Changes in Wage Differentials among College Graduates in , 1999-2008 (Eunmi Ko) 103

勞 動 經 濟 論 集 第34卷(1), 2011. 04, pp.103~138 ⓒ 韓 國 勞 動 經 濟 學 會

Changes in Wage Differentials among College Graduates in South Korea, 1999-2008*

1)

Eunmi Ko**

This paper examines the changes in relative wage of top 10 college graduates to the other college graduates among the age group of 26-28 years using Korean Labor and Income Panel Study (KLIPS). From 1999 to 2008, the wage differential between top 10 college graduates and the other college graduates increased in South Korea. This wage differential seems to persist along with their age. Within industry wage differential among college graduates also rose but in the late 2000s it became smaller than the wage differential within firm size and industry. Increase in elite college wage premium has led to recent changes in college wage premium.

- Key words: elite college wage premium, college graduates, wage differentials, relative demand for skilled workers

논문 접수일 : 2011년 2월 16일, 논문 수정일 : 2011년 4월 8일, 논문 게재확정일 : 2011년 4월 12일 * This paper is based upon my M.A. thesis at Seoul National University. I heartily thank Professor Chulhee Lee and two anonymous reviewers for helpful comments and suggestions. I am grateful to Professor Daeil Kim, Professor Bonggeun Kim, the commentators of the graduate student session in KEA joint conference and participants in the Seoul National University — University of Wollongong joint workshop. Also I am thankful to Professer Soyoung Kim and Professor Jae-Young Kim for encouragement. All remaining errors are my own. Financial support was provided by Brain Korea 21 program. ** Ph.D. candidate, Department of Economics, Seoul National University, Seoul, South Korea. ([email protected]) 104 勞動經濟論集 第34卷 第1號

Ⅰ. Introduction

Changes in the wage differentials, particularly focusing on the relative wage of college and high school graduates, and the fundamental causes of these changes have been studied by numerous researchers in South Korea.1) It is well documented that the college wage premium in South Korea decreased from the 1980s to the mid-1990s and subsequently began increasing from the late 1990s.2) In other words, the wage differential between college and high school graduates fell and turned to rise during the late twentieth century, thus forming a U-shaped curve. Many researchers have explained these changes in the framework of relative supply of and relative demand for college graduates as more skilled workers. More rapid growth in the relative demand for college graduates than relative supply of them has been found to be the primary factor for the increase in the college wage premium in the 2000s.3) However, it has been reported that since the mid-1990s, most of the high school graduates in South Korea entered colleges (see Table 1). Owing to this sharp increase in the college entrance rate, in the late 2000s, approximately 80 percent of the young

1) The extensive research on the fluctuations in college wage premium -the wage differential between college graduates and high school graduates- in the United States appears to have influenced these studies in South Korea. However, the economic environments of the two countries are different. For example, according to the Ministry of Strategy and Finance of South Korea, South Korea’s dependence on foreign trade was 82 percent in 2009. Therefore, the discussions for South Korea are distinguished from that of the United States. One class of South Korean researchers believes that globalization (particularly, extensive trade with China), rather than skill-biased technological change (STBC), was the primary reason for the recent increase in the college wage premium in South Korea. For more details, see section II of this paper. 2) See Choi and Jinho Jeong (2003), Choi and Jinook Jeong (2005), Shin (2007), and Wooyoung Kim (2008). It appears that there continue to be disagreements regarding the precise turning point when college wage premium began increasing during the 1990s. 3) Numerous explanations for the fundamental causes of these shifts are suggested in the studies conducted in South Korea as well as in the United States. Section II briefly presents these explanations of preceding research. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 105 potential workers with 1- or 2- years of experience were college graduates (including 2-year college graduates) and approximately 35 percent were 4-year college graduates (see Table A.1 and Table A.2). Reflecting this increase in the supply of college graduates, South Korea’s labor market of young workers was almost entirely composed of college graduates in the 2000s.4) Based on this phenomenon, it seems that different mechanisms, such as “whether graduating from elite college or not,” must operate in a labor market of young workers in order to distinguish the “more skilled” from the “less skilled” among college graduates who recently enter the labor market; thus leading to increasing the wage differentials among college graduates. In other words, for the young workers who enter the labor market, it is hypothesized that upward shifts in the relative demand for “more skilled and more educated” young workers favor only the elite college graduates and do not influence the other (non-elite) college graduates. Although there are numerous studies on the changes in college wage premium and some studies on the wage differential between elite college and the other college graduates in the view of cross-section, there exists no research on the changes in the wage premium of elite colleges in the view of time series5) and no study that focuses on the relative demand for elite college graduates. The limitations of the data might contribute to keeping researchers from the further studies about the wage differentials among college graduates. In this paper, I made the data which is comparable to

4) This kind of labor market for young workers, however, is not the case in the United States. According to the US Census Bureau, the proportion of 2-year college graduates and the higher educated in the United States was approximately 40 percent of the total population in the age group of 25-29 years. See “Educational Attainment in the United States: 2009 Current Population Survey” (http://www.census.gov/hhes/socdemo/education/data/cps/2009/tables.html). 5) The concept of elite college wage premium indicates the wage differential between the elite college graduates and the other college graduates in same industry and firm size. There are a variety of terms, such as “wage premiums for elite colleges” (Taejong Kim, 2004), “returns to university rank” (Jang, 2006), and other expressions, that have previously been used for representing similar wage differentials. Since “college wage premium” implies the wage differential between college and high school graduates, it is assumed that the term “elite college wage premium” is appropriate for expressing the elite and the other college graduate wage differential. 106 勞動經濟論集 第34卷 第1號 repeated cross-sectional one using longitudinal data and overcame the limitation; in detail, I took surveys from every three years (four times among ten year longitudinal survey data) and grouped individuals into the cohort of 26-28 years so that I could investigate different (not overlapped) cohorts in the each chosen (three year interval) survey. From this data it could be obtained that individuals' educational attainments, the college names who graduated from (if one did), and their real wages in order to analyze the wage differentials among workers who had nearly just graduated from their (4-year) colleges in the respective investigated year. First, I present that the elite college wage premium of young workers, who are in the age group of 26-28 years, increased from 1999 to 2008 in South Korea; the elite college wage premium for them rose from 3.82 percent to 23.0 percent during the 1999-2008 period. Second, I describe that the increase in the elite college wage premium were driven by the growth in the relative demand for elite college graduates

Changes in College Entrance Rates

College Entrance Rates 1995 51 1996 55 1997 60 1998 64 1999 67 2000 68 2001 71 2002 74 2003 80 2004 81 2005 82 2006 82 2007 83 2008 84 2009 82 * Source: Yearly Report for Statistic on Korean Education (retrieved from Center for Education Statistics [cesi.kedi.re.kr] on Nov. 2010) Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 107 by examining the relative wage of the top 10 college graduates to the top 14-50 college graduates.6) Third, I show that this relative wage rose along with their age. Moreover, it is indicated that the increase in the elite college wage premium played an important role in the rise of college wage premium throughout the 2000s. The paper is organized as follows: Section II documents the conventional results of the preceding research on the changes in the college wage premium in South Korea; they shows that the relative demand for the more skilled (more educated) have grown since the late 1990s. Section III describes the data from the Korean Labor and Income Panel Study (KLIPS) that has been used throughout the paper. Section IV presents the analyses of changes in the relative wages of top 10 college graduates in South Korea during the 1999-2008 period and indicates how the changes in the elite college wage premium lead to the increase in the college wage premium. Section V summarizes the conclusion.

II. An Overview of Preceding Research

In South Korea, a number of researchers have studied on the wage differentials by educational attainment. Kim and Topel (1995) suggested that the wage differentials between male college graduates and less educated male workers narrowed from the mid-1970s to the late 1980s. Using the relative demand and supply framework, Kim and Topel (1995) explained that this decrease resulted from the growth in the relative supply of college graduates in South Korea. In addition, Choi and Jeong (2005) indicated that while the wage differentials by education declined from 1983 to 1993, they turned to rise from 1993 to 2000. Moreover, they presented that skill-biased technological change widened the wage structure between male college and high school graduates from 1993 to 2000 in South Korea as it did in the United States in 1980s.

