HOW DOES HUMAN CAPITAL AFFECT ECONOMIC GROWTH? pdf

The first part of this paper classifies the education into primarysecondary and tertiary ones to test whether each educational level affect economic growth differently, and attempt to un

NATIONAL CHENG KUNG UNIVERSITY

Department of Economics

Master’s Thesis

How Does Human Capital Affect Economic Growth?

Advisor: Prof Chun-Li Tsai Student: Ming-Cheng Hung

June 2009

Abstract

Based on the empirical investigations and theory of endogenous growth, this paper examines the role of human capital on economic growth from 118 countries over period from 1980 to 2006 The first part of this paper classifies the education into primarysecondary and tertiary ones to test whether each educational level affect economic growth differently, and attempt to uncover this different effect among different countries development Then, we focus on the composition of human capital that be categorized into agriculture(agriculture), industry(science, manufacture), service(art, humanity, health, society, service) types to find which types of human capital has the greatest impact to economic growth and add the factor of country’s industrial structure to test whether the growth effect of education depends on the coordination between country’s education fields and its’ industrial structure of local economics

Empirical result indicates three conclusions First, tertiary education has the greatest contribution to economic growth for all countries development In particular, the less-developed nations need more tertiary human capital to catch up with the well-developed ones Second, we find out that only the industry skill has positive contribution to economic growth Finally, as including the factor of development and industrial organization, the human capital of industry skill only in developing countries, and the nations with highly-profession guiding in industry has significant growth effect

Keyword: economic growth, human capital

Contents

1 Introduction 1 

2 Literatures Review 8 

2.1 The human capital to economic growth 8 

2.2 The other variables to economic growth 14 

2.2.1 Investment, Fertility rate and economic growth 14 

2.2.2 Government expenditures and economic growth 15 

2.2.3 Openness and economic growth 16 

2.2.4 Political structure and economic growth 16 

2.2.5 Inflation and economic growth 17 

3 Methodology and Economic model 19 

3.1Methodology 19 

3.2 Economic model 21 

3.3 Data sources and usages 26 

3.4 Descriptive Statistics 28 

4 Estimation Results 31 

4.1 The human capital to economic growth 31 

4.1.1 the estimation result with whole data 32 

4.1.2 the estimation result in different country development 37 

4.2 The human capital allocation to economic growth 46 

4.2.1 the estimation result with whole data 47 

4.2.2 the estimation result in different country characteristics 48 

5 Conclusion 55 

6 Reference 57

7 Appendix 62

List of Tables

Table 2-1: the previous literatures about the growth effect of human capital .13 

Table 3-1: the predicted coefficient in control variables 26 

Table 3-2: the data resource 28 

Table 3-3: the descriptive statistics full sample 29 

Table 3-4: the descriptive statistics categorize by development 30 

Table 4-1: the effect of human capital on economic growth 34 

Table 4-2: the effect of human capital on economic growth full sample 35 

Table 4-3: the effect of human capital on economic growth well-developed nations 36 

Table 4-4: the effect of human capital on economic growth developing nations 39 

Table 4-5: the effect of human capital on economic growth less-developed nations 41 

Table 4-6: the effect of human caital on economic growth check robust 43 

Table 4-7: the effect of human capital on economic growth include dummy variable 44 

Table 4-8: the effect of human capital on economic growth different development .45 

Table 4-9: the growth effect of human capital composition 51 

Table 4-10: the growth effect of human capital composition different development 52 

Table 4-11: the growth effect of human capital composition different industry organization(1) 53

Table 4-12: the growth effect of human capital composition different industry organization(2) 54 

List of Figures

Figure 1-1: primary, secondary and tertiary school enrollment rate (1980-2006) 1 

Figure 1-2: primary school enrollment rate (1980-2006) 4 

Figure 1-3: secondary school enrollment rate (1980-2006) 4 

Figure 1-4: tertiary school enrollment rate (1980-2006) 4

Figure 1-5: the framework of this study 7

Figure 4-1: cross-sectional relationship between primary education and growth (2000) 31

Figure 4-2: cross-sectional relationship between secondary education and growth (2000) 32 Figure 4-3: cross-sectional relationship between tertiary education and growth (2000) 32

Figure 4-4: cross-sectional relationship between agriculture skill and growth (2003) 46

Figure 4-5: cross-sectional relationship between industry skill and growth (2003) 47

Figure 4-6: cross-sectional relationship between service skill and growth (2003) 47

