External and private return to education by using instrument variables approach evidence in vietnam with a panel data

In this paper, the Human Capital Earning Function Mincer, 1974 and instrument variables are adopted in order to estimate the external and private return to education.. While the external

UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM

HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES

VIETNAM THE NETHERLANDS

VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

EXTERNAL AND PRIVATE RETURN TO EDUCATION BY

USING OF INSTRUMENT VARIABLE APPROACH:

EVIDENCE IN VIETNAM WITH A PANEL DATA SET

BY

MR LE THANH HUNG

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, October 2016

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

HO CHI MINH CITY THE HAGUE

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

EXTERNAL AND PRIVATE RETURN TO EDUCATION BY USING OF INSTRUMENT VARIABLE APPROACH:

EVIDENCE IN VIETNAM WITH A PANEL DATA SET

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

MR LE THANH HUNG

Academic Supervisor:

PROF NGUYEN TRONG HOAI

HO CHI MINH CITY, October 2016

ACKNOWLEDGEMENT

To be able to finish this thesis, I have received the great supports from many people

Firstly, I would like to express my appreciation and special thanks to Prof Nguyen Trong

Hoai, my academic supervisor, who has given me many valuable guidance, advices, and

great encouragements for my thesis Secondly, I would like to express my gratitude to

Lecturers and Staff from Vietnam – Netherlands Program at University of Economics Ho

Chi Minh city Specially, I am indebted to Ph.D Truong Dang Thuy, who gave me

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

family and my friends, who gave me the greatest encouragements for my study

DECLARATION

I declare that “External and private return to Education by using of Instrument

Variable Approach: Evidence in Vietnam with a panel data set.” is my own work

This thesis is has not been submitted to any degree or examinations at any other

universities In addition, all the using sources are indicated by the completed references

ABSTRACT

The combination of Vietnam Household Living Standard Survey and Provincial Statistics

Yearbook from 2010 to 2014 provide a great opportunity for estimating the up to date

external and private return to education in Vietnam In this paper, the Human Capital

Earning Function (Mincer, 1974) and instrument variables are adopted in order to

estimate the external and private return to education The analysis suggests that not only

one additional schooling year have an impact on individual wage, but the increase in

proportion of skilled workers in the labor force also have an influence on the hourly wage

of individual

TABLE OF CONTENTS

ACKNOWLEDGEMENT i

DECLARE ii

ABSTRACT iii

TABLE OF CONTENTS iv

LIST OF TABLES vi

LIST OF FIGURES vii

LIST OF ACRONYMS viii

Chapter 1: Introduction 1

1.1 Introduction 1

1.2 Research objectives 3

1.3 Research scope 3

1.4 Structure of paper 3

Chapter 2: Literature review 5

2.1 Human capital theory 5

2.2 Returns to education 8

2.3 External return to education 10

2.4 Chapter remark 12

Chapter 3: Research methodology 14

3.1 Research methods 14

3.2 Endogeneity in Wage function 15

3.3 Data sources and measurement 16

3.4 Instrument variables 18

3.5 Additional provincial control variables and interact terms 20

3.6 Data description 21

3.7 Chapter remark 22

Chapter 4: Research results 24

4.1 Overview about Vietnam 24

4.2 Individual earning on individual characteristics 27

4.3 Individual earning on levels of education 28

4.4 Results on estimating returns to education 29

4.5 Returns to education classified by dummy variables 36

4.6 Chapter remark 41

Chapter 5: Main findings and recommendations 43

5.1 Main findings 43

5.2 Policy recommendations 45

5.3 Limitations 46

REFERENCE 47

APPENDIX 50

LIST OF TABLES

Table 3.3 Definitions and unit of individual-level variables 17

Table 3.4 Definitions and unit of provincial level variables 18

Table 3.5 Descriptive statistics for continuous variables 20

Table 3.6 Descriptive statistics for dummy variables 21

Table 4.2.1 Individual earning classified by gender 27

Table 4.2.2 Individual earning classified by marital status 28

Table 4.2.3 Individual earning classified by type of school 28

Table 4.3 Individual earning on levels of education 29

Table 4.4.1 OLS estimate 29

Table 4.4.2 Instrument variables estimate without control variables 31

Table 4.4.3a IV estimate with additional control variables 33

Table 4.4.3b External return to education for levels of education 34

Table 4.5 IV fixed effect with additional control variables for groups of individual’s characteristics 37

