0.237) (0.253) (0.319) (0.426) (0.197) (0.125) (0.082) (0.093) (0.065) (0.093)
STU-TEA RATIO -0.00 0.00 0.00 -0.00 -0.01 0.00 -0.00 0.02 0.02** 0.00 0.00 0.00
(0.014) (0.034) (0.014) (0.014) (0.021) (0.018) (0.016) (0.014) (0.007) (0.007) (0.010) (0.016)
SCHOOL CHOICE -0.14 -0.00 -0.16 -0.35 -0.41 -0.07 -0.21 -0.07 -0.08 -0.10 -0.10 -0.07
(0.213) (0.683) (0.280) (0.253) (0.304) (0.295) (0.238) (0.199) (0.068) (0.063) (0.099) (0.108)
PRESCORE 0.24*** 0.26** 0.26*** 0.27*** 0.24*** 0.23** 0.28*** 0.27*** 0.27*** 0.28*** 0.29*** 0.24***
(0.039) (0.090) (0.040) (0.050) (0.067) (0.077) (0.046) (0.042) (0.024) (0.018) (0.021) (0.026)
SELF-STUDY 0.17 0.52* 0.15 -0.10 0.20 -0.15 0.11 0.20* 0.01 0.03 0.27*** 0.08
(0.231) (0.262) (0.236) (0.253) (0.386) (0.425) (0.259) (0.092) (0.063) (0.110) (0.079) (0.117)
SELF-ESTEEM 0.24 0.30 0.19 0.22 0.28 0.23 0.11 0.26** 0.24*** 0.28*** 0.20** 0.25***
(0.130) (0.247) (0.213) (0.203) (0.174) (0.171) (0.152) (0.090) (0.061) (0.075) (0.066) (0.062)
PT in grade 10 0.03 0.21 0.23 0.11 -0.08 -0.21 0.09 0.67*** 0.36*** 0.36** 0.23* -0.00
(0.155) (0.297) (0.181) (0.205) (0.298) (0.327) (0.191) (0.179) (0.092) (0.111) (0.110) (0.151)
PT in grade 11 0.21 0.56 0.17 0.06 0.32 -0.29 -0.01 0.45*** 0.34*** 0.21* 0.27* 0.18
(0.158) (0.333) (0.142) (0.201) (0.285) (0.325) (0.189) (0.121) (0.064) (0.086) (0.124) (0.140)
PT in grade 12 0.36* -0.14 0.28 0.44 0.43 0.68* 0.45* 0.28* 0.29** 0.34*** 0.31*** 0.43**
(0.175) (0.416) (0.275) (0.257) (0.250) (0.289) (0.199) (0.131) (0.097) (0.094) (0.084) (0.134)
Constant 3.40*** 0.64 2.29*** 3.85*** 4.84*** 5.53*** 3.36*** 0.77 2.01*** 3.37*** 3.89*** 5.21***
(0.598) (2.029) (0.592) (0.610) (0.945) (0.783) (0.712) (0.638) (0.254) (0.221) (0.354) (0.602)

Observations 421 421 421 421 421 421 325 2,128 2,128 2,128 2,128 2,128
R-squared 0.181 0.180
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

298
Table N.3. Long-term heterogeneous effects of three years private tutoring in middle school on the CSAT achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of PT in middle school 0.04 0.27 0.02 -0.01 0.09 -0.12 0.48 0.28 0.09 0.01 0.50 0.64
(0.134) (0.431) (0.200) (0.144) (0.283) (0.343) (0.456) (0.641) (0.380) (0.464) (0.314) (0.490)
URBAN -0.03 -0.01 -0.00 -0.02 -0.02 -0.03 -0.15 -0.07 -0.11 -0.19 -0.08 0.07
(0.092) (0.176) (0.092) (0.126) (0.098) (0.115) (0.103) (0.165) (0.088) (0.125) (0.105) (0.146)
GENDER (1=female) 0.05 0.19 0.12 0.01 0.01 -0.28** 0.03 0.20 0.11 -0.04 -0.02 -0.12
(0.064) (0.132) (0.094) (0.060) (0.082) (0.104) (0.072) (0.139) (0.057) (0.072) (0.093) (0.100)
SES 0.31*** 0.20*** 0.25*** 0.36*** 0.39*** 0.38*** 0.25*** 0.16 0.23*** 0.31*** 0.31*** 0.28***
(0.043) (0.052) (0.053) (0.056) (0.067) (0.071) (0.051) (0.085) (0.043) (0.060) (0.045) (0.062)
SCHOOL TYPE 0.05 0.09 0.12 0.00 0.04 0.10 -0.04 0.07 0.06 0.01 0.03 0.09
(0.081) (0.152) (0.111) (0.096) (0.103) (0.130) (0.091) (0.114) (0.084) (0.085) (0.100) (0.131)
STU-TEA RATIO 0.00 0.01 0.02 -0.01 -0.01 0.00 0.01 0.02 0.02* 0.00 0.00 -0.00
(0.008) (0.015) (0.014) (0.009) (0.014) (0.014) (0.009) (0.012) (0.009) (0.007) (0.012) (0.013)
SCHOOL CHOICE -0.04 0.11 0.04 -0.01 -0.07 -0.15 -0.12 -0.07 -0.08 -0.10 -0.09 -0.07
(0.100) (0.217) (0.126) (0.103) (0.096) (0.135) (0.110) (0.216) (0.102) (0.144) (0.104) (0.116)
PRESCORE 0.26*** 0.21*** 0.24*** 0.28*** 0.29*** 0.23*** 0.26*** 0.27*** 0.27*** 0.28*** 0.29*** 0.24***
(0.020) (0.032) (0.020) (0.022) (0.028) (0.045) (0.022) (0.026) (0.018) (0.018) (0.020) (0.023)
SELF-STUDY 0.15 0.28** 0.14 0.13 0.21 -0.04 0.15 0.19 0.01 0.03 0.26*** 0.07
(0.087) (0.096) (0.091) (0.150) (0.111) (0.090) (0.095) (0.129) (0.061) (0.104) (0.078) (0.123)
SELF-ESTEEM 0.31*** 0.35*** 0.32*** 0.36*** 0.27** 0.40*** 0.28*** 0.26*** 0.23*** 0.28*** 0.18** 0.24***
(0.057) (0.084) (0.084) (0.088) (0.095) (0.087) (0.065) (0.068) (0.042) (0.050) (0.064) (0.060)
PT in grade 10 0.35*** 0.70* 0.33** 0.33* 0.31** 0.19 0.32** 0.64** 0.35* 0.36** 0.17 -0.07
(0.097) (0.328) (0.100) (0.139) (0.118) (0.174) (0.122) (0.212) (0.147) (0.114) (0.115) (0.099)
PT in grade 11 0.33*** 0.58*** 0.39*** 0.29* 0.29* 0.17 0.32** 0.44*** 0.34*** 0.21* 0.26** 0.18*
(0.087) (0.097) (0.097) (0.139) (0.113) (0.172) (0.101) (0.125) (0.067) (0.084) (0.081) (0.076)
PT in grade 12 0.29*** 0.28* 0.27** 0.30*** 0.28* 0.43*** 0.34*** 0.29* 0.29*** 0.34*** 0.32*** 0.45***
(0.072) (0.120) (0.092) (0.084) (0.113) (0.126) (0.079) (0.137) (0.065) (0.064) (0.080) (0.113)
Constant 3.24*** 0.68 2.06*** 3.32*** 4.10*** 5.94*** 2.95*** 0.69 1.98*** 3.36*** 3.73*** 5.00***
(0.316) (0.875) (0.474) (0.311) (0.374) (0.626) (0.455) (0.766) (0.415) (0.374) (0.335) (0.448)

Observations 1,452 1,452 1,452 1,452 1,452 1,452 1,190 2,128 2,128 2,128 2,128 2,128
R-squared 0.318 0.310
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

299
Table N.4. Long-term heterogeneous effects of one year verbal tutoring in middle school on the CSAT verbal achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of Verbal PT in middle school -0.17 -0.06 -0.14 -0.29 -0.24 -0.42** 0.52* -0.28* -0.19 -0.19 -0.22 -0.17
(0.106) (0.179) (0.178) (0.148) (0.180) (0.160) (0.256) (0.136) (0.174) (0.158) (0.173) (0.255)
URBAN -0.22 -0.19 -0.17 -0.32 -0.27 -0.26 -0.28 -0.08 -0.07 -0.09 0.11 -0.09
(0.158) (0.232) (0.307) (0.190) (0.217) (0.184) (0.180) (0.165) (0.141) (0.106) (0.109) (0.155)
GENDER (1=female) 0.20 0.33* 0.08 0.23 0.20 0.01 0.21 0.47*** 0.24*** 0.23** 0.11 -0.06
(0.107) (0.162) (0.177) (0.170) (0.175) (0.190) (0.121) (0.109) (0.069) (0.073) (0.100) (0.082)
SES 0.29*** 0.25* 0.28 0.36*** 0.26* 0.29*** 0.27*** 0.25** 0.18** 0.21*** 0.24*** 0.24***
(0.068) (0.128) (0.152) (0.072) (0.115) (0.074) (0.079) (0.078) (0.057) (0.044) (0.057) (0.066)
SCHOOL TYPE 0.01 -0.14 -0.03 -0.12 0.08 0.17 -0.17 -0.09 0.08 0.02 -0.01 0.02
(0.133) (0.192) (0.266) (0.195) (0.282) (0.259) (0.154) (0.175) (0.114) (0.090) (0.105) (0.104)
STU-TEA RATIO 0.02 0.04** 0.03 0.02 0.02 0.01 0.01 0.02 0.02* 0.01 0.00 0.01
(0.012) (0.012) (0.025) (0.016) (0.023) (0.011) (0.014) (0.014) (0.009) (0.010) (0.012) (0.016)
SCHOOL CHOICE 0.00 -0.07 0.06 -0.03 0.06 -0.05 -0.08 0.04 -0.00 0.00 0.07 -0.07
(0.174) (0.271) (0.386) (0.188) (0.269) (0.190) (0.198) (0.151) (0.121) (0.073) (0.112) (0.151)
PRESCORE 0.31*** 0.36*** 0.30*** 0.31*** 0.33*** 0.29*** 0.31*** 0.32*** 0.29*** 0.31*** 0.30*** 0.36***
(0.032) (0.075) (0.058) (0.045) (0.051) (0.056) (0.037) (0.041) (0.034) (0.032) (0.024) (0.028)
SELF-STUDY -0.06 -0.27 0.05 0.15 0.07 -0.19 -0.03 -0.21 0.10 0.16 0.02 0.06
(0.158) (0.263) (0.241) (0.158) (0.159) (0.176) (0.176) (0.135) (0.148) (0.109) (0.114) (0.132)
SELF-ESTEEM 0.33*** 0.43 0.35* 0.33*** 0.31** 0.24 0.28** 0.37*** 0.27*** 0.29*** 0.28** 0.15
(0.094) (0.317) (0.163) (0.093) (0.103) (0.127) (0.108) (0.082) (0.080) (0.069) (0.086) (0.095)
Verbal PT in grade 10 -0.10 -0.00 -0.12 -0.08 -0.00 -0.07 -0.18 0.16 0.08 0.04 0.05 -0.06
(0.115) (0.148) (0.177) (0.142) (0.150) (0.136) (0.136) (0.112) (0.090) (0.103) (0.121) (0.123)
Verbal PT in grade 11 0.13 0.04 0.00 0.16 0.26 0.41** 0.05 0.02 0.05 0.08 0.10 0.06
(0.143) (0.267) (0.299) (0.168) (0.266) (0.149) (0.163) (0.141) (0.100) (0.101) (0.098) (0.123)
Verbal PT in grade 12 0.57** 0.27 0.82 0.56 0.37 0.59** 0.53** 0.35* 0.18 0.39* 0.36* 0.20
(0.180) (0.247) (0.437) (0.302) (0.351) (0.200) (0.200) (0.148) (0.139) (0.158) (0.145) (0.211)
Constant 3.34*** 0.80 2.34* 3.50*** 4.25*** 5.88*** 3.35*** 1.01 2.30*** 3.27*** 4.55*** 5.60***
(0.492) (0.685) (1.167) (0.761) (0.782) (0.757) (0.568) (0.533) (0.396) (0.257) (0.353) (0.496)

Observations 791 791 791 791 791 791 635 2,178 2,178 2,178 2,178 2,178
R-squared 0.254 0.204
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

300
Table N.5. Long-term heterogeneous effects of two years verbal tutoring in middle school on the CSAT verbal achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of Verbal PT in middle school -0.21* -0.25 -0.13 -0.30** -0.39** -0.21 -0.28 -0.29 -0.20 -0.21 -0.23 -0.18
(0.103) (0.195) (0.123) (0.114) (0.143) (0.158) (0.222) (0.206) (0.169) (0.108) (0.217) (0.169)
URBAN -0.14 -0.24 -0.01 -0.43* -0.18 -0.18 -0.14 -0.09 -0.07 -0.09 0.10 -0.09
(0.144) (0.354) (0.278) (0.206) (0.202) (0.212) (0.160) (0.174) (0.158) (0.091) (0.155) (0.127)
GENDER (1=female) 0.40*** 0.44* 0.39* 0.38** 0.51** 0.25 0.39*** 0.45*** 0.23** 0.22*** 0.10 -0.07
(0.100) (0.185) (0.158) (0.129) (0.170) (0.166) (0.115) (0.125) (0.081) (0.065) (0.103) (0.102)
SES 0.26*** 0.38** 0.16 0.23* 0.18** 0.30** 0.24** 0.25** 0.19*** 0.21*** 0.24*** 0.24**
(0.067) (0.118) (0.131) (0.095) (0.063) (0.096) (0.075) (0.089) (0.056) (0.053) (0.046) (0.078)
SCHOOL TYPE 0.03 -0.21 0.05 -0.12 0.14 0.14 0.01 -0.08 0.08 0.02 -0.01 0.02
(0.128) (0.269) (0.272) (0.169) (0.188) (0.270) (0.144) (0.174) (0.084) (0.058) (0.114) (0.082)
STU-TEA RATIO 0.01 0.01 0.03* 0.01 0.00 -0.02 0.01 0.02 0.02 0.01 0.00 0.01
(0.011) (0.019) (0.012) (0.017) (0.010) (0.018) (0.013) (0.014) (0.012) (0.012) (0.010) (0.015)
SCHOOL CHOICE -0.14 -0.16 -0.02 -0.31 -0.12 -0.34 -0.14 0.04 -0.00 0.00 0.07 -0.07
(0.154) (0.315) (0.274) (0.211) (0.169) (0.237) (0.171) (0.215) (0.173) (0.097) (0.148) (0.136)
PRESCORE 0.31*** 0.27*** 0.32*** 0.30*** 0.32*** 0.35*** 0.31*** 0.32*** 0.29*** 0.31*** 0.30*** 0.36***
(0.029) (0.050) (0.039) (0.041) (0.048) (0.049) (0.033) (0.045) (0.022) (0.023) (0.026) (0.043)
SELF-STUDY -0.06 -0.19 0.11 0.06 -0.05 -0.31 -0.02 -0.20 0.11 0.17* 0.03 0.07
(0.141) (0.269) (0.291) (0.121) (0.175) (0.299) (0.156) (0.158) (0.139) (0.082) (0.119) (0.176)
SELF-ESTEEM 0.31*** 0.39* 0.47*** 0.34*** 0.26* 0.09 0.25* 0.37** 0.27*** 0.29*** 0.27** 0.15
(0.089) (0.189) (0.128) (0.088) (0.130) (0.134) (0.101) (0.114) (0.067) (0.064) (0.086) (0.102)
Verbal PT in grade 10 -0.01 -0.10 -0.11 0.14 0.01 -0.07 0.07 0.15 0.08 0.04 0.04 -0.06
(0.106) (0.210) (0.143) (0.135) (0.130) (0.132) (0.126) (0.118) (0.088) (0.093) (0.111) (0.116)
Verbal PT in grade 11 -0.02 0.04 -0.12 -0.01 0.20 -0.02 -0.11 0.02 0.05 0.08 0.10 0.06
(0.129) (0.244) (0.219) (0.197) (0.152) (0.192) (0.145) (0.126) (0.110) (0.064) (0.103) (0.087)
Verbal PT in grade 12 0.40* 0.56*** 0.39 0.23 0.36* 0.34 0.52** 0.33* 0.17 0.37** 0.34** 0.18
(0.167) (0.162) (0.259) (0.237) (0.178) (0.232) (0.181) (0.152) (0.141) (0.139) (0.129) (0.203)
Constant 3.38*** 1.74* 1.72* 3.82*** 4.32*** 6.29*** 3.38*** 1.01* 2.30*** 3.28*** 4.55*** 5.60***
(0.449) (0.884) (0.732) (0.569) (0.598) (0.821) (0.514) (0.467) (0.392) (0.307) (0.469) (0.425)

Observations 925 925 925 925 925 925 722 2,178 2,178 2,178 2,178 2,178
R-squared 0.222 0.214
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

301
Table N.6. Long-term heterogeneous effects of three years verbal tutoring in middle school on the CSAT verbal achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of Verbal PT in middle school -0.26** -0.13 -0.11 -0.26 -0.35* -0.39** -0.47* -0.41* -0.28 -0.29 -0.32 -0.25
(0.098) (0.219) (0.102) (0.138) (0.142) (0.148) (0.192) (0.202) (0.205) (0.152) (0.198) (0.246)
URBAN -0.13 -0.10 -0.16 -0.14 -0.09 -0.18 -0.14 -0.10 -0.08 -0.10 0.10 -0.10
(0.127) (0.194) (0.179) (0.114) (0.148) (0.227) (0.140) (0.201) (0.153) (0.121) (0.167) (0.193)
GENDER (1=female) 0.18* 0.19 0.26 0.21 0.04 -0.05 0.13 0.44*** 0.22** 0.21** 0.09 -0.08
(0.088) (0.196) (0.136) (0.123) (0.190) (0.176) (0.099) (0.097) (0.073) (0.064) (0.086) (0.113)
SES 0.19** 0.14 0.19* 0.10 0.25* 0.24** 0.19** 0.26** 0.19*** 0.21*** 0.25*** 0.25**
(0.060) (0.118) (0.075) (0.090) (0.102) (0.081) (0.066) (0.084) (0.054) (0.042) (0.041) (0.085)
SCHOOL TYPE -0.02 0.29 0.29* -0.02 -0.18 -0.22 -0.15 -0.07 0.09 0.03 0.00 0.03
(0.114) (0.205) (0.115) (0.129) (0.227) (0.145) (0.126) (0.175) (0.100) (0.084) (0.139) (0.123)
STU-TEA RATIO 0.02* 0.02 0.03* 0.02 0.01 0.02** 0.02* 0.02 0.02 0.02 0.00 0.01
(0.011) (0.015) (0.014) (0.012) (0.013) (0.009) (0.012) (0.013) (0.013) (0.010) (0.011) (0.017)
SCHOOL CHOICE -0.08 -0.17 -0.24 -0.05 0.12 0.07 -0.04 0.03 -0.01 -0.00 0.06 -0.08
(0.139) (0.229) (0.204) (0.166) (0.192) (0.166) (0.151) (0.177) (0.117) (0.130) (0.150) (0.178)
PRESCORE 0.29*** 0.29*** 0.26*** 0.27*** 0.33*** 0.35*** 0.30*** 0.32*** 0.29*** 0.31*** 0.30*** 0.36***
(0.027) (0.032) (0.038) (0.037) (0.034) (0.039) (0.030) (0.038) (0.028) (0.025) (0.025) (0.028)
SELF-STUDY -0.02 -0.17 0.15 0.03 0.01 -0.07 0.05 -0.21 0.10 0.16 0.02 0.06
(0.127) (0.252) (0.141) (0.174) (0.151) (0.232) (0.137) (0.121) (0.142) (0.117) (0.149) (0.139)
SELF-ESTEEM 0.43*** 0.52*** 0.45*** 0.50*** 0.34*** 0.23 0.38*** 0.39** 0.28** 0.30*** 0.29*** 0.16
(0.080) (0.151) (0.103) (0.070) (0.095) (0.146) (0.089) (0.118) (0.087) (0.052) (0.075) (0.093)
Verbal PT in grade 10 0.10 0.16 0.13 0.08 0.11 0.06 0.15 0.20 0.11 0.07 0.08 -0.04
(0.095) (0.182) (0.104) (0.172) (0.150) (0.135) (0.113) (0.146) (0.095) (0.090) (0.104) (0.128)
Verbal PT in grade 11 0.04 0.12 -0.07 0.06 0.15 0.17 0.02 0.02 0.05 0.08 0.10 0.06
(0.113) (0.202) (0.141) (0.170) (0.188) (0.164) (0.124) (0.136) (0.104) (0.093) (0.122) (0.113)
Verbal PT in grade 12 0.32* 0.11 -0.04 0.43* 0.39 0.30 0.36* 0.33 0.17 0.37*** 0.34*** 0.18
(0.141) (0.308) (0.272) (0.202) (0.269) (0.265) (0.150) (0.184) (0.095) (0.093) (0.091) (0.147)
Constant 3.52*** 1.76** 2.49*** 3.70*** 4.56*** 5.23*** 3.55*** 1.03* 2.32*** 3.29*** 4.56*** 5.61***
(0.416) (0.635) (0.661) (0.441) (0.689) (0.609) (0.467) (0.453) (0.409) (0.414) (0.393) (0.468)

Observations 1,114 1,114 1,114 1,114 1,114 1,114 909 2,178 2,178 2,178 2,178 2,178
R-squared 0.212 0.210
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05

302
Dependent variable: CSAT_V (verbal score on the CSAT)
Table N.7. Long-term heterogeneous effects of one year English tutoring in middle school on the CSAT English achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of English PT in middle school -0.18 -0.19 -0.15 -0.09 -0.31 -0.36 -0.02 0.36* 0.26* 0.33*** 0.30** 0.61*
(0.180) (0.460) (0.323) (0.254) (0.310) (0.419) (0.521) (0.161) (0.130) (0.084) (0.113) (0.261)
URBAN -0.44 -0.77 -0.33 -0.28 -0.23 -0.41 -0.62* 0.04 -0.23* -0.26** -0.13 0.02
(0.273) (0.567) (0.436) (0.263) (0.438) (0.484) (0.305) (0.230) (0.107) (0.083) (0.120) (0.172)
GENDER (1=female) 0.38* 0.28 0.40 0.31 0.53* 0.59 0.17 0.40*** 0.24*** 0.18** 0.06 0.01
(0.178) (0.419) (0.233) (0.165) (0.229) (0.312) (0.203) (0.110) (0.056) (0.066) (0.106) (0.111)
SES 0.15 0.09 0.22 0.36* 0.35 0.07 0.03 0.26* 0.31*** 0.43*** 0.44*** 0.36***
(0.139) (0.335) (0.167) (0.164) (0.215) (0.269) (0.167) (0.109) (0.066) (0.067) (0.052) (0.081)
SCHOOL TYPE 0.38 0.14 0.48 0.51 0.15 0.59 0.21 -0.00 0.20* 0.09 0.02 0.26
(0.234) (0.797) (0.299) (0.384) (0.384) (0.540) (0.262) (0.195) (0.092) (0.116) (0.136) (0.181)
STU-TEA RATIO 0.02 0.03 0.02 0.00 0.02 0.06 0.02 0.02 0.02 0.01 0.00 -0.00
(0.018) (0.045) (0.031) (0.021) (0.016) (0.034) (0.021) (0.019) (0.010) (0.008) (0.010) (0.017)
SCHOOL CHOICE -0.29 -0.73 -0.32 -0.35 -0.01 -0.04 -0.29 -0.00 -0.21* -0.21* -0.14 -0.26
(0.284) (0.496) (0.384) (0.378) (0.470) (0.570) (0.317) (0.241) (0.100) (0.097) (0.130) (0.150)
PRESCORE 0.25*** 0.18 0.28*** 0.23** 0.25** 0.34** 0.30*** 0.31*** 0.32*** 0.31*** 0.28*** 0.21***
(0.052) (0.106) (0.084) (0.070) (0.094) (0.129) (0.060) (0.052) (0.016) (0.024) (0.034) (0.042)
SELF-STUDY -0.13 -0.87 -0.28 0.00 0.13 -0.45 0.02 0.25 0.05 0.04 0.14 0.18
(0.323) (0.930) (0.471) (0.702) (0.410) (0.363) (0.378) (0.176) (0.127) (0.096) (0.148) (0.134)
SELF-ESTEEM 0.47** 0.32 0.50 0.52** 0.45* 0.41 0.32 0.39*** 0.29*** 0.32*** 0.30*** 0.29**
(0.156) (0.414) (0.262) (0.188) (0.179) (0.273) (0.179) (0.082) (0.067) (0.087) (0.086) (0.112)
English PT in grade 10 0.04 0.32 0.33 0.01 -0.08 -0.09 0.09 0.50** 0.26** 0.11 -0.06 -0.06
(0.188) (0.461) (0.206) (0.190) (0.249) (0.265) (0.221) (0.156) (0.085) (0.119) (0.138) (0.157)
English PT in grade 11 0.44* 0.89 0.28 0.46 0.38 0.62 0.42 0.51*** 0.19* 0.09 0.19 0.15
(0.216) (0.455) (0.383) (0.407) (0.361) (0.354) (0.266) (0.147) (0.083) (0.098) (0.112) (0.161)
English PT in grade 12 0.35 0.75 0.05 0.14 0.74* 0.34 0.15 0.20 0.29** 0.24** 0.13 -0.05
(0.289) (0.578) (0.450) (0.502) (0.363) (0.504) (0.338) (0.143) (0.095) (0.085) (0.071) (0.112)
Constant 3.16*** 2.13 1.92 3.73*** 3.44* 2.99 3.14** -0.33 1.75*** 2.92*** 4.30*** 5.67***
(0.819) (1.439) (1.117) (1.006) (1.423) (1.844) (0.959) (0.717) (0.249) (0.240) (0.318) (0.655)

Observations 343 343 343 343 343 343 260 2,172 2,172 2,172 2,172 2,172
R-squared 0.203 0.203
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

