II. Panel Data Models: Interactive Effects and Factor Instrumental Variables Are Options Mis-Priced? - Analysis of Option Return Premia
講題摘要:
The first part of the lecture begins with discussing classical factor analysis, in which the statistical theory is developed under a fixed N. We then go on to discuss high-dimensional factor analysis (modern factor analysis) where the number of variables (N) is comparable or even greater than the number of observations (T). Issues related to identification and estimation are discussed. In particular, the maximum-likelihood-based and principal components-based procedures for estimating factor models are investigated. We will also present, if time permitted, a few applications of factor models in economics and finance.
The second part of the lecture mainly focuses on panel data models with factor errors. Specifically we first present the panel data models with interactive fixed effects developed by Bai (2009,Econometrica). For ongoing research, we introduce a novel use of factor analysis controlling for endogeneity within the Panel Instrumental Variable Quantile Regression framework.