Table 4.5a External return to education for female 39

Table 4.5b External return to education for Married = 1 40

Table 4.5c External return to education for public school 41

LIST OF FIGURES

Figure 2.4: Analytical framework 13

Figure 4.1a: GDP across six key economic regions in Vietnam 2010-2014 25

Figure 4.1b: Educational system in Vietnam 26

LIST OF ACRONYMS

External and private return to Education

by using of Instrument Variable Approach:

Evidence in Vietnam with a panel data set

Chapter 1: Introduction

1.1 Introduction:

Human capital, physical capital, and development of technology contribute to the economic growth of a nation Specially, investing in human capital does not only enhance the productivity of labor force, but it also contributes to the improvement of social welfare, such as decreasing the criminal rate, upgrading the living standard, etc Aristotle

- a famous Greek philosopher – said “the fate of empires depend on the education of youth.” This statement convinces the major factor in the wealth of a nation of education, which is a well-known representing term for human capital

In Vietnam, the educational system has a great improvement from Doimoi policy since

1986 Before 1986, the illiteracy rate accounted more than ninety-five percent on total population, this rate decreased at 2.7% in 2015 according to General Statistics Office The government considers investment in education is one of the most priority policies and the proportion of this investment has to be at least 20% or higher on the total National Budget Understanding the benefits from education to social welfare provides valuable consultations for making policies in improving the wealth of nation The importance of education encourages me to study the educational benefits in Vietnam with

an up-to-date information from Vietnam Living Standard Survey 2010-2014

Nazier (2013) defined the private return to education as the productivity that individual gains from his own investing to education While the external return to education is considered as the productivity that individual gains from the local human capital In other words, the external return to education is the influence of the share of local educated labor force on individual’s wage The external return to education is the sum of spillover

effect and demand of labor force effect In which, the spillover effect enhance the workers’ productivity through the sharing ideas between workers, while demand of labor force effect speaks the higher competitive pressure on workers when increasing the skilled workers in the labor market In the modern economics theory, Moretti (2004) defined the social return to education as the sum of private return to education and external return to education

Moretti (2004) stated that for at least a century, economists have speculated that social return may exceed the private return to education The external return to education is one

of the most reasonable reasons for this speculation Understanding about human-capital externalities or external return to education brings a good guideline for policies making

or produce the better development strategy for any country

Economists have been paying so much in the private return to education, for instance the works of Doan (2011), Le (2014), and Wang and Cai (2014), while there are few of papers, which consider about external return to education This situation encourages me

to study the external return to education in Vietnam in order to contribute to estimate the relationship of the highly educational workers’ proportion on the individual wage by using the lasted dataset about Vietnam In addition, the studies conducing in Vietnamese education usually used one year Vietnam Household Living Standard Survey (VHLSS) in the period 1992-2008 While my paper is going to use the panel data from combining three years of VHLSS in 2010, 2012 and 2014 for individual level and the supporting of Provincial Statistics Yearbook of all 64 provinces/cities across Vietnam to analyze the returns to education

In this paper, some research questions are concerned:

- Does share of skilled labor force impact on individual’s wage?

- Is there any difference influence of schooling years on individual’s wage across the regions and the share of skilled labor force?

be bias due to the differences in policies, orientation in educational system between countries Therefore, the micro approach is adopted in this paper

1.4 Structure of paper:

This study will continue with four following chapters:

Chapter 2 illustrates in the definitions, literature, and methodologies of estimating the return to education The famous Mincer’s Human Capital Earning Function (1974), which dominates the approach of estimating the main purpose of this paper The previous papers in this field are summarized in this chapter, too

Chapter 3 describe the improvement from basic Mincer Human Capital Earning Function and using instrument variables approach in order to obtain a consistent external and private return to education

Chapter 4 expresses the results when adopting the methodologies, which are discussed in previous chapter The results are also represented by classifying the whole sample in gender, marital status, and types of school

Chapter 5 interprets the main findings in the paper and some recommendations in making polices

Chapter 2: Literature Review

This chapter includes four sections, which concentrate in human capital theory, some definitions, and related empirical studies on the human capital externalities The first part presents the history of human capital theory from the very early period to the present The next part aims to exploit the rate of return to education, including private and external return to education The third section focuses on the sign and size of external return to education in the previous researches The chapter remarks is in the final part

