UC Santa Cruz

UC Santa Cruz Electronic Theses and Dissertations

Title
Essays in Financial Development and Economic Growth

Permalink
https://escholarship.org/uc/item/62v5r34h

Author
Bun, Linh

Publication Date
2018

Peer reviewed|Thesis/dissertation

eScholarship.org Powered by the California Digital Library

University of California
 UNIVERSITY OF CALIFORNIA
SANTA CRUZ

ESSAYS IN FINANCIAL DEVELOPMENT AND ECONOMIC GROWTH

A dissertation submitted in partial satisfaction of the
requirements for the degree of

DOCTOR OF PHILOSOPHY

in

ECONOMICS

by

Linh Bun

March 2018

The Dissertation of Linh Bun

is approved:

Professor Nirvikar Singh, Co-Chair

Professor Joshua Aizenman (USC), Co-Chair

Professor Carl Walsh

Tyrus Miller
Vice Provost and Dean of Graduate Studies
Copyright c by

Linh Bun

2018
Table of Contents

List of Figures vi

List of Tables vii

Abstract ix

Dedication xii

Acknowledgments xiii

I Financial Development and Growth 1

1 Financial Development and Growth: Multi-dimensional View 2
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.4.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.5 Empirical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.5.1 Dynamic Panel Data Methodology . . . . . . . . . . . . . . . . 14
1.5.2 Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.3 Within Model: First Difference . . . . . . . . . . . . . . . . . 18
1.5.4 Financial Institutions Depth . . . . . . . . . . . . . . . . . . . 19
1.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.6.1 Estimation Model: Linear . . . . . . . . . . . . . . . . . . . . 21
1.6.2 Estimation Model: Quadratic . . . . . . . . . . . . . . . . . . . 31
1.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.8 Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

iii
 1.9 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

2 Heterogeneous Patterns of Financial Development: Implications for Asian

Financial Integration 36
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.2 Financial Integration and Financial Development . . . . . . . . . . . . 39
2.3 Measuring Financial Development: Level and Patterns . . . . . . . . . 47
2.3.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
2.3.2 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
2.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
2.6 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

II Structural Changes and Growth 83

3 Structural Changes and Labor Market Frictions 84
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
3.1.2 Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
3.1.3 Contribution . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
3.2 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.2.1 Production . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
3.2.2 Preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
3.3 Quantitative Analysis: Closed Economy . . . . . . . . . . . . . . . . . 99
3.3.1 Closed Economy Solution . . . . . . . . . . . . . . . . . . . . 99
3.3.2 Closed Economy Calibration . . . . . . . . . . . . . . . . . . . 102
3.3.3 South Korea . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.4 Quantitative Analysis: Open Economy . . . . . . . . . . . . . . . . . . 109
3.4.1 Open Economy Solution . . . . . . . . . . . . . . . . . . . . . 109
3.4.2 Open Economy Calibration . . . . . . . . . . . . . . . . . . . . 113
3.5 Sectoral Relative Prices: International Comparison Program (ICP) 2005 120
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.7 Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.8 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124

A Financial Development and Growth: Multi-dimensional View 143

A.1 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
A.1.1 Data Descriptions and Sources . . . . . . . . . . . . . . . . . . 143
A.1.2 Data Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 144
A.1.3 Estimation Results . . . . . . . . . . . . . . . . . . . . . . . . 150

iv
B Heterogeneous Patterns of Financial Development: Implications for Asian
Financial Integration 157
B.1 Appendix I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
B.2 Appendix II: Financial Development Index Pillars . . . . . . . . . . . . 163

C Structural Changes and Labor Market Frictions 166

C.1 Appendix I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
C.1.1 International Comparison Program 2005 . . . . . . . . . . . . . 166
C.2 Appendix II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
C.2.1 Closed Economy: Calibration for South Korea (Discussion &
Explanation) . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

v
List of Figures

2.1 Patterns of Financial Development, Developed Economies, 2014. . . . . 57

2.2 Patterns of Financial Development, Emerging Economies, 2014. . . . . 60
2.3 Level and Pattern Distance, FDIndex8, 2010 . . . . . . . . . . . . . . . 69
2.4 Level and Pattern Distance, FDIndex8, 2014 . . . . . . . . . . . . . . . 70
2.5 Level and Pattern Distance, FDIndex7, 2010 . . . . . . . . . . . . . . . 71
2.6 Level and Pattern Distance, FDIndex7, 2014 . . . . . . . . . . . . . . . 72
2.7 Level and Pattern Distance, FDIndex10, 2010 . . . . . . . . . . . . . . 73
2.8 Level and Pattern Distance, FDIndex10, 2014 . . . . . . . . . . . . . . 74
2.9 Level and Pattern Distance, FDIndex9, 2010 . . . . . . . . . . . . . . . 74
2.10 Level and Pattern Distance, FDIndex9, 2014 . . . . . . . . . . . . . . . 75
2.11 Level and Pattern Distance, FDIndex8, 2010 . . . . . . . . . . . . . . . 77
2.12 Level and Pattern Distance, FDIndex8, 2014 . . . . . . . . . . . . . . . 78

3.1 Closed Economy Model Simulation: Sectoral Employment Share . . . . 108

3.2 Sectoral Relative Prices (Manufacturing/Services) and Income . . . . . 121
3.3 Sectoral Relative Prices (Manufacturing/Agriculture) and Income . . . . 122

vi
List of Tables

1.1 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.2 Multi-dimensional View of Financial Development . . . . . . . . . . . 15
1.3 Panel OLS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.4 System GMM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.5 System GMM (With Interaction) . . . . . . . . . . . . . . . . . . . . . 30
1.6 Panel OLS: Quadratic . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.7 System GMM: Quadratic . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.1 Components of FDIndex, WEF Global Competitiveness Report. . . . . 51

2.2 Economies in the Analysis. . . . . . . . . . . . . . . . . . . . . . . . . 52
2.3 Financial Development Index Levels. . . . . . . . . . . . . . . . . . . 58
2.4 Regional Summary Statistics. . . . . . . . . . . . . . . . . . . . . . . . 61
2.5 Level and Pattern Differences FDIndex8 . . . . . . . . . . . . . . . . . 63
2.6 Level and Pattern Differences FDIndex7 . . . . . . . . . . . . . . . . . 64
2.7 Level and pattern differences FDIndex10. . . . . . . . . . . . . . . . . 66
2.8 Level and pattern differences FDIndex9. . . . . . . . . . . . . . . . . . 67
2.9 Clustering based on Levels of Financial Development . . . . . . . . . . 76
2.10 Clustering based on Patterns of Financial Development . . . . . . . . . 79

3.1 Calibrations for South Korea . . . . . . . . . . . . . . . . . . . . . . . 109

A.1 Data Description and Sources . . . . . . . . . . . . . . . . . . . . . . . 143

A.2 Credit to the Private Sector Over GDP (Selected Years) . . . . . . . . . 144
A.3 Bank Lending and Deposit Rate Spreads (Selected Years) . . . . . . . . 147
A.4 System GMM: 2nd and 3rd Lag (With Interaction) . . . . . . . . . . . 151
A.5 System GMM: 3rd Lag (With Interaction) . . . . . . . . . . . . . . . . 152
A.6 System GMM: 4th Lag (With Interaction) . . . . . . . . . . . . . . . . 153
A.7 Panel OLS (Comparison with Arcand et al. (2015a) Dataset) . . . . . . 154
A.8 System GMM (Comparison with Arcand et al. (2015a) Dataset) . . . . 155

vii
A.9 Panel OLS: Quadratic . . . . . . . . . . . . . . . . . . . . . . . . . . 156

B.1 Institutional Environment . . . . . . . . . . . . . . . . . . . . . . . . . 158

B.2 Business Environment . . . . . . . . . . . . . . . . . . . . . . . . . . 159
B.3 Additional Components and Variables . . . . . . . . . . . . . . . . . . 160
B.4 Level and Pattern Differences FDIndex14 . . . . . . . . . . . . . . . . 161
B.5 Level and Pattern Differences FDIndex13 . . . . . . . . . . . . . . . . 162
B.6 Clustering based on Levels of Financial Development . . . . . . . . . . 163
B.7 Financial Development Index: Category . . . . . . . . . . . . . . . . . 164
B.8 Financial Development Index: Category . . . . . . . . . . . . . . . . . 165

C.1 Basic Heading Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . 170

C.2 Sectoral Employment Notation . . . . . . . . . . . . . . . . . . . . . . 172

viii
 Abstract

Essays in Financial Development and Economic Growth

by

Linh Bun

Chapter 1 (Part I): Financial Development and Growth: Multi-dimensional

View: analyze the effects of ”financial development” on economic growth using a

multi-dimensional view. Existing literature uses a single dimension, financial depth,

as measure of financial development. Our approach is to consider a multi-dimensional

view of financial development and to take in to account other dimensions of financial

development. We categorize financial development (financial institutions and markets)

along three dimension: financial depth, access and efficiency. We find that for the

country sample analyzed, financial institutions depth and efficiency are compliment.

Chapter 2 (Part I): Heterogeneous Patterns of Financial Development: Impli-

cations for Asian Financial Integration studies detailed differences in patterns of fi-

nancial development across the major Asian economies, including three of the regions

largest economies (China, Japan and South Korea), to understand how these differ-

ences might affect possibilities for greater regional financial integration. In particular,

the paper argues that heterogeneous patterns of financial development, and not just dif-

ferences in levels of financial development, may present an economic challenge to re-

gional financial integration efforts, aside from possible political challenges. The paper

ix
provides background on the case for financial openness, Asian experiences with finan-

cial integration, and regional economic responses to external shocks. It also discusses

policy options, including regulatory reform and coordination, and possible risk man-

agement policies and institutions, in the context of heterogeneous patterns of financial

development.

Chapter 3 (Part II): Structural Changes and Labor Market Frictions ana-

lyzes a calibrated three-sector (agriculture, manufacturing and services) endogenous

growth model with learning-by-doing to study the process of structural transformation

in emerging and developing economies. We introduced a calibrated three-sector en-

dogenous growth model with sectoral labor adjustment costs. The model generates

a hump-shaped in manufacturing employment share. We introduce a key feature of

leaning-by-doing in manufacturing sector to a standard three-sector structural transfor-

mation model with labor market friction, which is new to the literature. Labor market

friction plays an important role in generating a novel feature of structural changes,

humped-shaped in manufacturing employment. In our model, opening to trade can ac-

celerate or decrease structural transformation (labor movement into the manufacturing

sector) depending on comparative advantage, whether the economy specializes in agri-

culture or non-agriculture goods after opening to trade. The calibrated model with labor

market friction generated the hump-shaped labor share in manufacturing for South Ko-

rea.

x
xi
To my wife and my parents.

xii
 Acknowledgments

I want to thank all members of my dissertation committee for their unconditional

support.

Chapter 1 (Part I): Financial Development and Growth: Multi-dimensional

View: I want to sincerely thank both of my advisors, Joshua Aizenman (USC) and

Nirvikar Singh, for their continuous support during this project. This work also bene-

fits greatly from my joint project (Bun and Singh, 2016) with Nirvikar Singh, published

at the International Economic Journal (DOI). I would like to thank the Center of Analyt-

ical Finance (CAFIN) at the University of California, Santa Cruz for generous financial

support in order to purchase the data.

Chapter 2 (Part I): Heterogeneous Patterns of Financial Development: Impli-

cations for Asian Financial Integration: I thank Nirvikar Singh for his support and

guidance throughout the process. This chapter is a version of our joint paper presented

at the Inaugural International Conference on Evolving Finance, Trade and Investment

in Asia, held at the Centre on Asia and Glottalization, Lee Kuan Yew School of Public

Policy, National University of Singapore, on September 16-17, 2015. I am grateful for

Nirvikar Singh’s presentation at the conference and valuable comments and suggestions

by the discussant, Siu Fung Yiu.

Chapter 3 (Part II): Structural Changes and Labor Market Frictions: this pa-

per is written during my internship at the International Monetary Fund (IMF). I thank

xiii
my supervisors at the IMF: Nikola Spatafora and Seung Mo Choi for their support dur-

ing my internship, Summer 2013, for me to work on this paper. I also thank George

Bulman, Johanna Francis, Grace Gu, Jon Robinson, Carl Walsh and seminar partic-

ipants at the UCSC Macroeconomics Workshop, UCSC Brown bag for very helpful

comments and discussions. I gratefully acknowledge the World Bank for granting me

access to the disaggregated International Comparison Program ICP 2005 dataset.

xiv
 Part I

Financial Development and Growth

1
Chapter 1

Financial Development and Growth:

Multi-dimensional View

1.1 Introduction

The recent global financial crisis, both in terms of depth and severity, has garnered

much attention to the functioning of the financial system. The analysis of the impact

of financial development on economic growth is becoming popular. While there are

many controversy both regarding the results as well as the definition of financial devel-

opment. There are many ways to define ”financial development” in the data and one

such measure of financial development could be in terms of financial depth proxied by

the percentage share of domestic private credit to GDP and with this definition, the re-

2
sults suggest that financial development has positive impact on growth only up to some

threshold (Arcand et al., 2015a; Law and Singh, 2014).

However, the lack of strong results with financial depth suggest the need for us-

ing quality of finance as a measure of financial development rather than quantity of

finance.1 Quality of finance could be defined broadly as the efficiency of the financial

system in allocating of credit to productive firms. Another way of measuring financial

development may be in terms of the efficiency of the financial system.

Financial integration also affects financial development. In particular, international

capital flows can have both negative and positive effects on economic growth through

its interaction with financial development. With large capital inflows, inefficient finan-

cial/banking system can misallocate resources and have harmful effects on long-term

growth, reallocating resources such as labor from productive tradable sector to less pro-

ductive non-tradable sector and resulting in the expansion of low productivity sector at

the expense of productive sector: as in Reis (2013) for the case of developed country

(Portugal)2 and developing countries in Buera et al. (2011).

1 Most existing literature uses financial depth as a measure of financial development, i.e. the ratio of
private credit to GDP and this broadly reflects quantity rather quality.
2 Reis (2013) focuses on explaining low productivity in non-tradable sector due to misallocation of

resources in the financial sector

3
1.2 Literature

Quantity and Quality of Finance: Aizenman et al. (2015) analyzes the impact of

financial development on sectoral output growth and they found that the impacts are

heterogeneous. In addition, they conclude that there is a non-linear effects of financial

deepening in some sectors (results are not robust in some sectors) as well as different

effects between Asia and Latin America. Interestingly, there is Dutch-disease effects

of financial development on the manufacturing sector, i.e. faster growth of the financial

sector hinder growth of the manufacturing sector and this results is elaborated Benigno

and Fornaro (2014b) and Benigno et al. (2015). Furthermore, they suggest the need for

measuring quality of finance (defined broadly as the efficiency of the financial system

in allocating of credit to productive firms). Existing literature uses mostly financial

depth as measure of financial development such as ratio of private credit to GDP.

In addition, Benigno et al. (2015) examines episodes of large capital inflows follow-

ing the recent literature especially the experiences of the Eurozone economies such as

Spain & Portugal after the introduction of the Euros. There are a few paper on the the-

oretical front linking large capital inflows and slow productivity growth (productivity

level?). The analysis focuses on medium/long-term study and they look at gross capital

flows. However, they abstract from using different measures of financial development.

Financial Integration: Financial Development & Growth: Herwartz and Walle

4
(2014b) analyze the role of financial integration on the finance-growth nexus using

aggregate macroeconomic variables. They find the asymmetric role of financial devel-

opment on economic growth and in particular, at very high level of financial openness

financial development has negative effects on growth. It is important to note that Her-

wartz and Walle (2014b) also emphasizes the different dimensions of financial develop-

ment and for example, measure of financial institutions such as private credit (by banks

and financial institutions) focuses on the role of the intermediaries in providing funds

(intermediating between savers and borrowers) as opposed to measure of financial mar-

kets.3 However, they do not analyze the role of sectoral reallocation and its linkage

with financial integration.

Literature on the real effects of financial integration Bonfiglioli (2008) points to

the potential impact of financial integration on growth through financial development

(competition effects and reallocation effects). However, they did not analyze the link-

ages at the sectoral level. On one level, the impact of financial development on eco-

nomic growth interacts with the level of financial openness and in particular, at a very

high level of financial integration financial development can have negative impact on

growth (Herwartz and Walle, 2014b). However, it is not clear how and which chan-

nels operates in Herwartz and Walle (2014b). There are several channels that financial

integration could impact aggregate economic growth through financial development:

3 The different in dimensions of financial development across countries could reflect stages of devel-

opment such as OECD where financial markets play a more prominent role whereas financial institutions
play a more active role in emerging and developing countries (banks are more important).

5
competition and reallocation effects (Bonfiglioli, 2008).

Financial Reform Effect on Finance-Growth Nexus: Demetriades and Rousseau

(2016) studies the impact of financial reforms (based on data from Abiad et al. (2010))

on the relationship between finance and growth. Following the literature that points to

the vanishing effect of finance on growth (Arcand et al., 2015a; Rousseau and Wachtel,

2011), which emphasizes the diminishing impact of financial depths on growth dimin-

ishes in recent years, their results show that financial reforms does not change the van-

ishing effect. In particular, based on measures of financial liberalization Abiad et al.

(2010),4 the impact of components of financial reforms (marginal effect) on finance-

growth nexus depends on level of banking regulation during recent years of financial

liberalizations. However, they do not analyze whether the impact of financial depth

on economic growth depends on the efficiency of the banking system or level of bank-

ing supervision (i.e. exploring the interaction between financial depth and financial

efficiency on finance-growth nexus).

In addition, Prati et al. (2012) uses unique and extensive financial reform index for

advanced and developing countries. They explore the association between financial

reform and growth. The financial reform index is based on Abiad et al. (2010). The

reform database in Abiad et al. (2010) provides indices between 0 and 1 for: openness

to international trade (real sector): average tariff rates and restrictions on the current
4 Financial reforms: strength of banking supervision, ease of bank entry, absence of distortions in

credit allocation (ease of credit controls), sophistication of securities markets, extent of privatization,
interest rate liberalization and capital account openness.

6
account transactions; product market structural reforms (real sector): agriculture and

electricity and telecommunications; domestic financial liberalization (financial sector):

banking sector: credit controls, interest rate controls, competition restrictions, degree of

state ownership, quality of bank supervision and regulation; capital/securities (equities

and bonds) markets and external capital account liberalization (financial sector).

Vanishing Effect of Finance on Growth: Rousseau and Wachtel (2011) presents

the ”vanishing effect” of finance on growth. In particular, they point out that the positive

relationship between finance and growth is weaker in recent years (after 1990). They

confirm their main hypothesis using cross-section, panel and dynamic panel methods.

They further investigate possible explanations: financial crises, recent financial liber-

alization and recent development of the equity market as a possible substitute. The

result indicates that the quality of finance could be an important factor (Rousseau and

Wachtel, 2011) but future research can extend the analysis to investigate whether there

is relevant threshold for quality of finance dimension as well.

Benigno and Fornaro (2014a) presents a two-sector model of the reallocation of

resources between tradable and non-tradable sectors during periods of large capital in-

flows. On the other hand, they ignore the role of domestic bank credit (total credit)

and they do not look at the particular mechanism in which large international capital

flows affect the reallocation process. They analyze capital flows (total: gross or net)

as oppose to investigating the types or sectoral destinations of capital flows. Samarina

7
and Bezemer (2016) emphasizes international capital flows into non-bank sector as de-

terminant of distribution of domestic bank credit (households and business). However,

Samarina and Bezemer (2016) does not analyze the impact on output and growth (pro-

ductivity). However, they do not link the declining business sector share of domestic

bank credit (or increasing household sector share) to growth (finance-growth nexus).

Threshold in Financial Development: Levine (2005) discusses the co-evolution of

finance and growth. The impact of technological advancement (productivity shocks or

improvement) on growth depends on level of financial development. In particular, this

argues for threshold of financial development in facilitating the impact of technological

improvement on growth. Technology itself may also directly affect financial develop-

ment. Levine (2005) has several remaining qualifications (still plague the literature)

regarding the finance-growth literature: different methodology has its own drawback

and measure of financial development does not match well with theory on the role of

financial development: reducing information frictions, corporate governance, realloca-

tion of resources.

Rajan and Zingales (1998) illustrated a mechanism in which financial development

helps economic growth: external financing. The key methodological contribution is in-

troducing the interaction term between financial development and measure of external

financing dependency. They hypothesize that the coefficient for the interaction term is

positive.5 Recent advancement in theoretical econometrics (panel data), Seo and Shin
5 Positive coefficient for the interaction term suggests that higher level of financial development in-

8
(2016) allows for both endogenous threshold and regressors. It is important to note that

Law and Singh (2014) use similar empirical framework but with only a single endoge-

nous regressor (financial development) and all the remaining covariates are assumed to

be either predetermined or strictly exogenous.

Financial Intermediary Development Indicators & Source of Economic Growth:

Benhabib and Spiegel (2000) found that different financial intermediary development

indicators affect growth through different channels/components (total factor productiv-

ity, physical and human capital investment/accumulation). In particular, indicators of

financial development (quantities and qualities of finance).

Financial Cycles: Booms and Contractions: Kannan (2012) studied the impact

of credit destruction during financial crises on sectoral output growth (dependence on

external finance à la Rajan and Zingales (1998)). However, clearly, the author only

focuses on periods of financial contractions as well as abstracting from impacts on

aggregate output (resource reallocation).6

Also, the linkages between financial sector services and sectoral growth could be

beyond just financing needs (external dependence) as illustrated in Claessens and Laeven

(2003). Moreover, Fisman and Love (2004) analyzes the growth opportunities of sec-

tor (generalized beyond that of Rajan and Zingales (1998)) and their method could be
duces larger impact on growth of sectors relying more on external financial. Hence, financial develop-
ment facilitates growth of sectors through efficient reallocation of resource to sectors need it most.
6 Future work could extend to analyze the impacts of financial cycles on the real sectoral output growth

and potentially, the impacts could be heterogeneous depending on external dependence and growth op-
portunity.

9
extended through time (temporal dimension) to analyze with panel data methodology.

Financial Integration, Financial Development and Institutions: Lucey and Zhang

(2011) provides evidences of the impacts of different form of financial integration

(credit/debt or equity) on firm corporate financing choices/structure (debts or equity).

However, they do not look at firm growth and they suggested that looking at firm-level

integration measures such as exposure to international sources (foreign assets or liabil-

ities) would be a good avenue for future research.

One strand of literature Claessens and Laeven (2003) linking financial development

and growth that emphasize the equally important role of legal rights (institutions) that

could impact the benefits of firm investment in intangible assets as opposed to tangible

assets as they benefits could be easier to be copied or stole. They follow the strategy of

Rajan and Zingales (1998) in construct measures of firms dependence on intangible (ra-

tio of intangible to tangible assets). Large literature on productivity and misallocation

Restuccia and Rogerson (2013) focuses on financial friction. However, institutional

frictions could be as equally important channel of misallocation as well.

Large capital inflows7 is shown to increase likelihood of financial crises, Caballero

(2016) and credit growth is important predictors as well as asset prices increases. Fo-

cusing on financial depth (size) as the main measure of financial development could
7 The McKinsey Global Institute report discusses the importance of financial development on eco-
nomic growth (hence, contributing to the literature). However, they also stress the importance of finance
beyond simple size measure (private domestic credit). In particular, they include measure of equity in
their financial development measure as well as distinguishing credit access by sector.

10
miss out on other important aspect of financial development that could affect economic

response to capital flows volatility. One potential candidate is quality of financial insti-

tutions. This could be linked with the literature on the asymmetric affects of financial

development on growth. In particular, financial booms and financial contractions have

asymmetric affects on the real economy Aizenman et al. (2013b).

Levchenko et al. (2009) uses difference-in-differences methodology (along the line

of non- liberalizing over liberalizing countries;8 whereas Rajan and Zingales (1998)

utilizes heterogeneity along external financing dependency) and Propensity Scores and

Matching (PSM) to find the effects of financial liberalization Kaminsky and Schmukler

(2008) on output, TFP growth and inputs (capital and labor).

However, they do not analyze the non-linearity of finance-growth nexus and the

role of financial development on the real impacts of financial integration or threshold in

international finance. They focus on measures of financial liberalization from Kamin-

sky and Schmukler (2008) not the increasing/decreasing nature of financial openness

(overtime) as in Herwartz and Walle (2014a).

Samarina and Bezemer (2016) provides evidence that the non-financial business

sector share of domestic bank credit is smaller for countries with larger international

capital flows into non-bank sector. In short, international capital flows to non-bank

competes with domestic banks in funding firms/businesses. This suggests a linkage

8 Popov (2014) employs similar estimation strategy, however, they make use all of the panel data

instead of dividing the data into pre-liberalization and post-liberalization.

11
between types of international capital flows and share of domestic bank credit (business

and households). The conclusion is along the same line as Lane and McQuade (2014).

However, they do not link the declining non-financial business sector share of domestic

bank credit (or equivalently, increasing household sector share) to growth (finance-

growth nexus).

Samarina and Bezemer (2016) emphasizes international capital flows into non-bank

sector as determinant of distribution of domestic bank credit (households and business).

However, Samarina and Bezemer (2016) does not analyze the impact on output and

growth (productivity). However, they do not link the declining business sector share of

domestic bank credit (or increasing household sector share) to growth (finance-growth

nexus). Davis et al. (2016) focuses on domestic credit as a whole and abstract from the

key role of the types of domestic credit: households and non-financial firms.

1.3 Motivation

Existing literature focuses on a single dimension of financial development, which

is mostly proxied by financial depth using private domestic credit, with the exception

of Aizenman et al. (2015).9 Our approach is to consider a multi-dimensional view of

financial development and while this multi-dimensional view of financial development

has been used in other literature (Sahay et al., 2015; Svirydzenka, 2016) we think that
9 They also discuss the important role of quality of financial development besides the standard mea-

sure in the literature

12
this is the first to apply this view on the analysis of financial development and economic

growth. Broadly speaking, similar to Sahay et al. (2015), we categorize financial de-

velopment (financial institutions and markets) along three dimension as in Table 1.2:

financial depth, access and efficiency.10

1.4 Data

Detailed data descriptions and sources are presented in appendix Table A.1). The

data are from the World Bank World Development Indicators (WDI) (World Bank,

2016) and Global Financial Development Database (GFDD) (Cihak et al., 2012) and

for benchmark comparison purpose, we also provide results with dataset from Arcand

et al. (2015a) (see appendix Table A). For all estimations, we employ panel data for the

period 1960-2015.11 In line with the finance-growth nexus literature, the data used in

regressions are averaged over five-year periods.12

[Add more detail data description (complements Table A.1 of appendix A) here]
10 Table 1.2 also provides a nice illustration of multi-dimensional view on financial development.
11 The list of countries along with data on domestic private credit and lending-deposits spreads are
provided in Appendix A (Table A.3 and A.2).
12 System GMM can be employed in a panel data with large cross-section observations, n, but small

number of time periods,t, (large n and small t). In addition, the five-year averaging smooths out business
cycle fluctuations and fits well with medium to long-run analysis

13
1.4.1 Summary

The summary is shown in Table 1.1. The detailed of private credit to GDP ratio

(selected years) is provided in Appendix A.2 and the bank lending-deposits spread rates

(selected years) is shown in Appendix A.3.

Table 1.1: Summary Statistics

Obs. Mean Std. Dev. Min. Max.

Growth (annual %) 1744 2.09 4.33 -42.62 69.00
Private domestic credit (% of GDP, in decimal
1588 0.36 0.34 0.00 2.47
points)
Lending-Deposit rate spread (in decimal points) 917 0.08 0.07 0.00 0.62
Education (average years) 1442 5.88 3.06 1.00 13.42
Gov’t consumption (% GDP) 1568 16.19 7.64 2.34 123.89
Inflation (%) 1486 42.51 342.63 0.00 8603.28
Trade openness (% of GDP) 1629 78.42 50.29 0.67 444.81
Source: World Bank WDI and GFDD

1.5 Empirical Analysis

1.5.1 Dynamic Panel Data Methodology

We employ estimation strategy using the system GMM estimator, which was pro-

posed by Arellano and Bover (1995) and Blundell and Bond (1998). As is standard in

the finance-growth nexus literature, we use the two-step procedure suggested by Arel-

lano and Bond (1991) with robust standard errors following Windmeijer (2005) finite

14
 Table 1.2: Multi-dimensional View of Financial Development

Financial Institutions Financial Markets

Private-sector credit (% of GDP) Stock market capitalization to GDP
Pension fund assets (% of GDP) Stocks traded to GDP
Depth

Mutual fund assets (% of GDP) International debt securities government (% of GDP)

Insurance premiums, life and non- Total debt securities of non-financial corporations (% of
life (% of GDP) GDP)
Total debt securities of financial corporations (% of GDP)
Branches (commercial banks) per Percent of market capitalization outside of top 10 largest
Access

100,000 adults companies

Total number of issuers of debt (domestic and external,
ATMs per 100,000 adults
non-financial corporations, and financial corporations)
Net interest margin Stock market turnover ratio (stocks traded/capitalization)
Efficiency

Lending-deposits spread
Non-interest income to total income
Overhead costs to total assets
Return on assets
Return on equity
Source: Sahay et al. (2015) and Svirydzenka (2016)

sample correction.13

1.5.2 Estimation

The growth-finance literature focuses on the following dynamic association be-

tween growth and financial depths and in particular, the dynamic panel estimation is

as follows:

0
Growthit = α + βFDi,t−1 + Xi,t−1 γ + µi + ηt + εi,t . (1.1)

0
where: Xi,t−1 is initial (lagged) key control variables (in logarithm) in the literature:

initial income (log of GDP per capita): lgd p pci,t−1 , initial schooling (Barro-Lee)
13 The
estimation is done using xtabond2 routine in STATA (Roodman, 2003, 2009a). For the discus-
sion of too many instruments issue, see Roodman (2009b).

