See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/363134171

Fiscal Policy, Human Capital and Economic Growth: Evidence from Vietnam

Article · August 2022

CITATION READS

1 722

2 authors, including:

Insang Hwang
International Christian University
15 PUBLICATIONS 59 CITATIONS

SEE PROFILE

All content following this page was uploaded by Insang Hwang on 31 August 2022.

The user has requested enhancement of the downloaded file.

 Fiscal Policy, Human Capital and Economic Growth:
Evidence from Vietnam

Insang Hwang* Duc Chinh Vu†

August 31, 2022

Abstract

Using Vietnamese provincial data from 2015 to 2020 and applying pooled OLS and fixed-
effects panel regression, the study examines how the fiscal policy and human capital influ-
enced Vietnam’s economic growth. The econometric results show that government con-
sumption expenditures have a significantly negative impact on economic growth, while
positive impact of the government expenditures on development and investment on eco-
nomic growth was found. The results imply that excessive public spending at a certain
level could harm economic growth. In order to investigate whether budget surplus or
deficit could support economic growth, the study used a dummy variable that represents
the deficit and surplus. The coefficients of the dummy variable have a negative sign on
economic growth. These results are possibly caused by unreasonable tax composition or
unproductive public expenditures. The study also sheds profound light on the role of hu-
man capital in Vietnam’s economy. Human capital variables, in this paper, represent the
average years of schooling of the active labor force, which can be considered the quality
of human capital. According to the econometric results, the human capital with higher
education harms economic growth. However, the human capital with lower education pos-
itively affects economic growth. These outcomes imply that the Vietnamese economy still
depends on the unskilled labor forces for development. Also, public spending on education
does not support economic growth. This finding conceivably proves that public expendi-
tures on education seems likely to be a consumption rather than an investment in educa-
tion. Thus, to improve the economic performance, education policy, including spending
on education, should be significantly reformed productively by focusing on the quality of
education instead of pursuing anecdotal degrees in education.

Keywords: Fiscal Policy, Human Capital, FDI, Openness, Economic Growth, Vietnamese
Economy, Panel Regression Analysis on the Vietnamese Provinces.

JEL Classifications: E62; F14; O11; O47; O53

* Corresponding author: International Christian University (ICU), Dept. of Economics and Business, Osawa
3-10-2, Mitaka-city, Tokyo, Japan. Email: hwang@icu.ac.jp.
†
General Department of Taxation of Vietnam, Email: chinhducvu@gmail.com.

1
1 Introduction

The overall objective of this study is to analyze the impact of tax revenue, public expenditure,
human capital on economic growth in Vietnam in 2015-2020. The specific objectives of the
study are; a) Analyze the relationship of fiscal variables, such as domestic tax revenue, public
expenditures including expenditure on development and investment, recurrent expenditure, and
public expenditure on education as another variable for human capital and economic growth.
b)Investigate the impact of human capital on economic growth. c) Policy implications are
drawn from research findings.

Fiscal policy has been recognized as a vital instrument of government to maintain sustainable
economic growth both in the long term and short term. The reasonable tax policy encour-
ages social investment, creates jobs, reduces the unemployment rate, then stimulates economic
development. The rational, effective allocation of expenditures could help reduce inequality,
expand production and business, create jobs, and promote economic development.

The expenditure structure in Vietnam is still unbalanced. Recurring expenditures that primarily
maintain the regular operation of state administrative agencies are still large and inefficient.
The administrative apparatus is cumbersome, with many civil servants who work inefficiently.
Current expenditures often account for nearly 20% of GDP and more than 70% of total pub-
lic expenditures while Expenditures on development investment are still low, overspreading,
accounting for just 26.4% of total public spending. Expenditures on education are classified
as and a portion of recurrent expenditures. The idea is that the expenditures on education in
Vietnam are more like public consumption than public investment.

Although a large number of extensive studies on fiscal policy and economic growth have been
conducted in the past years, few studies have examined the relationship between two sides of
fiscal aspects (revenue and expenditures), human capital, and economic growth in Vietnam.
In addition, the econometric model incorporates the fiscal policy variables with human capital
variables and the indicator of trade. Including the indicator of trade can improve the accuracy
of the fitted model. Significantly, the study also focuses on the impact of the quality of active
labor forces on economic growth in Vietnam instead of using the schooling enrollment rates
that proxy for human capital variables. Using schooling enrollment rates does not qualify to
measure the human capital, which can reflect the economic performance (Hanushek, 2013;
Hanushek & Kimko, 2000; Pritchett, 2006). The human capital measurement is calculated by
the average schooling of the active labor force, multiplying the number of the corresponding
labor forces. Using this measurement instead of schooling enrollments as a variable proxy for
human capital variables implies the quality of active labor forces that have not been conducted
in Vietnam literature.

Fiscal policy can be understood as government interventions to the tax system and government
spending to achieve macroeconomic goals such as economic growth, job creation, price stabil-
ity, and inflation. Fiscal policy is a crucial macroeconomic instrument used by governments
to effectively mobilize, distribute, and allocate to implement economic growth goals. Keynes
(1936) argued that public expenditures on infrastructure, education, health could support the
long-run economic growth. Fiscal policy affects economic growth both in the short and long
term. In the short term, fiscal policy affects actual output and inflation to stabilize the economy.
In the long term, the fiscal policy with the function of adjusting the allocation of the economic

2
resources is most important to achieve the goal of economic growth.

Humans can make a difference with knowledge. The knowledge economy is considered an
inevitable and objective trend of human society and the emergence of the fourth industrial
revolution. Human capital plays an increasingly important role. In particular, human capital
can affect: The growth of an individual’s wages, the productive capacity of enterprises and the
economic growth of a country (Schultz, 1961); core competencies or competitiveness of the
enterprise (Lepak & Snell, 1999); labor productivity of workers (Lucas Jr, 1988); the ability of
workers to find jobs (Vinokur et al., 2000).

2 Literature Review

2.1 Fiscal Policy and Economic Growth

Fiscal policy is a crucial instrument in governing macro-economic policy. It directly impacts

the budget balance and supports promoting sustainable economic growth. The government
uses fiscal policy, which includes spending and tax instruments, to intervene effectively in the
economy.

A considerable amount of literature has been published on the relationship between fiscal policy
and economic growth. In the theoretical frameworks, Keynes (1936) stated that fiscal policy
could either crowd-in or crowd-out the private investment depending on how the policy has
been designed and implemented. Solow (1956) built a growth model and assumed that the
effect of income tax on the saving-investment ratio depends on how the state uses it directly
in capital formation or consumption. Romer (1986) established another theoretical growth
model in which policy variables such as tax rates could conceivably have a long-run effect on
economic growth.

With the rise of growth theory in the 1980s, there were extensive empirical studies on public
expenditure and economic growth. Mankiw et al. (1992) used the simple model to illustrate the
potential of human capital as a share of GDP of education investment. The effect of education
investment supports per capita output and raises the steady-state stock of physical capital per
worker. However, this study has not clarified the role of government spending on growth. Be-
fore Mankiw et al. (1992), Kormendi and Meguire (1985) and Barro (1991) exploited data from
98 countries worldwide and computed growth rates over a long period. They used multivariable
regression analysis with many explanatory variables to account for differences in growth rates
across countries. However, these two studies showed different results. Kormendi and Meguire
(1985) show that government consumption spending does not affect growth, while Barro (1991)
shows that government consumption spending harms economic growth, and public investment
shows the reverse association with private investment. In addition, to explain the lower level of
growth rate than other countries in Africa and Latin America countries, he used dummy vari-
ables to proxy its continent and held other explanatory variables. The results reveal that both
dummy variables and fertility rates are significantly negative with economic performance, but
the investment ratios are also insignificantly negative. The unexplained relative weak growth
performance could be explained by missing regularities in his specification.

3
There are largely consistent results that Tax-financed-government expenditures significantly af-
fect economic growth but need to be restructured toward more productive or innovative rather
than increasing tax rates. Gupta et al. (2005) used a sample of 39 low-income countries during
the 1990s to test the impact of fiscal consolidation and composition of public expenditure on
economic growth. In this empirical study, the researchers indicated that fiscal consolidations
were not harmful to either long or short-run growth in 1990-2000. They reported that cutting
some public expenditures but protecting capital expenditures can promote higher growth rates
than adjustments based on increasing revenue. Consistent with Gupta et al. (2005), Angelopou-
los et al. (2007) used panel data and established an endogenous growth model to test the impact
of the composition of government spending and taxation on economic growth for the period
of 1970-2000 of 23 OECD countries. They found that non-productive expenditures and la-
bor income tax rates have a negative impact on economic growth, while capital and corporate
income tax rates have a positive relationship. They concluded that the government needs to re-
allocate public expenditures towards productive tax-financed spending, and tax reform can be
a substantial gain for growth. Using this approach, Afonso and Alegre (2011) applied the panel
data techniques to evaluate the effects of two sides of the budget (the allocation of taxes and
public expenditures) on economic growth in 15 European countries for the period 1971-2006.
They found that public investment has a crowd-in effect on private investment that provokes
economic growth through the revenue-financed public expenditures.

However, several studies indicated that tax variables and economic growth have a harmful link.
Romero-Avila and Strauch (2008), using time-series patterns over the last 40 years in 15 Eu-
ropean countries, investigated the relationship between fiscal variables and economic growth.
Their findings revealed that direct taxation harms the growth rate of GDP per capita and a
shred of robust evidence that distortionary taxation affects economic growth in the medium run
through its impact on private capital accumulation. The issue of direct taxation in relationship
with private investment and economic growth was dealt with, similarly, by Johansson et al.
(2008), who investigated the effects of tax structure changes on economic growth as GDP per
capita and suggested that direct taxes, such as corporate taxes, and personal income taxes are
most harmful to growth. However, in contrast with other empirical findings, indirect taxation
categorized into non-distortionary groups, such as the consumption tax, was found to have a
significantly negative relationship with growth. Simultaneously, by splitting data on the manu-
facturing sector and individual companies to analyze whether tax designs impact the significant
productivity gains of firms and then economic growth, they suggested that lowering corporate
tax rates for small firms does not seem to cause growth and that high marginal personal income
tax rates can minimize growth by reducing the investment activities of entrepreneurs.

Several studies categorized expenditures and taxes into productive or unproductive expen-
ditures, distortionary or non-distortionary taxation to investigate their impact on economic
growth. Most of them have a consistent outcome that productive expenditures or non-distortionary
taxation positively link with growth. Paparas et al. (2015) used panel data techniques for 15
European countries and ten years from 1995-2005. They employed the Origin Least Square
(OLS), fixed effects models, random-effects models, and GMM estimators to test the impact of
several types of spending or taxation on economic growth. The public expenditures and taxes
were divided into productivity/unproductive expenditures and distortionary/non-distortionary
taxation. The purpose of this categorization was to help to determine which variables im-
pact economic growth precisely. They found a positive effect on the economic growth from
infrastructure and education spending, which were categorized into productive expenditures.

4
However, they found a negative impact of distortionary taxation on economic growth while
no significant impact of deficit on economic growth was found. Additionally, no evidence of
the relationship between openness and growth was detected. Consistent with Paparas et al.
(2015), Hrnjic and Brankovic (2017) found a positive impact of productive expenditures on
economic growth, such as expenditures on human capital, property protection, and social in-
vestment, while a negative one from distortionary taxation. Unlike most studies mentioned
above, Stoilova (2017) categorized direct taxes in distortionary groups, such as personal in-
come tax and property tax, which support economic growth, while value-added tax negatively
impacts economic growth in EU-28 economies.