2.1 Human capital theory:

According to the empirical labor economics literature, Rauch (1991) defined human capital as the accumulation from two measurable components: education and experience that an individual could obtain in his lifetime In this case, the education is measured by the number of years for completing school, while experience is calculated by the current age minus schooling years minus the begin-attend-to-school age (or six)

In the serial book titled “The Wealth of Nations”, Adam Smith is the first economist who found out the relationship between skilled-worker and higher earning wage in 1776 In the 1960’s, Theodore W Schultz and Gary Becker developed the human capital theory

by considering knowledge and skills that people obtained from vocational and technical education as capital Theodore W Schultz (1961), a Nobel-winning-economist, established the term “human capital”, which mentioned about the productivity gain from investing in education and/or training-on-job According to him, this type of capital is one of the core factors for economic growth In 1964, Becker stated that education is a kind of investment in human capital and the individual would refer the future income from higher level of education than the opportunity cost from being in the labor market

In this research, Becker (1964) also provided the differences that influenced to the change

of individuals’ earnings through investment in human capital:

- Education at school: the knowledge that individual obtains from situation which

provides education as a product

- Training at work: the knowledge that individual “collect” from work place or learning-by-doing effect The distinction between general and specific training is

one of the most valuable contributions from Becker to human capital theory

o General training: the knowledge that workers can use in many firm However, workers have to pay all cost for this training or get the wage lower than their current productivities

particular firm, while the cost for this training is shared for both firm and workers

In 1974, Jacob Mincer used the findings of Human Capital (1964) from Becker as conceptual framework in order to estimate the return to education by using 1960 United State Census’ data In this work, Mincer did create a useful methodology – Human capital earning function – which dominated the way that economics estimate the return to education at the micro level This famous wage regression is developed from the human capital accumulation model (Ben Porath, 1967), which describe the relationship between the market wage and the skills-owned of individual in the competitive labor market It is:

Wt = Pt.Ht (2.1) Where:

In order to develop (2.1) into a econometric model, Mincer used number of schooling

Et+1 = Et + ct.pt = Et(1 +kt pt) (2.2) Where:

Et as the potential earnings at t period

Et = [ j p0)]E0 (2.3)

We have two period of a person in working life cycle, including in school and training, so

Hence, ρs is the private return to education and β1 is represented for the relationship between wage and experience

The contributions from Garry Becker and Jacob Mincer are considered as the basic conceptual framework for the modern human capital theory

2.2 Returns to education:

Lucas (1988) developed human capital to a higher level by introducing the externalities

of human capital In the first edition of Human capital, Becker (1962, p.45) also had a sentence about this: “An emphasis on human capital not only help explain differences in earning over time and among areas but also among persons or families within an area.” According to these points of view, Nazier (2013) defined two types of return to education

as following:

- Private return to education: the productivity that individual gains from his own investment in human capital This rate of return is usually estimated by the affect

of individual’s schooling years on his wage

- External return to education: the productivity, which spills over to the others in the same firm, region, industry, or nation In the other words, this type of return to education is measured by estimating the influence of the share of skilled labor force on the individual’s wage

There are two main approaches estimating the external return to education, which is distinguished by the scope of research:

- At the macro approach:

By using the macro-level-data, this approach aims to use the differences of input factors

to explain the differences of productivity across countries (Canton, 2007) In this type of approach, we could mention the working of Robert J Barro in 1991 “Economic growth in

a cross section of countries” Basing on the viewpoint of Nelson and Phelps in 1966,

which stated that a country with more human capital had tended to grow faster due to the ability to catch up the new technologies, Barro (1991) estimated the relationship between growth rate of Gross Domestic Product per Capita and school enrollment rate In this work, Barro used the data from Summer and Heston (1988) in 98 countries from 1960 to

1985 The school enrollment rate, which is used as the proxy for human capital, is defined as the number of students that enroll for primary and secondary school to the total population for the corresponding group of age His finding implied that one percent increased in the school enrollment rate led to the increase in growth rate of GDP per Capita at 2.5% for primary level and 3% for secondary level However, Barro found that his analysis difficult to explain the performance in the below-average-growth-rate nations In addition, Krueger and Lindahl (2001) implied the omitted variables in the macro approaching For instance, each country has their own way in order to improve educational system and the growth of economy concurrently; hence, this creates a reasonable omitted variable in estimating the return to education for cross-country data

- At the micro approach:

