【作者】 郭凤鸣；

【导师】 张世伟；

【作者基本信息】 吉林大学 ， 数量经济学， 2011， 博士

【摘要】 中国城镇劳动力市场中性别工资差异的经验研究劳动力市场中的性别工资差异一直是劳动经济学研究的主题之一。目前,由于微观数据的缺乏,关于中国劳动力市场中性别工资差异的相关研究相对较少。本文在对工资差异理论及其经验研究方法进行系统梳理的基础上,应用微观经济计量方法,对中国城镇劳动力市场中性别工资差异的表现特征及影响因素进行了系统的分析。在完全竞争的劳动力市场条件下,工资异质性是由劳动者之间影响生产率的能力差异或工作环境特征差异导致的。人力资本理论认为,人力资本差异将导致不同劳动者之间的工资差异;补偿性工资差异理论认为,工作环境特征的差异可能导致不同劳动者之间的工资差异。在不完全竞争的劳动力市场条件下,劳动者之间的工资差异可能不仅仅反映生产率的差异。歧视理论认为,劳动力市场中生产能力相同的个体,可能仅仅由于从属于不同的劳动力群体而得到不同的工资。因而,歧视可能是导致非竞争性劳动力市场中工资差异的一个重要因素。劳动力流动障碍有关的歧视理论认为,买方独家垄断、搜寻成本、拥挤和市场分割等影响劳动力流动的因素可能导致劳动力市场歧视;而统计性歧视理论认为,企业不能了解劳动者生产率的完全信息也可能导致劳动力市场歧视。依据劳动力市场工资差异理论,经济学者提出了不同劳动力群体间工资差异的分解方法。传统工资差异分解方法通常将工资均值差异分解为两部分,分别为个体特征差异导致的部分和劳动力市场歧视导致的部分。随着对工资差异研究的深入,经济学者认识到不同劳动力群体的工资分布通常存在明显差异,因而提出工资分布上工资差异分解方法,将工资差异的分解从工资分布的均值转移到工资分布的不同分位点之上,对不同收入群体中的工资差异及其影响因素进行分析和比较。上个世纪80年代初,经济学者结合拥挤假说,提出了考虑职业分割的工资差异分解方法,将工资差异分解为职业内工资差异和职业间工资差异,并进一步将这两部分差异分解成个体特征差异导致的部分和劳动力市场歧视导致的部分,从而为工资差异的分解提供了更加完善的框架。为了分析工资差异的变动,上个世纪90年代,经济学者将工资差异的分解进一步扩展到时间维度上,但由于没有将劳动力市场歧视的影响完全分离出来,这些方法并没有得到经济学者的广泛应用。基于工资差异分解方法,经济学者对不同国家不同劳动力市场中的性别工资差异进行分析,大部分研究结果表明性别歧视是性别工资差异的一个重要来源。本文结合中国实际,提出适合中国劳动力行为特征和劳动力市场环境的工资差异分解方法,并对性别工资差异的成因进行系统分析。首先,基于工资方程的传统模型和固定效应模型,提出工资差异变动的分解方法,并依据2002年CHIP数据和1991—2006年CHNS数据,分析了中国城镇劳动力市场中的平均性别工资差异及其变动。分析结果表明,中国城镇劳动力市场中男性和女性平均小时工资对数差异为0.1744,其中性别歧视导致的工资差异占77.59%,而性别特征差异导致的工资差异仅占22.41%。很显然,性别歧视是导致性别工资差异的主要原因。在1991年至2006年期间,性别工资差异并没有表现出明显的变动趋势,但性别歧视程度在这一阶段表现出明显上升的趋势,而性别间个体特征差异一直呈缩小趋势。这表明,虽然性别间个体特征差异的变化有助于性别工资差异的缩小,但是性别歧视程度的加剧抑制了性别工资差异的缩小。因此,与提高女性的人力资本水平相比,控制性别歧视程度不断上升的趋势,对降低性别工资差异相对更加重要。其次,基于工资方程的分位数回归模型,提出相应的工资差异分解方法,并依据2006年东北地区城市劳动力调查数据分析了工资分布不同分位数上的性别工资差异。分析结果表明,在工资分布的所有分位数上,男性工资水平均明显高于女性工资水平。在工资分布的底部(低收入群体),影响工资收入的男性个体特征优于女性个体特征,性别工资差异一部分是由个体特征差异造成的,一部分是由性别歧视造成的;在工资分布的中部以上(中高收入群体),影响工资收入的女性个体特征优于男性个体特征,性别工资差异完全是由性别歧视造成的。随着工资分布分位数的提升,性别工资差异不断缩小,性别工资歧视也不断缩小,但性别工资歧视对性别工资差异的解释能力却不断增强。因此,针对低收入群体,政府一方面应通过教育和培训提升低收入群体女性人力资本水平;另一方面应致力于设计和实施公平的就业制度和工资分配制度,逐渐消除针对女性的就业歧视和工资歧视。针对中高收入群体,由于影响工资收入的女性个体特征优于男性,提高女性的人力资本水平对减小性别工资差异的作用有限,因此政府应致力于公平就业制度和工资分配制度的制定和实施,逐渐消除针对中高收入水平女性的就业歧视和工资歧视,才能更加有效地缩小中高收入水平群体的性别工资差异。再次,提出基于工资方程多层模型的工资差异分解方法,并依据2002年CHIP城镇劳动力市场调查数据分析了区域经济环境对性别工资差异的影响。分析结果表明,在所有地区,性别歧视都是导致性别工资差异的主要原因。地区市场化水平的提高将导致性别工资歧视增大,而地区失业率的提高将导致性别工资歧视减小。不同地区的性别工资差异都是多种因素共同作用的结果,因而单纯的地区间的比较不能确定地区水平因素对工资差异的影响。无论从歧视水平还是歧视对性别工资差异的解释能力来看,东部地区的性别歧视程度较高,西部地区的性别歧视程度较低,而中部地区的性别歧视程度居中,表明发达地区的性别歧视程度明显高于欠发达地区的性别歧视程度。因此,政府应均衡各地区的发展,在提高欠发达地区的市场化水平和经济社会发展水平并扩大就业的同时,更应注重社会的公平性,加大各地区之间劳动力资源的流动性,对减小地区因素导致的性别工资歧视具有重要的作用。第四,提出基于工资方程双重样本选择模型的工资差异分解方法,并依据2006年东北地区城市劳动力市场调查数据分析了不同所有制部门中的性别工资差异。分析结果表明,国有部门中的性别工资差异低于非国有部门中的性别工资差异,表明随着市场化程度的提高,性别工资差异表现出扩大的趋势。在国有部门和非国有部门中,性别歧视都是导致性别工资差异的主要因素。从歧视水平来看,国有部门和非国有部门中性别歧视导致的工资差异分别为0.1295和0.3007,非国有部门中的性别歧视水平较高;从歧视对性别工资差异的解释能力来看,国有部门和非国有部门中性别歧视分别解释了性别工资差异的180.61%和104.96%,国有部门中的性别歧视对性别工资差异的解释能力更强。两部门中性别歧视的主要来源不同,在国有部门中,性别歧视的主要来自于部门内部的工资歧视,而在非国有部门中,性别歧视主要来自于劳动参与和部门选择带来的就业歧视。因此,设计和实施公平的工资制度,为女性提供公平的岗位竞争、职位晋升机会和工资水平,对于降低国有部门性别工资歧视程度,减小性别工资差异尤为重要;通过扩大劳动需求带动就业,消除女性劳动参与和部门选择的障碍,对于降低非国有部门的性别工资歧视程度则更为重要。第五,基于工资方程的分位数回归和反事实分析的方法,依据2006年东北地区城市劳动力市场调查数据,分析了教育对性别工资歧视的影响。分析结果表明,受教育程度越高的群体内部性别工资差异越小;在所有受教育水平群体中,都存在明显的针对女性的工资歧视,且在低教育水平群体中,性别歧视导致的工资差异较大,而在高教育水平群体中,性别歧视导致的工资差异较小。不同教育水平下歧视的分布曲线表明,教育有助于群体内部工资差异的缩小,大学及以上受教育程度女性所受工资歧视在个体间的差异较小,而高中与中专和初中及以下受教育程度的女性所受歧视在个体间存在明显差异。工资水平越高的女性受到的工资歧视越小,随着工资水平的提高,女性所受工资歧视在其预测工资条件下的分布曲线向负向移动。教育不仅可以通过提升女性个体的人力资本水平来增加其收入,还能从整体上扭转性别工资歧视的分布。因此,政府一方面应致力于设计和实施公平的就业制度和工资制度,逐渐消除针对女性的就业歧视和工资歧视;另一方面应通过大力发展教育和培训事业来提升女性的人力资本水平,尤其是工资水平较低女性的人力资本水平,将有利于缓解城市劳动力市场中的性别工资差异。最后,提出考虑职业分割的工资差异变动分解方法,并依据1995年和2002年CHIP城镇劳动力市场调查数据,分析了性别职业分割对性别工资差异变动的影响。分析结果表明,从1995年到2002年,性别职业分割存在加剧的趋势,性别职业分割的Duncan系数由0.1883提高为0.2238。在1995年,职业内和职业分割导致的职业间工资差异分别占总性别工资差异的69.57%和30.43%,职业内工资差异主要来源于工资歧视,而职业间工资差异主要来源于性别特征差异;在2002年,职业内工资差异和职业分割导致的职业间工资差异分别占性别工资差异的66.56%和33.44%,职业内工资差异完全来源于工资歧视,而职业间工资差异主要来源于就业歧视。1995年至2002年间,男性和女性小时工资对数差异提高了0.0480,占1995年总体性别工资差异的33.97%,性别歧视程度的加剧对性别工资差异的增大具有明显的促进作用。职业内性别特征差异的变动有利于缩小职业内部的性别工资差异,但工资歧视的明显提高使得职业内总体性别工资差异明显增大。职业间可解释的工资差异变动仅占职业间总工资差异变动的3.45%,其余96.55%的工资差异完全由就业歧视的变动导致。因此,政府应注重消除女性就业中所面临的众多障碍,减小就业歧视导致的性别职业分割,是减小性别工资差异的有效途径。本文的研究结果不仅有助于我们加深对劳动力市场运行规律的理解,而且有助于劳动力市场公共政策的评价与设计,具有理论和现实意义。更多还原

