Английски език
The first chapter of this dissertation introduces §the generalized minimum contrast estimator (GMC) §with §nonsmooth moment functions and studies its §asymptotic properties under mild conditions. A Chi-§square test based on GMC is discussed. The duality §between the GMC and the Generalized Empirical §Likelihood Estimator (GEL) is interpreted from both §a computational perspective and geometric §perspective. The second chapter investigates a §Bayesian approach to calculating the GMC. By §updating the concerned parameters and nuisance §parameters alternatively, this approach can converge §quickly and can be implemented easily. Its §performance is compared to existing methods based on §both just-identified conditions and over-identified §conditions. The third chapter employs a partial §linear model (PL) and geographically weighted §regression (GWR) to estimate housing prices. These §two semiparametric models provide flexible §consideration of spatial heterogeneity and spatial §correlation. Their performance is compared through §various simulation experiments. An empirical study §of the housing market in Connecticut is also §implemented.