303
Table N.8. Long-term heterogeneous effects of two years English tutoring in middle school on the CSAT English achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of English PT in middle school -0.13 -0.45 -0.30 -0.16 -0.37 -0.19 0.54 0.45 0.32 0.42** 0.38* 0.77*
(0.163) (0.403) (0.266) (0.167) (0.285) (0.283) (0.408) (0.272) (0.258) (0.136) (0.185) (0.362)
URBAN 0.03 0.63 -0.15 -0.15 -0.14 -0.22 -0.19 0.07 -0.21* -0.23* -0.11 0.07
(0.189) (0.389) (0.327) (0.175) (0.215) (0.388) (0.220) (0.203) (0.087) (0.099) (0.171) (0.217)
GENDER (1=female) 0.14 0.39 0.21 0.10 0.04 -0.28 0.18 0.40** 0.24** 0.18* 0.06 0.02
(0.132) (0.231) (0.117) (0.149) (0.100) (0.292) (0.154) (0.125) (0.092) (0.070) (0.059) (0.091)
SES 0.30** 0.38 0.54*** 0.32*** 0.36** 0.21 0.23* 0.23** 0.29*** 0.40*** 0.42*** 0.31**
(0.096) (0.231) (0.112) (0.091) (0.111) (0.142) (0.112) (0.081) (0.070) (0.052) (0.062) (0.112)
SCHOOL TYPE 0.16 -0.24 0.20 0.21 -0.03 0.39 0.03 -0.02 0.19* 0.07 0.01 0.23
(0.173) (0.337) (0.264) (0.242) (0.281) (0.391) (0.200) (0.142) (0.091) (0.098) (0.163) (0.162)
STU-TEA RATIO -0.00 -0.00 0.00 -0.00 -0.00 0.02 -0.01 0.02 0.02 0.01 0.01 0.00
(0.015) (0.031) (0.024) (0.014) (0.020) (0.028) (0.017) (0.020) (0.014) (0.011) (0.010) (0.011)
SCHOOL CHOICE -0.34 0.35 -0.51 -0.48** -0.73** -0.54* -0.49* 0.00 -0.21* -0.20* -0.13 -0.25
(0.201) (0.439) (0.350) (0.179) (0.272) (0.247) (0.233) (0.188) (0.103) (0.096) (0.177) (0.189)
PRESCORE 0.27*** 0.35*** 0.32*** 0.30*** 0.33*** 0.20** 0.28*** 0.30*** 0.31*** 0.30*** 0.28*** 0.20***
(0.039) (0.068) (0.035) (0.056) (0.056) (0.077) (0.047) (0.045) (0.025) (0.023) (0.028) (0.026)
SELF-STUDY 0.19 -0.00 0.13 0.25 0.13 0.24 0.05 0.25 0.05 0.04 0.14 0.17
(0.220) (0.405) (0.372) (0.284) (0.258) (0.349) (0.261) (0.157) (0.096) (0.063) (0.128) (0.152)
SELF-ESTEEM 0.30* 0.55* 0.35 0.18 0.24 0.35* 0.22 0.40** 0.29*** 0.32*** 0.30*** 0.30**
(0.119) (0.240) (0.188) (0.164) (0.131) (0.155) (0.143) (0.134) (0.051) (0.072) (0.079) (0.112)
English PT in grade 10 0.05 -0.02 0.03 0.14 0.01 0.16 0.01 0.47** 0.24* 0.09 -0.09 -0.11
(0.142) (0.304) (0.159) (0.186) (0.161) (0.272) (0.174) (0.181) (0.108) (0.070) (0.101) (0.160)
English PT in grade 11 0.42** 0.61 0.43** 0.39 0.39 0.31 0.34 0.51** 0.19 0.09 0.20 0.15
(0.154) (0.354) (0.161) (0.273) (0.258) (0.347) (0.184) (0.167) (0.106) (0.104) (0.123) (0.120)
English PT in grade 12 0.15 0.64 0.24 0.07 0.13 -0.07 0.12 0.20 0.29** 0.24* 0.13 -0.05
(0.183) (0.341) (0.172) (0.203) (0.284) (0.333) (0.213) (0.172) (0.099) (0.096) (0.129) (0.126)
Constant 3.69*** -0.10 2.79** 3.85*** 5.43*** 6.91*** 3.58*** -0.36 1.73*** 2.89*** 4.28*** 5.62***
(0.587) (1.178) (0.895) (0.537) (0.797) (1.020) (0.713) (0.766) (0.502) (0.409) (0.443) (0.584)

Observations 595 595 595 595 595 595 462 2,172 2,172 2,172 2,172 2,172
R-squared 0.204 0.173
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

304
Table N.9. Long-term heterogeneous effects of three years English tutoring in middle school on the CSAT English achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of English PT in middle school 0.09 -0.00 0.11 0.19 0.02 0.03 -0.47* -0.41* -0.28 -0.29 -0.32 -0.25
(0.132) (0.379) (0.219) (0.151) (0.169) (0.164) (0.192) (0.202) (0.205) (0.152) (0.198) (0.246)
URBAN 0.02 0.05 -0.12 -0.08 0.09 0.19 -0.14 -0.10 -0.08 -0.10 0.10 -0.10
(0.103) (0.253) (0.121) (0.132) (0.139) (0.174) (0.140) (0.201) (0.153) (0.121) (0.167) (0.193)
GENDER (1=female) 0.21** 0.32*** 0.30*** 0.25*** 0.06 0.08 0.13 0.44*** 0.22** 0.21** 0.09 -0.08
(0.071) (0.081) (0.072) (0.074) (0.088) (0.115) (0.099) (0.097) (0.073) (0.064) (0.086) (0.113)
SES 0.41*** 0.22 0.33*** 0.49*** 0.47*** 0.53*** 0.19** 0.26** 0.19*** 0.21*** 0.25*** 0.25**
(0.049) (0.118) (0.077) (0.068) (0.070) (0.081) (0.066) (0.084) (0.054) (0.042) (0.041) (0.085)
SCHOOL TYPE 0.07 0.04 0.22 0.08 0.10 -0.00 -0.15 -0.07 0.09 0.03 0.00 0.03
(0.090) (0.203) (0.147) (0.172) (0.142) (0.096) (0.126) (0.175) (0.100) (0.084) (0.139) (0.123)
STU-TEA RATIO 0.01 0.01 0.02 0.01 0.01 0.00 0.02* 0.02 0.02 0.02 0.00 0.01
(0.009) (0.016) (0.014) (0.013) (0.013) (0.019) (0.012) (0.013) (0.013) (0.010) (0.011) (0.017)
SCHOOL CHOICE -0.11 -0.18 -0.13 -0.06 -0.10 -0.25 -0.04 0.03 -0.01 -0.00 0.06 -0.08
(0.111) (0.270) (0.116) (0.114) (0.140) (0.207) (0.151) (0.177) (0.117) (0.130) (0.150) (0.178)
PRESCORE 0.28*** 0.29*** 0.29*** 0.31*** 0.26*** 0.18*** 0.30*** 0.32*** 0.29*** 0.31*** 0.30*** 0.36***
(0.023) (0.063) (0.028) (0.035) (0.040) (0.051) (0.030) (0.038) (0.028) (0.025) (0.025) (0.028)
SELF-STUDY 0.12 0.27 -0.02 -0.09 0.18 0.18 0.05 -0.21 0.10 0.16 0.02 0.06
(0.097) (0.161) (0.130) (0.110) (0.113) (0.169) (0.137) (0.121) (0.142) (0.117) (0.149) (0.139)
SELF-ESTEEM 0.40*** 0.51*** 0.45*** 0.39** 0.40*** 0.36** 0.38*** 0.39** 0.28** 0.30*** 0.29*** 0.16
(0.064) (0.130) (0.097) (0.118) (0.109) (0.131) (0.089) (0.118) (0.087) (0.052) (0.075) (0.093)
English PT in grade 10 0.13 0.37 0.18 0.13 -0.04 0.02 0.15 0.20 0.11 0.07 0.08 -0.04
(0.087) (0.232) (0.156) (0.112) (0.103) (0.146) (0.113) (0.146) (0.095) (0.090) (0.104) (0.128)
English PT in grade 11 0.14 0.31 0.11 0.01 0.01 0.09 0.02 0.02 0.05 0.08 0.10 0.06
(0.082) (0.182) (0.116) (0.099) (0.141) (0.141) (0.124) (0.136) (0.104) (0.093) (0.122) (0.113)
English PT in grade 12 0.23** 0.36*** 0.31*** 0.28** 0.18 0.13 0.36* 0.33 0.17 0.37*** 0.34*** 0.18
(0.082) (0.085) (0.066) (0.099) (0.108) (0.198) (0.150) (0.184) (0.095) (0.093) (0.091) (0.147)
Constant 3.05*** 1.03 1.89*** 2.80*** 4.60*** 6.17*** 3.55*** 1.03* 2.32*** 3.29*** 4.56*** 5.61***
(0.347) (0.596) (0.405) (0.347) (0.330) (0.613) (0.467) (0.453) (0.409) (0.414) (0.393) (0.468)

Observations 1,678 1,678 1,678 1,678 1,678 1,678 909 2,178 2,178 2,178 2,178 2,178
R-squared 0.257 0.210
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

305
Table N.10. Long-term heterogeneous effects of one year math tutoring in middle school on the CSAT math achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of Math PT in middle school -0.04 0.06 -0.24 -0.06 0.13 0.14 -0.34 0.03 0.14 0.19* 0.13 0.21
(0.192) (0.244) (0.339) (0.229) (0.249) (0.254) (0.475) (0.122) (0.105) (0.088) (0.072) (0.126)
URBAN -0.23 0.04 -0.19 -0.15 -0.60 -0.34 -0.47 -0.31 -0.31* -0.26* -0.19 -0.22
(0.291) (0.659) (0.363) (0.521) (0.417) (0.329) (0.322) (0.193) (0.135) (0.117) (0.135) (0.187)
GENDER (1=female) -0.52** -0.10 -0.33 -0.80*** -0.44* -0.64* -0.74*** -0.07 -0.22* -0.30** -0.41*** -0.26*
(0.192) (0.244) (0.267) (0.191) (0.220) (0.254) (0.216) (0.144) (0.100) (0.115) (0.102) (0.131)
SES 0.19 0.06 0.27 0.21 0.22* 0.45** 0.14 0.15 0.12 0.28*** 0.29*** 0.24*
(0.140) (0.292) (0.252) (0.201) (0.095) (0.161) (0.151) (0.079) (0.096) (0.071) (0.073) (0.102)
SCHOOL TYPE -0.11 -0.24 -0.60 -0.20 -0.02 0.50 -0.09 0.21 -0.05 -0.09 0.14 0.12
(0.265) (0.491) (0.350) (0.527) (0.484) (0.437) (0.290) (0.124) (0.092) (0.147) (0.146) (0.201)
STU-TEA RATIO 0.02 0.03 0.04 0.01 0.01 0.02 0.02 0.01 0.01 0.00 -0.00 0.02
(0.020) (0.018) (0.022) (0.026) (0.035) (0.031) (0.022) (0.011) (0.012) (0.011) (0.017) (0.014)
SCHOOL CHOICE 0.07 0.13 -0.03 0.14 -0.14 0.29 -0.17 -0.22 -0.17 -0.24* -0.16 -0.25
(0.306) (0.553) (0.343) (0.537) (0.625) (0.357) (0.335) (0.169) (0.133) (0.112) (0.132) (0.201)
PRESCORE 0.33*** 0.25* 0.27** 0.37*** 0.41*** 0.35*** 0.35*** 0.23*** 0.20*** 0.26*** 0.28*** 0.27***
(0.058) (0.104) (0.103) (0.054) (0.075) (0.092) (0.063) (0.047) (0.036) (0.026) (0.030) (0.032)
SELF-STUDY 0.06 0.02 0.22 -0.03 -0.03 -0.41 0.15 0.21 0.28* 0.26* 0.21 0.08
(0.316) (0.565) (0.524) (0.447) (0.487) (0.448) (0.359) (0.207) (0.127) (0.109) (0.133) (0.155)
SELF-ESTEEM 0.03 0.33 -0.13 -0.34 0.17 0.26 -0.12 0.21 0.11 0.03 0.07 0.22*
(0.173) (0.294) (0.225) (0.183) (0.246) (0.189) (0.194) (0.121) (0.103) (0.100) (0.097) (0.101)
Math PT in grade 10 0.13 0.29 0.42 0.10 0.12 0.11 0.33 0.47** 0.51*** 0.27* 0.16 -0.04
(0.206) (0.326) (0.260) (0.316) (0.264) (0.223) (0.239) (0.149) (0.137) (0.132) (0.109) (0.200)
Math PT in grade 11 0.66** 1.31*** 0.90* 0.85* 0.50 -0.15 0.69** 0.71*** 0.70*** 0.49*** 0.47*** 0.46**
(0.230) (0.342) (0.408) (0.362) (0.442) (0.242) (0.255) (0.097) (0.125) (0.094) (0.131) (0.153)
Math PT in grade 12 0.39 -0.21 -0.39 0.42 0.85* 0.87** 0.06 0.65*** 0.64*** 0.64*** 0.46*** 0.34**
(0.291) (0.465) (0.548) (0.416) (0.365) (0.293) (0.325) (0.155) (0.094) (0.093) (0.138) (0.117)
Constant 3.06*** 0.22 1.86 3.45** 4.33*** 5.33*** 3.89*** 1.21** 2.32*** 3.78*** 4.93*** 5.52***
(0.858) (1.167) (1.503) (1.323) (1.227) (1.091) (0.974) (0.440) (0.530) (0.536) (0.578) (0.510)

Observations 344 344 344 344 344 344 266 2,145 2,145 2,145 2,145 2,145
R-squared 0.235 0.258
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

306
Table N.11. Long-term heterogeneous effects of two years math tutoring in middle school on the CSAT math achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of Math PT in middle school 0.13 0.36 0.09 0.10 0.15 -0.05 0.13 0.05 0.22 0.31* 0.21 0.34
(0.166) (0.440) (0.184) (0.215) (0.331) (0.276) (0.442) (0.230) (0.201) (0.157) (0.166) (0.326)
URBAN -0.07 0.15 -0.10 -0.29 -0.23 0.21 -0.17 -0.31 -0.31 -0.26* -0.19 -0.22
(0.206) (0.525) (0.214) (0.215) (0.267) (0.300) (0.238) (0.190) (0.174) (0.103) (0.139) (0.191)
GENDER (1=female) 0.13 0.36 0.25 0.04 0.12 -0.01 -0.36* -0.07 -0.21* -0.30*** -0.40*** -0.25
(0.138) (0.306) (0.236) (0.188) (0.174) (0.272) (0.161) (0.145) (0.093) (0.088) (0.122) (0.137)
SES 0.05 -0.09 0.09 0.16 0.11 -0.05 0.14 0.15* 0.12 0.27*** 0.28*** 0.23**
(0.104) (0.348) (0.173) (0.158) (0.130) (0.147) (0.112) (0.069) (0.091) (0.061) (0.073) (0.083)
SCHOOL TYPE 0.31 -0.09 0.10 0.17 0.45 0.77 0.05 0.21 -0.06 -0.10 0.13 0.11
(0.175) (0.440) (0.237) (0.253) (0.319) (0.426) (0.208) (0.129) (0.079) (0.114) (0.132) (0.188)
STU-TEA RATIO -0.00 0.00 0.00 -0.00 -0.01 0.00 0.02 0.01 0.01 0.00 -0.00 0.02*
(0.014) (0.034) (0.014) (0.014) (0.021) (0.018) (0.018) (0.018) (0.013) (0.011) (0.012) (0.009)
SCHOOL CHOICE -0.14 -0.00 -0.16 -0.35 -0.41 -0.07 -0.07 -0.22 -0.17 -0.24 -0.16 -0.25
(0.213) (0.683) (0.280) (0.253) (0.304) (0.295) (0.243) (0.249) (0.177) (0.127) (0.147) (0.177)
PRESCORE 0.24*** 0.26** 0.26*** 0.27*** 0.24*** 0.23** 0.25*** 0.23*** 0.20*** 0.26*** 0.28*** 0.26***
(0.039) (0.090) (0.040) (0.050) (0.067) (0.077) (0.049) (0.035) (0.039) (0.021) (0.029) (0.027)
SELF-STUDY 0.17 0.52* 0.15 -0.10 0.20 -0.15 0.03 0.20 0.28* 0.25* 0.20 0.07
(0.231) (0.262) (0.236) (0.253) (0.386) (0.425) (0.259) (0.192) (0.130) (0.107) (0.143) (0.126)
SELF-ESTEEM 0.24 0.30 0.19 0.22 0.28 0.23 0.23 0.21 0.11 0.04 0.07 0.23*
(0.130) (0.247) (0.213) (0.203) (0.174) (0.171) (0.147) (0.128) (0.107) (0.096) (0.093) (0.102)
Math PT in grade 10 0.03 0.21 0.23 0.11 -0.08 -0.21 0.35 0.46* 0.48** 0.24 0.14 -0.08
(0.155) (0.297) (0.181) (0.205) (0.298) (0.327) (0.197) (0.213) (0.165) (0.148) (0.133) (0.120)
Math PT in grade 11 0.21 0.56 0.17 0.06 0.32 -0.29 0.41* 0.70*** 0.69*** 0.47** 0.46*** 0.44**
(0.158) (0.333) (0.142) (0.201) (0.285) (0.325) (0.194) (0.152) (0.136) (0.168) (0.119) (0.152)
Math PT in grade 12 0.36* -0.14 0.28 0.44 0.43 0.68* 0.89*** 0.65*** 0.63*** 0.63*** 0.45*** 0.33**
(0.175) (0.416) (0.275) (0.257) (0.250) (0.289) (0.214) (0.174) (0.116) (0.097) (0.110) (0.116)
Constant 3.40*** 0.64 2.29*** 3.85*** 4.84*** 5.53*** 3.30*** 1.21 2.29*** 3.75*** 4.91*** 5.48***
(0.598) (2.029) (0.592) (0.610) (0.945) (0.783) (0.730) (0.684) (0.433) (0.411) (0.536) (0.512)

Observations 421 421 421 421 421 421 466 2,145 2,145 2,145 2,145 2,145
R-squared 0.181 0.230
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

307
Table N.12. Long-term heterogeneous effects of three years math tutoring in middle school on the CSAT math achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of Math PT in middle school -0.15 -0.07 -0.04 -0.26 -0.37* -0.24 0.56 0.12 0.51 0.71 0.48 0.77
(0.143) (0.267) (0.250) (0.194) (0.185) (0.160) (0.461) (0.366) (0.415) (0.449) (0.556) (0.598)
URBAN -0.07 -0.27 -0.26 -0.02 0.05 0.06 -0.19 -0.31 -0.30 -0.25* -0.18 -0.21
(0.109) (0.184) (0.136) (0.123) (0.180) (0.211) (0.120) (0.196) (0.167) (0.110) (0.124) (0.151)
GENDER (1=female) -0.20** -0.12 -0.19 -0.22* -0.27** -0.35** -0.21* -0.07 -0.20* -0.28** -0.39*** -0.24*
(0.076) (0.122) (0.118) (0.087) (0.093) (0.130) (0.086) (0.145) (0.092) (0.100) (0.096) (0.102)
SES 0.23*** 0.20** 0.15 0.30*** 0.30** 0.36*** 0.19** 0.15* 0.11 0.26*** 0.28*** 0.22***
(0.052) (0.069) (0.089) (0.081) (0.104) (0.072) (0.060) (0.063) (0.088) (0.074) (0.079) (0.062)
SCHOOL TYPE -0.05 0.13 -0.07 -0.22 0.08 0.03 -0.13 0.21 -0.05 -0.09 0.14 0.12
(0.097) (0.131) (0.134) (0.179) (0.146) (0.211) (0.110) (0.137) (0.097) (0.101) (0.128) (0.178)
STU-TEA RATIO 0.01 -0.02* -0.00 0.00 0.01 0.04*** 0.01 0.01 0.01 -0.00 -0.01 0.02
(0.010) (0.011) (0.014) (0.013) (0.012) (0.010) (0.012) (0.019) (0.012) (0.014) (0.017) (0.019)
SCHOOL CHOICE -0.15 -0.35* -0.25 -0.09 -0.09 0.04 -0.26* -0.22 -0.16 -0.23 -0.16 -0.23
(0.119) (0.158) (0.143) (0.153) (0.194) (0.224) (0.130) (0.281) (0.168) (0.122) (0.146) (0.168)
PRESCORE 0.24*** 0.23*** 0.21*** 0.26*** 0.23*** 0.25*** 0.24*** 0.23*** 0.20*** 0.25*** 0.28*** 0.26***
(0.024) (0.046) (0.050) (0.040) (0.035) (0.048) (0.027) (0.035) (0.032) (0.032) (0.032) (0.030)
SELF-STUDY 0.21* 0.26 0.30* 0.25 0.27 0.04 0.18 0.20 0.26* 0.23 0.19 0.05
(0.103) (0.177) (0.115) (0.137) (0.189) (0.202) (0.113) (0.167) (0.132) (0.137) (0.160) (0.164)
SELF-ESTEEM 0.23*** 0.28** 0.18 0.15 0.27* 0.43*** 0.19* 0.21 0.10 0.02 0.06 0.21*
(0.069) (0.106) (0.093) (0.135) (0.108) (0.080) (0.077) (0.108) (0.074) (0.086) (0.104) (0.102)
Math PT in grade 10 0.42*** 0.76*** 0.58*** 0.41* 0.42* 0.15 0.38** 0.45* 0.44* 0.17 0.10 -0.15
(0.107) (0.226) (0.154) (0.164) (0.189) (0.247) (0.133) (0.200) (0.180) (0.167) (0.153) (0.183)
Math PT in grade 11 0.59*** 0.63*** 0.75*** 0.59*** 0.38** 0.50** 0.49*** 0.70*** 0.68*** 0.46*** 0.45*** 0.43**
(0.099) (0.155) (0.108) (0.105) (0.127) (0.159) (0.116) (0.162) (0.150) (0.087) (0.099) (0.151)
Math PT in grade 12 0.55*** 0.80*** 0.64*** 0.59*** 0.47*** 0.23* 0.55*** 0.66*** 0.64*** 0.64*** 0.47*** 0.35**
(0.087) (0.149) (0.120) (0.129) (0.130) (0.117) (0.096) (0.195) (0.094) (0.113) (0.129) (0.116)
Constant 3.56*** 1.98*** 2.68*** 3.55*** 4.86*** 5.02*** 3.31*** 1.17 2.13*** 3.52*** 4.76*** 5.24***
(0.374) (0.500) (0.436) (0.501) (0.426) (0.573) (0.491) (0.866) (0.538) (0.440) (0.526) (0.697)

Observations 1,668 1,668 1,668 1,668 1,668 1,668 1,369 2,145 2,145 2,145 2,145 2,145
R-squared 0.256 0.242
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

308
Table O.1. Long-term heterogeneous effects of private tutoring in grade 7 on the CSAT achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

PT in G7 0.14 0.28 0.17 0.13 0.09 -0.00 0.42 0.04 0.19 0.57 0.39 0.20
(0.087) (0.280) (0.144) (0.107) (0.119) (0.168) (0.274) (0.579) (0.383) (0.356) (0.451) (0.638)
URBANICITY -0.00 0.03 -0.03 -0.05 0.01 0.05 -0.03 0.06 -0.06 -0.12 -0.08 -0.10
(0.084) (0.164) (0.106) (0.137) (0.112) (0.114) (0.087) (0.175) (0.091) (0.121) (0.088) (0.135)
GENDER (1=female) 0.07 0.21* 0.14* 0.03 0.02 -0.14 0.06 0.20 0.14 0.02 -0.03 -0.17
(0.058) (0.100) (0.065) (0.085) (0.080) (0.100) (0.060) (0.107) (0.074) (0.076) (0.112) (0.118)
SES 0.28*** 0.18 0.24** 0.32*** 0.35*** 0.34*** 0.25*** 0.14 0.27*** 0.31*** 0.33*** 0.30***
(0.040) (0.091) (0.079) (0.068) (0.065) (0.049) (0.043) (0.078) (0.068) (0.076) (0.068) (0.085)
SCHOOL TYPE (1=private) 0.13 0.09 0.14 0.07 0.09 0.15 0.10 0.15 0.12 0.08 0.09 0.13
(0.074) (0.124) (0.078) (0.072) (0.103) (0.103) (0.077) (0.156) (0.092) (0.105) (0.111) (0.179)
STU-TEA RATIO 0.00 0.01 0.02* 0.00 -0.00 -0.01 0.01 0.01 0.02 0.00 0.00 -0.00
(0.007) (0.016) (0.008) (0.010) (0.009) (0.015) (0.007) (0.016) (0.009) (0.012) (0.012) (0.016)
SCHOOL CHOICE -0.04 0.14 0.02 -0.04 -0.04 -0.11 -0.05 0.10 -0.03 -0.11 -0.10 -0.20
(0.090) (0.201) (0.084) (0.117) (0.116) (0.113) (0.093) (0.201) (0.115) (0.136) (0.105) (0.125)
PRESCORE (G6) 0.25*** 0.21*** 0.24*** 0.28*** 0.27*** 0.23*** 0.25*** 0.22*** 0.25*** 0.26*** 0.28*** 0.23***
(0.018) (0.042) (0.026) (0.027) (0.028) (0.037) (0.019) (0.037) (0.026) (0.030) (0.029) (0.039)
STUDY HOURS (1= >10hrs /week) 0.13 0.31** 0.10 -0.04 0.26* -0.04 0.14 0.33*** 0.12 0.02 0.27* 0.01
(0.082) (0.111) (0.123) (0.167) (0.125) (0.164) (0.084) (0.079) (0.083) (0.100) (0.106) (0.131)
SELF ESTEEM 0.28*** 0.35*** 0.29*** 0.29*** 0.23*** 0.34*** 0.27*** 0.34*** 0.27*** 0.28*** 0.21** 0.30***
(0.052) (0.079) (0.060) (0.063) (0.050) (0.086) (0.054) (0.078) (0.073) (0.075) (0.064) (0.083)
PT in G8 0.17 0.39 0.17 0.13 0.17 0.26 0.14 0.54* 0.11 -0.00 0.06 0.19
(0.094) (0.212) (0.143) (0.139) (0.127) (0.204) (0.120) (0.218) (0.169) (0.114) (0.217) (0.258)
PT in G9 -0.11 -0.13 -0.11 -0.16 -0.12 -0.19 -0.22* -0.22 -0.13 -0.21 -0.27 -0.21
(0.093) (0.173) (0.093) (0.115) (0.155) (0.201) (0.104) (0.152) (0.106) (0.160) (0.149) (0.187)
PT in G10 0.19* 0.45* 0.27** 0.25** 0.13 -0.02 0.19* 0.50** 0.28** 0.27 0.16 0.05
(0.080) (0.180) (0.092) (0.089) (0.170) (0.121) (0.084) (0.185) (0.091) (0.151) (0.138) (0.134)
PT in G11 0.31*** 0.55*** 0.35*** 0.23* 0.27* 0.15 0.30*** 0.47** 0.36*** 0.25* 0.26** 0.13
(0.075) (0.116) (0.059) (0.091) (0.114) (0.103) (0.078) (0.154) (0.105) (0.104) (0.098) (0.174)
PT in G12 0.32*** 0.32** 0.30*** 0.36*** 0.29** 0.49*** 0.32*** 0.39** 0.29** 0.31*** 0.32*** 0.55***
(0.065) (0.102) (0.077) (0.086) (0.088) (0.117) (0.068) (0.120) (0.099) (0.083) (0.096) (0.124)
Constant 3.14*** 0.66 1.83*** 3.18*** 4.20*** 5.77*** 3.01*** 0.84 1.96*** 3.19*** 4.09*** 5.77***
(0.276) (0.827) (0.363) (0.364) (0.281) (0.558) (0.306) (0.720) (0.432) (0.479) (0.587) (0.655)

Observations 1,905 1,905 1,905 1,905 1,905 1,905 1,800 1,888 1,888 1,888 1,888 1,888
R-squared 0.296 0.287
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

309
Table O.2. Long-term heterogeneous effects of private tutoring in grade 8 on the CSAT achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

