Multi-dimensional labour market outcomes of higher education’ graduates: Evidence from a national survey in vietnam

Nội dung text: Multi-dimensional labour market outcomes of higher education’ graduates: Evidence from a national survey in vietnam

1. MULTI-DIMENSIONAL LABOUR MARKET OUTCOMES OF HIGHER EDUCATION’ GRADUATES: EVIDENCE FROM A NATIONAL SURVEY IN VIETNAM Bach Ngoc Thang National Economics University thangbn@neu.edu.vn Tran Ngoc Diep dieptran014@gmail.com Abstract Youth employment is a seriously concerned issue in developing countries, given recent moves to mass higher education. The limitations of unemployment rates as a guide to youth employment problems are widely recognised, but little has been known about other labour market outcomes of higher education’s graduates. This study takes advantages of a large-scale Vietnamese national survey on labour forces in 2015 to shed light on the effects of higher education on recent graduates’ multi- dimensional labour market outcomes. The findings show that college and university graduates outperform high school counterparts in terms of high-quality labour market outcomes, such as having higher wages and better employment protection, though the mere stance of being employed does not differ markedly among these graduate cohorts. The paper contributes to recent debates on higher education in a developing country like Vietnam, where the focus is on whether the move towards mass higher education might result in differed labour market outcomes for recent graduates. Key words: labour market outcomes, higher education, youth employment, selection bias, Vietnam. JEL classifications: I23, J21, J28. 1. Introduction Youth employment is a seriously-concerned issue the developing countries. The issue is even more acute given recent moves to mass higher education (Tran and Nørlund 2015; Ou and Zhao 2016). The study of youth employment is broad, and the limitations of unemployment rates as a guide to youth employment problems are widely recognised (Rees 1986; Ryan 2001). Various labour market outcomes of the youth are critical in their transition to a “transited” job - meeting a basic criterion of 631
2. permanency, providing the worker with a sense of security, or a job that worker feels personally satisfied with (Elder and Kring 2016). Therefore, the mere stance of being employed is not sufficient for a recent graduate, who just graduates and in the first years of employment in the labour market, to successfully move to a “transited” job. The other aspects, such as wages and the level of employment protection are equally important for a smooth transition process. In the area of youth employment, current works mostly focus on changing nature and factors affecting school-to-work transition (Quintini, Martin, and Martin 2007; Mavromaras et al. 2013; Sortheix, Chow, and Salmela-Aro 2015). For example, Quintini, Martin, and Martin (2007) found that OECD countries still have a high level of youth unemployment notwithstanding a declining young cohort with better education. Another strand of literature deals with the effects of Chinese higher education expansion on the youth’s labour market outcomes, but it uses unemployment rate as a sole indicator (Li, Whalley and Xing 2014; Yu 2014; Ou and Zhao 2016). In addition, among these studies, the higher education expansion’s effects on unemployment rates are mixed. Few international studies have so far examined the higher educational effects on the recent graduates’ various labour market outcomes, other than unemployment rates. In Vietnam, it has been widely circulated that there is over-supply of higher education, and there is a mismatch between supply and demand in the labour market (Nguyen and Chu 2014). According to a recent report on the employment and social trends in Vietnam, the youth’s employment rate was on the rise, from 5.5 per cent in 2002 to 7.7 per cent in the second quarter of 2017, which was much higher than the senior labour cohort of 35 years old and above, 1 per cent (Institute of Labour and Social Studies and International Labour Organization, 2018). Higher education has expanded rapidly since 2004; enrolment in higher education tripled in a 10-year time (World Bank 2016), but its contribution to the national economy is debatable. Besides, a large fraction of the labour force worked in the informal sector, without social insurance (Huong et al. 2013; Cook and Pincus 2014). In the context of globalization and industrialization, mostly young female workers joined the labour force with non-liveable wages and substandard working conditions (Tran and Nørlund 2015). Therefore, an in-depth study on labour market outcomes of recent graduates in a developing country like Vietnam is worth pursuing, given the recent move towards mass higher education, and mixed views of education’s contributions to the national economy with low levels of formality and of social protection. This paper seeks to address the following central research question: How does higher education affect the labour market outcomes of recent graduates, which cover not only employment/work status but also other multi-dimensional outcomes after 632