15
such as average years of schooling, ratio of government consumption expenditures to

GDP, inflation rate and ratio of exports plus imports to GDP, FDi,t−1 is lagged measure

of financial institutions depths, domestic private credit PCit , and financial institutions

0
efficiency, bank lending-deposits spreads LDSPREADit , Xi,t−1 is lagged measure of

controls discussed at above, µi is individual fixed effects (dummies): controls for un-

observed (omitted) country heterogeneity,14 ηt is time fixed effect (dummies), εit is

disturbances and the following equation:

Growthit = ∆ log (GDP Per Capita)it . (1.2)

In simplifying the notation, 1.1 above could be rewritten, using 1.5, as follows:

log (GDP Per Capita)it = α + (γ1 + 1) log (GDP Per Capita)i,t−1

0
+ βFDi,t−1 + Xi,t−1 γ + µi + ηt + εi,t . (1.3)

Hence, we reach the standard dynamic panel estimation equation (1.3). In 1.1,

the dependent variable denotes growth in GDP per capita, which is calculated as the

changes in log of GDP per capita. In this specification, the main variable of interest is

measure of financial development FDi,t−1 and it is proxied by financial depth, measured

as the ratio of private credit to GDP provided by deposit-taking institutions and other
14 Wooldridge (2010) and Angrist and Pischke (2009) provide further discussion on the fixed effect

model.

16
financial institutions. As in the standard growth and finance literature, 1.1 is a dynamic

0
model with lagged dependent variable as one of the covariates in Xi,t−1 . The covariates

0
Xi,t−1 also includes: initial schooling, general government consumption expenditures

as percentage of GDP, inflation rate and trade (exports plus imports as percentage of

GDP).

Equation 1.1 could be rewritten to explicitly include lagged dependent variable as in

1.3. Equation 1.3 includes time-invariant country-specific fixed effect µi (a set of coun-

try dummy variables) and this is included to control for unobserved country-specific

heterogeneous characteristics. This is one of the advantage of using a panel data analy-

sis as oppose to cross-country analysis. The equation also has a set of time fixed effect.

For quadratic specification, we will employ the following:

0
Growthit = α + βFDi,t−1 + δFD2i,t−1 + Xi,t−1 γ + µi + ηt + εi,t . (1.4)

0
where: Xi,t−1 is initial (lagged) key control variables (in logarithm) in the literature:

initial income (log of GDP per capita): lgd p pci,t−1 , initial schooling (Barro-Lee)

such as average years of schooling, ratio of government consumption expenditures to

GDP, inflation rate and ratio of exports plus imports to GDP, FDi,t−1 is lagged measure

of financial institutions depths, domestic private credit PCit , and financial institutions

0
efficiency, bank lending-deposits spreads LDSPREADit , Xi,t−1 is lagged measure of

controls discussed at above, µi is individual fixed effects (dummies): controls for un-

17
observed (omitted) country heterogeneity,15 ηt is time fixed effect (dummies), εit is

disturbances16

1.5.3 Within Model: First Difference

The dynamic equation 1.3 comprises of both omitted variable: time-invariant country-

specific fixed effect µi and lagged dependent variable log (GDP Per Capita)i,t−1 . In es-

timating the fixed effect model, we could use first-difference version of 1.3, removing

the time-invariant fixed effect µi , as follows:

∆ log (GDP Per Capita)it = α + (γ1 + 1) ∆ log (GDP Per Capita)i,t−1

0
+ β∆FDi,t−1 + ∆Xi,t−1 γ + ∆ηt + ∆εi,t .

For simplicity, let θ = γ1 + 1 and hence,

∆ log (GDP Per Capita)it = α + θ∆ log (GDP Per Capita)i,t−1

0
+ β∆FDi,t−1 + ∆Xi,t−1 γ + ∆ηt + ∆εi,t . (1.6)

15 Wooldridge (2010) and Angrist and Pischke (2009) provide further discussion on the fixed effect
model.
16 Similar to above, Growth is defined as follows:

Growthit = ∆ log (GDP Per Capita)it . (1.5)

18
1.5.4 Financial Institutions Depth

In this section, we will focus on financial institution depth and in particular, we

will use private domestic credit as a percentage of GDP as our measure of financial

development.17 Following empirical model in 1.1, we have the following estimation

equation:

Growthi,t = α + β1 PCi,t−1 + γ1 LGDPi,t−1 + γ2 LEDUi,t−1 + γ3 LGCi,t−1

+ γ4 LOPENi,t−1 + γ5 LINFLi,t−1 + µi + ηt + εi,t . (1.7)

where PCi,t is the level of financial institutions depth, FDi,t−1 , for country i at time t.

Also, for the quadratic specification case, we employ the following:18

2
Growthi,t = α + β1 PCi,t−1 + β2 PCi,t−1 + γ1 LGDPi,t−1 + γ2 LEDUi,t−1 + γ3 LGCi,t−1

+ γ4 LOPENi,t−1 + γ5 LINFLi,t−1 + µi + ηt + εi,t . (1.8)

Financial Institutions Depth and Growth (controlling for Financial Institutions

Efficiency (1.9)]): extending the results from non-linear impacts of the finance-growth

nexus, we provide preliminary evidence of the effect of efficiency on the finance-growth

17 This approach is standard in the finance-growth nexus literature.
18 where PCi,t is the level of financial institutions depth, FDi,t−1 , for country i at time t.

19
nexus in 1.9 and the coefficients of interest are β1 and β2 .

Depth
Eff
Growthi,t = α + β1 FIi,t−1 + β2 FIi,t−1 + γ1 LGDPi,t−1 + γ2 LEDUi,t−1 + γ3 LGCi,t−1

+ γ4 LOPENi,t−1 + γ5 LINFLi,t−1 + µi + ηt + εi,t . (1.9)

Interaction Between Measures of Financial Institutions Depth and Efficiency:

 
Depth Eff Depth Eff
Growthi,t = α + β1 FIi,t−1 + β2 FIi,t−1 + β3 FIi,t−1 × FIi,t−1

+ γ2 LEDUi,t−1 + γ3 LGCi,t−1 + γ4 LOPENi,t−1 + γ5 LINFLi,t−1

+ µi + ηt + εi,t . (1.10)

We will need to interpret the results including the interaction term and there are two
Depth
alternatives focusing on: financial institutions depth, FIit , and financial institution

efficiency, FIitEff .19

The partial effect20 of financial institutions depth on growth:

∂E(Growthi,t |Xi,t−1 ) Eff

Depth
= β1 + β3 FIi,t−1 . (1.11)
∂FIi,t−1
19 Rewriting (1.10) above, we have as follows:
 
Eff Depth Eff
Growthi,t = α + β1 + β3 × FIi,t−1 FIi,t−1 + β2 FIi,t−1 + γ1 LGDPi,t−1
+ γ2 LEDUi,t−1 + γ3 LGCi,t−1 + γ4 LOPENi,t−1 + γ5 LINFLi,t−1
+ µi + ηt + εi,t .

20 See Wooldridge (2010), page 67.

20
 Eff , LGDP
where other control variables, Xi,t−1 , are FIi,t−1 i,t−1 , LEDUi,t−1 , LGCi,t−1 ,

LOPENi,t−1 and LINFLi,t−1 .

Similarly, for financial institutions efficiency, we have the following estimation with

respective partial effect:

 
Depth Depth Eff
Growthi,t = α + β1 FIi,t−1 + β2 + β3 × FIi,t−1 FIi,t−1 + γ1 LGDPi,t−1

+ γ2 LEDUi,t−1 + γ3 LGCi,t−1 + γ4 LOPENi,t−1 + γ5 LINFLi,t−1

+ µi + ηt + εi,t .

∂E(Growthi,t |Xi,t−1 ) Depth

Eff
= β2 + β3 FIi,t−1 . (1.12)
∂FIi,t−1

Depth
where other control variables, Xi,t−1 , are FIi,t−1 , LGDPi,t−1 , LEDUi,t−1 , LGCi,t−1 ,

LOPENi,t−1 and LINFLi,t−1 .

1.6 Results

1.6.1 Estimation Model: Linear

1.6.1.1 Financial Institutions Depth and Efficiency

We will need to interpret the results including the interaction term and there are two
Depth
alternatives focusing on: financial depth, FIit , and financial efficiency, FIitEff . In the

21
analysis, we will focus on private domestic credit, PCit , for financial depth and bank

lending-deposits spreads, LDSPREADit , for financial (institutions) efficiency.

Depth
Financial Institutions Depth: FIi,t−1

Suppose we fix financial institutions efficiency, FIitEff , to its mean, FI Eff .21 Then,

we have the following expression:

 
Depth
Growthi,t = α + β1 + β3 × FI Eff FIi,t−1 + β2 FI Eff + γ1 LGDPi,t−1

+ γ2 LEDUi,t−1 + γ3 LGCi,t−1 + γ4 LOPENi,t−1 + γ5 LINFLi,t−1

+ µi + ηt + εi,t .

The impact of financial institutions depth could either increase or reduce growth

depending on β1 , β3 and FI Eff . The interaction term could also indicate whether finan-

cial institutions depth and efficiency are complement. In particular, from the dynamic
Depth
panel regression (Table 1.5 (all lags) below), the coefficient on FIitEff and FIit (spec-

ification 4) are β1 = −3.871 and β2 = −29.72 respectively and the coefficient on the

interaction terms is β3 = 47.93. Given the positive interaction term (β3 > 0), the result

indicates that financial institutions depth, PCi,t−1 , and efficiency, LDSPREADi,t−1 , are
21 Plugging FI Eff to the following equation:
 
Eff Depth Eff
Growthi,t = α + β1 + β3 × FIi,t−1 FIi,t−1 + β2 FIi,t−1
+ γ1 LGDPi,t−1 + γ2 LEDUi,t−1 + γ3 LGCi,t−1 + γ4 LOPENi,t−1
+ γ5 LINFLi,t−1 + µi + ηt + εi,t .

22
compliment.

Partial Effect of Financial Institutions Depth: the partial effect22 of financial

institutions depth on growth.

∂E(Growthi,t |Xi,t−1 ) Eff

Depth
= −3.871 + 47.93 × FIi,t−1 . (1.13)
∂FIi,t−1

Eff , LGDP
where other control variables, Xi,t−1 , are FIi,t−1 i,t−1 , LEDUi,t−1 , LGCi,t−1 ,

LOPENi,t−1 and LINFLi,t−1 .

With the interaction term, we are interested in the partial effect in 1.13 and in par-

ticular, the estimated coefficient β3 helps determine whether the effect of financial in-

Eff
stitutions depth on growth depends on efficiency. The benchmark case is when FIi,t−1

Eff → 0, and the partial

is arbitrarily small, approaching close to zero (efficient), FIi,t−1

Eff → β . With lower efficiency of financial in-

effect is arbitrarily close to β1 , FIi,t−1 1

Eff , the partial effect is β + β × FI Eff . For financial

stitutions, higher value of FIi,t−1 1 3 i,t−1

Depth
institutions depth, FIi,t−1 , to have growth enhancing role, one possible case is that the

estimated interaction coefficient β3 must be large enough to dominate the estimated

coefficient on depth β1 . Alternatively, with smaller estimated interaction coefficient β3 ,

Eff , must be large enough to offset

the threshold for financial institutions efficiency, FIi,t−1

coefficient for depth β1 for depth to have positive net effect on growth. The sign of

estimated coefficients seem to suggested that financial institutions depth and efficiency
22 See Wooldridge (2010), page 67.

23
are complement. The key coefficient of interest is β3 .23
Depth
For example, suppose that FI Eff is 5%, then the partial effect of FIi,t−1 on Growth:

∂E(Growthi,t |Xi,t−1 ) Eff

Depth
= β1 + β3 × FIi,t−1 .
∂FIi,t−1

= −3.871 + 47.93 × 5%.

= −1.47.

Hypothetically, in the benchmark case without interaction, we implicitly assumed

efficient financial institutions and for example, the bank lending and deposit spread is
∂E(Growthi,t |Xi,t−1 )
zero. The partial effect is then given by Depth = β1 . For the case with the
∂FIi,t−1

interaction term, we allow for the lending-deposit spread to be positive and the partial
∂E(Growthi,t |Xi,t−1 ) Eff .
effect is Depth = β1 + β3 × FIi,t−1
∂FIi,t−1
Eff
Financial Institutions Efficiency: FIi,t−1

From the estimation equation with interaction term between financial institutions
Depth
depth and efficiency (equation 1.10 above), we fix financial institutions depth, FIit ,

to its mean and we have the following effect of financial institutions efficiency on
23 Intuition: β3 > 0 : Larger impacts of financial institution depths on growth for less efficient financial
institution, i.e. larger lending-deposit spread. In particular, the results seem to suggest that for countries
with lower efficiency, financial institution efficiency improves the impact of financial institution depth
on growth.

24
growth.24

 
Eff
Growthi,t = α + β1 FI Depth + β2 + β3 × FI Depth FIi,t−1

+ γ1 LGDPi,t−1 + γ2 LEDUi,t−1 + γ3 LGCi,t−1 + γ4 LOPENi,t−1

+ γ5 LINFLi,t−1 + µi + ηt + εi,t .

Partial Effect of Financial Institutions Efficiency: similar to financial institutions

depth, the partial effect of efficiency on growth:

∂E(Growthi,t |Xi,t−1 ) Depth

Eff
= −29.72 + 47.93 × FIi,t−1 . (1.14)
∂FIi,t−1

Depth
where other control variables, Xi,t−1 , are FIi,t−1 , LGDPi,t−1 , LEDUi,t−1 , LGCi,t−1 ,

LOPENi,t−1 and LINFLi,t−1 .

As a benchmark, Table 1.3 provides panel OLS estimates of equation 1.1 using

financial depth, proxy with private credit measure, across four specifications covering

different time span: (1) for time span between 1960 − 1995, (2) for time span between

1960 − 2000, (3) for time span between 1960 − 2005 and (4) for time span between
24 Rewriting the estimation equation with interaction terms between financial institutions depth and
efficiency (equation 1.10 above):
 
Depth Depth Eff
Growthi,t = α + β1 FIi,t−1 + β2 + β3 × FIi,t−1 FIi,t−1
+ γ1 LGDPi,t−1 + γ2 LEDUi,t−1 + γ3 LGCi,t−1 + γ4 LOPENi,t−1
+ γ5 LINFLi,t−1 + µi + ηt + εi,t .

25
1960 − 2010. The key coefficient for financial depth is positive and significant for

specifications that exclude the recent years. Table 1.3 shows results across two different
Depth
data sources of private credit (financial institutions depth, FIit ).

Interactions (Table 1.3)

Controlling for efficiency, the estimated coefficient for private depth (financial in-

stitution depth) from Table 1.3 becomes negative and significant for specification in-

cluding later years. Also, the coefficient for the interaction term is positive.

26
 Table 1.3: Panel OLS

With Interaction Without Interaction

(1) (2) (3) (4) (1) (2) (3) (4)
LGDPt−1 -0.384∗ -0.363∗∗ -0.465∗∗∗ -0.411∗∗∗ -0.176 -0.143 -0.199 -0.178
(-1.76) (-2.08) (-3.10) (-3.12) (-1.09) (-1.04) (-1.62) (-1.63)
LEDUt−1 1.406∗∗∗ 1.732∗∗∗ 2.073∗∗∗ 1.842∗∗∗ 0.663∗∗ 0.838∗∗∗ 1.157∗∗∗ 1.166∗∗∗
(3.05) (4.47) (6.09) (5.97) (2.18) (3.12) (4.67) (5.11)
LGCt−1 -1.644∗∗∗ -1.084∗∗∗ -1.170∗∗∗ -1.235∗∗∗ -0.892∗∗ -0.768∗∗ -0.922∗∗∗ -1.023∗∗∗
(-3.34) (-2.72) (-3.32) (-3.90) (-2.38) (-2.39) (-3.14) (-3.88)
LOPENt−1 0.875∗∗ 0.545∗ 0.544∗∗ 0.497∗∗ 0.589∗∗ 0.519∗∗ 0.560∗∗∗ 0.536∗∗∗
(2.46) (1.91) (2.20) (2.24) (2.53) (2.55) (3.02) (3.17)
LINFLt−1 -0.154 -0.214∗ -0.284∗∗ -0.261∗∗ -0.255∗∗ -0.276∗∗∗ -0.357∗∗∗ -0.350∗∗∗
(-1.02) (-1.65) (-2.41) (-2.39) (-2.28) (-2.78) (-3.88) (-4.04)

27
PCt−1 -0.0264 -0.983 -1.482∗∗ -1.618∗∗∗ 2.163∗∗∗ 1.169∗∗ 0.403 -0.241
(-0.03) (-1.39) (-2.53) (-3.29) (3.29) (2.26) (0.92) (-0.68)
LDSPREADt−1 -0.230∗∗∗ -0.188∗∗∗ -0.180∗∗∗ -0.168∗∗∗
(-5.92) (-6.01) (-6.32) (-6.40)
PCt−1 × LDSPREADt−1 0.453∗∗∗ 0.355∗∗∗ 0.324∗∗∗ 0.323∗∗∗
(3.74) (3.75) (3.79) (4.17)
Constant 2.091 3.019∗ 4.705∗∗∗ 3.824∗∗∗ 0.0902 2.815∗∗ 3.189∗∗∗ 1.864∗
(1.08) (1.96) (3.27) (3.01) (0.07) (2.49) (3.04) (1.89)
Time Fixed Effect X X X X X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010 1960-1995 1960-2000 1960-2005 1960-2010
Observations 273 377 484 576 542 664 793 917
Dependent Variable: GDP % Growth 5 Year
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01
1.6.1.2 Dynamic Panel: System GMM

Private Credit [Arcand et al. (2015a) and World Bank GFDD] (Table A.8)

Table A.8 estimate a linear model with system-GMM for both measures of financial

institutions depth (Arcand et al. (2015a) and World Bank Global Financial Develop-

ment Database).
Table 1.4: System GMM

(1) (2) (3) (4)

LGDPt−1 -0.378 -0.465 -0.360 -0.0276
(-0.81) (-1.25) (-1.05) (-0.06)
LEDUt−1 0.386 1.280∗ 1.757∗∗ 1.589∗∗
(0.55) (1.78) (2.40) (2.49)
LGCt−1 -2.537∗∗ -2.377∗∗∗ -2.650∗∗∗ -2.626∗∗∗
(-2.43) (-2.78) (-3.39) (-3.34)
LOPENt−1 0.483 0.720 1.003 1.602∗∗
(0.58) (1.02) (1.62) (2.42)
LINFLt−1 -0.245 -0.377∗ -0.701∗∗∗ -0.644∗∗∗
(-0.95) (-1.80) (-3.15) (-3.14)
PCt−1 4.272∗∗∗ 2.542∗∗ 0.477 -1.102
(2.58) (2.23) (0.57) (-1.13)
Constant 9.100∗∗ 7.476∗∗ 6.308∗ 0.157
(2.17) (2.11) (1.81) (0.04)
Time Fixed Effect X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010
Observations 542 664 793 917
Sarganp 0.000000134 3.70e-09 6.20e-17 1.29e-18
Hansenp 0.839 0.995 1.000 1.000
AR1p 0.000349 0.0000204 0.0000288 0.00000242
AR2p 0.978 0.432 0.619 0.00819
Instruments 127 170 219 274
Dependent Variable: GDP % Growth 5 Year
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

The results for the estimation with interaction between financial depth and financial

28
efficiency are included in Table 1.5 (two and all lags included). Further results with

different lags are included in the appendix: 2 lags (Table A.4), 3 lags (Table A.5) and 4

lags (Table A.6).

29
 Table 1.5: System GMM (With Interaction)

1st and 2nd Lags All Lags

(1) (2) (3) (4) (1) (2) (3) (4)
LGDPt−1 -0.418 -0.561 -0.227 -0.609 -0.329 -0.140 -0.335 -0.116
(-0.68) (-0.93) (-0.36) (-1.19) (-0.73) (-0.33) (-1.10) (-0.32)
LEDUt−1 0.574 1.737 1.483 2.036∗∗ 0.638 1.921∗∗ 2.900∗∗∗ 2.190∗∗∗
(0.45) (1.36) (1.14) (2.03) (0.71) (2.25) (3.99) (3.21)
LGCt−1 -2.445∗ -1.670 -1.599 -1.361 -2.048∗ -2.017∗∗ -2.033∗∗ -1.934∗∗
(-1.85) (-1.36) (-1.41) (-1.36) (-1.66) (-2.35) (-2.52) (-2.48)
LOPENt−1 2.508∗∗ 2.212 1.665 1.994∗ 1.793∗ 0.781 0.859 1.309∗∗
(2.00) (1.14) (1.27) (1.86) (1.96) (1.16) (1.16) (2.20)
LINFLt−1 0.168 0.209 -0.00332 0.0541 0.0545 0.00579 -0.147 -0.252
(0.62) (0.86) (-0.01) (0.25) (0.28) (0.03) (-0.80) (-1.34)
PCt−1 1.106 -1.160 -2.972 -1.605 -0.151 -2.498 -3.943∗∗∗ -3.871∗∗∗
(0.27) (-0.40) (-1.34) (-1.01) (-0.07) (-1.46) (-3.03) (-4.55)

30
LDSPREADt−1 -37.71∗∗ -37.47∗∗ -33.79∗∗ -29.02∗∗∗ -35.70∗∗ -35.99∗∗∗ -37.48∗∗∗ -29.72∗∗∗
(-2.30) (-2.24) (-2.25) (-3.13) (-2.37) (-2.83) (-4.01) (-4.45)
PCt−1 × LDSPREADt−1 49.17 43.15 42.31 27.04 72.10∗∗ 57.38∗∗ 58.35∗∗∗ 47.93∗∗∗
(0.75) (0.96) (1.02) (0.82) (2.07) (2.04) (2.77) (3.25)
Constant 0.0888 -0.707 0.738 -1.257 0 0 0 0.410
(0.02) (-0.10) (0.15) (-0.26) (.) (.) (.) (0.16)
Time Fixed Effect X X X X X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010 1960-1995 1960-2000 1960-2005 1960-2010
Observations 273 377 484 576 273 377 484 576
Sarganp 0.000203 0.0000405 6.11e-08 1.13e-09 0.00000104 0.0000744 0.000000146 8.00e-09
Hansenp 0.219 0.272 0.284 0.381 0.888 0.976 1.000 1
AR1p 0.0145 0.00574 0.00250 0.000175 0.00995 0.00417 0.00223 0.000143
AR2p 0.834 0.260 0.681 0.00862 0.992 0.150 0.552 0.00304
Instruments 48 63 78 93 105 152 206 267
Dependent Variable: GDP % Growth 5 Year
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01
1.6.2 Estimation Model: Quadratic

1.6.2.1 Financial Depth and Efficiency

The standard benchmark panel OLS estimation for quadratic specification is in Ta-

ble 1.6. The results are for two dimensions of financial development separately, finan-

cial institutions depth, PCit , and financial institutions efficiency, LDSPREADit . The

results of system GMM estimation for quadratic case are in Table 1.7. Similarly, the

results include both dimensions of financial development separately.

31
 Table 1.6: Panel OLS: Quadratic

Private Credit Bank Lending-Deposits Spreads

(1) (2) (3) (4) (1) (2) (3) (4)
LGDPt−1 -0.248 -0.237∗ -0.293∗∗ -0.289∗∗∗ -0.164 -0.280∗ -0.436∗∗∗ -0.405∗∗∗
(-1.51) (-1.71) (-2.35) (-2.63) (-0.85) (-1.81) (-3.27) (-3.36)
LEDUt−1 0.632∗∗ 0.804∗∗∗ 1.109∗∗∗ 1.081∗∗∗ 1.562∗∗∗ 1.802∗∗∗ 2.077∗∗∗ 1.874∗∗∗
(2.09) (3.01) (4.50) (4.77) (3.38) (4.70) (6.15) (6.06)
LGCt−1 -0.993∗∗∗ -0.900∗∗∗ -1.033∗∗∗ -1.082∗∗∗ -1.676∗∗∗ -1.287∗∗∗ -1.344∗∗∗ -1.352∗∗∗
(-2.65) (-2.80) (-3.52) (-4.14) (-3.43) (-3.24) (-3.84) (-4.25)
LOPENt−1 0.499∗∗ 0.437∗∗ 0.465∗∗ 0.436∗∗ 0.539 0.311 0.371 0.315
(2.13) (2.15) (2.50) (2.58) (1.44) (1.04) (1.45) (1.37)
LINFLt−1 -0.246∗∗ -0.254∗∗ -0.329∗∗∗ -0.301∗∗∗ -0.125 -0.163 -0.214∗ -0.196∗
(-2.21) (-2.58) (-3.58) (-3.49) (-0.82) (-1.27) (-1.84) (-1.80)
PCt−1 5.736∗∗∗ 4.948∗∗∗ 4.041∗∗∗ 3.457∗∗∗

32
(3.60) (3.91) (3.66) (3.95)
2
PCt−1 -2.860∗∗ -2.750∗∗∗ -2.474∗∗∗ -2.195∗∗∗
(-2.46) (-3.27) (-3.59) (-4.60)
LDSPREADt−1 -0.307∗∗∗ -0.232∗∗∗ -0.192∗∗∗ -0.141∗∗∗
(-3.69) (-3.61) (-3.44) (-2.78)
2
LDSPREADt−1 0.00365∗∗ 0.00273∗∗ 0.00205∗ 0.00121
(2.08) (1.97) (1.65) (1.04)
Constant 0.600 3.544∗∗∗ 3.907∗∗∗ 2.674∗∗∗ 4.621∗∗ 4.045∗∗ 5.213∗∗∗ 4.400∗∗∗
(0.44) (3.10) (3.69) (2.70) (2.42) (2.53) (3.61) (3.30)
Time Fixed Effect X X X X X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010 1960-1995 1960-2000 1960-2005 1960-2010
Observations 542 664 793 917 306 414 523 616
Dependent Variable: GDP % Growth 5 Year
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01
 Table 1.7: System GMM: Quadratic

Private Credit Bank Lending-Deposits Spreads

(1) (2) (3) (4) (1) (2) (3) (4)
LGDPt−1 -0.348 -0.419 -0.378 -0.235 0.571 0.0326 -0.454 -0.366
(-0.84) (-1.12) (-1.05) (-0.61) (1.07) (0.07) (-1.31) (-1.42)
LEDUt−1 0.446 1.133 1.415∗∗ 1.376∗∗ 0.0289 1.709∗ 2.898∗∗∗ 2.426∗∗∗
(0.64) (1.62) (2.16) (1.97) (0.03) (1.86) (4.19) (3.52)
LGCt−1 -2.429∗∗∗ -2.146∗∗∗ -2.451∗∗∗ -2.526∗∗∗ -2.859∗∗ -2.269∗∗∗ -2.072∗∗ -2.268∗∗∗
(-2.60) (-2.70) (-3.45) (-3.52) (-2.54) (-2.86) (-2.39) (-3.18)
LOPENt−1 0.400 0.613 0.753 0.815 1.857 0.874 0.300 0.750
(0.56) (0.97) (1.42) (1.64) (0.94) (0.76) (0.33) (0.86)
LINFLt−1 -0.379 -0.406∗∗ -0.628∗∗∗ -0.585∗∗∗ -0.240 -0.104 -0.311 -0.408∗∗
(-1.59) (-2.20) (-3.62) (-3.50) (-1.01) (-0.47) (-1.53) (-2.08)
PCt−1 5.794∗∗ 4.259∗∗ 4.177∗∗ 3.289∗
(2.02) (2.00) (2.23) (1.90)

33
2
PCt−1 -2.320 -1.819 -2.586∗∗ -2.231∗∗∗
(-1.34) (-1.41) (-2.45) (-3.01)
LDSPREADt−1 -0.383∗∗ -0.277∗ -0.177 -0.172∗
(-2.10) (-1.92) (-1.44) (-1.69)
2
LDSPREADt−1 0.00760 0.00287 0.000340 0.00217
(1.64) (0.76) (0.10) (0.84)
Constant 8.812∗∗ 7.078∗∗ 6.803∗∗ 4.325 0.340 3.007 5.019 3.683
(2.23) (2.07) (2.11) (1.40) (0.04) (0.54) (1.24) (0.97)
Time Fixed Effect X X X X X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010 1960-1995 1960-2000 1960-2005 1960-2010
Observations 542 664 793 917 306 414 523 616
Sarganp 0.000000483 1.41e-08 3.16e-14 7.81e-15 2.19e-09 0.000000213 4.90e-11 3.14e-13
Hansenp 0.988 1.000 1.000 1 0.366 0.903 1.000 1.000
AR1p 0.000300 0.0000175 0.0000298 0.00000181 0.0114 0.000138 0.000287 0.00000322
AR2p 0.961 0.416 0.635 0.00748 0.778 0.252 0.584 0.00385
Instruments 147 197 254 318 94 138 189 247
Dependent Variable: GDP % Growth 5 Year
t statistics in parentheses; ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01
1.7 Conclusion

The main results suggest that financial institutions depth, using measures of domes-

tic private credit as share of GDP (PCit ), and financial institutions efficiency, proxy as

banks lending-deposits spreads (LDSPREADit ), are compliments. Although the system

GMM results are sensitive to the choice of lag lengths we believe that this provides an

avenue for further analysis of the multi-dimensional view of financial development in

the finance and growth literature.

1.8 Extensions

For future work, the multi-dimensional view of financial development, ”holistic”

view, can be extended into several directions. One potential avenue is analyzing the

impacts of different types of international capital flows on the finance-growth nexus

with multi-dimensional view. Another possibility is incorporating many other ”pil-

lars”/dimensions of financial development besides those employed in this paper, (see

Table 1.2) and we have been working on this direction with results for measures on

financial markets. In addition, it would be interesting to analyze whether institutional

quality plays an underlying role in affecting economic growth through financial in-

stitutions depth or efficiency or both.25 Moreover, there is a large literature studying

25 We are currently exploring this avenue using the International Country Risk Guide (ICRG) data for
constructing a panel of institutional quality across countries.