According to the argument of Lucas (1988) about the existence of externality from human capital investment and the human capital earning function, the second approach estimate the external return to education through micro-level-data The educational externality is often estimated by the influence of the average number of schooling years

in a firm, a region, or an industry, where the individual lives in or works for, on his wage Strawinski (2008) stated that individual’s wage was explained by his schooling years (private return), the average years of schooling in the related geographic area (external return), and addition instrument variables Some difficulties of this approach and the way how the previous studies used to estimate the external return to education is discussed in the next sections of this chapter

2.3 External return to education:

2.3.1 Negative external return to education:

The external return to education possibly has negative value if the supply of skilled labor exceeds the demand (Moretti 2004b) In this case, the number of high educational level workers leads to the decrease of the average wage in labor market and create the higher competitive pressure to the lower However, the negative value of external return to education is quite rare in the previous paper For instance, the work of Acemoglu and Angrist (2000) could be an example for negative external return to education

Acemoglu and Angrist (2001) argued that city with higher average educational level may also have higher wage In order to solve this issue, they decided to estimate the returns to education across the states in America due to the differences between the states from

1960 to 1990 in compulsory attendance laws and child labor laws Using the combination

of these laws as instrument variables (CLSs) for human capital, they found a little evidence for external return to education around 1-2% comparing to 7% in OLS estimating However, the interesting findings come from column 2 in table 11 in this paper there is the difference in external return to education when the authors allowed the vary in year, the external return is 2.8% by using Child Labor Law as instrument and be 1.7% by using Compulsory Attendance Law as instrument The external rate of return become negative at 1.8% when the author used Child Labor Law as instrument, while this rate of return is -3.00% when Compulsory Attendance Law instrument is adopted

2.3.2 Positive external return to education:

The external return to education could be positive due to two following reasons:

- According to standard neoclassical model, with the assumption that educated and uneducated workers are imperfect substitutes, there would be a growth of uneducated workers’ productivity when the proportion of educated workers is

increased In addition, Katz and Murphy (1992) proved the assumption, which is about imperfect substitutes of educated and uneducated workers

- The human spillover effect is existed (Moretti 2004b)

Rauch (1991) seems to be the first economist, who estimated the external return to education basing on the argument about the sharing knowledge and skills when individuals be working together In this paper, he used the average education level in 237 Standard Metropolitan Statistic Areas (SMSAs) and the logarithm of hourly wage of individuals in these areas to have the coefficient of external return to education at 2.8% in the United State

Moretti (2004) stated that the cities with higher proportion of educated labor force might have higher level of unobserved ability In order to control the differences among individuals in this type of ability and the differences in returns to skill across cities, Moretti used his owned version of National Longitudinal Survey of Youth (NLSY) He also used the Census data in 1980 and 1990 to compare with the coefficient in NLSY and these rates of return are similar In the first-differenced IV estimates in Moretti’s work stated that a one percent increases in the share of college graduates workers would lead to the increase in wage for high-school drop-out at 1.9%, for high-school graduates at 1.6%, for group did not finish college at 1.2% and at 0.4% for college graduates group In the early, almost economists focus on estimating the human capital externalities in one of the most powerful economic country – the United State

At the present, let us have a look on European economy Strawinski (2008) did applied Mincer’s Human capital earning function to estimate the external return to education in Poland By using main source of data from Households Budget Survey (HBS) in 1998 and 2005, Strawinski also contributed to the existence and significantly positive rate of external return to education Restricting empirical samples by age (16-65 for men and 16-

60 for women) helped to estimate external return in 1998-2005 period in the ordered of 0.7% to 1.3% for secondary education and 1.6% to 2.8% for tertiary level

2.3.3 Empirical studies in developing countries:

Bakis (2010) studied the external return to education in Turkey by Household Labor Survey in 2006, which just focused on workers in private sector He stated that the rate of return from human capital externalities is accounted for 2.4% with Instrument variables- OLS method and ranged from 1.3% to 3.5% with Instrument variables Quantile regression methodology

Liu (2007) seems to be the first economist who estimate the external return to education

in Chinese cities There is an interesting in this paper; the rate of external return from average schooling year across cities is range from 11% to 13%, while one percent of college share increase lead to the raise in individual’s wage only about 1%

In order to study the interesting findings from Liu (2007)’s work, Fan and Ma (2012) decided to estimate external return to education in a different way, such as using longitudinal data instead of cross-sectional data and estimating in both urban and rural areas in China In this paper, Fan and Ma use an instrument variable, named “211”, which represents for the Project 211 from Chinese Government aims to “label” the universities in this project as high-quality ones As the result, the rate of eternal return to education is range from 10% to 14% due to one percent increase in the share of college graduates