3. being employed? The multi-dimensional labour market outcomes comprise wages, level of employment, having employment protection from permanent labour contracts, social insurance, and working in establishments with formal business registration. This study takes advantages of a large-scale national survey on labour forces in 2015 to shed light on the effects of higher education on the multi- dimensional indicators of recent graduates’ labour market outcomes. The empirical findings show that college and university graduates outperform high school counterpart in terms of high-quality labour market outcomes, though the mere stance of being employed does not differ markedly among these graduate cohorts. Specifically, in reference to high school graduates, college and university counterparts are less inclined to underemployment, have marked higher probability of getting a job with permanent labour contracts, social insurance, and working in entities with formal business registration. These findings remain unchanged even after having controlled for selection bias that is inherent in the labour market studies. This study contributes to the related literature on youth labour market transition in the following three ways. First, it examines a number of multi- dimensional indicators relevant to recent graduates’ labour market outcomes, not just employment status. These multi-dimensional indicators are believed to yield a balanced view on youth employment in developing economies, where it is not uncommon to see uncompetitive wages, underemployment, and a large number of employees working in informal sectors, without social protection. A recent similar paper is Tran (2017), but it focuses merely on wage differences and other descriptive statistics of employment outcomes. Second, this study aims to produce consistent estimates of higher educational effects on labour market outcomes, once having controlled for selection bias that is inherent in labour market studies. For example, the outcomes relevant to wages or the level of employment (underemployment or not) might be severely affected by unobserved heterogeneity related to ability, motivation, and attitudes to work. Third, it contributes to a recent policy debate on the interplay between mass higher education and youth unemployment in a developing country like Vietnam. The rest of this paper is structured as follows. Section two is review of related strands of literature on the interplay between educational attainments and labour market outcomes of the youth. Section three deals with the contextual issue of higher education expansion and youth labour market outcomes in Vietnam. Section four mentions data source, summary statistics, and the benchmark empirical models. Empirical results and discussions are included in the following section. Finally, section six wraps up the paper with conclusions and venues for further studies. 633
4. 2. Literature review Educational attainment and employment Education and training is believed to improve the well-being of individuals since it increases access to job opportunities and the associated benefits that come from the quality of employment. Cain (1966) explained an individual with higher level of education is more likely to participate in the labour market because a more educated individual has higher income aspiration and expectation than his/her less- educated counterparts. Many studies have empirically supported this positive relationship between level of education and the rate of employment (Kennedy and Hedley 2003; Kennedy, Storey and Vance 2009; Faridi, Chaudhry and Basit 2010). This finding still holds when taking the gender and mode of studies variables into consideration (Berliner 1983; Jaumotte 2003; Laplagne, Glover, and Shomos 2007, and Kennedy, Storey, and Vance 2009). Yet another existing literature strand suggests education level and employment opportunities can be negatively correlated. International Labour Organization (2015) that examines the labour markets of more than 60 countries of various development stages shows a mixed picture of relationship between education levels and probability of being employed. In developed countries such as Canada and Germany, education can protect one from being jobless. For the most part, those with a higher education degree stand a much higher chance of getting a job compared to their less-educated peers. Whereas, in countries like Thailand, the Philippines or Cambodia, higher education does not necessarily protect graduates from unemployment. Highly-skilled workers seem to find harder to get a job than those with only primary education level. Other empirical research also report the existence of such trend (see Standing 1981; Breusch and Gray 2004; Lattimore 2007 and Laplagne, Glover and Shomos 2007). In search for explanation of this discrepancy, the nature of the economy has been employed. Since almost all developing countries rely on resource-driven growth models, and specialise in labour intensive products, mostly unskilled labourers, less- educated individuals have higher work options than their higher educated counterparts (Leamer 1995, Stolper and Samuelson 1941, and Standing 1981). ILO (2015) also points out that developed countries often have a higher rate of unemployment compared to developing countries. It is suggested that since these countries have sufficient social safety net, their citizens can afford to be unemployed while looking for a better job opportunity or learning a new skill set in transition to a different career direction. Citizens of developing countries, however, often do not have this privilege. Therefore, they have to take up jobs to pay the bills even if the jobs can be below their qualifications and expectation. That implies if a study or 634
