Nonlinear Time-Varying Panel Data Models: Theory and Practice

This thesis proposes a class of time-varying panel data models with interactive common factors to capture the nonlinearity in panel data. We first develop a novel recursive estimation algorithm for time-varying models with heterogeneity. Its unified ability in estimation is suitably illustrated according to flexible applications on both simulated and empirical studies. Another issue addressed in this thesis lies in time-varying panel data with semiparametric additive factor models. A profile marginal integration method is proposed to estimate unknown trending function, loadings, and factors jointly in a single step. The asymptotic distributions are established for the proposed estimators, including the factors and loadings.