| Abstract |
Treatment effect heterogeneity has attracted wide attention in recent studies, particularly in the scope of causal inference. This paper proposes a test for detecting the unobserved heterogeneous treatment effects within fuzzy regression discontinuity (FRD) designs. The null hypothesis of no unobserved heterogeneity is formulated to derive the test statistics. The proposed test allows the treatment effect function as a more general form than previous literature. The asymptotic properties of proposed test are established by the asymptotic analysis of U-process. The finite sample performance is validated by Monte Carlo experiments, exhibiting the appropriate size and power under both the null and fixed alternative hypothesis. |
| About the speaker |
Xiaojun Song is an associate professor of business statistics and econometrics at Guanghua School of Management of Peking University. His field of specialization is theoretical econometrics, focusing on nonparametric/semiparametric methods, hypothesis testing, and bootstrap methods. His articles have appeared in Biometrics, Econometric Theory, Journal of Applied Econometrics, Journal of Business & Economic Statistics, and Journal of Econometrics. He is currently serving as associate editor at the Economic Modelling.
|