5. policy takes unemployment rate as the only one factor of consideration to assess labour market outcomes of graduates, the study is anything but sufficient. Rather, it is important to look further into the quality of employment including the type of contracts and social safety net. This notification is particularly essential when it comes to study labour markets of developing countries where informal economy dominates (Cling, Razafindrakoto and Roubaud 2011). Youth employment Youth employment is usually lower than adult employment since it takes longer for young workers to find a job after leaving school. As such, youth employment is a serious problem globally, even in well-established capital economies (Quintini, Martin, and Martin 2007). For young workers in developing countries, the future can even be gloomier (Elder and Kring 2016). Since higher education started to expand across the globe, which now results in a concern of oversupply of higher education graduates, arguments on negative impacts of having too many graduates on employment prospective have been put forward (McGuinness 2006; Carroll and Tani 2013). Even graduates believed that when almost everyone has a college degree, that piece of paper has started losing its role in giving the owners advantages in earning a job (Tomlinson 2008). On the other hand, however, education attainment seems to currently exert positive influences on youth employment in transitional economies. In China, for instance, Ou and Zhao (2016) suggested that decreasing unemployment rates among males and high school graduates in China were attributed to the expansion of higher education. Kyui (2016) also reported a similar trend in Russia, albeit, a phenomenon of decreasing returns to higher education has appeared. In response to this concern, there have been policy proposals that promote vocational education instead of higher education and that focus on the school-to-work transition, arguing that vocational education provides students with employable skills and therefore prepares them better for work. Hanushek et al. (2017), however, warned that given technological changes in exponential rate, the gains today in youth employment “may be offset by less adaptability and diminished employment later in life”. Noticeably, studies of youth employment have to rise above using only unemployment rates as a compass to investigate youth employment problems (Rees 1986). Sortheix, Chow and Salmela-Aro (2015) suggested the inclusion of work values as they play a key motivational role in job selection and career development. Mortimer et al. (2008) argues studies have to take into account a delayed period for university graduates to go from “survival” jobs to “career” jobs that match their training at college level, which may result in a level of underemployment. Scurry and Blenkinsopp (2011) 635
6. suggests that research that try to understand graduate underemployment may benefit from relevant theoretical frameworks from career studies. Studies on youth employment in Vietnam are sparse. Pierre (2012) suggested that the youth labour market was dynamic and outcomes for youths had improved, but the overall quality of employment had suffered due to global crisis and economic volatility. Globalization and industrialization is found to not do any better to young, female workers who live on meagre earnings and below standard working conditions (Tran and Nørlund 2015). It is argued that there is mismatch between supply of higher education and labour market demand in Vietnam, as the unemployment rate among the high-skilled cohort is much higher than the average level (Nguyen and Chu 2014). Much remains unknown about multi-dimensional labour market outcomes of recent graduates in an economy characterised with low levels of formality and social protection, notwithstanding, over the past 30 years, Vietnam has experienced noticeable shifts of employment from informality (Schmillen and Packard 2016). 3. Higher education expansion and youth employment in Vietnam Higher education system in Vietnam has witnessed a period of considerable growth due to significant changes in the government’s policy in the last three decades. Before its economic innovation in 1986, a very small proportion of the tertiary-age population went to college. Those university-educated persons had been highly subsidised by the government during their studies as they would work for the public sector after graduation. After 1986, the government continued to be the single payer for university studies until the early 1990s when it could no longer afford to meet the increasing demand for colleges. As such, the government started to allow public universities to admit more students who did not pass the entrance exam yet were able to pay for their own studies, known as “the open-track students”, in addition to those who passed the exam and would be paid for by the government. Simultaneously, the government cautiously allowed few private universities to open, starting with vocationally-focused colleges. The expansion in terms of size and types of service providers has largely contributed to an increase in enrolment at tertiary level. In 1990, only 2.8 percent of tertiary-age population went to higher educational institutions. The figure rose to 9.7 percent in 2000 and then 22 percent in 2010. The number of students in the total population in 2011 was 245 students per 10,000 persons, ten times higher than those in 1987. The government of Vietnam is ambitious in getting more college-educated persons in the hope that this labour force will be a primary driver of the country’s economic growth. As one of the most dynamic emerging economies in the world, Vietnam has dramatically shifted from agriculture towards manufacturing and 636