The inconsistent outcomes with the above studies have been found on the relationship between
total government spending, the composition of government spending, and economic growth,
such as Devarajan et al. (1996) and Ghosh and Gregoriou (2008). With data collected from
43 developing countries in over 20 years, Devarajan et al. (1996) showed a rather remarkable
result: An increase in investment spending, such as capital investment, transport and communi-
cation, health, and education, have either negative or insignificant impact on economic growth,
while an increase in current expenditure has a positive effect. This result appears anomalously.
They implied that possibly excess productive expenditures in the developing countries could
not support growth productively. Ghosh and Gregoriou (2008), using the GMM method, with
data collected from 15 developing countries over 28 years, also gave quite consistent results
with the above results. According to the results of their empirical analysis, current government
expenditure has an essential contribution to economic growth, but not capital spending. On
the revenue side, tax and non-tax revenue significantly affect the growth rate, while surplus or
deficit seems to be statistically insignificant.

2.2 Human Capital and Economic Growth

In the past several years, major theoretical and empirical models have explored the role of hu-
man capital and economic performance. The contribution of human capital has been recognized
in the growth theories, such as the augmented neo-classical growth models and endogenous the-
ories. Furthermore, almost all empirical outcomes favor human capital as the important factor
supporting economic growth in the long run. However, the relative number of empirical stud-
ies showed an insignificant or ambiguous relationship between human capital and economic
growth. Human capital can be taken in several forms.

2.2.1 The Forms of Human Capital

In the early economic literature, the concept of human capital was first discussed by renowned
economists: Adam Smith and Alfred Marshall. The pioneer of political economy, Adam Smith
(1776), in his masterpiece The Wealth of Nations, stated that “Fixed capital consists of the
acquired and useful abilities of all the inhabitants or members of the society. The acquisition of
such talents, by the maintenance of the acquirer during his education, study, or apprenticeship,
always costs a real expense, which is a capital fixed and realized as it were, in his person.
Those talents, as they make a part of his fortune, so do they likewise of that of the society to
which he belongs. The improved dexterity of a workman may be considered in the same light

5
as a machine or instrument of trade, which, facilitates and abridges labor, and which, though
it costs a certain expense repays that expense with a profit.” In the Principles of Economics
which is the leading political economy textbook, Alfred Marshall (1890) included that “the
most valuable of all capital is that invested in human beings.”

In the research literature on economic growth related to human capital, human capital is a
broad concept. Schultz (1961), Mincer (1970), Barro (1991), Islam (1995), Baldacci et al.
(2003), Afridi (2016), and among others consider learning technology, school enrollment, or
average schooling year a component of human capital.

However, the role of human capital measured by formal schooling attainment has become con-
troversial. Pritchett (2006), Hanushek and Kimko (2000), and Hanushek (2013) argued that
schooling attainment does not guarantee economic performance. It thus does not entirely proxy
for human capital. This view makes sense by the fact that many developing countries have been
improving the level of school attainment but the much less contribution of education to growth
(Pritchett, 2001). Therefore, more empirical studies have used the quality of schooling as a
proxy for the human capital variable.

2.2.2 The Role of Human Capital and Economic Growth

The majority of previous sources have identified the positive impact of human capital on eco-
nomic growth. In the original neoclassical model by Solow (1956) and Swan (1956), they
explained that economic growth is augmented by the accumulation of physical capital, sav-
ings, and labor but with exogenous technological progress. The growth rate of the economy is
consistent with the growth rate of the labor force. Unlike the other early researchers who con-
sidered technological change the exogenous factors for returns, Romer (1986) had an advanced
step to emphasize the role of growth by the endogenous effects. He argued that knowledge
is encouraged by consumers’ output and only knowledge is an augmented factor rather than
the endogenous technology integrated into physical capital and human capital. Barro (1991)
used the post-war dataset of 98 countries in the period 1960-1985 to test the impact of school
enrollment rate (primary and secondary school) to the total population of the corresponding
age group, fertility rates that proxy for human capital variables and the growth. The results are
inconsistent with the previous evidence that the countries with a low level of per capita income
have a higher growth rate than the developed income level. However, if the human capital holds
constant, a positive correlation with the growth was found. He concluded that if the poorer
countries want to catch up with the developed countries, they must improve the human capital
through which accumulates the physical capital. The countries with higher initial human capi-
tal have higher fertility rates and higher physical investment to GDP. Sianesi and Van Reenen
(2000); Temple (2002) are two studies that summarized previous empirical studies. Sianesi and
Van Reenen (2000) mainly summarized the studies examining the relationship between formal
schooling and economic growth in general, while Temple (2002) summarized the studies on
formal education. The results confirmed the impact of both education and training and social
capital on economic growth in OECD countries. Zhu and Li (2017) used OLS estimation for
the years 1995-2000 of 216 countries to investigate the impact of economic complexity and hu-
man capital on economic growth in which secondary education and higher education as a proxy
for human capital. The results may imply that the countries with a higher initial human capital
level will have a more significant long-term growth rate. In addition, both secondary education

6
and higher education positively impact a country’s economic growth. However, the empirical
results show that secondary education has a relatively more substantial positive effect on long-
and short-term growth. The development of human capital is conducive to improving produc-
tion capabilities in an economy. Focusing more on cognitive skills and quality of schooling
rather than concentrating on the conventional measures of human capital as inputs, Hanushek
and Kimko (2000) estimated the data of 38 countries from 1965-1991 and measured the test
score labor-force on international mathematics and science test. They concluded that labor-
force quality has a strong, stable, and consistent relationship with economic growth. Quality
also has a causal impact on growth. However, higher quality through investing resources in
schools has no connection with growth.

However, considerable studies have explored the negative, ambiguous, or mixed results of hu-
man capital on economic growth. Islam (1995), instead of using a single cross-country re-
gression technique that contains omitted variables bias like Mankiw et al. (1992) had done,
employed the panel cross countries regression (LSDV – Least Squared Dummy Variables) to
examine the convergence by controlling the different aspects in the production function of
an individual country using the same Summers-Heston (1988) data set with MRW and Barro
(1989). The results show that the improvements in the country-effects, which mean the im-
provements in technological progress, institutions leads to the higher transitional growth rate.
These improvements have a conducive effect on population growth rate and saving rate. The
inclusion of human capital extended the statistic model. They found a positive correlation be-
tween human capital and the technical progress factor. However, by entering the schooling at
all levels as the human capital variable into the augmented production function, the outcome
was mixed and ambiguous across country samples. No impact of human capital on growth was
found. Even negative signs or the impact of human capital was minimal.

The negative effect of human capital on growth was also found by Benhabib and Spiegel (1994);
Pritchett (2001). In a similar vein, Nonneman and Vanhoudt (1996) extended the model used
by Mankiw et al. (1992) by adding the variable of endogenous accumulation of technological
know-how in physical capital and human capital, they only found a statistically significant in-
vestment shares in physical and technological know-how to growth. However, the influence of
human capital for OECD countries is less important. Kumar (2006) investigated the relation-
ship between human capital (schooling attainment) and economic growth using the neoclassical
framework and endogenizing technological progress as a function of the stock value of human
capital. Using the GMM method for the year: 1960-2000 of 94 non-oil countries, Both dif-
ference GMM and system GMM have the same result that the human capital as one of the
inputs has a positive impact on growth but is not significant. The understate of human capital
to growth may be explained by the heteroscedastic error, structure of the regression model,
missing values of countries’ samples, and the schooling attainment may not reflect the actual
quality of education of a country.

Some studies examine the role of human capital in the economic growth of a single country us-
ing subnational or time-series datasets. Zhang and Zhuang (2011) examine the effect of human
capital on economic growth at the Chinese city/province level. The GMM method shows that
higher education plays a more important role than primary and lower secondary education. Fur-
thermore, evidence suggests that the role of human capital components on economic growth by
region is related to the level of its development. Developed provinces benefit more from higher
education, while underdeveloped provinces depend on primary and lower secondary education.
Afridi (2016) examines the relationship between human capital and the economic growth of

7
Pakistan on a time series from 1972 to 2013, where human capital is represented by the pri-
mary school enrollment rate, birth, infant mortality rate. He found the important role of human
capital in promoting economic development and affirmed that education and health need to be
concentrated. While investing in the health and education sectors does not appear to work in
the short term, it does produce better long-term outcomes.

Recently, Affandi et al. (2019) employed average years of schooling and national examination
results proxy for human capital covering the period 1985–2014 for 30 provinces in Indone-
sia. The result shows that quantitative schooling at secondary and higher education (e.g., years
of schooling) is significantly important for growth. However, cognitive skills are more criti-
cal. The implication focuses on higher education infrastructure that enhances higher economic
growth.

Public expenditures on education are also treated as the human capital variable by numer-
ous authors, such as Al-Yousif (2008), Awaworyi Churchill et al. (2015), Haini (2020), and
Suwandaru et al. (2021). Al-Yousif (2008) employed the Granger-causality test to examine
the direction of the relationship between education expenditure as a proxy for human capital
and economic growth in six GCC (Gulf Cooperation Council) countries using time-series data
over the period 1977-2004. The analysis’s outcomes are mixed across countries. He concluded
that human capital as public expenditure on education and economic growth has a causal rela-
tionship contrary to much of existing studies that human capital has a unidirectional effect on
economic growth. The different outcomes made him suggest more studies in countries with
similar socio-economic conditions, and variables that proxy for human capital need to be in-
vestigated. Awaworyi Churchill et al. (2015), based on 306 estimates drawn from 31 primary
studies, conducted a meta-analysis to examine the relationship between expenditure on educa-
tion or health (GEH) and economic growth. The positive effect of GEH on growth was found.
However, the association between health and growth is negative. The combination of education
and health expenditure has a positive effect on economic growth. Haini (2020) examines the
effect of health and education investments on economic growth in Chinese provinces from 1996
to 2015 using a spatial autoregressive model. Both health and education investments are posi-
tive and significant to economic growth. However, the results show education investment has a
stronger effect than health investment in spurring economic growth. The most recent analysis
has examined the relationship between public expenditure in the educational sector and eco-
nomic growth by Suwandaru et al. (2021). They employed Autoregressive Distributed to test
the relationship between those variables in Indonesia over the time-series period 1986-2018.
The result shows that public expenditure on the educational sector is statistically insignificant
in both the long and short-run but by different directions, positive in the long run and negative
in the short run.

2.3 Critique on The Prior Studies

It is possible to identify some shortcomings of the empirical discussed studies. The majority
of discussed studies, including the studies conducted in the 1980s and 1990s, used the iden-
tical data set that was outdated when the studies were conducted. They attempted to explain
the empirical evidence consistent with the neoclassical growth theory, but the outcomes dif-
fered across studies. Systematic differences in parameters may explain the heterogeneity across
countries due to the missing-information data set from several sources, including low-income

8
and advanced developed countries.

Since studies were conducted relating to fiscal policy aspects with a certain homogeneity in
the database when using the data collected from the group of countries, such as OECD, Eu-
ropean, developing countries. The OLS model used in several studies above may cause the
measurement error of unobserved terms, producing biased estimates.

