作为人力资本形成的重要途径之一，教育是促进个人提升的内在发展动力，是决定个人收入水平和社会经济地位的重要因素。教育的代际流动水平是衡量教育资源分配和教育公平性的重要指标，影响着收入、职业及社会阶层等多方面的代际流动水平。婚姻选择过程中的教育匹配不仅对个体家庭资源和财富集聚产生影响，还通过代际传递影响子代的教育差距，扩大教育不平等，从而对整个社会结构产生重要影响。

本文利用1995年、2002年、2013年、2018年CHIP城镇微观数据，基于夫妻婚姻选择过程中的教育匹配视角，研究我国婚姻匹配模式及其演变，以及其对教育代际流动、子代教育机会不平等的影响。通过建立夫妻教育匹配列联表和构造相对教育匹配测度表，发现婚姻选择具有明显的同向教育匹配倾向，并且在城镇地区呈现增长趋势。即使控制了社会总体教育水平提升的变化，城镇地区婚姻的同向匹配也非常普遍。

建立实证模型分析婚姻教育匹配对子代教育水平的影响，得出相对于父母低教育水平-低教育水平的教育匹配模式，父母高教育水平-低教育水平、低教育水平-高教育水平、高教育水平-高教育水平的匹配模式使得子代教育年限分别提高了1.282年、1.197年、2.032年，父母高教育水平-高教育水平的匹配对子代受教育年限的影响效应最大，高教育水平的夫妻双方对子女的人力投资上会形成比较一致的观念，更加重视子女的教育投资。总体来看，婚姻匹配模式显著降低了教育代际流动增强的程度，扩大了教育不平等。结果进一步表明，高教育水平的婚姻匹配导致家庭资源增加，并且父母更可能陪伴子女共同生活，这是本文研究发现的潜在机制。此外，婚姻教育匹配的影响效应对城镇地区不同性别、不同户籍子代无显著差异，在中西部地区影响作用强于东部地区，中西部地区的教育机会不平等现象更加明显。相对于1995年和2002年，2013年和2018年婚姻匹配对子代教育水平的影响显著提高，但高教育水平的同向匹配对子代教育水平的正向影响在2018年有下降趋势。

婚姻教育匹配对子代升小学、小学升初中均无显著影响，但高教育水平的婚姻匹配显著提高子代获得中等和高等教育的可能性和就读重点高中的概率。同样地，婚姻匹配对子代升学率的影响无性别差异，但父母高教育水平的同向匹配模式对城镇户籍子代享有高等教育的影响大于农村户籍子代，对中西部地区的效应强于东部地区。

通过反事实分解，构造随机教育匹配下的子代教育分布，发现2013年、2018年的子代教育分布受婚姻教育匹配的影响更大，父母教育随机匹配使子代教育基尼系数减小了26%-28%。最后根据本文研究结果，提出提高教育代际流动，缓解教育不平等的政策建议。

Education, as one of the important ways to form human capital, is the internal driving force to promote individual’s development, and it is an important factor that determines the level of personal income and social and economic status. The level of intergenerational mobility in education is an important indicator to measure the distribution of educational resources and the equity of education, and it affects the level of intergenerational mobility in income, occupation and social class. The educational matching in the process of marriage selection not only affects family resources and wealth agglomeration, but also affects the children’s educational gap through intergenerational transmission. It expands educational inequality, and even has an important impact on the entire social structure.

This paper uses the CHIP data in urban areas for 1995, 2002, 2013, and 2018. We study the pattern of assortative marriage and its evolution in China, and its impact on intergenerational mobility in education and children’s educational opportunities inequality. By establishing a contingency table of educational distribution for married couples and constructing a relative education matching measure, we found that the proportion of spouses with the same education levels is striking and show an increasing trend in urban areas. Even if controlling the increasing education level in the overall, the same education level matching in urban areas is very common.

By establishing an empirical model to analyze the impact of marriage matching on the education level of children, we conclude that: compared with the education matching mode of parents with same low education level, the education matching mode of high education level of father and low education level of mother has increased the children’s education years by 1.282 years, and the education matching mode of low education level of father and high education level of mother has increased the children’s education years by 1.197 years, and the education matching mode of high education level of father and high education level of mother has increased the children’s education years by 2.032 years. The impact of the education matching mode of parents with same high education level on children’s education years is the biggest. Couples with the same high education level will form a relatively consistent concept in their children's human investment decisions, and they both value their children's education investment. It indicated that the marriage matching has significantly reduced the intergenerational mobility in education and has expanded educational inequality. The results further indicate that the education matching mode of parents with same high education level leads to increase family resources for children, and parents with same high education level are more likely to accompany their children to live together, which we interpret as potential mechanisms behind our findings. In addition, the effect of marriage matching has no significant difference for children of different genders and different household registrations in urban areas. The influence in the middle and western regions is stronger than it in the eastern regions. The inequality of education opportunities in the middle and western regions is more obvious. Compared with 1995 and 2002, the impact of marriage matching on the education level of children in 2013 and 2018 has increased significantly, but the positive effect of the education matching mode of high education level of father and high education level of mother on the education level of children has a downward trend in 2018.

The empirical analysis of marriage matching on the children's enrollment rate shows that it has no significant effect of matching on the promotion to elementary school and elementary school to junior high school. The marriage matching mode of parents with high education level has significantly increases the possibility of children's access to secondary and higher education, as well as the probability of attending key high school. There is no gender difference in the effect of marriage matching on the children's enrollment rate. The marriage matching mode of parents with high education level has a greater impact on the enjoyment of higher education by urban household registration children than rural household registration children, and has a stronger effect on the middle and western regions than eastern area.

Through counterfactual decomposition, the distribution of child education under random education matching was constructed. We found that the distribution of child education in 2013 and 2018 was more affected by marriage matching. Random matching reduces the Gini coefficient of children’s education by 26% -28%. Finally, based on the above research results, we propose political recommendations to improve the intergenerational mobility in education and alleviate educational inequality.