7. services. In 2015, less than 20 per cent of the country’s GDP derives from agriculture compared with 33 per cent from manufacturing and construction and 40 per cent from services (GSO 2015). Embracing an industrialised economy, Vietnam has to rely on highly-skilled human resources for growth. Accordingly, the government has encouraged further expansion of the higher education system. In 2005, the government issued the Higher Education Reform Agenda 2005-2020 (HERA) that set a vision for the higher education system for the next 15 years. One of the five goals of HERA are “continued rapid expansion of the higher education system that aims a rate of higher education participation by 2020 that is three to four times higher than the current level”. In Vietnam, it has been widely circulated that there is over-supply of higher education, and there is a mismatch between supply and demand in the labour market (Nguyen and Chu 2014). Recent labour statistics from the Labour Force Survey, which was conducted by Vietnam General Statistics Office (GSO), shows that, by the end of 2016, undergraduates and postgraduates accounted for the most part of unemployment in the cohort of skilled worker. Some even called for a shift in public investment from college and university to vocational schools, arguing that studies at this level is much less expensive and those who study at this level have a higher chance to get a job upon graduation compared to university-educated group (Tran 2016). Furthermore, a large fraction of the labour force worked in the informal sector, without social insurance (Huong et al. 2013). The majorities of informal employment are self-employment and household businesses. Except for some successful entrepreneurs, most informal employers earn poorly and are not covered by social security. Formal contracts rarely exist. Such economy often favours those who spend less years on education than those who spend more. In short, the debates on the employability of recent graduates with college and university education seem to reside in over supply and the mere stance of employment status, little has been known about the multi-dimensional labour market outcomes of this graduate cohort. This is even more acute in a transition country like Vietnam that is characterised with underemployment, large informal sectors, and low employment protection. 4. The data and empirical strategies The data This study uses the 2015 Labour Force Survey (LFS 2015) conducted by the General Statistics Office. LFS is an annual survey that collects individual information on the labour market across all 63 provinces/cities in Vietnam. Since 2011, the survey has been conducted on a quarterly basis to provide timely updates to changes in the 637
8. labour market. The survey is stratified into household where demographic information and related labour market outcomes of all individuals in a given household are collected using questionnaire-based interviews. It targets not only people in the labour force, the employed and unemployed, but also those not actively participating in economic activities, such as students, housewives and the retirees. Labour market outcomes indicators such as employment status, level of employment, social security, are measured against these groups, which makes this survey suitable for the research purpose of this study. Annually, there are more than 800,000 individuals taken part in this survey, with equal proportion in each of the four quarters. The sample selection method is two-stage stratification in order to ensure representativeness at regional level and provincial level. Stage one is to select enumeration areas - EAs, and based on the result of stage one, households are selected in stage two. Based on the 2014 Inter-censual Population and Housing Survey’s selected enumeration areas, a list of provincial enumeration areas is established. EAs are selected by the method of probability proportional to size (PPS). At stage two, a list of households at each abovementioned EA is selected on the basis of the upper/first and the lower/second half of the list. Fifteen households will then be selected from each half. In doing so, the sample represents different economic regions, gender, and age groups, and therefore ensures the validity and reliability of the data as well as its validity to make interferences for the population. In order to improve the design efficiency and ensure the reliability of survey sample, the households in each EAs are rotationally selected following the 2-2-2 rule. Households will be selected into sample in two adjacent quarters, excluded in the two following quarters, and selected again in the next two adjacent quarters. As such, each EA can be selected into the sample four times per year at most. Following the practice of the work transition surveys conducted by the International Labour Organization, this paper chooses the respondents aged from 15 to 29 in the labour force survey conducted in 2015. This age cohort provides ample time to capture diverse labour market outcomes of recent graduates upon graduation. The paper categorises three groups based on their educational attainment: those with high school graduates (tot nghiep pho thong co so), those with college graduates (tot nghiep cao dang) and those with university graduates and postgraduates (tot nghiep dai hoc va sau dai hoc). The study will compare labour market outcomes of the first group, or control group, with those of the other two. These three groups make up for 62,694 individuals nationwide. Among this total, high school graduates account for 66 per cent, college graduates 13 per cent, and university graduates 21 per cent. 638
9. Table 1: Personal and demographic information of the recent graduates No. of Standard Variable Mean Min Max observations deviation Gender (Female = 0, Male = 1) 62,694 0.473 0.499 0 1 Age (years old) 62,694 23 3 15 29 Marital status (Having married 62,661 0.363 0.481 0 1 = 1, 0 otherwise) Work experience (year cohorts) 42,237 2.967 0.880 1 5 Household head (head of 62,694 0.086 0.280 0 1 family = 1, 0 otherwise) Number of dependent persons 58,610 2.211 1.447 0 14 Source: Authors’ calculations from the Labour Force Survey 2015. Table 1 above show some personal and demographic information of the recent graduates in 2015. Among them, 47 percent were males, 36 percent were married, and 9 percent became household heads. On average, they lived in households with 2.2 economically dependent persons, but, in some few circumstances, they have up to 14 dependent persons. Their work experience ranges in the scale from 1 to 5, with the averaged scale, 2.97, indicating that most of them have the number of work experience from 1 to 5 years. Table 2: Snapshot of labour market outcomes of the recent graduates No. of Standard Variable Mean Min Max observations deviation Employment status (having 44,138 0.968 0.176 0 1 employed = 1, 0 otherwise) Ln(Wage) 33,736 8.204 0.635 3.912 13.312 Extra work demand (having 42,271 0.048 0.214 0 1 demand = 1, 0 otherwise) Having permanent labour contract 27,687 0.708 0.455 0 1 (having contract = 1, 0 otherwise) Having social insurance (having 42,903 0.434 0.496 0 1 insurance = 1, 0 otherwise) Working in entities with business registration (having registration = 38,912 0.732 0.443 0 1 1, 0 otherwise) Source: Authors’ calculations from the Labour Force Survey 2015. 639