0 20 40 60 80 100 120

1 Introduction

High level of education is commonly seen as one of the major prerequisite of the world’s current wealth, and as one of the major determinants of future economic growth and development In Fig.1-1, we present the world average enrollment rates1 of primary secondary and tertiary types that recognized as the different level of human capital2, we find the enrollment rate increase continually from 1980 to 2006 It indicates that the education becomes the more and more important resource and government policy all over the world recently Education also has numerous impacts on individual and social development According to Behrman, J & Stacey, N(1997), the higher education will not only contribute to economic growth but also has lots non-market effects and social benefits such as improving people’s health, reducing mortality, decreasing crime activity and mitigating wealth inequality

The gross school enrollment rate

Regarding to the relationship between human capital and economic growth, Nelson and Lucas(1966) recognize average educational attainment as an important factor to technology, and they find the worker with high level of education has tended to adopt productive innovations easily Then according to the “new growth theory” such as Lucass(1988) and Romer(1991), they added the variable of human capital into the growth model and provided new ways of conceptualizing how human capital might contribute to self-sustaining growth of per capita incomes In term of previous literatures, the education can increase the literacy rate and the labor’s productivity that can improve the efficiency of using technology In further, the higher level of education like university produces a lot of researches that be an important resource of new idea and advances in knowledge In addition, the education also has other channels to economic growth First, a spillover effect that raises not only the productivity of those of receiving education but also the productivity of those the work and interact with Second, Katarina & Keller(2007) demonstrate the indirect effect that education can reduce fertility rate and income inequality, those can foster the economic growth deeply

Empirically, many studies(Stephan,1997; Chatterji1998; Kwabena,2006) use different indexes3 of human capital and find the positive connection between higher education and economic growth However, the country’s development4 is also an important factor to the growth effect of human capital According to Fig.1-2,3,4 that present the enrollment rate

in different development countries from 1980 to 2006, we find out the well-developed nations have the highest enrollment rate in the primary, secondary and tertiary education

.However, the primary enrollment rate growths sharply in less-developed nations and catches up with the other countries’ level, but the tertiary education’s gap between well- and less-developed nations becomes large gradually Some literatures (Ruth Judson1998, Petrakis,2002 ; Katarina & Keller,2007) discuss the relationship between education, development and economic growth They demonstrate the empirical work that the education and development have the positive relationship and also suggest that the role of primary and secondary education seem to be more important in LDC nations, while growth

in well-developed economies depend mainly on tertiary education On the contrary, some literatures (Chatterji,1998 ; Kwabena,2006) demonstrate the different outcome that LDC nations need more tertiary education in order to foster the economic growth and catch up with the other well-developed countries Although there are no obvious theoretical issue about the link between education levels and different developments, those studies indicate that we can’t neglect the factor of country development as discussing the growth effect of human capital In contrast with previous literatures5 that always use the average-period data to discuss the issue about human capital and economic growth because of the missing value and compounding the measurement error in the data by emphasizing errors related to the timing of relationships, this paper uses unbalance panel methodology to solve the problem of missing value and include the year-fixed effect to eliminate the time shock

In recently, some literatures address some new idea about the education and economic growth Richard H Mattoon(2006) finds the relationship between education and growth

5

Barro (1998) consider the average period data, 1965-75, 1975-85, and 1985-95 in order to eliminate the missing value and accords with the growth theory, which do not attempt to explain short-run business fluctuations Therefore, the application of the theories to annual or other high-frequency observations would compound the measurement error in the data by emphasizing errors related to the timing of relationships

Figure 1-2: primary school enrollment rate (1980-2006)

Figure 1-3: secondary school enrollment rate (1980-2006)

Figure 1-4: tertiary school enrollment rate (1980-2006)

primary school enrollment rate

well-developed less-developed developing

0 20

secondary school enrollment rate

0 10

tertiary school enrollment rate

%

%

%

is not clear and addresses the conception that the best growth effect of education depend on what colleges and universities have to offer and what is happening to the local industrial structures of their economies Kitagawa F.(2004) also demonstrates the important connection between university and local industry

