Matrix Completion, Counterfactuals, and Factor Analysis of Missing Data

This article proposes an imputation procedure that uses the factors estimated from a tall block along with re-rotated loadings wide to impute missing values in panel of data. Assuming strong factor structure holds for full data and its sub-blocks, it is shown common component can be consistently at four different rates convergence without requiring regularization or iteration. An asymptotic analysis estimation error obtained. application our counterfactuals when potential outcomes have structure. We study average individual treatment effects on treated establish normal distribution theory useful hypothesis testing.