(Luận văn) external and private return to education by using instrument variables approach, evidence in vietnam with a panel data

t to UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM ng hi HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE NETHERLANDS ep w n lo ad y th VIETNAM – THE NETHERLANDS ju PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS yi pl n ua al n va ll fu m oi EXTERNAL AND PRIVATE RETURN TO EDUCATION BY USING OF INSTRUMENT VARIABLE APPROACH: EVIDENCE IN VIETNAM WITH A PANEL DATA SET at nh z z k jm ht vb om l.c MR LE THANH HUNG gm BY an Lu n va ey t re MASTER OF ARTS IN DEVELOPMENT ECONOMICS th HO CHI MINH CITY, October 2016 t to UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES ng HO CHI MINH CITY THE HAGUE hi ep VIETNAM THE NETHERLANDS w n lo ad VIETNAM - NETHERLANDS ju y th PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS yi pl n ua al n va EXTERNAL AND PRIVATE RETURN TO EDUCATION BY USING OF INSTRUMENT VARIABLE APPROACH: EVIDENCE IN VIETNAM WITH A PANEL DATA SET ll fu oi m at nh z z ht vb A thesis submitted in partial fulfilment of the requirements for the degree of k jm MASTER OF ARTS IN DEVELOPMENT ECONOMICS om MR LE THANH HUNG l.c gm By an Lu n ey t re PROF NGUYEN TRONG HOAI va Academic Supervisor: th HO CHI MINH CITY, October 2016 t to ACKNOWLEDGEMENT ng hi ep To be able to finish this thesis, I have received the great supports from many people w Firstly, I would like to express my appreciation and special thanks to Prof Nguyen Trong n lo Hoai, my academic supervisor, who has given me many valuable guidance, advices, and ad great encouragements for my thesis Secondly, I would like to express my gratitude to y th ju Lecturers and Staff from Vietnam – Netherlands Program at University of Economics Ho yi Chi Minh city Specially, I am indebted to Ph.D Truong Dang Thuy, who gave me pl ua al valuable support and comments for my thesis I am also grateful to Ph.D Pham Thi Bich Ngoc for her support in Stata’s commands in my thesis Finally, I am indebted to my n n va family and my friends, who gave me the greatest encouragements for my study ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re th Page |i t to ng hi DECLARATION ep w n I declare that “External and private return to Education by using of Instrument lo ad Variable Approach: Evidence in Vietnam with a panel data set.” is my own work y th This thesis is has not been submitted to any degree or examinations at any other ju universities In addition, all the using sources are indicated by the completed references yi pl n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re th P a g e | ii t to ng hi ABSTRACT ep w n The combination of Vietnam Household Living Standard Survey and Provincial Statistics lo ad Yearbook from 2010 to 2014 provide a great opportunity for estimating the up to date y th external and private return to education in Vietnam In this paper, the Human Capital ju Earning Function (Mincer, 1974) and instrument variables are adopted in order to yi pl estimate the external and private return to education The analysis suggests that not only al ua one additional schooling year have an impact on individual wage, but the increase in n proportion of skilled workers in the labor force also have an influence on the hourly wage n va of individual ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re th P a g e | iii t to ng hi TABLE OF CONTENTS ep ACKNOWLEDGEMENT i w n DECLARE ii lo ad ABSTRACT .iii y th ju TABLE OF CONTENTS iv yi pl LIST OF TABLES vi al n ua LIST OF FIGURES vii va LIST OF ACRONYMS viii n fu ll Chapter 1: Introduction oi m 1.1 Introduction at nh 1.2 Research objectives z z ht vb 1.3 Research scope k jm 1.4 Structure of paper l.c gm Chapter 2: Literature review 2.1 Human capital theory om an Lu 2.2 Returns to education 2.3 External return to education 10 ey t re Chapter 3: Research methodology 14 n va 2.4 Chapter remark 12 th 3.1 Research methods 14 P a g e | iv t to 3.2 Endogeneity in Wage function 15 ng hi 3.3 Data sources and measurement 16 ep 3.4 Instrument variables 18 w n 3.5 Additional provincial control variables and interact terms 20 lo ad 3.6 Data description 21 y th ju 3.7 Chapter remark 22 yi pl Chapter 4: Research results 24 al n ua 4.1 Overview about Vietnam 24 va 4.2 Individual earning on individual characteristics 27 n fu ll 4.3 Individual earning on levels of education 28 oi m 4.4 Results on estimating returns to education 29 at nh 4.5 Returns