34
the impacts of financial openness (liberalization) and financial development on growth

showing mixed results. However, there are few that look at financial development from

multi-dimensional view.

1.9 Appendix

The appendix for this chapter is provided at the end of the dissertation (Appendix

A).

35
Chapter 2

Heterogeneous Patterns of Financial

Development: Implications for Asian

Financial Integration

Written with second coauthor Nirvikar Singh, University of California, Santa Cruz,

(contact)

Published in the International Economic Journal (Bun and Singh (2016))

2.1 Introduction

The rapid post-war rise of Japan to developed country status, followed by economies

such as South Korea, Taiwan, Hong Kong and Singapore, began a process of making

36
East (and Southeast) Asia a significant contributor to global economic activity. Even

as Japanese growth slowed, the regional process has been accelerated by the growth

of China, and the creation of regional production networks that include many smaller

economies such as Malaysia and Vietnam in essential ways. These production net-

works have been an important aspect of openness to trade that characterized much of

the region and contributed to economic growth.1

Strong real growth has not been immune to normal business cycle fluctuations, as

well as to the negative impacts of international financial crises. The Asian financial

crisis of 1997-98 called attention to the differences between openness to trade and

openness to capital, and produced a particular set of policy responses in the region,

including ”self-insurance” through international reserve accumulation.2 The global fi-

nancial crisis of 2007-09 had an unavoidable negative impact on East Asia as well, but

it is arguable that the lessons of 1997-98 permitted the region to be better prepared for

this second, larger shock.

One of the issues brought to the fore by the global crisis was the benefits and costs

of financial openness and financial integration. Openness and integration can cover

several different aspects of economic interaction. The obvious example of financial

openness is liberalization of capital flows between countries. Integration has several

1 On regional production networks, trade, and growth, see Athukorala (2014) and Kaur (2014), as two
examples of a large and growing literature.
2 See Aizenman et al. (2013a) as one of several articles by those authors that document and analyze

this reserve accumulation.

37
connotations beyond openness, including harmonization of institutions such as finan-

cial market trading rules, and policy coordination for managing risks and instabilities

that are potentially associated with financial openness. The theoretical justification for

financial openness has been questioned much more than in the case of openness to

trade, but it remains a reality that has to be analyzed and managed. The last decade,

before and after the global financial crisis, has seen continued academic and policy

engagement with these questions.

A key complement to financial integration is the nature of domestic financial devel-

opment. It has been argued that financial openness and integration are more likely to

have positive outcomes in cases where the economies involved have adequate levels of

financial development.3 However, financial development is itself a multi-dimensional

concept. For example it can include various kinds of financial market institutions such

as banks and stock markets, but also broader legal and regulatory frameworks that cre-

ate the environment for financial decision-making by firms and households. This paper

is motivated by debates about financial integration and policy coordination in East Asia,

but focuses on financial development as a precondition for financial integration. In par-

ticular, it analyzes differences in patterns of financial development across the major

East Asian economies, including three of the largest economies of the region (China,

Japan and South Korea). In addition to differences in levels of financial development,

differences in patterns of financial development may present an economic challenge to

3 Arguments for this view are taken up in the next section, where references are provided.

38
regional financial integration efforts.

The next section briefly reviews arguments for and against financial openness, the

East Asian experience with financial crises and with financial integration, and how

economies in the region have responded to external shocks. It also considers the pos-

sible connection between financial integration and financial development. Section 3

discusses data and measurement of patterns of financial development, as well as overall

levels. A key idea here is that differences in various dimensions of financial develop-

ment can be relevant for financial integration, as well as differences in overall levels of

financial development. Section 4 presents the results of the paper, analyzing levels and

patterns for 14 economies in the region. A key finding is that comparisons of overall

levels of financial development convey somewhat different information than compar-

isons of patterns of financial development. This is borne out in a cluster analysis, which

also provides some direct insights on possibilities for future financial integration. Sec-

tion 5 concludes the paper, offering some possible implications of the analysis for future

thinking about financial integration in the region.

2.2 Financial Integration and Financial Development

The core aspect of financial integration is openness on the capital account, so that in-

ternational capital flows are unrestricted. In practice, a completely open capital account

does not lead to perfect financial integration, in the sense of unified financial markets.

39
An important reason for this imperfection is home bias in investment, reflected in a

positive correlation between domestic savings and investment Feldstein and Horioka

(1980). Investors do not view foreign and domestic assets as perfect substitutes, even

when they have the same objective characteristics. This can be due to regulatory differ-

ences, tax treatment, asymmetries of information, and other kinds of market or institu-

tional imperfections. Nevertheless, despite these perceptual and institutional barriers to

complete financial integration, the main defining feature of modern-day globalization

has been liberalization of restrictions on capital flows, allowing large amounts of cap-

ital to move swiftly between different countries. Furthermore, these capital flows now

mainly consist of private capital, rather than official government flows. Large flows

of capital, whether inward or outward, create challenges for the conduct of domestic

macroeconomic policies,4 and these are compounded by the volatility of these flows.

The post-World War II global economy was initially one of fixed exchange rates,

capital controls and monetary policy autonomy. This regime broke down in the 1970s,

and since then, theory and practice have swung back and forth between different policy
4 The main challenge for macroeconomic policy is encapsulated in the idea of the policy trilemma,
or impossible trinity, based on the Mundell-Fleming model of an open economy macroeconomic frame-
work. In the model, it is impossible for a government to simultaneously have monetary policy autonomy
(and hence the ability to control the domestic inflation rate) and a fixed exchange rate when the capital
account is completely unrestricted. Attempts to conduct an independent monetary policy will drive a
wedge between foreign and domestic interest rates, leading to continued capital inflows or outflows (de-
pending on the direction of the interest differential) in the absence of an equilibrating mechanism such as
exchange rate adjustment. Rey (2013) has advanced the proposition that the weight of capital flows in the
context of non-conventional monetary policies (essentially, what is known as QE or Quantitative Easing)
makes exchange rate flexibility insufficient for domestic inflation control unless there are also controls
on international capital flows. The theory and empirics of this view are still being debated. Aizenman
et al. (2013a) have been among the originators of quantitative approaches to measuring policy stances
with respect to the trilemma.

40
combinations. At one stage, the orthodoxy had coalesced on the desirability of flexible

exchange rates and openness of the capital account, the idea being that markets would

equilibrate to allocate resources efficiently around the globe. Few countries adopted this

policy mix, however, instead pursuing various combinations of partial capital controls,

partial exchange rate flexibility and partial monetary autonomy. The latest financial

crisis finally pushed the weight of expert opinion away from full capital account open-

ness.5 One of the most striking examples of this change was the near reversal of the

International Monetary Funds position on capital account liberalization, after the global

financial crisis.

Even before the global financial crisis of 2007-09, there was evidence that full cap-

ital account openness did not have identifiable positive effects on economic perfor-

mance. While the earlier 1997-98 financial crisis was still unfolding, Rodrik (1998)

argued against full capital account convertibility, pointing out that financial markets are

far from the textbook model of perfection, and subject to bubbles, panics and herd be-

havior in general, so that the theoretical case for capital account openness is difficult

to make convincingly.6 Looking at empirical evidence, Rodrik concluded, There is no

evidence in the data that countries without capital controls have grown faster, invested

more, or experienced lower inflation.7

5 For an academic statement of the changed thinking, see Ostry et al. (2012). Ghosh et al. (2008)
provide an earlier, more policy-focused take with the same perspective, also from the IMF.
6 Other prominent arguments against full capital account liberalization include Bhagwati (1998),

Cooper (1999), Stiglitz (2003), and Obstfeld (2009).

7 Kaur and Singh (2014) provide a detailed review of the empirical evidence on how different East

41
 More recently, Obstfeld (2009) offered a similarly cautious assessment, after an ex-

tensive literature review, Financial openness is not a panacea and it could be poison.

The empirical record suggests that its benefits are most likely to be realized when im-

plemented in a phased manner, when external balances and reserve positions are strong,

and when complementing a range of domestic policies and reforms to enhance stability

and growth.8 Addressing the obverse of the issue, Aizenman et al. (2007) found evi-

dence that domestic financial development might be more important for higher growth

than foreign capital. They constructed a self-financing measure, which was positively

correlated with growth, after controlling for the quality of domestic institutions. Even

earlier, Eichengreen (2003) had emphasized the importance of domestic financial devel-

opment, making that case in the context of criticizing the Chiang Mai Initiative (CMI)

designed in 2000 to provide regional swap lines as a means of preserving fixed ex-

change rates among the CMI group.

Despite the cautions emerging from work such as discussed above, financial integra-

tion, especially in East Asia as an economically dynamic region of growing importance,

has continued to receive considerable attention. Borensztein and Loungani (2011) used

cross-border equity and bond holdings, as well as equity returns and interest rates,

to argue that Asian (chiefly East Asian) financial integration had increased, but that
Asian economies reacted to the financial crises of 1997-98 and 2007-09, and the impacts of different
policy mixes with respect tot financial openness.
8 There is also evidence that the specific nature of capital flows matters. For example, equity flows

have a positive short-run impact on the host economy Henry (2007); Kose et al. (2009), as does foreign
direct investment Kose et al. (2009).

42
extra-regional connections remained stronger than intra-regional measures of integra-

tion. These results echoed earlier, similar studies Fung et al. (2008) and Garcı́a-Herrero

et al. (2008), with the latter paper providing an explanation of limited Asian financial

integration in terms of low liquidity in the regions financial markets. Another recent

study Lee et al. (2013) found no evidence for increased Asian financial integration after

the global crisis, but an up-to-date survey (Financial Services Institute of Australasia,

2015) suggested that regional financial integration has increased. This is also the con-

clusion of a recent IMF study (International Monetary Funds (IMF) (2015)), which

measures this increase in terms of intraregional financial flows, but nevertheless also

concludes that Home biasis particularly strong in Asia, limiting cross-border financial

transactions within the region. (p. 94)

Aside from measuring trends in financial integration, various studies have also tried

to estimate the potential costs and benefits. Hoxha et al. (2009) and Hoxha et al. (2013)

analyzed international financial integration, measured as foreign capital flows, and esti-

mated that these flows had had significant positive impacts on consumption and welfare.

Unfortunately, this analysis was performed before the global financial crisis. From a

somewhat different welfare perspective, Pongsaparn and Unteroberdoerster (2011) es-

timated that greater financial integration in Asia would support global rebalancing, and

hence financial stability. Park and Lee (2011) also implicitly assumed benefits from

greater financial integration in emerging Asia, through greater allocative efficiency, but

43
emphasized the need for more effective financial supervisory and regulatory mecha-

nisms, as well as improvements in various dimensions of financial development.

There is a large empirical literature on the impacts of financial development, with

the recent focus, post-global-crisis, being on the possibility that too much finance is in-

imical to economic growth. A common approach in this literature is to use a quadratic

term for the measure of financial development: Arcand et al. (2015a) provide the most

recent example of this specification. On the other hand, Law and Singh (2014) used an

endogenous threshold model to allow for negative effects of financial development on

growth.9 The typical conceptualization of financial development is in terms of finan-

cial depth, measured as a credit-to-GDP ratio, and the presumed channel of negative

impacts of financial depth on growth is volatility or financial crises Schularick and Tay-

lor (2012).10 Arcand et al. (2015a) used the ratio of total private sector credit to GDP

as the measure of financial depth, but their results are robust to using bank credit or

household credit instead. Law and Singh (2014) used private sector credit, liquid lia-

bilities and domestic credit as three possible measures of financial depth, with similar

results.

Law et al. (2013) examined the possibility that the finance-growth nexus depends

on institutional quality, and find that this is the case, using a threshold model. The insti-
9 Arcand et al. (2015a) also test a piecewise linear model with exogenous thresholds of financial depth.
10 Law and Singh (2014) survey other possible explanations for the negative impact of financial depth
on growth. Cline (2015) argues that all these results are merely a statistical artifact, and claims that the
results are explained by slower growth at higher per capita income levels, but this claim seems to have
been answered effectively by Arcand et al. (2015b).

44
tutional quality measures are generic capturing control of corruption, rule of law and

government effectiveness and not specific to the financial sector. Earlier, Demetriades

and Hook Law (2006) had explored a similar connection between financial depth and

institutional quality, using a specification with interactions, and a similar exercise is

carried out by Arcand et al. (2015a).

Most relevant to the current context, Herwartz and Walle (2014a) examined the im-

pacts of trade openness and financial openness on the finance-growth linkage. They

found that greater trade openness strengthened this linkage, whereas greater financial

openness eroded it. Their result is interpreted as providing a caution to the Rajan

and Zingales (2003) analysis that openness would promote financial development by

overcoming the resistance of domestic interest groups, since the financial development

achieved would come at the cost of a weaker finance-growth relationship. As in most

other studies, Herwartz and Walle (2014a) measure financial development as financial

depth, specifically, the private-sector credit-to-GDP ratio.

Outside this large literature on financial integration and financial development, and

their economic impacts, there has been some attention paid to the issue that financial

development is broader than just financial depth. From 2009-12, the World Economic

Forum (WEF) produced a Financial Development Index with seven components, as the

basis for its annual Financial Development Reports (FDRs), and the approach in that

exercise will be the starting point for the analysis of the current paper. Since 2013, the

45
World Bank has published annual Financial Development Reports, which are quite dif-

ferent from the WEFs documents, being focused on specific themes such as the role of

the State in finance (World Bank (2012)) and financial inclusion (World Bank (2014)).

These FDRs do not have an index of financial development, but they identify and pro-

vide data for four components or dimensions of overall financial development: depth,

access, efficiency and stability.

Adnan (2011), building on the work of Saci and Holden (2008), constructed an

index of financial development based on data for the banking and insurance sectors,

as well as stock and bond markets. Thirteen variables were used, capturing depth and

efficiency, with more in the former category. However, the index itself was derived

using principal components analysis. No further analysis was performed, beyond the

construction of the index.

More recently, Sahay et al. (2015) constructed an index of financial development

that captured financial markets as well as financial institutions, and access and effi-

ciency as well as depth. The authors used six indicators, with weights derived from

principal components analysis, to construct the index. They confirmed the non-linear,

sign-changing relationship between finance and growth, earlier found just for financial

depth, but now extended to this broader measure of financial development. However,

their results suggested that the too much finance result is driven by financial deepening

rather than greater efficiency or access, which is a plausible conclusion.

46
 Another significant recent paper is by Aizenman et al. (2015), who distinguished

between quantity and quality of financial intermediation, measuring the former by fi-

nancial depth (private bank credit to GDP ratio) and the latter by the lending-deposit

interest rate spread. They examined sectoral growth impacts, and found that quantity

and quality each had positive, negative or non-linear impacts, depending on the sector

and region considered. Their results, unlike Sahay et al. (2015), suggested that even

quality might have a non-linear impact in some circumstances.

The approach taken in the current paper also tries to decompose financial develop-

ment into its different aspects, but in a different manner than the Sahay et al. (2015) and

Aizenman et al. (2015) papers. In particular, it tries to isolate differences in patterns of

financial development across countries. The methodology and results are described in

the next two sections.

2.3 Measuring Financial Development: Level and Pat-

terns

The motivation for the analysis in this section and the next is the idea that pos-

sibilities for successful financial integration are a function of similarities in financial

development among the countries that are integrating. This is not the only way to think

about preconditions for successful financial integration. For example, dissimilarities in

47
financial development may also be positive in some circumstances: if one country in

a region has strong stock markets, they may serve the whole region better as financial

integration progresses. Nevertheless, proceeding with the idea of comparing financial

development across countries in a region such as East Asia, which is contemplating or

pursuing greater financial integration, it is straightforward to use a single measure or an

index of financial development as a basis for comparison.

As first suggested in Kaur and Singh (2014), however, comparing levels of financial

development via a single index misses possible differences across countries in the vari-

ous components of the index. One can compare levels of different components as well,

but this just generates additional information on level differences. Kaur and Singh

(2014) suggested a summary measure of differences in patterns of financial develop-

ment. The idea in this case is that financial development is inherently multidimensional,

and comparing a vector of indicators of financial development across countries can tell

one something about how similar or different the patterns of financial development are.

To illustrate, we use some of the calculations in Kaur and Singh (2014). They used

indices constructed from the WEF FDR of 2012. In the WEF methodology, the Finan-

cial Development Index (FDIndex) was constructed from seven underlying indicators,

each of which was itself built up from numerous base measures. The seven dimensions

of financial development in this framework were institutional environment, business

environment, banking financial services, non-banking financial services, financial sta-

48
bility, financial markets and financial access. Each dimension was scaled from 1 to 7,

and the overall index was an average of these seven numbers. Thus, the FDIndex was

5.10 for Singapore, 5.01 for Australia, 4.90 for Japan and 4.42 for South Korea, provid-

ing a numerical scaling as well as a ranking of levels of overall financial development.

Kaur and Singh (2014) proposed to complement this comparison of levels with a

comparison of patterns of financial development, as follows. Each country could be

thought of as characterized by a 7-vector of different aspects of financial development.

Calculating a correlation coefficient between these vectors for two countries could then

be a measure of how similar the two vectors were, and therefore a measure of similarity

in patterns of financial development.11 The correlation coefficient uses deviations from

the mean (the overall FDIndex for that country), so it would not be affected by the

similarity or difference of overall levels. For example, Singapore and Japan had a

correlation coefficient of 0.64, while Australia and Japan, slightly closer in level terms

than Singapore and Japan, had a correlation of 0.54. Korea had a correlation coefficient

of 0.13 with Japan, but a higher correlation of 0.22 with Australia, which was slightly

further away than Japan in level terms. In fact, Korea and China stood out among the 11

countries considered, for having patterns of financial development relatively different

from other countries in the group. This idea of comparing patterns as well as levels is

taken further in the analysis of this paper. The data is described next, followed by a

mathematical statement of the concepts used.

11 The interpretation of this calculation is discussed in more detail later in this section.

49
2.3.1 Data

The WEF did not publish a FDR after 2012. Instead, we use data from the WEF

Global Competitiveness Report, which calculates a Global Competitiveness Index. This

index includes a FDIndex as one of its components,12 but it is quite different from

the one used in the FDR. There are eight components, listed in Table 3.1. As in the

FDR, each component is scaled from 1 to 7. Besides being narrower than the index

constructed in the FDR (something we address in our calculations), the indicators in

Table 3.1 rely heavily on survey evaluations, rather than being constructed primarily

from quantitative measures (e.g., the percentage of households with bank accounts for

measuring availability of financial services). Reported numbers are averages across

a range of respondents. In terms of the Aizenman et al. (2015) dichotomy between

quality and quantity of financial services, all the components in Table 3.1 are primarily

quality indicators, although factors such as availability of financial services could be

viewed as being closer to measures of quantity.

The GCI data is available from 2006-07 to 2014-15, but the earliest year with all

eight of the components that are listed in Table 3.1 is 2010-11. Henceforth, we refer to

years by their first year. Hence, 2010-11 is termed 2010. Since there is no discernible

trend in the data over the period 2010 to 2014, we report calculations only for the first
12 More specifically, the term used is Financial Market Development, and it is one of 12 pillars of the

GCI.

50
 Table 2.1: Components of FDIndex, WEF Global Competitiveness Report.

Category Indicator Score

Affordability of financial services, 1-7 (best)
Availability of financial services, 1-7 (best)
Efficiency Financing through local equity market, 1-7 (best)
Ease of access to loans, 1-7 (best)
Venture capital availability, 1-7 (best)
Soundness of banks, 1-7 (best)
Trustworthiness and Confidence Regulation of securities exchanges, 1-7 (best)
Legal rights index, 1-7 (best)
Source: The World Economic Forum - Global Competitiveness Index

and last years for which there is complete comparable data, 2010 and 2014.

While the index constructed from the eight components listed in Table 3.1 has a

much narrower scope than the FDR index, the GCI database includes several variables

that are the same as, or close to, those used in the FDR index. We therefore constructed

two additional potential components for the FDIndex, capturing institutional environ-

ment and business environment. The underlying variables for these two additional

components are provided in the Appendix. The GCI data did not provide variables to

capture financial stability, or enable distinguishing of banking and non-banking finan-

cial services, so even with the additional two variables, substantial differences remain

from the FDR index, leading to somewhat different results. These differences do not

affect the validity of the methodology they merely highlight the sensitivity of compar-

isons among countries to how financial development is conceived of and measured. We

51
also constructed four other indices, measuring openness, technological sophistication,

business sophistication and other aspects of institutional environment. Calculations in-

volving these four indices or components of financial development are reported in the

Appendix.

Table 2.2: Economies in the Analysis.

Australia New Zealand

Cambodia Philippines
China Singapore
Hong Kong South Korea
Indonesia Taiwan
Japan Thailand
Malaysia Vietnam

Finally, with respect to the data, we focus on 14 economies from the Asia Pacific, or

East Asia region, and compare financial development across these economies. They are

listed in Table C.1, and include developed as well as emerging economies, with seven

economies in each of those two broad categories. In fact, the variation in per capita

GDP among these countries is substantial, with the richest being 15 times as well-off

as the poorest, even at purchasing power parity. The set includes all eight economies

that were analyzed in the East Asian Miracle study of the World Bank (1993), as well

as all 11 economies considered in Kaur and Singh (2014).

52
2.3.2 Methodology

We will denote the value of an indicator n for country i by xin . An indicator here

refers to a potential component of a financial development index (FDIndex), and we

assume that all indicators have already been converted to a common scale (1-7 in the

case of our data). For example, a weighted average of the components will yield an

index defined by
N
xi (w) = ∑ wn xin (1.1)
1

where the weights are assumed to sum to one. The left hand side of this equation

measures the level of financial development for country i, based on the index formula

chosen.

In this case, the Level Distance between country i and country j is simply xi (w) −

x j (w). This number may be positive or negative, of course. When considering regional

financial integration we may be interested in how close a single economy in the region

is to the other economies in the region, in terms of financial development. Note that

we are not measuring integration, but rather financial development as a precursor for

integration. With 14 economies in our set of analysis, we would have 13 bilateral

Level Distance. It is reasonable to summarize this information with an average over

the 13 distance measures. Simply adding up the individual distances and averaging will

cancel out positives and negatives. In order to avoid this, one can use absolute values

53
or squares, prior to summing up. We choose the latter, giving the formula

s
1  2
LDi = ∑ xi (w) − x j (w) (1.2)
13 j6=i

Hence, economies in the region with higher values of this measure are on average

further in levels of financial development from their regional counterparts.

In constructing this measure of average Level Distance, differences in patterns are

irrelevant, since they disappear into the aggregation. Instead, consider the entire vector

of indicators for economy i, before any aggregation. Denote this by xi = (xi1 , . . . , xiN ).

Differences in these vectors across countries can be thought of as capturing differences

in patterns of financial development. Therefore, we define the Pattern Distance be-

tween two economies by

PDi j = corr xi , x j (1.3)

In terms of the concept of distance, a higher correlation denotes a lower pattern dis-

tance.13 Hence, if two economies have a correlation of 1 in their vector measures of

financial development, their pattern difference would be the lowest possible. This is

incorporated into our formula by subtracting the correlation from 1: the pattern dis-

tance measure therefore ranges from 0 to 2. Note that using the correlation coefficient

removes the simple mean of the components of the vector. If the level index is con-
13 Implicitly, this calculation treats all the components of the index as having equal weights. This is
not an issue in our calculations, since the index we use weights components equally.

54
structed with equal weights, the index is the simple average of the components of the

vector of financial development measures. If we average across all such correlations for

country i, we obtain the average Pattern Distance for country i. Note that the correlation

coefficient does not have the interpretation usually associated with random variables it

is simply a convenient summary measure.14 For this calculation to make sense, all the

components should be on the same scale, which is the case with the GCI data. Also, the

calculation here can be contrasted with the correlations which are part of principal com-

ponents analysis, since those are calculated between vectors of individual components,

so that each countrys value for that component is an element of the ith vector of country

values. Therefore, differences in patterns of financial development as calculated here

are capturing something quite different than anything associated with principal compo-

nents.

2.4 Results

We begin by reporting the levels of financial development for the 14 countries in the

analysis. The individual country levels are reported in Table C.2, for two years (2010

and 2014) and four different specifications of the financial development index. FDIn-

dex8 is a slight modification of the measure of financial market development that is used

in the GCI report, with equal weights for all eight components, rather than the two-tier
14 In other words, the components of the vector are not draws from a single distribution, since they are

measures of different aspects of financial development.

55
weighting scheme used in that report. FDIndex7 removes Legal Rights of Investors

from the index calculation. FDIndex10 adds our measures of business environment and

institutional environment to the original eight components, while FDIndex9 adds these

two to the seven components, excluding legal rights.15 All the sub-components or vari-

ables included in these two additional components used in FDIndex10 and FDIndex9

are listed in tables in the Appendix. Removing the legal rights component reduces the

financial development index levels slightly, on average, but adding the two environment

components does not affect the overall levels. The range of variation of the index lev-

els across countries is not much affected by the differences in choice of components,

across the four indices.

The relative and absolute levels of financial development in Table C.2 are not sur-

prising. Overall values of the index are in a tight range, quite far from the extremes

of 1 and 7. There are strong associations between levels of financial development and

GDP per capita, although Korea stands out as a considerable exception on the low side,

while Malaysia has exceptionally high measures of financial development. Japan also

has somewhat of a low score, given its high-income status. We will not go into the con-

tribution of different components of the indices to the variations observed in the levels

in Table C.2. We mainly want to emphasize that the levels are fairly constant over this

short span of time, and especially, that they are not much affected by changes in the
15 The legal rights measure is actually included in the institutional environment component, but is only

one of several institutional factors in that case.

56
 Figure 2.1: Patterns of Financial Development, Developed Economies, 2014.

Affordability

Investor legal rights Availability

AUS

HKG

JPN

Reg of sec. exchanges Equity mkts KOR

NZL

SGP

TWN
Soundness of banks Access to loans

Venture capital

57
 Table 2.3: Financial Development Index Levels.

FDIndex8 FDIndex7 FDIndex10 FDIndex9

2010 2014 2010 2014 2010 2014 2010 2014

AUS 5.24 5.18 5.07 4.92 5.27 5.20 5.14 5.00

CHN 4.22 4.26 4.17 4.29 4.17 4.21 4.12 4.23
HKG 5.60 5.77 5.41 5.60 5.56 5.74 5.40 5.60
IDN 4.23 4.44 4.43 4.50 4.14 4.35 4.29 4.39
JPN 4.44 4.86 4.33 4.82 4.53 4.94 4.46 4.91
KHM 3.63 3.63 3.31 3.32 3.54 3.55 3.29 3.30
KOR 3.75 3.65 3.54 3.34 3.90 3.83 3.75 3.61
MYS 5.22 5.48 4.97 5.26 5.09 5.39 4.88 5.21
NZL 4.91 5.53 4.70 5.32 5.05 5.58 4.90 5.42
PHL 3.94 4.33 4.11 4.46 3.82 4.24 3.93 4.33
SGP 5.51 5.67 5.30 5.48 5.55 5.68 5.39 5.54
THA 4.35 4.54 4.49 4.62 4.27 4.43 4.37 4.48
TWN 4.73 4.88 4.92 5.01 4.76 4.91 4.92 5.02
VNM 4.03 3.65 3.77 3.35 3.96 3.63 3.75 3.39
Source: The World Economic Forum - Global Competitiveness Index

composition of the index, at least for the four alternatives considered here. However,

it is useful to illustrate the patterns of financial development graphically, which we do

in Figures 3.1 and 3.2, for FDIndex8 for 2014: Figure 3.1 shows the seven advanced

economies of the region in our sample, and Figure 3.2 shows the other seven.16 As

noted, Korea stands out in the first case, while Malaysia is an outlier in the second

set, though overall, the emerging economies seem to show greater variation in their

financial development patterns.

We next turn to the measurement of distances between economies of levels and pat-

terns of financial development. Each pair of the 14 economies can be compared in terms
16 We are indebted to our discussant, Siu Fung Yiu, for suggesting construction of these graphs.

58
of each of these measures, which implies 91 numbers for each measure. We report in-

stead the average for each economy, of 13 pairwise distances with the other economies

in the region that are in our set of analysis. Prior to examining these country averages,

regional averages are reported in Table 2.4. For the region as a whole, the first index

suggests that distances in financial development levels did not vary too much across

countries and years. However, average distances in patterns of financial development,

while similar when averaging across the region, showed considerably greater variation

across economies for FDIndex7 in 2014 versus 2010, as evidenced in the higher range.

Removing the legal rights component from the index had small effects on level dis-

tances, but large impacts on pattern distances the latter were much lower on average

(as a result of higher average correlations), and especially in the lower tail of the re-

gional distribution.17 Interestingly, adding in the two components measuring business

environment and institutional environment had small impacts on pattern distances in the

case where the legal rights index was included (FDIndex10), but large impacts when it

was excluded (FDIndex9). However, the pattern distances in the case of FDIndex9 are

lower than in the case of FDIndex8, implying that the differences in measured patterns

of financial development are driven significantly by the single component of legal rights

of investors.
17 In this context, the observations of the International Monetary Funds (IMF) (2015) report are of
interest, Differences in financial regulation between countries are important determinants of financial
integration, as investors may be reluctant to carry out financial transactions with entities in countries
whose regulations and institutions are very different from their own.