This chapter provides the foundational literature for estimating external return to education in this paper While, the contribution from human capital theory and human capital earning function is considered as the conceptual framework, the related studies supply valuable lessons in order to improve the quality of analysis Basing on the basic Mincer’s Earning Function (equation 2.7), and the previous studies about external return

to education, this paper would used the following model to estimate the external return to education in Vietnam, by adding the share of skilled workers in the local labor market (Hjt):

LnWijt = αSijt + βHjt + ᵹXijt + μjt + ꞓijt (2.8)

In which,

type of school)

According to equation 2.8, the private return to education (α) is defined as the wage of individual gaining by his owned schooling years, while the external return to education (β) is considered as the increase of individual’s wage due to the proportion of educated workers in his local area

Fig 2.4 Analytical framework

capital earning function

Private return to education

Returns to education

External return to education

The share of skilled workers

in the labor market Schooling years of

individual

Chapter 3: Research Methodology

This chapter is included six parts describe the steps using Mincer’s Human Capital Earning Function (HCEF) to estimate the external and private return to education; and how to calculate the variables and some arguments, which uses to apply the Instrument Variables for controlling the differences across the cities/provinces In the first section, basing on introducing HCEF in chapter 2, I discuss how to apply this function in estimating the external return to education The second part discuss the endogeneity in earning function The next part provides the sources of dataset and the approach calculating them in details The fourth one supplies more Instrument Variables in order to deal with the basic issues in earning function In addition, the next part illustrates the addition control variables, their definitions, and the way to calculate them In the next section, the whole sample used in this paper is described for two group, including continuous and dummy variables The final part of this chapter is chapter remark, which helps to summarize the important information

3.1 Research methods:

Moretti (2004) stated that the human capital spillovers would exist when we estimate the influence of the share of educated or trained labor force on the worker’s productivity Basing on that statement and the basic Mincer Human Capital Earning Function (2.7) in the chapter 2, I would add the share of local skilled human capital into the equation Hence, the external return to education across Vietnamese cities/provinces is estimated

by the influence of the local trained workers’ share on the individual’s hourly wage in that region Then, we have:

LnWijt = αSijt + βHjt + ᵹXijt + σPjt + μjt + ꞓijt (3.1)

In which,

t, including Age, Gender, Married Pjt denotes the provinces/cites characteristics While,

considered as the coefficient of capturing the effect of the skilled local human capital’s

In this paper, the proportion of trained workers on the total working employees in province/city j is consider as the proxy for the share of skilled labor force in that region According to Provincial Statistic Year Book (PSYB), trained worker has to satisfy two conditions Firstly, the worker has to be working in that region; even he comes from another region In addition, the worker is considered as trained when he has certification

or degree for short-term training, trade vocational, trade college, vocational school, college, university and over from a school or any educational organization This data provides an advantage for my estimate due to its first condition This condition helps to solve the problem of migration instead of using the share of provincial educated people For instance, a worker could be graduated at his hometown, but he moves to another province/city to work Hence, the share of trained worker is prefer than the share of

model (3.1) becomes as following:

LnWijt = αSijt + βEducatedjt + ᵹXijt + σPjt + μjt + ꞓijt (3.2)

3.2 Endogeneity in Wage function:

In order to have a Best Linear Unbiased Estimates, there are 2 issues that I have to deal with when using the earning function to estimate the returns to education The first is endogeneity of the individuals’ schooling years and the share of skilled human capital (trained workers) across cites/provinces

Endogeneity problems mentions about the presence of unobserved factors that are correlated with local and individual human capital and the differences in wage across regions, which lead the upward bias OLS estimates The previous studies focused on, including two main problems:

Firstly, the unobserved factor, ability, which is correlated with individual wage and his human capital or schooling years Basing on the idea about the individuals with higher ability often obtain higher educational level then the others with lower ability This results in the overestimate in private return to education when we use OLS without any additional control variables

Secondly, unobserved the factors, which are related to the differences across regions such

as amenities, weather, geographic location, or industrial structure, which also have an impact on the individual’s productivity in specific area

In order to deal with the endogeneity, the previous researches suggest Instrument Variables (IVs) to control the unobserved factors that leading to bias estimates In principle, these IVs should help to explain individual and local human capital while not

to be correlated to the unobserved factors influencing on wages

Secondly, the scale of this study is at provincial level, there would be some unobserved factors that affect to the local human capital and the productivity of workers in specific region Solutions for these issues will be discussed into details in the next sections