PT in G8 0.09 0.34 0.19 0.17 0.07 0.06 -0.18 -0.50 0.23 -0.00 0.01 -0.28
(0.086) (0.205) (0.158) (0.108) (0.106) (0.121) (0.293) (0.714) (0.465) (0.301) (0.349) (0.679)
URBANICITY -0.03 -0.12 0.05 -0.09 -0.04 0.09 -0.05 -0.12 0.05 -0.13 -0.03 -0.06
(0.075) (0.139) (0.106) (0.089) (0.096) (0.125) (0.078) (0.157) (0.095) (0.123) (0.110) (0.124)
GENDER (1=female) -0.11* 0.17 -0.01 -0.12* -0.26*** -0.26*** -0.13* 0.11 -0.03 -0.10 -0.27** -0.29**
(0.053) (0.122) (0.065) (0.052) (0.072) (0.068) (0.055) (0.093) (0.087) (0.061) (0.102) (0.106)
SES 0.18*** 0.14** 0.15** 0.18*** 0.25*** 0.18** 0.14*** 0.09 0.12* 0.17** 0.20** 0.10
(0.036) (0.045) (0.049) (0.039) (0.055) (0.059) (0.038) (0.085) (0.050) (0.055) (0.068) (0.065)
SCHOOL TYPE (1=private) 0.13* 0.17 0.07 0.19* 0.17 0.21 0.10 0.11 0.07 0.13 0.11 0.16
(0.067) (0.102) (0.083) (0.079) (0.092) (0.134) (0.069) (0.152) (0.065) (0.074) (0.103) (0.156)
STU-TEA RATIO -0.01 0.00 0.01 -0.01 -0.01 -0.00 -0.01 0.00 0.00 -0.01 -0.01 0.00
(0.007) (0.014) (0.012) (0.008) (0.008) (0.015) (0.007) (0.011) (0.011) (0.007) (0.007) (0.014)
SCHOOL CHOICE 0.04 0.04 0.14 0.01 0.08 0.02 0.03 0.01 0.13 -0.01 0.06 -0.09
(0.081) (0.134) (0.082) (0.100) (0.099) (0.128) (0.084) (0.138) (0.080) (0.095) (0.098) (0.086)
PRESCORE (G6) 0.95*** 0.92*** 0.93*** 0.97*** 0.98*** 0.96*** 0.97*** 1.00*** 0.92*** 0.99*** 0.99*** 1.02***
(0.039) (0.093) (0.051) (0.032) (0.048) (0.070) (0.041) (0.091) (0.061) (0.063) (0.046) (0.060)
STUDY HOURS 0.08 0.23 0.09 -0.01 0.12 -0.02 0.09 0.15 0.10 0.03 0.15 0.12
(1= >10hrs /week) (0.074) (0.124) (0.086) (0.106) (0.104) (0.153) (0.077) (0.144) (0.072) (0.101) (0.096) (0.156)
SELF ESTEEM 0.21*** 0.22* 0.23*** 0.18*** 0.23*** 0.30*** 0.20*** 0.12 0.23*** 0.18*** 0.19*** 0.23
(0.046) (0.093) (0.067) (0.040) (0.057) (0.083) (0.048) (0.080) (0.066) (0.046) (0.046) (0.131)
PT in G7 -0.06 0.12 -0.05 -0.14 -0.11 -0.16 0.02 0.43 -0.09 -0.11 -0.14 -0.08
(0.080) (0.167) (0.140) (0.100) (0.078) (0.156) (0.110) (0.248) (0.176) (0.130) (0.144) (0.198)
PT in G9 -0.12 -0.23 -0.14 -0.05 -0.21* -0.17 -0.04 0.10 -0.19 -0.08 -0.14 -0.08
(0.085) (0.160) (0.069) (0.093) (0.104) (0.154) (0.132) (0.329) (0.186) (0.149) (0.181) (0.346)
PT in G10 0.08 0.26 0.27* 0.11 0.00 -0.10 0.09 0.31* 0.24 0.14 0.01 -0.09
(0.073) (0.140) (0.112) (0.110) (0.108) (0.161) (0.077) (0.122) (0.142) (0.104) (0.108) (0.163)
PT in G11 0.30*** 0.41** 0.32*** 0.23* 0.16* 0.14 0.30*** 0.36* 0.32** 0.19* 0.13 0.10
(0.068) (0.133) (0.085) (0.095) (0.076) (0.141) (0.070) (0.142) (0.114) (0.095) (0.110) (0.156)
PT in G12 0.19** 0.19 0.10 0.20* 0.25*** 0.29*** 0.21*** 0.19 0.11 0.20** 0.30** 0.32*
(0.060) (0.154) (0.073) (0.090) (0.071) (0.084) (0.062) (0.100) (0.070) (0.065) (0.093) (0.129)
Constant 5.24*** 2.78*** 3.77*** 5.44*** 6.48*** 7.14*** 5.36*** 3.15*** 3.99*** 5.57*** 6.50*** 7.44***
(0.236) (0.606) (0.422) (0.293) (0.323) (0.513) (0.254) (0.373) (0.345) (0.281) (0.343) (0.493)

Observations 1,934 1,934 1,934 1,934 1,934 1,934 1,770 1,782 1,782 1,782 1,782 1,782
R-squared 0.405 0.403
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

310
Table O.3. Long-term heterogeneous effects of private tutoring in grade 9 on the CSAT achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

PT in G9 -0.07 -0.08 -0.05 -0.09 -0.27 -0.21 0.31 0.08 -0.19 0.19 0.40 0.79
(0.081) (0.183) (0.108) (0.110) (0.140) (0.124) (0.385) (0.532) (0.526) (0.301) (0.797) (0.596)
URBANICITY -0.04 0.10 0.05 0.04 -0.16 -0.17 -0.09 -0.01 0.02 -0.04 -0.11 0.02
(0.075) (0.143) (0.121) (0.107) (0.136) (0.105) (0.081) (0.155) (0.086) (0.071) (0.106) (0.115)
GENDER (1=female) -0.17** 0.07 -0.04 -0.21*** -0.26** -0.35*** -0.17** 0.02 -0.08 -0.24*** -0.27*** -0.33***
(0.052) (0.109) (0.074) (0.059) (0.083) (0.082) (0.059) (0.112) (0.068) (0.049) (0.047) (0.082)
SES 0.18*** 0.10 0.18*** 0.15** 0.21** 0.17** 0.17*** 0.08* 0.11* 0.09* 0.18* 0.13
(0.036) (0.085) (0.051) (0.049) (0.072) (0.060) (0.039) (0.042) (0.055) (0.044) (0.086) (0.076)
SCHOOL TYPE (1=private) 0.13 0.11 0.10 0.12** 0.15 0.18 0.10 0.16 0.08 0.06 0.11 0.11
(0.065) (0.101) (0.076) (0.040) (0.088) (0.119) (0.070) (0.109) (0.095) (0.060) (0.090) (0.134)
STU-TEA RATIO -0.00 0.00 0.00 -0.00 -0.01 0.00 -0.00 0.01 0.01 -0.00 -0.00 0.01
(0.006) (0.012) (0.011) (0.005) (0.009) (0.012) (0.007) (0.012) (0.010) (0.006) (0.010) (0.012)
SCHOOL CHOICE 0.02 0.11 0.06 0.08 -0.07 -0.01 -0.04 0.06 0.05 0.01 -0.05 0.08
(0.080) (0.122) (0.114) (0.116) (0.148) (0.120) (0.087) (0.137) (0.106) (0.082) (0.120) (0.140)
PRESCORE (G6) 0.84*** 0.87*** 0.85*** 0.89*** 0.88*** 0.80*** 0.82*** 0.91*** 0.88*** 0.87*** 0.87*** 0.79***
(0.036) (0.082) (0.057) (0.056) (0.050) (0.051) (0.038) (0.045) (0.055) (0.043) (0.051) (0.066)
STUDY HOURS 0.09 0.15 0.05 0.09 0.03 0.07 0.10 0.15 -0.07 -0.00 0.09 0.06
(1= >10hrs /week) (0.073) (0.110) (0.094) (0.064) (0.102) (0.092) (0.078) (0.079) (0.077) (0.065) (0.116) (0.095)
SELF ESTEEM 0.26*** 0.28*** 0.27*** 0.26*** 0.30*** 0.30*** 0.22*** 0.26** 0.29*** 0.25*** 0.24*** 0.20***
(0.045) (0.071) (0.069) (0.042) (0.062) (0.059) (0.051) (0.090) (0.055) (0.043) (0.060) (0.046)
PT in G8 0.02 0.12 0.03 0.05 0.01 -0.00 -0.17 -0.04 -0.01 -0.08 -0.31 -0.46
(0.081) (0.217) (0.133) (0.119) (0.132) (0.179) (0.178) (0.233) (0.248) (0.149) (0.409) (0.343)
PT in G10 0.16* 0.31*** 0.22** 0.20* 0.05 -0.06 0.16 0.42* 0.32* 0.21** 0.06 -0.15
(0.071) (0.086) (0.070) (0.087) (0.137) (0.111) (0.085) (0.170) (0.126) (0.077) (0.149) (0.184)
PT in G11 0.31*** 0.43** 0.38*** 0.21** 0.24** 0.34*** 0.27** 0.37** 0.31* 0.23*** 0.23** 0.30*
(0.066) (0.134) (0.086) (0.077) (0.078) (0.102) (0.083) (0.120) (0.155) (0.069) (0.089) (0.126)
PT in G12 0.22*** 0.27** 0.16 0.19** 0.29*** 0.20* 0.25*** 0.26*** 0.17 0.16* 0.22** 0.16*
(0.059) (0.098) (0.085) (0.069) (0.082) (0.082) (0.063) (0.065) (0.085) (0.074) (0.074) (0.071)
Constant 5.25*** 2.88*** 4.09*** 5.15*** 6.65*** 7.19*** 5.22*** 2.80*** 4.06*** 5.21*** 6.26*** 6.53***
(0.229) (0.345) (0.341) (0.242) (0.308) (0.389) (0.270) (0.416) (0.278) (0.249) (0.416) (0.438)

Observations 2,021 2,021 2,021 2,021 2,021 2,021 1,785 2,335 2,335 2,335 2,335 2,335
R-squared 0.390 0.378
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

311
Table O.4. Long-term heterogeneous effects of verbal tutoring in grade 7 on the CSAT verbal achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G7 -0.04 0.01 -0.05 -0.08 0.01 -0.14 -0.08 -0.58 0.07 -0.13 0.28 -0.11
(0.071) (0.140) (0.098) (0.088) (0.128) (0.147) (0.188) (0.542) (0.324) (0.336) (0.370) (0.587)
URBANICITY -0.01 -0.04 0.02 -0.05 0.19 -0.10 -0.01 -0.04 -0.02 -0.06 0.13 -0.09
(0.094) (0.197) (0.145) (0.127) (0.158) (0.190) (0.097) (0.152) (0.110) (0.091) (0.150) (0.250)
GENDER (1=female) 0.25*** 0.44*** 0.28*** 0.22** 0.18 0.04 0.24*** 0.41** 0.24* 0.21* 0.15 -0.06
(0.065) (0.106) (0.085) (0.085) (0.115) (0.110) (0.067) (0.132) (0.100) (0.101) (0.098) (0.083)
SES 0.23*** 0.29** 0.20** 0.24** 0.26** 0.25** 0.21*** 0.30*** 0.22*** 0.23*** 0.24*** 0.22*
(0.044) (0.092) (0.070) (0.077) (0.079) (0.080) (0.045) (0.065) (0.061) (0.069) (0.066) (0.089)
SCHOOL TYPE (1=private) 0.09 0.06 0.19 0.10 -0.01 0.05 0.08 0.17 0.20*** 0.12 0.02 -0.00
(0.081) (0.178) (0.108) (0.119) (0.147) (0.107) (0.084) (0.193) (0.060) (0.089) (0.114) (0.126)
STU-TEA RATIO 0.01 0.01 0.02 0.01 -0.00 0.01 0.01 0.01 0.02 0.01 0.00 0.01
(0.008) (0.015) (0.016) (0.012) (0.010) (0.013) (0.008) (0.013) (0.014) (0.010) (0.013) (0.017)
SCHOOL CHOICE 0.05 -0.03 0.09 0.03 0.24 0.02 0.05 -0.03 0.07 0.01 0.24 -0.00
(0.101) (0.191) (0.144) (0.134) (0.143) (0.173) (0.104) (0.190) (0.155) (0.122) (0.151) (0.224)
PRESCORE (G6) 0.29*** 0.28*** 0.29*** 0.29*** 0.30*** 0.33*** 0.29*** 0.28*** 0.28*** 0.29*** 0.29*** 0.34***
(0.020) (0.035) (0.027) (0.024) (0.035) (0.034) (0.021) (0.056) (0.029) (0.023) (0.034) (0.026)
STUDY HOURS (1= >10hrs /week) 0.03 -0.01 0.18 0.11 -0.02 -0.14 0.04 -0.04 0.19 0.15 0.02 -0.09
(0.092) (0.183) (0.145) (0.119) (0.130) (0.125) (0.094) (0.128) (0.108) (0.106) (0.140) (0.151)
SELF ESTEEM 0.30*** 0.41*** 0.28*** 0.36*** 0.29** 0.16* 0.30*** 0.47** 0.29** 0.35*** 0.28** 0.12
(0.058) (0.104) (0.061) (0.065) (0.090) (0.082) (0.060) (0.149) (0.091) (0.061) (0.090) (0.089)
VERBAL PT in G8 -0.02 -0.04 0.04 0.02 -0.17 -0.06 -0.03 0.16 0.00 0.01 -0.28 -0.13
(0.076) (0.112) (0.101) (0.113) (0.164) (0.164) (0.095) (0.181) (0.138) (0.162) (0.193) (0.240)
VERBAL PT in G9 -0.13 -0.00 -0.05 -0.16* -0.19 -0.27** -0.12 0.04 -0.09 -0.11 -0.22* -0.21
(0.075) (0.163) (0.096) (0.080) (0.124) (0.105) (0.081) (0.142) (0.129) (0.114) (0.101) (0.162)
VERBAL PT in G10 0.03 0.11 0.04 0.02 -0.05 -0.02 0.03 0.18 0.04 0.03 -0.05 -0.04
(0.068) (0.123) (0.086) (0.059) (0.098) (0.112) (0.071) (0.147) (0.092) (0.108) (0.117) (0.147)
VERBAL PT in G11 0.08 -0.05 0.03 0.05 0.16 0.20 0.07 -0.04 0.10 0.05 0.16 0.25
(0.081) (0.132) (0.107) (0.091) (0.166) (0.126) (0.083) (0.147) (0.086) (0.099) (0.155) (0.131)
VERBAL PT in G12 0.33*** 0.21 0.05 0.36** 0.46** 0.39** 0.35*** 0.29 0.13 0.42* 0.42 0.36
(0.101) (0.117) (0.103) (0.118) (0.146) (0.145) (0.102) (0.196) (0.148) (0.204) (0.229) (0.207)
Constant 3.32*** 1.25* 1.91*** 3.54*** 4.42*** 5.62*** 3.32*** 1.46** 2.13*** 3.49*** 4.36*** 5.66***
(0.301) (0.577) (0.510) (0.370) (0.373) (0.452) (0.313) (0.566) (0.573) (0.369) (0.402) (0.619)

Observations 2,032 2,032 2,032 2,032 2,032 2,032 1,921 2,013 2,013 2,013 2,013 2,013
R-squared 0.198 0.194
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

312
Table O.5. Long-term heterogeneous effects of verbal tutoring in grade 8 on the CSAT verbal achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G8 -0.04 0.05 -0.03 -0.09 -0.08 0.00 -0.33 -0.29 -0.52* -0.42* -0.47* -0.37
(0.071) (0.103) (0.113) (0.127) (0.120) (0.127) (0.176) (0.458) (0.261) (0.174) (0.196) (0.413)
URBANICITY -0.05 0.02 -0.02 -0.03 -0.00 -0.05 -0.04 0.00 0.03 -0.06 -0.06 -0.05
(0.088) (0.148) (0.082) (0.108) (0.147) (0.155) (0.090) (0.221) (0.210) (0.134) (0.147) (0.146)
GENDER (1=female) 0.07 0.30** 0.11 0.06 -0.06 -0.14 0.06 0.21 0.09 0.03 -0.10 -0.14
(0.061) (0.113) (0.088) (0.076) (0.095) (0.114) (0.063) (0.132) (0.111) (0.076) (0.083) (0.093)
SES 0.11** 0.10 0.10* 0.07 0.12* 0.11 0.07 0.11 0.09 0.04 0.10** 0.14*
(0.041) (0.102) (0.047) (0.055) (0.052) (0.073) (0.043) (0.098) (0.058) (0.064) (0.036) (0.068)
SCHOOL TYPE (1=private) 0.10 0.14 0.17 0.09 0.12 0.08 0.07 0.07 0.12 0.02 0.05 -0.01
(0.076) (0.110) (0.094) (0.106) (0.108) (0.130) (0.078) (0.148) (0.085) (0.096) (0.109) (0.122)
STU-TEA RATIO -0.00 -0.01 -0.00 -0.00 -0.01 -0.00 -0.00 -0.01 0.00 0.00 -0.01 -0.01
(0.007) (0.014) (0.013) (0.011) (0.014) (0.014) (0.008) (0.016) (0.014) (0.011) (0.008) (0.011)
SCHOOL CHOICE 0.14 0.01 0.12 0.21 0.23 0.17 0.15 0.08 0.11 0.17 0.15 0.22
(0.094) (0.180) (0.135) (0.108) (0.168) (0.142) (0.097) (0.221) (0.179) (0.093) (0.151) (0.163)
PRESCORE (G6) 0.95*** 1.04*** 0.97*** 0.99*** 0.97*** 0.89*** 0.96*** 1.03*** 0.96*** 1.01*** 0.98*** 0.91***
(0.043) (0.089) (0.057) (0.067) (0.065) (0.060) (0.045) (0.092) (0.079) (0.080) (0.080) (0.094)
STUDY HOURS (1= >10hrs /week) -0.02 -0.08 0.06 0.07 -0.01 -0.33** -0.00 -0.05 0.06 0.08 0.07 -0.27
(0.086) (0.198) (0.106) (0.088) (0.085) (0.117) (0.089) (0.160) (0.111) (0.111) (0.126) (0.141)
SELF ESTEEM 0.24*** 0.25** 0.22* 0.30*** 0.24** 0.23** 0.25*** 0.22 0.23* 0.27*** 0.23*** 0.20*
(0.053) (0.083) (0.087) (0.067) (0.086) (0.074) (0.055) (0.159) (0.090) (0.069) (0.068) (0.082)
VERBAL PT in G7 -0.12 -0.19 -0.08 -0.09 -0.08 -0.10 -0.03 -0.20 0.07 0.02 0.05 0.05
(0.067) (0.123) (0.096) (0.111) (0.091) (0.094) (0.084) (0.139) (0.133) (0.091) (0.127) (0.191)
VERBAL PT in G9 -0.08 -0.02 -0.01 -0.09 -0.20* -0.34** 0.03 0.07 0.13 0.04 -0.08 -0.23
(0.071) (0.127) (0.068) (0.078) (0.080) (0.104) (0.092) (0.172) (0.108) (0.128) (0.097) (0.171)
VERBAL PT in G10 -0.02 -0.05 0.01 0.00 -0.04 -0.05 -0.02 -0.12 0.06 0.07 -0.01 -0.05
(0.064) (0.115) (0.064) (0.107) (0.076) (0.118) (0.066) (0.109) (0.090) (0.082) (0.079) (0.125)
VERBAL PT in G11 0.10 0.15 -0.04 0.08 0.11 0.34** 0.08 0.09 -0.02 0.05 0.07 0.25
(0.075) (0.114) (0.096) (0.092) (0.099) (0.107) (0.077) (0.147) (0.095) (0.087) (0.130) (0.146)
VERBAL PT in G12 0.20* -0.03 0.12 0.21 0.25* 0.31* 0.23* 0.10 0.18 0.21* 0.26* 0.34*
(0.095) (0.145) (0.139) (0.117) (0.103) (0.122) (0.096) (0.177) (0.141) (0.095) (0.118) (0.135)
Constant 5.39*** 3.50*** 4.40*** 5.28*** 6.59*** 7.55*** 5.47*** 3.72*** 4.50*** 5.40*** 6.83*** 7.57***
(0.263) (0.499) (0.412) (0.265) (0.537) (0.491) (0.272) (0.480) (0.398) (0.292) (0.406) (0.316)

Observations 2,064 2,064 2,064 2,064 2,064 2,064 1,889 1,906 1,906 1,906 1,906 1,906
R-squared 0.282 0.278
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

313
Table O.6. Long-term heterogeneous effects of verbal tutoring in grade 9 on the CSAT verbal achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G9 -0.07 0.09 0.01 -0.10 -0.05 -0.32* 0.07 -0.19 0.05 -0.12 0.01 0.74
(0.072) (0.104) (0.093) (0.117) (0.152) (0.146) (0.290) (0.594) (0.350) (0.237) (0.431) (0.673)
URBANICITY -0.03 0.12 0.01 0.02 0.04 0.06 -0.05 0.06 0.02 0.02 0.05 0.04
(0.089) (0.158) (0.153) (0.125) (0.122) (0.170) (0.097) (0.124) (0.139) (0.089) (0.144) (0.164)
GENDER (1=female) 0.03 0.16 0.06 0.08 -0.09 -0.20 0.04 0.16 0.09 0.02 -0.22** -0.10
(0.063) (0.098) (0.075) (0.080) (0.091) (0.127) (0.072) (0.111) (0.055) (0.076) (0.075) (0.129)
SES 0.12** 0.14 0.11* 0.12* 0.14 0.18* 0.12* 0.13 0.07 0.08 0.08 0.07
(0.043) (0.092) (0.054) (0.050) (0.076) (0.083) (0.046) (0.080) (0.048) (0.049) (0.059) (0.073)
SCHOOL TYPE (1=private) 0.11 0.16 0.22* 0.07 0.11 0.04 0.08 0.11 0.14 0.05 0.15 0.00
(0.078) (0.132) (0.088) (0.108) (0.116) (0.176) (0.083) (0.169) (0.111) (0.109) (0.104) (0.116)
STU-TEA RATIO 0.00 -0.01 0.00 0.00 -0.01 0.01 0.00 -0.01 0.00 0.01 -0.00 0.00
(0.008) (0.015) (0.011) (0.007) (0.014) (0.013) (0.008) (0.016) (0.008) (0.009) (0.013) (0.014)
SCHOOL CHOICE 0.13 0.04 0.19 0.19 0.19 0.31 0.11 0.03 0.17 0.14* 0.15 0.14
(0.096) (0.181) (0.157) (0.109) (0.138) (0.170) (0.103) (0.159) (0.139) (0.062) (0.123) (0.133)
PRESCORE (G6) 0.83*** 0.95*** 0.93*** 0.85*** 0.81*** 0.78*** 0.82*** 1.00*** 0.92*** 0.90*** 0.85*** 0.87***
(0.042) (0.104) (0.072) (0.067) (0.070) (0.064) (0.045) (0.063) (0.062) (0.048) (0.051) (0.063)
STUDY HOURS (1= >10hrs /week) -0.01 -0.02 0.12 0.03 -0.05 -0.20 0.02 0.03 0.02 -0.04 -0.01 -0.25
(0.088) (0.234) (0.122) (0.093) (0.102) (0.182) (0.093) (0.153) (0.088) (0.098) (0.099) (0.141)
SELF ESTEEM 0.33*** 0.27** 0.30*** 0.35*** 0.36** 0.19** 0.29*** 0.25** 0.37*** 0.24*** 0.36*** 0.21*
(0.054) (0.104) (0.080) (0.084) (0.114) (0.072) (0.059) (0.091) (0.063) (0.062) (0.084) (0.089)
VERBAL PT in G7 -0.09 -0.06 -0.05 -0.02 -0.18 -0.11 -0.12 -0.02 -0.12 -0.09 -0.21 -0.29*
(0.069) (0.153) (0.109) (0.101) (0.120) (0.131) (0.078) (0.159) (0.104) (0.084) (0.119) (0.146)
VERBAL PT in G8 -0.06 -0.03 -0.03 -0.09 -0.11 -0.10 -0.10 -0.01 -0.04 0.02 -0.13 -0.49*
(0.073) (0.137) (0.097) (0.122) (0.144) (0.114) (0.133) (0.273) (0.161) (0.125) (0.172) (0.211)
VERBAL PT in G10 -0.02 -0.03 -0.01 0.03 -0.03 0.01 -0.02 -0.07 0.04 0.07 0.05 -0.10
(0.065) (0.119) (0.081) (0.084) (0.114) (0.104) (0.081) (0.158) (0.137) (0.112) (0.118) (0.125)
VERBAL PT in G11 0.10 0.03 0.14 0.09 0.20 0.20 0.09 -0.06 0.02 0.05 0.16 0.05
(0.077) (0.130) (0.108) (0.088) (0.105) (0.114) (0.082) (0.113) (0.137) (0.101) (0.107) (0.138)
VERBAL PT in G12 0.20* 0.01 0.06 0.11 0.25* 0.25 0.20* 0.04 0.20 0.08 0.20 0.23*
(0.097) (0.213) (0.144) (0.115) (0.122) (0.182) (0.103) (0.196) (0.168) (0.152) (0.169) (0.115)
Constant 5.40*** 3.57*** 4.16*** 5.20*** 6.59*** 7.34*** 5.36*** 3.74*** 4.17*** 5.25*** 6.77*** 7.28***
(0.268) (0.485) (0.397) (0.328) (0.476) (0.485) (0.298) (0.405) (0.404) (0.250) (0.379) (0.434)

Observations 2,044 2,044 2,044 2,044 2,044 2,044 1,812 2,273 2,273 2,273 2,273 2,273
R-squared 0.252 0.243
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

314
Table O.7. Long-term heterogeneous effects of English tutoring in grade 7 on the CSAT English achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G7 0.10 0.19 0.01 0.11 0.16 -0.07 0.65** 0.32 0.05 0.70*** 0.91** 0.98*
(0.085) (0.167) (0.108) (0.107) (0.120) (0.166) (0.251) (0.336) (0.453) (0.199) (0.302) (0.407)
URBANICITY 0.03 0.22 -0.11 -0.02 0.04 0.16 -0.01 0.12 -0.12 -0.09 -0.11 0.04
(0.090) (0.233) (0.121) (0.085) (0.119) (0.142) (0.094) (0.184) (0.099) (0.087) (0.128) (0.136)
GENDER (1=female) 0.24*** 0.33* 0.32*** 0.25* 0.12 0.07 0.23*** 0.29* 0.28*** 0.23*** 0.11 0.02
(0.062) (0.133) (0.096) (0.102) (0.092) (0.088) (0.065) (0.138) (0.083) (0.065) (0.093) (0.115)
SES 0.39*** 0.22* 0.36*** 0.48*** 0.44*** 0.43*** 0.37*** 0.28** 0.36*** 0.44*** 0.44*** 0.39***
(0.043) (0.093) (0.055) (0.064) (0.077) (0.077) (0.046) (0.100) (0.053) (0.051) (0.050) (0.069)
SCHOOL TYPE (1=private) 0.15 0.15 0.24 0.13 0.18 0.22 0.12 0.14 0.25** 0.07 0.09 0.18
(0.079) (0.180) (0.131) (0.126) (0.128) (0.180) (0.083) (0.168) (0.090) (0.088) (0.151) (0.136)
STU-TEA RATIO 0.01 0.00 0.02 0.01 0.01 0.00 0.01 0.00 0.02 0.02 0.01 0.00
(0.008) (0.016) (0.012) (0.008) (0.008) (0.012) (0.008) (0.018) (0.013) (0.010) (0.013) (0.013)
SCHOOL CHOICE -0.16 -0.10 -0.22 -0.10 -0.15 -0.24 -0.17 -0.09 -0.21 -0.12 -0.18 -0.26*
(0.097) (0.225) (0.166) (0.110) (0.141) (0.173) (0.101) (0.194) (0.126) (0.065) (0.102) (0.122)
PRESCORE (G6) 0.26*** 0.30*** 0.29*** 0.28*** 0.27*** 0.20*** 0.26*** 0.27*** 0.28*** 0.28*** 0.26*** 0.18***
(0.020) (0.050) (0.037) (0.026) (0.027) (0.033) (0.020) (0.039) (0.020) (0.035) (0.035) (0.049)
STUDY HOURS (1= >10hrs /week) 0.11 0.21 0.02 -0.06 0.10 0.15 0.12 0.25 0.05 -0.09 0.12 0.19
(0.088) (0.145) (0.117) (0.097) (0.132) (0.117) (0.091) (0.175) (0.127) (0.112) (0.144) (0.124)
SELF ESTEEM 0.37*** 0.50*** 0.40*** 0.40*** 0.32*** 0.31*** 0.35*** 0.42*** 0.36*** 0.36*** 0.29*** 0.33***
(0.056) (0.103) (0.112) (0.092) (0.064) (0.085) (0.058) (0.090) (0.073) (0.086) (0.074) (0.089)
ENGLISH PT in G8 0.22* 0.35* 0.19 0.17 0.31* 0.22 0.05 0.16 0.16 -0.07 0.03 -0.12
(0.097) (0.176) (0.151) (0.159) (0.157) (0.172) (0.124) (0.222) (0.185) (0.133) (0.193) (0.200)
ENGLISH PT in G9 -0.04 -0.02 0.04 0.05 -0.20 -0.04 -0.14 -0.12 -0.01 -0.01 -0.38* -0.28
(0.099) (0.181) (0.209) (0.174) (0.111) (0.159) (0.109) (0.204) (0.179) (0.101) (0.161) (0.152)
ENGLISH PT in G10 0.08 0.26 0.10 0.12 -0.04 0.03 0.06 0.43 0.17 0.06 -0.10 0.02
(0.073) (0.138) (0.104) (0.072) (0.097) (0.119) (0.077) (0.224) (0.149) (0.092) (0.103) (0.137)
ENGLISH PT in G11 0.24*** 0.37* 0.21** 0.13 0.11 0.20* 0.24** 0.46** 0.22** 0.12 0.13 0.18
(0.071) (0.186) (0.077) (0.089) (0.099) (0.080) (0.074) (0.171) (0.075) (0.096) (0.101) (0.127)
ENGLISH PT in G12 0.20** 0.35* 0.25** 0.21* 0.17* 0.04 0.20** 0.32 0.27* 0.24** 0.16 0.11
(0.075) (0.146) (0.092) (0.089) (0.082) (0.132) (0.077) (0.171) (0.107) (0.087) (0.091) (0.140)
Constant 3.08*** 0.25 1.84** 2.86*** 4.33*** 5.95*** 2.88*** 0.51 1.87*** 2.65*** 4.24*** 5.85***
(0.299) (0.680) (0.557) (0.351) (0.418) (0.409) (0.323) (0.462) (0.436) (0.374) (0.442) (0.500)

