本文利用1990、2000、2010年全国人口普查数据以及2005、2015年全国1%人口抽样调查数据，考察了教育水平与失业之间的关系及其变化趋势，并实证分析了这一关系对教育的期望回报率的影响。研究表明，提高教育水平能够显著降低失业的风险，只考虑工资收入的估计方式低估了教育的回报。降低失业的风险是教育为劳动者带来收益的重要方面，它影响到劳动者人力资本投资的预期回报，进而影响其人力资本的投资决策。 本文采用线性概率模型，估计了教育与失业风险之间的关系。结果显示，1990-2000年，教育降低失业的作用显著上升，2000年大专和大学教育的失业率分别比高中教育低6.9和9.3个百分点；而2005-2015年，教育水平对失业的影响则显著下降，到了2015年它们的失业率差别仅为1.5和3.5个百分点。这是由于高中教育的失业率在2000年以后开始下降，而大学教育的失业率则呈上升趋势。导致大学和大专毕业生与高中毕业生的失业率差距下降的重要原因是高校扩招，扩招改变了大学和大专毕业生以及高中毕业生的相对供给，使得两组人群的失业率差别降低。考虑到劳动力市场的分割，本文就教育对失业的影响还进行了异质性分析，发现两者间的关系随年龄、性别、城市规模以及户籍性质的不同而不同。 此外，本文结合中国家庭收入调查（CHIP）1995、2002、2007与2013年的数据估计得到的（事后）教育回报率，计算了考虑失业率后的教育期望回报率。结果表明，教育的就业回报（降低失业率）是构成期望回报的重要组成部分。我们还发现，2007年以前教育的期望回报率处于增长状态，尤其是1995-2002年这个阶段，教育回报率大幅增长。增长持续至2007年，之后开始呈下降趋势。对于年轻组别（22-29岁）来说，教育的期望回报率从2002年之后就开始下降。总之，无论是事后的教育回报率还是期望的教育回报率，近年来都呈下降趋势。这会降低人们进行人力资本投资的激励。 本文的研究结果对于全面了解我国转型时期的教育回报率以及家庭的教育决策变化具有重要意义。

Based on the national census and population survey data in 1990, 2000, 2005, 2010 and 2015, this paper investigates the relationship between education and unemployment and its changing trend, and analyzes the impact of this relationship on the expected return rate of education. The results show that improving the level of education can significantly reduce the risk of unemployment, and the return rate of education is underestimated. Reducing the risk of unemployment is an important aspect that education brings benefits to laborers. It affects the expected return of labor's human capital investment, and then influences the investment decision of labor's human capital. In this paper, we explore the mechanism that how education affect the unemployment risk from the perspective of human capital investment decision, and estimate the effect of education on unemployment, using linear probability model. The role of education has changed greatly from 1990 to 2015: between 1990 and 2000, the effect education in lowering unemployment risk increased significantly, and in 2005-2015, this effect declined significantly. Studies show that the unemployment rate for college and university is 6.9 and 9.3 percentage points lower than that for high school in 2000, but it dropped to 1.5 and 3.5 percentage points by 2015. It is because that the unemployment rate of university was rising while the unemployment rate of high school was falling. The main reason for the decrease of the unemployment rate gap between higher education and high school is College Open Their Doors Wider, which changes the relative labor force supply of higher education and high school graduates. Considering the segmentation of labor market, we also examined the heterogeneity of the impact of education on unemployment, including how these effects differ with age, gender, city size and household registration type. At the same time, in tandem with the China Household Income Project (CHIP) data for 1995, 2002, 2007 and 2013, which is used to estimate the (ex-post) return to education, we calculate the expected return of education. The results show that the expected rate of return on education increased from 1995 to 2007, and then decline afterwards. For the young group aged 22-29, the expected rate of return on education began to decline after 2002. To sum up, ex-post or expected return to education have declined in recent years. These reduce people’s incentives to invest in human capital. The research results are of great significance for a comprehensive understanding of the return on education and family education decisions in the transition period in China.