6) This helps in controlling the impact of the changes in the relative supply of elite college graduates. 108 勞動經濟論集 第34卷 第1號

Shin (2007) also illustrated that skill-biased technological change probably led to the increase in wage inequality, by decomposing the changes in the college wage premium. In contrast, Moon (2001) showed that changes in the trade pattern of South Korea throughout the 1990s, which induced the increase in the relative demand for more skilled workers, played a key role in widening the differences in the earnings between college graduates and high school graduates.7) In addition, Kim and Kang (2010) indicated that South Korea’s trade pattern with China largely explained the increases in the college wage premium, with complementing the skill-biased technological change hypothesis. In sum, these studies show that the college wage premium increased from the late 1990s in South Korea. It means that the relative demand for college graduates rose more sharply than the relative supply of them did. Fundamental causes of these shifts in relative demand for college graduates, however, seem to be more difficult to be found out in South Korea than in the United States because South Korea is a small open economy whose economic growth and labor demand heavily depend on net exports. Moreover, the fact that South Korea’s trade pattern has changed in recent decades (from less skilled to more skilled intense manufacturing) made the analysis of the fundamental causes even more difficult. Meanwhile, the wage structure among college graduates in South Korea has been studied from the perspective of returns to education with cross-sectional data. It was found that there is sizeable return on graduating from elite colleges. Using integrated data from 1998 to 2002, Jang (2006) indicated that the rate of return for students graduating from the top 5 colleges is approximately 24 percent more than for those graduating from the top 100 colleges. Moreover, Kim and Lee (2007) pointed out that meaningful wage differential exists between elite and the other college graduates who have the same underlying ability, which is measured by the College Scholastic Ability

7) However, Moon (2001) considered only the effect of changes in international trade on the fluctuations in wage inequality; thus the implications provided by him are limited. In fact, as long as I know, the analysis by the general equilibrium model was not conducted for the college wage premium in South Korea. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 109

Test (CSAT) score. They also suggested that this wage differential, which depending on the college rankings in South Korea, stem from not only the underlying ability of each college graduate but the educational quality, inside information obtained from alumni networks, and other factors. Although the changes in the wage differential between college and high school graduates have been extensively studied and the research on the wage inequality among college graduates in the view of cross-section has been conducted in South Korea, the changes in the wage differentials among college graduates, for example, the changes in the elite college wage premium throughout the 2000s, remain unknown. As

indicates, from the early 2000s, college graduates accounted for over 60 percent of the young workers who had recently entered the labor market. Rather than analyzing the wage differential between college and high school graduates, examining the wage differentials among college graduates appears to provide more plausible implications on the questions associated with relative demand for skills in the 2000s, such as what are fundamental causes of the increase in the relative demand for more educated and more skilled labor. Moreover, investigating the changes in the elite college wage premium and doing research on the causes of these changes might provide explanations regarding the underlying causes for the surges in shadow education8) in South Korea.

III. Data

The data used in this paper has been obtained from the Korean Labor and Income Panel Study (KLIPS); this survey began in 1998. KLIPS is a longitudinal study of representative Korean households in the urban areas of South Korea(excluding Jeju Island) and the individuals of those households. This survey has been conducted

8) “Shadow education” indicates private supplementary education. 110 勞動經濟論集 第34卷 第1號 annually since 1998 and was temporarily stopped in 2009. As this paper pursued to investigate the impact of the sharp increase in the number of college graduates since the early 2000s, the surveying time period of KLIPS corresponds to the purpose of this paper. KLIPS provides information regarding individuals’ employment such as the industry that the individuals are engaged in, their experience, average monthly wage, and regarding their secondary and territory education, for example, the names of the colleges from where the individuals have graduated.9) KLIPS provides the college names for all individuals only in the 1998 and 2001 surveys as well as in the 2000 additional survey. The college name information of individuals provided by every survey is limited to those who were newly included into the survey year. In order to identify the college names for the greater number of individuals, I merged all the KLIPS surveys that were conducted from 1998 to 2008 and assigned the college name codes to the respective personal IDs in each of the surveys. Of specific, if the college name for an individual was missing in any of the surveys, then it was replaced with the college name provided in the 1998 survey, the additional survey of 2000, or the 2001 survey for that personal ID. Priority was given to the last college name in each survey10) and the first reported college name among the three surveys (1998, 2000, and 2001)11). For identifying the elite colleges, only the top 10 colleges (total 13 colleges) were regarded as the elite colleges and the other colleges were regarded as the non-elite

9) The Korean Youth Panel also provides the college’s name; however, it has been surveyed only from 2001 to 2006, which is a much shorter time period (The observed individuals were changed from 2007 onwards in order to resolve the ageing problem since the purpose of the survey was to study “young” generations, who are in the age group of 15-29 years.). See http://survey.keis.or.kr/. Moreover, KLIPS provides richer information than that provided by the Korean Youth Panel. 10) In KLIPS, individuals can provide up to 4 college names in order of the year of entering the colleges. 11) For example, if an individual who informed that she had graduated from Seoul National University in 1998 and reported different college information in 2001, then the Seoul National University code was considered as the college name code. Since workers who belonged to the age group of 26-28 years, the first graduating college was more important than any other educational information in this paper. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 111 colleges. This criterion was used for the analyses from 1999 to 2008. In addition, a control group comprising the top 14-50 colleges was established in order to further investigations.12) According to Jin-Yeong Kim (2006), there were few changes in the hierarchy of (approximately) top 50 colleges — precisely top 30 percent of the colleges — from 1994 to 2003; this result supports the method for determining the college groups for comparison in this paper. Furthermore, in the industry variable, in order to reduce the number of missing industry codes, I replaced them with that of the preceding or succeeding survey’s industry codes. The individuals who engaged in human health service and educational service were excluded because it was assumed that employed teachers and doctors earn similar wages whether they graduate from an elite college or not. The wage measure employed in this paper is the logarithm of the average monthly real wages, adjusted by the consumer price index, of full-time workers provided by each of the KLIPS surveys.13) The elite college wage premiums were analyzed every three years from 199914) among the workers in the age group of 26-28 years. It was found that the 3-year time period was relevant for maintaining an appropriate sample size of the top 10

12) The following are the “Top 10 colleges” or “Elite colleges”: Seoul National University, , , Sogang University, Sung Kyun Kwan University, , Chung Ang University, , Han Kuk University of Foreign Studies, University of Seoul, Ewha Woman’s University, Korea Advanced Institute of Science and Technology (KAIST), and Pohang University of Science and Technology (POSTECH); this is a total of 13 universities. For the top 14-50 colleges, see

. The top 10 colleges were categorized using the university ranking data provided by Dae-Sung educational institution in 1999, by Jinhak-Sa in 2008, and the top 10 college data from 1994 to 1997 that was provided by Jin-Yeong Kim (2006). Although there are various data sources and they appear to be dispersed, the hierarchy of the colleges, except one or two colleges, appear to be similar; all the data sources identified the same group of top 10 colleges. Moreover, in creating a group of top 50 colleges from each data source, except with respect to girls’ colleges, all the data sources surprisingly provided the same list of colleges. 13) Bank of Korea annually offers South Korea’s consumer price indices; the consumer price index (2005=100) data has been used in this paper. See http://ecos.bok.or.kr/ 14) Thus, 1999, 2002, 2005, and 2008 were the target years. 112 勞動經濟論集 第34卷 第1號 college graduates and observing changes in the elite college wage premium throughout the 2000s. Despite the fact that information from KLIPS was available since 1998, this study considered data only from 1999 to 2008 in order to avoid the adverse effects of the 1997-1998 currency and financial crisis in Korea. Moreover, workers belonging to the age group of 26-28 years were only selected to examine wage differentials in this study because it was assumed that workers who had recently entered the labor market would experience direct effects of the changes in the relative supply of more educated workers.15) Of specific, for 1999, the analyzed cohort constituted young workers who were born between 1972-1974. For 2002, 2005, and 2008, the cohorts were composed with workers who were born between 1975-1977, 1978-1980, and 1981-1983, respectively. In order to estimate the wage differentials, the Mincerian wage equation, which was also employed in Kim and Kang (2010) as well as in numerous other studies, was used throughout this paper. The basic model that was employed in this paper is as follows:

                   

In this Mincerian equation, the  vector includes gender, experience, squared experience, age, squared age, dummy variable for engaged industry, and firm size dummy variable. The dummy variables for industry has the following 4 levels: relative high-technology manufacturing, relative low-technology manufacturing,16) service industry, and others. The dummy variable for firm size comprises the following 10

15) This paper focused on young male workers. Since 2-2.5 years of military service is mandatory for every South Korean man, almost all male workers who belong to the age group of 26-28 years are those workers who have entered Korea’s labor market recently. 16) The taxonomy of industry, which is associated with the relative technology level, is based on the Analysis on Business that has been published by Bank of Korea. See http://ecos.bok.or.kr. The cutting-edge technology, high-technology, and mid-technology industries have been included in the “relatively high-technology industry” classification. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 113 levels by the number of employees in the firms: 1-4, 5-9, 10-29, 30-49, 50-69,

70-99, 100-299, 300-499, 500-999, or over 1000 employees. The  represents an individual’s educational attainment, where high school graduates(k = 2)

17) being benchmarked. The coefficient of  indicates the average wage differential between non-elite college and high school graduates and that of  represents it between graduates from elite colleges and high schools.

IV. Results

1. Rising the Elite College Wage Premium During 1999- 2008

indicates the changes in the elite college wage premium. In 1999, elite college graduates earned 3.8 percent more than the non-elite college graduates who were employed in the same industries and firm sizes as the elite college graduates. The elite college wage premium presented rapid secular growth, and thus the young workers who had recently entered the labor market after graduating from elite colleges earned approximately 23 percent more than non-elite college graduates in 2008. This increase in the elite college wage premium suggests that wage inequality among college graduates continued to rise throughout the 2000s. In order to explain this phenomenon, it may be hypothesized that the rising relative demand for skills may lead to the increase in this wage differential. Furthermore, according to the discussion regarding college wage premium, the fundamental causes for this elite college wage premium might be related with net exports to newly industrialized economies such as China or skill-biased technological change.

17) For middle school graduates and less educated people, k = 1. For two-year college graduates, k = 3. For non-elite college graduates, k = 4 and for the elite (top 10) college graduates, k = 5. 114 勞動經濟論集 第34卷 第1號

Rising Elite College Wage Premium

(1) (2) (3) (4) 1999 2002 2005 2008 -0.0696 -0.103 -0.236 0.0469 Under High School Graduates (-0.93) (-1.06) (-2.11)* ( 0.37) 0.0624 0.0516 -0.0195 0.0934 2-Yr College Graduates (1.29) (1.16) (-0.39) (1.81) 0.205 0.146 0.180 0.252 Non-elite college College Graduates (4.10)*** (3.12)** (3.30)** (4.36)*** 0.251 0.270 0.417 0.541 Elite college College Graduates (2.91)** (3.21)** (4.64)*** (4.59)*** -0.282 -0.206 -0.253 -0.248 Sex (female) (-6.83)*** (-5.53)*** (-5.66)*** (-5.36)*** 50.160 -18.810 -30.650 26.350 Constant (1.83) (-0.69) (-1.00) (0.79) N 342 322 320 316 R-sq 0.317 0.229 0.335 0.328 adj. R-sq 0.272 0.175 0.288 0.280 Wage Premium for 4-yr Elite College 3.817 10.820 20.085 23.083 (%) Coefficients of dummy variables associated with age, experience and firm size are eliminated in this table. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001

2. Can Relative Supply Changes Explain the Increase in Elite College Wage Premium?

Nevertheless, another explanation for the increase in elite college wage premium can be provided. Excluding the possibility of the increase in the relative demand for skills, this rising elite college wage premium can be expected by considering the growth in relative supply for non-elite college graduates. The proportion of non-elite college graduates among young workers rapidly increased in the 2000s as a result of the deregulation on establishing colleges since 1996. As Jin-Yeong Kim (2006) illustrated, however, the hierarchy of the colleges and admission quota for elite colleges did not change. Therefore, the elite (top 10) college graduates became relatively rare in the Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 115 labor market of young workers whereas the supply of non-elite college graduates increased sharply. This might cause an increase in the relative wage of elite college graduates.18) In order to evaluate this relative supply side hypothesis, I expanded the basic model as follows:

                      

In this model,  stands for the top 10 college graduates,  denotes the top 14-50 college graduates, and  represents the college graduates from the remainder colleges. The impact of the decrease in the relative supply of graduates from elite colleges can be controlled by comparing the top 10 college graduates with the top 14-50 college graduates because the colleges that form the top 50 group hardly changed throughout the 1990s and the 2000s, and since the admission quota for each college was regulated so that there was no significant change during this time period.19)

18) On the other hand, as the proportion of non-elite college graduates increases, the relative supply of high school graduates could decrease. Thus, although the relative supply of the elite college graduates may not become scarce, their supply is steadier than that of the high school graduates. Considering that the wage differentials between the elite college and high school graduates also rapidly increased from 1999 to 2008 (from an increase of 27.8 percent to an increase of 59.8 percent), the growth in the relative demand for skills hypothesis appears to be more plausible. 19) The admission quota for each college has changed more or less; the number of entrants for colleges located in Seoul and its neighborhood (namely, in the Capital region) were regulated or slightly increased, whereas that for the colleges located at a distance from Seoul are known to have increased much more. However, the changes in the number of entrants or graduates were inspected for the following sample colleges that are located at a distance from Seoul and the data for which was accessible: Inha University, Kang Nung University (that renamed as Gangneung-Wonju National University), and Gyeongsang National University. Their prescribed number of entrants did not increase more than 10 percent over the last 10 years. On the basis of this information, the group comprising the top 14-50 colleges (which includes 15 colleges that are located at a distance from Seoul) can be regarded as a control group in the model. This control group aimed to control the supply of college graduates under the possibility that a proportion of the top 10 elite college wage premiums could be generated as a result of decreases in the relative supply of graduates from the top 10 colleges. The result in this analysis indicates that a meaningful amount of elite college 116 勞動經濟論集 第34卷 第1號

illustrates changes in wage premium of the top 10 college, subject to the relatively stable supply of top 50 college graduates. There was a very slight difference between the real earnings of the top 10 college graduates and those of the top 14-50

