| Via the two dimensional spaces of the time and the cross-section, panel data can reflect the regularity of the data variety, also have the characteristics of controlling the individual heterogeneity and reducing the multicollinearity among the regressors, so it is applied extensively to the economic research and becomes one of the hotspots in the econometric research field.This thesis takes the econometric models of panel data as the main, focusing on the linear panel models, the unit root test of panel data and the cointegration test of panel data, and used to the applications correspondingly.For the linear panel models, it emphases the fixed effects model and the random effects model popularly in the empirical study, and introduces the Hausman test based on the parameter estimates. Regarding the unit root test of panel data, we analyze several kinds of methods popularly in the empirical study. We also integrate the theory of panel unit root with the theory of structural break, and construct an unit root test method based on the structural break, then carry on the Monte Carlo Simulation to this method. Considering the factor of the structural break, it concludes that the power of the test is increasing significantly and the magnitude of the power of the test is no longer affected by the break degreeδ. The last, we apply this method to a study on the unemployment rate's stationary test among 15 representative OECD countries, and come to the conclusion that unemployment rate is a stationary process.This paper mainly introduced Kao DF test and ADF test, Pedroni method and McCoskey & Kao method in the chapter of the panel cointegration theory. Finally, using the normal exchange rate and the quarter data of the consumer price index of ten countries--Canada, Japan, France, Australia, Belgium, Finland, England, Ireland, Italy and Spain among 1978-2005, we adopt the panel cointegration theory to verify the purchasing- power-parity theory, and obtain the conclusion of support. |
PDF Full Text Request