Moreover, school enrollment rates are the most widely used proxy for human capital, which
does not adequately capture the actual human capital relative to the growth. The school enroll-
ment thus is an educational component of human capital. The human capital in this form is a
flow variable, the students who have been getting the education courses become the labor force
in the future. Therefore, the growth of educational attainment depends on not only the school
enrollment but also the rate of students entering and leaving the labor force (Pritchett, 2001).
The enrollment rates, which proxy for human capital variable, also have a flaw that contradicts
with the survey from many developing countries conducted by Pritchett (2001). Moreover, only
primary or/and secondary school enrollment are used by most of the studies but without any
clear or compelling explanation for why primary, secondary, and tertiary enrollment should be
included or excluded.

Finally, using the cross-countries data set with differences in socio-economic conditions can
lead the difficulties in making the policy prescriptions for specific countries.

3 Overview on the Vietnamese Macro-economy

The Vietnam’s economy has made remarkable growth in recent past years. The average growth
rate in the period 2010-2019 was around 6.31%. In the context that the economy of Vietnam
and the world are facing many difficulties and challenges, this is a relatively high growth rate,
the top growth rate in the world and the region. In particular, Vietnam achieved a growth rate

Figure 1

Real GDP Growth Rate of Vietnam from 2010 to 2020

Source: Authors’ calculation from the GSO data.

9
of 7.08% in 2018, the highest rate in the past ten years. The sharp decrease of the growth rate
in 2020 to 2.91% could have been caused by the spreading of the Covid-19 pandemic. The
growth rate of real GDP in Vietnam from 2010 to 2020 was represented in Figure 1.

Total public expenditures shown in Figure 2 accounted for 41.1% of GDP on average in 2010-
2020, a high level compared to the region and countries with similar development levels. The
structure of expenditures changes in the direction of current spending, accounting for an in-
creasing proportion. The current expenditures account for 19.7% on average as a share of GDP
and more than 70% of total public expenditures, while spending on development and invest-
ment accounts for 7.4% of GDP.

According to the Vietnamese Law on State Budget, the current expenditures are used to main-
tain the operation of the State apparatus, political organizations, socio-political organizations,
and other public spending. Therefore, the significant proportion of current expenditures in GDP
could have been explained by the failure to reduce the administrative cost and narrow the un-
necessary state management apparatus. Moreover, in the context of the economy facing many
difficulties, improvement of the state budget expenditure allocation should be considered, along
with budget-saving has been thoroughly applied.

Figure 3 represents the State revenue composition in 2019. The state budget revenue includes
revenues from taxes, fees, non-tax revenues collected from public services supplied by state
agencies, and grants. Thereby, tax revenue in 2019 accounts for the largest share of the total
state budget, about more than 70%, while non-tax revenue also accounts for a considerable size
of the total budget revenue.

Figure 4 describes the increase in the total domestic revenue as a percentage of GDP from
18% in 2010 to 26.8% in 2015, 34% in 2019, and 33.55% in 2020. The State budget revenue
depends more on domestic economic growth and development potentiality.

Figure 2

Total Public Expenditure in GDP at Constant Price 2010

Source: Author’s calculation from the Ministry of Finance data.

10
Figure 3

State Revenue Composition in 2019

Source: Author’s calculation from the Ministry of Finance data.

Figure 5 shows that corporate tax, value-added tax, and non-tax revenue account for the largest
share of GDP, at 7.24%, 9.7%, and 7.19%, respectively. Non-tax revenues can be considered
distortionary because businesses and citizens pay them for public or administrative services
supplied by state agencies. They often tend to affect firms’ profits directly or increase ad-

Figure 4

The Total Domestic Revenue in GDP from 2010 to 2020

Source: Author’s calculation from the Ministry of Finance data.

11
Figure 5

Tax Composition in 2019 (Percentage in GDP at constant price 2010)

Source: Author’s calculation from the Ministry of Finance data.

Figure 6

FDI Flows in Vietnam from 2010 to 2020 (Billion Dollars)

Source: Author’s calculation from the GSO data.

ministrative costs for citizens. However, this non-tax revenue accounts for almost the highest
proportion of GDP.

Vietnam has experienced sharp increases in attracting FDI flows over the years. The trends of
FDI flows are represented in Figure 6. There was a monotonically increase in FDI flows from
2011 to 2019 and a slight decrease in 2020, possibly due to the Covid-19 pandemic.

FDI, in general, made a remarkable growth in the period 2010-2020. It is attributed to the
commitment demonstrated by the government to promote the FDI flows. Although the average
registered FDI capital in the period 2010-2020 was 26.42 billion dollars, accounting for about

12
21.4% of GDP on average, implemented capital only accounted for 57.3% of the total registered
FDI capital.

Figure 7

International Trade Value in Vietnam from 2010 to 2020 (In percentage of GDP)

Source: Author’s calculation from the GSO data.

From Figure 7, the value of total import-export turnover in 2010-2020 accounts for about 86.2%
of GDP on average. The export value in GDP is approximately equivalent, 86.3% and 86% of
GDP, respectively. In 2017, there was high growth in international trade turnover in which
both export and import values increased, which can be attributed to the positive recovery of
the world economy. Vietnam’s national competitiveness also made a great stride compared to
previous years. However, the import trend increased slightly in 2020 compared to 2019. In
contrast, the export value did not grow in 2020, possibly due to the impact of the covid-19
pandemic.

4 Growth Model of Fiscal Policy and Human Capital

In the theoretical literature on investment and economic growth, the Neoclassical economists
maintained that the equilibrium is reached at the point when the level of labor is just enough to
ensure effective aggregate demand, economic efficiency will be optimized if there is a positive
impact on aggregate demand and the government must intervene in the economy to overcome
the imbalance, in order to increase the employment rate. From a theoretical point of view,
human capital is critical for long-term growth by supplying effective labor to the national econ-
omy with higher productivity. In this paper, the model employed Overlapping Generation
Model (OLG), which is identical to that in Glomm and Ravikumar (1997). In this model, the
state revenue is assumed unchanged over time, and the tax is only financed for government
expenditures. Hence, we consider the endogenous infrastructure and education expenditures
separately, respectively. In order to examine the relationship between government expenditures
on economic growth, the paper simply incorporates the individual behaviors, producer’s behav-

13
iors, and government behaviors and finds how the government policy will affect the incentive of
human capital and economic growth in the long run. The paper firstly describes the individual
sector.

4.1 Individual Behavior

Consider the overlapping-generation economy with the two generations live for two periods.
Each generation was born at t = 0, 1, 2, . . . . In the first period, the individuals have to work to
earn income wt , consumes c1 and saving their money for the future when they become old and
have no income. In the second period, the individual retires and consumes c2 . Each endowed
with physical capital k0 , human capital h0 . The individual preferences are represented in the
logarithmic form as follows:
Ut = ln(c1t ) + βln(c2t+1 ) (1)
0<β<1

The individual’s income is taxed by a uniform tax rate τ in order to finance the government
expenditures in infrastructure and education. The young and old generation at the time t faces
the budget constraints in each period:

c1t + s1t ≤ (1 − τ )wt ht (2)

c2t+1 ≤ [(1 − 1 − τ )rt+1 ]s1t (3)

c1t , c2t+1 ≥ 0

Where st is individuals’ savings. The individuals’ income is respected to tax at the a uniform
rate τ in the first period. In the second period, the generation will retire and have no income,
but they have an amount of savings that they loaned to firms and get the interest rt+1 and also
subjected to tax at the same tax rate τ .

By substituting the value of c1t and c2t+1 in equation (2), (3) into the utility function (1) and
maximize the utility of representative young generation, yields:

β(1 − τ )wt ht
st = (4)
1+β

The result (4) shows that the tax rate τ directly affects the individual’s savings rate. With the
τ ∈ (0, 1), the tax rate τ increases (decreases), the savings is decreases (increases) or vice
versa.

4.2 The Producer Behavior

The production function will take the form of Cobb-Douglas function as follows:

Y = AGθt ktα (nt ht )1−α (5)

14
Each firm hires capital and effective labors. The firm has to pay rental rate qt and effective
wages wt for rental capital and human labors, respectively. A represents augmented technology
which exhibits constant return to scale over the private factor so that the profits are zero in
equilibrium.

The private and public capital will be depreciated at the rates δk and δG , respectively. The
accumulation of private capital and public capital are described as follows:

kt+1 = ItP + (1 − δk )kt (6)

Gt+1 = ItG + (1 − δG )Gt (7)

Where ItP , ItG are private investment and public investment in the period t. The government’s
budget is assumed to be balanced.

Then, profit equation of a firm takes the form:

Π = AGθt ktα (nt ht )1−α − qt kt (8)

In an equilibrium, there are standard conditions are described by:

kt+1 = st , nt = 1

qt = rt + δk
ItG + Et = τ (wt ht + rt kt )

Given the initial k0 , h0 , and G0 , an equilibrium for the economy is the sequence of allocation:
{nt , kt , ht , ct , ct+1 , yt }∞ ∞
t=0 , and the price: {wt , rt , qt }t=0 . The agent’s savings equates to the
capital of the period t + 1, the total rental cost of capital is equal to the real interest rate and the
rate of capital depreciation. The public investment ItG and expenditures Et are balanced to the
total revenue as an assumption.

4.3 Public Expenditures on Infrastructures

Assuming that the public expenditures are financed through tax revenues where the revenue is
constant over time means the government has to sacrifice another spending to invest in infras-
tructures. Hence, this part considers H(ht , Et ) ≡ 1, and expenditures on human capital are
zero. In this model, long-run growth depends on the returns to the augmentable factors, public
capital G, and private capital k.

Profit maximization of firms respects to nt ht and kt , yields:

wt = (1 − α)AGθt ktα (9)

qt = αAGθt ktα−1 (10)

15
Thus, in an equilibrium, the condition kt+1 = st must be satisfied. Substituting kt+1 and the
result of wt in (9) into equation (4), yields.

β(1 − τ )
kt+1 = (1 − α)AGθt ktα (11)
1+β

For simplicity, suppose that δG = 1 and δk = 0, then:

Gt+1 = τ AGθt ktα (12)

Hence:
Gt+1 τ (1 + β)
= (13)
kt+1 β(1 − α)(1 − τ )

Assuming that α + θ = 1 and focus the balanced growth path. Taking the logarithm of (11) to
see the growth path of private capital.
βθ
ln(kt+1 ) − ln(kt ) = (1 − θ)ln( ) + lnA + (1 − θ)ln(1 − τ ) + θlnτ (14)
1+β

The growth rate of private capital is denoted by γ, taking the first derivative of γ respect to τ ,
the maximization point of γ at τ = θ, which is the output elasticity of public capital. Thus, the
income tax is not only impact the private incentives but also enhances the public capital which
helps for economic growth in the future.

4.4 Public Expenditures on Education

Considering the learning technology is the function as follows:

H(h, E) = Bhµ E 1−µ (15)

B > 0, µ ∈ (0, 1)

In this framework, set θ = 0 and ItG = 0 for all t. The long-run growth is the consequence of
both physical and human capital being augmentable.