Changes in Wage Differential between Top 10 and Top 14-50 College Graduates

(1) (2) (3) (4) 1999 2002 2005 2008 -0.0698 -0.105 -0.236 0.0453 Under High School Graduates (-0.93) (-1.08) (-2.12)* (0.35) 0.0624 0.0508 -0.0159 0.0934 2-Yr College Graduates (1.29) (1.14) (-0.32) (1.81) 0.163 0.117 0.134 0.236 Top 51- College Graduates (2.49)* (2.21)* (2.21)* (3.79)*** 0.243 0.198 0.272 0.297 Top 14-50 College Graduates (3.89)*** (3.01)** (3.55)*** (3.31)** 0.252 0.270 0.419 0.540 Top 10 College Graduates (2.91)** (3.21)** (4.67)*** (4.59)*** -0.283 -0.206 -0.258 -0.250 Sex (female) (-6.85)*** (-5.52)*** (-5.79)*** (-5.39)*** 48.53 -19.93 -30.82 24.79 Constant (1.77) (-0.73) (-1.00) (0.74) N 342 322 320 316 R-sq 0.319 0.232 0.341 0.329 adj. R-sq 0.272 0.176 0.293 0.279 Wage Premium for Top 10 to Top 50 0.724 6.010 11.557 18.736 (%) Coefficients of dummy variables associated with age, experience and firm size are eliminated in this table. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001

premium stems from the relative demand; however, the limitation and restriction of this interpretation continues to exist. On the other hand, owing to the Brain Korea 21 program, a reduction in the prescribed number of entrants for the group comprising the top colleges that participated in the program was enforced; however, this regulation was applicable only from 2002 and for this year the analyzed cohort comprised of workers who were born in 1983. In this paper, this regulation on the student numbers that stemmed from the Brain Korea 21 program was disregarded for the sake of convenience and clarity of the analysis. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 117 college graduates in 1999 (0.7 percent). The wage premium of the top 10 colleges, however, steadily increased during the past 10 years so that the graduates from them earned 18.7 percent more than the top 14-50 college graduates in 2008. This result implies that the increase in the relative supply of non-elite college graduates was not the primary reason for the rapid growth in elite college wage premium throughout the 2000s. Rather, this result suggests the possibility that the relative demand for skills may be responsible for the rapid increase in the elite college wage premium. Furthermore, I investigated wage differentials by education for males and females. By the mid-2000s, the coefficient of the elite college graduate dummy was only significant among males (see Table A.4) and although the coefficient of top 10 college graduate dummy variable was significant throughout the 2000s, that of top 14-50 college graduate dummy was not significant among males for the same time period(see Table A.6). In contrast, the coefficient of the non-elite college graduate and the top 14- 50 college graduate dummy were significant among females in the late 1990s in each model for female workers (see Table A.5 and Table A.7) at the 0.001 significance level. Since the mid-2000s, however, the elite college graduate dummies in those models were also significant for females at the 0.01 significance level. One possible explanation is that the changes in social norms for working women who were the minor group before the 2000s and the improvement in the mind-set of female elite college graduates20) led to this result abovementioned. Specifically, up to the early 2000s, South Korea’s labor market was male-dominated because it was influenced by the male-oriented social norm and by the re-structuring of the labor market during the 1997-1998 currency and financial Crisis in South Korea, which seem to have resulted in the laying off those who were loosely attached to the labor market (e.g., female workers). Since the mid-2000s, with the favorable changes in social norms for female workers and the presence of successful female workers as role models, the participation

20) It is known that the purpose of entering elite colleges for females was “marrying a prospective man, and being a wise mother and good wife” before the 2000s in South Korea. Since the 2000s, at least, that mind-set has faded away. 118 勞動經濟論集 第34卷 第1號 of (elite) female workers in the labor market might have increased. On the other hand, as indicated in

and
, most of the college wage premium in the labor market in 1999, when the labor market was dominated by male workers, seems to originate from the wage differential between the top 10 elite college graduates and high school graduates. After the mid-2000s, it appears that the preference for males in the labor market has lessened and that favoring better educated female workers has increased. However, it is necessary to conduct a following research to investigate further details for these hypotheses.

3. Comparison with Changes in ‘Within-industry’ Wage Differen- tials among College Graduates

and
show that the average wage differentials between college graduates within the same industrial category, respectively; the relative wage of top 10 college graduates to the other college ones and to the top 14-50 college ones. Within-industry wage differentials among college graduates also increased during 1999- 2008 in each case, but the both growth rates were less than the results from
and
; especially in the late 2000s, the relative wages of elite college graduates within industry became definitely smaller than those values within the industry and firm size. Because it is generally believed that the graduates from elite colleges are more likely to be employed by big conglomerates (which offer wages are higher than the others), within industry wage differentials among college graduates are expected to be larger. So these results seem to be not easily understandable. All the elite college graduates, however, do not match with big conglomerates; being compared to 'flight to quality' in the asset market, the job security become one of the most important factor after the 1997-1998 currency and financial crisis in South Korea. A number of elite college graduates choose secure jobs and working for public sector. In such case, because the firm size of the public sector are not quite large and their wages are generally not much high (instead of the security), ‘within industry’ Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 119 wage differentials between elite and non-elite college graduates could be smaller than those (elite college wage premium) ‘within industry and firm size’. Not only for job security but also for other qualitative benefit could make the graduates from elite colleges accept relatively lower wages. KLIPS provides limited information about job security and other qualitative factors so that the rigorous analysis for the smaller 'within industry' wage differentials was restricted; this job security hypothesis remains the one of possible explanations. However, the within industry wage differentials definitely grew throughout the 2000s and the rising relative demand for elite college graduates seems to have played an important role as mentioned above. Meanwhile, if the job security and other things equal, the elite college graduates earned higher wage than the others. The dummy variables for industry were not significant whereas the dummy variables for firm size were meaningful especially for large firms; elite college graduates who matching with large firms, mainly chaebol- affiliated ones, might lead to this result. After the restructuring during the 1997-1998 currency and financial crisis, the South Korean economy became the big firms- dominated one; it is well known that inequality among firms in South Korea widened in the 2000s. Large and export-oriented firms, including Samsung Electronics, have been more profitable.21) In addition, it has been emphasized that more creative, insightful and innovative workers are necessary to survive in the competitive market environments. Then it is highly expected that the more lucrative firms (which can afford to pay higher wages) demand the more qualified human capital -more and better educated, and more skilled workers, such as elite college graduates. Thus elite college graduates are more likely to match with the higher wage offering big conglomerates, while the other college graduates may work for the big firms which

21) The Macroeconomic Stability Report 2010 which is published by Ministry of Strategy and Finance referred the widening differentials between large and small businesses, and between exporting and domestic-market oriented firms based on the Financial Statement Analysis by Bank of Korea; according to Lee (2010), the chaebol-affiliated firms become more profitable than the other ones after the currency and financial crisis. 120 勞動經濟論集 第34卷 第1號

Rising Elite College Wage Premium (within the same industrial category)

(1) (2) (3) (4) 1999 2002 2005 2008 -0.0803 -0.0817 -0.296 0.00269 Under High School Graduates (-1.06) (-0.63) (-1.90) (0.02) 0.0631 0.0793 0.0289 0.125 2-Yr College Graduates (1.56) (2.04)* (0.64) (2.89)** 0.22 0.196 0.272 0.322 4-Yr Non-Elite College Graduates (4.52)*** (4.25)*** (4.96)*** (4.98)*** 0.278 0.322 0.493 0.598 4-Yr Elite College Graduates (2.44)* (3.56)*** (6.06)*** (4.92)*** -0.293 -0.207 -0.243 -0.264 Sex (female) (-6.59)*** (-5.47)*** (-5.18)*** (-5.64)*** 56.38 -19.86 -23.58 24.90 Constant (2.08)* (-0.73) (-0.81) (0.73) N 342 322 320 316 R-sq 0.297 0.197 0.256 0.239 adj. R-sq 0.271 0.166 0.227 0.209 Wage Premium for 4-Yr Within Industry 4.754 10.535 17.374 20.878 Elite College (%) From

3.817 10.820 20.085 23.083 Coefficients of dummy variables associated with age, experience and industry are eliminated in this table. Firm size dummy dropped in this model. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001 offer relatively lower wage. In such case, the wage differentials among college graduates within firm size and industry might result from the offered wage difference among big conglomerates. Moreover, as inequality among firms in South Korea widened in the 2000s, the wage differentials among college graduates within firm size and industry might increase. Combining above explanations, although some of elite college graduates seem to choose job security, rather than higher wage, wage differentials between elite and non-elite college graduates continuously rose. Also it is suggested that elite college graduates might have qualitative benefit such as job security as well as the quantitative wage premium, relative to non-elite college graduates. Then the wage differentials Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 121

Changes in Wage Differential between Top 10 and Top 14-50 College Graduates (within the same industrial category)