59
 Figure 2.2: Patterns of Financial Development, Emerging Economies, 2014.

Affordability

Investor legal rights Availability

KHM

CHN

IDN

Reg of sec. exchanges Equity mkts MYS

PHL

THA

VNM
Soundness of banks Access to loans

Venture capital

60
 Table 2.4: Regional Summary Statistics.

Level Distance Pattern Distance

Year
Average Range Min Max Average Range Min Max

FDIndex8

2010 0.25 0.17 0.18 0.35 0.37 1.35 0.01 1.36

2014 0.29 0.16 0.21 0.37 0.35 1.06 0.00 1.06

FDIndex7

2010 0.24 0.20 0.18 0.38 0.09 0.34 0.01 0.35

2014 0.30 0.21 0.22 0.43 0.11 0.63 0.01 0.64

FDIndex10

2010 0.26 0.16 0.19 0.35 0.40 1.24 0.02 1.26

2014 0.29 0.17 0.22 0.39 0.39 1.03 0.02 1.05

FDIndex9

2010 0.25 0.20 0.18 0.38 0.20 0.73 0.02 0.95

2014 0.29 0.21 0.22 0.43 0.22 0.71 0.02 0.93
Source: The World Economic Forum - Global Competitiveness Index

61
 Table 2.5 through 8 provide economy-level averages for the level and pattern dis-

tances, for each of the four financial development indices. The average level distances

depend on where the economy is in the ranking of levels, with high and low levels of

financial development tending to have higher average distances. However, the average

distance also depends on the overall distribution. Thus, in Table 2.5, Korea has a higher

average level distance than Indonesia, even though its level of financial development

is higher. This is because Indonesia is part of a cluster of economies in this regional

group with relatively low levels of financial development. If we look at average level

distances (again using Table 2.5 to illustrate), Indonesia was similar to several other

economies on this measure (China, Japan, New Zealand, Taiwan, and Thailand), but

very much an outlier in its pattern distance, which was much greater than any of the

other economies in the set. This illustrates a key general point, that economies can be

similar in levels of financial development, or in their closeness of levels on average, but

quite different in their specific patterns of financial development. This point is explored

further later in this section, through an explicit cluster analysis.

The pattern differences in Table 2.5 are not as large, on the whole, as those measured

in Kaur and Singh (2014), using date from the WEF FDR. This reflects the different

composition of the indices, based on different data and a somewhat different conceptu-

alization of financial development. Significantly, when only one of the components, the

index of legal rights of investors, is removed, the pattern distances almost disappear, as

62
 Table 2.5: Level and Pattern Differences FDIndex8

Level Distance Pattern Distance

Country Average Range Min Max Mean SD Min Max

Year 2010

AUS 0.27 1.98 -0.37 1.61 0.24 0.96 0.02 0.98

CHN 0.21 1.98 -1.38 0.59 0.22 0.68 0.08 0.76
HKG 0.35 1.89 0.09 1.98 0.26 1.09 0.04 1.13
IDN 0.20 1.98 -1.38 0.60 0.90 1.18 0.18 1.36
JPN 0.18 1.98 -1.16 0.82 0.23 0.76 0.06 0.82
KHM 0.32 1.85 -1.98 -0.13 0.37 1.31 0.05 1.36
KOR 0.29 1.98 -1.85 0.13 0.26 1.04 0.03 1.07
MYS 0.26 1.98 -0.38 1.59 0.34 1.30 0.03 1.33
NZL 0.21 1.98 -0.69 1.29 0.27 1.04 0.02 1.06
PHL 0.25 1.98 -1.66 0.32 0.42 0.68 0.01 0.69
SGP 0.33 1.98 -0.09 1.89 0.29 1.20 0.03 1.23
THA 0.19 1.98 -1.25 0.72 0.40 0.68 0.01 0.69
TWN 0.19 1.98 -0.87 1.11 0.58 0.81 0.10 0.91
VNM 0.24 1.98 -1.58 0.40 0.33 1.24 0.03 1.27

Year 2014

AUS 0.25 2.14 -0.59 1.55 0.23 0.53 0.00 0.53

CHN 0.25 2.14 -1.52 0.62 0.39 0.84 0.07 0.91
HKG 0.37 2.04 0.10 2.14 0.22 0.5 0.01 0.51
IDN 0.22 2.14 -1.34 0.80 0.44 0.95 0.04 0.99
JPN 0.21 2.14 -0.91 1.23 0.22 0.38 0.09 0.47
KHM 0.37 2.12 -2.14 -0.02 0.43 0.89 0.08 0.97
KOR 0.37 2.14 -2.12 0.02 0.33 0.78 0.03 0.81
MYS 0.31 2.14 -0.29 1.85 0.37 0.87 0.03 0.9
NZL 0.32 2.14 -0.24 1.90 0.22 0.51 0.00 0.51
PHL 0.24 2.14 -1.44 0.70 0.48 1.02 0.04 1.06
SGP 0.35 2.14 -0.10 2.04 0.23 0.54 0.02 0.56
THA 0.22 2.14 -1.23 0.91 0.35 0.79 0.04 0.83
TWN 0.22 2.14 -0.89 1.25 0.48 0.90 0.07 0.97
VNM 0.37 2.14 -2.12 0.02 0.49 1.01 0.05 1.06
Includes: All Eight Sub-Index of the 8th Pillar
Source: The World Economic Forum - Global Competitiveness Index

63
 Table 2.6: Level and Pattern Differences FDIndex7

Level Distance Pattern Distance

Country Average Range Min Max Mean SD Min Max

Year 2010

AUS 0.25 2.09 -0.33 1.76 0.92 0.05 0.78 0.98

CHN 0.20 2.09 -1.24 0.85 0.93 0.04 0.83 0.97
HKG 0.32 1.99 0.10 2.09 0.93 0.03 0.89 0.98
IDN 0.18 2.09 -0.98 1.11 0.87 0.08 0.65 0.95
JPN 0.18 2.09 -1.07 1.02 0.92 0.05 0.78 0.97
KHM 0.38 1.86 -2.09 -0.23 0.83 0.08 0.65 0.91
KOR 0.32 2.09 -1.86 0.23 0.92 0.06 0.75 0.99
MYS 0.23 2.09 -0.44 1.65 0.94 0.03 0.89 0.99
NZL 0.19 2.09 -0.70 1.39 0.89 0.07 0.71 0.98
PHL 0.21 2.09 -1.30 0.79 0.94 0.03 0.88 0.99
SGP 0.30 2.09 -0.10 1.99 0.92 0.05 0.82 0.97
THA 0.18 2.09 -0.92 1.17 0.94 0.04 0.86 0.99
TWN 0.22 2.09 -0.48 1.61 0.86 0.06 0.71 0.95
VNM 0.27 2.09 -1.63 0.46 0.91 0.06 0.75 0.99

Year 2014

AUS 0.24 2.28 -0.68 1.60 0.93 0.08 0.75 0.99

CHN 0.23 2.28 -1.30 0.97 0.89 0.08 0.66 0.97
HKG 0.36 2.16 0.12 2.28 0.94 0.08 0.68 0.99
IDN 0.22 2.28 -1.10 1.18 0.91 0.06 0.74 0.96
JPN 0.23 2.28 -0.78 1.50 0.92 0.09 0.66 0.99
KHM 0.43 2.26 -2.28 -0.02 0.71 0.14 0.36 0.89
KOR 0.42 2.28 -2.26 0.02 0.91 0.09 0.64 0.98
MYS 0.29 2.28 -0.33 1.94 0.92 0.07 0.74 0.98
NZL 0.30 2.28 -0.28 2.00 0.93 0.09 0.72 0.99
PHL 0.22 2.28 -1.13 1.14 0.94 0.07 0.77 0.99
SGP 0.33 2.28 -0.12 2.16 0.92 0.08 0.70 0.98
THA 0.22 2.28 -0.98 1.29 0.93 0.08 0.73 0.99
TWN 0.25 2.28 -0.59 1.69 0.88 0.11 0.54 0.98
VNM 0.42 2.28 -2.25 0.03 0.75 0.14 0.36 0.91
Excludes: Legal Right Index of the 8th Pillar
Source: The World Economic Forum - Global Competitiveness Index

64
can be seen in Table 2.6. Even when broader measures of institutional environment and

business environment are included, to bring the index somewhat closer to that calcu-

lated in the FDR, the pattern distances are higher than in the case of the original eight

components, though higher than the narrowest index of seven components (Tables 2.7

and 2.8). In all of these cases, level distances are relatively insensitive to changes in the

index construction, just as was the case for the levels themselves (Table C.2), although

there are exceptions.

Figures 3.3 through 2.10 are useful in visualizing and summarizing the data in the

tables. Each figure displays the combinations of level distance and pattern distance for

the economies in the regional group. Figures 3.3 and 2.4 present the results for FDIn-

dex8, corresponding to the data in Table 3.1. Clearly, Indonesia is an outlier in 2010, but

its pattern distance changes dramatically in 2014. Figure 2.4, in particular, drives home

the point that there is no obvious relationship between distances in levels of financial

development and distances in patterns of financial development. Thus, while Vietnam

is clearly furthest from other countries in the region on average, in both dimensions,

Hong Kong and Singapore, while having the highest levels of financial development,

and thus the greatest average distances from the other countries, have low levels of av-

erage distance in patterns of financial development. Taiwan and the Philippines, on the

other hand, have low average level distances, but high average pattern distances.

Figures 2.5 and 2.6 provide the scatter plot for the data from Table 2.6, where the

65
 Table 2.7: Level and pattern differences FDIndex10.

Level Distance Pattern Distance

Country Average Range Min Max Mean SD Min Max

Year 2010

AUS 0.28 2.01 -0.29 1.73 0.73 0.28 0.00 0.95

CHN 0.22 2.01 -1.39 0.62 0.74 0.16 0.31 0.89
HKG 0.35 2.01 0.00 2.01 0.71 0.32 -0.07 0.95
IDN 0.22 2.01 -1.42 0.59 0.11 0.39 -0.26 0.77
JPN 0.19 2.01 -1.02 0.99 0.72 0.22 0.11 0.90
KHM 0.34 1.74 -2.01 -0.27 0.61 0.38 -0.26 0.94
KOR 0.26 2.01 -1.66 0.35 0.62 0.31 -0.18 0.90
MYS 0.24 2.01 -0.47 1.55 0.62 0.36 -0.17 0.95
NZL 0.24 2.01 -0.50 1.51 0.66 0.32 -0.12 0.95
PHL 0.28 2.01 -1.74 0.27 0.56 0.22 0.35 0.98
SGP 0.35 2.01 0.00 2.01 0.68 0.37 -0.21 0.93
THA 0.20 2.01 -1.29 0.72 0.58 0.21 0.34 0.98
TWN 0.20 2.01 -0.79 1.22 0.38 0.28 0.04 0.83
VNM 0.25 2.01 -1.60 0.41 0.65 0.36 -0.19 0.95

Year 2014

AUS 0.26 2.18 -0.54 1.64 0.74 0.20 0.45 0.98

CHN 0.25 2.18 -1.53 0.65 0.59 0.27 0.10 0.93
HKG 0.37 2.13 0.05 2.18 0.76 0.18 0.50 0.97
IDN 0.23 2.18 -1.38 0.80 0.52 0.31 0.03 0.96
JPN 0.22 2.18 -0.80 1.38 0.72 0.14 0.50 0.90
KHM 0.39 2.11 -2.18 -0.08 0.56 0.29 0.01 0.91
KOR 0.33 2.18 -1.91 0.27 0.53 0.32 0.04 0.84
MYS 0.29 2.18 -0.35 1.83 0.59 0.30 0.05 0.91
NZL 0.33 2.18 -0.16 2.02 0.74 0.20 0.45 0.98
PHL 0.25 2.18 -1.50 0.68 0.51 0.35 -0.05 0.96
SGP 0.35 2.18 -0.05 2.13 0.74 0.21 0.41 0.98
THA 0.22 2.18 -1.31 0.88 0.62 0.26 0.18 0.96
TWN 0.22 2.18 -0.82 1.36 0.48 0.30 0.01 0.85
VNM 0.37 2.18 -2.11 0.08 0.49 0.37 -0.05 0.87
Includes: All Eight Sub-Index of the 8th Pillar, Includes: Two Additional Index
Source: The World Economic Forum - Global Competitiveness Index

66
 Table 2.8: Level and pattern differences FDIndex9.

Level Distance Pattern Distance

Country Average Range Min Max Mean SD Min Max

Year 2010

AUS 0.27 2.10 -0.25 1.85 0.84 0.11 0.54 0.94

CHN 0.21 2.10 -1.28 0.82 0.86 0.09 0.69 0.96
HKG 0.32 2.10 0.00 2.10 0.88 0.08 0.69 0.96
IDN 0.19 2.10 -1.11 0.99 0.60 0.19 0.25 0.83
JPN 0.18 2.10 -0.93 1.17 0.82 0.12 0.48 0.91
KHM 0.38 1.64 -2.10 -0.46 0.78 0.12 0.54 0.91
KOR 0.28 2.10 -1.64 0.46 0.70 0.17 0.25 0.90
MYS 0.22 2.10 -0.52 1.59 0.84 0.11 0.56 0.97
NZL 0.22 2.10 -0.49 1.61 0.76 0.16 0.35 0.94
PHL 0.24 2.10 -1.47 0.64 0.83 0.11 0.61 0.98
SGP 0.32 2.10 0.00 2.10 0.82 0.13 0.44 0.93
THA 0.18 2.10 -1.03 1.07 0.86 0.09 0.67 0.98
TWN 0.22 2.10 -0.48 1.62 0.76 0.10 0.54 0.87
VNM 0.27 2.10 -1.64 0.46 0.86 0.07 0.74 0.93

Year 2014

AUS 0.24 2.29 -0.60 1.69 0.85 0.09 0.72 0.97

CHN 0.24 2.29 -1.37 0.92 0.78 0.17 0.35 0.95
HKG 0.36 2.23 0.06 2.29 0.88 0.09 0.68 0.96
IDN 0.22 2.29 -1.20 1.09 0.76 0.19 0.27 0.96
JPN 0.23 2.29 -0.69 1.60 0.83 0.11 0.63 0.98
KHM 0.43 2.21 -2.29 -0.08 0.66 0.15 0.34 0.82
KOR 0.36 2.29 -1.99 0.30 0.58 0.20 0.27 0.79
MYS 0.28 2.29 -0.39 1.90 0.78 0.17 0.33 0.93
NZL 0.32 2.29 -0.18 2.11 0.84 0.10 0.70 0.98
PHL 0.23 2.29 -1.27 1.02 0.82 0.16 0.40 0.98
SGP 0.34 2.29 -0.06 2.23 0.85 0.09 0.69 0.98
THA 0.22 2.29 -1.12 1.17 0.83 0.13 0.50 0.98
TWN 0.25 2.29 -0.58 1.71 0.80 0.12 0.51 0.94
VNM 0.41 2.29 -2.21 0.08 0.68 0.15 0.34 0.90
Excludes: Legal Right Index of the 8th Pillar, Includes: Two Additional Index
Source: The World Economic Forum - Global Competitiveness Index

67
index of legal rights of investors has been removed from the overall index of financial

development. In this case, the pattern distance collapses for most of the countries in the

regional set (except for Cambodia and Vietnam). However, without these two outliers,

there is again no clear relationship between distance in levels and distance in patterns of

financial development. The increase in pattern distance from 2010 to 2014 for Cambo-

dia and Vietnam is noteworthy, in the context of increasing financial flows in the region

over this time period.

The scatter plots for the data in Tables 2.7 and 2.8 are presented in Figures 2.7

through 2.10. The results for these measures of financial development, using broader

sets of components, are somewhere in between the earlier two cases (Figures 3.3 and

2.4 versus Figures 2.5 and 2.6), and provide similar visualizations of the central point,

that measuring and comparing patterns of financial development provides different in-

formation than comparing levels of financial development. Figures 2.9 and 2.10, based

on FDIndex9, displays increases in pattern distances for Korea as well as Cambodia and

Vietnam, and a decrease for Indonesia. These heterogeneous movements in patterns of

financial development could have implications for further financial integration in the

region.

It is also helpful to illustrate the variations in level distance and pattern distance

against the actual levels of financial development.18 Since, as noted earlier, these levels

are quite insensitive to the choice of year of components of the index, we only provide
18 We are again indebted to our discussant, Siu Fung Yiu, for suggesting construction of these figures.

68
 Figure 2.3: Level and Pattern Distance, FDIndex8, 2010

1.00

0.90
IDN
0.80

0.70
Pattern Distance

0.60
TWN
0.50

0.40
PHL
THA
0.30 KHM
MYS
VNM
SGP
0.20 NZL KOR HKG
JPNCHN AUS

0.10

0.00
0.00 0.10 0.20 0.30 0.40 0.50
Level Distance

scatter plots for the case of 8 index components, and the year 2014. Figure 2.11 shows

how average level distances vary with levels for 2014: the pattern mostly reflects the

fact that economies with low or high levels of financial development tend to be fur-

ther away from their regional counterparts on average, as compared to economies with

intermediate levels of financial development.

Figure 2.12 shows how average pattern distances vary with levels, for FDIndex8 in

2014. There is some clustering of average pattern distances for five of the advanced

economies in the region, but the other nine economies have higher average pattern

distances, and there is no relationship between the pattern distances and the overall

69
 Figure 2.4:4:Level
Figure Leveland
andPattern
PatternDistance, FDIndex8,
Distance, FDindex82014
2014
0.5

0.45 TWN
PHL VNM

0.4 IDN KHM

0.35 CHN
MYS
Pattern Distance

THA
0.3 KOR

0.25

0.2 JPN AUS NZLSGP

HKG
0.15

0.1

0.05

0
0 0.1 0.2 0.3 0.4 0.5
Level Distance

levels of financial development for these economies. This reinforces the basic idea that

collapsing different dimensions of financial development into a single index can hide

underlying differences in financial development across economies.

As a final exercise, we conduct a cluster analysis to examine how closeness in levels

and patterns of financial development emerges among subgroups of the set of 14 coun-

tries.19 This is particularly pertinent for thinking about financial integration, although

the precise implications of closeness in financial development for financial integration

depend on the specifics of each: in some cases, specialization in aspects of finance may
19 Once more, we are indebted to our discussant, Siu Fung Yiu, for suggesting that we conduct a cluster

analysis.

70
 Figure 2.5: Level and Pattern Distance, FDIndex7, 2010

0.5

0.45

0.4

0.35
Pattern Distance

0.3

0.25

0.2
KHM
0.15
IDN TWN
NZL
0.1 VNM
JPN
CHN AUS SGP
KOR
HKG
THAPHL
MYS
0.05

0
0.00 0.10 0.20 0.30 0.40 0.50
Level Distance

support integration, whereas in other dimensions, integration may best occur when fi-

nancial development is similar across countries.

Our application of clustering analysis uses bilateral differences in levels and pat-

terns of financial development, rather than trade intensity or FDI intensity, which have

been used in previous applications.20 Otherwise, the procedure is the same: at the first

step, the two economies that are closest in distance are combined into a cluster, and this

cluster is treated as a single observation (with averaged values) for the second step. At

the next stage, either another economy is added to this cluster, or a second cluster is
20 For cluster analysis based on trade intensity, see Huang et al. (2006); for analysis based on FDI

intensity, see Yang and Huang (2014).

71
 Figure 2.6:6:Level
Figure Leveland
andPattern Distance,
Pattern FDIndex7,
Distance, 2014
FDIndex7
2014

0.5

0.45

0.4

0.35
Pattern Distance

0.3 KHM

0.25 VNM

0.2

0.15
TWN
CHN
0.1 IDN KOR
JPN
AUS MYS
THA NZLSGPHKG
PHL
0.05

0
0 0.1 0.2 0.3 0.4 0.5
Level Distance

formed, and so on. The sequence of steps is shown in Tables 2.9 and 2.10, for levels

and patterns, respectively.

The process of creating clusters based on levels of financial development is initially

as one would expect, with economies with high, medium and low levels of financial de-

velopment tending to be assigned to the same clusters (Table 2.9). The initial exceptions

to these clusters lining up with per capita incomes are the outliers, Korea and Malaysia.

The initial clusters tend to emerge in the middle and lower end of the distribution of

levels of financial development, with the most financially developed economies, Hong

Kong and Singapore, being clustered only in the seventh step of the process. At the end,

72
 Figure 2.7: Level and Pattern Distance, FDIndex10, 2010

0.9 IDN

0.8

0.7
Pattern Distance

TWN
0.6

0.5
PHL
THA
0.4 MYS
KOR KHM
VNM
NZL
SGP
0.3 JPN HKG
CHN AUS
0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5
Level Distance

the final two clusters are quite heterogeneous, in per capita incomes as well as levels of

financial development. Of course, the entire exercise can be very sensitive to the choice

of measure of financial development.21

21 For
example, we discovered that the clustering was quite sensitive to the way in which the original
measure of investor legal rights was converted to the 1-7 scale.

73
 Figure 2.8: Level and Pattern Distance, FDIndex10, 2014

0.9

0.8

0.7
Pattern Distance

0.6
TWN VNM
0.5 PHL
IDN KOR
KHM
0.4 CHNMYS
THA
0.3 JPN AUS NZL
SGP
HKG
0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5
Level Distance

Figure 2.9: Level and Pattern Distance, FDIndex9, 2010

0.5

0.45

0.4 IDN

0.35
Pattern Distance

0.3 KOR

0.25 NZL
TWN
KHM
0.2
JPN PHL SGP
MYS AUS
0.15 THA
CHN VNM
HKG
0.1

0.05

0
0 0.1 0.2 0.3 0.4 0.5
Level Distance

74
Figure 2.10: Level and Pattern Distance, FDIndex9, 2014

0.50

0.45
KOR
0.40

0.35 KHM
Pattern Distance

VNM
0.30

0.25 IDN
CHN MYS
0.20 TWN
PHL
THA
JPN NZL
0.15 AUS SGP
HKG
0.10

0.05

0.00
0 0.1 0.2 0.3 0.4 0.5
Level Distance

75
 Table 2.9: Clustering based on Levels of Financial Development

Clustering based on Levels of Financial Development

AUS AUS AUS AUS AUS AUS AUS AUS AUS AUS
KHM KHM KHM
CHN CHN CHN CHN CHN CP CP CP CPITh CPITh CPITh
HKG HKG HKG HKG HKG HKG HKG HS HS
IDN IDN IDN IDN IDN IDN ITh ITh
JPN JPN JTw JTw JTw JTw JTw JTw JTw JTw JTwA JTwACPITh JTwACPIThMNHS
KOR KoV KoV KoVKh KoVKh KoVKh KoVKh KoVKh KoVKh KoVKh KoVKh KoVKh KoVKh
MYS MYS MYS MYS MN MN MN MN MN MNHS MNHS MNHS
NZL NZL NZL NZL
PHL PHL PHL PHL PHL
SGP SGP SGP SGP SGP SGP SGP

76
TWN TWN
THA THA THA THA THA THA
VNM
 In the case of patterns of financial development (Table 2.10), the manner in which

clustering proceeds is quite different. The first three steps involve a clustering of Aus-

tralia, New Zealand, Singapore and Hong Kong. However, similarity of patterns of

financial development creates clusters that are quite different from those involving sim-

ilarity of levels, so that Korea and Malaysia are clustered,22 and Taiwan is added at step

nine to a cluster of economies with lower per capita incomes and lower financial devel-

opment measures. Japan, however, ends up in the higher income cluster, but the final

two clusters of economies with similar patterns of financial development are somewhat

different than those based on similarities of levels.

Figure 2.11: Level and Pattern Distance, FDIndex8, 2010

0.5
0.45
0.4
KOR HKG
0.35 VNM
KHM SGP
Level Distance

NZL
0.3
MYS
0.25 CHN AUS
0.2 PHL TWN
IDNTHA
JPN
0.15
0.1
0.05
0
0.00 2.00 4.00 6.00 8.00
Level

22 Vietnam, which is initially clustered with Korea in the levels analysis, later gets added to this cluster
based on patterns. Sahay et al. (2015) also discuss the comparison of South Korea and Vietnam, where
both countries have similar private credit but differ strongly along the financial access dimension.

77
Figure 2.12: Level and Pattern Distance, FDIndex8, 2014

0.5

0.45 VNM PHLTWN

0.4 KHM IDN

0.35 CHN
MYS
Pattern Distance

THA
0.3 KOR

0.25 SGP
0.2 JPN AUS HKG
NZL
0.15

0.1

0.05

0
0.00 2.00 4.00 6.00 8.00
Level

78
 Table 2.10: Clustering based on Patterns of Financial Development

AUS AN ANS ANSH ANSH ANSH ANSH ANSH ANSH ANSH ANSHJ ANSHJKoMV ANSHJKoMVKh
KHM KHM KHM KHM KHM KHM KHM KHM KHM KHM KHM KHM
CHN CHN CHN CHN CHN CHN CHN CHN
HKG HKG HKG
IDN IDN IDN IDN IP IP IPTh IPTh IPThC IPThCTw IPThCTw IPThCTw IPThCTw
JPN JPN JPN JPN JPN JPN JPN JPN JPN JPN
KOR KOR KOR KOR KOR KoM KoM KoMV KoMV KoMV KoMV
MYS MYS MYS MYS MYS
NZL
PHL PHL PHL PHL
SGP SGP
TWN TWN TWN TWN TWN TWN TWN TWN TWN

79
THA THA THA THA THA THA
VNM VNM VNM VNM VNM VNM VNM
2.5 Conclusion

Even after the global financial crisis, which included financial contagion across

national boundaries, there is a continued interest in financial integration, especially in

East Asia or the Asia Pacific region. Recent empirical analyses suggest that financial

integration within the region has been increasing, but still remains below what might

be most beneficial. It is difficult to weigh benefits and risks of financial integration, so

assertions of too much or too little financial integration have to be very tentative.

The literature on financial integration does make references to the need for adequate

levels of financial development, especially in the case of financial market regulation, but

typically does not go further in empirically associating financial integration and finan-

cial development. On the other hand, there is a large literature on financial development

and its impacts, especially on economic growth. This literature has tended to measure

financial development simply as financial depth, but recently broader-based indices of

financial development have begun to be constructed. In some cases, such as Aizenman

et al. (2015) and Sahay et al. (2015), there have been attempts to differentiate between

different dimensions of financial development, such as quantity versus quality, or depth

versus efficiency and access.

In this paper we have argued that even a sophisticated index is limited because it

seeks to reduce the complexities of financial development to a single dimension. As an

80
alternative, we have proposed and constructed a measure of differences in patterns of

financial development, and compared measured differences in patterns with differences

in levels of financial development. Note that our measure is only operative in capturing

the distance between two economies, as opposed to an index (or vector) that measures

financial development in a single economy. Measuring differences in patterns of fi-

nancial development (as opposed to differences in levels) extends the multidimensional

approach to characterizing financial development. One could potentially apply a mea-

sure of pattern difference to components of financial development that capture quantity

versus those that capture quality. This would refine the idea of measuring patterns of

financial development, and remains an avenue of future research.

For 14 economies of the Asia Pacific region, we have calculated differences in lev-

els and patterns of financial development for different years and different vectors of

components of overall financial development. We also illustrated how groupings of

countries can be constructed through cluster analysis, and how the clustering differed

when based on patterns versus levels of financial development. We suggest that these

kinds of calculations can be a useful preliminary tool for assessing prospects for ben-

eficial financial integration among a given set of economies. In particular, since the

components of financial development include various aspects of financial market insti-

tutions, as well as regulatory and governance institutions, focusing on patterns and sub-

patterns of financial development provides a more systematic way of assessing poten-

81
tial regulatory reform and coordination, and possible regional risk management policies

and institutions, both as precursors to, and aspects of, financial integration. Develop-

ing these linkages analytically might contribute to regional policy efforts to develop

bond markets in various regional member economies (e.g., Lim and Lim (2012)), and

to assess the balance between bank and non-bank financing channels, for example.

2.6 Appendix

The appendix for this chapter is provided at the end of the dissertation (Appendix

B).

82
 Part II

Structural Changes and Growth

83
Chapter 3

Structural Changes and Labor Market

Frictions

Written with Seung Mo Choi, International Monetary Fund (contact), and Nikola

Spatafora, International Monetary Fund (contact)

3.1 Introduction

3.1.1 Motivation

Stiglitz (1996) points out the experience of growth miracles in East Asian economies

such as South Korea, Singapore and Taiwan. In these economies, periods of high out-

put growth coincide with structural changes (i.e. labor reallocation from agriculture to

84
manufacturing and services sectors) led by export growth. This is a unique feature that

is not observed in other economies. Other models in the literature calibrate exogenous

productivity growth model and they abstract from features of endogenous growth model

such as learning-by-doing. Hence, they provide limited explanation on the engine of

growth during the growth miracle periods. Also, some models do not explicitly feature

tradable sector and have limited insight on the patterns of structural changes for many

growth miracles. In addition, these models assume frictionless labor market, which ab-

stracts from many sectoral labor reallocation costs that exist in these economies (Artuc

et al. (2015) and Buera and Kaboski (2009)).1

This paper analyzes a simple model of a small economy with agriculture, manufac-

turing and services sectors to study how trade opening and labor market frictions affects

structural transformation and output growth.2 Recent key empirical results presented by

McMillan and Rodrik (2011) provided important empirical underpinning of the results

of our model. In addition, Melitz and Trefler (2012) provides evident that incentive to

innovate is likely to increase in the present of trade and this links well will key features

in our model: endogenous growth (learning-by-doing) and sectoral knowledge transfer

(from manufacturing to agriculture).

The model is based on Matsuyama (1992) model of international trade and Choi

et al. (2014). The economy produces three types of goods: agricultural, manufacturing
1 With the exception of Cai (2015).
2 Papageorgiou and Spatafora (2012) also emphasize a broad connection among trade, structural trans-

formation, and growth.

85
and services good and they are produced with labor in respective sector and a fixed

input. We assume that manufacturing goods are tradable whereas services goods are

non-tradable. In addition, we introduce learning-by-doing in manufacturing sector, a

novel feature of the model that is new to the literature. In particular, the economy-wide

productivity growth is proportional to employment share in manufacturing and services

sectors. As in Duarte and Restuccia (2010), manufacturing sector productivity growth

directly have spillovers to agricultural productivity growth.

