Identification robust testing in linear factor models

This thesis investigates the identification robust tests in linear factor models used in empirical financial studies. It provides novel testing procedures for the parameters of interest in the asset pricing models, the risk premia (Chapters 3, 4), and the model specification (Chapter 2) under empirically relevant settings. We discuss identification robust inference, which provides the correct rejection rate under the null hypothesis regardless of the identification condition. We provide our proposed test procedures' asymptotic/exact finite sample behavior and run simulation exercises to verify their limited samples' performances. Our empirical applications document that the testing results, such as rejection rates, confidence sets, can differ substantially from conventional statistical tools via our proposed testing procedures.