Essays on weak instruments and dynamic panel data with applications to pollution regulation

This thesis consists of three essays on weak instruments methodology in a dynamic panel setting and their applications to labor economics and pollution regulation.; The Generalized Empirical Likelihood (GEL) methodology has been suggested to improve small sample performance of Generalized Methods of Moments (GMM). The GEL includes empirical likelihood (EL), exponential tilting (ET) and continuous updating (CU) estimators and test procedures. The first essay surveys the latest development in the GEL literature.; In the second essay, in order to improve the small sample performance of the Generalized Empirical Likelihood Kleibergen type tests (GELK), we propose to reweigh the variance of moments matrix with GEL probabilities. Our modification improves GELK significantly by cutting the size distortion in half. Using simulations, we compare the performance of our modified tests with Kleibergen (2002) and the original GELK tests in a dynamic panel setting. As an empirical application, we use the Arellano and Bond (1991) dynamic panel data to estimate labor demand. We compare our results with the traditional Wald test to illustrate the practical importance of using tests which are robust to weak instruments in a dynamic panel setting.; In the third essay, we study the issue of pollution abatement and "environmental justice" in a dynamic panel setting using the data for 236 plants of the US pulp and paper industry observed over the period 1985-1997. We suggest a model for the typical plant manager who incorporates the regulatory pressures into his calculations of the optimal amount of pollution and the actual pollution abatement which exhibits sluggish adjustment phenomenon. The model leads to the empirical AR(1) panel model. We estimate our dynamic model using both 'temporally lagged' and 'spatially lagged' economic instruments. We find that higher percentages of children and people below poverty line lead to higher pollution while higher percentage of elders leads to lower pollution. These findings confirm the existence of differential treatment based on population characteristics. However, it is not clear whether it comes from actively discriminating regulators or it is because the disadvantaged population groups lack the ability to impose political pressures on the regulators to achieve pollution reduction.