to education classified by dummy variables 36 z z ht vb 4.6 Chapter remark 41 k jm Chapter 5: Main findings and recommendations 43 l.c gm 5.1 Main findings 43 5.2 Policy recommendations 45 om an Lu 5.3 Limitations 46 REFERENCE 47 n va ey t re APPENDIX 50 th Page |v t to ng hi LIST OF TABLES ep Table 3.3 Definitions and unit of individual-level variables 17 w n Table 3.4 Definitions and unit of provincial level variables 18 lo ad Table 3.5 Descriptive statistics for continuous variables .20 y th ju Table 3.6 Descriptive statistics for dummy variables 21 yi pl Table 4.2.1 Individual earning classified by gender 27 al n ua Table 4.2.2 Individual earning classified by marital status 28 va Table 4.2.3 Individual earning classified by type of school 28 n fu ll Table 4.3 Individual earning on levels of education 29 oi m Table 4.4.1 OLS estimate 29 at nh Table 4.4.2 Instrument variables estimate without control variables 31 z z ht vb Table 4.4.3a IV estimate with additional control variables 33 k jm Table 4.4.3b External return to education for levels of education 34 l.c gm Table 4.5 IV fixed effect with additional control variables for groups of individual’s characteristics 37 om Table 4.5a External return to education for female 39 an Lu Table 4.5b External return to education for Married = .40 n va Table 4.5c External return to education for public school 41 ey t re th P a g e | vi t to ng hi LIST OF FIGURES ep Figure 2.4: Analytical framework 13 w n Figure 4.1a: GDP across six key economic regions in Vietnam 2010-2014 25 lo ad Figure 4.1b: Educational system in Vietnam 26 ju y th yi pl n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re th P a g e | vii t to ng hi LIST OF ACRONYMS ep w Gross domestic product n GDP lo Provincial Statistic Yearbook ju yi Human Capital Earning Function pl HCEF Vietnam Household Living Standard Survey y th PSY ad VHLSS n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re th P a g e | viii t to ng hi Table 4.5a External return to education for female ep Schooling year External return (%) w n 12 16 18 22 13.7 9.2 5.6 2.9 -0.7 -2.5 -6.1 lo ad For male: ju y th - yi pl The external return to education is not significant, while private return to education is al ua positive significant at p-value < 0.1 The SchYear coefficient is equal 0.163 – n 0.007*Educated + 0.0468*HighGDP The man’s private return to education become va n negative when the proportion of highly educational workers is higher than 23.3% when fu ll this man works in the province/city having GDP lower than average In province/city m oi having GDP higher than average, the private return to education of male become negative nh when the share of skilled worker in this province/city is higher than 30% at z z 4.5.2 By Married (column 1,2 in table 4.5): vb jm ht Interestingly, while the private and external return to education is positive significantly for individual living with husband/wife in period of survey, both two types of return to k l.c gm education are not significant for the remaining part For people living with husband or wife, the Educated coefficient, which is considered as om the external return to education, is equal 0.0947 – 0.008*SchYear, while the private an Lu return to education is equal 0.189 – 0.008*Educated + 0.0413*HighGDP Table 4.5b individual, who is living with husband or wife and not go to school, have an increase Page | 39 th graduating from primary and lower secondary schools, respectively ey in the labor force The increase in hourly wage is at 5.47%, 2.27% for individual t re in hourly wage at 9.47% when there is one percent increase in the share skilled workers n va helps to express the external return to education for each level of education In details, t to For individual with dummy Married variable equal 1, when there is one percent increase ng in the share of skilled workers in the local labor force, the hourly wage of individual, hi ep graduating from upper secondary, university and having master and doctor certification, decrease at 0.13%, 3.33%, 4.93% and 8.13% respectively w n lo Table 4.5b External return to education for Married = ad 12 16 18 22 5.47 2.27 -0.13 -3.33 -4.93 -8.13 ju y th 9.47 yi Schooling year External return (%) pl ua al n About private return to education, the rate is equal 0.2303 – 0.008*Educated for n va individuals living in provinces/cities with GDP higher than average and 0.189 – ll fu 0.008*Educated for individual living in province/cites