教育错配对工资的影响:来自CFPS的新证据
|
摘要
从资源配置效率的角度看,教育错配是劳动力资源配置效率低下的一种表现。本文利用中国家庭追踪调查数据研究了中国劳动力市场中的教育错配发生率,并借助异方差构造工具变量的思路估计了教育错配对劳动者工资的影响。研究发现,中国劳动力市场中存在比较普遍的垂直型教育错配现象;随着劳动者教育程度的提高,过度教育的发生率越来越高,在大专学历以上的劳动者当中有50%左右为过度教育;总体过度教育和教育不足发生率分别为37.8%和20%。其他条件相同时,过度教育的劳动者遭受到显著的工资惩罚,平均工资比恰好匹配的劳动者低22.7%;教育不足的劳动者获得一定工资溢价,平均工资比恰好匹配的劳动者高16.5%。
From the perspective of efficient resource allocation, educational mismatch may suggest evidence for inefficiencies in the labor market. In this paper, we estimate the incidence rate of educational mismatch and investigate the impact of educational mismatch on wages in China. Our empirical analysis, based on data from China Family Panel Studies(CFPS), indicate that vertical educational mismatch is a common phenomenon in the China's labor market. As workers become more educated, the incidence of vertical mismatch becomes higher. Specifically, the results show that 51.24% of college graduates are overeducated and the incidence rate of over-education and under-education in overall labor market is 37.8% and 20% respectively. To estimate the causal influence of educational mismatch, we solve the problem of endogeneity of the education and educational mismatches by method of using heteroscedasticity to identification. We find that over-educated individuals earn 22.7% less than individuals with similar levels of education but match with their job requirement. We also find evidence of under-education wage premium. Relative to workers with the same level of education, under-educated individuals earn 16.5% more than matched individuals.
From the perspective of efficient resource allocation, educational mismatch may suggest evidence for inefficiencies in the labor market. In this paper, we estimate the incidence rate of educational mismatch and investigate the impact of educational mismatch on wages in China. Our empirical analysis, based on data from China Family Panel Studies(CFPS), indicate that vertical educational mismatch is a common phenomenon in the China's labor market. As workers become more educated, the incidence of vertical mismatch becomes higher. Specifically, the results show that 51.24% of college graduates are overeducated and the incidence rate of over-education and under-education in overall labor market is 37.8% and 20% respectively. To estimate the causal influence of educational mismatch, we solve the problem of endogeneity of the education and educational mismatches by method of using heteroscedasticity to identification. We find that over-educated individuals earn 22.7% less than individuals with similar levels of education but match with their job requirement. We also find evidence of under-education wage premium. Relative to workers with the same level of education, under-educated individuals earn 16.5% more than matched individuals.
引文
[1] McGuinness S,Pouliakas K,Redmond P.Skills mismatch:Concepts measurement and policy approaches [J].Journal of Economic Surveys,2018,32(4):985-1015.
[2] McGuinness S.Overeducation in the labour market [J].Journal of Economic Surveys,2006,20(3):387-418.
[3] Handel M J,Alexandria V,Puerta M L S.Accounting for mismatch in low-and middle-income countries:Measurement,magnitudes,and explanations [M].Directions in Development.Washington,DC:World Bank,2016:37-78.
[4] 缪宇环.我国过度教育现状及其影响因素探究 [J].统计研究,2013(7):48-54.
[5] 颜敏,王维国.中国过度教育现状及其演变——来自微观数据的证据 [J].山西财经大学学报,2017(3):15-29.
[6] Lewbel A.Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor models [J].Journal of Business & Economic Statistics,2012,30(1):67-80.
[7] Lewbel A.Identification and estimation using heteroscedasticity without instruments:The binary endogenous regressor case [J].Economics Letters,2018,165:10-12.
[8] Millimet D L,Tchernis R.Estimation of treatment effects without an exclusion restriction:With an application to the analysis of the school breakfast program [J].Journal of Applied Econometrics,2013,28(6):982-1017.
[9] Leuven E,Oosterbeek H.Overeducation and mismatch in the labour market [C]// Hanushek E,Machin S,Woessmann L.Handbook of the Economics of Education (Volume 4).Elsevier,2011:283-326.