In our model, opening to international trade either accelerates structural transforma-

tion (labor reallocation from agriculture to manufacturing and services sectors) or re-

duces structural changes. In the case of the former, the economy exports non-agriculture

goods and for the latter case, the economy exports agricultural goods instead. Also,

since learning-by-doing exists in non-agricultural sectors exports in these sectors in-

tensify their aggregate productivity growth even more.3 Moreover, with international

trade, the economy specializes and exports either agriculture, manufacturing or ser-

vices goods depending on: (i) relative productivity between agriculture, manufactur-

ing and services sectors, (ii) international relative prices between agriculture and non-

agriculture goods (manufacturing and services) and (iii) trade policy. In addition, if

due to labor frictions the non-agriculture sectors (manufacturing or services) provides

higher value of marginal product then labor reallocates to those sectors to equalize the
3 This feature of the endogenous growth model with dynamic learning-by-doing in manufacturing

sector (engine of growth) is akin to that in Alberola and Benigno (2017) and Benigno and Fornaro
(2014b).

86
value of marginal product across all three sectors and frictions such as labor market

distortion accelerates structural transformation.

3.1.2 Literature

There is a large empirical literature, Sachs et al. (1995), Frankel and Romer (1999),

Alcalá and Ciccone (2004), and Felbermayr and Gröschl (2013), on the relationship

between international trade and real GDP growth provides positive evidence. And, on

theoretical front, several models were established to analyze how international trade

affect real GDP growth. However, the theoretical literature abstracts from endogenous

growth models setup, which could be important quantitatively. In particular, there are

few theoretical models with endogenous growth that connect the nexus between struc-

tural changes and growth.

Cai (2015) also models labor market friction (sectoral labor wedge) in model of

structural transformation and studies the role of relative sectoral productivity and la-

bor market distortions in accounting for structural transformation in the U.S., Brazil,

India and Mexico between 1960 and 2005. The paper finds that sectoral productiv-

ity growth plays more important role in the evolution of structural changes than labor

market frictions, using counterfactual on the model calibrated to match the data on em-

ployment and value-added shares in the data. In particular, the decline in agriculture

productivity contributes most to the reduction in that sector employment share. Similar

87
to Cai (2015), our model also introduces sectoral labor market friction to the literature

on structural changes. However, they do not take into account the role of international

trade and sectoral labor market distortions and productivity growth are assumed to be

independent.

Alberola and Benigno (2017) and Benigno and Fornaro (2014b) focuses on en-

dogenous growth model with dynamic learning-by-doing in tradable sector as engine

of growth. Ventura (1997) also studies the patterns of structural changes in miracle

economies during their fast growth periods with international trade. As a result, the

model in this paper could be viewed as alternative to that of Ventura (1997). How-

ever, Ventura (1997), as with others in the literature, abstract from endogenous growth

model, i.e. dynamic productivity growth with learning-by-doing in manufacturing sec-

tor as engine of growth. Capital accumulation plays a more important role as engine of

growth in Ventura (1997) as opposed to productivity growth.

Lucas (2009) links trade, growth and structural changes in a model where aggregate

productivity growth depends on the share of employment in non-agriculture sector and

the output gap with the leader economy. In particular, knowledge externalities is con-

centrated in urban setting (non-agriculture sector) and economies with higher share of

employment in non-agriculture sector benefits from international trade more in Lucas

(2009).

Even though this paper is closed related to Cai (2015) and Alberola and Benigno

88
(2017), the theoretical model is based on Matsuyama (1992), Lewis (1954) and Choi

et al. (2014). Matsuyama (1992) model could be as an extension of Lewis (1954) with

international trade. Matsuyama (1992) studies how structural transformation evolves at

a macroeconomic level focusing on total output and employment while Lewis (1954)

analyzes the patterns of structural changes at the micro level. However, Matsuyama

(1992) focus on theoretical aspect. This paper can be viewed as a quantitative extension

with distortion in the labor market.

Duarte and Restuccia (2010) investigate the role of sectoral labor productivity to

explain structural transformation. Restuccia et al. (2008) study the relationships among

sectoral productivity differences, labor market frictions, and the sectoral shares of em-

ployment. Also, Dolores Guilló et al. (2011) also provide a model of structural trans-

formation. All of them share some features with this paper, such as structural transfor-

mation and labor market frictions. However, they do not explicitly focus on the role of

international trade on structural transformation and eventually on output growth. Gollin

and Rogerson (2014) presents the case for labor market frictions between sectors in

developing countries.4 There are some micro-data evidence of labor market frictions

(between sectors) in developing countries (Artuc et al. (2013) and Artuc et al. (2015)).

Song et al. (2011), part of the literature on global imbalance trying to explain the

growing reserves in trade trade surplus countries, focuses on the role of financial market
4 There is tradition in the literature that argues for the role of labor market frictions and Gollin (2014)

discusses some detailed review.

89
imperfection in explaining the accumulation of international reserves in China. Their

model matches the experience of China’s patterns of economic transformation. Unlike

this paper, they emphasize friction in the financial market as opposed to the labor market

and international trade. Teignier (2018) also studies the link between structural changes

and international trade but with exogenous output growth.

3.1.3 Contribution

Our contribution is on generating the hump-shaped manufacturing employment

share, a robust empirical findings that also holds in terms of value added. The main

innovation introduces labor adjustment frictions between sectors. In addition, using

this framework of endogenous (learning-by-doing) growth model with labor market

frictions, we could study the impact of international trade and labor market distortions

on patterns of structural changes. We plan to contribute to the literature by incorpo-

rating labor market friction into an open economy à la Eaton and Kortum (2002) and

Caliendo and Parro (2015). Using this theoretical framework, we could analyze the

impact of structural changes and productivity shocks (agriculture, manufacturing and

service sectors) on the dynamics of the economy. In their model, Choi et al. (2014)

obtain closed-form solutions for many variables, which facilitates the understanding of

the main mechanism. Hence, the growth accounting can easily separate the contribution

of labor composition effect (i.e., moving from the less productive agricultural sector to

90
the more productive non-agricultural sector) from the contribution of learning by doing

(which is assumed to be a main endogenous source of productivity growth) and others,

if any, such as international knowledge transfer. We are not able to obtain closed-form

analytical solutions and we instead rely on numerical method in solving the model.

The remainder of the paper is organized as follows. Section 2 provides a theoretical

model. Section 3 studies a quantification of the model in a closed economy version.

Section 4 discusses the construction of the sectoral international relative prices (be-

tween agriculture, manufacturing and services). Section 5 extends the model to open

economy. Section 6 provides discussion on the ”Pseudo” Gini coefficients generated

from the model. Section 7 concludes.

3.2 Model

3.2.1 Production

There is a continuum of identical villages. One village has each of the following

three firms: A (agriculture) firm, M (manufacturing) firm, and S (services) firm. At

each period t, a fraction mt , 0 ≤ mt ≤ 1, of labor is allocated to the M firm, a fraction

st , 0 ≤ st ≤ 1, is allocated to the S firm, and the remaining fraction of labor, 1 − mt − st ,

is allocated to the A firm. The M firm’s goods production uses linear technology with

91
mt share labor unit in manufacturing:

YtM = Mt mt ,

where YtM is the output in units of M goods, and Mt is the country-wide productivity

in the M sector. The productivity, Mt , features learning-by-doing.5 To be specific, Mt

evolves externally according to the country-wide production, following

Mt+1 − Mt = µM mt Mt ,

for µM > 0, where mt is a country-wide fraction of employment in manufacturing

sector (so in equilibrium, mt = mt ). Rewriting the equation above to express in terms

of period percentage changes,

Mt+1 − Mt
= µM mt
Mt

Since an individual firm is small, its production decision does not affect the evolu-

tion of country-wide productivity Mt .6 In particular, the period growth rate of country-

wide productivity in manufacturing depends only on country-wide fraction of labor

employment in manufacturing, mt , and is independent of individual firm share, mt .

5 See Thompson (2010) for empirical survey on learning by doing.
6 Matsuyama (1992) and Lucas (2009) also assume external learning-by-doing effects. An alternative
set-up, to internalize the learning by doing, complicates the analysis without notable gains.

92
 The services sector, S, firm’s production function7 is

YtS = St st S ,
φ

where YtS is the output in units of S goods, and St is the country-wide productivity

in S sector. In addition, St evolves as

St+1 − St = µS st St ,

for µS > 0, where st is a country-wide level of st . Similar with manufacturing sector,

rewriting the equation above for country- wide services sector, St :

St+1 − St
= µS st
St

The agriculture sector, A, firm’s production function is

YtA = AMtθ (1 − mt − st )φ ,

where YtA is the output in units of agriculture goods. Here, A is a constant which cap-
7 Alternative S Sector Production Function: Possibly to account for spillover from M sector to S
sector, the use of intermediate inputs.

YtS = (AS St ) Mtθ st S ,

φ

93
tures cross-country differences in fertilities, etc... Also, country-wide manufacturing

productivity, Mt , affects output in agriculture and in particular, Mtθ reflects the knowl-

edge transfer from manufacturing sector, M, to agriculture sector, where θ ≥ 0 is a

parameter of the size of such transfer.8 This is to consider the fact that the growth of

per- capita GDP had been slower until the Industrial Revolution and the well-known

effect of technological advances toward agricultural sector.9

The instantaneous profits for a village is as follows:

ptM M ptS S
πt = YtA + Y + A Yt .
ptA t pt

Replacing output for agriculture, manufacturing, services sector, the instantaneous

profits is rewritten as:

ptM ptS  φS 
πt = AMtθ (1 − mt − st )φ + (M m
t t ) + St st .
ptA ptA

We introduce quadratic sectoral labor adjustment costs to the model and the instan-

taneous profits is rewritten as follows10 :

8 Lucas (2009) and Dolores Guilló et al. (2011) adopt similar approaches. See, for example, Johnson
and Evenson (1999) for empirical discussions on the knowledge transfer from manufacturing to agricul-
tural sector.
9 See, for example, Lucas (2009). Restuccia et al. (2008) shows an important role of technological

improvement of the manufacturing sector used as intermediate goods for agriculture production. A
parameter φ determines the curvature of A-good production, where a condition 0 < φ < 1 is imposed to
reflect decreasing marginal products due to fixed amount of lands, etcetera.
10 The main role of quadratic sectoral labor adjustment cost is to help generate the hump-shape in

manufacturing sector.

94
 ptM ptS  φS  1
πt = AMtθ (1 − mt − st )φ + (M m
t t ) + St st − κ (mt + st − mt−1 − st−1 )2 Mtθ .
ptA ptA 2

where κ is the quadratic labor adjustment costs parameter for labor reallocation in

and out of agriculture. The sectoral labor adjustment costs is proportional to agriculture

sector productivity, Mtθ , to remain on balanced growth path.

The firm maximization problem

 
ptM ptS
 
t AM θ (1 − m − s )φ + (M m ) + φ
∞ 
1  t t t ptA t t St st S + ... 
ptA
Max ∑  .
{mt ,st } t=0 1 + r  
... − 21 κMtθ (mt + st − mt−1 − st−1 )2

We derive the following FOCs:

With respect to manufacturing employment share, mt :

ptM
φAMtθ (1 − mt − st )φ−1 = Mt − κMtθ (mt + st − mt−1 − st−1 ) + ...
ptA
1
... + κM θ (mt+1 + st+1 − mt − st ) .
1+r t

The FOCs holds that the marginal product of labor ”benefits” in the A sector is equal

the marginal product of labor ”benefits” in the M sector net the marginal cost of labor

force adjustment in the M sector. The key implication is that due to the existence of

labor adjustment cost κ firms will smooth labor force fluctuation (adjustment).

95
 With respect to services employment share, st :

ptS φ −1
φAMtθ (1 − mt − st )φ−1 = A
φS St st S − κMtθ (mt + st − mt−1 − st−1 ) + ...
pt
1
... + κM θ (mt+1 + st+1 − mt − st ) .
1+r t

3.2.2 Preferences

The representative consumer at period t chooses consumption on agriculture, man-

ufacturing and services goods CtA ,CtM ,CtS respectively and solves the following maxi-

mization problem subject to total income ptAYtA + ptMYtM + ptSYtS given prices of agri-

culture, manufacturing and services goods ptA , ptM , ptS .

∞  
−ρt A M S
max ∑e U Ct ,Ct ,Ct
{CtA >γA ,CtM >0,CtS >0} t=0

for ρ > 0, βA > 0, βM > 0 and βS > 0, subject to:

ptACtA + ptMCtM + ptSCtS ≤ ptAYtA + ptMYtM + ptSYtS .

for all t = 0, 1, ...

We model the consumer utility as log utility with subsistence consumptions in agri-

culture and services sector. Preference is non-homothetic due to the existence of sub-

96
sistent consumption goods in agriculture and services.

 
U CtA ,CtM ,CtS = βA log(CtA − γA ) + βM log(CtM ) + βS log(CtS + γS )

where CtA , CtM , CtS are consumption in agriculture, manufacturing and services, γA

and γS are subsistent consumption in agriculture and services respectively. Rewriting

consumer optimization problem, we have as follows:

∞ h i
−ρt A M S
max ∑ e βA log(Ct − γA ) + β M log(Ct ) + βS log(Ct + γS ) ,
{CtA >γA ,CtM >0,CtS >0} t=0

for ρ > 0, βA > 0, βM > 0 and βS > 0, subject to

ptACtA + ptMCtM + ptSCtS ≤ ptAYtA + ptMYtM + ptSYtS .

for all t = 0, 1, ...

Here, CtM is the consumption of good M, CtS is the consumption of good S, and

CtA is the consumption of good A. In addition, γA and γS are the subsistence levels of

A consumption and S consumption. That is, the consumer first purchases γA units of

good A, and allocates βA /(βA + βM + βS ) of the remaining budget to consume A goods,

βM /(βA + βM + βS ) of the remaining budget to consume M goods, and βS /(βA + βM +

βS ) of the remaining budget to consume S goods and as income increases, the consumer

will allocate more resources into S sector consumption.

97
 Subsistence Consumption in Services Sector: We model the subsistence con-

sumption in the services sector γS as in Duarte and Restuccia (2010) and, as income

increases, more resources will be put into the services sector production as demand in-

creases. Intuitively, we think of γS as ”negative” subsistence consumption. Combining

the FOCs, we have the following:

ptA CtA − γA


ptMCtM pS (CS + γS )
= = t t . (3.1)
βM βA βS

In addition, γA and γS are the subsistence levels of A consumption and M con-

sumption. That is, the consumer first purchases γA units of good A, and allocates

βA /(βA + βM + βS ) of the remaining budget to consume A goods, βM /(βA + βM + βS )

of the remaining budget to consume M goods, and βS /(βA + βM + βS ) of the remaining

budget to consume S goods and as income increases, the consumer will allocate more

resources into S sector consumption. If the solution to this equation satisfies CtS < 0,

then the consumer allocates βM /(βA + βM + βS ) of the remaining budget to consume

M goods and βA /(βA + βM + βS ) of the remaining budget to consume A goods while

CtS = 0.11
11 As it becomes clear later, the levels of γA and γS affect the features of structural transformation in
a closed economy, but not in a small open economy. This is because the structural transformation in a
small open economy solely depends on its comparative advantage.

98
3.3 Quantitative Analysis: Closed Economy

3.3.1 Closed Economy Solution

A closed economy requires CtM = YtM ,CtS = YtS and CtA = YtA .

CtA = YtA = AMtθ (1 − mt − st )φ .

CtM = YtM = Mt mt .

CtS = YtS = St st S .
φ

Given the country-wide productivity, Mt and St , the above equations, becomes a system
pS ptA
of four equations with four unknowns: mt , st , pMt and ptM
. This system is rewritten here
t

for convenience:

99
 ptM
φAMtθ (1 − mt − st )φ−1 = Mt − κMtθ (mt + st − mt−1 − st−1 ) + ... (3.2)
ptA
1
... + κM θ (mt+1 + st+1 − mt − st ) .
1+r t
ptS φ −1
φAMtθ (1 − mt − st )φ−1 = A
φS St st S − κMtθ (mt + st − mt−1 − st−1 ) + ... (3.3)
pt
1
... + κMtθ (mt+1 + st+1 − mt − st ) .
 1 + r 
M pA AM θ (1 − m − s )φ − γ
pt (Mt mt ) t t t t A
= . (3.4)
βM βA
φ
ptM (Mt mt ) ptS (St st S + γS )
= . (3.5)
βM βS

In solving the system of equations (3.2-3.5) above, we combine equation 3.2 and 3.3
φ −1
ptM φS St st S
to get ptS
= Mt and from combining equation 3.4 and 3.5, we have the following:

" #
βM 1 1 γS
mt = st + . (3.6)
β S φS φS St stφS −1

With φS 6= 1, we have a non-linear relationship between mt and st and potentially,

this help generate the hump-shaped in the manufacturing sector. In particular, increases

in employment share of the services sector initially increase labor share of the manu-

facturing sector but decrease labor share of M sector later on. With φS = 1, then we

have
βM St st + γS
mt = .
βS St

100
Relationship between st and mt :

Taking total derivative of the relationship between mt and st above, we have

" #
dmt βM 1 1 − φS γS
= + .
dst βS φS φS St stφS

From equation above, we can get:

ptM βM AMtθ (1 − mt − st )φ − γA
Mt = .
ptA βA mt

ptM
Using the expression for M,
ptA t
we have:

βM AMtθ (1 − mt − st )φ − γA
φAMtθ (1 − mt − st )φ−1 − = ...
βA mt

1
... = κM θ (mt+1 + st+1 − mt − st ) − κMtθ (mt + st − mt−1 − st−1 ) . (3.7)
1+r t

And, we have another relationship between mt and st

φ
βM St st S + γS
mt = . (3.8)
βS φS St stφS −1

We have two equations with two unknowns {mt , st }

As a result, we have a non-linear second-order difference equation to solve for st as

101
a function of the calibrated parameters.

In order to solve the model, we will use the following simplifying approximation to

solve for {mt , st } in (3.7) and (3.8)

mt+1 + st+1 − (mt + st ) = mt + st − (mt−1 + st−1 ) .

Rewriting the expression,

at+1 + at−1
at = .
2

This paper will focus on the long-run trend in labor reallocation across sectors in

emerging and developing countries and we think that it is a ”reasonable” assumption. In

particular, sectoral employment shares (A, M & S) fluctuate slowly around the trend.12

3.3.2 Closed Economy Calibration

3.3.2.1 Production

The detailed explanation and discussion of the closed economy calibration.13 One

of the main feature of the model is the role of endogenous growth model with learning-

by-doing in agriculture and services sector and they are captured in {µM , µS }. From the
12 Note: we will need to check on the structural break in the WDI data on sectoral employment share
as some countries have significant and visible breaks in the labor shares time-series and hence, the
approximation will not likely hold. For example, for South Korea, in 2009, there is jumps in the WDI
data on manufacturing and services employment shares.
13 For detailed explanation and discussion of the closed economy calibration, please see the attached

Appendix B.

102
M sector, we have the following ratio:

T
YTM /mT
= ∏ (1 + µM mt ) .
Y0M /m0 t=0

Given the data on real value-added growth rates (normalized by respective employ-

ment shares) and employment shares mt for M sector, we get the learning-by-doing

parameters for M sector µM .14

Similarly, from the S sector, we have the following:

φ T
YTS /sTS
φ
= ∏ (1 + µS st ) .
Y0S /s0S t=0

Given the data on real value-added growth rates (normalized by respective employ-

ment shares) and employment shares st for S sector, we get the learning-by-doing pa-

rameters for S sector µS .

Manufacturing Spillover from Manufacturing to Agriculture {θ}: from the A

sector, we have the following:

YTA
  
log − φ log aaT0
Y0A
θ=  M  .
YT
log Y M − log mmT0
0

We use the sectoral real value added annual growth rates of each sector A, M and S
14 Similarto Duarte and Restuccia (2010), we assume that the growth rate of real value- added (nor-
malized by respective labor share) are good approximation to the changes in ”quantities”.

103
data from the World Bank WDI for k ∈ {A, M, S}:

RealVAtk
k
= 1 + gtk .
RealVAt−1

Initial Period Normalization: with Y1A = 1,Y1M = 1 and Y1S = 1, we get YTM ,YTM

and YTS using the sectoral real value added annual growth rates.

3.3.2.2 Preferences

Labor Adjustment Costs and Interest Rate: {κ, r}

From the FOCs, we have:

YtA pM Y M
φ = tA t − κMtθ (mt + st − mt−1 − st−1 ) + ...
at pt mt
 
1
... + κMtθ (mt+1 + st+1 − mt − st ) .
1+r

Subsistence Consumptions and Discount Factors: {βA , βM = 1, βS , γA , γM = 0 and γS }

Given the calibrated parameters {βA , βM = 1, βS }, we calibrate the subsistence con-

sumption parameters {γA , γM = 0, γS } as follows:

For closed economy equilibrium, we have

ptM
γA = AMtθ (1 − mt − st )φ − βA Mt mt . (3.9)
ptA

104
 Given the calibrated parameters {A, φ, βA , βM = 1, βS }, we will match the agricul-

ture subsistence consumption γA with the following: we first calibrate discount factor

βA and then we calibrate to match the data of the initial period employment shares for

agriculture (a0 = 1 − m0 − s0 ) and manufacturing (m0 ), initial period sectoral productiv-

ity for agriculture (AM0θ ) and manufacturing (M0 .) and the initial period relative price
 M
p
between manufacturing and agriculture p0A .
0

Similarly, given the parameters {βA , βM = 1, βS }, we calibrate the subsistence con-

sumption parameters {γA , γM = 0, γS } as follows:

For closed economy equilibrium, we have:

ptM φ
γS = S
βS Mt mt − St st S . (3.10)
pt

Given {φ, βA , βM = 1, βS }, we will match the services subsistence consumption γS

with the following: we first calibrate discount factor βS and using the initial period

employment shares in agriculture and services (m0 , s0 ) and initial period manufactur-

ing and services productivity (M0 , S0 ) as well as initial period relative price between
 M
p
manufacturing and services p0S .
0

Discount Factors {βA , βM = 1, βS }

105
Using the firms FOCs above, we have the expression for βS with βM = 1, we have

φ φ
St st S − S0 s0S
βS =  . (3.11)
pM

φ −1
φS St st S mt − p0S M0 m0
0

With φS = 1,

St st − S0 s0
βS = .
pM
St mt − 0
M m
pS0 0 0

We will use initial and end period data on sectoral employment shares and produc-

tivity to back out βS .

Given βS , we then get γS using

pM
0
γS = βS M0 m0 − S0 s0 .
pS0

Similarly for βA :

From the FOC above, we solve for βA ,

φ φ
AMtθ at − AM0θ a0
βA = . (3.12)
DA

Where
  pM
mt − 0A M0 m0 + ...
φ−1
DA = φAMtθ at
p0

106
 1
... + κMtθ (mt + st − mt−1 − st−1 ) mt − κM θ (mt+1 + st+1 − mt − st ) mt .
1+r t

With approximation,

mt+1 + st+1 − mt − st = mt + st − mt−1 − st−1 .

φ φ
AMtθ at − AM0θ a0
βA =  .
pM

φ−1 r
φAMtθ at mt + 1+r κMtθ (mt + st − mt−1 − st−1 ) mt − p0A M0 m0
0

Given βA , we then get γA using

φ pM
0
γA = AM0θ a0 − βA M0 m0 .
pA0

3.3.3 South Korea

For our quantitative analysis, we will focus on South Korea which is one of the

growth miracle economies illustrated in Stiglitz (1996) and shows hump-shape man-

ufacturing employment shares (middle left panel of Figure 3.1) and the parameters

calibration matching data from South Korea is shown in Table 3.1.

107
 Figure 3.1: Closed Economy Model Simulation: Sectoral Employment Share

(i) Data: Agriculture Employment Share (ii) Model: Agriculture Employment Share

0.25 0.25

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0 0
1985 1990 1995 2000 2005 2010 1985 1990 1995 2000 2005 2010

(iii) Data: Manufacturing Employment Share (iv) Model: Manufacturing Employment Share
0.4 0.35

0.35

0.3

0.25 0.3

0.2

0.15

0.1 0.25
1985 1990 1995 2000 2005 2010 1985 1990 1995 2000 2005 2010

(v) Data: Services Employment Share (vi) Model: Services Employment Share
0.8

0.7 0.65

0.6
0.6
0.55
0.5
0.5
0.4 0.45

0.3 0.4
1985 1990 1995 2000 2005 2010 1985 1990 1995 2000 2005 2010

108
 Table 3.1: Calibrations for South Korea

Description Parameter Value

Learning-by-doing for manufacturing µM 0.310
Learning-by-doing for services µS 0.053
Cross-country agriculture Productivity A 1
Agriculture Production φ 0.810
Services Production φS 0.450
Manufacturing Technology Spillover θ 0.716
Agriculture Discount Factor βA 0.047
Manufacturing Discount Factor βM 1
Services Discount Factor βS 8.999
Agriculture Subsistence Consumption γA 0.907
Manufacturing Subsistence Consumption γM 0
Services Subsistence Consumption γS 1.509
Interest Rate r 0.020
Sectoral Labor Adjustment Costs κ 2
Data sources: World Bank WDI and Groningen Growth and Development Center (GGDC) 10-Sector

3.4 Quantitative Analysis: Open Economy

3.4.1 Open Economy Solution

A unique currency is used in all economies. Assume a small open economy. As-

sume S goods are non-tradable. From the consumer FOCs above, we have:

pA CA − γA


ptMCtM
= t t
βM βA
ptS (CtS + γS )
= .
βS

109
 The open economy equilibrium condition for non-tradable services sector, we have

the following condition:

CtS = YtS = St st S .
φ

For tradable sectors (manufacturing and agriculture), we have the following:

ptA CtA − γA


ptMCtM
= .
βM βA
φ
pS (St st S + γS )
= t .
βS

Using the balanced trade condition, for tradable sectors in order to close out the

model we have as follows:

 
ptA Yt −Ct + ptM YtM −CtM = 0.
A A

The trade balance condition can be re-written as follows:

   
βA S βM S
ptA AMtθ at − ptA γA − M
φ φS φS
p (St st + γS ) + pt Mt mt − p (St st + γS ) = 0.
βS t βS t

The trade balance condition will give us a relationship between mt and st given

the parameters and the relative prices. Given the productivity, Mt and St , the above
pS ptA
equations, becomes a system of four equations with four unknowns: mt , st , pMt and ptM
.
t

This system is rewritten here for convenience:

110
 ptM
φAMtθ (1 − mt − st )φ−1 = Mt − κMtθ (mt + st − mt−1 − st−1 ) + ... (3.13)
ptA
1
... + κM θ (mt+1 + st+1 − mt − st ) ,
1+r t
ptS φ −1
φAMtθ (1 − mt − st )φ−1 = A
φS St st S − κMtθ (mt + st − mt−1 − st−1 ) + ... (3.14)
pt
1
... + κM θ (mt+1 + st+1 − mt − st ) ,
1+r t
pA CtA − γA

φ
ptMCtM ptS (St st S + γS )
= t = , (3.15)
βM βA βS

 
ptA Yt −Ct + ptM YtM −CtM = 0.
A A


(3.16)

Combining equations (3.13) and (3.14)15 :

φ −1
ptM φS St st S
= .
ptS Mt

Using the trade balance equation (3.16), we have the following:

βA + βM ptS h φS i pM
AMtθ (1 − mt φ
− st ) = γA + A
St st + γS − tA Mt mt .
βS pt pt

φ
Using CtM = YtM − XtM , YtM = Mt mt , and YtS = St st S we have as follows:
15 Similar with the literature, specifically Hsieh and Klenow (2007), relative prices is derived from the

model as the inverse of relative productivity.

111
We have
φ
XtM βM St st S + γS
mt = + .
Mt βS φS St stφS −1

From equation (3.15), we can get:

ptM βM AMtθ (1 − mt − st )φ − γA
Mt = .
ptA βA mt

ptM
Plugging in the expression for M,
ptA t
we have:

ptM
φAMtθ (1 − mt − st )φ−1 = A
Mt − κMtθ (mt + st − mt−1 − st−1 ) + ...
pt
1
... + κM θ (mt+1 + st+1 − mt − st ) ,
1+r t

With the simplifying approximation

mt+1 + st+1 − mt − st = mt + st − mt−1 − st−1 .

ptM r
φAMtθ (1 − mt − st )φ−1 = Mt − κM θ (mt + st − mt−1 − st−1 ) . (3.17)
ptA 1+r t

112
 From the balanced trade condition, we have another relationship between mt and st

βA + βM ptS φS ptM
AMtθ (1 − mt − st )φ = γA + (S s
t t + γ S ) − Mt mt . (3.18)
βS ptA ptA

We have two equations with two unknowns {mt , st }. As a result, we have a non-

linear second-order difference equation to solve for st as a function of the calibrated

parameters.

Simplifying Approximation: we will use the following simplifying approximation

to solve for {mt , st } in (3.17) and (3.18)

mt+1 + st+1 − (mt + st ) = mt + st − (mt−1 + st−1 ) .

Rewriting the expression,

at+1 + at−1
at = .
2

3.4.2 Open Economy Calibration

3.4.2.1 Open Economy Production

Endogenous Growth: Learning-by-Doing {µM , µS }: the calibration procedure

for µM and µS is the same as in the closed economy. From the M sector, we have the

following ratio:

113
 T
YTM /mT
= ∏ (1 + µM mt ) .
Y0M /m0 t=0

Given the data on real value-added growth rates (normalized by respective employ-

ment shares) and employment shares mt for M sector, we get the learning-by-doing

parameters for M sector µM 16 . Similarly, from the S sector, we have the following ratio:

φ T
YTS /sTS
φ
= ∏ (1 + µS st ) .
Y0S /s0S t=0

Given the data on real value-added growth rates (normalized by respective employ-

ment shares) and employment shares st for S sector, we get the learning-by-doing pa-

rameters for S sector µS .17

Manufacturing Spillover {θ}: from the A sector, we have the following:

YTA
   
aT
log − φ log
Y0A a0
θ=  . (3.19)
YM

log YTM − log mmT0
0

We use the sectoral real value added annual growth rates of each sector A, M and S data

from the World Bank WDI.