with GDP lower than average oi m When married individual living in the province/city with high GDP, the hourly wage will at nh decrease when the share of skilled workers in his working province/city is higher than 28.8% In addition, the share of skilled workers is higher than 23.6% will make the z z private return to education of individual living in provinces/cities having GDP lower than k jm ht vb average l.c gm 4.5.3 By SchType: om Due to the domination role of public schools in Vietnam, the observations in the whole an Lu sample, who study in public schools, are accounted for 93.25% basing on data description for dummy variables in chapter This may be the results of significantly positive n ey t re The Educated coefficient is equal 0.0927 – 0.009*SchYear, table 4.5c shows va external and private return to education for only individuals studying in public schools th Page | 40 t to Table 4.5c External return to education for public school ng hi ep Schooling year External return (%) 12 16 18 22 9.27 4.77 1.17 -1.53 -5.13 -6.93 -10.53 w n lo ad that the external return to education of individual graduating from public primary schools y th and lower secondary schools are at 4.77% and 1.17%, while individual not going to ju yi school have an increase at 9.27% of hourly wage if the share of skilled workers in local pl labor market increase one percent However, the external return to education become al n ua negative for people graduating from upper secondary and above For instance, individual n va graduating from public universities would have a decrease of hourly wage at 5.13% when ll fu the proportion of highly educational workers increase one percent oi m In addition, SchYear coefficient is equal 0.203 – 0.009*Educated + 0.0465*HighGDP at nh we could say the private return to education of individual living in province/city having GDP higher than average is higher 4.65% than individual living in city/province with z z GDP lower than 82,875 billion VND In addition, the private return to education of vb jm ht individual living in high-GDP province/cites become negative when the share of skilled workers in labor force is higher than 27.7% and the turning-point of skilled workers’ k Chapter remark: an Lu 4.6 om l.c gm share for province/city having GDP lower than average is 22.5% In this chapter, the overview of Vietnam in economy and education helps to make an Page | 41 th of school The returns to education are estimated on OLS model, then the apply of ey research, are described by classifying the whole sample in gender, marital status, and type t re in the paper The distributions of hourly wage, which is dependent variable in the n va introduction about Vietnam, which is related to the characteristics of whole sample using t to instrument variables in order to deal with endogeneity of Educated and SchYear ng Interaction terms are used in the final step of estimating the return rate of education hi ep w n lo ad ju y th yi pl n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re th Page | 42 t to ng hi Chapter 5: Main findings and recommendations ep This chapter includes three parts, including main findings, recommendations, and w limitations While the main findings section helps to summarize the results of paper in n lo aiming estimate the relationship between the shares of highly educational workers on the ad individuals’ hourly wage, the recommendations provides some policies in order to y th ju improve Vietnamese performance The final part express the limitations of this paper in yi estimating the external and private return to education in Vietnam pl n ua al 5.1 Main findings: va The main purpose of this paper is up to date in estimating the external and private return n to education in Vietnam The combination of Vietnam Household Living Standard fu ll Survey from 2010 to 2014 and Provincial Statistics Yearbook in these years creates an m oi interesting opportunity for this paper’s aim The Ordinary Least Squares, Instrument nh at variable Two Stage Least Squares and interact terms are the main methodologies in this z paper when adopting the Human Capital Earning Function (Mincer, 1974) z vb jm ht According to the results expressing in the previous chapter, the paper states that there is positive external return to education in Vietnamese labor market, which is difference in k gm every educational level However, an increase one percent in the share of trained workers l.c only makes