According to the above-mentioned, they address two fundamental elements influencing the growth effect of human capital: human capital allocation and countries’ local economics With regard to the composition of human capital, the similar and

empirical literatures studied by Murphy et al(1991) and Tiago(2007) showing some

evidence that the students majoring in engineer in college have contribution to growth greater than the students that major in another fields, Colombo and Grilli(2005) also find the human capital from scientific and technical fields have significant positive effect to firm’s growth However, the previous studies only put emphasis on the composition of human capital but they don’t include the country’s development and local industry dummy into their growth model to examine this issue The concerning literature studied by Tiago(2003), which finds the nonlinear relationship between GDP per capita and S&E skill, and concludes that the different country developments need different types of human capital to support their local economics Therefore our study considers the education allocation and country’s local economics simultaneously in order to demonstrate whether the relationship between education and economic growth depends on the field of human capital and country’s characteristics

In conclusion, our paper has two issues about the education and economic growth First, we use panel data including 118 countries from 1980-2006 Differently to other literatures, we use individual-period data not the average-period data to reexamine this

issue Second, although previous studies have demonstrated the relationship between skill allocation and growth, they don’t find this effect in different type of industry organization and development Therefore, we try to prove the conception of Richard H Mattoon(2006) and add the factor of industry organization and development to find whether the growth effect of tertiary education depends on the country’s different characteristics

The paper is structured as follow (Fig.1-5) Section I explain the motivation and purpose of this study and address the contribution that different from other literatures Section II reviews the previous studies discussing the relationship between human capital and economic growth Section III then build the model and introduce our econometric method first Second, we explain our data selection and describe the definition and the descriptive statistics of the important variables Third, we present the outcome of the growth effect of the human capital and compare each contribution among different developed countries Finally, this paper discusses the issue about the allocation of human capital Some conclusion and future work are drawn in section IV

Figure1-5: the framework of this study

Motivation and purpose

Literatures review

Issue (I):

The relationship between human

capital and economic growth

Issue (II):

The relationship between allocation

of human capital and economic growth

Classify the data by different

development

Classify the data by different industrial organization and development

Conclusion

2 Literatures Review

2.1The human capital to economic growth

Much studies demonstrate theoretically and empirically education’s importance to economic growth through diverse mechanisms Nelson and Lucas(1966) recognize average educational attainment as an investment to human capital, and address the conception that the man with higher education adapts the technology more effectively than the others Then basing on the “new growth theory ” such as Lucass(1988) and Romer(1991), they added the variable of education into the growth model and place the emphasis on the characteristic of non-decreasing returns to scale in human capital that explain the self-sustaining growth of per capita incomes Mankiw and Romer(1992) augment the solow model by including accumulation of human capital as well as physical capital, they seem the enrollment rate of secondary level as the stock of human capital and find the education not only has positive effect to growth but also improves the power of explanation in the solow growth model Norman(1995) also finds the same conclusion ,but adds the additional variable of change quantity in education that can provide useful additional dynamic information on the contribution of human capital to economic growth

In empirically, lots papers want to prove the relationship between human capital and economic growth Although they discuss this issue by different indexes of education, they indicate the similar conclusion that investment in education is a key factor in the rapid growth rates Barro(1991) makes use of the primary and secondary enrollment rate in1960

to confirm the positive growth effect of human capital, and find the secondary education has the largest effect to economic growth Therefore, much coming literatures recognized the secondary education as the variable of human capital In recently, Chatterji(1998)

reexamines this issue by the variable of secondary and tertiary enrollment rate from 1960

to 1985, and demonstrates the tertiary education does displace secondary education as the major driver of growth Stephan(1997) also finds similar conception that tertiary education has the greatest economic impact on output In conclusion, the higher level of education plays more and more important role to describe the accumulation of human capital recently

Closely related papers in the literature to this research are studies by Stephan(1997),

Ruth Judson(1998) , Petrakis(2002) Katarina & Keller(2007) They further examine this relationship is ambiguous by different level of education and in different country developments They mostly argue that the low income countries rely heavily on primary education and moderately on secondary education while higher education seems to be more profitable in wealthy countries On the contrary, Chatterji(1998) uses the enrollment rate as the accumulation of human capital and find that tertiary education is important for countries that attempt to `catch up’ with the well-developed countries, but less important for the world leaders since their tertiary enrolment rates are already near the maximum The similar paper studied by Kwabena(2006) adopts 34 African countries over the 1960–2000 period and measure the education variables as the average number of years of education attained by the adult population In contrast with other literature6 finding no significant effect, they demonstrate that the higher tertiary education has positive and statistically significant effects on the growth rate as well as the

