Job Market Paper. [PDF] [Slides]
Abstract: From 1999 to 2020, China undertook one of the world’s largest expansions of higher education, increasing annual enrollment from 4 to 32 million and contributing 18% of global tertiary enrollment growth. This paper examines how this reform affected wages and employment at both the worker and local labor market levels. Using newly assembled city-level data, I exploit the staggered openings of new colleges and universities across cities, which capture a sustained local expansion of higher education and produce more than 118,000 additional graduates per city over a decade. About one-third of the graduates were absorbed locally into high-skill services, manufacturing, and state-owned sectors. Although manufacturing firms grew substantially in scale, there is little evidence of short-run productivity improvements per firm, and the cumulative growth of city-average wages and per capita income was 20 percent lower in treated than in control cities. Worker-level evidence shows that the wage effects were concentrated among younger college graduates: the college premium fell sharply for post-reform entrants but remained stable for earlier cohorts. Using a labor demand model and a wage decomposition framework, I find that college cohorts compete in relatively segmented markets, as implied by a very low elasticity of substitution of 1.5 to 2. The supply effect of larger cohort size, combined with the declining cohort quality, has more than offset the positive demand shifts. The reform effectively closed the college–noncollege wage gap, but also created an emerging wage gap between younger and older workers within the educated workforce.
The Human Capital Effects of Early Childhood Exposure to Community Violence
[Slides] Draft available soon.
Abstract: This paper examines the causal effects of early life exposure to community violence on human capital accumulation, leveraging a unique historical dataset on violence intensity during the Cultural Revolution (1966-1976) in China, matched with individual data from contemporary census and health survey. My identification strategy exploits city-by-cohort variation in the number of unnatural deaths attributable to political violence from 1966-1971. I find considerable detrimental impacts of early life violence exposure on timely school enrollment and finalized educational attainment in adulthood. The adverse effects on educational attainment of a one-standard deviation increases in violence intensity over a 12-months period exposed before age 3 is equivalent to more than 1.16 years of reduction in mother’s years of schooling. This effect is most pronounced for exposure during the intrauterine period and is diminishing with age in early childhood. My results also reveal a significant gender gap, where the negative impacts are mostly driven by girls and are estimated precisely zeros for boys. To explore potential mechanisms, I show suggestive evidence that increased violence exposure at early ages resulted in deteriorated physical health, depression, and some cognitive functions in later life.
Firm Networks, Technology Complexity, and Patent Litigation in the United States
with Ilan Vertinsky and Yong Li
Draft available upon request.
Abstract: This research examines the determinants of patent infringement litigation in the United States through the lens of firm networks and technological complexity. We assemble a novel firm-dyadic dataset that records inter-firm litigation histories and collaboration networks, along with detailed information on patent portfolios and firm characteristics. Using a matched case-control design, we compare actual plaintiff–defendant dyads with observationally similar firm pairs to estimate how a plaintiff firm’s technological complexity and relative position in collaboration networks affect the likelihood of filing a patent litigation case. Our results show that, beyond traditional determinants of litigation such as direct competition and patent similarity, inter-firm collaboration networks play an important role in reducing information search costs associated with increasing patent complexity and ownership fragmentation, thereby moderating litigation risks.