[10] Carroll D,Tani M.Over-education of recent higher education graduates:New Australian panel evidence [J].Economics of Education Review,2013,32:207-218.
[11] Carroll D,Tani M.Job search as a determinant of graduate over-education:Evidence from Australia [J].Education Economics,2015,23(5):631-644.
[12] 胥秋.从过度教育现象看我国高等教育的结构失衡 [J].高教发展与评估,2006(1):10-13.
[13] 李锋亮,岳昌君,侯龙龙.过度教育与教育的信号功能 [J].经济学(季刊),2009(2):569-582.
[14] Korpi T,Tahlin M.Educational mismatch,wages,and wage growth:Overeducation in Sweden,1974—2000 [J].Labour Economics,2009,16(2):183-193.
[15] Mavromaras K,McGuinnes S,O’Leary N,et al.Job mismatches and labour market outcomes:Panel evidence on university graduates [J].Economic Record,2013,89(286):382-395.
[16] Kleibrink J.Inept or badly matched?—Effects of educational mismatch in the labor market [J].Labour,2016,30(1):88-108.
[17] Montt G.Field-of-sudy mismatch and overqualification:Labour market correlates and their wage penalty [J].Iza Journal of Labor Economics,2017,6(1):2-22.
[18] Lamo A,Messina J.Formal education,mismatch and wages after transition:Assessing the impact of unobserved heterogeneity using matching estimators [J].Economics of Education Review,2010,29(6):1086-1099.
[19] 武向荣.过度教育的经验研究——基于对3家企业的调查 [J].教育与经济,2005(2):4-8.
[20] Ren W,Miller P W.Changes over time in the return to education in urban China:Conventional and ORU estimates [J].China Economic Review,2012,23(1):154-169.
[21] Wu Na,Wang Qunyong.Wage penalty of overeducation:New micro-evidence from China [J].China Economic Review,2018,50:206-217.
[22] 颜敏,王维国.教育错配对工资的惩罚效应——来自中国微观面板数据的证据 [J].财经研究,2018(3):84-96.
[23] Duncan G,Hoffmann S.The incidence and wage effects of overeducation [J].Economics of Education Review,1981,1(1):75-86.
[24] Li I W,Miller P W.Overeducation and earnings in the Australian graduate labour market:An application of the Vahey model [J].Education Economics,2015,23(1):63-83.
(1)事实上,该文表5说明工具变量在第一阶段回归中都不显著,且原文也没有详细介绍弱工具变量及过度识别的检验结果。
(2)和其他识别策略依赖的假定无法检验一样,这里的假定也无法检验。不过,式(4)和式(5)是比较容易满足的,Lewbel[6]详细论证了其中的理由。事实上,在微观计量分析中异方差问题是比较常见的。所以,即便式(4)和式(5)无法检验,但可以从经验上判断出本文对模型的假定并不强。
(3)本文没有使用面板数据的另外两个重要原因是2016年的收入信息缺失较多并且2016年的认知能力测试分数与2014年不可比,而这两个变量都是本文的关键变量。事实上,考虑收入信息和其他变量缺失的情况下,我们只成功匹配了546位受访者,但利用这一面板数据估计并没有得到非常稳健可靠的结论。
(4)我们还对每种估计方法做了平衡性检验,结果都说明处理组与控制组的匹配效果较好,但限于控制变量较多导致平衡性检验的详细结果篇幅太大,故未在文中汇报。
[2] McGuinness S.Overeducation in the labour market [J].Journal of Economic Surveys,2006,20(3):387-418.
[3] Handel M J,Alexandria V,Puerta M L S.Accounting for mismatch in low-and middle-income countries:Measurement,magnitudes,and explanations [M].Directions in Development.Washington,DC:World Bank,2016:37-78.
[4] 缪宇环.我国过度教育现状及其影响因素探究 [J].统计研究,2013(7):48-54.
[5] 颜敏,王维国.中国过度教育现状及其演变——来自微观数据的证据 [J].山西财经大学学报,2017(3):15-29.