Profit maximization by firms respects to nt ht and kt , yields:

wt ht = (1 − α)Aktα ht1−α (16)

qt kt = αAktα h1−α
t (17)

In an equilibrium:
β(1 − τ )(1 − α)Aktα ht1−α
kt+1 = (18)
1+β

16
Assuming δk = 0, the public expenditures on education is:

Et = τ Aktα h1−α
t (19)

The accumulation of human capital will become:

ht+1 = Bhµt (τ Aktα h1−α

t )1−µ (20)

Thus, the ratio between human capital and physical capital is according to:

ht+1 B(1 + β)τ 1−µ ht

= µ
( )αµ (21)
kt+1 β(1 − τ )(1 − α)A kt

The ratio converges monotonically to a unique steady state x⋆ given by:

1 B(1 + β) 1−µ 1
lnx⋆ = ln µ
+ lnτ − ln(1 − τ ) (22)
1 − αµ β(1 − α)A 1 − αµ 1 − αµ

By taking the logarithm of (18), the long-run growth rate is given by:

β(1 − α)A ht
lnkt+1 − lnkt = ln( ) + ln(1 − τ ) + (1 − α)ln( ) (23)
1+β kt

Substituting equation (22) into equation (23), yields:

β(1 − α)A 1 B(1 + β) 1−µ 1

γ = ln( )+ln(1−τ )+(1−α){ ln µ
+ lnτ − ln(1−τ )}
1+β 1 − αµ β(1 − α)A 1 − αµ 1 − αµ
(24)

Taking the first derivative of γ to find the maximization of the tax rate τ = 1 − α. The growth
maximizing tax rate is τ = 1 − α, which is output elasticity of human capital. Therefore,
high tax rate could theoretically disincentive human capital and diminish the benefits of public
expenditures on human capital investment.

5 Data and Econometric Methodology

5.1 Data Collection

The data was collected mainly from the General Statistical Office of Vietnam (GSO), the For-
eign Investment Agency, the Ministry of Plan and Investment (FIA, MPI), and the Ministry
of Industry and Trade (MOIT). The data will be constructed as the balanced panel data at the
subnational level over the six-year interval from 2015 to 2020. Because the data are from Viet-
namese sources in the local currency (Vietnam dong), they were transferred to the data into US
dollars. The definition of variables used in this paper is described in Table 1.

17
5.2 Definition of Variables

The study used seven main variables, which include per capita Regional Gross Domestic Prod-
uct, population growth rate, public expenditures on investment and development, current spend-
ing, expenditures on education, and two human capital variables. The two additional variables
are the registered foreign direct investment and international trade variable. Regional gross
domestic per capita is the real provincial domestic product at the constant price of 2010 and
divided by the total population. The population growth rate is calculated by the natural log-
arithm of t time minus the logarithm of the population at the t − 1 time. Under the State
budged Law, the expenditures on investment and development includes expenditures on funda-
mental construction and other investment described by Law. Current expenditures are financed
to maintain the State apparatus, political organizations, and other regular tasks. The spending
on education is a portion of current expenditures.

The HC1 measures the quality of active labor, which is calculated by the average years of
schooling in under tertiary multiplied by the number of active labors with responding level of
education, then divided by the total population. HC2 measures the quality of active labor, which
is calculated by the average years of schooling in tertiary or above multiplied by the number of
active labors with responding level of education, then divided by the total population.

According to Zhang and Zhuang (2011), the average years of schooling are decomposed into
two different levels of schooling – tertiary and before tertiary. The average years of schooling
at tertiary and above is 15.5 multiplies the number of population who gained the junior college
degree or over (15.5 ∗ L15.5 ) while the average years of schooling before tertiary is 6, 9, 12
(elementary school, junior high school, senior high school) multiplies the number of population
with the level of schooling respectively (6 ∗ L6 + 9 ∗ L9 + 12 ∗ L12 ). The study used the same
technique to calculate the human capital variable at the higher education (HC2) with the average
years of schooling of the active labors with tertiary and above (15.5). There is little difference
compared to Zhang and Zhuang (2011). Instead of using the average years of schooling at the
age of six and above in the population, the study uses the average years of schooling in the
active labor forces. However, due to the unavailability of data on the active labor forces with
different levels of schooling in Vietnam, human capital with lower education (HC1) is assumed
to be 9 for the average year of schooling of active labors. The human capital variables are thus
given by:
9⋆ L9
HC1 =
P OP
15.5⋆ L15.5
HC2 =
P OP
(9 ⋆ L9 ) + 15.5 ⋆ L15.5
HC =
P OP

Additionally, The foreign investment variable is registered foreign direct investment capital as
a percentage of GRDP. The international trade variable is the total export and import value as a
share of GRDP. The variables used in this study are described in Table 1.

18
Table 1

Describe The Variables Used in The Models

Variable name Notation Data Source Unit
Gross regional domestic product per capita PC GRDP GSO USD

Population Growth rate POP.growth GSO Percentage

Public expenditures on Investment and De- EDI GSO % GRDP

velopment

Current Expenditures CEX GSO % GRDP

Public expenditures on Education EDU GSO % GRDP

Total domestic revenue DREV GSO % GRDP

Human capital at lower education HC1 GSO % POP

Human capital at higher education HC2 GSO % POP

Total of human capital stock HC GSO % POP

The registered foreign investment capital FDI FIA, MPI % GRDP

Trade open Openness MOIT % GRDP

5.3 Econometric Methods and Data Analysis

The study employs the pooled OLS and fixed-effect panel regression method to test the impact
of fiscal variables, human capital, and some other explanatory variables on economic growth.
The reasons for employing the fixed-effect panel regression method are its advantages: (a) this
econometric is widely used and not quite complicated compared to other methods, such as the
Generalized Moment Method (GMM), which is very sensitive to using instruments; (b) the
regression results are produced-consistent across regressions; (c) it allows more observations,
more cross-sectional information, and more degrees of freedom; (d) It incorporates changes
within a province as well as changes across provinces; (e) Finally, it accounts for the impact of
province-specific attributes that possibly cause the OLS estimates bias.
yit = β1 Xit + β2 Zit + β3 Hit + αi + δt + ϵit

Where yit is GDP per capita in the logarithmic form, Xit denotes the basic variables (population
growth rate, public investment), Zit proxies for the fiscal variables, and Hit denotes the human
capital variables, αi is province-fixed effects, δt is year specific effects; i and t represent the
province and time interval, respectively. The ϵit is idiosyncratic error term which is assumed
stochastic (i.e., errors is not correlated with independent variables) and identically, indepen-
dently distributed: ϵit ∼ iid(0, δϵ2 ). In addition, the inclusion of other explanatory variables
could be added to the model, such as Openness, FDI, and the interaction variables.

19
In order to take the government budget constraint into account, the study includes the deficit
dummy variable, which is calculated by the total domestic tax revenue (DREV) minus the total
public expenditures (CEX + EDI). However, using the budget balance in the models may incur
the potentially perfect collinearity (Ghosh & Gregoriou, 2008), then the deficit dummy variable
can be used to avoid this issue. In the economic sense, the structural budget deficit can promote
a higher demand in expenditures and then encourage economic growth. Hence, in this case, the
study uses 1 proxy for deficit and 0 if surplus. The coefficient of the dummy variable of budget
balance (deficit) is expected to be significantly positive in the regression models. Also, to test
the impact of interaction terms on economic growth, the study uses several variables.

Table 2

Descriptive Statics

Statistic N Mean St. Dev. Min Max

PC GRDP 378 1, 700.490 1, 295.440 609.419 10, 779.110
POP.growth 315 0.009 0.009 −0.012 0.052
DREV 378 0.170 0.073 0.060 0.613
EDI 378 0.105 0.067 0.015 0.458
CEX 378 0.220 0.141 0.025 0.860
EDU 378 0.082 0.060 0.007 0.311
HC1 378 4.723 1.168 0.890 12.811
HC2 378 1.244 0.337 0.203 3.875
HC 378 5.967 1.487 1.093 16.686
FDI 378 1.477 1.809 0.000 10.880
Openness 378 1.573 2.210 0.003 13.487

From Table 2, the descriptive statics shows the high volatility of most variables across provinces
in Vietnam, such as per capita income, the labor forces at both levels of education, public expen-
diture, FDI, and Openness. The per capita income, fiscal indicators, FDI, and trade openness
across provinces are relatively uneven, reflecting that growth in some provinces outperforms
others.

The value of public expenditures on development and investment in the sample is quite high.
The highest is 45.8% of GRDP, the lowest is 1.5% of GRDP, and the average for the whole
study period is 1.05% of GRDP. The highest indices of current expenditures in GRDP range
from 2.5%-86%, while the public expenditures on education range from 0.7% to 31.1% of
GRDP.

The final indicators that also affect economic growth are the openness of the economy and
foreign direct investment. The openness of the economy is very high, 157.3% of GDP on
average, and FDI accounts for 147.7% of GDP, which proves that Vietnam’s economy has
deeply integrated into the world economy and this is a bright spot in attracting foreign direct
investment as an additional source of capital for growth. Clearly, there is an existing association
between FDI and trade open from the descriptive analysis.

20
6 Empirical Results

In this Section, the study estimates a system of simultaneous equations. The variables in the
equations are endogenized population growth rates integrated with fiscal policy, human capital
and other variables such as foreign direct investment and international trade open variables.
Since the joint estimation of the system equation generally gives more efficient estimates than
the estimation of every single equation separately, all variables were included to estimate the
growth model.

6.1 Fiscal Policy and Economic Growth

From Table 3, strong evidence of public expenditures on development and investment (EDI)
was found when testing the effect of fiscal variables and economic growth. The contribution
of EDI to the economic growth ranges from nearly 10% to 17.4% in both OLS and Panel
regression models. The statistically significant negative coefficient of current expenditures
demonstrates that excess spending at a certain level could harm economic growth. This makes
intuitive sense when the provincial governments of Vietnam spend a considerable share of
gross regional domestic product (GRDP), up to 22% on average, and the highest level is 86%
of GRDP, much higher than the average of the world (17.69%)1 .

To compare the difference between surplus and deficit by using a dummy variable that repre-
sents the deficit is significantly negative in most cases. The negative signs against economic
growth possibly prove that the composition of domestic tax revenue or unproductive expendi-
tures is possibly inappropriate, which can harm economic growth.

In columns 5 and 6, the model includes an interaction term of current expenditure and deficit
dummy variable to examine whether deficit complements government consumption in promot-
ing growth. The estimates show positively in both variables and significant at 10% and 1%
respectively in the OLS regression model but insignificantly positive in the panel regression.
However, the estimates omit many other variables, such as province-specific characteristics
that can potentially explain economic growth. Failing to control for these variables and also the
correlation of both variables can lead the biased regression.

Finally, the population growth rate is significantly positive in the pooled OLS estimates but
insignificantly negative in panel regression. However, the omitted variables could potentially
affect the growth that can incur the same problem of biased estimates. The negative sign of the
coefficient of the population growth rate could be plausible when it makes the GDP decrease at
the per capita level.