(1) (2) (3) (4) 1999 2002 2005 2008 -0.0797 -0.0823 -0.295 0.00131 Under High School Graduates (-1.05) (-0.63) (-1.91) (0.01) 0.0633 0.0789 0.0312 0.124 2-Yr College Graduates (1.56) (2.02)* (0.69) (2.87)** 0.171 0.170 0.220 0.295 Top 51- College Graduates (3.32)** (3.03)** (3.15)** (3.77)*** 0.265 0.244 0.368 0.397 Top 14-50 College Graduates (3.88)*** (4.40)*** (6.01)*** (5.36)*** 0.278 0.321 0.492 0.596 Top 10 College Graduates (2.44)* (3.54)*** (6.04)*** (4.90)*** -0.293 -0.206 -0.249 -0.268 Sex (female) (-6.59)*** (-5.46)*** (-5.34)*** (-5.89)*** 54.80 -20.96 -24.57 22.64 Constant (2.04)* (-0.77) (-0.85) (0.65) N 342 322 320 316 R-sq 0.300 0.200 0.264 0.242 adj. R-sq 0.272 0.166 0.233 0.209 Wage Premium for Within Industry 1.028 6.190 9.064 14.245 Top 10 to Top 50 (%) From

0.724 6.010 11.557 18.736 Coefficients of dummy variables associated with age, experience and industry are eliminated in this table. Firm size dummy dropped in this model. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001. among college graduates in this paper may under-estimate the inequality of labor market.

4. Persistence of the Increase in Elite College Wage Premium along with Age

It is well-known that the increase in college wage premium is persistent so that the wage differentials by education rise with age or experience. In order to examine the 122 勞動經濟論集 第34卷 第1號 changes in elite college wage premium during their life-time, using the property of longitudinal survey, the individuals were traced every three years during 1999-2008; and the changes in their elite college wage premium along with their age were examined.

presents the changes in the relative wages by education along with age for the cohort who were 26-28 years old in 1999. The elite college wage premium seems to be persistent within their 10-year work lives. In particular, the elite college wage premium of this group increased when they were 29-31 years old, and then it decreased when 32-34 years old but was much larger than that in 26-28 years old over 10 percent. It was insignificant in 2008 survey. Before providing an explanation of this result, the description for people constituting the sample seems to be necessary. In this analysis, the sample size of model reduced sharply from 342 in 1999 to 174 in 2002. The missing sample was examined; in 1999, among the 342 sample individuals, 219 were male (123 were female). In 2002, among a total of 174 sample individuals, 121 were male (53 were female). Among the 168 individuals that were missing in the sample in 2002, 98 were male and 61 were male high school graduates. On the other hand, 70 females who were included in the 1999 sample size were missing in the 2002 sample; among them, 34 were high school graduates. In sum, the attrited samples from 1999 to 2002 comprised both females and males in the ratio of 10 to 7; moreover, most of the missing individuals were high school graduates.22) The sharp decrease in the total sample size appears to be a limitation in this analysis; however, the component ratio by education, particularly that of the elite college and non-elite college graduates, and that by gender continued to be similar in each survey. Therefore, it may be suggested that elite college wage premium are persistent.

22) A hypothesis can be suggested as follows: the high school graduates, who are hired and observe college wage premium which is larger than expected, might leave the labor market in order to enter colleges. However, there is lack of information to analyze this explanation in the data. It can be investigated by subsequent studies. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 123

Persistency of Elite College Wage Premium

(1) (2) (3) (4) 1999 2002 2005 2008 Birth in 1972-1974 cohort (26-28 yrs old) (29-31 yrs old) (32-34 yrs old) (35-37 yrs old) -0.0698 -0.277 -0.195 -0.468 Under High School Graduates (-0.93) (-2.79)** (-1.69) (-2.86)** 0.0624 0.165 0.192 0.172 2-Yr College Graduates (1.29) (2.55)* (2.96)** (1.92) 0.163 0.298 0.307 0.109 Top 51- College Graduates (2.49)* (3.04)** (3.46)*** (0.80) 0.243 0.351 0.436 0.154 Top 14-50 College Graduates (3.89)*** (3.91)*** (4.92)*** (1.06) 0.252 0.608 0.596 0.337 Top 10 College Graduates (2.91)** (4.58)*** (3.33)** (1.62) -0.283 -0.234 -0.251 -0.156 Sex (female) (-6.85)*** (-3.67)*** (-4.09)*** (-1.59) 48.53 42.18 -25.77 42.81 Constant (1.77) (0.89) (-0.47) (0.45) N 342 174 151 131 R-sq 0.319 0.407 0.548 0.473 adj. R-sq 0.272 0.320 0.471 0.366 Wage Premium for Top 10 to Top 0.724 19.023 11.142 N.A. 50 (%) Coefficients of dummy variables associated with age, experience and firm size are eliminated in this table. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001.

5. Effects on U-shaped Curve of College Wage Premium

As many researchers have shown, the college wage premium began increasing since the late 1990s at least. It is hypothesized that the increase in the elite college wage premium may contribute to the growth in the college wage premium. In order to examine this hypothesis, the weighted wage premium, which can be termed the total returns of human capital stock, were employed. The human capital stock can be measured using the number of workers by educational attainment; multiplying their 124 勞動經濟論集 第34卷 第1號 average wage for each group.23) Since KLIPS was surveyed using random sampling, it can be regarded that the sample size of KLIPS in each survey reflected the employment levels by education in the entire nation. In order to decompose the college wage premium, an model was employed as follows. It is assumed that all the workers comprise male graduates from high schools or 4-year colleges. For college graduates, they are assumed to be categorized as either 4-year non-elite college or 4-year elite college graduates. Note that the subjects (of human capital stock computation) are qualified to be college graduate workers that belong to the age group of 26-28 years in order to examine the effects of “the increase in elite college wage premium” on “the changes in college wage premium” using the obtained results. Let  stand for the proportion of the elite college  graduates’ employment to the entire college graduates employment and  represent the rate of returns from more (non-elite college/high school) and better (elite college/high school) educational attainment for college graduates comparing with high school graduates (The subscript 1 and 2 index graduates from non-elite colleges and elite colleges respectively.).24) The real rate of return to human capital stock for college graduates in this economy may be expressed as follows:

       (1)

 Based on the assumptions of this model,    can be interpreted as the employment-weighted wage differential between non-elite college graduates and high school graduates and it implies the contribution of the “non-elite” college factor on the  total returns to the human capital of college graduates.  can be considered as the employment-weighted elite college wage premium. It reflects the influence of the “elite” college factor on the overall return to the college graduates’ human capital.25)

23) Weil (2009) suggested this kind of computation on human capital stock. See chapter 6: Human Capital. 24) X and W can be compared to the amount of issued stocks and its rate of return, respectively. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 125

In this way, college wage premium can be decomposed into the wage differential between elite college and high school graduates, and that between non-elite college and high school graduates. Another labor market is assumed where the elite college wage premium does not increase and remains the same during ten years. In other words, it is assumed that the growth rate in the relative wage of the elite college graduates to high school ones is equal to that of non-elite college graduates to high school ones in this labor market.

               (2) 

  , the growth rate of the estimated college premium in this assumed  economy, where the elite college wage premium remains the same from 1999 to 2008,  was compared with , the growth rate of the simulated college premium by  equation (1). During this time period, the average proportion of the elite college (male) workers () is 0.23.26) It is applied that this level of employment did not change during 1999 −2008; therefore the growth in the college wage premium in the assumed labor   market, , from 1999 to 2008 equals that of the non-elite college/high school   wage differential, 22.9 percent. However, the simulated college wage premium, ,  is 47.7 percent. Approximately half of the growth in college wage premium can be explained by the increase in the elite college wage premium throughout the 2000s(see Table 7).