6

Appiah and McMahon (2002)find that the sub-Saharan African countries invest more in education but allocate large percentages of education budgets toward higher education ‘‘where the returns are the lowest, emigration the largest, and tuitions remain highly subsidized.’

education of primary and secondary level Although there are no obvious theoretical issue about the link between education level and different developments, those studies indicate that we can’t neglect the factor of country development as discussing the growth effect of human capital Because of missing value of some education variable, they all use the average-period data to solve this problem (see Table2-1) However, this method can eliminate some information from data and they overlook the year fixed effect as discussing the relationship between human capital and economic growth Therefore, we adopt individual period data and add the factor of year and country fixed effect to reexamine this topic

Moreover, some studies emphasize the issue of human capital allocation7 Base on the endogenous growth theory, they indicate that not all kind of education fields can drive the technology and economic growth Therefore, the effect of human capital composition

on growth and development is a more recent field in the economic literatures.The first paper in this issue was the one from Murphy et al.(1991), which argues the rent seeker8such as lawyer rewarding talent more than the producer and innovator, and show the result that the engineering students in college have positive effect to economic growth but the law students does not The coming literature studied by Tiago(2007) findnig the positive relationship between high-tech skill and economic growth by adding the factor of the composition of human capital into growth model, and also demonstrate this effect empirically in OECD countries There are also some relating literatures addressing the importance of engineering employment to firm’s growth Almus and Nerlinger(1999) on

newly established West German firms They provide evidences that firms started by the skill in science and engineering fields rewards greater growth rates in high-, medium-, and low-tech industries On the contrary, business education has a significantly positive impact only for low-tech firms The similar literature is studied by Colombo and Grilli(2005) They examine the role of specific human capital in scientific and technical fields vice versa other education fields of entrepreneurs for growth of new firms in high-tech sectors They find that the average years of founders’ education has no significant effect whereas education in S&E9 field has large positive effects

According to the above-mentioned, they only demonstrate which field of human capital has contribution to economic growth but don’t consider the factor of country’s industry structure Richard H Mattoon(2006) reviews some previous literatures and brings

up the conception that the tertiary education is most successful in influencing economic growth as they are attuned to the economic structure of their local economies In others words, the degree of education’s growth effect depends on whether the field of human capital from universities can offer local industry new technology and innovation or upgrade the existing industry Kitagawa F.(2004) also sheds light on the collaborative mechanics between university and local industry In contrast with other literatures, it seems fruitful to investigate the use of different knowledge bases within particular industrial and learning not only looking at high-tech industry and high-skilled labor but also other sectors

of activities economy However, those two literatures only address the simple conception about the connection between human capital and their industrial characteristics, but they

9

the UNESCO classify the tertiary into the fields of education, humanities and art, social science, science, engineering, agriculture, health, and service S&E skill means the sum of graduate in science and engineering as a share of total graduate in tertiary education

don’t prove this result empirically Therefore, we add the dummy variables that represent the different country industrial organization and development to examine whether the country’s characteristics can impact the relationship between human capital and economic growth

Table 2-1: the previous literatures about the the growth effect of human capital

author period country econometric data profile the variable of human capital outcome

Barro(1991) 1960~1984 98 OLS(cross-section) one period data primary secondary

enrolloment rate

+

Mankiw & Romer1992 1960~1985 75 OLS(cross-section) average period secondary enrolloment rate + Norman (1995) 1960~1985 75 OLS(cross-section) change in enrollment rate secondary, tertiary enrollment + Chatterji(1998) 1960~1985 98 OLS(cross-section) change in enrollment rate secondary, tertiary enrollment + Stephan(1997) 1985 77 OLS(cross-section) one period data average years of school in primary

secondary and tertiary

+

Pritchett (1997) 1965~1984 77 OLS(cross-section) change in average years

of school

average years of school Uncertan

Temple(1999) 1965~1985 64 LTS(cross-section) change in average years

of school

average years of school +

Kwabena et al(2006) 1960~2000 34 African system GMM average period average years of school +

Katarina(2007) 1970~2000 75 panel, country fixed

effect

average period enrollment rate ,public education

expenditures

Uncertan

this paper 1980~2006 118 panel,system GMM individual data enrolloment rate

2.2 The other variables to economic growth

In order to reduce the mean square error10 and the bias of estimators, we add other control variables that can explain the economic growth The most variables that we obtain are similar to Barro(1991,1998) such as real GDP per capita fertility rate the growth of consumer price index government expenditure invest in physical capital openness and political right In addition, adding those control variables can test the robustness11 of the relation between education and economic growth

