Research

China's 2021 regulation virtually eliminated weekday online gaming for minors and limited play to one hour on weekends/holidays. Using nationally representative CFPS data (2020, 2022) in a difference-in-differences design, I find sharp declines in minors' gaming and total Internet use but no robust improvements in study effort, academic performance, or physical health, with suggestive declines in mental well-being. A complementary Regression Kink (RK) design using city mock-exam data around the age-18 cutoff likewise shows no robust exam effects.

This paper studies how China's 2021 gaming restriction for minors reshaped digital markets using app-level downloads and revenues. Difference-in-differences and event-study estimates show sharp declines in youth-oriented games, smaller effects for adult-oriented titles, and growth in substitutes such as e-books and social media. Usage shifts away from weekdays and toward permitted weekend hours, consistent with strategic adaptation by users and firms. Market responses are heterogeneous, including rebranding, monetization changes, and advertising re-targeting. Ongoing work examines longer-run firm dynamics (entry/exit) and innovation responses.

We study whether highly visible academic "superstars" generate local human-capital spillovers by shifting beliefs and inducing schooling responses. Using microdata from China's 2010 Population Census, we construct county-by-cohort exposure to the first emergence of a provincial Gaokao top scorer (1990--2005) based on the top scorer's high-school location. Exploiting staggered timing across counties, we estimate event-study and staggered difference-in-differences models. Exposure increases educational attainment for cohorts plausibly able to respond -- students one to two years younger than the top scorer. In this window, exposure to a science-track top scorer reduces dropout, raises years of schooling by 0.267 years, and increases tertiary attainment, with four-year college completion rising by 2.9 percentage points.

In 2011, China's Supreme People's Court strengthened individual property rights over premarital housing, reducing expected redistribution upon divorce. We exploit this legal shock to estimate how property rights affect marriage timing in the presence of high housing costs. Using 2010 and 2015 census microdata, we implement a difference-in-differences design combined with entropy balancing. We find a distinct gender asymmetry: the reform significantly reduces marriage entry for young women in high-housing-price cities, while men's outcomes remain largely unchanged. Women in high-price areas experience an approximately 3 percentage point decline in marriage rates relative to low-price cities.

We study how lowering administrative barriers to legal marriage affects marriage timing and match composition in urban China. The 2003 Regulations on Marriage Registration removed key documentary requirements, sharply reducing registration frictions. Using 2010 China Population Census microdata and a monthly regression discontinuity in time design, we find a sizable increase in first-marriage registration at the cutoff of about 3.7 per 1,000 unmarried urban residents per month, with effects persisting beyond the implementation month. The reform also shifts match outcomes: spousal age and education gaps change and cross-jurisdiction marriages rise.

Using China's 1999 higher-education expansion as a cohort-based supply shock, we study rural–urban occupational mobility. We construct an LLM-based occupational socioeconomic-status score from free-text job descriptions in the 2015 1% Population Sample Survey. Post-expansion rural cohorts gain 0.25 points in occupational status relative to urban cohorts (closing approximately 17% of the raw gap). IV evidence supports a causal education channel: each additional year of schooling raises occupational status by about 0.21 points, with upgrading concentrated in moves from low-skill manual work into skilled trades and basic services.