【Abstract】 Gender Wage Differentials in China’s Urban Labor MarketGender wage differential in labor markets has always been one of the main subjects of labor economics. At present, due to the lack of micro dataset, the number of related research on Gender wage differentials in China’s labor market is relatively small. Based on systematic study of wage differential theories and empirical methods and by using micro-econometric methods, this paper systematically analyzes the performance characteristics and influencing factors of gender wage differentials in China’s urban labor market.Under the conditions of perfect competitive labor market, wage heterogeneity is due to workers’ability differences or differences in working environment characteristics which affect productivity. Human capital theory suggests that differences in human capital will lead to wage differentials between different workers; compensating wage differential theories suggest that differences in working environment characteristics may lead to wage differentials between different workers. Under the conditions of imperfect competitive market, wage differentials between workers may not only reflect differences in productivity. Discrimination theory suggests that individuals with the same productivity may be paid differently only because they belong to different labor groups. Thus, discrimination may be an important factor of wage differentials in noncompetitive labor market. Discrimination theory related to labor mobility barriers suggests that monopoly, searching costs, crowding, market segmentation and other factors which affect labor mobility may result in labor market discrimination; statistical discrimination theory suggests that the fact that employers cannot acquire full information of employees’productivity may also lead to labor market discrimination.According to the theories of wage differential in labor market, economists have developed some decomposition methods of wage differentials between different labor groups. Traditional wage differential decomposition methods usually divide wage differential into two parts: one part is due to the differences in individual characteristics and the other is due to labor market discrimination. With further research on wage differentials, economists recognized that wage distributions of different labor groups are usually significantly different from each other, so they proposed decomposition methods of wage differential on wage distribution. The core of these methods is to transfer the decomposition of wage differential from the mean to different quantiles of wage distribution and then analyze and compare the wage differential and influencing factors of different income groups. In the early 1980s, based on the crowding hypothesis, economists proposed a new method of wage differential decomposition by taking into consideration of occupational segmentation. In this method, wage differential is decomposed into wage differences within and between occupations and these two differences are further decomposed into differences in individual characteristics and labor market discrimination. This method provides a more perfect framework for wage differential decomposition. In 1990s, in order to analyze changes in the wage differential, economists extended the decomposition methods of wage differential to the time dimension. But these methods have not been widely used by economists because the effects of labor market discrimination cannot be completely separated from other factors.Based on decomposition methods of wage differential, economists analyzed