2.2.1 Investment, Fertility rate and economic growth

According to the model, which proposed theoretically by Robert Solow in1956, the

investment and population growth are the important factors to impact the economic growth Empirically finding studied by Mankiw(1992), Long and Summers(1991, 1992) improve this connection between investment and economic growth According to some previous literatures, there are two mechanics through which fertility, as measured by birth rates, might affect per capita economic growth.The first effectis through changes in the share of the population of working age When the fertility rate decline, the population of working age and the effective labor input would rise, and further foster the economic growth The second effect is through the per capita investment rate The fertility involves an increase in the value of parent’s time and thereby a rise in the cost of raising children and tend to reduce the investment to physical capital

But some literatures argue that population may have a scale effect that is beneficial to economic growth Kemer(1993) demonstrate that technological progress is an increasing function to population size The reasoning is simple: the larger the population, the more people there are to make discoveries, and the more rapidly knowledge accumulation So he recognized the growth of population as an important element to economic growth Boucekkine(2002) also find that the long-term relationship between fertility and per-capita growth is hump-shaped When the fertility rate is low below the threshold, increasing in fertility would raise the effective labor force, and thus foster the economic growth On the other hand, when the fertility rate is high above the threshold, increasing in fertility would rise the parent’s cost of raising children, and thus reduce the investment and labor input to the economics In conclusion, the relation between fertility rate and economic growth is ambiguous But according to empirical literatures, the effect of fertility mostly have negative to per-capita growth Brander, James and Steve Dowrick(1993) use a

107 country panel data set covering 1960-85, and find that high birth rates appear to reduce economic growth Li Hongbin and Junsen Zhang(2007) use a panel data set of 28 provinces in China over twenty years by the method of GMM estimation and also find the same result

2.2.2 Government expenditure and economic growth

Barro(1991) use a 98 country data set covering 1970-85 and find the ratio of real government consumption expenditure to real GDP has a negative association with economic growth The argument is that the government expenditures cause the increasing

in tax and thus reduces the saving rate and growth. Landau Daniel(1983) also demonstrates the same conclusion But the growth effect of government consumption is not always negative, when we classify the expenditure into several types According to Peter Nijkampa, Jacques

Pootb (2004), they arrange the sample of 93 published studies and sort out the policy into five categories: general government consumption, tax rates, education expenditures, defense, and public infrastructure and find the evidence for a positive effect of conventional fiscal policy on growth is rather weak, but the commonly identified importance of education and infrastructure is confirmed In addition, the relationship between government expenditures and growth is different in different level of development country. Landau Daniel(1983) finds the government consumption cause negative effect in developed country and positive but insignificant effect in less developed country

2.2.3Openness and economic growth

New growth theory (Romer,1991) has provided important insights into an understanding of the positive relationship between trade and growth For example, if growth is driven by R&D activities, the trade activity is taken for the incentive and access

to transit technological knowledge between country and country Further trade allows company to have more resources to expand its scale and investment of R&D According to this issue, Harrison(1993), Yanikkaya, H(2003), Chen, P and Gupta, R (2006) find the positive relationship between openness and economic growth empirically In general, the growth effect of openness in less-developed country is more positive and significant than the country that is well-developed The reason is that developing countries obtain the idea and intermediate input that is relatively new and productive to their economic and production, so when the trade is more frequent and unobstructed, the resource of knowledge accumulation from well-developed countries is more quickly and easily

2.2.4 Political structure and economic growth

Many literatures discuss the relationship between democracy and economic growth,

but the direction of effect is unclear12 According to previous studies, two mechanics that affect this relationship are opposite First, democracy may cause distortion and inefficiency

to society and economic environment: Chong, Alberto(2002) demonstrates the effect of democracy that exacerbates the social inequality in the poor and highly unequal countries Moreover, political and civil freedom makes it harder for government to take tough but necessary decisions (World Bank, 1991) On the other hand, authoritarian regimes are able

to implement the kinds of policies that are necessary for rapid economic growth Second, the region of democracy has some indirect effect to economic growth positively For example, Gerring, John (2004) regards democracy as an important institutional factor in the development of human capital, as measured by declining fertility rates and improvements in education, public health, and life expectancy Hellwell(1994) also finds the positive relationship between democracy and investment that offset the negative effect

of democracy on subsequent economic growth In conclusion, the democracy that depends

on the two mechanics is ambiguous to the GDP per capita growth

2.2.5 Inflation and economic growth

In many studies, inflation can cause lots of social cost and negative effects to economic growth The first cost of inflation that can be easily identified is that it distorts the tax system and further reduces people’s wealth and saving that weaken the incentive of investment (Barro,1995) Second, higher inflation induces more inefficiency in market transaction because of price variability that can disturb the long-term relationship among business transaction Many literatures(Barro,1995; Bruno Michael,1998; Alexander 1997)