Observations 2,302 2,302 2,302 2,302 2,302 2,302 2,179 2,281 2,281 2,281 2,281 2,281
R-squared 0.247 0.229
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

315
Table O.8. Long-term heterogeneous effects of English tutoring in grade 8 on the CSAT English achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G8 0.13 0.10 0.17 0.16 0.09 0.03 0.39 0.33 0.39 1.04* 0.20 0.60
(0.090) (0.164) (0.129) (0.102) (0.125) (0.170) (0.317) (0.507) (0.437) (0.430) (0.389) (0.467)
URBANICITY -0.02 -0.13 -0.12 -0.09 0.02 0.19 -0.05 -0.17 -0.12 -0.17 -0.02 0.03
(0.083) (0.196) (0.114) (0.115) (0.116) (0.155) (0.086) (0.171) (0.110) (0.107) (0.118) (0.145)
GENDER (1=female) 0.03 0.13 0.18* 0.03 -0.06 -0.12 0.04 0.12 0.17 0.07 -0.03 -0.14
(0.058) (0.099) (0.085) (0.073) (0.085) (0.077) (0.060) (0.139) (0.099) (0.064) (0.073) (0.097)
SES 0.25*** 0.13* 0.27*** 0.27*** 0.37*** 0.29*** 0.20*** 0.14 0.22** 0.18* 0.31*** 0.20**
(0.040) (0.063) (0.043) (0.065) (0.049) (0.057) (0.043) (0.080) (0.074) (0.088) (0.065) (0.075)
SCHOOL TYPE (1=private) 0.14* 0.19 0.25** 0.18* 0.06 0.02 0.13 0.11 0.21* 0.15 0.09 -0.01
(0.073) (0.158) (0.085) (0.089) (0.080) (0.135) (0.076) (0.122) (0.096) (0.085) (0.072) (0.139)
STU-TEA RATIO -0.00 -0.00 -0.01 -0.00 -0.01 -0.01 -0.00 0.00 -0.01 -0.00 -0.02 -0.01
(0.007) (0.014) (0.008) (0.007) (0.009) (0.012) (0.008) (0.017) (0.010) (0.008) (0.009) (0.014)
SCHOOL CHOICE -0.06 -0.21 -0.08 -0.09 -0.01 -0.13 -0.08 -0.20 -0.05 -0.13 -0.02 -0.24
(0.089) (0.168) (0.126) (0.124) (0.107) (0.139) (0.093) (0.195) (0.093) (0.125) (0.120) (0.188)
PRESCORE (G6) 0.99*** 1.14*** 1.04*** 1.05*** 0.95*** 0.90*** 1.03*** 1.13*** 1.07*** 1.07*** 0.97*** 1.00***
(0.042) (0.067) (0.080) (0.066) (0.067) (0.105) (0.045) (0.067) (0.055) (0.068) (0.061) (0.082)
STUDY HOURS (1= >10hrs /week) 0.06 0.01 0.10 -0.02 0.05 0.06 0.07 -0.04 0.14 -0.00 0.06 0.12
(0.081) (0.127) (0.107) (0.092) (0.136) (0.122) (0.084) (0.154) (0.133) (0.074) (0.155) (0.118)
SELF ESTEEM 0.27*** 0.35*** 0.35*** 0.25*** 0.20** 0.19* 0.24*** 0.32** 0.32*** 0.24*** 0.23*** 0.09
(0.051) (0.096) (0.064) (0.065) (0.077) (0.087) (0.053) (0.111) (0.071) (0.060) (0.062) (0.122)
ENGLISH PT in G7 -0.08 0.16 -0.14 -0.09 -0.13 -0.14 -0.09 0.11 -0.18 -0.31 -0.04 -0.22
(0.079) (0.163) (0.089) (0.101) (0.085) (0.180) (0.107) (0.183) (0.150) (0.167) (0.123) (0.164)
ENGLISH PT in G9 -0.00 0.12 -0.03 -0.03 -0.01 0.02 -0.11 0.07 -0.01 -0.40** -0.13 -0.17
(0.091) (0.203) (0.115) (0.099) (0.141) (0.227) (0.148) (0.295) (0.215) (0.152) (0.180) (0.301)
ENGLISH PT in G10 0.04 0.23* 0.09 0.03 -0.01 -0.02 0.03 0.23* 0.05 -0.02 0.01 -0.03
(0.067) (0.099) (0.064) (0.086) (0.086) (0.081) (0.071) (0.109) (0.090) (0.114) (0.081) (0.102)
ENGLISH PT in G11 0.21** 0.24 0.18** 0.13 0.15 0.14 0.18* 0.22 0.11 0.08 0.13 0.05
(0.065) (0.153) (0.062) (0.075) (0.094) (0.095) (0.069) (0.116) (0.080) (0.092) (0.068) (0.125)
ENGLISH PT in G12 0.15* 0.26 0.11 0.18 0.12 0.08 0.17* 0.27* 0.13 0.20* 0.12 0.06
(0.069) (0.149) (0.099) (0.108) (0.122) (0.175) (0.072) (0.104) (0.091) (0.083) (0.098) (0.161)
Constant 5.13*** 3.00*** 4.29*** 5.16*** 6.48*** 7.39*** 5.05*** 2.78*** 4.22*** 5.11*** 6.46*** 7.48***
(0.258) (0.357) (0.320) (0.377) (0.420) (0.333) (0.287) (0.691) (0.376) (0.426) (0.452) (0.415)

Observations 2,339 2,339 2,339 2,339 2,339 2,339 2,148 2,161 2,161 2,161 2,161 2,161
R-squared 0.342 0.342
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

316
Table O.9. Long-term heterogeneous effects of English tutoring in grade 9 on the CSAT English achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G9 -0.05 0.09 -0.01 -0.08 -0.10 -0.05 0.51 0.17 0.59 0.66 0.64 0.35
(0.092) (0.175) (0.137) (0.116) (0.161) (0.217) (0.389) (0.483) (0.684) (0.477) (0.693) (0.793)
URBANICITY 0.01 0.17 -0.06 -0.06 -0.07 0.17 -0.02 0.00 -0.14 -0.11** 0.00 0.25*
(0.083) (0.135) (0.098) (0.071) (0.115) (0.140) (0.089) (0.150) (0.074) (0.043) (0.110) (0.120)
GENDER (1=female) -0.05 0.02 0.02 -0.02 -0.07 -0.28* -0.05 -0.04 0.04 -0.05 -0.13 -0.27*
(0.059) (0.127) (0.079) (0.088) (0.089) (0.141) (0.064) (0.115) (0.096) (0.054) (0.066) (0.118)
SES 0.26*** 0.25** 0.24*** 0.22*** 0.31*** 0.35*** 0.25*** 0.19** 0.23*** 0.20*** 0.28*** 0.34***
(0.041) (0.077) (0.044) (0.047) (0.063) (0.069) (0.044) (0.073) (0.040) (0.046) (0.061) (0.063)
SCHOOL TYPE (1=private) 0.15* 0.01 0.25* 0.21** 0.17* 0.09 0.14 -0.02 0.18* 0.13 0.12 0.07
(0.073) (0.109) (0.098) (0.076) (0.075) (0.140) (0.079) (0.144) (0.076) (0.076) (0.115) (0.179)
STU-TEA RATIO -0.00 0.00 0.01 0.00 -0.01 0.00 -0.00 0.01 0.01 -0.00 -0.01 -0.00
(0.007) (0.014) (0.018) (0.011) (0.012) (0.015) (0.008) (0.012) (0.011) (0.010) (0.009) (0.009)
SCHOOL CHOICE -0.06 0.02 -0.02 -0.03 -0.24 -0.07 -0.10 -0.08 -0.04 -0.09 -0.15 -0.04
(0.090) (0.133) (0.108) (0.094) (0.130) (0.181) (0.096) (0.158) (0.077) (0.084) (0.150) (0.164)
PRESCORE (G6) 0.92*** 1.17*** 0.98*** 0.95*** 0.89*** 0.79*** 0.91*** 1.18*** 0.96*** 0.91*** 0.89*** 0.77***
(0.040) (0.080) (0.084) (0.057) (0.063) (0.096) (0.043) (0.080) (0.079) (0.046) (0.033) (0.057)
STUDY HOURS (1= >10hrs /week) 0.08 0.23 0.05 0.05 0.07 0.07 0.09 0.18 0.06 0.03 0.05 0.17
(0.081) (0.142) (0.073) (0.082) (0.106) (0.127) (0.087) (0.130) (0.093) (0.118) (0.118) (0.154)
SELF ESTEEM 0.33*** 0.36*** 0.40*** 0.36*** 0.29*** 0.18* 0.30*** 0.37*** 0.37*** 0.36*** 0.34*** 0.18*
(0.051) (0.099) (0.080) (0.057) (0.060) (0.076) (0.056) (0.102) (0.048) (0.058) (0.055) (0.079)
ENGLISH PT in G7 -0.03 0.14 0.04 -0.05 -0.10 -0.11 -0.09 0.06 -0.10 -0.19 -0.25 -0.16
(0.079) (0.189) (0.125) (0.105) (0.086) (0.160) (0.092) (0.185) (0.080) (0.100) (0.156) (0.152)
ENGLISH PT in G8 0.12 -0.01 -0.02 0.21 0.10 0.07 -0.10 -0.13 -0.20 -0.05 -0.21 -0.22
(0.090) (0.194) (0.189) (0.146) (0.127) (0.232) (0.167) (0.250) (0.214) (0.208) (0.242) (0.298)
ENGLISH PT in G10 0.07 0.21 0.15 -0.00 0.05 0.03 0.06 0.23 0.10 0.01 -0.01 0.04
(0.068) (0.134) (0.113) (0.092) (0.126) (0.174) (0.081) (0.171) (0.138) (0.063) (0.112) (0.148)
ENGLISH PT in G11 0.22*** 0.16 0.18* 0.26*** 0.16 0.10 0.21** 0.29* 0.19** 0.22** 0.17 0.13
(0.066) (0.168) (0.089) (0.051) (0.110) (0.100) (0.073) (0.129) (0.072) (0.068) (0.096) (0.121)
ENGLISH PT in G12 0.17* 0.25 0.21* 0.17* 0.17* 0.03 0.16* 0.18 0.25*** 0.18** 0.12 -0.10
(0.069) (0.140) (0.084) (0.073) (0.079) (0.140) (0.074) (0.151) (0.066) (0.062) (0.091) (0.116)
Constant 5.18*** 2.79*** 4.01*** 5.08*** 6.71*** 7.51*** 5.07*** 3.00*** 3.87*** 5.10*** 6.48*** 7.34***
(0.259) (0.372) (0.467) (0.243) (0.330) (0.553) (0.303) (0.450) (0.356) (0.316) (0.414) (0.422)

Observations 2,319 2,319 2,319 2,319 2,319 2,319 2,062 2,277 2,277 2,277 2,277 2,277
R-squared 0.337 0.326
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

317
Table O.10. Long-term heterogeneous effects of math tutoring in grade 7 on the CSAT math achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G7 0.11 0.36** 0.12 0.06 -0.02 0.14 -0.13 0.03 -0.26 -0.54 -0.06 2.62*
(0.091) (0.128) (0.200) (0.134) (0.166) (0.112) (0.450) (0.731) (0.593) (0.515) (0.765) (1.164)
URBANICITY -0.05 -0.17 -0.14 -0.04 0.06 -0.03 -0.07 -0.22 -0.13 -0.07 -0.02 -0.29
(0.097) (0.175) (0.130) (0.147) (0.156) (0.146) (0.102) (0.180) (0.087) (0.135) (0.165) (0.206)
GENDER (1=female) -0.21** -0.03 -0.25** -0.26** -0.34*** -0.31* -0.25*** -0.10 -0.24 -0.30*** -0.41*** -0.21*
(0.067) (0.088) (0.096) (0.086) (0.087) (0.127) (0.071) (0.142) (0.140) (0.087) (0.109) (0.100)
SES 0.21*** 0.15** 0.13 0.27*** 0.34*** 0.29*** 0.21*** 0.17** 0.20** 0.30*** 0.30*** 0.21**
(0.046) (0.058) (0.075) (0.049) (0.079) (0.070) (0.051) (0.066) (0.075) (0.054) (0.065) (0.074)
SCHOOL TYPE (1=private) 0.05 0.11 -0.09 -0.07 0.22 0.26* 0.07 0.18 -0.05 -0.03 0.14 0.17
(0.085) (0.106) (0.131) (0.154) (0.127) (0.119) (0.090) (0.104) (0.106) (0.110) (0.126) (0.139)
STU-TEA RATIO 0.00 -0.00 0.02 0.01 -0.00 0.02 0.01 -0.00 0.02 0.01 0.00 0.01
(0.008) (0.021) (0.017) (0.012) (0.010) (0.012) (0.009) (0.016) (0.012) (0.010) (0.012) (0.011)
SCHOOL CHOICE -0.09 -0.18 -0.07 -0.04 -0.05 -0.10 -0.13 -0.24 -0.02 -0.11 -0.10 -0.25
(0.104) (0.145) (0.163) (0.143) (0.131) (0.157) (0.108) (0.180) (0.101) (0.158) (0.144) (0.224)
PRESCORE (G6) 0.23*** 0.22*** 0.21*** 0.25*** 0.23*** 0.24*** 0.22*** 0.21*** 0.18*** 0.24*** 0.26*** 0.23***
(0.021) (0.023) (0.032) (0.043) (0.034) (0.034) (0.022) (0.039) (0.045) (0.026) (0.029) (0.031)
STUDY HOURS 0.14 0.26 0.24 0.19 0.18 -0.03 0.16 0.26 0.28* 0.18 0.13 -0.09
(1= >10hrs /week) (0.094) (0.164) (0.130) (0.103) (0.114) (0.126) (0.097) (0.172) (0.112) (0.101) (0.145) (0.118)
SELF ESTEEM 0.24*** 0.33*** 0.16* 0.14 0.24* 0.38*** 0.23*** 0.32** 0.17 0.14 0.15 0.34***
(0.060) (0.089) (0.073) (0.072) (0.104) (0.079) (0.062) (0.105) (0.102) (0.081) (0.112) (0.071)
MATH PT in G8 -0.04 0.07 0.03 0.01 -0.05 -0.37* 0.04 0.23 0.13 0.19 -0.03 -1.21**
(0.104) (0.184) (0.151) (0.146) (0.220) (0.164) (0.176) (0.336) (0.251) (0.249) (0.356) (0.459)
MATH PT in G9 -0.16 -0.29 -0.10 -0.27* -0.27** -0.02 -0.17 -0.41* -0.05 -0.12 -0.30 -0.40**
(0.107) (0.196) (0.184) (0.113) (0.096) (0.132) (0.131) (0.202) (0.189) (0.198) (0.191) (0.154)
MATH PT in G10 0.30*** 0.44* 0.47*** 0.25 0.30** -0.02 0.33*** 0.60*** 0.47*** 0.31* 0.32* -0.06
(0.088) (0.192) (0.112) (0.151) (0.112) (0.146) (0.092) (0.152) (0.115) (0.138) (0.132) (0.184)
MATH PT in G11 0.62*** 0.77*** 0.77*** 0.69*** 0.45** 0.50*** 0.65*** 0.77*** 0.87*** 0.68*** 0.54*** 0.50***
(0.083) (0.152) (0.112) (0.127) (0.140) (0.149) (0.088) (0.166) (0.114) (0.092) (0.095) (0.150)
MATH PT in G12 0.55*** 0.75*** 0.62*** 0.62*** 0.52*** 0.42*** 0.58*** 0.77*** 0.62*** 0.71*** 0.54*** 0.35**
(0.078) (0.162) (0.141) (0.105) (0.142) (0.117) (0.082) (0.158) (0.132) (0.137) (0.115) (0.129)
Constant 3.57*** 1.13* 2.10*** 3.54*** 5.19*** 5.82*** 3.74*** 1.44* 2.30*** 3.85*** 5.03*** 5.52***
(0.320) (0.466) (0.366) (0.618) (0.543) (0.554) (0.354) (0.600) (0.483) (0.465) (0.537) (0.810)

Observations 2,294 2,294 2,294 2,294 2,294 2,294 2,172 2,265 2,265 2,265 2,265 2,265
R-squared 0.238 0.236
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

318
Table O.11. Long-term heterogeneous effects of math tutoring in grade 8 on the CSAT math achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G8 -0.07 0.04 -0.02 0.05 -0.13 -0.32 0.01 -0.84 -0.29 0.15 -0.05 -0.21
(0.098) (0.216) (0.121) (0.127) (0.159) (0.171) (0.362) (0.562) (0.374) (0.650) (0.700) (0.633)
URBANICITY -0.09 -0.24 -0.20 -0.01 -0.02 -0.16 -0.12 -0.23 -0.14 -0.09 -0.03 -0.12
(0.091) (0.186) (0.176) (0.110) (0.142) (0.144) (0.094) (0.132) (0.150) (0.107) (0.110) (0.215)
GENDER (1=female) -0.39*** -0.17 -0.22** -0.42*** -0.51*** -0.49*** -0.41*** -0.18 -0.29** -0.46*** -0.53*** -0.54***
(0.065) (0.113) (0.082) (0.080) (0.076) (0.087) (0.067) (0.106) (0.099) (0.087) (0.080) (0.118)
SES 0.10* -0.00 0.02 0.13* 0.17* 0.11 0.08 -0.01 0.02 0.12 0.15* 0.10
(0.045) (0.065) (0.085) (0.054) (0.072) (0.060) (0.047) (0.088) (0.089) (0.072) (0.072) (0.089)
SCHOOL TYPE (1=private) 0.09 0.17 -0.02 0.06 0.13 0.19 0.06 0.10 -0.05 0.04 0.09 0.13
(0.081) (0.097) (0.117) (0.127) (0.112) (0.166) (0.084) (0.119) (0.119) (0.104) (0.131) (0.158)
STU-TEA RATIO -0.00 -0.01 -0.00 -0.00 -0.01 -0.01 -0.00 -0.01 0.00 -0.01 -0.01 -0.01
(0.008) (0.010) (0.013) (0.008) (0.012) (0.014) (0.008) (0.018) (0.019) (0.011) (0.013) (0.016)
SCHOOL CHOICE -0.04 -0.11 -0.14 0.12 -0.06 -0.17 -0.06 -0.06 -0.11 0.04 -0.05 -0.17
(0.099) (0.212) (0.194) (0.170) (0.146) (0.207) (0.102) (0.218) (0.167) (0.133) (0.157) (0.214)
PRESCORE (G6) 0.87*** 0.73*** 0.81*** 0.89*** 0.94*** 0.90*** 0.90*** 0.74*** 0.89*** 0.90*** 0.98*** 0.96***
(0.048) (0.087) (0.083) (0.079) (0.059) (0.096) (0.050) (0.108) (0.080) (0.087) (0.065) (0.052)
STUDY HOURS 0.06 0.20 0.09 0.11 0.05 0.09 0.03 0.11 0.14 0.06 0.03 0.07
(1= >10hrs /week) (0.089) (0.116) (0.119) (0.125) (0.126) (0.124) (0.093) (0.123) (0.127) (0.109) (0.135) (0.182)
SELF ESTEEM 0.15** 0.27* 0.08 0.12* 0.15 0.25** 0.14* 0.26 0.03 0.16* 0.10 0.24**
(0.056) (0.108) (0.093) (0.062) (0.084) (0.087) (0.058) (0.133) (0.083) (0.075) (0.086) (0.077)
MATH PT in G7 -0.03 0.27 0.11 -0.10 -0.17 0.09 -0.06 0.46* 0.14 -0.18 -0.22 -0.04
(0.087) (0.166) (0.186) (0.106) (0.149) (0.163) (0.126) (0.200) (0.189) (0.228) (0.234) (0.200)
MATH PT in G9 -0.12 -0.10 -0.11 -0.13 -0.22* 0.05 -0.19 0.21 -0.17 -0.26 -0.23 -0.16
(0.102) (0.182) (0.150) (0.136) (0.104) (0.122) (0.159) (0.278) (0.219) (0.245) (0.237) (0.228)
MATH PT in G10 0.18* 0.28* 0.45*** 0.13 0.11 -0.12 0.18* 0.46* 0.48*** 0.13 0.05 -0.20
(0.084) (0.129) (0.125) (0.125) (0.150) (0.172) (0.089) (0.179) (0.127) (0.104) (0.149) (0.161)
MATH PT in G11 0.63*** 0.74*** 0.69*** 0.63*** 0.52*** 0.59*** 0.61*** 0.74*** 0.61*** 0.66*** 0.53*** 0.62***
(0.079) (0.179) (0.157) (0.090) (0.092) (0.171) (0.083) (0.166) (0.151) (0.106) (0.079) (0.140)
MATH PT in G12 0.38*** 0.67*** 0.48*** 0.45*** 0.34*** 0.22 0.38*** 0.60*** 0.44*** 0.42*** 0.32*** 0.17
(0.075) (0.131) (0.111) (0.084) (0.077) (0.134) (0.078) (0.121) (0.109) (0.089) (0.090) (0.155)
Constant 5.44*** 2.90*** 3.87*** 5.22*** 7.06*** 8.00*** 5.54*** 2.97*** 4.15*** 5.58*** 7.21*** 8.35***
(0.284) (0.524) (0.559) (0.475) (0.418) (0.541) (0.309) (0.648) (0.557) (0.585) (0.455) (0.619)

Observations 2,330 2,330 2,330 2,330 2,330 2,330 2,142 2,153 2,153 2,153 2,153 2,153
R-squared 0.298 0.297
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

319
Table O.12. Long-term heterogeneous effects of math tutoring in grade 9 on the CSAT math achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G9 -0.14 0.06 -0.21 -0.20 -0.19 -0.14 0.44 0.10 0.27 0.53 0.48 0.62
(0.102) (0.209) (0.180) (0.170) (0.131) (0.154) (0.437) (0.452) (0.613) (0.621) (0.489) (0.644)
URBANICITY -0.06 -0.08 -0.07 -0.04 -0.01 -0.08 -0.12 -0.11 -0.14 -0.15 -0.05 -0.13
(0.092) (0.139) (0.106) (0.129) (0.114) (0.175) (0.098) (0.198) (0.136) (0.133) (0.136) (0.188)
GENDER (1=female) -0.46*** -0.22 -0.31*** -0.48*** -0.55*** -0.65*** -0.47*** -0.22 -0.31*** -0.53*** -0.56*** -0.67***
(0.066) (0.156) (0.090) (0.090) (0.127) (0.120) (0.072) (0.125) (0.085) (0.070) (0.075) (0.095)
SES 0.10* 0.04 -0.00 0.10 0.11 0.24** 0.09 0.04 0.04 0.09 0.11 0.17*
(0.045) (0.095) (0.064) (0.078) (0.081) (0.087) (0.048) (0.084) (0.084) (0.076) (0.074) (0.081)
SCHOOL TYPE (1=private) 0.08 0.15 0.04 0.04 0.12 0.18 0.05 0.08 -0.05 0.01 0.10 0.26*
(0.081) (0.150) (0.125) (0.123) (0.105) (0.165) (0.086) (0.144) (0.115) (0.145) (0.096) (0.120)
STU-TEA RATIO -0.00 -0.01 0.00 -0.00 -0.00 0.01 -0.00 -0.01 0.01 -0.01 -0.01 0.01
(0.008) (0.012) (0.010) (0.011) (0.009) (0.014) (0.009) (0.012) (0.015) (0.014) (0.014) (0.014)
SCHOOL CHOICE -0.04 -0.06 -0.05 0.02 -0.05 0.07 -0.14 -0.03 -0.06 -0.11 -0.11 -0.13
(0.099) (0.150) (0.143) (0.168) (0.098) (0.184) (0.106) (0.196) (0.146) (0.122) (0.162) (0.164)
PRESCORE (G6) 0.81*** 0.75*** 0.79*** 0.89*** 0.93*** 0.76*** 0.79*** 0.71*** 0.70*** 0.88*** 0.93*** 0.76***
(0.045) (0.092) (0.085) (0.094) (0.083) (0.078) (0.048) (0.070) (0.073) (0.077) (0.082) (0.049)
STUDY HOURS 0.08 0.27 0.13 0.11 0.08 -0.03 0.09 0.17 0.15 0.12 0.11 0.07
(1= >10hrs /week) (0.090) (0.145) (0.121) (0.131) (0.117) (0.162) (0.097) (0.237) (0.152) (0.162) (0.111) (0.110)
SELF ESTEEM 0.21*** 0.22*** 0.21** 0.14 0.25** 0.28** 0.16** 0.19 0.17* 0.09 0.20** 0.19*
(0.056) (0.064) (0.071) (0.088) (0.078) (0.099) (0.062) (0.101) (0.086) (0.073) (0.070) (0.076)
MATH PT in G7 0.01 0.28* 0.09 -0.13 -0.10 0.08 -0.09 0.11 -0.04 -0.22 -0.20* -0.07
(0.087) (0.133) (0.167) (0.117) (0.122) (0.178) (0.103) (0.141) (0.132) (0.136) (0.100) (0.226)
MATH PT in G8 -0.10 -0.09 0.14 0.02 -0.19 -0.41* -0.25 -0.10 -0.07 -0.23 -0.31 -0.62**
(0.099) (0.147) (0.141) (0.169) (0.178) (0.189) (0.175) (0.209) (0.222) (0.237) (0.205) (0.192)
MATH PT in G10 0.20* 0.36* 0.41** 0.15 0.05 -0.10 0.17 0.28 0.37 0.12 -0.08 -0.22
(0.085) (0.159) (0.156) (0.127) (0.168) (0.173) (0.105) (0.156) (0.199) (0.162) (0.129) (0.117)
MATH PT in G11 0.65*** 0.68*** 0.69*** 0.66*** 0.57*** 0.48** 0.61*** 0.68*** 0.62*** 0.54*** 0.59*** 0.47***
(0.079) (0.135) (0.150) (0.145) (0.144) (0.170) (0.090) (0.140) (0.151) (0.070) (0.086) (0.134)
MATH PT in G12 0.38*** 0.55** 0.44*** 0.39*** 0.32** 0.20 0.37*** 0.64*** 0.52*** 0.39*** 0.29* 0.27
(0.075) (0.191) (0.110) (0.116) (0.103) (0.103) (0.080) (0.128) (0.071) (0.097) (0.126) (0.139)
Constant 5.45*** 2.82*** 3.77*** 5.49*** 7.00*** 7.86*** 5.47*** 2.88*** 3.61*** 5.72*** 6.95*** 7.94***
(0.285) (0.416) (0.400) (0.501) (0.346) (0.507) (0.332) (0.627) (0.489) (0.481) (0.420) (0.514)

Observations 2,310 2,310 2,310 2,310 2,310 2,310 2,055 2,252 2,252 2,252 2,252 2,252
R-squared 0.296 0.284
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