gender wage differentials in different labor markets of different countries and most of the results show that gender discrimination is an important source of the gender wage differentials. In this paper, by taking into consideration of China’s reality, we propose some wage differential decomposition methods which are suitable for the behavioral characteristics of Chinese labor force and labor market environment.Firstly, we propose decomposition methods of changes in the wage differential based on traditional model and fixed effects model of wage equation, and then analyze the average gender wage differential and its changes in China’s urban labor market by using 2002 CHIP data and 1991-2006 CHNS data. The results show that in China’s urban labor market, the average log hourly wage differentials between male and female workers is 0.1744 and gender discrimination can accounts for 77.59% of this difference while gender characteristic differences accounts for only 22.41% of the difference. Clearly, gender discrimination is the main source of gender wage differential. From 1991 to 2006, the gender wage differentials did not show significant changes in trend, but gender discrimination showed a clear upward trend, while gender characteristic differences has shown a narrowing trend. This suggests that although changes in gender characteristic differences can help narrowing the gender wage differential, but the increase of gender discrimination will prevent narrowing the gender wage differential. Thus, compared to the enhancing of female’s human capital level, controlling the increase trend of gender discrimination will be more important in reducing gender wage differential.Secondly, based on quantile regression model of the wage equation, we propose the corresponding wage differential decomposition method, and analyze gender wage differentials on different quantiles of the wage distribution by using labor force survey data in the Northeast cities of China in 2006. The results show that on all the quantiles of the wage distribution, the wage of male workers is significantly higher than that of female. At the bottom of the wage distribution (low-income group), characteristics which affect wage of male are better than that of female, and gender wage differentials are partly caused by individual characteristics and partly caused by gender discrimination; over the middle of the wage distribution and above (middle and high income groups), characteristics which affect wage of female are better than that of male, and gender wage differentials are entirely caused by gender discrimination. With the increasing of quantiles on the wage distribution, gender wage differential and gender wage discrimination are narrowing, but the explanatory power of gender discrimination to the gender wage differential is expanding. Therefore, in order to gradually eliminate employment discrimination and wage discrimination against women in low-income group, on the one hand, the government should enhance female human capital level through education and training; on the other hand, the government should be dedicated to the design and implementation of fair employment system and wage distribution system. Because female characteristics which affect wage are better than that of male in middle and high income groups, the role of enhancing women’s human capital level to decrease gender wage differential is limited, so in order to gradually eliminate employment discrimination and wage discrimination against women and decrease gender wage differential more effectively in these groups, the government should be dedicated to the design and implementation of fair employment system and wage distribution system.Thirdly, we propose a wage differential decomposition method based on multilevel models of wage equation, and