Semiparametric dynamic panel data model: An application to cross -country income dynamic

My dissertation mainly explores two areas of economics: one is in the field of econometric methodology of dynamic panel models, while the other deals with a currently debated topic in growth economics, convergence. It seeks to extend the range of empirical applicability of dynamic panel data models in general and to the data driven nonparametric (NP) and semiparametric (SP) modeling in particular. Chapter 2 of this thesis starts with a systematic study of existing literature in parametric dynamic panel models To allow for more realistic and complicated patterns of dynamic panel model, chapter 3 develops some nonparametric and semiparametric consistent estimators with desirable asymptotic properties to deal with fixed effect (FE) and random effect (RE) models. Due to lack of prior information on the functional form of the dynamic model, it is realistic to assume that the model contains explanatory variables including lagged dependent variable in unknown functional form and my approach is to estimate these functions by local linear nonparametric methods. A range of both time specific and cross-section specific FE and RE nonparametric and semiparametric models has been proposed, and their consistency and asymptotic normality along with their small sample properties have also been established in this chapter. The second part of my dissertation studies the issue of cross-country income dynamics over time in general and speed of convergence in particular. While most studies in the literature on growth and convergence have controlled for heterogeneity in the intercept given the panel nature of the data, they failed to take care of the functional form problem and the heterogeneity in terms of country specific long run equilibrium in general. Chapter 4 presents a study of the speed of convergence using the newly proposed SP estimators for dynamic panel models. Empirical findings in this study show that there is a tendency for fast convergence (faster than 2%) across countries towards their own steady state level of income per capita. And it's different across countries. I found enough evidence to reject the assumption of common growth rate and the believe that government expenditure on per capita output is always negative.