320
Table P.1. Heterogeneous effects of one year of private tutoring in high school on the CSAT overall achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of PT in high school 0.06 0.22 0.17 0.19* -0.00 -0.27 0.36 0.41 0.57*** 0.46*** 0.35 0.26
(0.104) (0.152) (0.125) (0.090) (0.145) (0.140) (0.376) (0.246) (0.157) (0.107) (0.198) (0.255)
URBAN -0.12 -0.20 -0.16 0.02 -0.11 -0.01 -0.27 -0.01 -0.10 -0.01 -0.12 0.04
(0.159) (0.200) (0.220) (0.169) (0.302) (0.229) (0.189) (0.120) (0.104) (0.099) (0.117) (0.074)
GENDER (1=female) -0.15 -0.02 0.03 -0.10 -0.28* -0.27 -0.23 0.18 0.04 -0.11 -0.20 -0.20
(0.100) (0.161) (0.113) (0.128) (0.135) (0.157) (0.122) (0.114) (0.051) (0.079) (0.134) (0.144)
SES 0.11 0.00 0.14 0.24* 0.06 0.23* 0.09 0.13 0.14** 0.15*** 0.25*** 0.25***
(0.080) (0.194) (0.132) (0.097) (0.104) (0.104) (0.100) (0.088) (0.052) (0.041) (0.068) (0.061)
SCHOOL TYPE 0.03 0.11 -0.09 0.16 -0.00 -0.14 0.09 0.19* 0.14 0.15* 0.21* 0.22**
(0.105) (0.153) (0.124) (0.111) (0.164) (0.160) (0.124) (0.094) (0.086) (0.071) (0.098) (0.086)
SCHOOL TRACK 0.34* 0.51* 0.68*** 0.44** 0.36* 0.20 0.46* 1.26*** 0.87*** 0.47*** 0.52*** 0.54**
(0.141) (0.228) (0.133) (0.143) (0.152) (0.247) (0.181) (0.177) (0.239) (0.132) (0.146) (0.195)
STU-TEA RATIO -0.01 0.04 -0.02 -0.03 -0.01 -0.00 -0.02 -0.05** -0.03* -0.04** -0.03 -0.03
(0.020) (0.047) (0.029) (0.020) (0.018) (0.021) (0.024) (0.020) (0.014) (0.011) (0.022) (0.027)
SCHOOL CHOICE -0.06 0.14 -0.02 0.05 -0.06 -0.31 -0.23 -0.19 -0.15 -0.15 -0.18 -0.20
(0.155) (0.258) (0.156) (0.173) (0.264) (0.235) (0.185) (0.133) (0.099) (0.109) (0.142) (0.116)
PRESCORE (G9) 0.82*** 0.90*** 0.75*** 0.80*** 0.89*** 0.86*** 0.80*** 0.77*** 0.70*** 0.76*** 0.69*** 0.66***
(0.071) (0.074) (0.075) (0.082) (0.107) (0.082) (0.085) (0.104) (0.051) (0.046) (0.063) (0.058)
SELF-STUDY 0.46*** 0.56** 0.45** 0.34*** 0.35 0.61*** 0.35* 0.27* 0.17* 0.31*** 0.37*** 0.40***
(0.121) (0.176) (0.157) (0.095) (0.207) (0.159) (0.144) (0.121) (0.068) (0.066) (0.102) (0.098)
SELF-ESTEEM 0.28*** 0.13 0.20 0.27* 0.34** 0.29 0.34** 0.47*** 0.46*** 0.41*** 0.41*** 0.28*
(0.084) (0.156) (0.124) (0.121) (0.128) (0.158) (0.104) (0.079) (0.085) (0.049) (0.072) (0.117)
PT in middle school -0.06 -0.06 0.01 -0.08 -0.11 -0.20* -0.10 0.13 0.05 -0.03 -0.05 -0.09
(0.047) (0.046) (0.065) (0.055) (0.078) (0.089) (0.065) (0.085) (0.077) (0.055) (0.077) (0.062)
Constant 5.09*** 2.20* 3.79*** 5.12*** 6.37*** 7.69*** 5.56*** 2.81*** 3.80*** 5.36*** 6.40*** 7.13***
(0.449) (1.011) (0.667) (0.436) (0.653) (0.675) (0.561) (0.566) (0.415) (0.287) (0.356) (0.440)

Observations 639 639 639 639 639 639 465 1,571 1,571 1,571 1,571 1,571
R-squared 0.327 0.319
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

321
Table P.2. Heterogeneous effects of two years of private tutoring in high school on the CSAT overall achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of PT in high school 0.29** 0.45** 0.52*** 0.37** 0.09 0.14 0.51 0.51 0.71*** 0.57*** 0.44 0.32
(0.101) (0.149) (0.122) (0.117) (0.152) (0.194) (0.304) (0.285) (0.162) (0.127) (0.241) (0.326)
URBAN -0.22 -0.14 -0.16 -0.10 -0.29 -0.13 -0.20 -0.03 -0.12 -0.03 -0.14 0.03
(0.119) (0.154) (0.127) (0.139) (0.188) (0.210) (0.137) (0.166) (0.097) (0.087) (0.129) (0.109)
GENDER (1=female) -0.12 -0.01 -0.08 -0.14 -0.21 -0.26* -0.08 0.17 0.02 -0.12 -0.21 -0.21
(0.083) (0.115) (0.103) (0.114) (0.119) (0.123) (0.097) (0.109) (0.078) (0.083) (0.163) (0.131)
SES 0.10 0.03 0.06 0.12** 0.16 0.24* 0.08 0.11 0.11 0.12** 0.23*** 0.23**
(0.054) (0.092) (0.061) (0.039) (0.129) (0.100) (0.064) (0.056) (0.058) (0.046) (0.068) (0.083)
SCHOOL TYPE 0.05 0.08 0.07 0.11 0.06 -0.02 0.07 0.19 0.13 0.14 0.20* 0.22*
(0.085) (0.182) (0.128) (0.081) (0.123) (0.143) (0.098) (0.110) (0.095) (0.083) (0.097) (0.087)
SCHOOL TRACK 0.54*** 0.95*** 0.57*** 0.53*** 0.61* 0.24 0.66*** 1.20*** 0.79** 0.40* 0.47* 0.50
(0.134) (0.265) (0.155) (0.120) -0.254 (0.341) (0.170) (0.214) (0.242) (0.177) (0.186) (0.277)
STU-TEA RATIO -0.03 -0.00 -0.03 -0.03 -0.03 -0.05 -0.05* -0.05* -0.03* -0.04* -0.03 -0.03
(0.018) (0.033) (0.025) (0.028) (0.035) (0.043) (0.021) (0.028) (0.016) (0.015) (0.028) (0.029)
SCHOOL CHOICE -0.29* -0.20 -0.17 -0.19 -0.45** -0.39* -0.26 -0.18 -0.13 -0.13 -0.16 -0.19
(0.123) (0.178) (0.121) (0.175) (0.158) (0.195) (0.148) (0.210) (0.081) (0.120) (0.160) (0.144)
PRESCORE (G9) 0.74*** 0.81*** 0.81*** 0.79*** 0.72*** 0.69*** 0.70*** 0.77*** 0.70*** 0.77*** 0.69*** 0.66***
(0.056) (0.116) (0.069) (0.079) (0.118) (0.092) (0.065) (0.096) (0.045) (0.050) (0.075) (0.074)
SELF-STUDY 0.40*** 0.22 0.40*** 0.39*** 0.31* 0.42* 0.42*** 0.26 0.16 0.30*** 0.36** 0.39**
(0.093) (0.137) (0.071) (0.103) (0.134) (0.203) (0.109) (0.144) (0.101) (0.082) (0.130) (0.132)
SELF-ESTEEM 0.31*** 0.50*** 0.28*** 0.21* 0.27** 0.29** 0.35*** 0.47*** 0.46*** 0.41*** 0.41*** 0.28**
(0.067) (0.131) (0.061) (0.082) (0.090) (0.090) (0.080) (0.056) (0.084) (0.059) (0.099) (0.106)
PT in middle school 0.02 0.04 0.03 -0.03 0.02 -0.04 0.03 0.11 0.01 -0.05 -0.07 -0.10
(0.047) (0.085) (0.064) (0.056) (0.082) (0.091) (0.071) (0.095) (0.056) (0.055) (0.069) (0.075)
Constant 5.51*** 2.95*** 4.31*** 5.37*** 6.81*** 8.16*** 5.44*** 2.90*** 3.92*** 5.46*** 6.48*** 7.18***
(0.386) (0.588) (0.530) (0.567) (0.514) (1.025) (0.459) (0.576) (0.417) (0.326) (0.482) (0.610)

Observations 826 826 826 826 826 826 627 1,571 1,571 1,571 1,571 1,571
R-squared 0.407 0.402
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

322
Table P.3. Heterogeneous effects of three years of private tutoring in high school on the CSAT overall achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of PT in high school 0.59*** 0.56* 0.72*** 0.64*** 0.48*** 0.45** 0.51* 0.54* 0.75*** 0.61*** 0.46* 0.34
(0.105) (0.245) (0.166) (0.173) (0.140) (0.175) (0.251) (0.268) (0.184) (0.160) (0.214) (0.329)
URBAN -0.08 -0.04 -0.03 -0.11 -0.18 -0.04 -0.11 -0.01 -0.10 -0.01 -0.13 0.04
(0.108) (0.238) (0.141) (0.172) (0.238) (0.113) (0.120) (0.132) (0.112) (0.095) (0.167) (0.100)
GENDER (1=female) -0.04 0.19* 0.04 -0.04 -0.15 -0.25 -0.05 0.18 0.04 -0.11 -0.20 -0.20**
(0.072) (0.091) (0.093) (0.094) (0.132) (0.161) (0.082) (0.126) (0.058) (0.062) (0.144) (0.075)
SES 0.13** 0.07 0.12 0.15* 0.22* 0.27** 0.13* 0.10 0.10 0.12** 0.23** 0.23**
(0.048) (0.088) (0.081) (0.066) (0.086) (0.086) (0.058) (0.087) (0.060) (0.036) (0.072) (0.077)
SCHOOL TYPE 0.21** 0.20 0.18 0.32** 0.25** 0.26* 0.23** 0.18 0.11 0.13 0.20** 0.21*
(0.074) (0.122) (0.101) (0.111) (0.087) (0.111) (0.082) (0.107) (0.066) (0.069) (0.076) (0.094)
SCHOOL TRACK 0.52*** 0.91*** 0.36* 0.60** 0.56** 0.55*** 0.75*** 1.20*** 0.79*** 0.40* 0.47** 0.50
(0.132) (0.260) (0.173) (0.191) (0.201) (0.153) (0.165) (0.194) (0.223) (0.174) (0.151) (0.422)
STU-TEA RATIO -0.05** -0.03 -0.03 -0.07* -0.07** -0.08** -0.04* -0.05 -0.03 -0.04* -0.03 -0.03
(0.016) (0.024) (0.027) (0.027) (0.024) (0.030) (0.019) (0.033) (0.019) (0.017) (0.018) (0.023)
SCHOOL CHOICE -0.11 -0.04 -0.02 -0.01 -0.18 -0.14 -0.17 -0.16 -0.11 -0.11 -0.15 -0.18
(0.115) (0.217) (0.128) (0.173) (0.200) (0.149) (0.129) (0.144) (0.080) (0.113) (0.155) (0.126)
PRESCORE (G9) 0.73*** 0.87*** 0.82*** 0.75*** 0.69*** 0.62*** 0.68*** 0.76*** 0.68*** 0.75*** 0.68*** 0.66***
(0.051) (0.097) (0.088) (0.067) (0.058) (0.072) (0.057) (0.078) (0.055) (0.042) (0.067) (0.063)
SELF-STUDY 0.26** 0.29** 0.13 0.25* 0.33* 0.27* 0.16 0.25* 0.15* 0.30*** 0.36*** 0.39**
(0.081) (0.097) (0.124) (0.120) (0.139) (0.116) (0.090) (0.104) (0.078) (0.080) (0.104) (0.137)
SELF-ESTEEM 0.43*** 0.46*** 0.46*** 0.43*** 0.44*** 0.31* 0.48*** 0.48*** 0.47*** 0.41*** 0.41*** 0.29**
(0.061) (0.083) (0.102) (0.084) (0.095) (0.122) (0.068) (0.076) (0.076) (0.063) (0.083) (0.100)
PT in middle school -0.03 0.10 0.01 -0.05 -0.00 -0.11 0.03 0.11 0.01 -0.05 -0.07 -0.10
(0.050) (0.115) (0.097) (0.078) (0.066) (0.096) (0.071) (0.073) (0.058) (0.041) (0.081) (0.092)
Constant 5.44*** 2.59*** 4.13*** 5.55*** 6.77*** 8.03*** 5.24*** 2.91*** 3.93*** 5.46*** 6.48*** 7.19***
(0.356) (0.666) (0.534) (0.455) (0.527) (0.854) (0.407) (0.624) (0.427) (0.360) (0.440) (0.624)

Observations 985 985 985 985 985 985 782 1,571 1,571 1,571 1,571 1,571
R-squared 0.453 0.442
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

323
Table P.4. Heterogeneous effects of one year of verbal tutoring in high school on the CSAT verbal achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of Verbal PT in high school 0.01 -0.11 -0.05 0.06 0.08 -0.09 0.54 0.7 0.36 0.83* 0.88* 0.55
(0.071) (0.092) (0.078) (0.073) (0.090) (0.108) (0.464) -0.687 -0.429 -0.337 -0.366 -0.599
URBAN -0.03 -0.12 -0.02 -0.10 -0.03 -0.14 -0.09 -0.03 0.01 -0.15 -0.00 -0.03
(0.101) (0.195) (0.152) (0.130) (0.158) (0.188) (0.124) (0.167) (0.134) (0.132) (0.158) (0.140)
GENDER (1=female) 0.12 0.20 0.18* 0.11 -0.01 0.02 0.03 0.28* 0.22* 0.03 -0.21* -0.22
(0.072) (0.188) (0.093) (0.074) (0.121) (0.157) (0.090) (0.112) (0.091) (0.105) (0.103) (0.124)
SES 0.08 0.01 0.12 0.16** 0.05 0.03 0.04 0.05 0.11 0.09 0.02 0.11
(0.047) (0.135) (0.073) (0.049) (0.051) (0.041) (0.060) (0.106) (0.064) (0.052) (0.060) (0.096)
SCHOOL TYPE 0.08 0.08 0.12 0.04 0.07 0.09 0.03 0.06 0.10 0.00 0.03 -0.06
(0.072) (0.141) (0.120) (0.086) (0.089) (0.148) (0.084) (0.129) (0.089) (0.083) (0.076) (0.163)
SCHOOL TRACK 0.58*** 0.59*** 0.91*** 0.76*** 0.49*** 0.36 0.68*** 0.86*** 1.05*** 0.92*** 0.53* 0.68
(0.119) (0.168) (0.168) (0.146) (0.118) (0.226) (0.151) (0.219) (0.179) (0.176) (0.239) (0.379)
STU-TEA RATIO -0.02 -0.02 -0.02 -0.03* -0.03 -0.01 -0.02 -0.05 -0.03 -0.04 -0.00 -0.05
(0.016) (0.024) (0.027) (0.014) (0.019) (0.029) (0.020) (0.033) (0.020) (0.021) (0.024) (0.030)
SCHOOL CHOICE -0.01 -0.01 -0.03 -0.05 0.04 -0.22 0.03 0.07 0.04 -0.05 0.15 -0.06
(0.105) (0.158) (0.126) (0.096) (0.170) (0.225) (0.126) (0.206) (0.180) (0.140) (0.139) (0.145)
PRESCORE (G9) 0.77*** 0.87*** 0.66*** 0.78*** 0.76*** 0.81*** 0.76*** 0.69*** 0.63*** 0.76*** 0.75*** 0.71***
(0.048) (0.089) (0.057) (0.070) (0.052) (0.058) (0.057) (0.123) (0.062) (0.040) (0.053) (0.069)
SELF-STUDY 0.11 0.14 0.12 -0.04 0.10 0.16 0.11 0.13 0.05 0.01 0.14 0.24
(0.078) (0.127) (0.104) (0.108) (0.125) (0.162) (0.092) (0.131) (0.117) (0.083) (0.108) (0.167)
SELF-ESTEEM 0.31*** 0.31* 0.37*** 0.31*** 0.31*** 0.25* 0.34*** 0.56*** 0.53*** 0.38*** 0.31*** 0.21
(0.060) (0.146) (0.087) (0.084) (0.092) (0.109) (0.071) (0.141) (0.064) (0.040) (0.076) (0.125)
Verbal PT in middle school -0.08* -0.01 -0.08* -0.08** -0.13** -0.11* -0.13* -0.08 -0.10* -0.15** -0.20*** -0.22*
(0.031) (0.057) (0.035) (0.028) (0.044) (0.042) (0.052) (0.079) (0.052) (0.047) (0.041) (0.092)
Constant 5.26*** 3.21*** 4.06*** 5.42*** 6.59*** 7.52*** 5.07*** 3.07*** 3.73*** 5.37*** 6.09*** 8.13***
(0.333) (0.616) (0.474) (0.321) (0.434) (0.609) (0.448) (0.769) (0.554) (0.514) (0.437) (0.691)

Observations 1,596 1,596 1,596 1,596 1,596 1,596 1,227 1,672 1,672 1,672 1,672 1,672
R-squared 0.256 0.230
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

324
Table P.5. Heterogeneous effects of two years of verbal tutoring in high school on the CSAT verbal achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of Verbal PT in high school 0.24** 0.43** 0.15 0.23* 0.20 0.28* 0.34 0.38 0.19 0.45 0.47* 0.30
(0.093) (0.161) (0.131) (0.117) (0.139) (0.129) (0.240) (0.429) (0.278) (0.232) (0.206) (0.363)
URBAN -0.17 -0.13 -0.09 -0.30 -0.15 -0.09 -0.12 -0.04 0.01 -0.15 -0.01 -0.03
(0.121) (0.195) (0.147) (0.204) (0.221) (0.187) (0.139) (0.168) (0.159) (0.172) (0.166) (0.169)
GENDER (1=female) 0.14 0.12 0.22* 0.17 0.11 -0.16 0.11 0.31* 0.24* 0.06 -0.17 -0.20
(0.081) (0.121) (0.099) (0.115) (0.142) (0.144) (0.093) (0.155) (0.097) (0.067) (0.097) (0.154)
SES 0.10 0.02 0.13* 0.11* 0.09 0.07 0.09 0.04 0.11 0.08 0.02 0.11
(0.053) (0.109) (0.052) (0.049) (0.086) (0.121) (0.066) (0.166) (0.057) (0.049) (0.062) (0.075)
SCHOOL TYPE -0.01 0.02 0.06 0.01 -0.05 -0.19 -0.04 0.07 0.10 0.02 0.05 -0.05
(0.083) (0.136) (0.102) (0.084) (0.127) (0.174) (0.095) (0.168) (0.114) (0.093) (0.110) (0.126)
SCHOOL TRACK 0.64*** 0.76** 0.83*** 0.72*** 0.53** 0.42 0.81*** 0.89*** 1.06*** 0.96*** 0.58*** 0.71*
(0.138) (0.257) (0.210) (0.174) (0.190) (0.248) (0.168) (0.241) (0.208) (0.179) (0.118) (0.327)
STU-TEA RATIO -0.01 -0.03 -0.03 -0.04 -0.01 0.01 -0.02 -0.05 -0.03 -0.04* -0.01 -0.06
(0.018) (0.031) (0.027) (0.019) (0.024) (0.033) (0.021) (0.038) (0.021) (0.018) (0.026) (0.037)
SCHOOL CHOICE -0.08 0.03 -0.00 -0.19 0.04 -0.10 -0.04 0.05 0.03 -0.07 0.13 -0.07
(0.123) (0.175) (0.173) (0.170) (0.206) (0.175) (0.137) (0.153) (0.151) (0.164) (0.135) (0.160)
PRESCORE (G9) 0.66*** 0.79*** 0.68*** 0.71*** 0.64*** 0.66*** 0.66*** 0.68*** 0.63*** 0.75*** 0.75*** 0.70***
(0.055) (0.082) (0.091) (0.057) (0.049) (0.118) (0.063) (0.105) (0.068) (0.049) (0.064) (0.088)
SELF-STUDY 0.20* 0.11 0.11 0.09 0.23 0.39** 0.20 0.14 0.06 0.02 0.15 0.25*
(0.089) (0.171) (0.132) (0.090) (0.143) (0.148) (0.102) (0.126) (0.086) (0.095) (0.112) (0.100)
SELF-ESTEEM 0.42*** 0.35* 0.49*** 0.50*** 0.44*** 0.29*** 0.45*** 0.57*** 0.53*** 0.40*** 0.33*** 0.22*
(0.065) (0.162) (0.089) (0.089) (0.105) (0.089) (0.075) (0.095) (0.081) (0.060) (0.072) (0.098)
Verbal PT in middle school -0.10** -0.05 -0.16*** -0.07* -0.15* -0.15* -0.12* -0.06 -0.09* -0.12* -0.17** -0.20**
(0.036) (0.066) (0.045) (0.037) (0.060) (0.066) (0.046) (0.061) (0.038) (0.059) (0.061) (0.063)
Constant 5.27*** 3.52*** 4.35*** 5.72*** 6.13*** 7.44*** 5.15*** 3.24*** 3.82*** 5.58*** 6.31*** 8.27***
(0.381) (0.702) (0.700) (0.476) (0.722) (0.614) (0.444) (0.663) (0.466) (0.537) (0.388) (0.623)

Observations 1,208 1,208 1,208 1,208 1,208 1,208 938 1,672 1,672 1,672 1,672 1,672
R-squared 0.273 0.275
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

325
Table P.6. Heterogeneous effects of three years of verbal tutoring in high school on the CSAT verbal achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of Verbal PT in high school 0.23 0.32 0.01 0.16 0.46* 0.51* 0.23 0.42 0.21 0.50 0.53** 0.33
(0.143) (0.269) (0.188) (0.175) (0.221) (0.245) (0.258) (0.549) (0.346) (0.309) (0.173) (0.304)
URBAN -0.16 -0.34 -0.11 -0.19 0.07 -0.08 -0.09 -0.03 0.02 -0.14 0.01 -0.02
(0.136) (0.276) (0.180) (0.193) (0.198) (0.279) (0.155) (0.152) (0.157) (0.166) (0.138) (0.169)
GENDER (1=female) 0.17 0.15 0.21 0.25* 0.20 -0.16 0.15 0.31 0.24* 0.06 -0.18* -0.20
(0.092) (0.197) (0.127) (0.118) (0.179) (0.141) (0.107) (0.169) (0.113) (0.082) (0.089) (0.115)
SES 0.07 -0.06 0.01 0.14* 0.05 -0.01 0.08 0.06 0.11 0.10 0.04 0.12
(0.061) (0.178) (0.108) (0.068) (0.075) (0.103) (0.073) (0.145) (0.063) (0.053) (0.049) (0.095)
SCHOOL TYPE -0.08 -0.01 -0.02 -0.09 -0.12 -0.11 -0.11 0.07 0.10 0.02 0.05 -0.05
(0.093) (0.206) (0.135) (0.097) (0.125) (0.184) (0.107) (0.149) (0.088) (0.081) (0.096) (0.122)
SCHOOL TRACK 0.66*** 0.65* 0.78*** 0.83*** 0.54** 0.49 0.82*** 0.91*** 1.07*** 0.98*** 0.59*** 0.72
(0.149) (0.320) (0.197) (0.162) (0.191) (0.301) (0.187) (0.219) (0.188) (0.134) (0.155) (0.485)
STU-TEA RATIO -0.03 -0.02 -0.04 -0.04* -0.04 0.02 -0.02 -0.05 -0.03 -0.04** -0.01 -0.06
(0.020) (0.040) (0.029) (0.022) (0.036) (0.031) (0.024) (0.045) (0.022) (0.015) (0.025) (0.033)
SCHOOL CHOICE -0.19 -0.24 -0.24 -0.19 0.09 -0.16 -0.13 0.05 0.03 -0.08 0.12 -0.08
(0.137) (0.236) (0.160) (0.173) (0.179) (0.260) (0.154) (0.211) (0.181) (0.149) (0.165) (0.184)
PRESCORE (G9) 0.72*** 0.89*** 0.71*** 0.75*** 0.66*** 0.64*** 0.66*** 0.67*** 0.62*** 0.74*** 0.73*** 0.69***
(0.063) (0.112) (0.086) (0.064) (0.130) (0.103) (0.074) (0.112) (0.067) (0.052) (0.061) (0.075)
SELF-STUDY 0.14 0.24 0.07 -0.01 0.20 0.20 0.17 0.14 0.06 0.02 0.16 0.25*
(0.103) (0.192) (0.106) (0.103) (0.175) (0.164) (0.118) (0.202) (0.088) (0.088) (0.103) (0.117)
SELF-ESTEEM 0.41*** 0.33* 0.51*** 0.40*** 0.43*** 0.36** 0.43*** 0.57*** 0.53*** 0.40*** 0.33*** 0.22
(0.074) (0.151) (0.105) (0.073) (0.090) (0.134) (0.087) (0.134) (0.083) (0.070) (0.090) (0.129)
-
Verbal PT in middle school -0.07 -0.04 -0.11 -0.06 -0.12 -0.12 -0.06 -0.04 -0.08 -0.11* -0.15** 0.19***
(0.039) (0.064) (0.075) (0.045) (0.060) (0.068) (0.047) (0.071) (0.057) (0.044) (0.050) (0.055)
Constant 5.56*** 3.69*** 4.86*** 5.63*** 6.30*** 7.29*** 5.22*** 3.26** 3.83*** 5.60*** 6.33*** 8.28***
(0.422) (0.941) (0.540) (0.553) (0.549) (0.674) (0.498) (1.078) (0.599) (0.424) (0.566) (0.546)

Observations 988 988 988 988 988 988 760 1,672 1,672 1,672 1,672 1,672
R-squared 0.290 0.270
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

326
Table P.7. Heterogeneous effects of one year of English tutoring in high school on the CSAT English achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of English PT in high school -0.12 -0.22 -0.15 -0.17 -0.06 0.00 0.41 0.61* 0.57** 0.47* 0.31 0.12
(0.090) (0.174) (0.107) (0.104) (0.120) (0.184) (0.349) (0.264) (0.194) (0.210) (0.175) (0.279)
URBAN 0.04 0.08 0.02 -0.01 0.09 0.37 -0.03 0.05 -0.17 -0.05 0.02 0.26*
(0.128) (0.281) (0.178) (0.179) (0.185) (0.231) (0.150) (0.258) (0.139) (0.104) (0.107) (0.132)
GENDER (1=female) 0.16 0.42*** 0.31** 0.14 0.05 0.00 0.14 0.30* 0.19* 0.07 -0.01 -0.19*
(0.088) (0.127) (0.113) (0.106) (0.117) (0.130) (0.104) (0.122) (0.082) (0.069) (0.067) (0.076)
SES 0.26*** 0.10 0.22*** 0.26* 0.36*** 0.32*** 0.24** 0.20 0.23** 0.29*** 0.31*** 0.33***
(0.066) (0.065) (0.059) (0.104) (0.061) (0.091) (0.082) (0.143) (0.083) (0.062) (0.087) (0.083)
SCHOOL TYPE 0.09 0.36 0.18 -0.00 0.04 0.10 0.21 0.42* 0.20* 0.13 0.22* 0.22*
(0.091) (0.198) (0.106) (0.131) (0.142) (0.128) (0.106) (0.166) (0.099) (0.089) (0.098) (0.106)
SCHOOL TRACK 0.57*** 0.97*** 0.99*** 0.48*** 0.60** 0.64* 0.67*** 1.05*** 0.87*** 0.68** 0.76*** 0.90***
(0.139) (0.185) (0.224) (0.141) (0.190) (0.316) (0.182) (0.148) (0.204) (0.220) (0.153) (0.179)
STU-TEA RATIO -0.02 -0.05* -0.04 -0.03 -0.02 -0.03 -0.04 -0.04 -0.04 -0.06* -0.05* -0.04
(0.018) (0.023) (0.029) (0.022) (0.026) (0.041) (0.022) (0.021) (0.021) (0.023) (0.023) (0.029)
SCHOOL CHOICE -0.13 0.12 -0.15 -0.29 -0.03 0.05 -0.22 -0.17 -0.30* -0.20 -0.12 0.05
(0.131) (0.266) (0.187) (0.183) (0.224) (0.180) (0.154) (0.238) (0.130) (0.109) (0.131) (0.157)
PRESCORE (G9) 0.79*** 1.03*** 0.84*** 0.85*** 0.79*** 0.64*** 0.80*** 0.92*** 0.81*** 0.78*** 0.73*** 0.55***
(0.059) (0.093) (0.115) (0.090) (0.118) (0.130) (0.072) (0.094) (0.079) (0.064) (0.060) (0.069)
SELF-STUDY 0.41*** 0.44** 0.30** 0.53*** 0.28 0.35 0.28* 0.19 0.27** 0.36*** 0.34** 0.28
(0.100) (0.142) (0.099) (0.113) (0.158) (0.187) (0.119) (0.143) (0.086) (0.094) (0.112) (0.157)
SELF-ESTEEM 0.44*** 0.48*** 0.47*** 0.46*** 0.37*** 0.43*** 0.46*** 0.51*** 0.50*** 0.44*** 0.37*** 0.44***
(0.074) (0.114) (0.067) (0.072) (0.107) (0.093) (0.089) (0.116) (0.061) (0.050) (0.075) (0.096)
English PT in middle school 0.02 0.07 0.09 0.00 -0.03 -0.06 -0.02 0.09 0.05 -0.00 -0.07 -0.05
(0.043) (0.063) (0.055) (0.048) (0.053) (0.069) (0.059) (0.051) (0.049) (0.073) (0.059) (0.063)
Constant 4.78*** 1.94** 3.41*** 5.29*** 5.69*** 6.64*** 4.88*** 2.02** 3.69*** 5.36*** 6.31*** 6.96***
(0.402) (0.691) (0.602) (0.565) (0.607) (0.600) (0.491) (0.697) (0.466) (0.461) (0.456) (0.666)