RealVAtk
k
= 1 + gtk
RealVAt−1
16 Similar to Duarte and Restuccia (2010), we assume that the growth rate of real value-added (nor-
malized by respective labor share) are good approximation to the changes in ”quantities”.
17 Similar to footnote description on µ above.
M

114
 where k ∈ {A, M, S}

Initial Period Normalization: with Y1A = 1,Y1M = 1 and Y1S = 1, we get YTM ,YTM

and YTS using the sectoral real value added annual growth rates.

3.4.2.2 Open Economy Preference

Labor Adjustment Costs and Interest Rate {κ, r}: from the FOCs, we have:

YtA pM Y M
φ = tA t − κMtθ (mt + st − mt−1 − st−1 ) + ...
at pt mt
 
1
... + κMtθ (mt+1 + st+1 − mt − st ) (3.20)
1+r

Subsistence Consumptions and Discount Factors: {βA , βM = 1, βS , γA , γM = 0 and γS }

Trade Balance Condition: using the trade balance equations, we have as follows:

 
ptA Yt −Ct + ptM YtM −CtM = 0.
A A

Sectoral Net Exports Xti : from the data, we define sectoral net exports Xti for


i ∈ {A, M} as follows:

Xti = Yti −Cti .

115
 Hence, the value of sectoral net exports are as follows:

pti Xti = pti Yti −Cti .

World Bank WDI Data for Net Exports Xti : we will use imports and exports


data on (as % of Merchandise imports and exports): agriculture raw materials, food

and manufactures and in order to get the value of imports and exports, we will need the

value of: merchandise imports (current US$) and merchandise exports (current US$)

We calculate imports and exports as % of GDP for manufacturing (manufactures)

and agriculture (agriculture raw materials and food/trade balance condition require-

ment) from the data above. We have the sectoral net exports as a percentage of GDP as

follows (for i ∈ {A, M}):

pti Xti
xti = .
ptAYtA + ptMYtM + ptSYtS
XtM
xtM = S . (3.21)
ptA A M + pt Y S
M
pt
Yt +Yt pt t
M

Similarly, for agriculture, we have the following

XtA
xtA = M S . (3.22)
YtA + pptA YtM + ppAt YtS
t t

116
We will also impose trade balance condition on xtA once we have xtM . In particular,

xtA = −xtM .

Real Manufacturing Net Exports XtM :



ptA A ptS S
 
XtM = xtM M
Y +Yt + M Yt .
ptM t pt
 Ah
ptS h φS i

M pt
i
M θ φ
Xt = xt AMt at + Mt mt + M St st .
ptM pt

t=T
Using the data on sectoral real value added growth rates gtA , gtM , gtS

t=2
(initial pe-

riod sectoral value added normalization to 1), we calculate all the sequence of sectoral
t=T
real value added YtA ,YtM ,YtS t=1 .


n A Mo
p p
Relative Prices pMt , ptS : for the relative price between agriculture and manu-
t t
 A
p
facturing, we calculate and aggregate pMt from the detailed level of the World Bank
t

ptM
international price (ICP 2005 edition) and ptS
is calibrated using the first-order condi-
φ −1
ptM φS St st S
tion: ptS
= Mt .

Manufacturing: we have the following equations

βM S
ptM XtM = ptM Mt mt −
φ
pt (St st S + γS ).
βS

Solving for γS , we have

117
 βS ptM
Mt mt − XtM − St st S .

φ
γS = S
βM pt

Set

βM = 1.

ptM 
γS = S βS Mt mt − XtM − St st S .
 φ
(3.23)
pt

Agriculture: solving for γA , we have

h i pM 
γA = AMtθ at − XtA − tA βA Mt mt − XtM .
φ 
(3.24)
pt

Subsistence Consumption {γA , γM = 0, γS }: given the parameters {βA , βM = 1, βS },

we calibrate the subsistence consumption parameters γA using (3.24). Given the cal-

ibrated parameters {A, φ, βA , βM = 1, βS }, we match the agriculture subsistence con-

sumption using: calibrated discount factor βA , initial period employment shares for

agriculture and manufacturing (a0 = 1 − m0 − s0 , m0 ), initial period net exports in agri-

culture and manufacturing X0A , X0M , initial period sectoral productivity for agriculture

and manufacturing AM0θ , M0 and initial period relative price between manufacturing
 M
p
and agriculture p0A .
0

Similarly, given the calibrated parameters {βA , βM = 1, βS }, we calibrate the sub-

118
sistence consumption parameters γS using (3.23). Given the calibrated parameters

{A, φ, βA , βM = 1, βS }, we match the services subsistence consumption γS with the fol-

lowing:

• Calibrated discount factor βS .

• Initial period employment shares for manufacturing m0 and services s0 .

• Initial period net exports in manufacturing X0M .

• Initial period sectoral productivity for manufacturing M0 and services S0 .

pM
• Initial period relative price between manufacturing and services 0
pS0
.

Discount Factors {βA , βM = 1, βS }: we will use equations (3.23) and (3.24) above

to calibrate the discount factors. Solving for βS , we will use initial and end period data

on sectoral employment shares and productivity to back out βS .

φ φ
ST sTS − S0 s0S
βS = . (3.25)
XM pM 
h i
φ −1
φS ST sTS mT − MTT − p0S M0 m0 − X0M
0

Given βS , we then get γS using

pM
0
βS M0 m0 − X0M − S0 s0S .
  φ
γS = S
(3.26)
p0

119
 Solving for βA , with the simplifying approximation,

mt+1 + st+1 − mt − st = mt + st − mt−1 − st−1 .

We will use initial and end period data on sectoral employment shares and produc-

tivity to back out βA .

YTA − XTA − Y0A − X0A

 
βA = .
YTM −XTM pM 
h i
φ−1 r
MT φAMTθ aT + 1+r κMTθ (mT + sT − mT −1 − sT −1 ) − p0A Y0M − X0M
0
(3.27)

Given βA , we then get γA using

h i pM 
γA = Y0A − X0A − 0A βA Y0M − X0M .

(3.28)
p0

3.5 Sectoral Relative Prices: International Comparison

Program (ICP) 2005

The sectoral relative prices are constructed based on the detail level data from the

International Comparison Program (ICP) 2005.18 Figure 3.2 displays the relative price

of manufacturing to services sector against PPP real GDP per capita (in natural log
18 Thediscussion on the construction of sectoral relative prices (World Bank ICP) is based on: Kural-
bayeva and Stefanski (2013), Adamopoulos (2009), The World Bank International Comparison Program
Handbook (World Bank (2005)), Buera et al. (2011) and Deaton and Hestor (2010).

120
term) and Figure 3.3 plots relative price of manufacturing to agriculture sector over

PPP real GDP per capita (in natural log term).

Figure 3.2: Sectoral Relative Prices (Manufacturing/Services) and Income

3.6 Conclusion

We introduced a calibrated three-sector endogenous growth model with sectoral

labor adjustment costs. The model generates a hump-shaped in Manufacturing em-

ployment share. We introduce a key feature of leaning-by-doing in Manufacturing

121
Figure 3.3: Sectoral Relative Prices (Manufacturing/Agriculture) and Income

122
sector to a standard three-sector structural transformation model with labor market fric-

tion, which is new to the literature. Labor market friction plays an important role in

generating a novel feature of structural changes, humped-shaped in Manufacturing em-

ployment. In our model, opening to trade can accelerate or decrease structural trans-

formation (labor movement into the Manufacturing sector) depending on comparative

advantage, whether the economy specializes in Agriculture or non-agriculture goods

after opening to trade.

3.7 Extensions

For future work, we plan to apply the model in the context of income distribution

and inequality. In particular, it would be interesting to analyze the interaction between

labor market frictions (barriers to labor movement across sectors) and international

trade. We hypothesize that opening to trade could increase or decrease income in-

equality depending on specialization upon trade opening.19 It is important to note that

income inequality, dispersion in expenditure shares, is generated via preferences in our

model and it may also be generated by the supply side McCalman (2018). In addition,

the model’s key feature is the dynamic productivity growth in manufacturing sector via

learning-by-doing and international trade could have implication for aggregate growth
19 For
sectoral wages (compensations) data, we plan to utilize data from the IPUMS International of
Minnesota Population Center at the University of Minnesota as well as EU KLEMS at the Groningen
Growth and Development Centre.

123
depending on weather trade expands or depress the dynamic engine of growth sector.

3.8 Appendix

The appendix for this chapter is provided at the end of the dissertation (Appendix

C).

124
Bibliography

Abiad, A., Detragiache, E., and Tressel, T. A new database of financial reforms. IMF

Staff Papers, 57(2):281–302, 2010. ISSN 1564-5150.

Adamopoulos, T. Do richer countries have higher distribution margins? Unpublished

Manuscript, York University, 2009.

Adnan, N. Measurement of financial development: a fresh approach. Eight interna-

tional conference on islamic economics and finance., University of Surrey and Com-

sats Institute of Information Technology, United Kingdom and Islamabad, Pakistan,

2011.

Aizenman, J., Pinto, B., and Radziwill, A. Sources for financing domestic capital is

foreign saving a viable option for developing countries? Journal of International

Money and Finance, 26(5):682 – 702, 2007. ISSN 0261-5606.

Aizenman, J., Chinn, M. D., and Ito, H. The impossible trinity hypothesis in an era of

125
 global imbalances: Measurement and testing. Review of International Economics,

21(3):447–458, 2013a. ISSN 1467-9396.

Aizenman, J., Pinto, B., and Sushko, V. Financial sector ups and downs and the real

sector in the open economy: Up by the stairs, down by the parachute. Emerging

Markets Review, 16:1 – 30, 2013b. ISSN 1566-0141.

Aizenman, J., Jinjarak, Y., and Park, D. Financial development and output growth in

developing asia and latin america: A comparative sectoral analysis. Working Paper

20917, National Bureau of Economic Research, January 2015.

Alberola, E. and Benigno, G. Revisiting the commodity curse: A financial perspective.

Journal of International Economics, 108(Supplement 1):S87 – S106, 2017. ISSN

0022-1996. 39th Annual NBER International Seminar on Macroeconomics.

Alcalá, F. and Ciccone, A. Trade and productivity. The Quarterly Journal of Eco-

nomics, 119(2):613–646, 2004.

Angrist, J. D. and Pischke, J.-S. Mostly Harmless Econometrics: An Empiricist’s Com-

panion. Number 8769 in Economics Books. Princeton University Press, 12-2014

2009. ISBN ARRAY(0x757654e0).

Arcand, J. L., Berkes, E., and Panizza, U. Too much finance? Journal of Economic

Growth, 20(2):105–148, 2015a. ISSN 1381-4338.

126
Arcand, J. L., Berkes, E., and Panizza, U. Too Much Finance or Statistical Illusion:

A Comment. IHEID Working Papers 12-2015, Economics Section, The Graduate

Institute of International Studies, June 2015b.

Arellano, M. and Bond, S. Some tests of specification for panel data: Monte carlo

evidence and an application to employment equations. The Review of Economic

Studies, 58(2):277–297, 1991.

Arellano, M. and Bover, O. Another look at the instrumental variable estimation of

error-components models. Journal of Econometrics, 68(1):29 – 51, 1995. ISSN

0304-4076.

Artuc, E. and McLaren, J. A structural empirical approach to trade shocks and labor

adjustment: An application to turkey. In Porto, G. and Hoekman, B. M., editors,

Trade adjustment Costs in Developing Countries : Impacts, Determinants and Policy

Responses. Washington, DC, 2009.

Artuc, E., Bet, G., Brambilla, I., and Porto, G. Trade shocks, firm-level investment

decisions, and labor market responses. 2013.

Artuc, E., Lederman, D., and Porto, G. A mapping of labor mobility costs in the

developing world. Journal of International Economics, 95(1):28 – 41, 2015. ISSN

0022-1996.

127
Athukorala, P.-C. Global production sharing and trade patterns in east asia. In Kaur, I.

and Singh, N., editors, The Oxford Handbook of the Economics of the Pacific Rim,

pages 333–361. Oxford University Press, New York, 2014.

Benhabib, J. and Spiegel, M. M. The role of financial development in growth and

investment. Journal of Economic Growth, 5(4):341–360, 2000. ISSN 1573-7020.

Benigno, G. and Fornaro, L. The financial resource curse. The Scandinavian Journal

of Economics, 116(1):58–86, 2014a. ISSN 1467-9442.

Benigno, G. and Fornaro, L. The financial resource curse*. The Scandinavian Journal

of Economics, 116(1):58–86, 2014b. ISSN 1467-9442.

Benigno, G., Converse, N., and Fornaro, L. Large capital inflows, sectoral allocation,

and economic performance. Journal of International Money and Finance, 55:60 –

87, 2015. ISSN 0261-5606. Macroeconomic and financial challenges facing Latin

America and the Caribbean after the crisis.

Bhagwati, J. The capital myth: the difference between trade in widgets and dollars.

Foreign Affairs, 77(3):7–12, 1998.

Blundell, R. and Bond, S. Initial conditions and moment restrictions in dynamic panel

data models. Journal of Econometrics, 87(1):115 – 143, 1998. ISSN 0304-4076.

128
Bonfiglioli, A. Financial integration, productivity and capital accumulation. Journal of

International Economics, 76(2):337 – 355, 2008. ISSN 0022-1996.

Borensztein, E. and Loungani, P. Asian Financial Integration; Trends and Interruptions.

IMF Working Papers 11/4, International Monetary Fund, January 2011.

Buera, F. J. and Kaboski, J. P. Can traditional theories of structural change fit the data?

Journal of the European Economic Association, 7(2-3):469–477, 2009.

Buera, F. J., Kaboski, J. P., and Shin, Y. Finance and development: A tale of two

sectors. American Economic Review, 101(5):1964–2002, 2011.

Bun, L. and Singh, N. Heterogeneous patterns of financial development: Implica-

tions for asian financial integration. International Economic Journal, 30(2):243–271,

2016.

Caballero, J. A. Do surges in international capital inflows influence the likelihood of

banking crises? The Economic Journal, 126(591):281–316, 2016. ISSN 1468-0297.

Cai, W. Structural change accounting with labor market distortions. Journal of Eco-

nomic Dynamics and Control, 57:54 – 64, 2015. ISSN 0165-1889.

Caliendo, L. and Parro, F. Estimates of the trade and welfare effects of nafta. The

Review of Economic Studies, 82(1):1–44, 2015.

129
Caselli, F. and Coleman II, W. J. The us structural transformation and regional conver-

gence: A reinterpretation. Journal of Political Economy, 109(3):584–616, 2001.

Choi, S. M., Kim, H., and Ma, X. Trade, structural transformation, and growth, imf

working paper. Technical report, WP, January, 2014.

Cihak, M., Demirgüç-Kunt, A., Feyen, E., and Levine, R. Benchmarking financial

development around the world. World Bank Policy Research Working Paper, 6175:

1–58, 2012.

Claessens, S. and Laeven, L. Financial development, property rights, and growth. The

Journal of Finance, 58(6):2401–2436, 2003. ISSN 1540-6261.

Cline, W. R. Too Much Finance, or Statistical Illusion? Policy Briefs PB15-9, Peterson

Institute for International Economics, June 2015.

Cooper, R. N. Should Capital Controls be Banished? Brookings Papers on Economic

Activity, 30(1):89–142, 1999.

Davis, J. S., Mack, A., Phoa, W., and Vandenabeele, A. Credit booms, banking crises,

and the current account. Journal of International Money and Finance, 60:360 – 377,

2016. ISSN 0261-5606.

Deaton, A. and Hestor, A. Understanding ppps and ppp-based national accounts. Amer-

ican economic journal. Macroeconomics, 2(4):1–35, 2010.

130
Demetriades, P. and Hook Law, S. Finance, institutions and economic development.

International Journal of Finance & Economics, 11(3):245–260, 2006. ISSN 1099-

1158.

Demetriades, P. O. and Rousseau, P. L. The changing face of financial development.

Economics Letters, 141:87 – 90, 2016. ISSN 0165-1765.

Dolores Guilló, M., Papageorgiou, C., and Pérez-Sebastián, F. A unified theory of

structural change. Journal of Economic Dynamics and Control, 35(9):1393–1404,

2011.

Duarte, M. and Restuccia, D. The role of the structural transformation in aggregate

productivity. The Quarterly Journal of Economics, 125(1):129–173, 2010.

Eaton, J. and Kortum, S. Technology, geography, and trade. Econometrica, 70(5):

1741–1779, 2002. ISSN 1468-0262.

Eichengreen, B. What to do with the chiang mai initiative. Asian Economic Papers, 2

(1):1–49, 2003.

Felbermayr, G. and Gröschl, J. Natural disasters and the effect of trade on income:

A new panel iv approach. European Economic Review, 58(Supplement C):18 – 30,

2013. ISSN 0014-2921.

131
Feldstein, M. and Horioka, C. Domestic saving and international capital flows. The

Economic Journal, 90(358):pp. 314–329, 1980. ISSN 00130133.

Fisman, R. and Love, I. Financial development and intersectoral allocation: A new

approach. The Journal of Finance, 59(6):2785–2807, 2004. ISSN 1540-6261.

Frankel, J. A. and Romer, D. Does trade cause growth? American economic review,

pages 379–399, 1999.

Fung, L., Tam, C.-s., and Yu, I.-w. Assessing the integration of asia’s equity and bond

markets. BIS paper, (42), 2008.

Garcı́a-Herrero, A., Yang, D.-y., and Wooldridge, P. Why is there so little regional

financial integration in asia? Press & Communications CH 4002 Basel, Switzerland,

page 38, 2008.

Ghosh, A., Goretti, M., Joshi, B., Ramakrishnan, U., Thomas, A., and Zalduendo,

J. Capital Inflows and Balance of Payments Pressures: Tailoring Policy Responses

in Emerging Market Economies. IMF Policy Discussion Papers 08/2, International

Monetary Fund, June 2008.

Gollin, D. The lewis model: A 60-year retrospective. Journal of Economic Perspec-

tives, 28(3):71–88, 2014.

132
Gollin, D. and Rogerson, R. Productivity, transport costs and subsistence agriculture.

Journal of Development Economics, 107(0):38 – 48, 2014. ISSN 0304-3878.

Henry, P. B. Capital account liberalization: Theory, evidence, and speculation. Journal

of Economic Literature, 45(4):887–935, 2007.

Herwartz, H. and Walle, Y. M. Determinants of the link between financial and economic

development: Evidence from a functional coefficient model. Economic Modelling,

37:417 – 427, 2014a. ISSN 0264-9993.

Herwartz, H. and Walle, Y. M. Openness and the finance-growth nexus. Journal of

Banking & Finance, 48:235 – 247, 2014b. ISSN 0378-4266.

Hoxha, I., Kalemli-Ozcan, S., and Vollrath, D. How big are the gains from interna-

tional financial integration? Working Paper 14636, National Bureau of Economic

Research, January 2009.

Hoxha, I., Kalemli-Ozcan, S., and Vollrath, D. How big are the gains from international

financial integration? Journal of Development Economics, 103(Supplement C):90 –

98, 2013. ISSN 0304-3878.

Hsieh, C.-T. and Klenow, P. J. Relative prices and relative prosperity. American Eco-

nomic Review, 97(3):562–585, 2007.

133
Huang, D.-S., Huang, Y.-Y., and Sun, Y.-C. Production specialization and trade blocs.

Journal of Economic Integration, 21(3):474–495, 2006. ISSN 1225651X.

International Monetary Funds (IMF). Regional economic outlook: Asia and pacific

stabilizing and outperforming other regions. World economic and financial surveys,

april, International Monetary Funds (IMF), 2015.

Johnson, D. K. N. and Evenson, R. E. R&d spillovers to agriculture: measurement

and application. Contemporary Economic Policy, 17(4):432–456, 1999. ISSN 1465-

7287.

Kaminsky, G. L. and Schmukler, S. L. Short-run pain, long-run gain: Financial liberal-

ization and stock market cycles. Review of Finance, 2008.

Kannan, P. Credit conditions and recoveries from financial crises. Journal of Interna-

tional Money and Finance, 31(5):930 – 947, 2012. ISSN 0261-5606.

Kaur, I. and Singh, N. Financial integration and financial development in east asia.

Millennial Asia, 5(1):1–22, 2014.

Kaur, I. N. Are the geese still flying? catch-up industrialization in a changing in-

ternational economic environment. In Kaur, I. and Singh, N., editors, The Oxford

Handbook of the Economics of the Pacific Rim, pages 333–361. Oxford University

Press, New York, 2014.

134
Kose, M. A., Prasad, E. S., and Terrones, M. E. Does openness to international financial

flows raise productivity growth? Journal of International Money and Finance, 28

(4):554 – 580, 2009. ISSN 0261-5606. Emerging Market Finance.

Kuralbayeva, K. and Stefanski, R. Windfalls, structural transformation and specializa-

tion. Journal of International Economics, 90(2):273–301, 2013.

Lane, P. R. and McQuade, P. Domestic credit growth and international capital flows*.

The Scandinavian Journal of Economics, 116(1):218–252, 2014. ISSN 1467-9442.

Law, S. H. and Singh, N. Does too much finance harm economic growth? Journal of

Banking & Finance, 41:36 – 44, 2014. ISSN 0378-4266.

Law, S. H., Azman-Saini, W., and Ibrahim, M. H. Institutional quality thresholds and

the finance growth nexus. Journal of Banking & Finance, 37(12):5373 – 5381, 2013.

ISSN 0378-4266.

Lee, H.-H., Yi, I., and Park, D. Impact of the global financial crisis on the degree of

financial integration among east asian countries. Global Economic Review, 42(4):

425–459, 2013.

Levchenko, A. A., Rancire, R., and Thoenig, M. Growth and risk at the industry level:

The real effects of financial liberalization. Journal of Development Economics, 89

(2):210 – 222, 2009. ISSN 0304-3878. New Approaches to Financial Globalization.

135
Levine, R. Chapter 12 finance and growth: Theory and evidence. volume 1, Part A of

Handbook of Economic Growth, pages 865 – 934. Elsevier, 2005.

Lewis, W. A. Economic development with unlimited supplies of labour. The manch-

ester school, 22(2):139–191, 1954.

Lim, M.-H. and Lim, J. Asian initiatives at monetary and financial integration: A

critical review. Background paper no. 4, unctad program on the rise of the south and

new paths of development in the 21st century, march, 2012.

Lucas, R. E. Trade and the diffusion of the industrial revolution. American Economic

Journal: Macroeconomics, 1(1):1–25, 2009.

Lucey, B. M. and Zhang, Q. Financial integration and emerging markets capital struc-

ture. Journal of Banking & Finance, 35(5):1228 – 1238, 2011. ISSN 0378-4266.

Matsuyama, K. Agricultural productivity, comparative advantage, and economic

growth. Journal of economic theory, 58(2):317–334, 1992.

McCalman, P. International trade, income distribution and welfare. Journal of Interna-

tional Economics, 110(Supplement C):1 – 15, 2018. ISSN 0022-1996.

McMillan, M. S. and Rodrik, D. Globalization, structural change and productivity

growth. Working Paper 17143, National Bureau of Economic Research, June 2011.

136
Melitz, M. J. and Trefler, D. Gains from trade when firms matter. Journal of Economic

Perspectives, 26(2):91–118, 2012.

Obstfeld, M. International finance and growth in developing countries: What have we

learned? IMF Staff Papers, 56(1):63–111, print 2009.

Ostry, J. D., Ghosh, A. R., Chamon, M., and Qureshi, M. S. Tools for managing

financial-stability risks from capital inflows. Journal of International Economics, 88

(2):407 – 421, 2012. ISSN 0022-1996. {NBER} Global.

Papageorgiou, C. and Spatafora, N. Economic diversification in lics: Stylized facts

and macroeconomic implications, imf staff discussion note. Technical report,

SDN/12/13, 2012.

Park, C.-Y. and Lee, J.-W. Financial integration in emerging asia: Challenges and

prospects. Asian Economic Policy Review, 6(2):176–198, 2011. ISSN 1748-3131.

Pongsaparn, R. and Unteroberdoerster, O. Financial integration and rebalancing in asia.

IMF Working Papers, pages 1–31, 2011.

Popov, A. Credit constraints, equity market liberalization, and growth rate asymmetry.

Journal of Development Economics, 107:202 – 214, 2014. ISSN 0304-3878.

Porto, G. and Hoekman, B. M. Trade adjustment costs in developing countries : Im-

137
 pacts, determinants and policy responses. Technical report, World Bank, Washing-

ton, DC, 2010.

Prati, A., Onorato, M. G., and Papageorgiou, C. Which reforms work and under what

institutional environment? evidence from a new data set on structural reforms. Re-

view of Economics and Statistics, 95(3):946–968, 2016/11/21 2012.

Rajan, R. G. and Zingales, L. Financial dependence and growth. The American Eco-

nomic Review, 88(3):pp. 559–586, 1998. ISSN 00028282.

Rajan, R. G. and Zingales, L. The great reversals: the politics of financial development

in the twentieth century. Journal of Financial Economics, 69(1):5 – 50, 2003. ISSN

0304-405X. Tuck Symposium on Corporate Governance.

Reis, R. The Portugese Slump and Crash and the Euro Crisis. Brookings Papers on

Economic Activity, 44(1 (Spring):143–210, 2013.

Restuccia, D. and Rogerson, R. Misallocation and productivity. Review of Economic

Dynamics, 16(1):1 – 10, 2013. ISSN 1094-2025. Special issue: Misallocation and

Productivity.

Restuccia, D., Yang, D. T., and Zhu, X. Agriculture and aggregate productivity: A

quantitative cross-country analysis. Journal of Monetary Economics, 55(2):234 –

250, 2008. ISSN 0304-3932.

138
Rey, H. Dilemma not trilemma: the global cycle and monetary policy independence.

Proceedings - Economic Policy Symposium - Jackson Hole, pages 1–2, 2013.

Rodrik, D. Who needs capital-account convertibility? in should the imf pursue capital-

account convertibility, essays in international finance, 207, princeton university. ro-

drik, dani (1999): Why do more open economies have bigger governments. Survival,

35:27–47, 1998.

Roodman, D. How to do xtabond2: An introduction to difference and system gmm in

stata. Stata Journal, 9(1):86–136(51), 2009a.

Roodman, D. XTABOND2: Stata module to extend xtabond dynamic panel data esti-

mator. Statistical Software Components, Boston College Department of Economics,

November 2003.

Roodman, D. A note on the theme of too many instruments*. Oxford Bulletin of

Economics and Statistics, 71(1):135–158, 2009b. ISSN 1468-0084.

Rousseau, P. L. and Wachtel, P. What is happening to the impact of financial deepening

on economic growth? Economic Inquiry, 49(1):276–288, 2011. ISSN 1465-7295.

Sachs, J. D., Warner, A., Åslund, A., and Fischer, S. Economic reform and the process

of global integration. Brookings papers on economic activity, pages 1–118, 1995.

139
Saci, K. and Holden, K. Evidence on growth and financial development using principal

components. Applied Financial Economics, 18(19):1549–1560, 2008.

Sahay, R., Cihak, M., N’Diaye, P., Barajas, A., Pena, D. A., Bi, R., Gao, Y., Kyobe, A.,

Nguyen, L., Saborowski, C., Svirydzenka, K., and Yousefi, R. Rethinking financial

deepening: Stability and growth in emerging markets. IMF Staff Discussion Notes

15/8, International Monetary Fund, May 2015.

Samarina, A. and Bezemer, D. Do capital flows change domestic credit allocation?

Journal of International Money and Finance, 62:98 – 121, 2016. ISSN 0261-5606.

Schularick, M. and Taylor, A. M. Credit booms gone bust: Monetary policy, leverage

cycles, and financial crises, 1870-2008. American Economic Review, 102(2):1029–

61, 2012.

Seo, M. H. and Shin, Y. Dynamic panels with threshold effect and endogeneity. Journal

of Econometrics, 195(2):169 – 186, 2016. ISSN 0304-4076.

Song, Z., Storesletten, K., and Zilibotti, F. Growing like china. American Economic

Review, 101(1):196–233, 2011.

Stiglitz, J. E. Some lessons from the east asian miracle. The World Bank research

observer, 11(2):151 – 177, 1996.

Stiglitz, J. E. Globalization and Its Discontents. Norton, New York, 2003.

140
Svirydzenka, K. Introducing a new broad-based index of financial development. 2016.

Teignier, M. The role of trade in structural transformation. Journal of Development

Economics, 130(Supplement C):45 – 65, 2018. ISSN 0304-3878.

Thompson, P. Learning by doing. Handbook of the Economics of Innovation, 1:429–

476, 2010.

Ventura, J. Growth and interdependence. The Quarterly Journal of Economics, 112(1):

57–84, 1997.

Windmeijer, F. A finite sample correction for the variance of linear efficient two-step

gmm estimators. Journal of Econometrics, 126(1):25 – 51, 2005. ISSN 0304-4076.

Wooldridge, J. M. Econometric Analysis of Cross Section and Panel Data, volume 1 of

MIT Press Books. The MIT Press, March 2010. ISBN 9780262232586.

World Bank. The east asian miracle : economic growth and public policy. Technical

report, World Bank, New York, New York, 1993.

World Bank. Global purchasing power parities and real expenditures: 2005 interna-

tional comparison program. Methodological handbook, World Bank, Washington,

DC, 2005.

World Bank. Global financial development report 2013: Rethinking the role of the state

in finance. Technical report, World Bank, Washington, DC, 2012.

141
World Bank. Global financial development report 2014: Financial inclusion. Technical

report, World Bank, Washington, DC, 2014.

World Bank. World Development Indicators 2016. The World Bank, 2016. URL

https://data.worldbank.org/products/wdi.

Yang, T.-H. and Huang, D.-S. Multinational enterprises, foreign direct investment, and

the east asian economic integration. In Kaur, I. and Singh, N., editors, The Oxford

Handbook of the Economics of the Pacific Rim, pages 333–361. Oxford University

Press, New York, 2014.

