| dc.description.abstract | In the first chapter, the hypothesis that workers are fully compensated in wages for differences in cost of living is tested for four groups of workers with different levels of educational attainment for the years of 2000, 2006, and 2011. Unlike previous studies, based on the 2SLS method, I find that the wage-price elasticity generally increases with educational attainment in the more recent years. The empirical results support the full compensation hypothesis for workers with educational attainment equal to or higher than some college in the more recent years, while workers whose educational attainment is lower than high-school are found to be incompletely compensated in all years. In the second chapter, the Panel Analysis of Nonstationarity in Idiosyncratic and Common components (PANIC) approach with recursive mean adjustment method (RMA) is applied to investigate regional economic convergence in China. Unlike previous research which finds evidence in favor of intra-region economic convergence, based on a panel set of real per capita GDP for 28 provincial unit in China from 1978-2012, after bias reduction, common factors are found to be nonstationary for China and its three sub-regions. Most of the idiosyncratic components are also found to have a unit root. My results show then that regional economic clubs do not exist in China; thus, reflecting the problem of provincial growth divergence in China. In the third chapter, a new G2SLS approach proposed by Lee (2007) and developed by Bramoulle et al. (2009) is first applied to investigate spillover effects of counties' employment growth, initial fiscal policy variables and initial employment density for the U.S. Unlike the conventional Spatial-Durbin IV model, according to the results of the G2SLS approach, no evidence of spillover effects of employment growth is found; positive spillover effects on employment growth is found for initial safety expenditures in 2000-2007 and 2000-2010, and for initial high-tech employment share in 2000-2010. Initial log of county employment density is found to lower its own employment growth rate. The Monte Carlo simulation results imply that, based on the group interaction structure in the sample, the G2SLS approach provides credible identification of the model parameters. | |