25) This can be compared to decomposing the “market capital growth rate” into contribution of common stock and the role of the preferred stock. 26) The ratio of employment of elite college graduates to college graduates, α has changed without a trend from 1999 to 2008. 126 勞動經濟論集 第34卷 第1號

The Contribution of the Changes in the Elite College Wage Premium to the Growth in College Wage Premium

Proportions of Human Capital Stock (Elite [top 10] college vs. Non-elite colleges) (1) (2) (3) (4) (5) (6) (7) (8) (9)               (fixed ) (fixed )  1999 0.7442 0.2050 0.2558 0.2510 0.2510 0.2156 0.2156 0.2168 0.2168 2008 0.7966 0.2520 0.2034 0.5410 0.3085 0.2650 0.3185 0.2635 0.3108 Growth 0.2293 1.1554 0.2293 0.2293 0.4773 0.2156 0.4337 Rate Note: (1) and (3) were obtained from KLIPS. (2) W1 and (4) W2 are from author's computation (see Table 2). (5) is elite college/high school wage differential in created labor market in this paper. (8) and (9) were calculated by fluctuated “”. However, (6) and (7) were calculated by fixed “”; and these are used for the analysis in this paper.

Furthermore, this pattern was observed as well even after the changes in the supply  of college graduates were controlled. In the expanded model,  now denotes the rate of return from more (top 14-50 colleges) and better (top 10 colleges) educational attainment of college graduates comparing with high school graduates. Under the fixed average employment ratio of the top 10 college graduates to the top 14-50 college graduates () of 0.23,27) the growth rate of the college wage premium in the assumed   economy, , from 1999 to 2008 was 22.2 percent, which is equivalent to that  of the wage differential between the top 14-50 colleges and the high schools.  Nevertheless, in case of the simulated college wage premium, is 44.2 percent.  Half of the increases in college wage premium can be interpreted by the rising elite college wage premium (see Table 8). These results evidence that the increase in the elite college wage premium could play an important role in the growth of college wage premium. Therefore the

27) The same ratio was assumed between the elite versus non-elite labor market between Table 7 and Table 8. On an average, the employment ratio of the top 10 college graduates to the top 50 college graduates is 0.42. However, this employment ratio increased from 0.35 to 0.47 during this time period. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 127

The Contribution of the Changes in the Elite College Wage Premium to the Growth in College Wage Premium (top 10 vs. top 14-50)

Proportions of Human Capital Stock (Top 10 college vs. Top 14-50 colleges) (1) (2) (3) (4) (5) (6) (7) (8) (9)                (fixed ) (fixed )  1999 0.7442 0.2430 0.2558 0.2510 0.2510 0.2448 0.2448 0.2450 0.2450 2008 0.7966 0.2970 0.2034 0.5410 0.3068 0.2992 0.3531 0.2990 0.3466 Growth 0.2222 1.1554 0.2222 0.2222 0.4422 0.2204 0.4147 Rate Note: (1) and (3) are assumed ratio, which were obtained from KLIPS for

(elite/non-elite college graduates). In order to investigate the effect of controlled supply or quality of graduates, these were plugged in this model for (1) and (3). (2) W1 and (4) W2 is from author's computation (see Table 3). (5) is elite college/high school wage differential in the model labor market in this paper. (8) and (9) were calculated by fluctuated “”. However, (6) and (7) were calculated by fixed “”; and these are used for the analysis in this paper. hypothesis that shifts in the relative demand for “more skilled and more educated” young workers occur in favor of elite college graduates rather than non-elite college graduates appears to be persuasive. Moreover, the growth in the relative demand for labor favoring elite college graduates — as more skilled workers among other college graduates — seems to induce the more rapid increase in the overall college wage premium throughout the 2000s.

V. Conclusion

1. The Increase in the Elite College Wage Premium Throughout the 2000s

In South Korea, the elite college premium rose in the 2000s. The average wage of the top 10 college graduates increased relative to that of top 14-50 college graduates 128 勞動經濟論集 第34卷 第1號 as well as relative to that of overall non-elite college graduates. These results support the hypothesis that since the labor market consists of college graduates, the relative demand for skills may raise the demand for elite college graduates, rather than the overall demand for college graduates. In addition, the rising the elite college wage premium seems to have led to the increase in the college wage premium of young workers. This increase in the elite college wage premium implies that the wage inequality among college graduates continued to widen throughout the 2000s. Reinterpreting Lee (2008), this widening wage inequality among college graduates may have occurred since the late 1990s.28) According to Goldin and Katz (2007), similar phenomenon had also existed in the United States during the late 1980s and it was accompanied with the growth in the relative supply of college graduates. The reasons for these phenomena — expanding wage differentials among college graduates — seem to be related with the fundamental causes of the increase in the relative demand for skills.

2. The Reason for the Increase in Elite College Wage Premium

In the paper, however, Goldin and Katz (2007) interpreted the rising wage inequality within college graduates in the United States since the late 1980s as a result of the polarization of wage distribution. They explained that this polarization of the wage structure was influenced by the bifurcated labor demand,29) i.e., the demand for a high-wage abstract task or a low-wage task, which cannot be performed by a computer. They hypothesized that computer-based technological change or offshoring to emerging economies could drive these shifts in the relative demand for labor, which

28) Lee (2008) indicated that the wage inequality had increased in the right tail of the income distribution in South Korea from 1996 to 2000. Since a majority of the top-end components can be assumed to be college graduates, this research may be reinterpreted to imply the rising inequality within college graduates. 29) Autor, Katz, and Kearney (2008) provided explanations based on this concept. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 129 were against the workers with a medium level of education but in favor of workers with higher or lower levels of education. In this paper, the wage distribution of the right-tail was not studied in detail; additional research is required for examining the polarization of wage distribution in South Korea. However, at least for the top-end wage distribution, an interpretation that polarizing demand for skills may occur in favor of elite college graduates can be suggested in order to explain the increase in the elite college wage premium. On the other hand, other explanations associated with globalization would seem to be more likely for the increase in elite college wage premium in South Korea. Dae-Il Kim (2006) indicated that offshoring, making direct investments into China (to exploit the opportunity of employing cheaper workers than those in South Korea, thus leading to a decrease in the demand for South Korean low wage workers — the unskilled ones — ), or growing trade volumes with China (which drives South Korea to specialize in higher technology embedded manufacturing) may reduce the relative labor demand for unskilled workers. According to this, the impact of emerging relationships with China and other factors associated with globalization or international trade may increase the relative demand for elite college graduates as more skilled workers because they are required for abstract and higher technological tasks. Choi and Jeong (2005), and Shin (2007) indicated that skill-biased technological change may have contributed to increasing the relative demand for skills in South Korea. For the polarizing skill-biased technological change hypothesis, it have not been examined in South Korea although Autor, Katz, and Kearney (2008) presented that skill-biased technological change drives the relative demand for labor polarization at the expense of median skilled workers in the United States. Nevertheless, the overall skill-biased technological change may at least contribute to increasing the relative demand for “elite college graduates” as the more skilled workers. In conclusion, since the relative supply of college graduates has grown rapidly in the young workers’ labor market, the elite college graduates are considered to be “more skilled workers” than the other college graduates; and thus the relative demand 130 勞動經濟論集 第34卷 第1號 for graduates from elite colleges increased. As Kim and Kang (2010) indicated, the effects of growing trade with China could have been a major source of shifts in the relative labor demand for elite college graduates as skills, with complementing the (polarizing) skill-biased technological change hypothesis

3. The Educational Policy Implication

In addition, this increase in the elite college wage premium might contribute to a sharp rise in the demand for shadow education which helps students to enter elite colleges. As an increase in the expected rate of return to an asset raise the demand for it,30) the rising elite college wage premium and its life-time persistence would be one of the incentives to enter elite colleges for students. Under the rank-based grading system for entering colleges, this incentive may contribute to increasing the demand for shadow education. Although it is well known that the education market is closely related with labor market, the mechanism of the impact of rising elite college wage premium on shadow education appears not to be regarded seriously. Even some previous educational policies, such as the policies for suppressing the demand for shadow education, seem to be made without considering these changes in the labor market. Admittedly, the distorted education system in South Korea is definitely problematic because it makes students and their parents rely on shadow education much more than public education. If the quality of public institutional education and other factors improve, students' academic performance via official education can be better; thus demand for shadow education might decrease. However, that demand may remain high in order to obtain better academic performance than other students, as long as elite college wage premium increases and there exists the rank-based grading system to evaluate students. Then wiser policies for shadow education are necessary which steer

30) Lee (2007) suggested that the demand for shadow education may be generated by the students’ incentives to enter elite colleges. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 131 those demands in the right direction to the robust economic growth. In sum, when establishing educational policies, it seems to be required to consider the changes in labor market as well as the education market itself in the view of general equilibrium model.

