我国农民工与城镇职工就业质量差异及其分解——基于RIF无条件分位数回归的分解法

Job Quality Gap between Rural Migrants and Urban Workers in China and Its Decomposition——Evidence from RIF Quantile Regression and Decomposition Method

本文利用CHIP2002和CHIP2013数据,采用基于RIF回归的分解法,考察不同时间维度和不同分位数下,我国农民工与城镇职工就业质量差异、影响因素及其变动趋势。结果显示：我国农民工与城镇职工就业质量的差距不断缩小,且中高分位数的变化幅度较大。尽管如此,农民工的就业质量仍大大低于城镇职工,且其差距在中高分位数上较大。农民工与城镇职工就业质量的差距是个体特征和制度性歧视共同导致的,在中低分位数上,特征效应是造成就业质量差异的主因;在高分位数上,系数效应是造成就业质量差异的主因。从时间维度看,城镇低端劳动力市场上对农民工的制度性歧视有所减弱,但中高端劳动力市场上的制度性歧视有所增强。提高农民工人力资本,消除城镇劳动力市场的制度性歧视,是我国城镇化健康发展的前提和基础。

The paper employed unconditional quantile decomposition method based on RIF regression,to examine the job quality gap and its determinants in time dimension and in different quantiles. The result shows： job quality gap between rural migrant and urban residents reduced since 2002,greater variations are found in the top than in the median and bottom of the job quality distribution. Yet,job quality of migrants is far lower than urban workers in 2013,and the gap are found to be higher in the top than in the bottom of the job quality distribution both in 2002 and2013. Furthermore,we decomposed the distribution of job quality differentiates between the rural migrants and urban workers into characteristic effect explained by differences in productivity characteristics and coefficient effect（ discrimination effect） explained by the unequal returns to these characteristics. We found that in the median and low quantiles of the job quality gap distribution,characteristic effect is the main reason of job quality differentials; but in the top end of the job quality gap distribution,coefficient effect plays an important role. It means discrimination against rural migrants in the higher labor market is greater. In the time dimension,discrimination in the lower labor market are weakened since 2002,yet it ＇s strengthened in other markets. Thus,rapidly increase the investment and accumulating in rural human capital,diminish the discrimination in urban labor market,is the vital foundation of urbanization in China.