3.3 Data sources and Measurement:

The individual-level data of this paper is from Vietnam Household Living Standard Survey (VHLSS) in 2010, 2012 and 2014 The combination of these surveys provides 109,748 observations across 64 provinces/cities in Vietnam This dataset provides the information about individuals’ characteristics, including wage, age, gender, schooling

Table 3.3 Definitions and unit of individual-level variables

SchHead (year) The number of schooling years of the head of

household which the individual i live in

VHLSS

living with husband or wife = 1 , not = 0

VHLSS

= 0

VHLSS

SchType The type of school that the individual i studied in, in

which public school = 1, not = 0

VHLSS

which graduates upper-secondary school or below =

1, not = 0

VHLSS

which higher than upper-secondary school and lower than university =1, not =0

VHLSS

which at university or above =1, not = 0

Adopting the equation (3.2), which specific variables in this paper, we have the empirical model:

In which,

3.4 Instrument Variables:

Arcand (2005) used parents’ education and smoking habit as demand-side variation in schooling instrument variables and proximity to primary school and colleges as supply-side variation in schooling instrument variables, to estimate the private return to education in Vietnam In addition, Sripad and Lars (2011) explained the demand-side variation of schooling as the characteristics of individual and family have an impact on individual’s schooling choice Sripad and his partner also argued about the supply-side variation of schooling as the availability and quality of school have an influence on the schooling choice

Table 3.4 Definitions and unit of provincial level variables

Educated (%) The share of people, who have degree of certification

for short-term training, trade vocational, trade college, vocational school college, university and over; and be working, to the total workers in a specific province/city

GDP higher than the average = 1, not = 0

PSY

The most concentrating issue in this paper is endogeneity of main independent variables: SchYear explaining for private return to education and Educated explaining for external return to education In this paper, I use instrument variables for both SchYear and Educated Basing on two conditions for a valuable instrument variable, including correlated to instrumented variable and not correlated to the residual in earning function

- For demand-side variation of schooling:

The schooling year the master’s household (SchHead), in which the individual lives The SchHead could be correlated to individual’s schooling year, because the master of household with higher educational level tends to encourage the members in his family to study While, there is no direct relationship between individual’s wage and unobserved factors affecting wage, ability for instance

The share of poor household at a province could relative to the share of highly education workers For low-income households, the expenditure for uniform, materials, or studying tools is a big abstract for them to go to school Therefore, the share of poor household in

a province/city have an effect on the skilled labor force proportion in that province/city

- For supply-side variation of schooling:

(StuLecRatio)

The number of lecturers in universities/college helps to describe the quality or the supply for professional technique/knowledge training and teaching, while it does not affect to the unobserved factors influencing on individual’s earning In addition, the ratio of number

of students on a lecturer could express the quality of universities and colleges in province

j at t period, which have an impact on schooling choice and not influence in the individual’s wage

3.5 Additional provincial control variables and interact terms:

Although this paper uses Instrument variables to deal with the unobserved factors affecting to individual wage per hour, my estimate could be bias due to the unobserved time-varying factors In order to strengthen the estimate, I add more addition provincial variables and interact terms, including: GDP, interact term of share of skilled workers in the labor market and schooling years (EduSchYear) and the interaction of schooling year and dummy variable for provinces/cities with GDP higher than average in the whole sample (SchYearHighGDP)

Table 3.5 Descriptive statistics for continuous variables

These addition variables helps to explain more about the differences across 64 provinces/cities in the whole sample

The Instrument variables and addition control variables are collected from the Provincial Statistic Year Book in 2010, 2012 and 2014 The definitions and measurement unit of these variables are listed on table 3.4

3.6 Data description:

The data description for continuous variables is available at table 3.5 According to this table, there is a big gap in hourly wage of individual across 64 provinces/cities in Vietnam The minimum value of HrWag is at 194.44 VND, while the maximum value is

at 539,750 VND and average wage of Vietnamese is 21,617.84 VND per hour SchYear provides the information about the educational level of whole sample; the maximum value is 22 years, which is represented for Ph.D or Doctor Degree, while the average value is 6.9 years The proportion of trained workers on the total working people in a province/city is range from 5.1% to 38.4% for whole sample

Table 3.6 Descriptive statistics for dummy variables