the hourly wage of individuals who graduates from lower secondary and om below For instance, people who never go to school get the highest external return to an Lu education at 7.7% and respectively at 3.7% and 0.5% for primary graduates and lower secondary graduates Due to the supply effect in labor market, for people who graduates va from upper secondary and above, the external return to education becomes negative n for people has schooling year at 12, at (-5.1%) for people Fan and Ma (2012) about the external return to education low and high educational Page | 43 th Doctoral Certification holders The results from this paper have the similar argument with ey graduating from universities, at (-6.7%) for master degree and decreases to (-9.9%) for t re significant at (-1.9%) t to workers Fan and Ma (2012) stated that the external return to education is strong positive ng significantly for low-education workers at 17.4% while, the external rate of return for hi ep Chinese graduating from college and above is not significant when using IV fixed-effect and at (-4.5%) when dropping out instrument variable w n The private return to education is an additional reward in this paper This rate of return is lo ad based on two interact terms, including the pressure of competition and economic y th development in the province/city where individual works The first interact term is ju yi between dummy variable HighGDP (province/city having GDP higher the average = 1) pl and schooling year (SchYearHighGDP), which helps to describe the differences of al n ua educational level across the development level of provinces/cites The second interaction va variable belongs to share of highly educational workers in the labor market and schooling n year (EduSchYear) This interact term is added in order to express the differences of fu ll skilled workers’ supply on the education of individual oi m nh From estimating the whole sample, the private return to education is equal 0.184 – at 0.008*Educated + 0.037*HighGDP In other words, when the share of skilled workers in z z the province/city having GDP higher than average increase one percent, an additional vb jm ht schooling year makes the hourly wage of individual increase 21.3% This rate of return is reasonable when compared to the private return of Asian at 20% from Psacharopoulos k gm and Patrinos (2004) Comparing to Tinh (2011), who found the private return to l.c education is about 17% on average when using Vietnam data set in 2008, this paper om expresses an improvement in schooling benefit from 2008 to 2014 in Vietnam For an Lu instance, the private return to people working in cites or provinces having GDP higher than 82,875 billion VND will become negative when the share of highly educational va workers is higher than 27.65% While the turning point to negative value of SchYear in n ey t re city or province, whose GDP is lower than average, is higher than 23% of highly educational labor force th In addition, the results from classifying the whole sample by gender, marital status, and type of school express some interesting findings The external return to education is only Page | 44 t to significant for female living with husband and studying in public school For individual ng living with husband/wife, an increase in share of skilled workers in labor market, the hi ep hourly of this person is increase 9.47% for who not go to school, 2.27% for who graduating from lower secondary school and this rate of return become negative when w n this person graduating from upper secondary school and above lo ad For female, the external return to education is higher than estimating whole sample y th (13.7% compared to 7.7%) for who not go to school While whole sample have ju yi external return to education at (-9.9%) for individual have Doctor Certification, one pl percent increased in share of skilled labor force make the hourly wage decrease 6.1% for al n ua female, who is Doctor Certificate holder About the private return to education, an va additional year of schooling have a positive benefit for male when the share of skilled n worker is not greater than 23.3% and 22.4% for female ll fu oi m at nh 5.2 Policy Recommendations: z According to the main findings and some abstractions discussing in the previous sections, z - jm ht vb this paper suggests some recommendations in the following: The province or city with better economic performance and living standard often k gm