Observations 1,157 1,157 1,157 1,157 1,157 1,157 878 1,887 1,887 1,887 1,887 1,887
R-squared 0.327 0.298
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

327
Table P.8. Heterogeneous effects of two years of English tutoring in high school on the CSAT English achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of English PT in high school 0.22* 0.11 0.22 0.26 0.14 0.10 0.18 0.61 0.57*** 0.47* 0.31 0.12
(0.090) (0.238) (0.129) (0.160) (0.141) (0.162) (0.221) (0.336) (0.153) (0.227) (0.181) (0.177)
URBAN -0.21 0.16 -0.38* -0.18 -0.32* 0.10 -0.20 0.05 -0.16 -0.05 0.03 0.26
(0.118) (0.359) (0.189) (0.159) (0.155) (0.174) (0.134) (0.209) (0.159) (0.129) (0.119) (0.151)
GENDER (1=female) 0.02 0.02 0.09 0.13 0.02 -0.05 0.02 0.30* 0.19* 0.07 -0.01 -0.19
(0.081) (0.137) (0.127) (0.099) (0.151) (0.127) (0.093) (0.140) (0.095) (0.055) (0.092) (0.111)
SES 0.18*** 0.15 0.13** 0.17* 0.30*** 0.29*** 0.17** 0.17 0.19** 0.26*** 0.29*** 0.32***
(0.052) (0.097) (0.051) (0.075) (0.072) (0.074) (0.063) (0.110) (0.061) (0.046) (0.072) (0.071)
SCHOOL TYPE 0.11 -0.05 0.15 0.12 0.17 0.13 0.17 0.42*** 0.20* 0.14 0.22 0.22
(0.083) (0.193) (0.096) (0.119) (0.118) (0.138) (0.095) (0.125) (0.102) (0.101) (0.131) (0.138)
SCHOOL TRACK 0.62*** 0.54* 0.80*** 0.55*** 0.69*** 0.93** 0.76*** 1.06*** 0.88*** 0.69*** 0.76*** 0.90***
(0.140) (0.226) (0.190) (0.128) (0.192) (0.334) (0.166) (0.150) (0.168) (0.158) (0.179) (0.200)
STU-TEA RATIO -0.04 -0.01 -0.04 -0.04 -0.05 -0.07*** -0.05* -0.04 -0.04 -0.06* -0.05 -0.05
(0.018) (0.021) (0.024) (0.025) (0.034) (0.016) (0.022) (0.026) (0.032) (0.027) (0.027) (0.027)
SCHOOL CHOICE -0.31* -0.02 -0.41* -0.36*** -0.41* -0.21 -0.29* -0.17 -0.31 -0.21 -0.12 0.05
(0.121) (0.324) (0.187) (0.095) (0.168) (0.142) (0.137) (0.168) (0.171) (0.138) (0.107) (0.152)
PRESCORE (G9) 0.79*** 1.09*** 0.89*** 0.75*** 0.69*** 0.72*** 0.79*** 0.90*** 0.79*** 0.77*** 0.72*** 0.54***
(0.057) (0.151) (0.090) (0.057) (0.084) (0.094) (0.064) (0.080) (0.052) (0.045) (0.083) (0.060)
SELF-STUDY 0.42*** 0.38 0.41* 0.44*** 0.48*** 0.33* 0.36*** 0.15 0.24* 0.33* 0.32*** 0.28**
(0.090) (0.257) (0.168) (0.115) (0.144) (0.164) (0.105) (0.141) (0.103) (0.130) (0.092) (0.106)
SELF-ESTEEM 0.40*** 0.54*** 0.40*** 0.41*** 0.33* 0.24 0.41*** 0.53*** 0.51*** 0.45*** 0.37*** 0.45***
(0.066) (0.136) (0.091) (0.082) (0.134) (0.132) (0.076) (0.083) (0.053) (0.060) (0.088) (0.071)
English PT in middle school 0.01 0.11 0.11 -0.01 0.00 -0.09 0.05 0.07 0.04 -0.01 -0.08 -0.05
(0.045) (0.100) (0.078) (0.059) (0.074) (0.069) (0.060) (0.085) (0.066) (0.044) (0.065) (0.055)
Constant 5.48*** 2.58** 4.27*** 5.47*** 6.77*** 7.73*** 5.45*** 2.18*** 3.83*** 5.48*** 6.39*** 6.99***
(0.394) (0.837) (0.611) (0.390) (0.523) (0.513) (0.461) (0.550) (0.422) (0.394) (0.415) (0.559)

Observations 1,228 1,228 1,228 1,228 1,228 1,228 956 1,887 1,887 1,887 1,887 1,887
R-squared 0.363 0.352
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

328
Table P.9. Heterogeneous effects of three years of English tutoring in high school on the CSAT English achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of English PT in high school 0.36*** 0.37 0.49*** 0.44** 0.19 0.06 0.29 0.49* 0.45*** 0.37 0.25 0.10
(0.100) (0.231) (0.121) (0.142) (0.116) (0.166) (0.164) (0.221) (0.130) (0.213) (0.181) (0.245)
URBAN -0.16 -0.25 -0.31* -0.09 0.04 0.20 -0.16 0.05 -0.16 -0.05 0.03 0.26
(0.133) (0.278) (0.137) (0.127) (0.120) (0.256) (0.146) (0.229) (0.148) (0.123) (0.128) (0.153)
GENDER (1=female) 0.10 0.17 0.19 0.14 0.04 0.00 0.09 0.29* 0.18 0.06 -0.02 -0.19
(0.088) (0.174) (0.122) (0.090) (0.138) (0.125) (0.100) (0.116) (0.114) (0.082) (0.107) (0.100)
SES 0.25*** 0.18 0.21** 0.21* 0.34** 0.38*** 0.23** 0.18 0.21*** 0.27*** 0.30*** 0.32***
(0.064) (0.154) (0.078) (0.092) (0.105) (0.097) (0.077) (0.094) (0.060) (0.065) (0.067) (0.045)
SCHOOL TYPE 0.09 0.23 0.05 -0.04 0.12 0.13 0.16 0.42** 0.20* 0.13 0.22** 0.22***
(0.092) (0.213) (0.096) (0.108) (0.116) (0.209) (0.103) (0.161) (0.089) (0.073) (0.073) (0.066)
SCHOOL TRACK 0.54*** 0.69** 0.59* 0.45* 0.60** 0.23 0.66*** 1.10*** 0.92*** 0.72*** 0.78*** 0.91***
(0.146) (0.239) (0.284) (0.202) (0.185) (0.283) (0.174) (0.244) (0.197) (0.182) (0.145) (0.258)
STU-TEA RATIO -0.04* -0.01 -0.01 -0.05* -0.06* -0.05 -0.04 -0.04 -0.03 -0.06 -0.05* -0.04
(0.019) (0.043) (0.030) (0.026) (0.028) (0.043) (0.022) (0.029) (0.029) (0.030) (0.022) (0.027)
SCHOOL CHOICE -0.20 -0.07 -0.22 -0.16 -0.07 0.02 -0.25 -0.19 -0.32* -0.22 -0.13 0.05
(0.135) (0.258) (0.158) (0.148) (0.169) (0.224) (0.149) (0.231) (0.162) (0.147) (0.142) (0.147)
PRESCORE (G9) 0.80*** 1.00*** 0.86*** 0.84*** 0.81*** 0.65*** 0.77*** 0.89*** 0.79*** 0.76*** 0.72*** 0.54***
(0.061) (0.143) (0.109) (0.083) (0.096) (0.129) (0.068) (0.111) (0.083) (0.046) (0.056) (0.071)
SELF-STUDY 0.18 0.20 0.09 0.14 0.18 0.30 0.03 0.18 0.26* 0.35*** 0.34*** 0.28**
(0.099) (0.281) (0.155) (0.117) (0.165) (0.193) (0.111) (0.177) (0.110) (0.093) (0.088) (0.108)
SELF-ESTEEM 0.51*** 0.48*** 0.48*** 0.48*** 0.41*** 0.54*** 0.54*** 0.52*** 0.50*** 0.44*** 0.37*** 0.45***
(0.074) (0.145) (0.081) (0.099) (0.110) (0.141) (0.083) (0.126) (0.086) (0.074) (0.077) (0.075)
English PT in middle school 0.03 0.12 0.09 0.00 0.00 -0.04 0.09 0.10 0.07 0.01 -0.06 -0.04
(0.048) (0.098) (0.094) (0.069) (0.089) (0.097) (0.057) (0.064) (0.057) (0.057) (0.079) (0.054)
Constant 5.32*** 2.36** 3.83*** 5.61*** 6.38*** 7.46*** 5.29*** 2.19*** 3.85*** 5.49*** 6.40*** 7.00***
(0.416) (0.877) (0.594) (0.579) (0.621) (0.942) (0.481) (0.462) (0.525) (0.553) (0.397) (0.602)

Observations 963 963 963 963 963 963 757 1,887 1,887 1,887 1,887 1,887
R-squared 0.399 0.377
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

329
Table P.10. Heterogeneous effects of one year of math tutoring in high school on the CSAT math achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

1 year of Math PT in high school 0.20 0.26 0.34 0.24 0.12 -0.01 0.56 0.75** 1.08*** 1.03*** 0.91*** 0.70***
(0.111) (0.230) (0.181) (0.141) (0.197) (0.168) (0.374) (0.276) (0.193) (0.208) (0.163) (0.168)
URBAN -0.05 -0.10 -0.11 0.18 -0.16 0.10 -0.15 -0.20 0.03 -0.03 -0.01 -0.16
(0.162) (0.344) (0.215) (0.199) (0.235) (0.383) (0.190) (0.140) (0.171) (0.124) (0.164) (0.211)
GENDER (1=female) -0.41*** -0.12 -0.16 -0.46*** -0.56*** -0.44* -0.50*** -0.29* -0.37*** -0.53*** -0.49*** -0.52***
(0.107) (0.198) (0.169) (0.131) (0.124) (0.173) (0.128) (0.125) (0.086) (0.091) (0.098) (0.125)
SES 0.00 -0.15 -0.06 0.07 0.01 0.10 -0.06 0.01 0.05 0.06 0.18* 0.21**
(0.079) (0.223) (0.152) (0.086) (0.115) (0.082) (0.099) (0.114) (0.071) (0.091) (0.084) (0.074)
SCHOOL TYPE 0.17 0.11 0.12 0.19 0.39* 0.18 0.19 0.16 0.27** 0.06 0.21* 0.28*
(0.110) (0.282) (0.244) (0.164) (0.174) (0.162) (0.128) (0.095) (0.085) (0.121) (0.105) (0.135)
SCHOOL TRACK 0.25 0.37* 0.21 0.57* 0.36 -0.01 0.28 0.71** 0.44 0.29 0.37* 0.40
(0.156) (0.177) (0.188) (0.244) (0.200) (0.202) (0.200) (0.227) (0.338) (0.246) (0.186) (0.290)
STU-TEA RATIO -0.01 -0.03 -0.01 -0.02 -0.02 0.01 -0.02 -0.05 -0.02 -0.04 -0.04 -0.03
(0.022) (0.048) (0.039) (0.020) (0.026) (0.025) (0.026) (0.028) (0.036) (0.023) (0.029) (0.023)
SCHOOL CHOICE 0.01 0.05 0.09 0.26 -0.07 -0.23 -0.11 -0.55** -0.13 -0.17 -0.16 -0.33
(0.162) (0.409) (0.155) (0.187) (0.197) (0.286) (0.192) (0.173) (0.128) (0.132) (0.160) (0.205)
PRESCORE (G9) 0.82*** 0.73*** 0.84*** 0.90*** 0.91*** 0.69*** 0.78*** 0.79*** 0.75*** 0.86*** 0.78*** 0.66***
(0.072) (0.103) (0.134) (0.105) (0.141) (0.099) (0.084) (0.098) (0.090) (0.054) (0.053) (0.075)
SELF-STUDY 0.72*** 0.76** 0.74*** 0.72*** 0.64** 0.86*** 0.68*** 0.36* 0.39*** 0.49*** 0.44*** 0.45***
(0.125) (0.239) (0.215) (0.098) (0.214) (0.132) (0.147) (0.170) (0.113) (0.091) (0.107) (0.113)
SELF-ESTEEM 0.29** 0.35* 0.31* 0.28 0.19 0.40** 0.37*** 0.41*** 0.25* 0.29** 0.22** 0.23**
(0.088) (0.137) (0.156) (0.145) (0.176) (0.138) (0.105) (0.072) (0.098) (0.097) (0.076) (0.086)
Math PT in middle school -0.08 0.06 -0.03 -0.13* -0.13 -0.17* -0.14* 0.06 -0.01 -0.09 -0.17** -0.08
(0.050) (0.099) (0.070) (0.051) (0.075) (0.077) (0.067) (0.077) (0.071) (0.054) (0.066) (0.062)
Constant 4.97*** 2.34 3.32*** 4.55*** 6.61*** 7.52*** 5.36*** 3.63*** 3.60*** 5.79*** 6.91*** 7.95***
(0.483) (1.220) (0.671) (0.657) (0.735) (0.719) (0.597) (0.563) (0.619) (0.528) (0.577) (0.648)

Observations 910 910 910 910 910 910 680 1,899 1,899 1,899 1,899 1,899
R-squared 0.259 0.247
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

330
Table P.11. Heterogeneous effects of two years of math tutoring in high school on the CSAT math achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

2 years of Math PT in high school 0.73*** 0.74*** 0.89*** 0.68** 0.60** 0.42* 1.01*** 0.81** 1.17*** 1.12*** 0.98*** 0.76**
(0.116) (0.211) (0.186) (0.220) (0.225) (0.202) (0.283) (0.288) (0.210) (0.254) (0.258) (0.260)
URBAN -0.21 -0.43* -0.29* -0.26 -0.16 -0.17 -0.24 -0.25* -0.03 -0.09 -0.06 -0.20
(0.140) (0.217) (0.137) (0.232) (0.189) (0.245) (0.158) (0.118) (0.141) (0.138) (0.199) (0.186)
GENDER (1=female) -0.36*** -0.16 -0.31* -0.44*** -0.39* -0.40** -0.31** -0.27* -0.33* -0.49*** -0.46*** -0.49***
(0.097) (0.228) (0.128) (0.133) (0.169) (0.139) (0.111) (0.130) (0.133) (0.095) (0.087) (0.148)
SES -0.00 -0.08 -0.09 -0.02 0.06 0.14 -0.02 -0.01 0.02 0.03 0.16 0.19*
(0.065) (0.087) (0.084) (0.074) (0.105) (0.093) (0.078) (0.094) (0.063) (0.066) (0.090) (0.076)
SCHOOL TYPE 0.11 0.14 0.17 0.03 0.16 0.13 0.11 0.17 0.29** 0.09 0.23 0.30**
(0.099) (0.249) (0.127) (0.145) (0.135) (0.140) (0.113) (0.146) (0.101) (0.122) (0.139) (0.114)
SCHOOL TRACK 0.44** 0.81* 0.38 0.67** 0.31 0.03 0.50* 0.67** 0.37 0.23 0.31 0.35
(0.163) (0.316) (0.200) (0.252) (0.223) (0.282) (0.206) (0.228) (0.334) (0.245) (0.247) (0.311)
STU-TEA RATIO -0.05* -0.04 -0.04 -0.04 -0.06 -0.04 -0.06* -0.05 -0.01 -0.04* -0.04* -0.03
(0.021) (0.035) (0.032) (0.030) (0.032) (0.032) (0.025) (0.027) (0.032) (0.018) (0.015) (0.028)
SCHOOL CHOICE -0.27 -0.59* -0.18 -0.27 -0.32 -0.42 -0.36* -0.54*** -0.12 -0.16 -0.15 -0.32*
(0.143) (0.245) (0.171) (0.269) (0.215) (0.251) (0.165) (0.145) (0.150) (0.183) (0.150) (0.162)
PRESCORE (G9) 0.78*** 0.70*** 0.84*** 0.87*** 0.78*** 0.65*** 0.70*** 0.77*** 0.72*** 0.83*** 0.75*** 0.63***
(0.066) (0.145) (0.104) (0.083) (0.106) (0.114) (0.075) (0.079) (0.092) (0.081) (0.066) (0.077)
SELF-STUDY 0.58*** 0.43 0.57*** 0.57*** 0.67*** 0.65** 0.50*** 0.35 0.38*** 0.48*** 0.43** 0.44**
(0.106) (0.222) (0.165) (0.126) (0.160) (0.239) (0.125) (0.185) (0.099) (0.113) (0.141) (0.170)
SELF-ESTEEM 0.25** 0.28 0.14 0.17 0.20 0.43* 0.27** 0.44*** 0.30** 0.34*** 0.26*** 0.26*
(0.078) (0.184) (0.154) (0.129) (0.114) (0.168) (0.091) (0.085) (0.112) (0.068) (0.070) (0.124)
Math PT in middle school -0.05 0.02 -0.09 -0.07 -0.01 -0.06 -0.06 0.04 -0.03 -0.11 -0.20** -0.10
(0.053) (0.076) (0.059) (0.073) (0.088) (0.135) (0.072) (0.079) (0.073) (0.070) (0.069) (0.081)
Constant 5.81*** 3.51*** 4.50*** 5.83*** 7.22*** 8.56*** 5.98*** 3.68*** 3.67*** 5.86*** 6.96*** 7.99***
(0.443) (0.672) (0.505) (0.775) (0.809) (0.620) (0.517) (0.549) (0.497) (0.511) (0.536) (0.738)

Observations 1,083 1,083 1,083 1,083 1,083 1,083 832 1,899 1,899 1,899 1,899 1,899
R-squared 0.324 0.308
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

331
Table P.12. Heterogeneous effects of three years of math tutoring in high school on the CSAT math achievement
OLS 2SLAD
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

3 years of Math PT in high school 1.18*** 1.32*** 1.35*** 1.21*** 0.98*** 0.93*** 1.08*** 0.91* 1.32*** 1.26*** 1.10*** 0.85**
(0.121) (0.233) (0.185) (0.176) (0.233) (0.245) (0.246) (0.363) (0.228) (0.269) (0.333) (0.327)
URBAN -0.03 -0.09 0.03 0.06 -0.13 0.15 -0.03 -0.20 0.04 -0.02 0.00 -0.15
(0.133) (0.165) (0.188) (0.190) (0.153) (0.242) (0.148) (0.181) (0.122) (0.109) (0.168) (0.185)
GENDER (1=female) -0.30*** -0.05 -0.17 -0.36** -0.45*** -0.44*** -0.36*** -0.26* -0.32** -0.49*** -0.45*** -0.49**
(0.090) (0.129) (0.175) (0.125) (0.129) (0.123) (0.102) (0.128) (0.104) (0.096) (0.096) (0.152)
SES 0.06 0.01 -0.11 0.04 0.18* 0.11 0.08 -0.02 -0.01 0.01 0.14 0.17
(0.061) (0.098) (0.104) (0.080) (0.076) (0.096) (0.075) (0.122) (0.062) (0.078) (0.102) (0.109)
SCHOOL TYPE 0.06 -0.24 0.04 0.13 0.17 0.37** 0.04 0.15 0.26** 0.05 0.20* 0.28**
(0.092) (0.172) (0.141) (0.118) (0.126) (0.137) (0.103) (0.143) (0.095) (0.091) (0.104) (0.106)
SCHOOL TRACK 0.39* 0.12 0.16 0.50 0.49 0.20 0.52** 0.69*** 0.40 0.25 0.33* 0.37
(0.157) (0.270) (0.227) (0.292) (0.329) (0.258) (0.193) (0.200) (0.296) (0.174) (0.167) (0.249)
STU-TEA RATIO -0.03 -0.02 0.01 -0.03 -0.07* -0.05 -0.02 -0.05 -0.01 -0.04 -0.04 -0.03
(0.021) (0.035) (0.041) (0.032) (0.026) (0.038) (0.024) (0.030) (0.023) (0.022) (0.021) (0.030)
SCHOOL CHOICE -0.10 -0.38 0.06 0.14 -0.28 -0.11 -0.20 -0.49* -0.05 -0.09 -0.09 -0.27
(0.141) (0.207) (0.192) (0.233) (0.203) (0.224) (0.160) (0.198) (0.146) (0.157) (0.201) (0.192)
PRESCORE (G9) 0.72*** 0.74*** 0.79*** 0.82*** 0.71*** 0.62*** 0.66*** 0.74*** 0.68*** 0.79*** 0.72*** 0.61***
(0.065) (0.168) (0.089) (0.086) (0.091) (0.087) (0.074) (0.087) (0.100) (0.073) (0.084) (0.085)
SELF-STUDY 0.46*** 0.31 0.45** 0.58*** 0.36** 0.27 0.33** 0.33 0.34** 0.45*** 0.40*** 0.42**
(0.100) (0.217) (0.170) (0.116) (0.138) (0.177) (0.114) (0.168) (0.130) (0.098) (0.097) (0.153)
SELF-ESTEEM 0.32*** 0.37** 0.42*** 0.29*** 0.27* 0.34** 0.37*** 0.44*** 0.29*** 0.33*** 0.25*** 0.25*
(0.075) (0.133) (0.115) (0.052) (0.109) (0.113) (0.083) (0.103) (0.087) (0.084) (0.072) (0.099)
Math PT in middle school -0.12* -0.02 -0.02 -0.20* -0.16 -0.16 -0.09 0.05 -0.02 -0.10 -0.19** -0.09
(0.055) (0.082) (0.117) (0.083) (0.101) (0.099) (0.070) (0.066) (0.059) (0.053) (0.070) (0.079)
Constant 5.40*** 3.43*** 3.08*** 5.06*** 7.63*** 8.13*** 5.42*** 3.65*** 3.62*** 5.82*** 6.93*** 7.96***
(0.440) (0.783) (0.881) (0.670) (0.692) (0.846) (0.500) (0.594) (0.585) (0.580) (0.602) (0.696)

Observations 1,081 1,081 1,081 1,081 1,081 1,081 854 1,899 1,899 1,899 1,899 1,899
R-squared 0.401 0.386
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

332
Table Q.1. Effects of private tutoring in grade 10 on the CSAT average achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

PT in G10 0.07 0.26* 0.22* 0.08 -0.05 -0.14 0.20 0.09 0.43 0.37 -0.18 0.01
(0.073) (0.115) (0.102) (0.094) (0.104) (0.151) (0.335) (0.579) (0.630) (0.289) (0.409) (0.476)
URBANICITY -0.06 -0.05 -0.05 -0.04 -0.11 0.04 -0.09 -0.14 -0.06 -0.07 -0.17 0.04
(0.076) (0.119) (0.094) (0.121) (0.145) (0.136) (0.082) (0.127) (0.098) (0.107) (0.134) (0.135)
GENDER (1=female) -0.07 0.09 0.00 -0.07 -0.18* -0.25** -0.10 0.07 -0.03 -0.15* -0.18 -0.31**
(0.052) (0.107) (0.076) (0.078) (0.080) (0.081) (0.060) (0.115) (0.069) (0.065) (0.106) (0.116)
SES 0.13*** 0.07 0.10* 0.16*** 0.19*** 0.22*** 0.13*** 0.05 0.09 0.15*** 0.21** 0.24***
(0.035) (0.053) (0.042) (0.032) (0.056) (0.057) (0.039) (0.049) (0.049) (0.035) (0.079) (0.067)
SCHOOL TYPE (1=private) 0.17** 0.12 0.15 0.22* 0.19* 0.11 0.17** 0.13 0.13 0.20* 0.18 0.15
(0.053) (0.089) (0.082) (0.087) (0.095) (0.066) (0.058) (0.125) (0.083) (0.087) (0.099) (0.107)
SCHOOL TRACK 0.56*** 0.91*** 0.73*** 0.53*** 0.53*** 0.49** 0.59*** 1.07*** 0.65* 0.53* 0.58*** 0.48*
(1=general) (0.094) (0.184) (0.158) (0.131) (0.132) (0.180) (0.121) (0.213) (0.262) (0.220) (0.168) (0.221)
STU-TEA RATIO -0.03** -0.03 -0.03 -0.04*** -0.03* -0.04 -0.04** -0.03 -0.02 -0.06*** -0.03 -0.04
(0.012) (0.026) (0.019) (0.013) (0.014) (0.022) (0.013) (0.022) (0.016) (0.015) (0.019) (0.025)
SCHOOL CHOICE -0.09 -0.10 -0.02 -0.03 -0.13 -0.16 -0.14 -0.21 -0.05 -0.14 -0.24 -0.16
(0.080) (0.125) (0.088) (0.125) (0.141) (0.163) (0.088) (0.117) (0.083) (0.095) (0.134) (0.131)
PRESCORE (G9VEM_Z) 0.74*** 0.84*** 0.78*** 0.78*** 0.71*** 0.65*** 0.72*** 0.85*** 0.73*** 0.75*** 0.69*** 0.64***
(0.036) (0.085) (0.051) (0.053) (0.065) (0.053) (0.039) (0.089) (0.067) (0.050) (0.062) (0.068)
STUDY HOURS 0.34*** 0.28*** 0.30*** 0.30*** 0.42*** 0.41*** 0.33*** 0.25*** 0.20* 0.32*** 0.42*** 0.37***
(1= >10hrs /week) (0.057) (0.081) (0.064) (0.060) (0.084) (0.093) (0.062) (0.073) (0.079) (0.085) (0.090) (0.075)
SELF ESTEEM 0.35*** 0.40*** 0.33*** 0.34*** 0.37*** 0.26* 0.34*** 0.37*** 0.35*** 0.34*** 0.36*** 0.29***
(0.044) (0.097) (0.048) (0.056) (0.079) (0.100) (0.048) (0.084) (0.076) (0.059) (0.065) (0.086)
TOTAL PT in middle school -0.01 0.05 0.03 -0.02 -0.04 -0.10 -0.02 0.06 0.02 -0.04 -0.03 -0.11
(0.033) (0.086) (0.046) (0.052) (0.053) (0.064) (0.044) (0.087) (0.083) (0.044) (0.068) (0.076)
PT in G11 0.25*** 0.30* 0.27** 0.19** 0.16 0.19 0.22* 0.31 0.27 0.11 0.21 0.08
(0.067) (0.116) (0.093) (0.072) (0.114) (0.125) (0.102) (0.194) (0.149) (0.113) (0.137) (0.190)
PT in G12 0.26*** 0.16 0.17* 0.28*** 0.37*** 0.32** 0.26*** 0.23* 0.19 0.28*** 0.35*** 0.35***
(0.059) (0.108) (0.077) (0.085) (0.100) (0.099) (0.067) (0.100) (0.099) (0.072) (0.091) (0.096)
Constant 5.14*** 2.85*** 3.87*** 5.26*** 6.29*** 7.50*** 5.28*** 3.04*** 3.72*** 5.63*** 6.49*** 7.61***
(0.257) (0.544) (0.380) (0.366) (0.286) (0.673) (0.293) (0.375) (0.390) (0.287) (0.425) (0.503)

