A0191
Title: Panel data estimation and inference: Homogeneity versus heterogeneity
Authors: Fei Liu - Nankai University (China) [presenting]
Abstract: An underlying data-generating process that allows for different magnitudes of cross-sectional dependence is defined, along with time series autocorrelation. This is achieved via high-dimensional moving average processes of infinite order (HDMA($\infty$)). The setup and investigation integrate and enhance homogeneous and heterogeneous panel data estimation and testing in a unified way. To study HDMA($\infty$), the Beveridge-Nelson decomposition is extended to a high-dimensional time series setting, and a complete toolkit set is derived. Homogeneity versus heterogeneity is examined using Gaussian approximation, a prevalent technique for establishing uniform inference. For post-testing inference, central limit theorems are derived through Edgeworth expansions for both homogeneous and heterogeneous settings. Additionally, the practical relevance of the established asymptotic properties is showcased by revisiting the common correlated effects (CCE) estimators and a classic nonstationary panel data process. Finally, the theoretical findings are verified via extensive numerical studies using both simulated and real datasets.
