| This thesis has two major parts. In the first part of the thesis, we propose new semiparametric estimation methods and test statistics to estimate and test econometric models. Specifically (i) We propose the nearest neighbor method to estimate a partially linear model and we establish the root-n (n is the sample size) normality of our proposed estimator. (ii) We propose a consistent test statistic to test a parametric single index model versus a semiparametric single index model. The second part of the thesis contains two empirical applications. The first is the application of semiparametric estimation techniques to cross-country data to examine the so called "convergence hypothesis" which says that all countries, rich or poor, will eventually reach the same growth rates. The second application uses 1989 Labour Market Activity Survey (LMAS) data to study wage-offer and labour supply equations. In particular we are interested in knowing whether semiparametric and parametric estimation results are significantly different from each other. |