Observations 1,940 1,940 1,940 1,940 1,940 1,940 1,680 1,701 1,701 1,701 1,701 1,701
R-squared 0.421 0.407
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

333
Table Q.2. Effects of private tutoring in grade 11 on the CSAT average achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

PT in G11 0.25*** 0.30* 0.27** 0.19* 0.16 0.19 0.36 2.86 1.51 1.69 0.00 1.03
(0.067) (0.120) (0.102) (0.078) (0.098) (0.137) (0.283) (2.018) (1.321) (0.906) (1.533) (2.368)
URBANICITY -0.06 -0.05 -0.05 -0.04 -0.11 0.04 -0.06 -0.17 -0.05 -0.00 -0.17 0.05
(0.076) (0.121) (0.051) (0.070) (0.110) (0.107) (0.078) (0.141) (0.098) (0.101) (0.142) (0.133)
GENDER (1=female) -0.07 0.09 0.00 -0.07 -0.18* -0.25*** -0.08 0.15 -0.01 -0.04 -0.16 -0.22*
(0.052) (0.055) (0.059) (0.047) (0.079) (0.069) (0.054) (0.109) (0.062) (0.063) (0.098) (0.105)
SES 0.13*** 0.07 0.10* 0.16*** 0.19** 0.22*** 0.12*** -0.05 0.07 0.12** 0.20* 0.19
(0.035) (0.053) (0.046) (0.039) (0.059) (0.031) (0.037) (0.096) (0.056) (0.042) (0.087) (0.103)
SCHOOL TYPE (1=private) 0.17** 0.12 0.15 0.22** 0.19 0.11 0.18*** 0.11 0.13 0.15** 0.18* 0.13
(0.053) (0.103) (0.077) (0.072) (0.104) (0.099) (0.055) (0.094) (0.076) (0.054) (0.075) (0.098)
SCHOOL TRACK (1=general) 0.56*** 0.91*** 0.73*** 0.53*** 0.53*** 0.49*** 0.56*** 0.75** 0.56* 0.41** 0.50*** 0.40
(0.094) (0.266) (0.170) (0.109) (0.120) (0.117) (0.099) (0.258) (0.246) (0.158) (0.149) (0.254)
STU-TEA RATIO -0.03** -0.03 -0.03 -0.04** -0.03* -0.04* -0.03** -0.03 -0.03 -0.05*** -0.03** -0.04
(0.012) (0.021) (0.018) (0.015) (0.017) (0.017) (0.012) (0.027) (0.023) (0.013) (0.013) (0.019)
SCHOOL CHOICE -0.09 -0.10 -0.02 -0.03 -0.13 -0.16 -0.10 -0.18 -0.05 -0.06 -0.21 -0.17
(0.080) (0.149) (0.085) (0.108) (0.115) (0.130) (0.082) (0.127) (0.121) (0.112) (0.157) (0.122)
PRESCORE (G9VEM_Z) 0.74*** 0.84*** 0.78*** 0.78*** 0.71*** 0.65*** 0.72*** 0.68*** 0.73*** 0.73*** 0.69*** 0.60***
(0.036) (0.093) (0.051) (0.040) (0.061) (0.041) (0.038) (0.093) (0.077) (0.060) (0.071) (0.074)
STUDY HOURS 0.34*** 0.28** 0.30*** 0.30*** 0.42*** 0.41*** 0.31*** 0.16 0.19** 0.25*** 0.40*** 0.32***
(1= >10hrs /week) (0.057) (0.085) (0.059) (0.053) (0.075) (0.090) (0.059) (0.111) (0.060) (0.056) (0.074) (0.094)
SELF ESTEEM 0.35*** 0.40*** 0.33*** 0.34*** 0.37*** 0.26** 0.37*** 0.40*** 0.35*** 0.34*** 0.39*** 0.30***
(0.044) (0.073) (0.054) (0.066) (0.059) (0.078) (0.045) (0.077) (0.052) (0.059) (0.070) (0.080)
TOTAL PT in middle school -0.01 0.05 0.03 -0.02 -0.04 -0.10 -0.03 -0.11 -0.05 -0.12 -0.02 -0.15
(0.033) (0.088) (0.044) (0.027) (0.053) (0.052) (0.041) (0.115) (0.074) (0.079) (0.114) (0.141)
PT in G10 0.07 0.26 0.22 0.08 -0.05 -0.14 0.04 -0.51 -0.06 -0.31 0.00 -0.29
(0.073) (0.156) (0.120) (0.094) (0.100) (0.159) (0.107) (0.630) (0.401) (0.240) (0.472) (0.653)
PT in G12 0.26*** 0.16* 0.17* 0.28*** 0.37** 0.32*** 0.23* -0.50 -0.18 -0.19 0.41 0.03
(0.059) (0.077) (0.074) (0.060) (0.112) (0.080) (0.095) (0.563) (0.407) (0.283) (0.471) (0.693)
Constant 5.14*** 2.85*** 3.87*** 5.26*** 6.29*** 7.50*** 5.16*** 2.80*** 3.79*** 5.26*** 6.45*** 7.32***
(0.257) (0.481) (0.333) (0.327) (0.263) (0.423) (0.267) (0.596) (0.528) (0.301) (0.439) (0.568)

Observations 1,940 1,940 1,940 1,940 1,940 1,940 1,827 1,883 1,883 1,883 1,883 1,883
R-squared 0.421 0.413
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

334
Table Q.3. Effects of private tutoring in grade 12 on the CSAT average achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

PT in G12 0.26*** 0.16 0.17 0.28** 0.37*** 0.32*** 0.16 0.10 0.17 0.10 0.07 0.09
(0.059) (0.084) (0.088) (0.086) (0.086) (0.074) (0.145) (0.199) (0.182) (0.243) (0.243) (0.227)
URBANICITY -0.06 -0.05 -0.05 -0.04 -0.11 0.04 -0.06 -0.02 -0.07 -0.02 -0.08 0.11
(0.076) (0.105) (0.076) (0.088) (0.086) (0.090) (0.080) (0.127) (0.095) (0.104) (0.158) (0.120)
GENDER (1=female) -0.07 0.09 0.00 -0.07 -0.18* -0.25*** -0.06 0.15 0.04 -0.06 -0.22* -0.21**
(0.052) (0.066) (0.080) (0.045) (0.085) (0.072) (0.055) (0.091) (0.072) (0.069) (0.095) (0.078)
SES 0.13*** 0.07 0.10** 0.16*** 0.19*** 0.22*** 0.13*** 0.07 0.12 0.16** 0.20* 0.25***
(0.035) (0.058) (0.035) (0.039) (0.049) (0.045) (0.037) (0.054) (0.075) (0.049) (0.084) (0.067)
SCHOOL TYPE (1=private) 0.17** 0.12 0.15 0.22*** 0.19 0.11 0.15** 0.15 0.17* 0.24*** 0.20* 0.14*
(0.053) (0.069) (0.077) (0.065) (0.098) (0.089) (0.056) (0.079) (0.077) (0.058) (0.080) (0.070)
SCHOOL TRACK (1=general) 0.56*** 0.91*** 0.73*** 0.53*** 0.53*** 0.49* 0.61*** 0.94*** 0.69*** 0.57*** 0.58*** 0.42
(0.094) (0.228) (0.196) (0.077) (0.125) (0.193) (0.104) (0.190) (0.200) (0.150) (0.139) (0.267)
STU-TEA RATIO -0.03** -0.03 -0.03* -0.04*** -0.03* -0.04** -0.03** -0.02 -0.03 -0.05** -0.04* -0.03*
(0.012) (0.020) (0.015) (0.013) (0.014) (0.013) (0.012) (0.017) (0.022) (0.017) (0.016) (0.016)
SCHOOL CHOICE -0.09 -0.10 -0.02 -0.03 -0.13 -0.16 -0.12 -0.09 -0.03 -0.06 -0.15 -0.13
(0.080) (0.175) (0.099) (0.101) (0.115) (0.092) (0.084) (0.172) (0.109) (0.106) (0.174) (0.149)
PRESCORE (G9VEM_Z) 0.74*** 0.84*** 0.78*** 0.78*** 0.71*** 0.65*** 0.73*** 0.83*** 0.76*** 0.77*** 0.70*** 0.68***
(0.036) (0.092) (0.046) (0.043) (0.063) (0.056) (0.037) (0.068) (0.049) (0.049) (0.068) (0.059)
STUDY HOURS 0.34*** 0.28*** 0.30*** 0.30*** 0.42*** 0.41*** 0.34*** 0.31** 0.25** 0.33*** 0.41*** 0.44***
(1= >10hrs /week) (0.057) (0.065) (0.067) (0.067) (0.088) (0.090) (0.060) (0.115) (0.091) (0.087) (0.088) (0.092)
SELF ESTEEM 0.35*** 0.40*** 0.33*** 0.34*** 0.37*** 0.26*** 0.38*** 0.50*** 0.34*** 0.38*** 0.39*** 0.27**
(0.044) (0.085) (0.059) (0.057) (0.054) (0.072) (0.046) (0.063) (0.055) (0.052) (0.091) (0.087)
TOTAL PT in middle school -0.01 0.05 0.03 -0.02 -0.04 -0.10 -0.01 0.06 0.03 0.01 -0.05 -0.08
(0.033) (0.074) (0.046) (0.035) (0.046) (0.067) (0.036) (0.078) (0.059) (0.038) (0.051) (0.056)
PT in G10 0.07 0.26 0.22** 0.08 -0.05 -0.14 0.06 0.27* 0.26 0.08 -0.02 -0.22
(0.073) (0.147) (0.072) (0.105) (0.092) (0.139) (0.078) (0.134) (0.144) (0.097) (0.109) (0.185)
PT in G11 0.25*** 0.30* 0.27* 0.19** 0.16 0.19* 0.31*** 0.32* 0.27 0.30* 0.32* 0.26*
(0.067) (0.136) (0.121) (0.075) (0.090) (0.089) (0.086) (0.129) (0.157) (0.127) (0.163) (0.130)
Constant 5.14*** 2.85*** 3.87*** 5.26*** 6.29*** 7.50*** 5.16*** 2.61*** 3.72*** 5.20*** 6.39*** 7.35***
(0.257) (0.488) (0.333) (0.351) (0.327) (0.394) (0.272) (0.581) (0.526) (0.381) (0.372) (0.516)

Observations 1,940 1,940 1,940 1,940 1,940 1,940 1,752 1,752 1,752 1,752 1,752 1,752
R-squared 0.421 0.422
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_avg (average score on the CSAT)

335
Table. Q.4. Effects of verbal tutoring in grade 10 on the CSAT verbal achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G10 0.00 0.07 0.00 -0.02 0.00 -0.16 0.11 0.15 -0.21 0.05 -0.08 -0.10
(0.064) (0.124) (0.084) (0.081) (0.064) (0.120) (0.291) (0.958) (0.585) (0.760) (0.311) (0.574)
URBANICITY -0.04 0.08 -0.01 -0.14 -0.04 -0.11 -0.05 -0.06 0.07 -0.08 -0.03 -0.08
(0.090) (0.102) (0.125) (0.122) (0.125) (0.158) (0.102) (0.157) (0.110) (0.148) (0.122) (0.187)
GENDER (1=female) 0.10 0.17 0.18* 0.11 -0.00 -0.03 0.04 0.26 0.17 -0.00 -0.15 -0.12
(0.062) (0.107) (0.082) (0.075) (0.076) (0.094) (0.069) (0.138) (0.109) (0.108) (0.122) (0.126)
SES 0.08* 0.01 0.12* 0.10* 0.09* 0.10* 0.08 0.02 0.13 0.12 0.09 0.15
(0.040) (0.138) (0.059) (0.042) (0.044) (0.049) (0.047) (0.109) (0.088) (0.073) (0.074) (0.079)
SCHOOL TYPE (1=private) 0.09 0.02 0.13 0.12 0.06 0.03 0.07 0.03 0.17 0.06 0.07 0.01
(0.063) (0.117) (0.077) (0.080) (0.095) (0.110) (0.068) (0.104) (0.091) (0.107) (0.102) (0.150)
SCHOOL TRACK (1=general) 0.66*** 0.70*** 0.98*** 0.86*** 0.51*** 0.40 0.72*** 0.83** 1.04*** 0.98*** 0.55*** 0.48*
(0.110) (0.091) (0.147) (0.163) (0.094) (0.204) (0.126) (0.281) (0.218) (0.262) (0.138) (0.214)
STU-TEA RATIO -0.03* -0.04 -0.03 -0.05* -0.03 -0.02 -0.03 -0.04 -0.02 -0.05* -0.02 -0.02
(0.014) (0.036) (0.017) (0.020) (0.018) (0.019) (0.015) (0.037) (0.021) (0.024) (0.019) (0.019)
SCHOOL CHOICE 0.02 0.18 0.04 -0.02 0.10 -0.08 0.01 0.05 0.09 0.00 0.09 -0.14
(0.094) (0.154) (0.129) (0.128) (0.118) (0.166) (0.102) (0.189) (0.135) (0.163) (0.097) (0.177)
PRESCORE (G9VEM_Z) 0.72*** 0.87*** 0.67*** 0.75*** 0.71*** 0.76*** 0.72*** 0.78*** 0.64*** 0.73*** 0.73*** 0.71***
(0.041) (0.095) (0.063) (0.049) (0.053) (0.074) (0.045) (0.082) (0.071) (0.047) (0.064) (0.101)
STUDY HOURS 0.17* 0.15 0.09 0.10 0.20* 0.30* 0.18* 0.09 0.12 0.10 0.18 0.21*
(1= >10hrs /week) (0.068) (0.111) (0.081) (0.100) (0.092) (0.120) (0.073) (0.110) (0.071) (0.107) (0.123) (0.096)
SELF ESTEEM 0.34*** 0.35*** 0.42*** 0.35*** 0.28** 0.24** 0.32*** 0.37** 0.44*** 0.37*** 0.29** 0.25*
(0.051) (0.091) (0.050) (0.063) (0.087) (0.082) (0.057) (0.140) (0.089) (0.074) (0.103) (0.118)
VERBAL PT in middle school -0.08** -0.04 -0.09* -0.05 -0.13*** -0.11** -0.11** -0.07 -0.07 -0.09 -0.14** -0.16**
(0.028) (0.048) (0.035) (0.029) (0.031) (0.038) (0.038) (0.086) (0.041) (0.070) (0.047) (0.053)
VERBAL PT in G11 0.13 0.25 0.04 0.10 0.14 0.20* 0.11 0.18 0.12 0.07 0.14 0.22
(0.075) (0.131) (0.092) (0.091) (0.099) (0.102) (0.102) (0.347) (0.208) (0.213) (0.195) (0.214)
VERBAL PT in G12 0.21* 0.03 0.06 0.26** 0.31** 0.38* 0.23* 0.18 0.07 0.25 0.34** 0.40***
(0.094) (0.151) (0.106) (0.092) (0.105) (0.161) (0.102) (0.191) (0.123) (0.137) (0.116) (0.107)
Constant 5.27*** 3.15*** 3.99*** 5.47*** 6.47*** 7.52*** 5.31*** 3.20*** 3.77*** 5.56*** 6.51*** 7.87***
(0.296) (0.706) (0.409) (0.467) (0.393) (0.428) (0.329) (0.692) (0.479) (0.500) (0.317) (0.331)

Observations 2,070 2,070 2,070 2,070 2,070 2,070 1,790 1,807 1,807 1,807 1,807 1,807
R-squared 0.269 0.261
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

336
Table Q.5. Effects of verbal tutoring in grade 11 on the CSAT verbal achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G11 0.13 0.25* 0.04 0.10 0.14 0.20 0.05 1.45 -0.67 1.49 2.64 -4.08
(0.075) (0.100) (0.095) (0.074) (0.122) (0.171) (0.279) (3.226) (3.329) (3.780) (2.371) (3.274)
URBANICITY -0.04 0.08 -0.01 -0.14 -0.04 -0.11 -0.02 -0.11 0.03 -0.16 -0.12 0.09
(0.090) (0.138) (0.120) (0.093) (0.167) (0.203) (0.094) (0.208) (0.154) (0.200) (0.211) (0.238)
GENDER (1=female) 0.10 0.17 0.18* 0.11 -0.00 -0.03 0.09 0.32 0.12 0.17 0.10 -0.25
(0.062) (0.110) (0.082) (0.083) (0.080) (0.081) (0.064) (0.198) (0.199) (0.202) (0.159) (0.169)
SES 0.08* 0.01 0.12* 0.10 0.09* 0.10 0.08 0.08 0.15 0.04 -0.01 0.27
(0.040) (0.105) (0.057) (0.053) (0.044) (0.080) (0.043) (0.187) (0.178) (0.137) (0.085) (0.140)
SCHOOL TYPE (1=private) 0.09 0.02 0.13 0.12 0.06 0.03 0.11 0.05 0.13 0.09 0.04 0.03
(0.063) (0.103) (0.073) (0.085) (0.080) (0.114) (0.065) (0.129) (0.115) (0.071) (0.076) (0.100)
SCHOOL TRACK (1=general) 0.66*** 0.70*** 0.98*** 0.86*** 0.51*** 0.40 0.68*** 0.75*** 0.98*** 0.83*** 0.48** 0.50*
(0.110) (0.210) (0.197) (0.164) (0.105) (0.204) (0.114) (0.203) (0.172) (0.190) (0.158) (0.205)
STU-TEA RATIO -0.03* -0.04 -0.03 -0.05** -0.03 -0.02 -0.03 -0.04 -0.04 -0.03 0.00 -0.05
(0.014) (0.025) (0.017) (0.017) (0.024) (0.022) (0.014) (0.036) (0.033) (0.038) (0.026) (0.031)
SCHOOL CHOICE 0.02 0.18 0.04 -0.02 0.10 -0.08 0.03 0.05 -0.07 0.00 0.19 -0.26
(0.094) (0.152) (0.145) (0.119) (0.169) (0.208) (0.098) (0.211) (0.187) (0.212) (0.260) (0.314)
PRESCORE (G9VEM_Z) 0.72*** 0.87*** 0.67*** 0.75*** 0.71*** 0.76*** 0.71*** 0.76*** 0.68*** 0.75*** 0.70*** 0.76***
(0.041) (0.090) (0.052) (0.051) (0.052) (0.077) (0.043) (0.101) (0.055) (0.050) (0.053) (0.078)
STUDY HOURS 0.17* 0.15 0.09 0.10 0.20* 0.30* 0.16* 0.05 0.04 0.02 0.08 0.41*
(1= >10hrs /week) (0.068) (0.127) (0.085) (0.076) (0.085) (0.123) (0.069) (0.167) (0.151) (0.160) (0.137) (0.199)
SELF ESTEEM 0.34*** 0.35*** 0.42*** 0.35*** 0.28*** 0.24** 0.35*** 0.41** 0.44*** 0.36*** 0.30*** 0.25*
(0.051) (0.097) (0.066) (0.063) (0.047) (0.092) (0.053) (0.132) (0.058) (0.059) (0.060) (0.108)
VERBAL PT in middle school -0.08** -0.04 -0.09*** -0.05 -0.13** -0.11*** -0.07* -0.05 -0.07 -0.09 -0.15* -0.06
(0.028) (0.036) (0.025) (0.027) (0.038) (0.024) (0.029) (0.078) (0.063) (0.085) (0.067) (0.092)
VERBAL PT in G10 0.00 0.07 0.00 -0.02 0.00 -0.16 0.00 -0.19 0.13 -0.23 -0.42 0.60
(0.064) (0.105) (0.069) (0.068) (0.089) (0.121) (0.080) (0.565) (0.544) (0.618) (0.406) (0.583)
VERBAL PT in G12 0.21* 0.03 0.06 0.26* 0.31** 0.38 0.23 -0.38 0.32 -0.26 -0.57 1.84
(0.094) (0.136) (0.065) (0.115) (0.110) (0.203) (0.130) (1.158) (1.138) (1.374) (0.871) (1.088)
Constant 5.27*** 3.15*** 3.99*** 5.47*** 6.47*** 7.52*** 5.25*** 3.16** 4.46*** 5.00*** 5.52*** 8.92***
(0.296) (0.555) (0.551) (0.396) (0.446) (0.522) (0.313) (1.169) (1.239) (1.270) (0.897) (1.200)

Observations 2,070 2,070 2,070 2,070 2,070 2,070 1,952 2,010 2,010 2,010 2,010 2,010
R-squared 0.269 0.259
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

337
Table Q.6. Effects of verbal tutoring in grade 12 on the CSAT verbal achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

VERBAL PT in G12 0.21* 0.03 0.06 0.26** 0.31** 0.38* 0.09 0.14 -0.22 0.21 0.23 -0.38
(0.094) (0.207) (0.126) (0.087) (0.106) (0.178) (0.260) (0.214) (0.280) (0.450) (0.403) (0.357)
URBANICITY -0.04 0.08 -0.01 -0.14 -0.04 -0.11 -0.02 0.01 0.04 -0.10 0.04 -0.03
(0.090) (0.183) (0.125) (0.133) (0.134) (0.251) (0.094) (0.192) (0.109) (0.083) (0.117) (0.143)
GENDER (1=female) 0.10 0.17 0.18 0.11 -0.00 -0.03 0.09 0.22 0.19** 0.07 -0.05 -0.00
(0.062) (0.120) (0.114) (0.083) (0.104) (0.092) (0.065) (0.146) (0.070) (0.071) (0.107) (0.160)
SES 0.08* 0.01 0.12 0.10 0.09 0.10 0.07 0.01 0.12 0.08 0.10 0.10
(0.040) (0.086) (0.064) (0.061) (0.052) (0.075) (0.043) (0.122) (0.086) (0.074) (0.053) (0.078)
SCHOOL TYPE (1=private) 0.09 0.02 0.13 0.12 0.06 0.03 0.08 -0.01 0.13 0.15* 0.11 -0.01
(0.063) (0.091) (0.084) (0.109) (0.085) (0.148) (0.066) (0.108) (0.069) (0.063) (0.085) (0.113)
SCHOOL TRACK (1=general) 0.66*** 0.70*** 0.98*** 0.86*** 0.51** 0.40 0.71*** 0.76*** 0.96*** 0.87*** 0.65*** 0.54*
(0.110) (0.128) (0.133) (0.115) (0.165) (0.265) (0.121) (0.225) (0.193) (0.239) (0.190) (0.233)
STU-TEA RATIO -0.03* -0.04 -0.03 -0.05** -0.03 -0.02 -0.03* -0.04 -0.02 -0.05** -0.04 -0.04*
(0.014) (0.023) (0.021) (0.015) (0.024) (0.020) (0.014) (0.032) (0.024) (0.017) (0.024) (0.020)
SCHOOL CHOICE 0.02 0.18 0.04 -0.02 0.10 -0.08 0.02 0.10 0.05 0.03 0.12 -0.12
(0.094) (0.160) (0.129) (0.128) (0.144) (0.283) (0.099) (0.190) (0.146) (0.089) (0.113) (0.140)
PRESCORE (G9VEM_Z) 0.72*** 0.87*** 0.67*** 0.75*** 0.71*** 0.76*** 0.72*** 0.77*** 0.67*** 0.77*** 0.72*** 0.77***
(0.041) (0.066) (0.054) (0.036) (0.052) (0.070) (0.044) (0.067) (0.054) (0.050) (0.070) (0.077)
STUDY HOURS 0.17* 0.15 0.09 0.10 0.20 0.30** 0.18* 0.09 0.13 0.07 0.24 0.35**
(1= >10hrs /week) (0.068) (0.101) (0.102) (0.065) (0.107) (0.097) (0.072) (0.157) (0.092) (0.089) (0.150) (0.113)
SELF ESTEEM 0.34*** 0.35** 0.42*** 0.35*** 0.28*** 0.24** 0.38*** 0.45** 0.48*** 0.40*** 0.30** 0.26**
(0.051) (0.134) (0.059) (0.075) (0.071) (0.088) (0.054) (0.138) (0.081) (0.114) (0.104) (0.084)
VERBAL PT in middle school -0.08** -0.04 -0.09* -0.05 -0.13** -0.11** -0.08** -0.06 -0.07 -0.06 -0.12** -0.09*
(0.028) (0.035) (0.039) (0.033) (0.039) (0.042) (0.030) (0.044) (0.041) (0.039) (0.043) (0.047)
VERBAL PT in G10 0.00 0.07 0.00 -0.02 0.00 -0.16 -0.00 0.07 -0.02 -0.01 -0.04 -0.14
(0.064) (0.110) (0.081) (0.058) (0.072) (0.116) (0.068) (0.142) (0.112) (0.100) (0.101) (0.109)
VERBAL PT in G11 0.13 0.25 0.04 0.10 0.14 0.20* 0.15 0.19 0.12 0.14 0.14 0.37
(0.075) (0.128) (0.114) (0.086) (0.073) (0.086) (0.094) (0.151) (0.144) (0.155) (0.160) (0.190)
Constant 5.27*** 3.15*** 3.99*** 5.47*** 6.47*** 7.52*** 5.31*** 3.24*** 3.89*** 5.48*** 6.56*** 7.79***
(0.296) (0.524) (0.418) (0.300) (0.437) (0.718) (0.311) (0.784) (0.501) (0.420) (0.430) (0.622)

Observations 2,070 2,070 2,070 2,070 2,070 2,070 1,870 1,870 1,870 1,870 1,870 1,870
R-squared 0.269 0.272
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_V (verbal score on the CSAT)

338
Table Q.7. Effects of English tutoring in grade 10 on the CSAT English achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G10 0.03 0.16 0.14 0.03 -0.00 0.07 0.26 0.72 0.62 0.37 -0.08 0.45
(0.067) (0.129) (0.105) (0.097) (0.080) (0.143) (0.267) (0.638) (0.367) (0.420) (0.358) (0.366)
URBANICITY -0.02 0.01 -0.17 -0.11 0.07 0.14 -0.05 -0.03 -0.17 -0.12 0.05 0.21
(0.083) (0.175) (0.100) (0.093) (0.112) (0.126) (0.091) (0.139) (0.095) (0.114) (0.091) (0.158)
GENDER (1=female) 0.07 0.17 0.19** 0.06 0.00 -0.10 0.05 0.16 0.17** 0.05 -0.04 -0.15
(0.058) (0.131) (0.067) (0.078) (0.076) (0.091) (0.063) (0.130) (0.067) (0.072) (0.072) (0.105)
SES 0.24*** 0.14 0.18*** 0.25*** 0.31*** 0.35*** 0.25*** 0.15 0.21*** 0.25*** 0.32*** 0.31***
(0.039) (0.075) (0.042) (0.046) (0.065) (0.051) (0.044) (0.115) (0.050) (0.047) (0.036) (0.082)
SCHOOL TYPE (1=private) 0.14* 0.25 0.12 0.10 0.18* 0.18 0.19** 0.33** 0.14 0.10 0.22* 0.24**
(0.059) (0.151) (0.078) (0.080) (0.072) (0.095) (0.064) (0.117) (0.083) (0.118) (0.102) (0.088)
SCHOOL TRACK (1=general) 0.63*** 0.77*** 0.75*** 0.62*** 0.72*** 0.68*** 0.67*** 0.73*** 0.82*** 0.58*** 0.75*** 0.85***
(0.106) (0.222) (0.181) (0.137) (0.163) (0.206) (0.122) (0.179) (0.174) (0.160) (0.179) (0.217)
STU-TEA RATIO -0.03* -0.02 -0.02 -0.05** -0.03* -0.05* -0.04** -0.04 -0.02 -0.05* -0.04 -0.05**
(0.013) (0.016) (0.020) (0.018) (0.015) (0.022) (0.015) (0.026) (0.021) (0.023) (0.019) (0.017)
SCHOOL CHOICE -0.15 -0.07 -0.22* -0.24** -0.07 -0.06 -0.18 -0.11 -0.23 -0.27* -0.13 -0.04
(0.088) (0.198) (0.100) (0.086) (0.107) (0.139) (0.095) (0.125) (0.140) (0.123) (0.114) (0.162)
PRESCORE (G9VEM_Z) 0.73*** 0.96*** 0.80*** 0.74*** 0.69*** 0.60*** 0.72*** 0.95*** 0.78*** 0.72*** 0.70*** 0.55***
(0.039) (0.079) (0.066) (0.048) (0.057) (0.082) (0.042) (0.106) (0.083) (0.062) (0.070) (0.065)
STUDY HOURS (1= >10hrs /week) 0.38*** 0.40*** 0.34*** 0.42*** 0.33*** 0.26 0.36*** 0.31*** 0.30*** 0.39*** 0.35*** 0.21
(0.063) (0.103) (0.078) (0.063) (0.074) (0.145) (0.068) (0.093) (0.069) (0.090) (0.095) (0.150)
SELF ESTEEM 0.43*** 0.47*** 0.48*** 0.42*** 0.39*** 0.49*** 0.42*** 0.37*** 0.43*** 0.42*** 0.37*** 0.44***
(0.048) (0.081) (0.047) (0.054) (0.076) (0.079) (0.053) (0.110) (0.053) (0.064) (0.088) (0.075)
ENGLISH PT in middle school 0.02 0.08 0.03 0.01 -0.01 -0.05 0.01 0.03 0.04 -0.03 -0.02 -0.06
(0.034) (0.089) (0.047) (0.052) (0.046) (0.048) (0.043) (0.093) (0.070) (0.092) (0.063) (0.060)
ENGLISH PT in G11 0.18** 0.30 0.19** 0.14 0.10 0.02 0.13 0.28 0.04 0.06 0.11 -0.07
(0.065) (0.169) (0.059) (0.078) (0.060) (0.102) (0.093) (0.191) (0.111) (0.117) (0.111) (0.173)
ENGLISH PT in G12 0.19** 0.09 0.25* 0.29** 0.09 0.03 0.17* 0.11 0.21* 0.21** 0.09 -0.00
(0.068) (0.114) (0.110) (0.094) (0.085) (0.135) (0.075) (0.177) (0.096) (0.073) (0.122) (0.118)
Constant 4.95*** 2.24*** 3.68*** 5.43*** 5.91*** 7.29*** 4.99*** 2.45*** 3.37*** 5.54*** 6.12*** 7.04***
(0.284) (0.595) (0.457) (0.344) (0.439) (0.436) (0.314) (0.566) (0.516) (0.505) (0.316) (0.405)

