Large dimensional latent factor modeling with missing observations and applications to causal inference

This paper develops the inferential theory for latent factor models estimated from large dimensional panel data with missing observations. We propose an easy-to-use all-purpose estimator a model by applying principal component analysis to adjusted covariance matrix partially observed data. derive asymptotic distribution factors, loadings and imputed values under approximate general patterns. The key application is estimate counterfactual outcomes in causal inference unobserved control group modeled as values, which are inferred model. allows us test individual treatment effects at any time adoption patterns where units can be affected factors.