attract many workers, which make the return to schooling decrease and could l.c become negative Therefore, the policies, which help to arrange the distribution of - an Lu worker across the country, are necessary om labor force in city or province to balance the proportion of highly educational The negative external return to education could express the lack of high-skill- va require jobs in the economy In order to improve this issue, some policies helps to n The external return to education has the highest value for people never going to Page | 45 th school and be positive for the groups of people graduating from lower secondary ey - t re increase the demand for highly skilled labor force are very important t to and below This implies that investments in the primary and lower secondary ng schools bring many benefits for human capital in our nation hi ep - There should be some polices that encourage people improve the educational level, especially illiterates w n lo ad ju y th 5.3 Limitations: yi pl This paper deals with some abstractions, which may not explain the educational effects al n ua on the individuals’ wage The first issue comes from the interviewees in the Vietnam va Household Living Standard Survey This survey is conducted in the regions where people n usually participate in the economics by farming activities Farming activities does not fu ll require a good or high certification in education and it is very difficult to collect the m oi information about the wage that these people get from their jobs nh at In addition, Vietnam has a great gap between 64 provinces/cites According to Provincial z z Statistics Yearbook (2014), there are only five provinces/cities with GDP higher than vb jm ht 100,000 billion VND and the highest GDP is over 650,000 billion VND for Ho Chi Minh City Due to this difference, the returns to education varies and fluctuates widely k gm The lack of empirical for external return to education is also a difficulty, then the issues l.c in this type of estimating about the model and approaches is not provided many om experiences in dealing with them an Lu n va ey t re th Page | 46 t to REFERENCE ng hi Acemoglu, D., & Angrist, J (2001) How large are human-capital externalities? Evidence ep from compulsory-schooling laws In NBER Macroeconomics Annual 2000, Volume 15 (pp 9-74) MIT Press w n lo Altonji, J G., & Card, D (1991) The effects of immigration on the labor market ad outcomes of less-skilled natives In Immigration, trade, and the labor market (pp 201- y th ju 234) University of Chicago Press yi pl Bakis, O., Davutyan, N., Levent, H., & Polat, S (2010, April) External returns to higher al n ua education in Turkey In Economic Research Forum Working Paper Series (No 517) va Barro, R J (1991) A cross-country study of growth, saving, and government n In National saving and economic performance (pp 271-304) University of Chicago ll fu oi m Press at political economy, 9-49 nh Becker, G S (1962) Investment in human capital: A theoretical analysis The journal of z z vb Becker, G M., DeGroot, M H., & Marschak, J (1964) Measuring utility by a jm ht single‐response sequential method Behavioral science, 9(3), 226-232 k Ben-Porath, Y (1967) The production of human capital and the life cycle of l.c gm earnings The Journal of Political Economy, 352-365 an Lu 449-468 om Canton, E (2007) Social returns to education: Macro-evidence De Economist, 155(4), va Doan, T T (2011) Labour market returns to higher education in Vietnam Economics n Discussion Paper, (2011-4) ey t re Fan, W., & Ma, Y (2012) Estimating the External Returns to Education: Evidence from th China Page | 47 t to Katz, L F., & Murphy, K M (1991) Changes in relative wages, 1963-1987: Supply and ng demand factors (No w3927) National Bureau of Economic Research hi ep Le, P V (2014) More Schooling Is Not Always Better: Evidence from an Instrumental Variables Approach to Educational Reform in Vietnam w n lo Lindahl, M., & Krueger, A B (2001) Education for Growth: Why and for ad Whom? Journal of Economic Literature, 39(4), 1101-1136 y th ju Liu, Z (2007) The external returns to education: Evidence from Chinese cities Journal yi pl of Urban Economics, 61(3), 542-564 al n va economics, 22(1), 3-42 n ua Lucas, R E (1988) On the mechanics of economic development Journal of monetary ll fu Mincer, J (1974) Schooling, Experience, and Earnings Human Behavior & Social oi m Institutions No nh Moock, P R., Patrinos, H