Observations 2,344 2,344 2,344 2,344 2,344 2,344 2,032 2,049 2,049 2,049 2,049 2,049
R-squared 0.348 0.336
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

339
Table Q.8. Effects of English tutoring in grade 11 on the CSAT English achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G11 0.18** 0.30 0.19* 0.14 0.10 0.02 0.33 2.45 1.11 0.57 0.13 0.25
(0.065) (0.173) (0.093) (0.124) (0.089) (0.108) (0.212) (1.566) (0.588) (0.947) (0.691) (1.346)
URBANICITY -0.02 0.01 -0.17 -0.11 0.07 0.14 -0.02 0.11 -0.17 -0.11 0.04 0.22**
(0.083) (0.213) (0.094) (0.128) (0.147) (0.197) (0.086) (0.196) (0.158) (0.110) (0.090) (0.083)
GENDER (1=female) 0.07 0.17 0.19*** 0.06 0.00 -0.10 0.07 0.22 0.22* 0.07 0.05 -0.15
(0.058) (0.112) (0.047) (0.070) (0.086) (0.079) (0.060) (0.131) (0.089) (0.083) (0.094) (0.086)
SES 0.24*** 0.14 0.18*** 0.25*** 0.31*** 0.35*** 0.21*** 0.05 0.14* 0.23*** 0.32*** 0.31*
(0.039) (0.084) (0.048) (0.053) (0.060) (0.073) (0.041) (0.133) (0.064) (0.065) (0.078) (0.122)
SCHOOL TYPE (1=private) 0.14* 0.25* 0.12 0.10 0.18* 0.18 0.16** 0.27 0.18* 0.10 0.18 0.17
(0.059) (0.106) (0.071) (0.082) (0.090) (0.109) (0.061) (0.146) (0.080) (0.088) (0.103) (0.107)
SCHOOL TRACK (1=general) 0.63*** 0.77*** 0.75*** 0.62*** 0.72*** 0.68*** 0.64*** 0.87*** 0.73*** 0.58*** 0.67*** 0.69**
(0.106) (0.202) (0.191) (0.150) (0.124) (0.166) (0.110) (0.208) (0.155) (0.144) (0.161) (0.231)
STU-TEA RATIO -0.03* -0.02 -0.02 -0.05** -0.03 -0.05* -0.03* -0.03 -0.02 -0.05* -0.04** -0.05
(0.013) (0.020) (0.015) (0.018) (0.020) (0.024) (0.014) (0.022) (0.017) (0.020) (0.015) (0.028)
SCHOOL CHOICE -0.15 -0.07 -0.22 -0.24* -0.07 -0.06 -0.14 -0.00 -0.20 -0.22* -0.06 -0.04
(0.088) (0.211) (0.134) (0.116) (0.102) (0.149) (0.091) (0.201) (0.137) (0.113) (0.092) (0.125)
PRESCORE (G9VEM_Z) 0.73*** 0.96*** 0.80*** 0.74*** 0.69*** 0.60*** 0.71*** 0.94*** 0.79*** 0.76*** 0.67*** 0.59***
(0.039) (0.086) (0.056) (0.059) (0.053) (0.068) (0.041) (0.072) (0.059) (0.053) (0.067) (0.070)
STUDY HOURS (1= >10hrs /week) 0.38*** 0.40*** 0.34*** 0.42*** 0.33*** 0.26* 0.34*** 0.20 0.27*** 0.35*** 0.32*** 0.19
(0.063) (0.121) (0.094) (0.079) (0.097) (0.108) (0.065) (0.140) (0.065) (0.087) (0.067) (0.109)
SELF ESTEEM 0.43*** 0.47*** 0.48*** 0.42*** 0.39*** 0.49*** 0.45*** 0.47*** 0.47*** 0.45*** 0.42*** 0.47***
(0.048) (0.096) (0.065) (0.065) (0.057) (0.075) (0.050) (0.099) (0.059) (0.050) (0.061) (0.085)
ENGLISH PT in middle school 0.02 0.08 0.03 0.01 -0.01 -0.05 0.01 -0.01 0.01 0.02 -0.00 -0.05
(0.034) (0.066) (0.061) (0.054) (0.046) (0.057) (0.037) (0.103) (0.060) (0.072) (0.042) (0.074)
ENGLISH PT in G10 0.03 0.16 0.14 0.03 -0.00 0.07 0.01 -0.36 -0.08 -0.06 0.01 0.09
(0.067) (0.175) (0.100) (0.099) (0.091) (0.168) (0.086) (0.426) (0.188) (0.228) (0.184) (0.344)
ENGLISH PT in G12 0.19** 0.09 0.25** 0.29** 0.09 0.03 0.14 -0.61 -0.04 0.12 0.07 -0.03
(0.068) (0.147) (0.080) (0.110) (0.081) (0.088) (0.094) (0.490) (0.204) (0.323) (0.231) (0.442)
Constant 4.95*** 2.24*** 3.68*** 5.43*** 5.91*** 7.29*** 4.91*** 1.87*** 3.40*** 5.40*** 5.98*** 7.28***
(0.284) (0.653) (0.520) (0.316) (0.388) (0.517) (0.296) (0.453) (0.351) (0.418) (0.465) (0.696)

Observations 2,344 2,344 2,344 2,344 2,344 2,344 2,212 2,273 2,273 2,273 2,273 2,273
R-squared 0.348 0.335
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

340
Table Q.9. Effects of English tutoring in grade 12 on the CSAT English achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

ENGLISH PT in G12 0.19** 0.09 0.25* 0.29*** 0.09 0.03 0.14 0.39* 0.18 -0.04 0.14 -0.24
(0.068) (0.120) (0.114) (0.087) (0.068) (0.085) (0.144) (0.178) (0.241) (0.193) (0.208) (0.243)
URBANICITY -0.02 0.01 -0.17 -0.11 0.07 0.14 -0.03 -0.05 -0.17 -0.08 0.05 0.22
(0.083) (0.200) (0.121) (0.109) (0.088) (0.134) (0.088) (0.175) (0.125) (0.084) (0.119) (0.122)
GENDER (1=female) 0.07 0.17 0.19* 0.06 0.00 -0.10 0.07 0.20* 0.16* 0.08 -0.03 -0.08
(0.058) (0.153) (0.081) (0.067) (0.093) (0.089) (0.061) (0.088) (0.076) (0.067) (0.056) (0.108)
SES 0.24*** 0.14 0.18*** 0.25*** 0.31*** 0.35*** 0.23*** 0.15 0.17** 0.29*** 0.31*** 0.33***
(0.039) (0.107) (0.047) (0.055) (0.054) (0.067) (0.042) (0.092) (0.054) (0.078) (0.043) (0.072)
SCHOOL TYPE (1=private) 0.14* 0.25** 0.12 0.10 0.18* 0.18* 0.14* 0.26* 0.12* 0.07 0.17* 0.19*
(0.059) (0.096) (0.089) (0.119) (0.076) (0.090) (0.062) (0.127) (0.061) (0.087) (0.066) (0.091)
SCHOOL TRACK (1=general) 0.63*** 0.77*** 0.75*** 0.62*** 0.72*** 0.68*** 0.69*** 1.07*** 0.80*** 0.60*** 0.80*** 0.87**
(0.106) (0.197) (0.165) (0.116) (0.135) (0.191) (0.116) (0.159) (0.186) (0.131) (0.117) (0.283)
STU-TEA RATIO -0.03* -0.02 -0.02 -0.05** -0.03 -0.05 -0.03* -0.02 -0.02 -0.04 -0.04 -0.05
(0.013) (0.020) (0.016) (0.015) (0.018) (0.025) (0.014) (0.016) (0.023) (0.023) (0.022) (0.026)
SCHOOL CHOICE -0.15 -0.07 -0.22 -0.24* -0.07 -0.06 -0.19* -0.23 -0.26 -0.27* -0.10 -0.02
(0.088) (0.205) (0.139) (0.109) (0.097) (0.126) (0.093) (0.209) (0.140) (0.114) (0.111) (0.107)
PRESCORE (G9VEM_Z) 0.73*** 0.96*** 0.80*** 0.74*** 0.69*** 0.60*** 0.72*** 0.89*** 0.82*** 0.73*** 0.71*** 0.57***
(0.039) (0.087) (0.054) (0.058) (0.059) (0.076) (0.041) (0.100) (0.074) (0.050) (0.049) (0.062)
STUDY HOURS (1= >10hrs /week) 0.38*** 0.40** 0.34*** 0.42*** 0.33** 0.26 0.38*** 0.34** 0.35*** 0.45*** 0.35*** 0.26*
(0.063) (0.153) (0.100) (0.095) (0.104) (0.133) (0.066) (0.129) (0.089) (0.075) (0.078) (0.112)
SELF ESTEEM 0.43*** 0.47*** 0.48*** 0.42*** 0.39*** 0.49*** 0.45*** 0.53*** 0.50*** 0.44*** 0.36*** 0.42***
(0.048) (0.125) (0.072) (0.062) (0.068) (0.088) (0.051) (0.106) (0.069) (0.077) (0.074) (0.118)
ENGLISH PT in middle school 0.02 0.08 0.03 0.01 -0.01 -0.05 0.02 0.05 0.03 0.02 -0.04 -0.02
(0.034) (0.081) (0.049) (0.062) (0.056) (0.055) (0.037) (0.071) (0.061) (0.057) (0.057) (0.063)
ENGLISH PT in G10 0.03 0.16 0.14* 0.03 -0.00 0.07 0.03 0.12 0.18 -0.05 -0.01 0.06
(0.067) (0.162) (0.062) (0.115) (0.080) (0.124) (0.071) (0.124) (0.111) (0.084) (0.072) (0.103)
ENGLISH PT in G11 0.18** 0.30* 0.19* 0.14 0.10 0.02 0.20* 0.14 0.22* 0.29** 0.08 0.09
(0.065) (0.150) (0.083) (0.093) (0.064) (0.112) (0.077) (0.164) (0.093) (0.113) (0.105) (0.131)
Constant 4.95*** 2.24** 3.68*** 5.43*** 5.91*** 7.29*** 4.96*** 2.27*** 3.68*** 5.32*** 6.16*** 7.07***
(0.284) (0.696) (0.347) (0.311) (0.465) (0.603) (0.300) (0.624) (0.521) (0.553) (0.414) (0.551)

Observations 2,344 2,344 2,344 2,344 2,344 2,344 2,114 2,114 2,114 2,114 2,114 2,114
R-squared 0.348 0.349
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_E (English score on the CSAT)

341
Table Q.10. Effects of math tutoring in grade 10 on the CSAT math achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G10 0.14 0.29 0.33** 0.14 -0.04 -0.07 0.14 -0.23 0.47 0.80 0.01 0.26
(0.082) (0.162) (0.112) (0.125) (0.093) (0.126) (0.322) (0.718) (0.423) (0.501) (0.309) (0.694)
URBANICITY -0.08 -0.21 -0.04 -0.01 -0.05 -0.04 -0.10 -0.18 -0.05 -0.05 -0.03 -0.13
(0.089) (0.180) (0.130) (0.080) (0.114) (0.179) (0.096) (0.138) (0.108) (0.123) (0.100) (0.265)
GENDER (1=female) -0.35*** -0.16 -0.23* -0.43*** -0.44*** -0.45*** -0.39*** -0.25 -0.28** -0.49*** -0.49*** -0.48***
(0.063) (0.155) (0.115) (0.099) (0.085) (0.097) (0.069) (0.159) (0.102) (0.107) (0.079) (0.118)
SES 0.04 -0.03 -0.02 0.05 0.11** 0.17** 0.04 -0.03 -0.03 0.00 0.11 0.12
(0.042) (0.082) (0.048) (0.071) (0.039) (0.061) (0.047) (0.097) (0.076) (0.090) (0.063) (0.066)
SCHOOL TYPE (1=private) 0.17** 0.05 0.19 0.17 0.17* 0.31*** 0.16* 0.12 0.21 0.14 0.17* 0.26**
(0.064) (0.128) (0.098) (0.093) (0.082) (0.092) (0.068) (0.174) (0.114) (0.085) (0.076) (0.084)
SCHOOL TRACK (1=general) 0.41*** 0.52* 0.37* 0.49* 0.47*** 0.35 0.41** 0.89*** 0.23 0.34 0.44* 0.23
(0.116) (0.206) (0.154) (0.199) (0.133) (0.218) (0.139) (0.199) (0.150) (0.182) (0.191) (0.295)
STU-TEA RATIO -0.04* -0.04 -0.03 -0.03 -0.05*** -0.03 -0.04** -0.06* -0.03 -0.04 -0.06** -0.02
(0.014) (0.025) (0.020) (0.022) (0.013) (0.020) (0.016) (0.028) (0.023) (0.023) (0.017) (0.020)
SCHOOL CHOICE -0.13 -0.35 -0.08 0.03 -0.18 -0.20 -0.21* -0.50*** -0.10 -0.07 -0.23* -0.30
(0.094) (0.208) (0.155) (0.149) (0.145) (0.169) (0.101) (0.149) (0.101) (0.149) (0.117) (0.293)
PRESCORE (G9VEM_Z) 0.75*** 0.75*** 0.80*** 0.83*** 0.76*** 0.63*** 0.71*** 0.69*** 0.74*** 0.75*** 0.72*** 0.63***
(0.043) (0.087) (0.067) (0.074) (0.062) (0.076) (0.046) (0.077) (0.072) (0.050) (0.068) (0.061)
STUDY HOURS 0.50*** 0.37* 0.44*** 0.54*** 0.50*** 0.55*** 0.49*** 0.28 0.43** 0.50*** 0.50*** 0.48***
(1= >10hrs /week) (0.068) (0.170) (0.088) (0.071) (0.111) (0.073) (0.074) (0.166) (0.136) (0.103) (0.130) (0.135)
SELF ESTEEM 0.26*** 0.31** 0.24* 0.27*** 0.24*** 0.30*** 0.26*** 0.34** 0.22* 0.25*** 0.23** 0.31***
(0.052) (0.097) (0.120) (0.064) (0.065) (0.075) (0.056) (0.113) (0.087) (0.063) (0.089) (0.076)
MATH PT in middle school -0.05 0.05 0.02 -0.08 -0.08 -0.02 -0.05 0.15 -0.00 -0.13** -0.12* -0.09
(0.037) (0.075) (0.076) (0.060) (0.051) (0.079) (0.047) (0.092) (0.050) (0.050) (0.051) (0.076)
MATH PT in G11 0.55*** 0.53*** 0.50*** 0.55*** 0.58*** 0.52*** 0.58*** 0.71*** 0.53*** 0.41** 0.57*** 0.47*
(0.076) (0.123) (0.058) (0.102) (0.091) (0.134) (0.105) (0.149) (0.099) (0.142) (0.100) (0.240)
MATH PT in G12 0.43*** 0.57*** 0.49*** 0.39*** 0.37*** 0.27** 0.45*** 0.67*** 0.51*** 0.44*** 0.40*** 0.28*
(0.072) (0.141) (0.076) (0.097) (0.077) (0.098) (0.080) (0.161) (0.091) (0.119) (0.111) (0.133)
Constant 5.40*** 3.23*** 3.75*** 5.28*** 6.96*** 7.57*** 5.67*** 3.44*** 3.95*** 5.41*** 7.31*** 7.77***
(0.305) (0.472) (0.457) (0.528) (0.348) (0.479) (0.336) (0.672) (0.446) (0.493) (0.465) (0.791)

Observations 2,338 2,338 2,338 2,338 2,338 2,338 2,033 2,053 2,053 2,053 2,053 2,053
R-squared 0.345 0.334
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

342
Table Q.11. Effects of math tutoring in grade 11 on the CSAT math achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G11 0.55*** 0.53*** 0.50*** 0.55*** 0.58*** 0.52** 0.63** 1.02* 1.32 1.92*** 1.30 0.51
(0.076) (0.143) (0.095) (0.091) (0.111) (0.166) (0.241) (0.476) (0.701) (0.545) (0.795) (1.029)
URBANICITY -0.08 -0.21 -0.04 -0.01 -0.05 -0.04 -0.09 -0.22 0.01 -0.02 -0.10 -0.10
(0.089) (0.125) (0.114) (0.117) (0.153) (0.200) (0.092) (0.149) (0.119) (0.126) (0.110) (0.137)
GENDER (1=female) -0.35*** -0.16 -0.23* -0.43*** -0.44*** -0.45*** -0.38*** -0.26* -0.28* -0.34*** -0.41*** -0.47***
(0.063) (0.134) (0.108) (0.084) (0.073) (0.117) (0.065) (0.117) (0.116) (0.085) (0.115) (0.122)
SES 0.04 -0.03 -0.02 0.05 0.11 0.17* 0.04 -0.14 -0.02 0.04 0.12* 0.18
(0.042) (0.093) (0.070) (0.064) (0.072) (0.082) (0.044) (0.100) (0.073) (0.059) (0.057) (0.098)
SCHOOL TYPE (1=private) 0.17** 0.05 0.19* 0.17* 0.17 0.31** 0.18** 0.11 0.16 0.10 0.22* 0.21*
(0.064) (0.116) (0.087) (0.075) (0.088) (0.102) (0.065) (0.148) (0.088) (0.093) (0.090) (0.095)
SCHOOL TRACK (1=general) 0.41*** 0.52 0.37** 0.49* 0.47*** 0.35 0.39** 0.43* 0.29 0.32 0.42 0.49**
(0.116) (0.289) (0.125) (0.212) (0.114) (0.191) (0.121) (0.177) (0.243) (0.259) (0.239) (0.185)
STU-TEA RATIO -0.04* -0.04 -0.03 -0.03 -0.05** -0.03 -0.04* -0.03 -0.02 -0.03 -0.04 -0.04
(0.014) (0.027) (0.015) (0.021) (0.017) (0.026) (0.015) (0.041) (0.028) (0.020) (0.019) (0.022)
SCHOOL CHOICE -0.13 -0.35 -0.08 0.03 -0.18 -0.20 -0.17 -0.39* -0.02 -0.03 -0.19 -0.22
(0.094) (0.177) (0.134) (0.124) (0.159) (0.218) (0.098) (0.185) (0.110) (0.113) (0.116) (0.183)
PRESCORE (G9VEM_Z) 0.75*** 0.75*** 0.80*** 0.83*** 0.76*** 0.63*** 0.75*** 0.75*** 0.79*** 0.78*** 0.75*** 0.66***
(0.043) (0.084) (0.072) (0.085) (0.081) (0.097) (0.045) (0.097) (0.074) (0.065) (0.071) (0.075)
STUDY HOURS 0.50*** 0.37** 0.44*** 0.54*** 0.50*** 0.55*** 0.48*** 0.33* 0.48*** 0.40*** 0.39*** 0.53***
(1= >10hrs /week) (0.068) (0.138) (0.070) (0.066) (0.077) (0.069) (0.071) (0.141) (0.107) (0.081) (0.112) (0.115)
SELF ESTEEM 0.26*** 0.31*** 0.24** 0.27*** 0.24*** 0.30*** 0.28*** 0.40*** 0.22** 0.30*** 0.21** 0.27**
(0.052) (0.069) (0.088) (0.075) (0.068) (0.056) (0.054) (0.099) (0.068) (0.078) (0.064) (0.097)
MATH PT in middle school -0.05 0.05 0.02 -0.08 -0.08 -0.02 -0.07 -0.01 -0.10 -0.15* -0.14 -0.06
(0.037) (0.084) (0.058) (0.056) (0.048) (0.066) (0.042) (0.079) (0.085) (0.059) (0.082) (0.110)
MATH PT in G10 0.14 0.29 0.33** 0.14 -0.04 -0.07 0.12 0.08 0.08 -0.24 -0.28 -0.03
(0.082) (0.152) (0.111) (0.084) (0.129) (0.157) (0.103) (0.176) (0.261) (0.168) (0.246) (0.315)
MATH PT in G12 0.43*** 0.57*** 0.49*** 0.39*** 0.37*** 0.27* 0.40*** 0.53* 0.19 -0.01 0.15 0.26
(0.072) (0.157) (0.100) (0.095) (0.094) (0.104) (0.097) (0.256) (0.239) (0.153) (0.203) (0.314)
Constant 5.40*** 3.23*** 3.75*** 5.28*** 6.96*** 7.57*** 5.56*** 3.31*** 3.68*** 5.16*** 6.79*** 7.83***
(0.305) (0.700) (0.482) (0.488) (0.540) (0.495) (0.319) (0.713) (0.359) (0.275) (0.428) (0.552)

Observations 2,338 2,338 2,338 2,338 2,338 2,338 2,209 2,271 2,271 2,271 2,271 2,271
R-squared 0.345 0.341
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

343
Table Q.12. Effects of math tutoring in grade 12 on the CSAT math achievement
OLS IV
(1) (2) (3) (4) (5) (6) (1) (2) (3) (4) (5) (6)
VARIABLES Average q10 q25 q50 q75 q90 Average q10 q25 q50 q75 q90

MATH PT in G12 0.43*** 0.57*** 0.49*** 0.39** 0.37*** 0.27* 0.42** 0.40 0.73** 0.48** 0.61* 0.45
(0.072) (0.140) (0.109) (0.129) (0.103) (0.108) (0.157) (0.456) (0.271) (0.161) (0.290) (0.387)
URBANICITY -0.08 -0.21 -0.04 -0.01 -0.05 -0.04 -0.07 -0.21 -0.06 0.00 -0.15 -0.10
(0.089) (0.136) (0.122) (0.111) (0.099) (0.165) (0.095) (0.162) (0.147) (0.142) (0.127) (0.134)
GENDER (1=female) -0.35*** -0.16 -0.23** -0.43*** -0.44*** -0.45*** -0.34*** -0.19 -0.23 -0.38** -0.44*** -0.50***
(0.063) (0.136) (0.089) (0.081) (0.087) (0.107) (0.066) (0.114) (0.119) (0.115) (0.095) (0.101)
SES 0.04 -0.03 -0.02 0.05 0.11* 0.17** 0.04 -0.03 -0.06 0.02 0.09 0.15*
(0.042) (0.078) (0.062) (0.069) (0.051) (0.060) (0.045) (0.083) (0.063) (0.071) (0.059) (0.067)
SCHOOL TYPE (1=private) 0.17** 0.05 0.19 0.17 0.17 0.31* 0.19** 0.07 0.17 0.24** 0.22* 0.25*
(0.064) (0.139) (0.117) (0.093) (0.115) (0.126) (0.067) (0.156) (0.099) (0.085) (0.090) (0.127)
SCHOOL TRACK (1=general) 0.41*** 0.52* 0.37* 0.49** 0.47*** 0.35 0.45*** 0.67* 0.37** 0.53*** 0.54*** 0.40*
(0.116) (0.215) (0.167) (0.169) (0.123) (0.212) (0.127) (0.288) (0.127) (0.140) (0.120) (0.186)
STU-TEA RATIO -0.04* -0.04 -0.03 -0.03 -0.05** -0.03 -0.04* -0.07 -0.03 -0.04* -0.05* -0.03
(0.014) (0.029) (0.026) (0.020) (0.016) (0.020) (0.015) (0.041) (0.026) (0.017) (0.023) (0.018)
SCHOOL CHOICE -0.13 -0.35* -0.08 0.03 -0.18 -0.20 -0.16 -0.44* -0.09 0.02 -0.22 -0.19
(0.094) (0.142) (0.101) (0.131) (0.123) (0.150) (0.100) (0.209) (0.192) (0.165) (0.174) (0.137)
PRESCORE (G9VEM_Z) 0.75*** 0.75*** 0.80*** 0.83*** 0.76*** 0.63*** 0.75*** 0.81*** 0.79*** 0.87*** 0.77*** 0.68***
(0.043) (0.082) (0.086) (0.096) (0.078) (0.100) (0.045) (0.084) (0.089) (0.063) (0.062) (0.080)
STUDY HOURS 0.50*** 0.37* 0.44*** 0.54*** 0.50*** 0.55*** 0.47*** 0.42** 0.35*** 0.47*** 0.44*** 0.51***
(1= >10hrs /week) (0.068) (0.158) (0.117) (0.079) (0.104) (0.098) (0.072) (0.156) (0.093) (0.108) (0.096) (0.107)
SELF ESTEEM 0.26*** 0.31*** 0.24** 0.27*** 0.24** 0.30*** 0.29*** 0.31*** 0.28** 0.27*** 0.27** 0.28**
(0.052) (0.071) (0.091) (0.062) (0.078) (0.082) (0.055) (0.076) (0.091) (0.063) (0.093) (0.109)
MATH PT in middle school -0.05 0.05 0.02 -0.08 -0.08 -0.02 -0.06 0.01 -0.03 -0.08 -0.11 -0.08
(0.037) (0.080) (0.059) (0.049) (0.059) (0.079) (0.040) (0.075) (0.059) (0.061) (0.072) (0.091)
MATH PT in G10 0.14 0.29* 0.33** 0.14 -0.04 -0.07 0.11 0.45** 0.33** 0.07 -0.08 -0.17
(0.082) (0.135) (0.106) (0.121) (0.109) (0.153) (0.088) (0.143) (0.110) (0.110) (0.148) (0.147)
MATH PT in G11 0.55*** 0.53*** 0.50*** 0.55*** 0.58*** 0.52** 0.60*** 0.53* 0.54*** 0.67*** 0.58*** 0.54***
(0.076) (0.152) (0.119) (0.096) (0.088) (0.161) (0.094) (0.228) (0.134) (0.127) (0.165) (0.162)
Constant 5.40*** 3.23*** 3.75*** 5.28*** 6.96*** 7.57*** 5.43*** 3.64*** 3.97*** 5.17*** 7.06*** 7.76***
(0.305) (0.674) (0.548) (0.552) (0.348) (0.487) (0.323) (0.711) (0.632) (0.438) (0.576) (0.652)

Observations 2,338 2,338 2,338 2,338 2,338 2,338 2,117 2,117 2,117 2,117 2,117 2,117
R-squared 0.345 0.343
Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05
Dependent variable: CSAT_M (math score on the CSAT)

344