在高校扩招和金融危机的双重背景下，大学生就业难问题成为当前中国的热点，然而理论界尚缺乏深入系统的微观分析。同时，大学毕业生劳动力市场方兴未艾，但其运行机制未能得到学术界的足够重视。经典的人力资本理论不能够解释为何在人力资本同质的情况下仍然存在不同群体的工资差距现象，新兴的劳动力市场分割、歧视等理论可以提供部分答案。本书主要探讨人力资本投资基本同质的大学毕业生的初始工资结构问题，借以寻找当前大学生就业难的根源所在。全书共分为8个章节，各章的主要内容安排如下：

第1章阐述本书的研究背景和意义，简要介绍全书的研究思路与创新处，并概括说明调查数据来源、处理过程以及基本的描述性统计特征，从直觉上给出研究动机。

第2章对国内外相关研究文献进行系统的梳理与回顾，并作出相应评述。本章先从工资结构理论和方法入手，扼要介绍人力资本和劳动力市场分割等理论，并梳理了研究方法脉络。然后，本章系统回顾了性别、行业、部门、城市规模等领域的国内外工资结构相关文献资料，最后作了总结和评述，以便找准以往研究空白和不足，进一步突出本书研究创新和价值所在。

第3章是大学毕业生性别工资差异研究部分。本章首先建立了性别工资差异分析框架，并推理出性别工资差异的重要原因是歧视与能力因素以及女性集中的行业存在较低的工资收入与较大的性别工资差异等假说。随后利用中国2008届大学毕业生求职与工作能力调查数据验证了这样的看法。文章还就大学毕业生性别工资差异和歧视作了国际对照，发现当前我国大学生性别工资差异远低于发达国家和地区水平，但性别歧视程度相对非常严重。最后从计量方法上作了相关讨论与扩展，证实大学毕业生行业内差距和性别歧视比重很大。本章结论说明，女大学毕业生提升工作能力并挤入男生集中的行业是增加工资收入的重要途径。

第4章分析大学毕业生高、低收入行业的就业选择与工资收入的影响因素，并测算行业工资差异的大小及构成。通过考察2008届全国大学本科毕业生抽样数据发现：男性、211院校毕业或有良好家庭背景的大学毕业生更容易进入高收入行业工作，能力对行业选择与工资收入影响显著；行业工资差异高达41%，其中超过80%的差异归因于行业分割，余下差异中地区因素居主导地位；去除垄断性行业能够降低高、低收入行业工资差异与分割比重，并提升能力解释力。这些经验事实表明，目前大学毕业生劳动力市场就业机会不均等现象同样普遍存在，行政垄断等非市场性因素直接导致行业分割问题非常严重。

第5章在考虑就业参与及部门选择双重选择性偏差的前提下，利用大学生就业数据进一步研究了不同部门就业的大学毕业生工资差异大小及构成。结果发现，211院校毕业、有良好家庭背景或来自中西部地区的大学毕业生更容易进入公有部门，专业为法学、理学或来自中西部不发达地区的大学毕业生更不容易找到工作；选择偏差项对公有部门没有影响，但对私有部门影响显著为正；部门工资差异仅为18%，但分割效应占61%，男女子样本略有区别。这表明，尽管目前大学毕业生公私部门工资差异很小，但大学毕业生劳动力市场就业机会不公平和分割现象仍然存在，能够部分解释“公务员热”等现象。

第6章实证考察不同规模城市的工资差异并检验能力群分、生活成本差异或者城市集聚经济等多种理论假说。研究发现，各种假说均能不同程度地解释直辖市与副省级城市相对于地级城市的工资溢价，但城市集聚经济假说最为重要，其次是生活成本差异假说，这与发达国家能力群分和集聚效应明显的经验事实相区别。将城市类型变量替换为城市人口度量的稳健性检验进一步支持实证分析结果。

第7章从高校毕业生劳动力市场产出的角度考察教育质量的高低。利用部分抽样调查数据，本章通过OLS、IV、MTE等多种计量方法实证分析211院校毕业生的工资溢价来源。结果发现，在控制个人能力和家庭背景后，采用各省211高校个数作为毕业学校类型工具变量的教育质量回报为47.3%，ATE方法则高达39.4%—94.1%。这一结论对我国财政支持高等教育方向与途径具有重要的参考价值。

第8章总结全书研究结论，概述现实启示，并指明研究缺陷与改进方向。

本书的主要结论可以进一步概括为以下两点：

（一）在狭义的人力资本几乎同质的情况下，性别、行业、部门、城市规模、教育质量方面的工资结构问题仍然明显存在，并且更多地归结为歧视、分割、财力等非市场性因素。这对经典经济理论解释提出新的挑战和更高要求。