6.2 Fiscal Policy, Human Capital, and Economic Growth

Regarding the fiscal policy in Table 4 below, the regression results are consistent with the pre-
vious section. The expenditures on development and investment is almost significantly positive
1
see: https://data.worldbank.org/indicator/NE.CON.GOVT.ZS

21
Table 3

Fiscal Policy Regression Results

Dependent variable: log(PC GRDP)
OLS panel OLS panel OLS panel
linear linear linear
(1) (2) (3) (4) (5) (6)
POP.growth 12.461∗∗∗ −2.131 8.779∗∗∗ −1.846 5.983∗∗∗ −1.999
(1.404) (2.507) (1.394) (2.461) (1.502) (2.456)

log(EDI) 0.130∗∗∗ 0.177∗∗∗ 0.094∗∗∗ 0.176∗∗∗ 0.094∗∗∗ 0.174∗∗∗

(0.029) (0.016) (0.028) (0.016) (0.027) (0.016)

log(CEX) −0.701∗∗∗ −0.279∗∗∗ −0.564∗∗∗ −0.250∗∗∗ −0.745∗∗∗ −0.351∗∗∗

(0.027) (0.067) (0.032) (0.067) (0.052) (0.093)

dummy acc −0.288∗∗∗ −0.079∗∗∗ 0.242∗ 0.182

(0.039) (0.024) (0.128) (0.169)

log(CEX)*
dummy acc 0.252∗∗∗ 0.132
(0.058) (0.085)

Constant 6.311∗∗∗ 6.717∗∗∗ 6.315∗∗∗

(0.051) (0.073) (0.117)

Observations 315 315 315 315 315 315

R2 0.798 0.329 0.828 0.357 0.838 0.363
Adjusted R2 0.796 0.154 0.826 0.186 0.835 0.190
Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01

to the growth. In a similar vein, the current expenditures are significantly negative. The coeffi-
cients of the current spending ranging from 20% to 71% negatively impact GDP.

In order to examine the impact of human capital, the model includes the human capital vari-
ables. As shown in Table 4, columns (1), (2), (5), and (6), the labor with tertiary and above has
a significant and negative impact on GDP growth while the labor under tertiary has a signifi-
cant and positive impact on GDP growth, and becomes insignificantly positive in column (6)
when including the interaction terms. In other words, the unskilled labors contribute positively
to economic growth rather than the skilled labors in Vietnam. The results are in contrast to
the findings from prior studies. Zhang and Zhuang (2011) concluded that tertiary education is
more important than before tertiary education to GDP growth. Affandi et al. (2019) claimed
that both secondary and higher education play an important role in economic growth. However,
cognitive skills are more critical. The implication focuses on higher education infrastructure
that enhances higher economic growth.

22
 Table 4

Fiscal and Human Capital Regression Results

Dependent variable: log(PC GRDP)
OLS panel OLS panel OLS panel
linear linear linear
(1) (2) (3) (4) (5) (6)
∗∗∗
POP.growth 12.064∗∗∗ −2.324 12.328 −2.855 11.438∗∗∗ −2.191
(1.359) (1.537) (1.400) (2.202) (1.327) (1.694)
log(EDI) 0.070∗∗ 0.019 0.129∗∗∗ 0.132∗∗∗ −0.022 0.006
(0.031) (0.013) (0.029) (0.015) (0.027) (0.012)
log(CEX) −0.683∗∗∗ −0.203∗∗∗ −0.711∗∗∗ −0.314∗∗∗
(0.028) (0.042) (0.028) (0.059)
log(EDU) −1.547∗∗∗ 0.094
(0.351) (0.230)
log(HC1) 0.837∗∗∗ 0.712∗∗∗ 2.742∗∗∗ 0.317

23
(0.171) (0.103) (0.765) (0.465)
log(HC2) −0.711∗∗∗ −0.809∗∗∗ −2.471∗∗∗ −0.565∗∗∗
(0.160) (0.041) (0.599) (0.177)
log(HC) 0.095∗ −0.878∗∗∗
(0.050) (0.101)
log(EDU)*log(HC1) 0.706∗∗∗ −0.147
(0.249) (0.152)
log(EDU)*log(HC2) −0.627∗∗∗ 0.083
(0.209) (0.060)
Constant 5.040∗∗∗ 6.127∗∗∗ 1.805
(0.262) (0.110) (1.097)
Observations 315 315 315 315 315 315
R2 0.813 0.750 0.801 0.485 0.830 0.751
Adjusted R2 0.810 0.683 0.798 0.348 0.826 0.680
∗
Note: p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01
Additionally, the model includes interaction terms between public expenditure on education
and labors with tertiary or above, spending on education, and lower tertiary labor to investigate
whether education spending complementing labor with different levels of education has a pos-
itive impact on economic growth. The result shows a significant and negative coefficient of the
interaction term of education spending and the higher level of educated labor while the positive
and significant impact on economic growth in the OLS model. However, the insignificant effect
of interaction terms on economic growth in the panel regression (column 6). Including those
interaction terms could induce multicollinearity problems, making the estimate biased.

6.3 Human Capital, Foreign Direct Investment, and Economic Growth

The empirical results from Table 5 confirm that the HC1 represents the labors with lower ter-
tiary has a significant and positive effect on economic growth. In contrast, the labors with a
higher level of education (HC2) is a significantly positive relationship with economic growth.

The coefficients of the FDI variable is positive and significant in the first model of OLS but
insignificant positive in that of the panel model. The FDI coefficient in the second model
(column (4)) loses its significance when the interaction term between HC1, HC2, and FDI is
included. The coefficients of FDI is a statistically significant and positive impact on economic
growth in the OLS model but a statistically insignificant and positive impact on economic
growth in the panel model. Although the FDI data in this study use registered foreign direct
investment instead of implemented capital, the data of FDI can partly explain its contribution to
GDP growth. Besides this, according to a Vietnamese paper2 , the FDI sector contributed about
25.7% to Vietnam’s economic growth in the period from 2011 to 2019, accounting for about
13% of GDP in 2010 and 19.6% of GDP in 2019.

2
See more at https://tapchitaichinh.vn/su-kien-noi-bat/phat-huy-vai-tro-cua-fdi-trong-phat-trien-kinh-te-dat-
nuoc-333279.html

24
 Table 5

Human capital and FDI Regression

Dependent variable: log(PC GRDP)
OLS panel linear OLS panel linear OLS panel linear OLS panel linear
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
POP.growth 16.381 ∗∗∗ −3.429∗∗ 14.589∗∗∗ −3.634 ∗∗ 15.274∗∗∗ −3.417∗∗ 11.872∗∗∗ −3.333 6.986 ∗∗∗ −2.991
(2.127) (1.594) (2.088) (1.562) (2.281) (1.584) (1.386) (2.211) (1.487) (2.229)
log(EDI) −0.400∗∗∗ −0.010 −0.395∗∗∗ −0.010 −0.425∗∗∗ −0.008 0.038 0.101∗∗∗ 0.048∗ 0.104∗∗∗
(0.036) (0.012) (0.035) (0.012) (0.039) (0.012) (0.027) (0.014) (0.025) (0.014)
log.FDI 0.086∗∗∗ 0.005 0.563∗∗∗ 0.027 −0.018 0.130∗∗ −0.162∗∗∗ 0.016
(0.011) (0.008) (0.191) (0.071) (0.038) (0.059) (0.027) (0.056)
log(HC1) 1.575∗∗∗ 0.780∗∗∗ 1.570∗∗∗ 0.742∗∗∗ 1.527∗∗∗ 0.797∗∗∗
(0.266) (0.107) (0.263) (0.106) (0.283) (0.106)
log(HC2) −1.518∗∗∗ −0.838∗∗∗ −1.446∗∗∗ −0.860∗∗∗ −1.555∗∗∗ −0.827∗∗∗
(0.246) (0.043) (0.244) (0.042) (0.262) (0.043)
log.FDI*log(HC1) −0.325∗∗ −0.011

25
(0.137) (0.048)
log.FDI*log(HC2) 0.186 −0.046∗∗
(0.131) (0.022)
dummy FDI 0.167∗∗∗ 0.038∗
(0.044) (0.019)
log(EDU) −0.595∗∗∗ −0.191∗∗∗ −0.657∗∗∗ −0.195∗∗∗
(0.027) (0.038) (0.026) (0.038)
log(HC) −0.044 −0.905∗∗∗ 0.134∗∗ −0.842∗∗∗
(0.050) (0.107) (0.053) (0.106)
log.FDI*log(HC) 0.012 −0.068∗∗
(0.020) (0.033)
log.FDI*log(EDU) −0.070∗∗∗ 0.002
(0.011) (0.021)
Constant 4.097∗∗∗ 4.113∗∗∗ 4.021∗∗∗ 5.751∗∗∗ 5.311∗∗∗
(0.410) (0.401) (0.435) (0.111) (0.124)
Observations 313 313 313 313 315 315 313 313 313 313
R2 0.538 0.730 0.571 0.746 0.473 0.731 0.811 0.495 0.834 0.487
Adjusted R2 0.530 0.656 0.561 0.674 0.464 0.658 0.808 0.355 0.831 0.344
Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01
Therefore, in order to investigate the positive impact of FDI on the GDP of Vietnam, the model
used the dummy variable for FDI. Where the FDI as a share of GRDP is more than 1.5 is 1, and
others are 0. Including a dummy variable of FDI is more comfortable rather than categorizing
provinces into groups. Using a dummy variable for FDI is to confirm whether FDI has a positive
effect on economic growth. Interestingly, the results in columns (5) and (6) indicate a strongly
significant and positive impact on economic growth.

In model 3 (columns (3) and (4)), the interaction term between FDI and two levels of human
capital variables are included to investigate whether the levels of educated labor support the
FDI sector and then economic growth. The outcome shows the significant negative impact
of the interaction term between the skilled labors and FDI while the insignificant and slightly
negative impact of the interaction term between the unskilled labors and FDI on economic
growth. The interaction term between total human capital stock and FDI significantly harms
economic growth. It makes sense that most FDIs in Vietnam operate in the labor-intensive sec-
tors, such as textile, manufacturing, and assembly sectors. According to the report3 conducted
by ManpowerGroup Vietnam (2021), the proportion of low-skilled workers in the automobile
and motorcycle assembly industry accounts for 71%, 59% in the garment industry, 49% in
electronics, and 34% in the chemical/furniture industry. According to this report, the quality of
employees working in FDI enterprises still has a particular gap compared with the expectations
of foreign investors.

6.4 Human Capital, Openness, and Economic Growth

From Table 6, a comparison of the two results reveals that the quality of human capital relative
to the levels of educated labor forces has the likely same direction as FDI and human capital
levels composition. The coefficient of the lower level of education is significantly positive for
economic growth, while the higher level of educated labors has a negative relationship with
economic growth. It is consistent with the descriptive statics that the volume of trade is higher
in provinces with higher FDI attractive capital. Therefore, the lower educated labors support
more in the export and import situation and then economic growth.

The Openness represents the total volume of imports and exports as a percentage of GRDP has
a significant positive impact on economic growth in most of models. The positive coefficients
range from 3.3% to 17%.

Columns (5), (6), (7), and (8) show that public education spending does not support growth
productively. It can be explained that government expenditures on education are classified as
a portion of recurrent expenditures. Thus, the idea is that the expenditures on education in
Vietnam are more like a public consumption than a public investment. The proportion of ex-
penditures on people (salary expenses, salary-based items) possibly accounts for a large propor-
tion of the total recurrent expenditures at educational institutions. However, when combining
the openness and education spending, total labors with different levels of education (HC), and
openness, the coefficient of these variables becomes positive. It makes sense that public edu-
cation spending and total labors are important factors to facilitate the diffusion of trade open,
and in turn, trade open has a positive contribution to economic growth.