10. Table 2 indicates snapshot of labour market outcomes of the recent graduates, ranging from employment status (having employed or not) to wage income, level of employment, and employment protection. On average, 97 percent of the recent graduates were employed, with wage of 8.2 in the natural logarithmic scale. However, not all of the employed satisfied with the current level of employment; namely, five percent expressed the need to seek additional jobs. With respect to employment protection, less than half of the recent graduates had social insurance, and about 70 percent had permanent labour contracts and worked in entities with formal business registration. Empirical strategies The labour market outcomes of recent graduates are estimated with the following regression equation (1). To control for selection bias, the following selection equation (2) is used. The regression equation: 푖 = 훼 + 훽. 표푙푙푒𝑔푒푖 + 훾. 푛𝑖푣푒 푠𝑖푡 푖 + 훿. 푿푖 + 휎풑 + 휃푖 (1) Where: 푖 is the labour market outcomes of individual 𝑖 in province , which could be either employment status or the other multi-dimensional outcomes once being employed. The latter comprises of wages, demand for extra works (underemployment or not), having permanent contracts, social insurance, and working in entities with formal business registration. 표푙푙푒𝑔푒푖 is a dummy variable, taking value of 1 if individual 𝑖 in province has the highest level of college education, and 0 otherwise. College education is more skill-oriented than the university level, and takes three years to accomplish in Vietnam. 푛𝑖푣푒 푠𝑖푡 푖 is a dummy variable, taking value of 1 if individual 𝑖 in province has the highest levels of undergraduate or postgraduate education, and 0 otherwise. University education normally takes four years to accomplish, and postgraduate education normally take 2 years to accomplish. 푿푖 is a vector of other covariates of individual 𝑖 in province that might affect his or her labour market outcomes. These include demographic variables like gender, age, marital status, working experience, whether a given individual is 640
11. the household head, and the number of economically dependent individuals in the household. The number of economically dependent individuals in a given household is counted by the number of people not taking part in any economic activities. They could be children, the elderly, non-paid housewives or handicapped people in the households. 휎풑 is unobservabble province-specific effects, which denote for differed economic and geographic conditions across provinces. The inclusion of these effects is important, as the labour market outcomes of a given individual is differed greatly across provinces due to socio-economic conditions. 훼, 훽, 훾, 휹 are the coefficients to be estimated. The key coefficients of interest are 훽 and 훾, which represent the effects of college and university education on the labour market outcomes, with reference to the group of higher secondary education. 휃푖 is the usual random error, which is independently and identically distributed. The selection equation 풁푖 + 휏푖 > 0 (2) Where: 풁푖 is a vector of covariates that might guarantee the labour market outcome of individual 𝑖 in province is observed or not. In this paper, we include gender, marital status, and the number of economically dependent people in a given household as the covariates of interest. In one classic example, women might not choose to work when they are married, and prefer to do household works. Also, the number of economically dependent variables might represent economic burden constraining the motivation to work of a given individual. The inclusion of the selection equation is important when an individual’s work status is not made randomly due to unobserved heterogeneity. 휏푖 is the usual random error, which is independently and identically distributed. The labour market outcomes in equation (1) is not always observed. Rather, it is observed if equation (2) holds. The ordinary regression for the labour market outcomes in equation (1) is biased if corr(휃푖 , 휏푖 ) ≠ 0. Given corr(휃푖 , 휏푖 ) ≠ 0, the above Heckman procedure provides consistent, asymptotically efficient estimates for all the parameters in such models. For the dependent variable of wages, equation (1) is estimated with the standard OLS estimator. For the other dependent variables 641
12. related to employment status, level of employment, and employment protection, equation (1) is estimated with the probit estimator. 5. Results and discussion Table 3 delivers the estimated results of the labour market outcomes of recent graduates in 2015. The variables of interest are 표푙푙푒𝑔푒 and 푛𝑖푣푒 푠𝑖푡 , whose estimated coefficients account for the impacts of educational levels on the labour market outcomes. All the specifications from (1) to (11) control for the other demographic and individual variables that might affect the labour market outcomes. They include 𝑔푒푛 푒 , 𝑔푒, 𝑖푡 푙 푠푡 푡 푠, 푤표 푒 푒 𝑖푒푛 푒, ℎ표 푠푒ℎ표푙 ℎ푒 , and 푒 푒푛 푒푛푡 푒 푠표푛푠. In addition, all the specifications control for unobservable province-specific effects. The coefficients are all reported as the average marginal effects, and they could be interpreted as the probability gaps with respect to the specifications with binary outcomes, (1) and (4) to (11). In specification (1), the probability of being employed does not differ greatly among college and university graduates, as the estimated coefficients for 표푙푙푒𝑔푒 and 푛𝑖푣푒 푠𝑖푡 are of the same magnitude. However, the positive estimates of these coefficients indicate that the probability of being employed for college and university graduates is slightly higher than that of high school graduates. 642