12

Leblang(1997) Gerring John(2005) find the positive effect and John F hellwell(1994) Barro(1995) find the negative effect between democracy and economic growth

find the negative effect between growth and inflation However, Burdekin(2000) offers further evidence that the effects of inflation on growth are negative, nonlinear and varying

by the country type, and demonstrate the cost of inflation can be much higher for industrial than for developing countries

3 Methodology and Economic model

3.1Methodology

The previous and frequent estimation methodology that used to analyze the issue of economic growth is the cross-section13 While, simple ordinary least square (OLS) was applied to a growth model, in which growth in the whole period or the log of per capita income at the one period year is used as the dependent variable This method is advantageous since it allows one to estimate coefficients on time invariant variables like discussing the issue among lots countries or areas Although the methodology of cross-section has an advantageous position in collecting data, the problems associated with cross-section estimations are often serious The country-specific aspect of economic growth that is ignored in a single cross-section regression may be correlated with independent variables, causing an omitted variable bias In addition, one condition needed

to produce consistent estimates in a cross-section framework is that explanatory variables should be strictly exogenous

Therefore, we use the panel data14 including the formation of cross-section and time series and add the country fixed effect that capture the stable difference between countries and year fixed effect that capture the influence of shocks affecting the growth rate in multiple countries at the same time Example might include major technology, the innovation of computer and internet Using panel data has lots advantages that superior to

the methodology of cross-section First, it could control the unobserved country and year characteristic by considering the fixed effect Second, the panel data offer more information, less collinearity, more efficiency and more observation

Although panel fixed effect methodology has improve the drawback of cross-section data, many variables used in growth regressions are potentially endogenous to the dependent variable and not strictly exogenous to the control variables It can cause the correlation between explanatory variables and error tem, and induce the problem of endogeneity and estimation bias Although one can use an instrumental method15 to solve this problem, it is hard to find the absolutely exogenous instrument that can be associated only with the explanatory variable, and not with the error terms The dynamic panel estimators are provided by Arellano-Bond(1991) and Arellano-Bover (1995)/Blundell-Bond(1998) in order to solve the problem of endogeneity In addition, the GMM methodology is designed for the situations with few time periods and large countries, the dynamic dependent variable that depend on its own past term, the independent variables that are not strictly exogenous, fixed individual effect In order to reform the small sample bias that characterizes the first-differenced GMM, Arellano-Bover (1995)/Blundell-Bond (1998) address the system GMM methodology In addition, we use two criteria, namely the over-identification test and the test for second order serial correlation of the residuals in the differenced equation (AR2) to evaluate our model that is correctly specified The overidentification test is the sargan/Hansen test that tests whether the instrument variables have the correlation with error terms If the p-value is large that

15

The previous estimation methodology that solve the problem of endogeneity is 2SLS which distinguishes between

regressors and instruments while allowing the two categories to overlap The main drawback of using 2SLS is that when the instrument variables are more than the explanatory ones, the endogenous problem can not be solved

can’t reject the null hypothesis of no relationship between instruments and error terms, our instruments are exogenous and correctly specified by the overidentification test The Arellano-Bond tests whether the differenced residuals have the second-order serial correlation If the null hypothesis of no autocorrelation is not rejected, we pass the test of AR(2)

In this paper, we use the fixed-effect panel as well as system GMM method to correct for unobserved country heterogeneity, omitted variable bias, measurement error, and potential endogeneity In order to mitigate the problem of few time periods in graduated fields’ data, we only use the system GMM method to discuss the relationship between composition of human capital and economic growth In addition, we also use sargan test and AR(2) to confirm our model that is correctly specified