3
See ManpowerGroup Vietnam (2021).

26
 Table 6

Human Capital and Openness Regression Results

Dependent variable: log(PC GRDP)
OLS panel OLS panel OLS panel OLS panel
linear linear linear linear
(1) (2) (3) (4) (5) (6) (7) (8)
POP.growth 12.114∗∗∗ −2.624∗ 12.168 ∗∗∗ −2.524 11.430∗∗∗ −1.980 11.988∗∗∗ −1.862
(2.380) (1.553) (2.465) (1.546) (1.605) (2.108) (1.474) (2.145)
log(EDI) −0.414∗∗∗ −0.006 −0.414∗∗∗ −0.004 0.036 0.094∗∗∗ 0.040 0.095∗∗∗
(0.038) (0.011) (0.038) (0.011) (0.026) (0.013) (0.026) (0.014)
log(HC1) 1.430∗∗∗ 0.763∗∗∗ 1.412∗∗∗ 0.633∗∗∗
(0.280) (0.104) (0.375) (0.123)
log(HC2) −1.449∗∗∗ −0.775∗∗∗ −1.444∗∗∗ −0.701∗∗∗
(0.260) (0.044) (0.313) (0.050)
log(EDU) −0.592∗∗∗ −0.209∗∗∗ −0.591∗∗∗ −0.134∗∗∗
(0.026) (0.050) (0.024) (0.039)

27
log(HC) −0.040 −0.714∗∗∗ −0.236∗∗ −0.762∗∗∗
(0.049) (0.100) (0.101) (0.121)
Openness 0.049∗∗∗ 0.033∗∗∗ 0.037 −0.167∗∗ 0.020 0.170∗∗∗ −0.113∗∗ 0.013
(0.010) (0.008) (0.190) (0.073) (0.045) (0.052) (0.054) (0.048)
log(HC1)*Openness 0.008 0.135∗∗∗
(0.135) (0.050)
log(HC2)*Openness −0.003 −0.064∗∗∗
(0.116) (0.021)
log(EDU)*Openness 0.004 0.036∗∗
(0.014) (0.016)
log(HC)*Openness 0.068∗∗ 0.025
(0.031) (0.028)
Constant 4.184∗∗∗ 4.212∗∗∗ 5.737∗∗∗ 6.093∗∗∗
(0.430) (0.563) (0.114) (0.191)
Observations 315 315 315 315 315 315 315 315
R2 0.490 0.744 0.490 0.753 0.813 0.535 0.816 0.527
Adjusted R2 0.482 0.674 0.478 0.683 0.810 0.407 0.813 0.396
Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01
6.5 Regression with All Variables

In this Section, the study reports the Regression results from the fixed-effect model. The Haus-
man test can help choose the appropriate models between the fixed-effects and random-effects
models. The null hypothesis is preferred to the random-effects model and the alternative hy-
pothesis is preferred to the fixed-effect model. The fixed-effect model will be the best if the
p-value is significant (<0.05). Essentially, the test examines whether there is a correlation be-
tween the errors and regressors. The p-values of the Hausman test are mostly close to zero
across models. Therefore, the fixed-effects models would be chosen, and the Pooled OLS
models become biased.

Comparing the results when regressing variables separately and all variables, it can be seen that
the outcomes are likely identical in most of the coefficients. The regression results are reported
in Table 7.

For fiscal variables, while public spending on development investment positively affects eco-
nomic growth, recurrent expenditure harms economic growth. The finding is consistent with
several previous studies such as Barro (1991), Gupta et al. (2005), Angelopoulos et al. (2007),
and Mai (2012).

One unanticipated finding is that the coefficients of FDI are not significant though positive.
Therefore, to identify whether FDI has a significant effect on economic growth, the model in-
cludes the dummy variable for FDI. Interestingly, the dummy variable proxies for provinces
with a higher percentage of registered foreign direct capital in GRDP have a positive and sig-
nificant relationship to the growth.

Regarding the human capital variables, there was evidence that public education spending has
a negative relationship with economic growth across regressions. The labor with tertiary or
above has a negative impact on economic growth, while the labors with lower tertiary have a
positive impact on economic growth. In the meantime, the total human capital stock does not
support economic growth. The result could be explained that the effect of education on labor
forces in Vietnam does not stimulate the quality of human capital.

Similar to the previous analytical results, the regression coefficients of the interaction terms
between human capital at both levels of educated labor and FDI have a negative sign but in-
significance. Besides this, the coefficients of the interaction terms between human capital at
both education levels and Openness confirm that the unskilled labor forces support the trade
openness more than the skilled labor forces. The variations in per capita GRDP are explained
by close to 80% (except column 4) across models.

28
 Table 7

Fixed-Effect Panel Regression with All Variables

Dependent variable: log(PC GRDP)

(1) (2) (3) (4) (5) (6)

POP.growth −1.867 −2.146 −1.705 −1.790 −1.695 −2.249
(1.506) (1.534) (1.502) (2.066) (1.478) (1.500)
log(EDI) 0.020 0.001 0.022∗ 0.121∗∗∗ 0.022∗ 0.016
(0.012) (0.012) (0.012) (0.014) (0.012) (0.013)
log(CEX) −0.194∗∗∗ −0.189∗∗∗ −0.280∗∗∗ −0.178∗∗∗ −0.179∗∗∗
(0.041) (0.041) (0.056) (0.040) (0.041)
log(EDU) −0.098∗∗∗
(0.028)
log.FDI 0.005 0.003 0.005 0.006 0.056
(0.008) (0.008) (0.008) (0.008) (0.068)
log(HC1)*log.FDI −0.032
(0.046)
log(HC2)*log.FDI −0.021

29
(0.021)
log(HC1) 0.666∗∗∗ 0.714∗∗∗ 0.529∗∗∗ 0.665∗∗∗ 0.642∗∗∗
(0.102) (0.103) (0.121) (0.100) (0.102)
log(HC2) −0.756∗∗∗ −0.769∗∗∗ −0.691∗∗∗ −0.743∗∗∗ −0.780∗∗∗
(0.043) (0.044) (0.049) (0.042) (0.043)
Openness*log(HC1) 0.130∗∗∗
(0.048)
Openness*log(HC2) −0.056∗∗∗
(0.021)
log(HC) −0.713∗∗∗
(0.098)
Openness 0.031∗∗∗ 0.023∗∗∗ −0.161∗∗ 0.061∗∗∗ 0.031∗∗∗ 0.027∗∗∗
(0.008) (0.008) (0.071) (0.010) (0.008) (0.008)
dummy FDI 0.048∗
(0.025)
dummy acc −0.047∗∗∗
(0.014)
Observations 313 313 313 315 313 313
R2 0.767 0.758 0.775 0.556 0.777 0.774
Adjusted R2 0.701 0.689 0.708 0.433 0.712 0.708
Note: ∗ p<0.1; ∗∗ p<0.05; ∗∗∗ p<0.01
7 Summary and Policy Implications

7.1 Fiscal Policy

One of the aims of this study was to identify the relationship between fiscal variables and
economic growth. The current study found that government consumption expenditure harms
economic growth. The coefficient of current expenditure ranges from 20%-70% and is signifi-
cant at a 1%-5% level across regressors. The negative impact of consumption expenditures on
growth is in line with several empirical evidence, such as Barro (1991), Gupta et al. (2005),
Angelopoulos et al. (2007), Mai (2012), and Paparas et al. (2015). This result conceivably
reflects excessive recurrent spending that can harm economic development dramatically. The
description statics show the share of recurrent expenditure in GRDP such high as 22% on aver-
age and up to 86% at maximum, much higher than the average of the world (17.69%) [footnote
1].

The positive contribution of expenditures on development and investment to economic growth

ranges from nearly 10 percent to 17.4 percent in both OLS and Panel regression models. The
findings are consistent with Mai (2012), Paparas et al. (2015), and Hrnjic and Brankovic (2017).

Regarding the dummy variable of budgetary constraint (dummy acc), with the assumption that
total domestic revenue equates to the total state revenue and public expenditure on development
and investment, current expenditure is total public spending. This dummy variable is defined
by total domestic revenue over GRDP minus total public expenditure over GRDP. Where the
result from this calculation is negative, the dummy is 1, and the positive the dummy is 0.
The coefficient of the dummy variable is significantly negative to growth across regressions
which is in contrast to the theory of Ricardian Equivalence. The view of Ricardian Equivalence
argues that the current deficits will have equivalent effects on the overall economy in the way of
stimulating economic growth through accumulated demand from government spending. This
negative coefficient of the deficit possibly proves that the composition of domestic tax revenue
is possibly inappropriate. Also, The result implies that the distortionary taxes in the design of
tax structures can harm the growth. An increase in unproductive government spending may
also be the reason for the negative sign of the coefficient of a deficit dummy variable.

In a nutshell, increasing in unproductive government spending and the inappropriate tax struc-
tures may be the reasons for the negative sign of the coefficient of deficit dummy variable and
the current expenditures.

7.2 Human Capital

Prior studies that have noted the importance of human capital on economic growth reported
that human capital as one of the inputs positively impacts development. Still, higher education
is more critical to economic growth. Adversely, the results of this study indicate that human
capital represents the labor with tertiary level or above harms growth. In contrast, the human
capital variable representing the labor with lower education has a strong positive and significant
effect on growth. This finding is consistent with Hanushek and Kimko (2000), who found that
higher education in the labor force has no connection with growth but is contrary to that of

30
Zhang and Zhuang (2011), Affandi et al. (2019). They found that higher-skilled workers play
more role in economic development.

Although the inclusion of interaction variables, such as both level education of human capital
variables with FDI, public education spending, and trade open, did not make the discrepancy
of direction with the coefficients of two human capital variables (HC1, HC2) on economic
growth. This finding implies that Vietnam’s economy still depends more on simple labor than
on high-tech or skilled labor. Also, the combination between public education spending and
FDI has a negative impact on economic growth in OLS model while no impact was found in
panel regression. This result indicates that public education spending is not efficient for the FDI
sector. This is also consistent with the results reported by ManpowerGroup Vietnam (2021) that
the rate of labor retraining in FDI enterprises is up to nearly 80% to over 80% across sectors.

Public spending as a portion of current expenditures on education is not conducive to economic

growth. The finding makes sense that expenditures on education in Vietnam are more like
public consumption than public investment. However, when combining openness and education
spending; human capital with all levels of education (HC) and Openness, the coefficient of
these variables becomes positive. The results imply that education spending and total labors
are important factors in facilitating the diffusion of Openness, and in turn, Openness has a
positive contribution to economic growth. It is therefore likely that connections exist between
human capital and FDI or trade open in Vietnam.

7.3 Foreign Direct Investment and Openness

Contrary to expectations, this study did not find a consistent result of foreign direct investment
with growth across regressions. In reality, FDI plays an important role in the economic devel-
opment of developing countries, especially Vietnam. It accounts for more than 19% of GDP
in 2019. Therefore, the dummy variable for FDI was used to confirm the positive impact of
FDI on the GDP of Vietnam. The regression results show that FDI has a positive relationship
with economic growth. This outcome is consistent with that of Su and Liu (2016), who found
evidence for the positive contribution of FDI to Chinese economic growth.

Openness has an important influence on economic growth. It will significantly affect economic
growth, especially in developing countries. The study’s result indicates that trade substantially
contributes to the Vietnamese economy. This finding is consistent with that of Mai (2012),
who found the positive impact of Openness on economic growth over the twenty years in Viet-
nam. According to the statistics data collected, the share of export value accounts for 70% of
provincial GDP in Vietnam from 2015-to 2020. This level is higher than the average level of
the world (27%) and ASEAN countries in the same period (64%)4 .

