Research

Meritocracy in Charitable Giving:
How College Selectivity Shapes Medical Crowdfunding Outcomes

Working Paper | Download Slides

In medical crowdfunding, college selectivity helps attract donations by increasing the recipient’s perceived deservingness.

First in Line: The Name Order Effect in Top-Down Political Selection

Working Paper | Download Slides

Behavioral and institutional biases appear to favor Chinese officials with early-listed surnames in promotion.

Social Learning in Policy Making

Joint with Yiming Cao

Using large language models to categorize policy documents and exploiting a unique social networking setup among policymakers at China’s Party School, we investigate the diffusion of policy ideas.

Advance Only: Reporting Bias and Perception Manipulation in Official Statistical Communication

Joint with Shengqi Ni | Working Paper | Download Slides | Beamer Template

Statistical manipulation can take the form of selective reporting. We assemble a novel dataset of official annual reports summarizing local socioeconomic performance of Chinese cities, and measure reporting patterns in major indicators using Large Language Models. The authorities avoid explicitly reporting year-over-year declines in favorable indicators (e.g., GDP growth) and increases in adverse indicators (e.g., air pollution). Marginal improvements are approximately 10 percentage points more likely to be explicitly reported than marginal deteriorations. Selective reporting is more pronounced for politically salient indicators。