Reference

Autor, David H., Katz, Lawrence F., and Kearney, Melissa S. “Trends in U.S. Wage Inequality: Revising the Revisionists.” Review of Economics and Statistics 90 (2) (May 2008): 300-323. Ban, Ga Woon. “The Shift to the Service Economy and the Characteristic of the Structural Change since Financial Crisis in Korea -Focused on the Analysis of Manufacturing and Service Industry Using Productivity Decomposition Methodology and International Comparison-.” Korean Journal of Labor Economics 33 (1) (April 2010): 85-107. Bank of Korea, ed. Financial Statement Analysis. Seoul, Korea: Bank of Korea, 2009. (http://ecos.bok.or.kr). Bank of Korea, “Consumer Price Index.” Seoul, Korea: Bank of Korea, 1998-2009. (http://ecos.bok.or.kr). Choi, Kang-Shik, and Jeong, Jinook. “Technological Change and Wage Premium in a Small Open Economy: The Case of Korea.” Applied Economics 37 (1) (January 2005): 119-131. , and Jeong, Jinho. “An Analysis of the Causes of Wage Differentials in Korea.” Kukje Kyungje Yongu 9 (3) (December 2003): 183-208. Dae-Sung Educational Institute, ed. The Table of College Ranking with College Scholastic Ability Test Score of Admitted Students. Seoul, Korea: Dae-Sung Educational Institute, 1999. Goldin, Claudia, and Katz, F. Lawrence. “Long-Run Changes in the Wage Structure: 132 勞動經濟論集 第34卷 第1號

Narrowing, Widening, Polarizing.” Brookings Papers on Economic Activity 2007 (2) (September 2007): 135-165. Chensavasdijai, Varapat, Feyzioglu, Tarhan, Miniane, Jacques, Semblat, Romuald, Frydl, Eduward, Kang, Kenneth, and Kim, Song-Yi. “Republic of Korea: Selected Issues.” IMF Country Report. Washington, D.C.: International Monetary Fund, 2006. Jang, Soomyung. “An Analysis on Economic Return to University Rank.” Journal of Korean Education 33 (2) (July 2006). Jinhak-Sa, ed. The Table of Matching College Ranking with College Scholastic Ability Test Score. Seoul, Korea: Jinhak-Sa, 2008. (www.jinhak.com.) Kim, Dae-Il. “The Impacts on Korea's Labor Market of Rising China.” Structural Changes in the Korean Economy after the Economic Crisis, 2006, pp. 195-244. , and Topel, Robert H.. eds. “Labor Markets and Economic Growth: Lessons from Korea's Industrialization, 1970-1990.” In Differences and Changes in Wage Structures. edited by Richard Barry Freeman and Lawrence F. Katz, Chicago: University of Chicago Press, 1995. Kim, Hi Sam, and Lee, Sam Ho. “Analysis on the Effects of Higher Education on Labor Market and Its Hierarchy Structure.” Korea Development Institute Policy Study Series Vol. 2007-08. Seoul, Korea: Korea Development Institue, August 2007. Kim, Jin-Yeong. “Changes in the Hierarchy among Korean Universities Form 1994 to 2003.” Public Economics 11 (1) (May 2006): 121-153. Kim, Minseong, and Kang, Eun Young. “Effects of Korea's Exports to China on College Wage Premium.” Kukje Kyungje Yongu 16 (2) (August 2010): 31-49. Kim, Taejong. “Is There a Premium for Elite College Education: Evidence from a Natural Experiment in Japan.” Unpublished Manuscript. 2004. Kim, Wooyoung. “Changes in Employment Patterns and Wage Differentials by Education.” Studies for Finance and Economics (Bank of Korea working paper series) 344. Seoul, Korea: Bank of Korea, September 2008. Lee, Chulhee. “Rising Household Income Inequality in Korea, 1996-2000 -Impacts of Changing Wages, Labor Supply, and Household Structure.” Korean Journal of Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 133

Labor Economics 31 (2) (August 2008): 1-34. Lee, Jay Min. “The Performance of the Structural Adjustment after the Crisis in Korea.” Kukje Kyungje Yongu 16 (1) (April 2010): 79-105. Lee, Soojeong. “An Socio-Psychological Approach to the Cause of Shadow Education in South Korea.” The Journal of Educational Administration 25 (4) (December 2007): 455-484. Ministry of Education, Science and Technology. “Yearly Report for Statistic on Korean Education.” Seoul, Korea: Center for Education Statistics (http://cesi.kedi.re.kr). 1995-2009. Ministry of Strategy and Finance. Macroeconomic Stability Report. Seoul, Korea: Ministry of Strategy and Finance. 2009, 2010 (http://www.mosf.go.kr). Moon, Seong Hyeok. “The Skill Wage Differential of Korean after the 1980s: The Effects of Supply, Demand and International Trade, 1981-99.” Graduate School of Seoul National University, 2001. Shin, Sukha. “Technology and the Demand for Unskilled Labor after the Economic Crisis.” Korea Development Review 29 (1) (June 2007): 1-39. Statistics Korea. “Population Projections for Korea.” 1995-2009. (http://kosis.kr). Weil, David N. Economic Growth. 2d ed. Massachusetts: Addison Wesley, 2008, 1st ed. Massachusetts: Addison Wesley, 2005. 134 勞動經濟論集 第34卷 第1號

Appendix

The Proportion of 2-year and 4-year College Graduates in Young Workers (1) (2) (3) (4) (5) (6) (7) The amount of 2 yr Three times 24-26 years 26-28 years Young workers The proportion of and 4 yr college of column old Female old Male (estimated by the sum college graduates graduates who 2 population population of column 4 and 5) in young workers graduated in this year 1998 478590 1435770 1287046 1366965 2654011 0.5410 1999 511984 1535952 1234433 1365648 2600081 0.5907 2000 542196 1626588 1177305 1352996 2530301 0.6428 2001 584340 1753020 1121824 1313824 2435648 0.7197 2002 603282 1809846 1095752 1253566 2349318 0.7704 2003 633730 1901190 1105178 1201324 2306502 0.8243 2004 627270 1881810 1135415 1171772 2307187 0.8156 2005 634523 1903569 1165091 1173614 2338705 0.8139 2006 638259 1914777 1175833 1201943 2377776 0.8053 2007 642987 1928961 1149938 1233498 2383436 0.8093 2008 646092 1938276 1091665 1245949 2337614 0.8292 2009 636628 1909884 1020947 1221434 2242381 0.8517 * Source: Yearly Report for Statistic on Korean Education (retrieved from Center for Education Statistics [cesi.kedi.re.kr] on Nov. 2010) and Population projections for Korea(retrieved from Statistics Korea [kosis.kr] on Nov. 2010). Populations from 2006 to 2009 are estimates. * Note that the proportion of college graduates to young economically active population would be higher. * “Young” means the age group of 26-28 years in this paper.