[6] Lewbel A.Using heteroscedasticity to identify and estimate mismeasured and endogenous regressor models [J].Journal of Business & Economic Statistics,2012,30(1):67-80.
[7] Lewbel A.Identification and estimation using heteroscedasticity without instruments:The binary endogenous regressor case [J].Economics Letters,2018,165:10-12.
[8] Millimet D L,Tchernis R.Estimation of treatment effects without an exclusion restriction:With an application to the analysis of the school breakfast program [J].Journal of Applied Econometrics,2013,28(6):982-1017.
[9] Leuven E,Oosterbeek H.Overeducation and mismatch in the labour market [C]// Hanushek E,Machin S,Woessmann L.Handbook of the Economics of Education (Volume 4).Elsevier,2011:283-326.
[10] Carroll D,Tani M.Over-education of recent higher education graduates:New Australian panel evidence [J].Economics of Education Review,2013,32:207-218.
[11] Carroll D,Tani M.Job search as a determinant of graduate over-education:Evidence from Australia [J].Education Economics,2015,23(5):631-644.
[12] 胥秋.从过度教育现象看我国高等教育的结构失衡 [J].高教发展与评估,2006(1):10-13.
[13] 李锋亮,岳昌君,侯龙龙.过度教育与教育的信号功能 [J].经济学(季刊),2009(2):569-582.
[14] Korpi T,Tahlin M.Educational mismatch,wages,and wage growth:Overeducation in Sweden,1974—2000 [J].Labour Economics,2009,16(2):183-193.
[15] Mavromaras K,McGuinnes S,O’Leary N,et al.Job mismatches and labour market outcomes:Panel evidence on university graduates [J].Economic Record,2013,89(286):382-395.
[16] Kleibrink J.Inept or badly matched?—Effects of educational mismatch in the labor market [J].Labour,2016,30(1):88-108.
[17] Montt G.Field-of-sudy mismatch and overqualification:Labour market correlates and their wage penalty [J].Iza Journal of Labor Economics,2017,6(1):2-22.
[18] Lamo A,Messina J.Formal education,mismatch and wages after transition:Assessing the impact of unobserved heterogeneity using matching estimators [J].Economics of Education Review,2010,29(6):1086-1099.
[19] 武向荣.过度教育的经验研究——基于对3家企业的调查 [J].教育与经济,2005(2):4-8.
[20] Ren W,Miller P W.Changes over time in the return to education in urban China:Conventional and ORU estimates [J].China Economic Review,2012,23(1):154-169.
[21] Wu Na,Wang Qunyong.Wage penalty of overeducation:New micro-evidence from China [J].China Economic Review,2018,50:206-217.
[22] 颜敏,王维国.教育错配对工资的惩罚效应——来自中国微观面板数据的证据 [J].财经研究,2018(3):84-96.
[23] Duncan G,Hoffmann S.The incidence and wage effects of overeducation [J].Economics of Education Review,1981,1(1):75-86.
[24] Li I W,Miller P W.Overeducation and earnings in the Australian graduate labour market:An application of the Vahey model [J].Education Economics,2015,23(1):63-83.
(1)事实上,该文表5说明工具变量在第一阶段回归中都不显著,且原文也没有详细介绍弱工具变量及过度识别的检验结果。
(2)和其他识别策略依赖的假定无法检验一样,这里的假定也无法检验。不过,式(4)和式(5)是比较容易满足的,Lewbel[6]详细论证了其中的理由。事实上,在微观计量分析中异方差问题是比较常见的。所以,即便式(4)和式(5)无法检验,但可以从经验上判断出本文对模型的假定并不强。
(3)本文没有使用面板数据的另外两个重要原因是2016年的收入信息缺失较多并且2016年的认知能力测试分数与2014年不可比,而这两个变量都是本文的关键变量。事实上,考虑收入信息和其他变量缺失的情况下,我们只成功匹配了546位受访者,但利用这一面板数据估计并没有得到非常稳健可靠的结论。
(4)我们还对每种估计方法做了平衡性检验,结果都说明处理组与控制组的匹配效果较好,但限于控制变量较多导致平衡性检验的详细结果篇幅太大,故未在文中汇报。