7.4 Policy Implications

This section aims to provide policy implications that will be relevant to the context of Vietnam
to pursue the goals of economic development. Based on theoretical and empirical analysis,
4
Calculated from world Bank data at The WorldBank Group

31
some policy implications are derived as follows:

◦ With the essential function of state budget revenue, Macro-regulating the economy, and re-
sources redistribution to ensure social equality or to overcome market imbalances, the appro-
priate budgetary policy is apparently recognized as an essential driver of economic growth. Re-
structure tax design and public spending compositions toward more productive by minimizing
the taxation considered distortionary indices and unproductive government expenditures should
be tightly controlled, especially public consumption. Improve the efficiency of public invest-
ment by strictly controlling investment capital, focusing on efficiently productive projects, and
narrowing the development gap between provinces by investing more in infrastructure projects
that could support economic growth. Moreover, the theoretical model analyzes that when tax is
absorbed efficiently in productive spending such as infrastructure expenditures could not only
impact the private incentives and enhance the public capital, then economic growth in the fu-
ture. Also, an appropriate tax structure probably raises agents’ savings, which economically
will, in turn, support economic growth by absorbing savings into economic activities.

◦ The empirical results show that human capital plays a vital role in economic growth. In
this study, strong evidence that unskilled labor is significantly positive with growth. However,
the negative impact of human capital proxied by the active labor force with higher education
indicates that Vietnam’s economy still depends on the unskilled labor force. Also, the interac-
tion terms between human variables and FDI or Openness show that public education spending
is only productive for the lower quality of the labor force. Therefore, education and public
spending for education should be seriously considered when designing education policies.

Overall, this study strengthens the idea that the size of government expenditures should be
appropriately restructured more productively. The contribution of this study has been to con-
firm the existing literature that excess government expenditures could harm economic growth.
Similarly, if public spending financed by tax revenue is used efficiently and absorbed well
in economic activities, both tax and spending favor the private sectors, enhancing economic
growth. Therefore, tax composition should be taken into account when redesigning a more
productive tax system.

Also, the study provides an essential insight into the role of human capital in economic growth
in Vietnam. Both theoretical and empirical studies have confirmed the importance of human
factors as inputs for development, especially in the long run. Thus, to improve the economic
performance, education, including expenditure on education, should be significantly reformed
based on a more productive, skilled labor force by focusing on the quality of education instead
of pursuing anecdotal degrees in education.

7.5 Limitations

Although the study has successfully demonstrated the strong evidence of fiscal policy and hu-
man capital to economic growth, it has certain limitations in terms of the data sample. The
sample data was provincially representative of GRDP and other variables but would tend to
miss the Centre level of variables that could measure the impact of variables of interest more
properly. The number of time periods is relatively smaller than the number of observations,
possibly produce the fixed-effect coefficients less reliable (Hill et al., 2020). Also, lacking the

32
real implemented foreign direct capital as an FDI variable could cause the likelihood of the
effects of FDI in the regression models to be imprecise.

Besides, it is unfortunate that the study did not include the composition of taxes that could be
explained whether tax types categorized as distortionary/non-distortionary groups affect eco-
nomic growth. The unobserved heterogeneity of province-fixed effect could not be measured
and ignored due to the time-varying characteristics of provinces in the fixed effect model re-
gressions, which could produce bias estimates (Hill et al., 2020).

Therefore, these issues are an intriguing ones that could be usefully explored in further research.

8 Conclusion

Using the data collected mainly from GSO for six years interval from 2015 to 2020, the study
employed pooled OLS, fixed effect panel regression method to investigate the impact of fiscal
policy, human capital, and economic growth in Vietnam from 2015-to 2020. Also, including
FDI and international trade value variables in the models could make the regression results
precise.

The findings clearly indicate that fiscal policy and human capital play an important role in
economic growth. The investigation of fiscal policy variables has shown that the current ex-
penditures have a significant negative impact on economic growth. In contrast, expenditures
on development and investment significantly support economic growth. The dummy variable
was used proxy for the budget constraint is against economic growth, implying that the tax rev-
enue composition might be inappropriate with a significant portion of distortionary taxes, or the
allocation of the unproductive expenditure can harm economic growth. Therefore, both govern-
ment expenditures and tax revenue and their compositions should be appropriately restructured
more productively.

Regarding the relationship between human capital and economic growth, the study used the
levels of education of the active labor force proxy for human capital variables that can be
defined as the quality of human capital rather than schooling enrollment rates as human capital
variables. The human capital as the labor force with tertiary or above impacts significantly and
negatively on economic growth, while the human capital as the labor force with below tertiary
level has a positive effect on economic growth. The combination of human capital at each level
of education has the same direction of coefficient of two human capital variables. The findings
conceivably prove that the Vietnamese economy still depends on unskilled labor for economic
growth. Besides this, public expenditure on education has a significantly negative sign on
growth. It makes sense that public expenditure on education as a portion of current expenditure
mostly pays for salary and administrative governance and seems likely a consumption than an
investment in education.

Additionally, the models included registered foreign direct investment capital and trade open-
ness, which positively impact economic growth. The interaction term between FDI and two
different levels of education of human capital have opposite signs. The combination between
higher education (tertiary or above) human capital and FDI has a negative while between lower
education (below tertiary) human capital has a positive impact on growth. The results reflect

33
the fact that most FDI firms operate in the labor-intensive sectors in Vietnam. Similarly, the
interaction term between openness and human capital at all levels of education, openness, and
education spending support economic growth. There is likely an existing connection between
the total labor force, education spending, openness, also FDI.

This study has shed light on the importance of various factors on economic growth and thereby
provides a basis for deciding the scope and focus of public-sector activities in Vietnam. How-
ever, it faces some limitations, such as sample size and lacking several types of taxes or fees that
could explain the models more precisely. Also, the study did not control the unmeasured time-
variant effects, which might cause some bias. Therefore, further research should be undertaken
in these areas.

References
Affandi, Y., Anugrah, D. F., & Bary, P. (2019). Human capital and economic growth across
regions: A case study in indonesia. Eurasian Economic Review, 9(3), 331–347.
Afonso, A., & Alegre, J. G. (2011). Economic growth and budgetary components: A panel
assessment for the eu. Empirical Economics, 41(3), 703–723.
Afridi, A. H. (2016). Human capital and economic growth of pakistan. Business & Economic
Review, 8(1), 77–86.
Al-Yousif, Y. K. (2008). Education expenditure and economic growth: Some empirical evi-
dence from the gcc countries. The Journal of Developing Areas, 69–80.
Angelopoulos, K., Economides, G., & Kammas, P. (2007). Tax-spending policies and economic
growth: Theoretical predictions and evidence from the oecd. European Journal of Po-
litical Economy, 23(4), 885–902.
Awaworyi Churchill, S., Yew, S. L., & Ugur, M. (2015). Effects of government education and
health expenditures on economic growth: A meta-analysis.
Baldacci, E., Hillman, A., & Kojo, N. (2003). Growth, governance, and fiscal policy in low-
income countries: Understanding the transmission channels. IMF Working Paper.
Barro, R. J. (1989). The ricardian approach to budget deficits. Journal of Economic perspec-
tives, 3(2), 37–54.
Barro, R. J. (1991). Economic growth in a cross section of countries. The quarterly journal of
economics, 106(2), 407–443.
Benhabib, J., & Spiegel, M. M. (1994). The role of human capital in economic development
evidence from aggregate cross-country data. Journal of Monetary economics, 34(2),
143–173.
Devarajan, S., Swaroop, V., & Zou, H.-f. (1996). The composition of public expenditure and
economic growth. Journal of monetary economics, 37(2), 313–344.
Ghosh, S., & Gregoriou, A. (2008). The composition of government spending and growth: Is
current or capital spending better? Oxford Economic Papers, 60(3), 484–516.
Glomm, G., & Ravikumar, B. (1997). Productive government expenditures and long-run growth.
Journal of Economic Dynamics and control, 21(1), 183–204.
Gupta, S., Clements, B., Baldacci, E., & Mulas-Granados, C. (2005). Fiscal policy, expenditure
composition, and growth in low-income countries. Journal of International Money and
Finance, 24(3), 441–463.

34
Haini, H. (2020). Spatial spillover effects of public health and education expenditures on eco-
nomic growth: Evidence from china’s provinces. Post-Communist Economies, 32(8),
1111–1128.
Hanushek, E. A. (2013). Economic growth in developing countries: The role of human capital.
Economics of education review, 37, 204–212.
Hanushek, E. A., & Kimko, D. D. (2000). Schooling, labor-force quality, and the growth of
nations. American economic review, 90(5), 1184–1208.
Hill, T. D., Davis, A. P., Roos, J. M., & French, M. T. (2020). Limitations of fixed-effects
models for panel data. Sociological Perspectives, 63(3), 357–369.
Hrnjic, M., & Brankovic, A. (2017). Endogenous growth model: Evidence from east european
countries. Economic Review: Journal of Economics and Business, 15(1), 33–46.
Islam, N. (1995). Growth empirics: A panel data approach. The quarterly journal of economics,
110(4), 1127–1170.
Johansson, Å., Heady, C., Arnold, J. M., Brys, B., & Vartia, L. (2008). Taxation and economic
growth.
Keynes, J. M. (1936). The general theory of employment, interest and money (1936). The
Collected Writings of John Maynard Keynes, 7, 1971–9.
Kormendi, R. C., & Meguire, P. G. (1985). Macroeconomic determinants of growth: Cross-
country evidence. Journal of Monetary economics, 16(2), 141–163.
Kumar, C. S. (2006). Human capital and growth empirics. The Journal of Developing Areas,
153–179.
Lepak, D. P., & Snell, S. A. (1999). The human resource architecture: Toward a theory of human
capital allocation and development. Academy of management review, 24(1), 31–48.
Lucas Jr, R. E. (1988). On the mechanics of economic development. Journal of monetary eco-
nomics, 22(1), 3–42.
Mai, D. L. (2012). The impact of fiscal decentralization on vietnam’s economic growth (in
vietnamese). Ph.D. dissertation.
Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A contribution to the empirics of economic
growth. The quarterly journal of economics, 107(2), 407–437.
Mincer, J. (1970). The distribution of labor incomes: A survey with special reference to the
human capital approach. Journal of economic literature, 8(1), 1–26.
Nonneman, W., & Vanhoudt, P. (1996). A further augmentation of the solow model and the
empirics of economic growth for oecd countries. The Quarterly Journal of Economics,
111(3), 943–953.
Paparas, D., Richter, C., & Paparas, A. (2015). Fiscal policy and economic growth, empirical
evidence in european union. Turkish economic review, 2(4), 239–268.
Pritchett, L. (2001). Where has all the education gone? The world bank economic review, 15(3),
367–391.
Pritchett, L. (2006). Does learning to add up add up? the returns to schooling in aggregate data.
Handbook of the Economics of Education, 1, 635–695.
Romer, P. M. (1986). Increasing returns and long-run growth. Journal of political economy,
94(5), 1002–1037.
Romero-Avila, D., & Strauch, R. (2008). Public finances and long-term growth in europe: Evi-
dence from a panel data analysis. European Journal of Political Economy, 24(1), 172–
191.
Schultz, T. W. (1961). Investment in human capital. The American economic review, 51(1), 1–
17.