142
Appendix A

Financial Development and Growth:

Multi-dimensional View

A.1 Appendix

A.1.1 Data Descriptions and Sources

Table A.1: Data Description and Sources

Variables Descriptions and Sources

Annual percentage growth rate of GDP at market prices based on constant local cur-
rency. Aggregates are based on constant 2010 U.S. dollars. GDP is the sum of gross
value added by all resident producers in the economy plus any product taxes and minus
Growth any subsidies not included in the value of the products. It is calculated without mak-
ing deductions for depreciation of fabricated assets or for depletion and degradation of
natural resources. Code: GDP growth (annual %)(NY.GDP.MKTP.KD.ZG). Source:
WDI (World Bank, 2016)
(Continued on next page)

143
 Table A.1 – (Continued from previous page)
Variables Descriptions and Sources
The financial resources provided to the private sector by domestic money banks as a
share of GDP. Domestic money banks comprise commercial banks and other financial
PC institutions that accept transferable deposits, such as demand deposits. Code: Private
credit by deposit money banks to GDP (%)(GFDD.DI.01). Source: GFDD (Cihak
et al., 2012)
Difference between lending rate and deposit rate. Lending rate is the rate charged
by banks on loans to the private sector and deposit interest rate is the rate of-
LDSPREAD
fered by commercial banks on three-month deposits. Code: Bank lending-deposit
spread(GFDD.EI.02). Source: GFDD (Cihak et al., 2012)
Average years of total schooling of males and females, 25+, total is the average years
of education completed among people over age 25. Code: Barro-Lee: Average years
LEDU
of total schooling, age 25+, total(BAR.SCHL.25UP) (analyses use log of the variable).
Source: WDI (World Bank, 2016)
General government final consumption expenditure as a percentage of GDP: includes
all government current expenditures for purchases of goods and services (including
compensation of employees). It also includes most expenditures on national defense
LGC
and security, but excludes government military expenditures that are part of govern-
ment capital formation. Code: General government final consumption expenditure (%
of GDP) (analyses use log of the variable). Source: WDI (World Bank, 2016)
Trade openness: sum of exports and imports of goods and services measured as a share
LOPEN of GDP (analyses use log of the variable). Code: Trade (% of GDP). Source: WDI
(World Bank, 2016)
Inflation as measured by the consumer price index reflects the annual percentage
change in the cost to the average consumer of acquiring a basket of goods and ser-
vices that may be fixed or changed at specified intervals, such as yearly. The Laspeyres
formula is generally used (analyses use log of the variable). Code: Inflation, consumer
LINFL prices (annual %)(FP.CPI.TOTL.ZG). Following Arcand et al. (2015a), we drop all
observations for which inflation is less than 10% and observations with negative infla-
tion values are 
set to zero. Also, we apply
 the inverse hyperbolic sine transformation
 √
LINFL = log INFL + INFL2 + 1 . Source: WDI (World Bank, 2016)

A.1.2 Data Summary

Table A.2: Credit to the Private Sector Over GDP (Selected Years)

Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
Afghanistan 0.03 0.06 Albania 0.04 0.06 0.38
(Continued on next page)

144
 Table A.2 – (Continued from previous page)
Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
Algeria 0.55 0.26 0.09 0.15 American Samoa
Andorra Angola 0.03 0.19
Antigua and Barbuda 0.36 0.48 0.63 0.73 Argentina 0.19 0.13 0.17 0.12
Armenia 0.21 0.07 0.36 Aruba 0.40 0.48 0.60
Australia 0.25 0.59 0.87 1.29 Austria 0.72 0.88 0.90 0.94
Azerbaijan 0.06 0.06 0.20 Bahamas, The 0.40 0.53 0.57 0.81
Bahrain 0.38 0.49 0.50 0.97 Bangladesh 0.07 0.15 0.26 0.39
Barbados 0.36 0.45 0.54 Belarus 0.18 0.08 0.27
Belgium 0.27 0.56 0.66 0.55 Belize 0.31 0.37 0.56 0.58
Benin 0.29 0.14 0.12 0.23 Bermuda
Bhutan 0.02 0.06 0.12 0.48 Bolivia 0.12 0.32 0.51 0.41
Bosnia and 0.33 0.52 Botswana 0.13 0.12 0.16 0.28
Herzegovina
Brazil 0.27 0.32 0.29 0.67 Brunei Darussalam 0.62 0.34
Bulgaria 0.60 0.18 0.69 Burkina Faso 0.20 0.12 0.12 0.20
Burundi 0.10 0.14 0.25 0.18 Cambodia 0.03 0.06 0.35
Cameroon 0.27 0.19 0.08 0.13 Canada 0.75 0.92 1.47
Cape Verde 0.03 0.08 0.39 0.62 Cayman Islands
Central African 0.11 0.06 0.05 0.11 Chad 0.15 0.06 0.03 0.05
Republic
Channel Islands Chile 0.50 0.50 0.68 0.99
China 0.76 1.09 1.26 Colombia 0.30 0.24 0.22 0.45
Comoros 0.13 0.15 0.09 0.20 Congo, Dem. Rep. 0.00 0.00 0.01 0.04
Congo, Rep. 0.15 0.13 0.04 0.08 Costa Rica 0.20 0.12 0.26 0.47
Cote d’Ivoire 0.47 0.33 0.14 0.17 Croatia 0.25 0.38 0.69
Cuba Curacao
Cyprus 0.56 1.04 1.57 2.47 Czech Republic 0.69 0.35 0.49
Denmark 0.23 0.41 1.29 1.87 Djibouti 0.43 0.26 0.30
Dominica 0.32 0.48 0.46 0.54 Dominican Republic 0.29 0.18 0.28 0.22
Ecuador 0.20 0.15 0.23 0.25 Egypt, Arab Rep. 0.21 0.22 0.52 0.28
El Salvador 0.28 0.23 0.76 1.02 Equatorial Guinea 0.17 0.03 0.09
Eritrea 0.25 0.14 Estonia 0.11 0.41 0.78
Ethiopia 0.15 0.15 0.21 Faroe Islands
Fiji 0.21 0.36 0.37 0.71 Finland 0.47 0.85 0.55 0.90
France 0.69 0.92 0.76 0.95 French Polynesia
Gabon 0.15 0.12 0.10 0.10 Gambia, The 0.21 0.11 0.08 0.15
Georgia 0.07 0.34 Germany 0.83 0.91 1.11 0.84
Ghana 0.02 0.04 0.11 0.14 Greece 0.40 0.31 0.50 1.16
Greenland Grenada 0.32 0.49 0.59 0.79
Guam Guatemala 0.17 0.13 0.22 0.27
Guinea 0.04 0.03 0.06 Guinea-Bissau 0.03 0.02 0.09
Guyana 0.25 0.20 0.58 0.39 Haiti 0.12 0.12 0.14 0.15
Honduras 0.29 0.26 0.35 0.50 Hong Kong SAR, 1.35 1.48 1.93
China
Hungary 0.16 0.35 0.34 0.53 Iceland 0.28 0.44 1.03 1.33
(Continued on next page)

145
 Table A.2 – (Continued from previous page)
Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
India 0.22 0.23 0.29 0.48 Indonesia 0.11 0.43 0.19 0.30
Iran, Islamic Rep. 0.35 0.28 0.29 0.50 Iraq 0.06
Ireland 0.45 0.44 0.81 1.18 Isle of Man
Israel 0.45 0.54 0.77 0.75 Italy 0.50 0.57 0.64 0.92
Jamaica 0.25 0.21 0.18 0.27 Japan 1.24 1.73 1.89 1.79
Jordan 0.52 0.65 0.72 0.69 Kazakhstan 0.16 0.16 0.36
Kenya 0.23 0.19 0.25 0.32 Kiribati
Korea, Dem. Rep. Korea, Rep. 0.47 0.51 0.82 1.04
Kosovo 0.05 0.34 Kuwait 0.63 0.30 0.57 0.65
Kyrgyz Republic 0.04 0.13 Lao PDR 0.04 0.07 0.19
Latvia 0.14 0.28 0.72 Lebanon 1.34 0.45 0.79 0.83
Lesotho 0.12 0.16 0.12 0.16 Liberia 0.04 0.15
Libya 0.19 0.13 Liechtenstein
Lithuania 0.14 0.17 0.48 Luxembourg 1.16 1.10 0.76 0.91
Macao SAR, China 0.58 0.69 0.60 0.55 Macedonia, FYR 0.37 0.17 0.46
Madagascar 0.19 0.15 0.08 0.11 Malawi 0.18 0.11 0.05 0.13
Malaysia 0.57 0.86 1.19 1.08 Maldives 0.15 0.10 0.18 0.43
Mali 0.19 0.11 0.17 0.20 Malta 0.35 0.82 1.05 1.12
Marshall Islands Mauritania 0.27 0.32 0.27
Mauritius 0.21 0.35 0.58 0.93 Mexico 0.14 0.23 0.16 0.26
Micronesia, Fed. Sts. 0.24 0.20 Moldova 0.04 0.15 0.36
Monaco Mongolia 0.08 0.13 0.46
Montenegro 0.10 0.58 Morocco 0.16 0.21 0.46 0.68
Mozambique 0.13 0.12 0.24 Myanmar 0.05 0.05 0.07 0.10
Namibia 0.29 0.42 0.46 Nepal 0.08 0.13 0.26 0.53
Netherlands 0.62 0.81 1.14 1.16 New Caledonia
New Zealand 0.18 0.76 1.03 1.39 Nicaragua 0.39 0.24 0.23 0.27
Niger 0.17 0.11 0.05 0.13 Nigeria 0.16 0.11 0.13 0.13
Northern Mariana Norway 0.49 0.74 0.80
Islands
Oman 0.14 0.21 0.37 0.41 Pakistan 0.21 0.22 0.22 0.17
Palau Panama 0.48 0.45 0.86 0.69
Papua New Guinea 0.18 0.22 0.15 0.30 Paraguay 0.18 0.18 0.19 0.38
Peru 0.13 0.08 0.22 0.27 Philippines 0.41 0.25 0.35 0.32
Poland 0.55 0.18 0.27 0.50 Portugal 0.63 0.49 1.15 1.51
Puerto Rico Qatar 0.27 0.39
Romania 0.45 0.50 0.09 0.43 Russian Federation 0.06 0.16 0.46
Rwanda 0.05 0.06 0.09 Samoa 0.07 0.15 0.32 0.64
San Marino Sao Tome and 0.07 0.32
Principe
Saudi Arabia 0.34 0.53 0.53 0.51 Senegal 0.33 0.25 0.18 0.28
Serbia 0.22 0.46 Seychelles 0.16 0.10 0.26 0.23
Sierra Leone 0.06 0.03 0.02 0.06 Singapore 0.84 0.87 1.08 1.11
Sint Maarten (Dutch Slovak Republic 0.49 0.38 0.46
part)
(Continued on next page)

146
 Table A.2 – (Continued from previous page)
Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
Slovenia 0.24 0.37 0.77 Solomon Islands 0.25 0.20 0.21 0.29
Somalia South Africa 0.57 0.82 1.23 1.45
South Sudan 0.02 Spain 0.72 0.77 0.97 1.56
Sri Lanka 0.18 0.11 0.27 0.27 St. Kitts and Nevis 0.43 0.63 0.54 0.64
St. Lucia 0.41 0.51 0.72 1.08 St. Martin (French
part)
St. Vincent and the 0.37 0.40 0.50 0.52 Sudan 0.13 0.05 0.03 0.10
Grenadines
Suriname 0.32 0.28 0.11 0.25 Swaziland 0.23 0.16 0.13 0.21
Sweden 0.78 1.15 0.82 1.26 Switzerland 1.06 1.46 1.43 1.63
Syrian Arab 0.06 0.08 0.08 0.20 Tajikistan 0.11 0.15
Republic
Tanzania 0.02 0.11 0.06 0.15 Thailand 0.44 0.91 1.02 1.31
Timor-Leste 0.06 0.12 Togo 0.25 0.24 0.14 0.27
Tonga 0.15 0.29 0.40 0.34 Trinidad and Tobago 0.33 0.41 0.36 0.37
Tunisia 0.45 0.58 0.66 0.71 Turkey 0.15 0.14 0.14 0.54
Turkmenistan Turks and Caicos
Islands
Tuvalu Uganda 0.03 0.03 0.07 0.15
Ukraine 0.01 0.15 0.71 United Arab 0.27 0.41 0.36 0.64
Emirates
United Kingdom 0.31 1.00 1.26 1.68 United States 0.90 1.16 1.66 1.74
Uruguay 0.41 0.21 0.48 0.23 Uzbekistan
Vanuatu 0.34 0.35 0.36 0.66 Venezuela, RB 0.56 0.21 0.10 0.20
Vietnam 0.14 0.40 0.95 Virgin Islands (U.S.)
West Bank and Gaza 0.22 0.25 Yemen, Rep. 0.04 0.05 0.06
Zambia 0.05 0.06 0.14 Zimbabwe 0.13 0.23 0.18

Table A.3: Bank Lending and Deposit Rate Spreads (Selected Years)

Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
Afghanistan Albania 0.03 0.09 0.06
Algeria 0.04 0.03 0.06 American Samoa
Andorra Angola 0.59 0.12
Antigua and Barbuda 0.04 0.07 0.07 0.07 Argentina 0.02 0.08 0.03
Armenia 0.13 0.08 Aruba 0.05 0.07 0.07
Australia 0.01 0.03 0.05 0.03 Austria
Azerbaijan 0.08 0.08 Bahamas, The 0.04 0.02 0.02 0.03
Bahrain 0.06 0.07 0.05 Bangladesh 0.03 0.05 0.08 0.03
Barbados 0.05 0.05 0.06 0.06 Belarus 0.33 0.13 0.00
Belgium 0.08 0.06 0.05 Belize 0.05 0.06 0.08 0.07
Benin 0.08 0.09 Bermuda
(Continued on next page)

147
 Table A.3 – (Continued from previous page)
Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
Bhutan 0.10 0.09 0.09 0.10 Bolivia 0.14 0.25 0.12 0.09
Bosnia and 0.09 0.04 Botswana 0.04 0.02 0.06 0.07
Herzegovina
Brazil 0.42 0.27 Brunei Darussalam 0.04 0.05
Bulgaria 0.32 0.07 0.07 Burkina Faso 0.08 0.09
Burundi 0.08 Cambodia
Cameroon 0.06 0.10 0.14 Canada 0.03 0.04 0.04 0.02
Cape Verde 0.06 0.08 0.07 Cayman Islands
Central African 0.05 0.09 0.14 Chad 0.06 0.10 0.14
Republic
Channel Islands Chile 0.12 0.06 0.04 0.04
China 0.01 0.01 0.03 0.03 Colombia 0.10 0.07 0.07
Comoros 0.07 0.09 0.09 Congo, Dem. Rep. 0.24
Congo, Rep. 0.04 0.09 0.14 Costa Rica 0.05 0.13 0.14 0.12
Cote d’Ivoire 0.08 0.09 Croatia 0.16 0.09 0.08
Cuba Curacao
Cyprus 0.03 0.03 0.03 Czech Republic 0.07 0.04 0.04
Denmark 0.05 0.05 0.05 Djibouti 0.05 0.09 0.10
Dominica 0.04 0.07 0.07 0.06 Dominican Republic 0.14 0.10 0.08
Ecuador 0.11 0.08 Egypt, Arab Rep. 0.04 0.07 0.05 0.05
El Salvador 0.02 0.04 0.05 Equatorial Guinea 0.10 0.14
Eritrea Estonia 0.13 0.04 0.05
Ethiopia 0.03 0.04 Faroe Islands
Fiji 0.04 0.06 0.04 Finland 0.01 0.05 0.03
France 0.05 0.05 0.04 French Polynesia
Gabon 0.05 0.09 0.14 Gambia, The 0.10 0.13 0.12 0.14
Georgia 0.17 0.04 Germany 0.05 0.06 0.07
Ghana 0.07 Greece 0.06 0.09 0.05
Greenland Grenada 0.05 0.07 0.07 0.07
Guam Guatemala 0.03 0.10 0.10 0.08
Guinea 0.03 0.12 Guinea-Bissau 0.26
Guyana 0.03 0.06 0.11 0.12 Haiti 0.19 0.11
Honduras 0.06 0.10 0.09 0.09 Hong Kong SAR, 0.04 0.04 0.05
China
Hungary 0.07 0.03 0.03 Iceland 0.05 0.06 0.07
India Indonesia 0.04 0.05 0.05
Iran, Islamic Rep. 0.05 0.00 Iraq
Ireland 0.04 0.06 0.04 Isle of Man
Israel 0.53 0.09 0.04 0.03 Italy 0.06 0.08 0.05
Jamaica 0.05 0.11 0.11 0.14 Japan 0.03 0.03 0.02 0.01
Jordan 0.03 0.06 0.05 Kazakhstan
Kenya 0.03 0.05 0.13 0.09 Kiribati
Korea, Dem. Rep. Korea, Rep. 0.02 0.02 0.02
Kosovo Kuwait 0.01 0.01 0.03 0.03
Kyrgyz Republic 0.23 0.17 Lao PDR 0.09 0.22 0.20
(Continued on next page)

148
 Table A.3 – (Continued from previous page)
Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
Latvia 0.38 0.05 0.06 Lebanon 0.04 0.18 0.05 0.02
Lesotho 0.06 0.07 0.11 0.07 Liberia 0.09 0.08 0.14 0.10
Libya 0.02 0.01 0.04 0.04 Liechtenstein
Lithuania 0.09 0.06 0.04 Luxembourg 0.03 0.02
Macao SAR, China 0.04 0.05 0.05 Macedonia, FYR 0.42 0.08 0.03
Madagascar 0.06 0.12 0.45 Malawi 0.08 0.07 0.22 0.24
Malaysia 0.03 0.04 0.02 Maldives 0.06 0.07
Mali 0.08 0.09 Malta 0.03 0.04 0.02
Marshall Islands Mauritania 0.07 0.07 0.14 0.10
Mauritius 0.03 0.07 0.12 0.02 Mexico 0.03 0.06 0.03
Micronesia, Fed. Sts. 0.13 0.14 Moldova 0.08 0.06
Monaco Mongolia 0.20 0.20 0.07
Montenegro Morocco 0.01 0.01 0.08
Mozambique 0.10 0.06 Myanmar 0.07 0.07 0.06 0.05
Namibia 0.09 0.07 0.04 Nepal 0.03 0.03 0.04 0.04
Netherlands 0.06 0.08 0.01 New Caledonia
New Zealand 0.02 0.02 Nicaragua 0.08 0.09 0.11
Niger 0.08 0.09 Nigeria 0.03 0.07 0.08 0.09
Northern Mariana Norway 0.09 0.04 0.02
Islands
Oman 0.01 0.03 0.05 0.03 Pakistan 0.06 0.06
Palau Panama 0.04 0.05 0.05
Papua New Guinea 0.03 0.06 0.10 0.10 Paraguay 0.10 0.20 0.17
Peru 0.62 0.19 0.16 Philippines 0.04 0.05 0.04 0.04
Poland 0.09 0.05 Portugal 0.02 0.07
Puerto Rico Qatar 0.04 0.04 0.04 0.04
Romania 0.33 0.17 0.06 Russian Federation 0.12 0.04
Rwanda 0.07 0.09 0.07 0.10 Samoa 0.07 0.07
San Marino Sao Tome and 0.03 0.20 0.14
Principe
Saudi Arabia Senegal 0.08 0.09
Serbia 0.18 0.08 Seychelles 0.06 0.06 0.09
Sierra Leone 0.05 0.14 0.14 0.12 Singapore 0.03 0.03 0.04 0.05
Sint Maarten (Dutch Slovak Republic 0.06 0.05
part)
Slovenia 0.23 0.05 Solomon Islands 0.03 0.06 0.13 0.11
Somalia South Africa 0.05 0.04 0.05 0.03
South Sudan 0.11 Spain 0.03 0.04 0.02
Sri Lanka 0.02 0.06 0.06 0.03 St. Kitts and Nevis 0.03 0.06 0.07 0.05
St. Lucia 0.04 0.07 0.08 0.07 St. Martin (French
part)
St. Vincent and the 0.05 0.08 0.07 0.06 Sudan
Grenadines
Suriname 0.05 0.13 0.05 Swaziland 0.05 0.06 0.07 0.06
Sweden 0.05 0.07 0.03 Switzerland 0.02 0.02 0.03 0.03
(Continued on next page)

149
 Table A.3 – (Continued from previous page)
Country 1980 1990 2000 2010 Country 1980 1990 2000 2010
Syrian Arab 0.05 0.05 0.03 0.04 Tajikistan 0.12 0.16
Republic
Tanzania 0.08 0.13 0.07 Thailand 0.04 0.02 0.05 0.05
Timor-Leste 0.15 0.11 Togo 0.08 0.09
Tonga 0.04 0.06 0.06 0.07 Trinidad and Tobago 0.05 0.08 0.08 0.06
Tunisia 0.04 Turkey
Turkmenistan Turks and Caicos
Islands
Tuvalu Uganda 0.05 0.05 0.13 0.11
Ukraine 0.39 0.17 0.06 United Arab 0.02 0.04
Emirates
United Kingdom 0.02 0.02 United States
Uruguay 0.28 0.31 0.07 Uzbekistan
Vanuatu 0.07 0.11 0.07 0.04 Venezuela, RB 0.08 0.07 0.02
Vietnam 0.10 0.04 0.03 Virgin Islands (U.S.)
West Bank and Gaza Yemen, Rep. 0.05 0.06
Zambia 0.04 0.13 0.20 0.07 Zimbabwe

A.1.3 Estimation Results

150
 Table A.4: System GMM: 2nd and 3rd Lag (With Interaction)

(1) (2) (3) (4)

LGDPt−1 -0.0342 -0.220 -0.263 -0.433
(-0.05) (-0.38) (-0.45) (-1.00)
LEDUt−1 0.0658 1.582 2.135∗ 2.305∗∗∗
(0.05) (1.37) (1.73) (2.77)
LGCt−1 -2.657∗∗ -1.414 -2.337∗∗ -1.925∗∗
(-2.08) (-1.20) (-2.18) (-2.12)
LOPENt−1 1.846∗ 0.722 1.015 0.902
(1.82) (0.66) (1.10) (1.13)
LINFLt−1 -0.196 -0.0410 -0.248 -0.169
(-0.79) (-0.17) (-0.93) (-0.77)
PCt−1 0.777 -2.217 -4.458∗∗ -2.597∗
(0.26) (-0.93) (-2.07) (-1.67)
LDSPREADt−1 -27.74∗∗ -33.78∗∗ -41.51∗∗∗ -32.18∗∗∗
(-2.10) (-2.34) (-2.96) (-3.36)
PCt−1 × LDSPREADt−1 43.70 42.36 63.92 35.67
(0.86) (0.87) (1.40) (1.11)
Constant 0 0 0 4.153
(.) (.) (.) (0.90)
Time Fixed Effect X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010
Observations 273 377 484 576
Sarganp 0.000688 0.000228 0.000000246 1.92e-09
Hansenp 0.354 0.144 0.218 0.484
AR1p 0.0252 0.00670 0.00740 0.000431
AR2p 0.580 0.163 0.462 0.00397
Instruments 66 88 110 132
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

151
 Table A.5: System GMM: 3rd Lag (With Interaction)

(1) (2) (3) (4)

LGDPt−1 0.335 -0.275 -0.604 -0.701
(0.41) (-0.40) (-0.98) (-1.50)
LEDUt−1 -0.899 1.892 2.764∗∗ 2.294∗∗
(-0.54) (1.48) (2.27) (2.22)
LGCt−1 -3.285∗∗ -1.408 -1.673 -2.479∗∗
(-2.36) (-0.89) (-1.13) (-2.00)
LOPENt−1 -0.0906 -0.258 -0.225 0.365
(-0.05) (-0.19) (-0.19) (0.38)
LINFLt−1 0.00463 0.0117 -0.0639 0.0101
(0.01) (0.05) (-0.27) (0.04)
PCt−1 -3.702 -3.603 -4.394∗ -3.390∗∗∗
(-0.94) (-1.32) (-1.76) (-2.70)
LDSPREADt−1 -38.20∗ -36.40∗∗ -46.52∗∗∗ -40.44∗∗∗
(-1.73) (-2.25) (-2.87) (-3.26)
PCt−1 × LDSPREADt−1 85.52 59.16 72.06∗∗ 79.37∗∗∗
(1.42) (1.52) (2.08) (2.71)
Constant 0 7.370 9.463 9.443
(.) (1.10) (1.61) (1.52)
Time Fixed Effect X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010
Observations 273 377 484 576
Sarganp 0.0100 0.00200 0.00101 0.0000810
Hansenp 0.276 0.254 0.420 0.390
AR1p 0.0120 0.00390 0.00512 0.000311
AR2p 0.668 0.175 0.563 0.00917
Instruments 40 55 70 85
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

152
 Table A.6: System GMM: 4th Lag (With Interaction)

(1) (2) (3) (4)

LGDPt−1 0.145 -1.081 -1.420 -0.837
(0.10) (-1.06) (-1.27) (-1.40)
LEDUt−1 0.276 2.985∗ 5.215∗∗ 3.753∗∗∗
(0.12) (1.78) (2.48) (2.96)
LGCt−1 -2.829∗ -1.803 -0.999 -1.820
(-1.67) (-1.08) (-0.65) (-1.57)
LOPENt−1 2.015 -0.294 -1.061 0.436
(1.11) (-0.18) (-0.74) (0.34)
LINFLt−1 -0.551 -0.122 -0.404 -0.0566
(-1.16) (-0.32) (-0.86) (-0.16)
PCt−1 -1.460 -2.330 -4.397∗∗ -4.837∗∗∗
(-0.42) (-0.83) (-2.07) (-3.19)
LDSPREADt−1 -20.84 -35.92∗∗ -41.33∗∗∗ -39.61∗∗∗
(-0.73) (-2.18) (-2.59) (-3.36)
PCt−1 × LDSPREADt−1 40.09 60.47 76.14 66.20∗
(0.57) (1.02) (1.40) (1.73)
Constant 1.129 11.26 11.66 5.868
(0.08) (1.11) (1.54) (0.80)
Time Fixed Effect X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010
Observations 273 377 484 576
Sarganp 0.000269 0.0375 0.0278 0.00469
Hansenp 0.0163 0.0517 0.0712 0.309
AR1p 0.0277 0.00572 0.00873 0.000656
AR2p 0.299 0.146 0.369 0.0168
Instruments 28 41 56 71
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

153
 Table A.7: Panel OLS (Comparison with Arcand et al. (2015a) Dataset)

Arcand et al. (2015a) World Bank GFDD

(1) (2) (3) (4) (1) (2) (3) (4)
LGDPt−1 -0.177 -0.138 -0.148 -0.118 -0.176 -0.143 -0.199 -0.178
(-1.13) (-1.05) (-1.24) (-1.11) (-1.09) (-1.04) (-1.62) (-1.63)
LEDUt−1 0.988∗∗∗ 1.133∗∗∗ 1.324∗∗∗ 1.288∗∗∗ 0.663∗∗ 0.838∗∗∗ 1.157∗∗∗ 1.166∗∗∗
(3.34) (4.36) (5.55) (5.82) (2.18) (3.12) (4.67) (5.11)
LGCt−1 -1.157∗∗∗ -0.982∗∗∗ -1.044∗∗∗ -1.132∗∗∗ -0.892∗∗ -0.768∗∗ -0.922∗∗∗ -1.023∗∗∗
(-3.27) (-3.26) (-3.82) (-4.59) (-2.38) (-2.39) (-3.14) (-3.88)
LOPENt−1 0.486∗∗ 0.380∗ 0.410∗∗ 0.411∗∗ 0.589∗∗ 0.519∗∗ 0.560∗∗∗ 0.536∗∗∗
(2.11) (1.91) (2.29) (2.52) (2.53) (2.55) (3.02) (3.17)
LINFLt−1 -0.0381 -0.0806 -0.138 -0.116 -0.255∗∗ -0.276∗∗∗ -0.357∗∗∗ -0.350∗∗∗

154
(-0.36) (-0.86) (-1.58) (-1.41) (-2.28) (-2.78) (-3.88) (-4.04)
PCt−1 1.663∗∗∗ 0.902∗ 0.119 -0.440
(2.72) (1.86) (0.29) (-1.31)
PCt−1 2.163∗∗∗ 1.169∗∗ 0.403 -0.241
(3.29) (2.26) (0.92) (-0.68)
Constant 3.464∗∗∗ 3.137∗∗∗ 3.224∗∗∗ 1.490 0.0902 2.815∗∗ 3.189∗∗∗ 1.864∗
(2.84) (2.90) (3.23) (1.59) (0.07) (2.49) (3.04) (1.89)
Time Fixed Effect X X X X X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010 1960-1995 1960-2000 1960-2005 1960-2010
Observations 549 675 798 917 542 664 793 917
Dependent Variable: GDP % Growth 5 Year
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01
 Table A.8: System GMM (Comparison with Arcand et al. (2015a) Dataset)

Arcand et al. (2015a) World Bank GFDD

(1) (2) (3) (4) (1) (2) (3) (4)
LGDPt−1 -0.688∗ -0.828∗∗ -0.800∗∗ -0.770∗∗ -0.378 -0.465 -0.360 -0.0276
(-1.83) (-1.98) (-2.43) (-2.26) (-0.81) (-1.25) (-1.05) (-0.06)
LEDUt−1 1.343∗ 2.008∗∗∗ 2.780∗∗∗ 2.833∗∗∗ 0.386 1.280∗ 1.757∗∗ 1.589∗∗
(1.78) (2.81) (4.26) (4.46) (0.55) (1.78) (2.40) (2.49)
LGCt−1 -3.208∗∗∗ -2.625∗∗∗ -1.722∗∗∗ -1.744∗∗∗ -2.537∗∗ -2.377∗∗∗ -2.650∗∗∗ -2.626∗∗∗
(-4.07) (-3.61) (-2.97) (-3.09) (-2.43) (-2.78) (-3.39) (-3.34)
LOPENt−1 1.590∗∗ 1.615∗∗∗ 1.444∗∗∗ 1.666∗∗∗ 0.483 0.720 1.003 1.602∗∗
(2.15) (2.72) (2.67) (3.07) (0.58) (1.02) (1.62) (2.42)
LINFLt−1 0.0747 -0.0138 -0.262 -0.229 -0.245 -0.377∗ -0.701∗∗∗ -0.644∗∗∗
(0.39) (-0.08) (-1.49) (-1.24) (-0.95) (-1.80) (-3.15) (-3.14)
PCt−1 5.429∗∗∗ 3.652∗∗∗ 1.063 0.0725
(3.46) (2.95) (1.43) (0.10)

