Estimating Parameters And Testing For Threshold Effect Of Threshold Cointegration Model

The analysis of threshold cointegration model in nonlinear time series has been regarded as one of the core areas of research in statistics and econometrics. Recently, numerous studies related to threshold cointegration model have been appear in theory and application.This paper studies parameter estimation and hypothesis testing for threshold effect of threshold cointegration model, and researches the relationship between China' imports and exports. We obtain several results as follows:In chapter 2, we propose two new cointegration models, level threshold vector error correction model and general threshold vector error correction model. Optimal algorithm and maximum likelihood estimation combine to estimate the parameters of threshold cointegration model. The simulated experiment shows genetic simulated annealing algorithm works quite well.Chapter 3 tests for threshold effect of the cointegrating vector and the adjustment vector in the vector error correction model with an unknown threshold. We adopt a Sup-LM statistics for the presence of a threshold, which asymptotic distribution under null hypothesis is derived to be the function of a standard Brownian.In chapter 4, we test the long-run equilibrium between China' import and export according to the relevant data 1998.1-2004.9. Import is the Granger causality of export in cointegrating systems.