The paper "Inference in Partially Identified Panel Data Models with Interactive Fixed Effects" co-authored by Associate Professor Yaqi Wang from the School of Finance at Central University of Finance and Economics, Professor Liangjun Su from Tsinghua University, and Associate Professor Shengjie Hong from Renmin University of China, was published online in the top international journal of econometric theory, Econometric Theory, in February 2024.

In recent years, the use of interactive fixed effects in panel data modeling has become widespread in empirical research for two main reasons: (i) compared to single-dimensional individual or time effects, interactive fixed effects offer a more flexible way to capture individual heterogeneity, thus reducing the reliance on instrumental variables to some extent; (ii) interactive fixed effects have emerged as the primary modeling approach for panel data showing strong cross-sectional dependence.

This study's key contribution is in relaxing the conditions for applying interactive fixed effects models, allowing for partial identification and considering the presence of endogenous explanatory variables. Specifically, the innovations in this study include: (i) effectively handling high-dimensional parameter vectors' factor loadings under certain normalization rules and specifying a set of conditional moment restrictions involving finite-dimensional factor parameters and infinite-dimensional nonparametric components; (ii) introducing a panel data analysis method suitable for large N and fixed T, applicable to various enterprise or household survey panel data. It is possible to extend the current theoretical framework to conditional moment inequality models through the introduction of some slackness parameter.This will greatly broaden the scope of the current paper. We leave the extensions for future research.