Essays in Econometrics

This thesis consists of four independent chapters that focus on two increasingly popular models in the field of econometrics, namely Mixed Data Sampling (MIDAS) models and spatial econometric models. Chapter 1 introduces the MIDAS approach to a panel data model. It presents the asymptotic distributions of the nonlinear least squares (NLS) estimator for the panel-MIDAS model and the least squares (LS) estimator for the standard LS model based on predetermined weights. The utility of the panel-MIDAS model is demonstrated by a Monte Carlo simulation and an empirical application. Chapter 2 presents an application of the MIDAS model to study the economic policy uncertainty (EPU) effect on mortality in the US. The results indicate a significant negative correlation between EPU and mortality. The accompanying analysis reveals that changes in risky health behaviors are consistent with fluctuations in mortality. Chapter 3 proposes a spatial quasi-limited information maximum likelihood (SQLIML) estimation for a cross-sectional spatial lag model (SLM) with additional endogenous variables X. It derives the asymptotic properties of the SQLIML estimator and compares them with the spatial two-stage least squares (S2SLS) estimator. It also introduces a test of the null that X is exogenous in the SLM model. The above theoretical aspects are explored via a Monte Carlo study and an empirical application. Chapter 4 presents an application of the spatial econometric model to study the minimum wage effect on the gender wage gap across income levels in the US. The results reveal that minimum wage increases in a given state narrow the 10th-60th percentile gender wage gap in its own state and its neighboring states. However, there is no evidence that the gender wage gap in the upper tail of the wage distribution is associated with minimum wages.