analyze the impact of regional economic environment on gender wage differential by using 2002 CHIP dataset. The results show that in all regions, gender discrimination is the main cause of gender wage differential. Raising the level of regional marketization will increase gender wage discrimination, while the increase of unemployment rates will reduce gender wage discrimination. Gender wage differentials in different regions are the result of several factors, so a simple comparison between regions cannot determine the impact of regional factors on wage differentials. As we can see from the discrimination level and the explanatory power of discrimination on gender wage differentials, eastern region has the highest level of gender discrimination and western region has the lowest level of gender discrimination while the gender discrimination in central region is at the middle level, indicating that the gender discrimination in developed areas is clearly higher than less developed areas. Therefore, government should balance regional development. At the same time of increasing marketization, improving economic and social development and increasing employment in less developed areas, government should pay more attention to social justice and increase labor mobility among various regions, which will play an important role in reducing the gender wage discrimination which is caused by regional factors.Fourthly, we present a wage differential decomposition method based on double sample selection model of wage equation, and analyze the gender wage differentials of different ownership sectors by using labor force survey data in the Northeast cities of China in 2006. The results show that the gender wage differential in the state sector is lower than that in the non-state sector, which suggests that with the improvement of the marketization, gender wage differential shows a trend of expansion. In both sectors, gender discrimination is the main cause of gender wage differentials. In terms of discrimination level, wage differential in the state sector and non-state sector due to gender discrimination are 0.1295 and 0.3007 and gender discrimination in non-state sector is higher. In terms of explanatory power of discrimination on gender wage differentials, gender discrimination in the state public sector and non-state sector explains 180.61% and 104.96% of the gender wage differentials respectively and the explanatory power of discrimination on gender wage differential in state sector is higher. The main source of gender discrimination in two sectors is different. In the state sector, gender discrimination is mainly from wage discrimination within sectors, while in the non-state sector, gender discrimination is mainly from labor participation and employment discrimination when selecting sectors. Therefore, designing and implementing fair wage system and providing fair competitive positions, job promotion opportunities and wage levels for women will help reducing the level of gender wage discrimination and decreasing gender wage differential in state sector; increasing employment through expanding labor demand and eliminating female labor participation and sector selection barriers will be more important in reducing the level of gender wage discrimination in non-state sector.Fifthly, based on quantile regression of wage equation and counterfactual analysis, we analyze the influence of education on gender wage discrimination by using labor force survey data in the Northeast cities of China in 2006. The results show that gender wage differentials within group is smaller when the group’s educational level is higher; there is significant wage discrimination against women in all groups, and in the groups with low level of education, wage differential due to gender discrimination is larger, while