3.2 Economic model

First, we use panel data from 118 countries over period from 1980 to 2006 and examine the relationship between human capital and economic growth The dependent variable is real GDP per capita growth rates Enrollment rate each for primary secondarytertiary are recognized as the human capital and lagged by one period16 Control variable added are similar to Barro(1991,1998,2000) and Katarina & Keller(2007) According to

this literature, the variable such as log of GDP per capital(Y) log of fertility rate(lnf) the

growth of consumer price index(cpi) government non-education expenditure(G/Y)

16

Sylwester(2000) observes public education expenditures concurrent to growth being negative, short lags insignificantly positive, and longer lags significantly positive He suggests the negative short-term effects are due to large inflows of people substituting public education for work, distortionary taxation, and delayed benefits

investment in physical capital(I/Y) openness(T/Y) and political right(PR) are the important

factors to economic growth First, we discuss the relationship between three levels of human capital and economic growth in model (I) The equation 3.1 represents the basic growth model including different level of education and GDP per capita, which considers the factor of conditional convergence The equation 3.2 adds other control variables in order to check robust and reduce the mean square error

Model(

(3.1) (3.2)

, , : primary, secondary and tertiary gross enrollment rate in period

t -1 for each country, the gross enrollment rate expressed as a percentage of the population in the theoretical age group for the same level of education

, , : tertiary graduates as percentage of all graduates in agriculture,

industry and service17

17

The categorization is made according to the following specification: the graduate from agriculture field is recognized

as the human capital of “agriculture skill”, the “industry skill” are defined by Tiago(2007) who adds the science and engineering person, which is considered as high-tech human capital, the “service “one: we include other fields with humanity, art, service, health care, social science, and education

: the natural log of real GDP per capita(based on 2000 year US dollar) in period t -1 for each country

: investment in physical capital as share of GDP in period t for each country

: the natural log of fertility rate in period t for each country

: the openness (T=export + import of goods and services) variable in periodt for each country

: the government non-education18 expenditures in period t for each country

: the growth rate of the consumer price index(base year:2000) in period t for

each country

: political right(Political Rights are measured on a one-to-seven scale, with one

representing the highest degree of Freedom and seven the lowest.) in period t for

each country

: country fixed effect

: year fixed effect

: the error term in period t for each country

According to the previous papers that we review in chapter 2, the relationship between economic growth and some control variables are predicted in Table 3-1 We find the gross enrollment rate, investment in physical capital, openness and the human capital with industry skill have positive impact to economic growth On the contrary, natural log

of GDP per capita, fertility rate, government non-education expenditures and inflation have negative contribution However, the political right and the human capital with the non-industry skill are ambiguous because of diversified mechanics and lacking the

discussion in previous literatures

Table3-1: the predicted coefficient in control variables

3.3 Data sources and usages

The paper uses the usual sources for most of the data used in the regression analysis Details on the data sources are provided in Table 3-2 Most of the data for real GDP per capita, fertility rate, trade, investment, and so on are from the World Development Indicators (WDI) 2008 The variable of political right that represents the degree of democratic freedom is provided by freedom house For the data on education variable, we relied on the statistics from United Nations Educational, Scientific and Cultural Organization (UNESCO), and adopt the gross enrollment rate as the measurement of human capital in accordance with previous studies19

19

Barro(1991) Mankiw & Romer(1992), Norman (1995), Chatterji(1998), Barro(2000), Petrakis(2002),

Katarina(2007), Keun Lee et Al(2008)

variable The expected effect to economic growth

the tertiary human capital with industry skill +

The tertiary human capital with service skill +/-

Although there are many measurements in human capital such as education expenditures, average years of school and literacy rate20, the enrollment rate only has complete data in individual period and with large spans of years

The data that we adopt includes 118 countries and over period from 1980 to 2006 Although the variable of enrollment rate begin in 1970 from UNESCO, we exclude the period before 1980 because of measuring in five-year periods and lacking observations, and due to some countries with many missing value, this paper drop the country from 170 to118 including 25 advanced, 74 developing and 19 less-developed countries that be classified by International Monetary Fund (IMF) Concerning the variable of composition

of human capital that we discuss in model III, some previous studies21 use the ratio of the enrollment in industry to total enrollment person in tertiary education as the human capital with industry skill However, this data in individual period has large missing value, and must be calculated by author because of no direct data in UNESCO Thus, this paper uses the variable of tertiary graduates as percentage of all graduates in agriculture, industry and service to proxy the different skill of human capital This variable are provided from UNESCO while considering the restriction from UNESCO data base and the amount of missing value, we only adopt the data including 73 countries and over period from 1999 to