（二）大学生就业问题同自身能力和整个劳动力市场大环境均有密切关系，故解决大学生就业难的问题需要学生、学校和政府多管齐下。

本书的创新点主要体现在研究视角、研究方法和研究数据三个方面。

第一，从研究视角上讲，本书选取大学生就业的课题，从初始工资结构角度探讨大学毕业生劳动力市场运行特征，以剖析大学生就业难的根源，对政府决策和理论文献都有重要的研究意义。一方面，高校扩招政策和金融危机冲击对高校毕业生就业问题产生重大的负面影响，他们的就业状况关系国计民生与社会稳定，系统研究大学生就业难根源具有很强的现实针对性和政策含义。另一方面，以往工资结构研究无法剔除受教育程度、工作经验、家庭婚姻等因素对不同群体工资收入的影响，故得到的结论值得质疑，本书以毕业半年后的大学生工资结构为研究对象，能够很大程度上回避上述特征的干扰，更为纯粹地考察人力资本以外的工资差距影响因素与原因，这在理论上具有明显的优势和创新之处。

第二，本书研究方法前沿，具有很强的创新性。比如，虽然国内外有很多文献研究公有部门和私有部门之间的工资差距，但几乎都忽略未就业状态的劳动者群体引起的选择性偏差问题。本书利用Meng and Schimidt（1985）提供的双重选择思路来分析大学毕业生公私部门的工资差异，具有较强的方法创新性，所得结论令人更为信服。再如，Appleton分解虽然复杂，但能够解决指数基准和样本选择问题；边际处理效应不仅更一致地估计高等教育质量回报，还能刻画教育回报的异质性，相对于平均处理效应具有明显的优势。本书尝试分别用来分析大学毕业生性别工资差异和211院校毕业生的工资溢价。

第三，本书所用数据独特，具有良好的统计性质。以往大学生就业相关文献大多停留在宏观现象描述和宏观数据分析层次上。然而众所周知，加总后的宏观数据不能够觉察出经济现象微观机制所在，甚至会得出某些误解或错误结论。虽然也有部分文献使用大学生工资微观数据，但他们的调查往往局限于某些地区、某些时段，样本量和随机性受到很大限制，得到的结论不具有很强的推断性，政策含义有限。本书使用麦可思的“中国大学毕业生求职与工作能力2009年度调查”数据，调查面几乎囊括全国各省区高校，调查时间是本科毕业半年且基本处于经济危机影响最严重时段，调查问卷中参照国际标准尤其设计了工作能力测度，能够在很大程度上扩展以往大学生就业数据的深度与广度，从而分析出更为科学可靠的实证结果。

关键词：大学毕业生 工资结构 均值分解 能力 双重选择 边际处理效应

Abstract

In the contexts of both college expansion and financial crisis，college graduates employment issues have become pressing；however，there is little deep and systematic analysis at a micro level in this academic field.In the meanwhile，current labor market of graduates is promising，but researchers have paid not enough attention to its mechanism.Classic human capital theory cannot explain why wage differentials between groups with homogenous human capital still exist.New theories such as Segmented Labor Markets Theory or discrimination arguments provide several answers to this phenomenon.My thesis investigates initial wage structures for the graduates with almost same human capital investments in order to find out more reasons for difficulty of college graduates’employments.The study includes 8 chapters，and the main idea of each chapter is listed as follows：

Chapter 1 illustrates the research background and influence，introduces the thoughts and contributions，and gives a brief idea about data sources，data processing，and the descriptive characteristics so as to provide the research motivation intuitively.

The second chapter reviews the relevant literatures at both home and abroad，and gives some comments.It first introduces the theories such as human capital and segmented labor market and common methods for the wage structure research.Then，the chapter takes a systematic review about wage structure from gender，industry，sector，and city size view，respectively.It finally concludes and makes some comments so that I can find out the research empties and caveats and also further highlight the creation and contribution of the whole thesis.

Chapter 3 is the research about the gender wage differentials for graduates.Firstly，it builds up an analysis framework for gender wage gaps，and proposes several hypotheses.For example，gender wage gap stems from discrimination and ability issues，and industries with more women usually have lower wages and larger gender wage differentials.Then，I testify above arguments using a survey data of Placement and Ability of Chinese College Graduates graduated in 2008.In addition，we conduct a comparison about gender wage gaps and discriminations across the world，and it turns out that the gender wage gap in current Chinese graduate labor market is relatively lower than that in the developed countries and districts while the gender discrimination is the highest among them.Finally，I provide relevant discussions and expansions in an econometric method view，and confirm that intra-industry wage differential and gender discrimination is very serious.The results infer that，it is quite necessary for female graduates to lift up working abilities and to crowd into male-dominated industries in order to earn more wages.

Chapter 4 analyzes the determinants of employment choice and wages in different income industries，and estimates the magnitude and composition of wage differentials.Based on the same survey data，I find that，those who are male，from 211 universities or having good family backgrounds seem easier to enter high-paying industries，and individual ability plays an important role in industrial choice and wage in come.Industry wage differential is as high as 41%，over 80% of which can be attributable to industrial segmentation while geographic factors dominate the rest portion.High-low paying wage differential and segmentation portion will go down while ability explainable portion will rise if getting rid of several monopoly industry subsamples.These empirical evidences indicate that the labor market of graduates also has traits of unequal opportunity and severe segmentation，and the nonmarket factors such as state monopoly affect the industry segmentation directly and heavily.