35
Sianesi, B., & Van Reenen, J. (2000). The returns to education: A review of the macro-economic
literature.
Solow, R. M. (1956). A contribution to the theory of economic growth. The quarterly journal
of economics, 70(1), 65–94.
Stoilova, D. (2017). Tax structure and economic growth: Evidence from the european union.
Contadurıa y administración, 62(3), 1041–1057.
Su, Y., & Liu, Z. (2016). The impact of foreign direct investment and human capital on eco-
nomic growth: Evidence from chinese cities. China Economic Review, 37, 97–109.
Suwandaru, A., Alghamdi, T., & Nurwanto, N. (2021). Empirical analysis on public expendi-
ture for education and economic growth: Evidence from indonesia. Economies, 9(4),
146.
Swan, T. W. (1956). Economic growth and capital accumulation. Economic record, 32(2), 334–
361.
Temple, J. (2002). Growth effects of education and social capital in the oecd countries. Histor-
ical Social Research/Historische Sozialforschung, 5–46.
Vinokur, A. D., Schul, Y., Vuori, J., & Price, R. H. (2000). Two years after a job loss: Long-
term impact of the jobs program on reemployment and mental health. Journal of occu-
pational health psychology, 5(1), 32.
Zhang, C., & Zhuang, L. (2011). The composition of human capital and economic growth: Ev-
idence from china using dynamic panel data analysis. China Economic Review, 22(1),
165–171.
Zhu, S., & Li, R. (2017). Economic complexity, human capital and economic growth: Empirical
research based on cross-country panel data. Applied Economics, 49(38), 3815–3828.

36
A Appendix A

Individual Behavior

Consider the overlapping-generation economy with the two generations live for two periods.
Each generation was born at t = 0, 1, 2, . . . . In the first period, the individuals have to work to
earn income wt , consumes c1 and saving their money for the future where they become old and
have no income. In the second period, the individual retires and consumes c2 . Each endowed
with physical capital k0 , human capital h0 . The individual preferences are represented in the
logarithmic form as follows:
Ut = ln(c1t ) + βln(c2t+1 ) (25)
0<β<1

The individual’s income is taxed by a uniform tax rate τ in order to finance the government
expenditures in infrastructure and education. The young and old generation at the time t faces
the budget constraints in each period:

c1t + s1t ≤ (1 − τ )wt ht (26)

c2t+1 ≤ [(1 − 1 − τ )rt+1 ]s1t (27)

c1t , c2t+1 ≥ 0
Where st is individuals’ savings.

The individuals’ income is respected to tax at the a uniform rate τ in the first period. In the
second period, the generation will retire and have no income, but they have an amount of
savings that they loaned to firms and get the interest rt+1 and also subjected to tax at the same
tax rate τ .

By substituting the value of c1t and c2t+1 in equation (26), (27) and maximize the utility of
representative young generation, yields:

max
1
ln[(1 − τ )wt ht − s1t ] + βln[1 + (1 − τ )rt+1 ]s1t
st

−1 1 + (1 − τ )rt+1
F OC = 1
+β
(1 − τ )wt ht − st [1 + (1 − τ )rt+1 ]s1t
−1 1
⇒ 1
+β 1 =0
(1 − τ )wt ht − st st
⇔ β(1 − τ )wt ht − st − βst = 0
⇔ β(1 = τ )wt ht = (1 + β)st
β(1 − τ )wt ht
⇒ st = (28)
1+β

37
The Producer Behavior

The production function will take the form of Cobb-Douglas function as follows:

Y = AGθt ktα (nt ht )1−α (29)

Each firm hires capital and effective labors. The firm has to pay rental rate qt and effective
wages wt for rental capital and human labors, respectively. A represents augmented technol-
ogy which exhibits constant return to scale over the private factor so that the profits are zero in
equilibrium. The private and public capital will be depreciated at the rates δk and δG , respec-
tively. The accumulation of private capital and public capital are described as follows:

kt+1 = ItP + (1 − δk )kt (30)

Gt+1 = ItG + (1 − δG )Gt (31)

Where ItP , ItG are private investment and public investment in the period t. The government’s
budget is assumed to be balanced.

Then, profit equation of a firm takes the form:

Π = AGθt ktα (nt ht )1−α − qt kt (32)

In an equilibrium, there are standard conditions are described by:

kt+1 = st , nt = 1

qt = rt + δk
ItG + Et = τ (wt ht + rt kt )

Given the initial k0 , h0 , and G0 , an equilibrium for the economy is the sequence of allocation
{nt , kt , ht , ct , ct+1 , yt }∞ ∞
t=0 , and the price {wt , rt , qt }t=0 . The agent’s savings equates to the cap-
ital of the period t + 1, the total rental cost of capital is equal to the real interest rate and the
rate of capital depreciation. The public investment ItG and expenditures Et are balanced to the
total revenue as an assumption.

Public Expenditures on Infrastructures

Profit maximization of firms respects to nt ht and kt , yields:

max Π = AGθt ktα (nt ht )1−α − wt nht − qt kt

nh,k

∂Π
= (1 − α)AGθt ktα = 0
∂nt ht

(nt ht is unity as assumed)

38
 ⇒ wt = (1 − α)AGθt ktα (33)
∂Π
= αAGθt ktα−1 − qt = 0
∂kt
⇒ qt = αAGθt ktα−1 (34)

Thus, in an equilibrium, the condition kt+1 = st must be satisfied. Substituting kt+1 and the
result of wt in (33) into equation (28), yields.

β(1 − τ )
kt+1 = (1 − α)AGθt ktα (35)
1+β

For simplicity, suppose that δG = 1 and δk = 0, then:

Gt+1 = τ AGθt ktα (36)

Hence:
Gt+1 τ (1 + β)
= (37)
kt+1 β(1 − α)(1 − τ )

Assuming that α + θ = 1 and focus the balanced growth path. Taking the logarithm of (35) to
see the growth path of private capital.

β(1 − τ )
lnkt+1 = ln[(1 − α)AGθt ktα ]
(1 + β)

βθ
= ln( ) + lnA + ln(1 − τ ) + (1 − θ)lnkt + θlnGt
1+β
βθ
lnkt+1 = ln( ) + lnA + ln(1 − τ ) + lnkt − θlnkt + θlnGt
1+β
βθ Gt
lnkt + 1 − lnkt = ln( ) + lnA + ln(1 − τ ) + θln( )
1+β kt

At the steady state Gt+1 = Gt and kt+1 = kt , hence:

βθ τ (1 + β)
lnkt+1 − lnkt = ln( ) + lnA + ln(1 − τ ) + θln( )
1+β βθ(1 − τ )

βθ
= ln( ) + lnA + ln(1 − τ ) + θlnτ + θln(1 + β) − θln(1 − τ ) − θln(βθ)
1+β
βθ βθ
= ln( ) + lnA + (1 − θ)ln(1 − τ ) + θlnτ − θln( )
1+β 1+β
βθ
⇒ ln(kt+1 ) − ln(kt ) = (1 − θ)ln( ) + lnA + (1 − θ)ln(1 − τ ) + θlnτ (38)
1+β

39
Public Expenditures on Education

Profit maximization by firms respects to nt ht and kt :

∂Π
= (1 − α)Aktα h−α
t − wt = 0
∂nt ht

nt is unity as assumed.
⇒ wt = (1 − α)Aktα h−α
t

⇒ wt ht = (1 − α)Aktα h1−α
t (39)
∂Π
= αAktα−1 hα−1
t − qt = 0
∂kt
⇒ Aktα−1 hα−1
t = qt
⇒ qt kt = Aktα hα−1
t (40)

Thus:
wt ht = (1 − α)Aktα ht1−α (41)
qt kt = αAktα h1−α
t (42)

In an equilibrium:
β(1 − τ )wt ht
kt+1 =
(1 + β)

Substituting wt ht into the equilibrium equation, yields:

β(1 − τ )(1 − α)Aktα ht1−α

kt+1 = (43)
1+β

Assuming δk = 0, the public expenditures on education is:

Et = τ Aktα h1−α
t (44)

The accumulation of human capital will become:

ht+1 = Bhµt (τ Aktα h1−α

t )1−µ (45)

Thus, the ratio between human capital and physical capital is according to:

ht+1 Bhµt (τ Aktα h1−α

t )1−µ (1 + β)
=
kt+1 β(1 − τ )(1 − α)Aktα ht1−α

ht+1 Bτ 1−µ hµt htαµ−µ kt−αµ A−µ (1 + β) B(1 + β)τ 1−µ hαµ
t
= =
kt+1 β(1 − τ )(1 − α) β(1 − τ )(1 − α)Aµ ktαµ
ht+1 B(1 + β)τ 1−µ ht
⇒ = µ
( )αµ (46)
kt+1 β(1 − τ )(1 − α)A kt

40
 In the steady state where kt+1 = kt and ht+1 = ht . Taking the logarithm of equation above and
find the steady state x∗ .

h B(1 + β)τ 1−µ h αµ

ln( ) = ln{ ( ) }
k β(1 − τ )(1 − α)Aµ k

h B(1 + β) h
ln( ) = ln µ
+ lnτ 1−µ − ln(1 − τ ) + ln( )αµ
k β(1 − α)A k
h h B(1 + β)
⇔ ln − αµln = ln + (1 − µ)lnτ − ln(1 − τ )
k k β(1 − α)Aµ
h B(1 + β)
⇔ (1 − αµ)ln = ln + (1 − µ)lnτ − ln(1 − τ )
k β(1 − α)Aµ
h 1 B(1 + β) 1−µ 1
⇒ ln = ln + lnτ − ln(1 − τ )
k 1 − αµ β(1 − α)Aµ 1 − αµ 1 − αµ

Clearly, the ratio converges monotonically to a unique steady state x⋆ given by:

1 B(1 + β) 1−µ 1
lnx∗ = ln µ
+ lnτ − ln(1 − τ )
1 − αµ β(1 − α)A 1 − αµ 1 − αµ

By taking the logarithm of equation A.19, the long-run growth rate is given by:

β(1 − α)A
lnkt+1 − lnkt = ln + ln(1 − τ ) + αlnkt − lnkt + (1 − α)lnht
1+β

β(1 − α)A ht
lnkt+1 − lnkt = ln + ln(1 − τ ) + (1 − α)ln (47)
1+β kt

Substituting lnx∗ into equation (47), yields:

β(1 − α)A 1 B(1 + β) 1−µ 1

γ = ln +ln(1−τ )+(1−α){ ln µ
+ lnτ − ln(1−τ )}
1+β 1 − αµ β(1 − α)A 1 − αµ 1 − αµ

Taking the first derivative of γ to find the maximization of the tax rate τ :
′ −1 1−µ 1
γ = + (1 − α)[ + ]=0
1−τ (1 − αµ)τ (1 − αµ)(1 − τ )

−1 (1 − µ)(1 − α) 1−α
⇒ + + =0
1−τ (1 − αµ)τ (1 − αµ)(1 − τ )
⇒ (αµ − 1)τ + (1 − µ)(1 − α)(1 − τ ) + (1 − α)τ = 0
αµτ − τ + (1 − µ)(1 − α) − τ + µτ + ατ − αµτ + τ − ατ = 0
⇒ (1 − µ)(1 − α) − (1 − µ)τ = 0
⇒τ =1−α

41

View publication stats