A1284
Title: Panel data with high-dimensional factors with application to peer-effects analysis in networks
Authors: Yike Wang - London School of Economics and Political Science (United Kingdom) [presenting]
Taisuke Otsu - London School of Economics (United Kingdom)
Abstract: Factor models are widely used in economics to capture unobserved aggregate shocks and individual reactions to the shocks. While the existing literature focuses on models with a small and fixed number of factors, a new method is developed to allow for a large and growing number of factors under sparsity assumptions on the factor loadings. The new approach is called the High-Dimensional Interactive Fixed Effects (HD-IFE) estimator. The conditions under which the new estimator is consistent and asymptotically normal are provided. Then, the HD-IFE estimator is applied to address a common endogeneity issue in peer-effects estimation caused by missing nodes and connections in the sampled network data. The sparsity conditions of the HD-IFE estimator are plausible when networks have sparse links. Empirically, the existence of tacit collusion on price in the Houston gasoline retail market is examined, for which different findings are obtained by using the new estimator and low-dimensional ones.