Adjusting double selection biases from both working participation and sectoral choice，the fifth chapter investigates the magnitude and composition of wage differentials between public and private sectors based on the microdata for graduates employment.The results show that those who once graduated from 211 universities，having good family backgrounds or from the middle or western areas of China seem more easily to enter public sector while those who majored in law or science or from the poor areas in the middle or western of China are less likely to be employed in the labor market.Selection terms seem not significant to the public sector while have a significantly positive impact on wages of the private.The sectoral wage differential is merely 18%，61% of which can be attributable to segmentation effect，with performance varying by gender.These show that the new labor market has traits of unequal opportunity and segmentation as well，though the public-private sector wage gap for graduates seems small，which may partially explain the phenomenon of“civil servant fever”.

Chapter 6 empirically analyzes the wage differentials between differ ent levels of cities，and tests the reasons for that，such as ability sorting，living cost differences，or urban agglomeration economy.I find that，the above hypotheses can，to different degrees，explain the wage premia in municipalities and sub-provincial cities compared to prefec tures，especially the agglomeration economy and living cost hypotheses，which is different from the facts that ability sorting and urban agglomeration effects are important in developed counties.The robustness checks substituting category variable to urban population measure further support the above results.

Chapter 7 investigates the magnitude of education quality from the view of labor market outcome of college graduates.Using a part of above survey data，I empirically analyze the sources of wage premium for 211 university graduates from various methods such as OLS，IV and MTE.I find that，controlling working ability and family background，education quality return is 47.3% when numbers of 211 universities in each province is acted as instruments of dummy variable for type of graduated university.The return is as high as 39.4%—94.1% using ATE method.The conclusion is valuable for the direction and measurement of fiscal flow for hither education.

The last chapter draws out the conclusions，gives relevant implications，and points out caveats and directions for further study.

The main conclusion of the thesis can further come into the following 2 points：

First，under the condition of almost homogenous human capital in a narrow sense，wage structure issues in gender，industry，sector，city size or education quality aspects still exist，and can attribute more to the nonmarket factors such as discrimination，segmentation，or fiscal funds.This is，of course，providing more challenge and requirement to the classic economic theory and explanation.

Second，the employment of graduates is closely related to their abilities and the whole environment of current labor market.Therefore，it is quite necessary for all graduates，universities，and governments to solve this issue.

The contributions of the thesis are reflected in such three aspects as research angle，methodology，and dataset.

Firstly，from the research angle，I select the employment of graduates as study topic，and investigate the mechanism and characteristics of the new labor market in view of initial wage structure in order to find out the answer to difficulty of graduates’employment.This has important value for both the government guide and the academic field.On the one hand，college expansion policy and financial crisis shock have a heavily negative impact on the employment of graduates，and thus systematically studying the reasons for their employments has strong reality value and policy implication.On the other hand，previous studies on the wage structure cannot get rid of the influence of issues such as education attainments，working experiences，and family responsibility on the wage incomes，thus the conclusions can be questioned in a strict sense.Taking the wage structures of graduates graduated after half a year as study object，this paper can，to a great degree，avoid the above contamination to find out the factors and reasons except human capital，which is apparent in the innovative view.

Secondly，the methodology is advanced and creative.For example，although there are many literatures both in home and abroad about the wage differentials between public and private sectors，they almost ignore the selection bias arising from the non-participants.The paper uses the double selection framework proposed by Meng and Schimidt（1985）to analyze the wage gap between sectors for graduates，which is more convincing to us.For another example，Appleton decomposition is compli cated，but it can deal with the index number problem and sample selection issues.Marginal treatment effect is not only more consistent to estimate higher education quality return，but also describe its heterogeneity，which is advantageous to the traditional method such as average treatment effect.Hence，I try to use the above two methods to investigate the gender wage gap for graduates and wage premium of 211 graduates in the thesis，respectively.

Last but not the least，the dataset is unique and has good statistic properties.Most previous literatures stay the level of macro phenomenon description and data analysis.As is known，the aggregated data cannot notice the micro mechanism of economic phenomena，which may often draw out misunderstanding or even wrong conclusions.Although the microdata are also employed by several studies，they are often restricted to certain districts or periods，which cannot infer more and thus have not enough policy implications.In contrast，2009 survey data of Placement and Ability of Chinese College Graduates is used in the thesis，which goes across nearly the whole provinces in China and the survey date is after leaving school for half a year and basically in the severe stage of economic crisis.It is worthwhile to point out that advanced questionnaire for ability measurement at international standard is constructed and more accurate results can be drawn out，which，to a great extent，broaden the depth and breadth of literatures on the employment of graduates.

Key Words：College Graduate；Wage Structure；Mean Decomposition；Ability；Double Selection；Marginal Treatment Effect