13. Table 3: Education attainment and labour market outcomes (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Permanent Permanent Dependent variable: Work extra work extra work Social Social Business Business Ln(wage) Ln(wage) labour labour status demand demand insurance insurance registration registration contract contract College education 0.003 0.120 0.119 -0.008 -0.010 0.174 0.176 0.239 0.231 0.189 0.188 (0.001) (0.022) (0.022) (0.003) (0.004) (0.012) (0.012) (0.013) (0.013) (0.009) (0.009) University education 0.003 0.262 0.262 -0.018 -0.021 0.271 0.275 0.348 0.335 0.268 0.267 (0.001) (0.021) (0.021) (0.004) (0.005) (0.010) (0.010) (0.013) (0.013) (0.013) (0.013) Gender 0.001 0.160 0.159 0.009 -0.005 -0.080 -0.079 -0.079 -0.070 -0.057 -0.049 (0.001) (0.014) (0.014) (0.003) (0.005) (0.009) (0.010) (0.009) (0.010) (0.008) (0.008) Age 0.000 0.034 0.034 -0.002 -0.002 0.006 0.006 0.015 0.014 0.013 0.013 (0.000) (0.004) (0.004) (0.000) (0.001) (0.001) (0.001) (0.002) (0.002) (0.001) (0.001) Marital status -0.002 0.009 0.010 -0.015 -0.027 0.048 0.053 0.034 0.038 -0.008 -0.003 (0.001) (0.012) (0.012) (0.003) (0.005) (0.007) (0.008) (0.009) (0.010) (0.008) (0.009) Work experience 0.000 0.030 0.030 -0.010 -0.012 0.138 0.141 0.068 0.066 -0.025 -0.024 (0.000) (0.020) (0.020) (0.007) (0.002) (0.005) (0.005) (0.008) (0.007) (0.006) (0.006) Household head 0.000 0.025 0.025 0.007 0.010 0.006 0.005 -0.031 -0.032 -0.018 -0.021 (0.002) (0.026) (0.026) (0.006) (0.007) (0.014) (0.015) (0.013) (0.013) (0.013) (0.013) Dependent persons -0.002 0.000 -0.001 0.006 0.027 -0.008 -0.019 -0.047 -0.055 -0.037 -0.051 (0.000) (0.005) (0.006) (0.001) (0.004) (0.003) (0.009) (0.003) (0.003) (0.003) (0.003) Province-fixed effect Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Control for selection bias Yes Yes Yes Yes Yes Number of observations 30,245 33,646 36,816 41,568 44,699 27,611 30,800 41,558 44,689 38,374 41,515 R-squared 8.2%a 23.1% 9.0%a 22.0%a 24.4%a 18.5%a Wald test of independent equations (rho = 0) 0.249 0.000 0.1414 0.0007 0.000 Notes: Robust standard errors in parentheses. Standard errors are robust to the cross-sectional dependence within provinces. p<0.01, p<0.05, * p<0.1. a Pseudo R-squared. 643
14. In the specifications from (2) to (11), each indicator of labour market outcome is first estimated with the equation (1) in Section 4 above, and then equation (2) that controls for the selection bias. In the last row of Table 3, the Wald tests of independent equations show that specifications (5), (9), and (11) are more appropriate in determining the labour market outcomes of recent graduates. These tests indicate that the selection biases are inherent with the conventional OLS estimators for the dependent variables of extra work demand, having social insurance, and formal business registration. The Heckman two-step procedure thus yield unbiased estimates for these dependent variables. Although college and university graduates do not differ markedly to high school graduates in terms of work status, they differ remarkably in terms of wages. The OLS estimator for the monthly wage in specification (2) shows that college graduates have 12 percent higher wages than the reference group of high school graduates. Similarly, university graduates have 26 percent higher wages than high school graduates. These differed wages are basically of the same magnitude in specification (3), which controls for selection bias with the Heckman procedure. The demand for extra work while being employed might indicate the level of underemployment among recent graduates. The negative estimated coefficients for 표푙푙푒𝑔푒 and 푛𝑖푣푒 푠𝑖푡 in the specifications (4) and (5) show that these cohorts demand for less extra work while being employed. Among them, university graduates demand for even less extra works than college graduates. Specifically, in specification (5), the university and college graduates have respectively 2.1 percent and 1 percent lower probability of extra work demand than the reference group of high school graduates. The other control variables in these two specifications also indicate some interesting results about the level of underemployment among recent graduates. Specifically, male graduates and those with more economically dependent person demand for more extra works, while age and marital status are negatively correlated with the demand for extra work. The finding that the number of economically dependent persons induces demand for extra works is particularly interesting, as this might indicate the level of economic burden encountered by recent graduates in a given household. For economic reasons, recent graduates would have higher motivation of looking for extra works if their families are more economically disadvantageous. Besides, the negative impact of age on the demand for extra work is understandable given graduates’ more stable and transited jobs over time. Specifications from (6) to (11) in Table 3 further indicate the impacts of college and university education on the labour market outcomes of recent graduates. The estimated coefficients for these variables are much greater in magnitude than those appeared in the previous specifications. The probability of obtaining permanent contracts, having social insurance, and being employed in establishments with formal 644
