Research

Working Papers in Economics

1. Dynamic Matching Mechanism and Matching Stability in College Admissions: Evidence from Inner Mongolia (with Wei Ha, Marty Haoyuan Chen, and Yuhao Deng)

3rd Round Revise & Resubmit at Journal of Economic Behavior & Organization

We present the first large-scale empirical evidence on the effects of adopting a dynamic matching mechanism on matching stability in college admissions in China. Initially introduced in 2007 and fully developed by 2010, Inner Mongolia’s “Real-time Dynamic Mechanism” allowed college applicants to modify their choices multiple times during a restricted time interval while observing tentative admission outcomes after each adjustment. Using administrative data on the National College Entrance Exam (NCEE) takers from 2005 to 2011, we construct measures of justified envy, an indicator of matching stability. Using a generalized difference-in differences framework, we find that students justifiably envy a smaller proportion of competing applicants under the real-time dynamic mechanism than under the Immediate Acceptance (IA) mechanism. However, no evidence suggests the dynamic mechanism is better than IA in reducing the likelihood of having justified envy and the degree of envy across various dimensions. Furthermore, we show that the real-time dynamic mechanism is less effective in eliminating justified envy than the parallel mechanism, a hybrid of IA and the Deferred Acceptance (DA) mechanism, which is now widely adopted in other provinces in China. 

2. Matching Mechanisms, Justified Envy, and College Admissions Outcomes (with Wei Ha, Yang Song, and Sen Zhou)

Matching mechanisms are crucial in centralized school choice and college admissions. This paper examines Chinese college admissions, the worlds largest matching market, and compares the Immediate Acceptance (IA) mechanism with the parallel mechanism, a variant of Deferred Acceptance. Using administrative data and staggered provincial reforms, we nd that switching from IA to the parallel mechanism reduced unfair assignments, measured by justi ed envy. Students from low socioeconomic backgrounds experienced higher levels of justi ed envy, but the reform closed this gap. Additionally, the reform led to a reduction in retake rates, a revealed preference measure of student dissatisfaction with admission outcomes. 

3. The Value of Late Specialization in Human Capital Accumulation: Evidence from China’s Meta-major Reform (with Di Wang, Xu Wei, Xiaoyang Ye, and Yi Zhou)

The timing of academic specialization in higher education plays a crucial role in human capital accumulation. However, policymakers often lack sufficient information about students' preferences for different specialization systems and the trade-offs associated with each system to make appropriate decisions. Using administrative data from China’s National College Entrance Examination, we examine a nationwide meta-major reform to study whether students prefer late academic specialization. We find that admission scores for specialties within meta-majors are significantly higher, revealing students’ strong preferences for late specialization. Preferences are stronger for meta-majors offered by prestigious institutions and covering more subjects. Male and urban students exhibit stronger preferences for meta-majors. Meta-major students show greater satisfaction with their majors and schools, are less willing to transfer, and expect higher wages than those in single-subject programs. These findings suggest late specialization reduces student-major mismatch and enhance human capital accumulation. 

4. Gone with the Aid: How Paired Assistance Program Affects High-Stakes College Choices (with Shuo Li, Qinrui Xiahou, and Peng Zhang)

This paper examines how interregional partnership programs influence migration intentions. Leveraging China’s 2008 Sichuan Earthquake reconstruction initiative as a quasi-random experiment, we employ a pair-level difference-in-differences design that matches affected counties with exogenously assigned donor provinces. This approach identifies the causal effect of exogenously assigned aid partnerships on skilled migration intentions, proxied by college enrollment decisions. We find that students from aided counties became 40% more likely to enroll in donor-province universities following the aid program, with an immediate enrollment spike that gradually diminishes over time. The effect is especially strong among rural students and those pursuing unfamiliar majors or attending unfamiliar universities, suggesting that improved information access plays a key role. Back-of-the-envelope calculations show that attending donor-province universities raises the likelihood of graduates working in these regions, resulting in a 26.43% increase in projected lifetime earnings. These results highlight how temporary interregional partnerships can systematically redirect human capital through education, with lasting implications for the redistribution of economic outcomes.

5. Unveiling Educational Inequality: The Divergent Effects of School Consolidation in Rural China (with Naijia Guo, Shuangxin Wang, and Peng Zhang)

This paper examines how school closures in rural China affect children from different socioeconomic backgrounds. Using a difference-in-differences approach that leverages cross-county variations in expected school closures, we find that high-SES counties experience short-term increases in primary and middle school enrollment and long-term educational gains, while children from low-SES counties experience immediate declines in enrollment and reduced attainment. The analysis shows that closures improve educational quality but also raise costs. High-SES families can better afford these costs and benefit from quality improvements, whereas low-SES families face higher financial burdens that limit their access. Overall, school closures may inadvertently worsen regional inequalities. 

6. Information Access and College Admission Outcomes (with Li Chen and Yang Song)

This paper provides quasi-experimental evidence on the effects of information access in centralized college admissions. The variation in access to information stems from a batched dynamic matching mechanism adopted in the Chinese province of Inner Mongolia. Students are grouped into batches based on test scores and face different deadlines to finalize their choice submission: students just above the cutoffs (JACs) finalize their choices along with their higher-scoring peers and thus only observe the tentative choices, whereas students just below the cutoffs (JBCs) can observe the finalized choices of higher-scoring peers. Using administrative data linked with university post-graduation outcomes, we find that JBCs are more likely to apply for colleges considered reach options and are admitted to significantly more selective colleges. These colleges are more likely to be out-of-province and with graduates having higher starting salaries and higher graduate school attendance rates. The effects are more pronounced near batch cutoffs where access to such information is more important. 

7. From Rural Schools to City Factories: Assessing the Quality of Chinese Rural Schools (with Eric A. Hanushek, Xueying Li, and Lei Zhang)

The changing pattern of quality in China’s rural schools across time and province is extracted from the differential labor market earnings of rural migrant workers. Variations in rates of return to years of schooling across migrant workers working in the same urban labor market but having different sites of basic education provide for direct estimation of provincial school quality. Corroborating this approach, these school quality estimates prove to be highly correlated with provincial cognitive skill test scores for the same demographic group. Returns to quality increase with economic development level of destination cities. Importantly, quality appears higher and provincial variation appears lower for younger cohorts, indicating at least partial effectiveness of more recent policies aimed at improving rural school quality across provinces. Surprisingly, however, provincial variations in quality are uncorrelated with teacher-student ratio or per student spending. 

8. Government Bans, Household Spending, and Academic Performance: Evidence from Private Tutoring in China  (with Marty Haoyuan Chen, Wei Lu, Yi Wei, and Jingyi Xing)

This paper studies China’s nationwide ban on private academic tutoring for students in compulsory education. Using a novel nationally representative household panel survey on education expenditure from 2017 to 2023, we find that the ban was associated with a substantial decline in private tutoring spending, but a larger increase in in-school education spending. Households in the lower and middle tiers of the income distribution responded to the ban with lower private academic tutoring spending and students from these families were less likely to be ranked at the top of the class. Further analysis reveals that (i) 87.5 percent of private academic tutoring facilities exited the market after the ban, (ii) the declines in private tutoring expenditures were concentrated among urban households, and (iii) highly educated parents may turn to home tutoring as a substitute to private tutoring. Our findings highlight the substitution toward alternative forms of academic support under the ban and the heterogeneous impacts of the policy.