155
PCt−1 4.272∗∗∗ 2.542∗∗ 0.477 -1.102
(2.58) (2.23) (0.57) (-1.13)
Constant 2.956 2.257 0.264 -1.292 9.100∗∗ 7.476∗∗ 6.308∗ 0.157
(0.90) (0.71) (0.09) (-0.40) (2.17) (2.11) (1.81) (0.04)
Time Fixed Effect X X X X X X X X
Period 1960-1995 1960-2000 1960-2005 1960-2010 1960-1995 1960-2000 1960-2005 1960-2010
Observations 549 675 798 917 542 664 793 917
Sarganp 0.00000794 0.000000540 4.94e-13 2.89e-15 0.000000134 3.70e-09 6.20e-17 1.29e-18
Hansenp 0.978 1.000 1.000 1.000 0.839 0.995 1.000 1.000
AR1p 0.000171 0.00000913 0.000000608 8.28e-08 0.000349 0.0000204 0.0000288 0.00000242
AR2p 0.657 0.265 0.385 0.0879 0.978 0.432 0.619 0.00819
Instruments 127 170 219 274 127 170 219 274
Dependent Variable: GDP % Growth 5 Year
t statistics in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01
 Table A.9: Panel OLS: Quadratic

Private Credit Lending-Deposits Spread

LGDPt−1 -0.235 -0.366
(0.388) (0.258)
LGDPt−1 1.376** 2.426***
(0.698) (0.690)
LGCt−1 -2.526*** -2.268***
(0.718) (0.714)
LOPENt−1 0.815 0.750
(0.497) (0.870)
LINFLt−1 -0.585*** -0.408**
(0.167) (0.196)
PCt−1 3.289*
(1.734)
2
PCt−1 -2.231***
(0.742)
LDSPREADt−1 -0.172*
(0.102)
2
LDSPREADt−1 0.00217
(0.00260)
Time Fixed Effect X X
Observations 917 616
Number of Countries 132 124
Dependent Variable: GDP % Growth 5 Year
Standard errors in parentheses
∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01

156
Appendix B

Heterogeneous Patterns of Financial

Development: Implications for Asian

Financial Integration

B.1 Appendix I

157
 Table B.1: Institutional Environment

Efficacy of corporate boards

Reliance on professional management
Willingness to delegate authority
Strength of auditing and reporting standards
Ethical behavior of firms
Protection of minority shareholders interests
Burden of government regulation
Regulation of securities exchanges
Property rights
Intellectual property protection
Diversion of public funds
Public trust in politicians
Legal rights index
Judicial independence
Irregular payments and bribes
Source: The World Economic Forum - Global Competitiveness Index

158
 Table B.2: Business Environment

Quality of management schools

Quality of math and science education
Extent of staff training
Availability of research and training services
Secondary education enrollment, gross
Tertiary education enrollment, gross
Quality of the education system
Quality of overall infrastructure
Quality of electricity supply
Individuals using Internet,
Fixed broadband Internet subscriptions/100 pop.
Fixed telephone lines/100 pop.
Mobile telephone subscriptions/100 pop.
No. procedures to start a business
No. days to start a business
Source: The World Economic Forum - Global Competitiveness Index

159
 Table B.3: Additional Components and Variables

Category Component

Local supplier quality

Local supplier quantity
State of cluster development
Business Sophistication Nature of competitive advantage
Value chain breadth
Control of international distribution
Production process sophistication Extent of marketing

Intl Internet bandwidth, kb/s per user

Technology Sophistication
Mobile broadband subscriptions/100 pop.*

Prevalence of trade barriers

Trade tariffs,
Openness Prevalence of foreign ownership
Business impact of rules on FDI
Burden of customs procedures Imports as a percentage of GDP

Favoritism in decisions of government officials

Wastefulness of government spending
Efficiency of legal framework in settling disputes
Institutional Environment2 Efficiency of legal framework in challenging regs
Transparency of government policymaking
Effect of taxation on incentives to invest
Total tax rate,
Source: The World Economic Forum - Global Competitiveness Index

160
 Table B.4: Level and Pattern Differences FDIndex14

Level Distance Pattern Distance

Country Average Range Min Max Average Range Min Max

Year 2010

AUS 0.24 1.93 -0.42 1.51 0.72 0.21 0.20 0.93

CHN 0.20 1.93 -1.35 0.58 0.75 0.12 0.42 0.86
HKG 0.33 1.93 -0.04 1.89 0.70 0.27 0.07 0.93
IDN 0.21 1.93 -1.39 0.54 0.28 0.33 -0.10 0.81
JPN 0.18 1.93 -0.91 1.01 0.67 0.12 0.38 0.90
KHM 0.32 1.74 -1.93 -0.18 0.60 0.32 -0.10 0.91
KOR 0.23 1.93 -1.52 0.41 0.64 0.19 0.15 0.90
MYS 0.21 1.93 -0.59 1.34 0.65 0.26 0.09 0.92
NZL 0.22 1.93 -0.54 1.39 0.65 0.27 0.02 0.93
PHL 0.27 1.93 -1.74 0.18 0.61 0.19 0.40 0.98
SGP 0.34 1.89 0.04 1.93 0.67 0.31 -0.06 0.93
THA 0.19 1.93 -1.29 0.64 0.62 0.18 0.39 0.98
TWN 0.19 1.93 -0.76 1.16 0.47 0.26 0.14 0.86
VNM 0.24 1.93 -1.57 0.36 0.68 0.28 0.04 0.93

Year 2014

AUS 0.22 2.06 -0.63 1.43 0.74 0.15 0.55 0.96

CHN 0.23 2.06 -1.46 0.60 0.60 0.24 0.14 0.91
HKG 0.35 2.03 0.03 2.06 0.74 0.14 0.57 0.95
IDN 0.22 2.06 -1.35 0.71 0.59 0.27 0.13 0.94
JPN 0.22 2.06 -0.68 1.38 0.65 0.12 0.40 0.77
KHM 0.36 2.01 -2.06 -0.05 0.53 0.25 0.10 0.81
KOR 0.27 2.06 -1.69 0.37 0.48 0.24 0.12 0.74
MYS 0.26 2.06 -0.42 1.64 0.64 0.17 0.35 0.83
NZL 0.29 2.06 -0.27 1.79 0.70 0.15 0.52 0.95
PHL 0.24 2.06 -1.48 0.58 0.56 0.30 0.07 0.96
SGP 0.34 2.06 -0.03 2.03 0.71 0.18 0.45 0.96
THA 0.21 2.06 -1.30 0.76 0.64 0.22 0.27 0.96
TWN 0.20 2.06 -0.81 1.25 0.55 0.26 0.10 0.88
VNM 0.35 2.06 -2.01 0.05 0.51 0.29 0.07 0.81
Includes: All Eight Sub-Index of the 8th Pillar, Includes: Six Additional Index
Source: The World Economic Forum - Global Competitiveness Index

161
 Table B.5: Level and Pattern Differences FDIndex13

Level Distance Pattern Distance

Country Average Range Min Max Average Range Min Max

Year 2010

AUS 0.23 1.98 -0.41 1.58 0.80 0.08 0.65 0.91

CHN 0.19 1.98 -1.27 0.71 0.83 0.08 0.70 0.92
HKG 0.31 1.98 -0.04 1.94 0.83 0.09 0.66 0.92
IDN 0.18 1.98 -1.17 0.81 0.66 0.16 0.40 0.89
JPN 0.17 1.98 -0.85 1.14 0.74 0.08 0.61 0.90
KHM 0.34 1.60 -1.98 -0.39 0.72 0.11 0.49 0.87
KOR 0.23 1.98 -1.50 0.49 0.70 0.11 0.48 0.90
MYS 0.19 1.98 -0.63 1.35 0.82 0.10 0.60 0.95
NZL 0.20 1.98 -0.54 1.45 0.72 0.14 0.40 0.92
PHL 0.24 1.98 -1.56 0.43 0.83 0.11 0.66 0.98
SGP 0.32 1.94 0.04 1.98 0.78 0.12 0.46 0.92
THA 0.18 1.98 -1.11 0.87 0.84 0.10 0.67 0.98
TWN 0.20 1.98 -0.54 1.44 0.77 0.09 0.57 0.86
VNM 0.25 1.98 -1.60 0.39 0.84 0.05 0.75 0.91

Year 2014

AUS 0.21 2.12 -0.68 1.45 0.79 0.12 0.61 0.94

CHN 0.22 2.12 -1.34 0.79 0.71 0.19 0.20 0.93
HKG 0.34 2.10 0.03 2.12 0.80 0.12 0.51 0.92
IDN 0.21 2.12 -1.22 0.90 0.72 0.20 0.23 0.95
JPN 0.22 2.12 -0.59 1.53 0.72 0.10 0.47 0.85
KHM 0.38 2.07 -2.12 -0.06 0.54 0.10 0.39 0.72
KOR 0.29 2.12 -1.72 0.40 0.45 0.19 0.14 0.75
MYS 0.25 2.12 -0.45 1.67 0.69 0.21 0.14 0.93
NZL 0.28 2.12 -0.29 1.84 0.74 0.16 0.41 0.94
PHL 0.22 2.12 -1.31 0.81 0.79 0.18 0.34 0.97
SGP 0.33 2.12 -0.03 2.10 0.76 0.13 0.53 0.94
THA 0.20 2.12 -1.17 0.95 0.79 0.14 0.44 0.97
TWN 0.22 2.12 -0.64 1.48 0.78 0.12 0.45 0.89
VNM 0.37 2.12 -2.07 0.06 0.61 0.11 0.40 0.82
Excludes: Legal Right Index of the 8th Pillar, Includes: Six Additional Index
Source: The World Economic Forum - Global Competitiveness Index

162
B.2 Appendix II: Financial Development Index Pillars

B.2.0.1 Clustering based on Levels of Financial Development

Table B.6: Clustering based on Levels of Financial Development

8.01 Availability of financial services, 1-7 (best)

8.02 Affordability of financial services, 1-7 (best)
8.03 Financing through local equity market, 1-7 (best)
8.04 Ease of access to loans, 1-7 (best)
8.05 Venture capital availability, 1-7 (best)
8.06 Soundness of banks, 1-7 (best)
8.07 Regulation of securities exchanges, 1-7 (best)
8.08 Legal rights index, 1-7 (best)
Source: The World Economic Forum - Global Competitiveness Index

163
 Table B.7: Financial Development Index: Category

Alternative Specification I
Category Indicator Score

Availability of financial services, 1-7 (best)

Affordability of financial services, 1-7 (best)
Efficiency Financing through local equity market, 1-7 (best)
Ease of access to loans, 1-7 (best)
Venture capital availability, 1-7 (best)

Soundness of banks, 1-7 (best)

Trustworthiness and Confidence Regulation of securities exchanges, 1-7 (best)
Legal rights index, 1-7 (best)

Local supplier quantity, 1-7 (best)

Local supplier quality, 1-7 (best)
State of cluster development, 1-7 (best)
Nature of competitive advantage, 1-7 (best)
Business Sophistication
Value chain breadth, 1-7 (best)
Control of international distribution, 1-7 (best)
Production process sophistication, 1-7 (best)
Extent of marketing, 1-7 (best)

Intl Internet bandwidth, kb/s per user*

Technology Sophistication
Mobile broadband subscriptions/100 pop.*
Source: The World Economic Forum - Global Competitiveness Index

164
 Table B.8: Financial Development Index: Category

Alternative Specification II
Category Indicator Score

Availability of financial services, 1-7 (best)

Affordability of financial services, 1-7 (best)
Efficiency Financing through local equity market, 1-7 (best)
Ease of access to loans, 1-7 (best)
Venture capital availability, 1-7 (best)

Soundness of banks, 1-7 (best)

Trustworthiness and Confidence
Regulation of securities exchanges, 1-7 (best)

Local supplier quantity, 1-7 (best)

Local supplier quality, 1-7 (best)
State of cluster development, 1-7 (best)
Nature of competitive advantage, 1-7 (best)
Business Sophistication
Value chain breadth, 1-7 (best)
Control of international distribution, 1-7 (best)
Production process sophistication, 1-7 (best)
Extent of marketing, 1-7 (best)

Intl Internet bandwidth, kb/s per user*

Technology Sophistication
Mobile broadband subscriptions/100 pop.*
Source: The World Economic Forum - Global Competitiveness Index

165
Appendix C

Structural Changes and Labor Market

Frictions

C.1 Appendix I

C.1.1 International Comparison Program 2005

Relative Prices (ICP) 2005: There is a large literature on sectoral relative prices

calculation.1 The international comparison program (ICP) 2005 provides detailed in-

formations on:

• Expenditures at the basic headings in local currency for all countries in the sam-
1 The notes is based on Kuralbayeva and Stefanski (2013), and Adamopoulos (2009): The Interna-

tional Comparison Program Handbook (2005). Other papers in the literature that utilize the disaggre-
gated ICP data: Buera et al. (2011) and Deaton and Hestor (2010).

166
 ple.

• PPP at the basic headings level (129 basic headings) for all countries in the sam-

ple..

• Respective exchange rate (per US dollars) for all countries in the sample.

Hence, the expenditures is converted to in terms of US dollars. The base country is the

US. The ICP data also gives us the detailed expenditure data in PPP exchange rate at

disaggregated level (129 basic headings) .

Aggregating PPP Exchange Rates from ICP (2005) Data:

Calculating the Sectoral PPP Exchange Rates from the ICP 2005: for each

sector j ∈ {A, M & S}, we match each basic headings n = 1, ..., N j in the ICP(2005)

data with the corresponding sector j where N j is the number of basic headings level

in the respective sector j and N is the total number of basic headings level in the ICP

(2005) data. We have that

NA + NM + NS = N.

where A is for Agriculture, M is for Manufacturing and S is for Services.

Let i or k = 1, ..., K be the country in the sample.

The Geary-Khamis Aggregation Method: we will follow the literature and use

the Geary-Khamis aggregation method. A desirable property of the Geary-Khamis

167
aggregation method is that it preserves the additive property and hence, we calculate

the share of component of GDP or sectoral PPP exchange rates for the sectoral relative

prices. The Geary-Khamis aggregation method involves solving a system of simulta-

neous equations.

K " #
qin pin

πn = ∑ . (C.1)
i=1 ∑K k
k=1 qn Pij

for country i or k = 1, ..., K and basic headings level n = 1, ..., N in sector j ∈

{A, M & S} We think of πn as the weighted average ”international” prices for each
qin
basic headings n with K qk as the corresponding weight. where:
∑k=1 n

Eni
qin = . (C.2)
pin

i ∑N i i
n=1 pn qn
P = N
. (C.3)
∑ πn qin
n=1

for country i = 1, ..., K in the sample and basic heading level n = 1, ..., N in sector

j ∈ {A, M & S}

where: E ij is sector j expenditures for country i in US dollars and is the sum of

all expenditures at the basic headings within the sector. qin or qkn is the ”notional”

quantities and is calculated using the detailed ICP data and they are defined as the ratio

168
of expenditure in US dollars and PPP at the basic headings n for country i. pin is the

PPP exchange rate for basic headings level n of country i (it is given in the ICP (2005)

data). Pij is the PPP exchange rate for sector j ∈ {A, M & S} of country i (relative to

the base country US (2005)). Pi is the GDP PPP exchange rate for country i (relative

to the base country US in (2005)).

Since we are interested in the relative price of Agriculture, Manufacturing and Ser-

vices we will need to aggregate the ICP basic headings level data to match our three-

sector model specification.

n o
i
We will solve for πn , Pj using two key systems of equations (C.1) and (C.3) above.

Notations: Aggregate GDP Level

• PPP Exchange Rate: Pi for country i = 1, ..., K in the sample

Aggregate GDP Level Symbol

PPP Exchange Rate Pi

Sectoral Level

• PPP Exchange Rate: Pij for sector j ∈ {A, M & S} of country i = 1, ..., K in the

sample

Sectoral Level Symbol

PPP Exchange Rate Pij

Basic Headings Level

169
• PPP Exchange Rate: pin for basic headings n = 1, ..., N of country i = 1, ..., K in

the sample

• International Prices: πn for basic headings n = 1, ..., N in the sample

• Nominal Expenditure in Local Currency: Eni

• ”Notional/Real” Quantities of Expenditure (deflated using pin ): qin

Table C.1: Basic Heading Levels

Levels Symbols
Basic Headings Level
PPP Exchange Rate pin
Expenditure in Local Currency Eni
”Real” Expenditure (deflated using pin ) qin
International Prices πn
Sectoral Level
PPP Exchange Rate Pij
Aggregate GDP Level
PPP Exchange Rate Pi −

1. ICP (2005) Data at Basic Headings Level

(a) Nominal Expenditures in Local Currency: Eni

(b) Basic Headings PPP Exchange Rate: pin

(c) Nominal USD Market Exchange Rate: ei

2. Calculating the Basic Headings Real Expenditure qin (deflated using basic head-

ings PPP exchange rate pin )

Eni
qin = , (C.4)
pin

170
3. The Geary-Khamis Aggregation Method

The system of simultaneous equations

 

K
 i  i 
 qn  pn
πn = ∑   K  Pi .
 (C.5)
i=1 
∑ qkn

k=1

for country i or k = 1, ..., K and basic headings level n = 1, ..., N

Intuition:

We think of the International Prices πn as the weighted average of each country’s

PPP prices (normalized by country i respective aggregate PPP exchange rate Pi )

qin
for each basic headings n = 1, ..., N with K
as the corresponding weight for
∑ qkn
k=1
country i.
N
∑ pinqin
n=1
Pi = N
. (C.6)
∑ πn qin
n=1

for country i = 1, ..., K in the sample and basic heading level n = 1, ..., N

Intuition:

The Aggregate PPP Exchange Rates Pi is the ratio of GDP in domestic/local

currency and total expenditures in International Prices πn .

4. Solving for πn , Pi using (C.5) and (C.6)



171
 (a) Once we obtain the solution for the international prices πn , we get the ag-

gregate GDP PPP exchange rate Pi using (C.6)

(b) With the solution for the international prices πn , we also get the sectoral

PPP exchange rate Pij in sector j ∈ {A, M & S} for country i = 1, ..., K

5. Calculating the Sectoral PPP Exchange Rate Pij

For basic headings level n = 1, ..., N j in sector j ∈ {A, M & S} in country i =

1, ..., K in the sample, we divide the basic headings level to match our three-

sector environment and framework:

With N total of number basic headings, we have:

NA + NM + NS = N.

Table C.2: Sectoral Employment Notation

Sectors Numbers of Basic Headings

A Agriculture NA
M Manufacturing NM
S Services NS

Using the international prices πn calculated above, we get the sectoral PPP ex-

change rate for each sector (Agriculture, Manufacturing and Services) j ∈ {A, M & S}
 
Nj
as the ratio of respective sectoral expenditure in local currency ∑n=1 pin qin and
N
!
j
sectoral expenditures in international prices ∑ πnqin
n=1

172
 Nj
∑n=1 pin qin
Pij = Nj
.
∑n=1 πn qin

The Sectoral Relative Prices: we define the sectoral relative prices for country

i = 1, ..., K in the sample as:

i PAi
SectPA/M = i . (C.7)
PM

i PSi
SectPS/M = i
. (C.8)
PM

The nominal expenditures in US dollars is calculated by converting the nominal expen-

ditures in local currency using the nominal exchange rate provided by the ICP data.

Intuition: we derive the implicit relative price level from the ratio of nominal ex-

penditures in US dollars and expenditure in PPP exchange rate.

The nominal expenditures in US dollars is converted from the nominal expenditure in

domestic currency using the nominal exchange rate provided in the ICP data.

Relative Prices (ICP 2005)

The International Comparison Program (ICP) 2005 provides detailed informations

on

• Expenditures at the basic headings in local currency for all countries in the sam-

ple.

173
 • PPP at the basic headings level (129 basic headings) for all countries in the sam-

ple..

• Respective exchange rate (Per US dollars) for all countries in the sample.

Hence, the expenditures is converted to in terms of US dollars. We will follow Kural-

bayeva and Stefanski (2013) (Table 9) in aggregating and matching the basic headings

with the three-sector environment (Agriculture ”A”, Manufacturing ”M” and Services

”S”). The base country is the US. The ICP data also gives us the detailed expendi-

ture data in PPP exchange rate at disaggregated level (129 sectors) . We will define the

following:

Sectors Symbols
Nominal expenditures in US dollars E ij
Expenditures in PPP exchange rate E i,PPP
j

International Comparison Program (2005): it is worth noting that the sectoral

relative price calculated from the ICP (2005) are based on producer prices of expen-

diture data. It is not the basic price level that consumers pay and hence, it potentially

include transportation costs, tax and subsidies etc.

174
C.2 Appendix II

C.2.1 Closed Economy: Calibration for South Korea (Discussion

& Explanation)

We will use data from two main database: the World Bank World Development

Indicator (WDI) data and the Groningen Growth and Development Center (GGDC) 10-

Sector dataset. The ranges for the data for South Korea: the initial period 0 is 1980 and

the last period T is 2010.

For Korea, the World Bank WDI data on sectoral employment shares cover the

1979 − 2010 period and the World Bank WDI data on sectoral (constant 2005 US$) real

value added covers the 1970 − 2011 period (the sectoral value added data is truncated

to match the shorter time series for sectoral employment shares). For GGDC 10-Sector

data, the sectoral employment shares is from 1963 to 2005.

Production Parameters: φ, φS , θ, µM , µS

Agriculture Production from Manufacturing to Agriculture {φ}: following the

literature Caselli and Coleman II (2001) as in Choi et al. (2014), we assume φ = 0.81

and this implies that 19% of A production belongs to the owners of fixed inputs (such

as lands).

Services Section Production {φS }: φS = 0.5.

175
 Manufacturing Technology Spillover to Agriculture {θ}: The technology spillover

from the Manufacturing sector to the Agriculture sector (such as the use of imported

intermediate goods) is governed by the spillover parameter θ. The calibrated value for

θ is similar to that in Choi et al. (2014). In calculating θ, we use the ratio of initial and
aT 1−mT −sT mT
last period Agriculture and Manufacturing labor shares a0 = 1−m0 −s0 and m0 , which

are taken from the data and the initial period is 0 and the last period is T . Given the

time series sectoral real value added in A and M (in constant 2005 US$) for Korea, we

calculate the annual growth rates of sectoral real value added adjusted for population

growth rates. Normalizing the initial period A and M real value added to 1, Y0A = 1 and

Y0M = 1, the last period sectoral real value added YTA and YTM are calculated using the

annual growth rates recovered from the data with Y0A = 1 and Y0M = 1. In particular, we
YTA
   
1−m −s
log −φ log 1−mT −sT
Y0A 0 0
derive the spillover parameter θ as follows: θ = 
YTM
  
mT
and we have
log M −log m
Y0 0

that θ = 0.7158.2

Endogenous Growth (Learning-by-Doing) {µM , µS }: the learning-by- doing pa-

rameters for the Manufacturing µM and Services sector µS .

YTM /mT
Manufacturing: The learning-by-doing parameter µM is backed out using Y0M /m0
=
T
∏ (1 + µM mt ) . In particular, we find the annual growth rates of Manufacturing real
t=0
value added (in constant 2005US$) from the World Bank WDI data for Korea and by

normalizing the initial period Y0M to 1, we find the last period YTM using the M sector
2 Note:
alternatively, we also calibrate the parameter using the Groningen Growth and Development
Center (GGDC) 10-Sector database.

176
real annual growth rates. An implicit assumption is that the growth rates of Manu-

facturing real value added (indices in constant 2005 US$) is a good approximation

for the changes in ”quantities” of Manufacturing output. This approach of calibrating

the sequence of sectoral productivity for M that will be put into the model using data

on sectoral real value added per worker is similar to those in the literature Duarte and

Restuccia (2010) and µM = 0.3100.

Services: Similarly for the learning-by-doing parameter µS . µS is backed out using

φ T
YTS /sTS
φ = ∏ (1 + µS st ) . In particular, we find the annual growth rates of Services real
Y0S /s0S t=0
value added (in constant 2005US$) from the World Bank WDI data for Korea and by

normalizing the initial period Y0S to 1, we get the last period YTS using the S sector real

annual growth rates. An implicit assumption is that the growth rates of Services real

value added (indices in constant 2005US$) is a good approximation for the changes in

”quantities” of Services output. This approach of calibrating the sequence of sectoral

productivity for S that will be put into the model using data on sectoral real value added

per worker is similar to that of Duarte and Restuccia (2010) and µS = 0.0530.

Preference Parameters:

{r, κ, βA , βM = 1, βS , γA , γM = 0, γS }

• Sectoral labor adjustment cost {κ}: The empirical literature3 estimates for the
3 Empirical literature on sectoral labor adjustment costs in Artuc et al. (2013), Porto and Hoekman

177
 sectoral labor adjustment costs κ = 5.

• The rate of interest {r}: The rate of interest r is taken to be 2% and hence,

r = 0.02.

The Discount Factors Parameter {βA , βM = 1, βS }: in calibrating the discount

factor parameters for Agriculture βA and Services βS (we normalized Manufacturing

βM to 1), we will make use of two equations for the subsistence parameters for A sector

γA and S sector γS by evaluating at the initial period data below, (the two expressions

are derived from the consumer FOC):

φ pM
0
γA = AM0θ a0 − βA M0 m0 .
pA0
pM
0
γS = βS M0 m0 − S0 s0 .
pS0

In particular, in matching the subsistence consumption parameters for A sector γA

and S sector γS with initial period data on the sectoral employment shares {a0 , m0 , s0 },
n M Mo
p p
the relative productivity for A, M and S as well as the relative prices p0A , p0S , we
0 0

follow the method similar to that of Duarte and Restuccia (2010) for their simple two

sectors model. Once we calibrated βA and βS , the subsistence parameters for A sector

γA and S sector γS is calculated by simply plugging in the two equations above. Using

the firm FOC, we derive the two expressions for calibrating βA and βS as follows:
(2010) and Artuc and McLaren (2009)

178
 1 St st +γS
Agriculture {βA }: we will solve for the expression for βA using mt = βS St and
φ pM
γA = AM0θ a0 − 0
β M m .
pA0 A 0 0
Using the expression:

φ φ
AMTθ aT − AM0θ a0
βA =  .
pM

φ−1 r
φAMTθ aT mT + 1+r κMTθ (mT + sT − mT −1 − sT −1 ) mT − p0A M0 m0
0

We get the A sector discount factor parameter βA using the initial period data on the
pM
sectoral employment shares {a0 , m0 }, the relative prices 0
pA0
, relative productivity for

A, M sector as well as the last period data on sectoral employment shares {aT , mT , sT }.
φ−1
pM
0 pM
0 φAM0θ a0
The initial period relative price pA0
is implied by the model using pA0
= M0 +
r θ
1+r κM0 (m0 +s0 −m−1 −s−1 )
M0 , from the firms FOC and initial period data {a0 , m0 , s0 }. With

calibrated {r = 0.02, κ = 5}, we get βA = 0.0473. Note: we simplify the equation using

an approximation mt+1 + st+1 − mt − st = mt + st − mt−1 − st−1 .

1 St st +γS
Services {βS }: we will solve for the expression for βS using mt = βS St and
pM ST sT −S0 s0
γS = 0
β M m − S0 s0 .
pS0 S 0 0
Using the expression βS = pM
, we get the S sec-
ST mT − 0 M m
0 0
pS0
tor sector discount factor parameter βS using the initial period data on the sectoral
pM
employment shares {m0 , s0 }, the relative prices 0
pA0
, relative productivity for M and S

sector as well as the penultimate and last period data on sectoral employment shares
pM
{mT , mT −1 , sT , sT −1 }. In calculating the initial period relative price 0
pS0
, from the firm
φ −1
pM
0 φS S0 s0S
FOC, we use pS0
= M0 , which is the inverse of relative productivity between M

and S evaluated using the initial period data. Plugging in, we get βS = 8.9993.

179
 The subsistence consumption{γA , γM = 0, γS }: given the calibrated parameters

{βA , βM = 1, βS } , the subsistence consumption parameters {γA , γM = 0, γS } are cali-

brated as follows:

Agriculture {γA }: following Duarte and Restuccia (2010), we calibrate the Agri-

culture sector subsistence consumption parameter γA to match the initial period data
pM
on Agriculture and Manufacturing employment share {a0 , m0 }, and relative price 0
pA0
 pM
and A, M productivity AM0θ , M0 . Similarly, the initial period relative price 0
pA0
is
ptM
calculated as above. Moreover, for time t, the relative price between M and A ptA
=
φ−1 r θ
φAMtθ at 1+r κMt (mt +st −mt−1 −st−1 )
Mt + Mt , is equal to the inverse of relative productivity M

and A plus sectoral labor adjustment costs. Given the calibrated parameters {βA , r, κ},
φ pM
we get γA using γA = AM0θ a0 − 0
β M m ,
pA0 A 0 0
and γA = 0.9071.

Services{γS }: similarly, we will calibrate the Services sector subsistence con-

sumption parameter γS to match the initial period data on Manufacturing and Services
pM
employment share m0 and s0 , and relative price p0S and M and S productivity M0 and S0 .
0
φ −1
ptM φS St st S
Moreover, for time t, the relative price between M and S ptS
= Mt , is equal to the

inverse of relative productivity M and S. Similarly, we derive the initial period relative
pM
0
price pS0
, from the firm FOC, using initial period data as above. Given the calibrated
pM
βS , we get γS using γS = 0
β M m − S0 s0 ,
pS0 S 0 0
and γS = 1.5086.

180