15. business registration is much higher than that of high school graduates. Among recent graduates with college and university education, the former has smaller probability of attaining the just-mentioned labour market outcomes after being employed. For example, the probability of university graduates obtaining permanent labour contracts is 27 percent more than that of high school graduates, which is higher than the similar gap between college and high school graduates, 17 percent. All of the control variables related to demographic and individual characteristics have sign impacts as expected, except for gender and the number of economically dependent persons. For example, the persistently negative estimates coefficients for gender might indicate the more readiness of male graduates in working in the informal sectors that do not offer permanent contracts, social insurance, and not have business registration. However, it is cautious to yield concrete conclusions on the gender effects on these labour market outcomes, as we could not completely control for the unobserved heterogeneity that is correlated with gender and potentially affect his or her performances in the labour market. In short, the empirical results in Table 3 reiterates high-quality labour market outcomes achieved by recent graduates with college and university education, compared to high school counterparts. Although these three cohorts of graduates do not defer remarkably in terms of employment status, having employed or not, college and university graduates have relatively higher wages, better employment protection than high school counterparts. Specifically, the former has higher chances of working in formal sectors, and of securing a job with permanent labour contract and social insurance. By studying these multi-dimensional educational effects on recent graduates, this paper provides a more balanced view on the labour market outcomes of recent graduates in a developing country. It gives out evidence against recent debates on very high unemployment rates among high-skilled employees with college and university education. Higher education thus benefits recent graduates in terms of high-quality jobs, not just decent ones. The challenge confronting college and university graduates is not finding a decent job, but the one that could guarantee them with better employment projection that is crucial for their smoother school-to- work transition. 6. Conclusion This study examines the labour market outcomes of recent college and university graduates in reference to high school counterparts. It not only looks into graduates’ employment status, being employed or not, but also other multi- dimensional labour market outcomes after being employed; namely, monthly wages, level of employment, the chances of having permanent labour contracts, social insurance, and of working in establishments with formal business registration. The 645
16. empirical findings show that college and university graduates outperform high school counterparts in terms of these high-quality labour market outcomes, though the chance of being employed does not differ markedly among these graduate cohorts. These findings remain unchanged even after having controlled for selection bias that is inherent in the labour market studies. The paper would contribute to recent debates in higher education in the transition economies like Vietnam, where the focus is on whether the move towards mass higher education might result in differed labour market outcomes for recent graduates. This paper provides concrete evidences for a better school-to-work transition for recent graduates, where the high-quality labour market outcomes play a more important role than the mere stance of being employed or not. A number of venues for further studies could be departed from this study. Among them is the impact of over education on skill utilization, and scarring effects in the labour markets of a developing country like Vietnam. Acknowledgements This research is funded by Vietnam National Foundation for Science and Technology Development (NAFOSTED) under grant number 502.01-2018.12. It is also under the research grant No. KTQD/E2018.02 provided National Economics University. The authors would like to thank the Foundation and the University for this support. References 1. Bai, Limin. “Graduate Unemployment: Dilemmas and Challenges in China's Move to Mass Higher Education”. The China Quarterly no. 185 (2006): 128-144. 2. Berliner, Joseph S. “Education, Labour-force Participation, and Fertility in the USSR”. Journal of comparative economics no. 7 (1983): 131-157. 3. Breusch, Trevor and Edith Gray. “New Estimates of Mothers Forgone Earnings Using HILDA data”. Australian Journal of Labour Economics no. 7 (2004): 125-150. 4. Cain, Glen G. Married Women in the Labour Force: An Economic Analysis. Chicago: University of Chicago Press, 1966. 5. Carroll, David and Massimiliano Tani. “Over-education of Recent Higher Education Graduates: New Australian Panel Evidence”. Economics of Education Review no. 32 (2013): 207-218. 6. Cling, Jean, Mireille Razafindrakoto and François Roubaud. The Informal Economy in Vietnam. Hanoi: International Labour Office, 2011. 646
