This article studies gender gap patterns in language and math during primary school periods in China, and investigates candidate theories to explain the observed patterns. Using an administrative test dataset, we find that girls on average outperform boys in language and have advantage in math in primary school. The gaps for language in all quantiles widen consistently over time. Female’s advantages in math in most quantiles, except the lowest, shrink over time. Using a survey on students in primary schools, we find that noncognitive skills can effectively reduce the gender gap in both Chinese and math. This effect is most pronounced for students in the bottom quantile group.

We examine the effect of media persuasion on educational choice, and find that Chi- nese newspaper articles on violence against doctors influence students’ decisions to study medicine at college. We match articles from over 1200 newspapers with an administrative dataset on college entrance enrollment from 2005 to 2011, and find that one additional article on anti-doctor violence leads to a 0.6% decrease in the number of students enrolled in medicine-related majors, and this effect is more pronounced for physician and nurs- ing majors. We perform a series of checks to ensure that the effect is driven by exposure to violence-related news rather than violent incidents themselves. An instrumental vari- able approach that exploits plausibly exogenous variations in local political turnover and province-wide violent incidents helps establish causality. Moreover, we find that exposure to violence-related news reduces the quality of medical students, measured by their rank in the college entrance examination. Our findings suggest that media coverage can change individuals’ perception of career risks and affect their educational choices.

This paper provides empirical evidence on how China’s transition from the Boston mechanism to the Chinese parallel mechanism (a simplified version of the Deferred Acceptance mechanism), along with changes to the information available to students on their entrance exam performance when they submit their college preferences, affect the academic match between colleges and students. Using data on students admitted to Chinese colleges from 2005 to 2011, we characterize the general patterns of mismatch between colleges and students based on students’ scores on China’s National College Entrance Exam and find evidence of substantial overmatch and undermatch. Results from a generalized difference-in-differences model indicate that switching from the Boston mechanism to the Chinese parallel mechanism lowered the probability of mismatch by approximately 6%. Allowing students to submit their college preferences after learning their exam scores rather than before the exam reduced the probability of mismatch by 18%.

Substituting generic for brand name drugs whenever possible has been proposed to control prescription drug expenditure growth in the United States. This work investigates two types of state laws that regulate the procedures under which pharmacists substitute bioequivalent generic versions of brand name drugs. Mandatory substitution laws require pharmacists to use the generic as a default, and presumed consent laws allow them to assume that the patient agrees to the substitution. Both situations can be overruled by the patient. Using plausibly exogenous changes in states’ laws, we use differencein‐ differences and a discrete choice model to show that although the mandatory switching laws have little effect, the presumed consent laws reduce consumers’ probability of purchasing brand name drugs by 3.2% points. The differential effectiveness of the laws is likely caused by pharmacists’ profit motives. These results offer important implications for policies that seek to reduce drug expenditures by incentivizing the use of generic drugs.

This paper examines households’ extrapolative behaviors and developers’ strategic responses in China’s land and housing markets. Our stylized model predicts that when households extrapolate in the housing market, developers with nearby land holdings strategically bid higher, anticipating rising housing prices to boost the value of their existing stock. Using the exogenous timing of land auctions, we show that higher land premiums increase transaction prices and volumes of nearby houses. Furthermore, developers with nearby land stocks consistently pay higher premiums in the land auctions. We exploit a policy reform and implement various tests to support the hypothesis that developers strategically influence homebuyers’ expectations.

We conduct a laboratory school-choice experiment in which admission cut-offs arise endoge- nously, testing correlation neglect in school choice settings. Relative to Rees-Jones et al. (2024) which fixes thresholds ex ante, we show that correlation neglect persists in complex school-choice environments. We evaluate three debiasing tools. First, we provide a reminder highlighting that admission outcomes are correlated. Second, we offer an algorithmic aid that supplies personal- ized admission probabilities, reducing the burden of Bayesian updating. Third, we implement an iterative deferred-acceptance mechanism that reveals rejection outcomes sequentially, sim- plifying contingent reasoning. The reminder has no effect. Providing personalized probabilities substantially reduces aggressive applications, but the reduction does not vary with the severity of correlation neglect. In contrast, the iterative mechanism directly attenuates correlation-neglect errors, especially when correlated risks are large. These findings inform the ongoing global reform of centralized admissions systems and suggest that both informational and structural interventions deserve policy attention.

This paper examines college application behavior in a centralized matching mar- ket where students face demand uncertainty. Using a large field dataset from Chinese college admissions, we provide evidence of demand uncertainty by exploiting varia- tion in assignment mechanisms and other institutional features. We estimate a model of student decision-making with heterogeneous strategies and find that, relative to the benchmark with correct common priors, about 15% of students applied more pes- simistically and 24% more optimistically. The welfare effects of assignment mechanisms are also heterogeneous: optimistic and pessimistic students with high priority benefited from the transition from the Immediate Acceptance mechanism to the Chinese Parallel mechanism.