Research

Social media have been a potential source of bias because much of the information has been heavily correlated across different outlets. Are people able to correct for the bias even if they are fully aware of the correlation? In a field experiment on Weibo, we exposed Chinese college students to redundant negative posts about civil service jobs. They developed a 0.16 standard deviation more negative view of civil service jobs and were 11% less likely to register for civil service exams, compared to the control group. Our study demonstrates that correlation neglect can affect beliefs and real-life outcomes, including career choice.

Why do people believe in authoritarian leaders, and who tend to do so? I use the World Value Survey to measure people’s preferences towards authoritarianism. I find that, at both the aggregate and the individual levels, when people have greater confidence in the current political institutions, they express higher support for authoritarian leadership of army and/or religious figure, controlling for demographics and the actual quality of institutions. The result is consistent with a simple model that, when people have greater trust to the political system and the elites, they are more willing to delegate their decision rights to the elites.

As part of the recent debate on collectivism vs individualism, this paper argues that the clan culture in China has a positive effect in innovations, development, and income equality. We set up a general equilibrium growth model to show that a preference for accumulating clan assets leads to higher savings, more innovation, higher income and lower inequality. Empirically, we compile a panel data set for 227 prefectures across 17 years in China to measure yearly variation in clan culture using genealogical records. We then use a shift-share instrument to show that a predicted rise in clan culture leads to higher income, lower inequality, and a lower interest rate, controlling for time and region fixed effects. As predicted by the model, most of the effects are through higher savings and lower interest rates.

The literature recognizes social image concern as an important factor, but it is unclear when and how people develop such concern in equilibrium. I set up a model of gift exchange: individuals are matched in pairs to help each other, and they are matched by believed types, which are inferred from their previous behaviors. The theory shows that social image concern arises from the demand to cooperate, and diversity in types increases social image concern. I am currently developing measurements for social image concern in observational data.