17. 7. Cook, Sarah and Jonathan Pincus. “Poverty, Inequality and Social Protection in Southeast Asia: An Introduction”. Journal of Southeast Asian Economies no. 31 (2014): 1-17. 8. Elder, Sara and Sriani Kring. Young and Female – A Double Strike? Gender Analysis of School-to-Work Transition Surveys in 32 Developing Economies. Geneva: International Labour Office, 2016. 9. Faridi, Muhammad Zahir, Imran Chaudhry and A. B Basit. “An Analysis of the Determinants of Male Labour Force Participation and Employment Status in Pakistan: The Case of Bahawalpur District”. Pakistan Journal of Social Sciences no. 29 (2010): 189-208. 10. GSO. Statistical Yearbook of Vietnam 2015. Hanoi: General Statistics Office, 2015. 11. GSO. Statistical Handbook of Vietnam 2016. Hanoi: General Statistics Office, 2016. 12. Hanushek, Eric A., Guido Schwerdt, Ludger Woessmann and Lei Zhang. “General Education, Vocational Education, and Labour-Market Outcomes over the Lifecycle”. Journal of Human Resources no. 52 (2017): 48-87. 13. Nguyen, Thi Lan Huong, Luu Quan Tuan, Matthias Meissner, Bui Sy Tuan, Dang Do Quyen and Nguyen Hai Yen. “Social Protection for the Informal Sector and the Informally Employed in Vietnam: Literature and Data Review”. IEE Working Paper no. 199. Institut fuer Entwicklungsforschung und Entwicklungspolitik, Ruhr-Universitaet Bochum, 2013. 14. ILO. Key Indicators of the Labour Market. Geneva: International Labour Office, 2015. 15. Institute of Labour and Social Studies and International Labour Organization. The Employment and Social Trends in Vietnam During 2012-2017. Hanoi: The Youth’s Publishing House, 2018. 16. Kyui, Natalia. Expansion of Higher Education, Employment and Wages: Evidence from the Russian Transition. Labour Economics no. 39 (2016): 68-87. 17. Jaumotte, Florence. “Female Labour Force Participation: Past Trends and Main Determinants in OECD Countries”. OECD Working Paper no. 376. Paris: Organisation for Economic Cooperation and Development, 2003. 18. Kennedy, Steven and David Hedley. “A Note on Educational Attainment and Labour Force Participation in Australia”. Australian Treasury Working Paper no. 3, 2003. 19. Kennedy, Steven, Nicolas Store and Leo Vance. “Labour Force Participation and the Influence of Educational Attainment”. Australia: The National Centre for Vocational Education Research, 2009. 647
18. 20. Laplagne, Patrick, Maurice Glover, and Anthony Shomos. “Effects of Health and Education on Labour Force Participation”. Australian Government Productivity Commission Working Paper, 2007. 21. Lattimore, Ralph. “Men Not at Work: An Analysis of Men Outside the Labour Force”. Australian Government Productivity Commission Working Paper, 2007. 22. Leamer, Edward. The Heckscher-Ohlin Model in Theory and Practice. Princeton: Princeton University Press, 1995. 23. Li, Shi, John Whalley and Chunbing Xing. “China's Higher Education Expansion and Unemployment of College Graduates”. China Economic Review no. 30 (2014): 567-582. 24. Mavromaras, Kostas, Seamus McGuinness, Nigel O'Leary, Peter Sloane, and Zhang Wei. “Job Mismatches and Labour Market Outcomes: Panel Evidence on University Graduates”. Economic Record no. 89 (2013): 382-395. 25. McGuinness, Seamus. “Overeducation in the Labour Market”. Journal of Economic Surveys no. 20 (2006): 387-418. 26. Mortimer, Jeylan T., Jeremy Staff, Mike Vuolo, Sara Wakefield and Wanling Xie. “Tracing the Timing of “Career” Acquisition in a Contemporary Youth Cohort”. Work and Occupations no. 35 (2008): 44-84. 27. Nguyen, Ba Ngoc and Chu Thu Lan. “Thi Truong Lao Dong Chuyen Mon Ky Thuat Trinh Do Cao o Viet Nam” [High-skilled Labour Markets in Vietnam]. Journal of Economics and Development, no. 201 (2014): 72-78. 28. Ou, Dongshu and Zhong Zhao. “Higher Education Expansion and Labour Market Outcomes for Young College Graduates”. Discussion Paper no. 9643. German: Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labour, 2016. 29. Pierre, Gaëlle. “Recent Labour Market Performance in Vietnam Through a Gender Lens”. World Bank Policy Research Working Paper no. 6056. Washington, 2012. 30. Quintini, Glenda, John P. Martin and Sébastien Martin. “The Changing Nature of the School-to-Work Transition Process in OECD Countries”. Discussion Paper no. 2582. German: Forschungsinstitut zur Zukunft der Arbeit Institute for the Study of Labour, 2007. 648
19. 31. Rees, Albert. “An Essay on Youth Joblessness”. Journal of Economic Literature no. 24 (1986): 613-628. 32. Ryan, Paul. “The School-to-Work Transition: A Cross-national Perspective”. Journal of economic literature no. 39 (2001): 34-92. 33. Schmillen, Achim Daniel and Truman G. Packard. “Vietnam's Labour Market Institutions, Regulations, and Interventions: Helping People Grasp Work Opportunities in a Risky World”. World Bank Policy Research Working Paper. Vietnam, 2016. 34. Scurry, Tracy and John Blenkinsopp. “Under-employment Among Recent Graduates: A Review of the Literature”. Personnel Review no. 40 (2011): 643-659. 35. Sortheix, Florencia, Angela Chow and Katariina Salmela-Aro. “Work Values and the Transition to Work Life: A Longitudinal Study”. Journal of Vocational Behavior no. 89 (2015): 162-171. 36. Standing, Guy. Labour Force Participation and Development. Geneva: International Labour Office, 1981. 37. Stolper, Woflgang and Paul A. Samuelson. “Protection and Real Wages”. Review of Economic Studies no. 9 (1941): 58-73. 38. Tomlinson, Michael. “The Degree Is Not Enough: Students’ Perceptions of the Role of Higher Education Credentials for Graduate Work and Employability”. British Journal of Sociology of Education no. 29 (2008): 49-61. 39. Tran, Angie Ngoc and Irene Nørlund. “Globalisation, Industralisation, and Labour Markets in Vietnam”. Journal of Asia Pacific Economy no. 20 (2014): 143-163. 40. Tran, Thi Tuyet. “Youth Transition to Employment in Vietnam: A Vulnerable Path”. Journal of Education and Work, no. 31 (2017): 59-71. 41. UNESCO. Global Education Digest 2007. Montreal: The United Nations Educational, Scientific and Cultural Organization, 2007. 42. World Bank. Vietnam: Higher Education and Skills for Growth. Washington DC: The World Bank, 2008. 43. World Bank. World Development Indicator. 2016. . 44. Yu, Yanmin. “Higher Education Expansion in China and its Impacts on the Labour Market Outcomes of College Graduates”. Master’s thesis, Department of Economics, Dalhousie University, Canada, 2014. 649