A., & Venkataraman, M (2003) Education and earnings in a at z transition economy: the case of Vietnam Economics of Education Review, 22(5), 503- z ht vb 510 jm Moretti, E (2004) Estimating the social return to higher education: evidence from k longitudinal and repeated cross-sectional data Journal of econometrics,121(1), 175- l.c gm 212 an Lu in Management and Applied Economics, 3(2), 193 om Nazier, H (2013) Higher Education Externalities in Egyptian Labor Markets.Advances va Nelson, R R., & Phelps, E S (1966) Investment in humans, technological diffusion, and n economic growth The American economic review, 56(1/2), 69-75 Page | 48 th 1122 ey and returns to education in Vietnam Journal of Comparative Economics, 41(4), 1106- t re Phan, D., & Coxhead, I (2013) Long-run costs of piecemeal reform: wage inequality t to Rauch, J E (1991) Productivity gains from geographic concentration of human capital: ng evidence from the cities (No w3905) National Bureau of Economic Research hi ep Psacharopoulos, G., & Patrinos*, H A (2004) Returns to investment in education: a further update Education economics, 12(2), 111-134 w n lo Schultz, T W (1961) Investment in human capital The American economic review, 1- ad 17 y th ju Smith, A (1776) The wealth ofnations New York: The Modern Library yi pl Smith, A., & Nicholson, J S (1887) An Inquiry Into the Nature and Causes of the al n ua Wealth of Nations T Nelson and Sons va Stock, J H., & Yogo, M (2005) Testing for weak instruments in linear IV n ll fu regression Identification and inference for econometric models: Essays in honor of oi m Thomas Rothenberg nh Strawinski, P (2008) Changes in return to higher education in Poland 1998- at z 2005 Available at SSRN 1159044 z ht vb Summers, R., & Heston, A (1988) A new set of international comparisons of real k wealth, 34(1), 1-25 jm product and price levels estimates for 130 countries, 1950–1985 Review of income and gm l.c Tien, N D (2014) An analysis of labour market returns to education in Vietnam: an Lu the International Labour Organization Working Paper, om evidence from the national labour force survey 2012 International Training Centre of va Wang, Q., & Liu, N C (2014) Higher education research institutes in Chinese n universities Studies in Higher Education, 39(8), 1488-1498 ey t re th Page | 49 t to APPENDIX ng hi Appendix 1: Hausman test for OLS estimate ep w n lo ad ju y th yi pl n ua al n va ll fu oi m at nh z z Appendix 2: Hausman test for IV estimate without addional control variables k jm ht vb om l.c gm an Lu n va ey t re th Page | 50 t to Appendix 3: Hausman test for IV estimate with additional control variables ng hi ep w n lo ad ju y th yi pl n ua al n va ll fu oi m at nh z z vb variables k jm ht Appendix 4: Tests for Instrument variables in IV estimate with additional control om l.c gm an Lu n va ey t re th Page | 51 t to Due to the collinear issue, the SchHead instrument variable is dropped out of IV estimate, ng hence two endogenous variables (Educated and SchYear) are instrumented by NumLec, hi ep ShPHo and StuLecRatio The following tests are the test of validity for the remaining instrument variables w n In the underidentification test, p-value = 0.000, which means that the instruments lo ad variables are relevance to the endogenous variables y th ju While, the weak identification test provides the critical value of 16.27, which is higher yi than 10% maximal IV size for the case of two endogenous and three instrument variables, pl ua al it means the instruments variable are not weak n In the overidentification test of all instruments, p-value is higher than 0.05, which implies va n that the set of instrument variables are exogenous fu ll Combining three of tests in Appendix 4, the set of instrument variables, including oi m NumLec, StuLecRatio and ShPho, is valid at nh z z k jm ht vb Appendix 5: First-stage regression of Educated variable om l.c gm an Lu n va Page | 52 th StuLecRatio and ShPHo have a significant impact on this endogenous variable ey t re In the first-stage regression of Educated variable, the two instrument variables t to Appendix 6: First-stage regression of SchYear variable ng hi ep w n lo ad ju y th yi pl ua al n There are two instruments variables, including NumLec and ShPHo have a very strong va relevance to SchYear, while the StuLecRatio does not have a significant influence on this n ll fu endogenous variable oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re th Page | 53