The Proportion of 4-year College Graduates in Young Workers (1) (2) (3) (4) (5) (6) (7) The amount of 4 yr 24-26 years 26-28 years Young workers The proportion of Three times college graduates who old Female old Male (estimated by the sum 4-yr college graduates of column 2 graduated in this year population population of column 4 and 5) in young workers 1998 196566 589698 1287046 1366965 2654011 0.2222 1999 204390 613170 1234433 1365648 2600081 0.2358 2000 214498 615454 1177305 1352996 2530301 0.2432 2001 239702 658590 1121824 1313824 2435648 0.2704 2002 244852 699052 1095752 1253566 2349318 0.2976 2003 258126 742680 1105178 1201324 2306502 0.3220 2004 267058 770036 1135415 1171772 2307187 0.3338 2005 268833 794017 1165091 1173614 2338705 0.3395 2006 270546 806437 1175833 1201943 2377776 0.3392 2007 277858 817237 1149938 1233498 2383436 0.3429 2008 282670 831074 1091665 1245949 2337614 0.3555 2009 279059 837177 1020947 1221434 2242381 0.3733 * Source: Yearly Report for Statistic on Korean Education(retrieved from Center for Education Statistics [cesi.kedi.re.kr] on Nov. 2010) and Population projections for Korea(retrieved from Statistics Korea [kosis.kr] on Nov. 2010). Populations from 2006 to 2009 are estimates. * Note that the proportion of college graduates to young economically active population would be higher. * “Young” means the age group of 26-28 years in this paper. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 135

The List of Top 50 Colleges

Top 10 (Elite) colleges Seoul National University, Korea University, Yonsei University, Sogang University, Sung Kyun Kwan University, Hanyang University, Chung Ang University, Kyung Hee University, Han Kuk University of Foreign Studies, University of Seoul, Ewha Womans University, Korea Advanced Institute of Science and Technology (KAIST), and Pohang University of Science and Technology (POSTECH); total 13 colleges. Top 14-50 colleges Pusan National University, Kyungpook National University, Dongguk University, , , , Soongsil University, Inha University, Kwangwoon University, , Sookmyung Women's University, Kookmin University, Myongji University, Korea Aerospace University, Incheon University, Chonnam University, The Catholic University of Korea, Sangmyung University, Chungnam National University, Sejong University, Pukyong National University, Chonbuk National University, , Hansung University, Kyungwon University, Dong-A University, Yeungnam University, Chungbuk National University, The , Kangwon National University, Seo Kyeong University, Sahmyook University, Kongju Nationl University, Anyang University, Hanshin University, Yong In University, Kangnam University, Chosun University, Sungshin Women's University, and Dongduk Women's University; total 37 colleges

Elite College Wage Premium (for male)

(1) (2) (3) (4) male 1999 2002 2005 2008 -0.0201 -0.125 -0.225 0.0482 Under High School Graduates (-0.23) (-1.24) (-1.81) (0.32) 0.0413 0.0754 -0.0148 0.0913 2-Yr College Graduates (0.74) (1.35) (-0.22) (1.25) 0.103 0.0538 0.145 0.246 Non-elite college College Graduates (1.53) (0.85) (1.93) (2.99)** 0.401 0.126 0.390 0.491 Elite college College Graduates (3.72)*** (1.26) (3.39)*** (3.18)** 15.50 -71.70 -48.77 32.46 Constant (0.46) (-2.07)* (-1.10) (0.69) N 219 201 189 194 R-sq 0.258 0.207 0.330 0.273 adj. R-sq 0.184 0.119 0.251 0.189 Coefficients of dummy variables associated with age, experience, industry and firm size are eliminated in this table. This analysis is only for males. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001. 136 勞動經濟論集 第34卷 第1號

Elite College Wage Premium (female)

(1) (2) (3) (4) female 1999 2002 2005 2008 -0.118 -0.0727 -0.366 0.0155 Under High School Graduates (-0.82) (-0.20) (-1.05) (0.04) 0.181 0.0190 -0.0520 0.0949 2-Yr College Graduates (1.93) (0.25) (-0.64) (1.20) 0.330 0.175 0.230 0.264 Non-elite college College Graduates (4.20)*** (2.11)* (2.73)** (2.99)** 0.0610 0.519 0.476 0.632 Elite college College Graduates (0.41) (3.20)** (2.98)** (2.86)** 73.07 78.47 -3.743 19.27 Constant (1.51) (1.53) (-0.08) (0.37) N 123 121 131 122 R-sq 0.446 0.316 0.423 0.356 adj. R-sq 0.337 0.179 0.318 0.229 Coefficients of dummy variables associated with age, experience, industry and firm size are eliminated in this table. This analysis is only for females. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001.

Wage Differential between Top 10 and Top 14-50 College Graduates (male) (1) (2) (3) (4) male 1999 2002 2005 2008 -0.0202 -0.129 -0.222 0.0475 Under High School Graduates (-0.23) (-1.28) (-1.79) (0.31) 0.0413 0.0754 -0.00946 0.0910 2-Yr College Graduates (0.74) (1.35) (-0.14) (1.24) 0.0918 0.00507 0.100 0.236 Top 51- College Graduates (0.99) (0.07) (1.22) (2.70)** 0.112 0.136 0.253 0.288 Top 14-50 College Graduates (1.34) (1.57) (2.31)* (1.93) 0.401 0.124 0.393 0.491 Top 10 College Graduates (3.71)*** (1.24) (3.43)*** (3.17)** 14.93 -75.50 -44.90 31.76 Constant (0.44) (-2.18)* (-1.01) (0.67) N 219 201 189 194 R-sq 0.259 0.215 0.338 0.273 adj. R-sq 0.180 0.123 0.254 0.184 Coefficients of dummy variables associated with age, experience, industry and firm size are eliminated in this table. This analysis is only for males. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001. Changes in Wage Differentials among College Graduates in South Korea, 1999-2008 (Eunmi Ko) 137

Wage Differential between Top 10 and Top 14-50 College Graduates (female) (1) (2) (3) (4) female 1999 2002 2005 2008 -0.121 -0.0608 -0.361 0.0189 Under High School Graduates (-0.84) (-0.17) (-1.04) (0.05) 0.179 0.0173 -0.0578 0.0966 2-Yr College Graduates (1.91) (0.22) (-0.71) (1.22) 0.278 0.150 0.194 0.243 Top 51- College Graduates (2.81)** (1.61) (1.92) (2.50)* 0.376 0.221 0.275 0.308 Top 14-50 College Graduates (3.93)*** (1.98) (2.54)* (2.57)* 0.0600 0.519 0.468 0.636 Top 10 College Graduates (0.41) (3.19)** (2.91)** (2.86)** 73.51 80.75 -7.233 15.86 Constant (1.52) (1.57) (-0.16) (0.31) N 123 121 131 122 R-sq 0.450 0.319 0.426 0.358 adj. R-sq 0.336 0.174 0.315 0.224 Coefficients of dummy variables associated with age, experience, industry and firm size are eliminated in this table. This analysis is only for females. T statistics is in parentheses. * p<0.05 ** p<0.01 *** p<0.001. 138 勞動經濟論集 第34卷 第1號

abstract

1999∼2008년 한국에서 대졸자 간 임금격차의 변화

고 은 미

본 논문은 노동패널을 이용하여 4년제 대학을 졸업한 26∼28세 임금근로자 간 (특 히 명문대와 다른 대학 출신 졸업자 사이의) 임금격차를 분석한다. 산업과 기업규모 를 통제했을 때 다른 4년제 대학 출신 근로자에 비하여 상위 10위권 대학 졸업생의 상대임금이 꾸준히 상승해 왔고, 경력 기간 동안 지속되는 것으로 보인다. 이들의 산 업 내 임금격차도 상승하나 2000년대 후반에는 그 크기가 산업과 기업규모 통제 시 보다 상대적으로 작아졌다. 나아가 최근 대졸-고졸 임금격차의 상승에 명문대 임금프 리미엄의 증가가 상당 부분 기여하고 있는 것으로 나타났다.

- 주제어: 신규 대졸 취업자, 임금격차, 숙련근로자에 대한 상대수요