in the groups with high level of education, wage differential due to gender discrimination is smaller. The distribution curve of discrimination at different educational level shows that education can help narrowing wage differential within groups; difference of wage discrimination suffered by women with universities and higher educational level is small between individuals, while difference of wage discrimination suffered by women with senior high school, secondary school, junior high school and lower educational level is significant between individuals. Women with higher wage suffer smaller wage discrimination. With the increase of wage, wage discrimination against women moves to the negative direction under the predicted wage distribution curve. Education cannot only increase women’s income by enhancing their human capital level, but also reverse the overall distribution of gender wage discrimination. Therefore, on the one hand, the government should be dedicated to the design and implementation of fair employment system and wage system in order to gradually eliminate the employment discrimination and wage discrimination against women; on the other hand, the government should enhance human capital level of women, especially those at the low wage level, by developing education and training with greater efforts, which will help alleviate gender wage differential in urban labor market.Finally, we propose a method of decomposing changes in the wage differential by considering occupational segregation, and analyze the impact of gender occupational segregation on gender wage differential changes according to the 1995 and 2002 CHIP data. The results show that from 1995 to 2002, there is an increasing trend of gender occupational segregation. The Duncan coefficient, which is an indicator of gender occupational segmentation, increases from 0.1883 to 0.2238. In 1995, within-occupation and between-occupation wage differential resulted from occupation segmentation account for 69.57% and 30.43% of total gender wage differential respectively. Wage differential within occupations is mainly from wage discrimination while wage differential between occupations is mainly from gender characteristic differences. In 2002, within-occupation and between-occupation wage differential resulted from occupation segmentation account for 66.56% and 33.44% of total gender wage differential respectively. Wage differential within occupations is all from wage discrimination, while wage differential between occupations is mainly from employment discrimination. From 1995 to 2002, log hourly wage differential between male and female is increased by 0.0480, which accounts for 33.97% of the overall gender wage differential in 1995. The rising of gender discrimination plays a significant role in the increasing of gender wage differential. Change in gender characteristic differential within occupations is in favor of narrowing the gender wage differential within occupations, but a significant increase of wage discrimination makes the overall gender wage differential within occupations increase significantly. The explained change in wage differential within occupations only accounts for 3.45% of the total change in wage differential, and the remaining 96.55% of the wage differential is entirely caused by changes in employment discrimination. Therefore, the government should focus on the elimination of obstacles confronted in female employment and reduce gender occupational segregation due to employment discrimination, which is an effective way of reducing gender wage differential.The results of this paper will not only help deepening our understanding of labor market operation rules, but also help the evaluation and design of labor market public policy, and thus have theoretical and practical significance.更多还原