A0244
Title: Buffered vector error-correction models
Authors: Renjie Lu - The University of Hong Kong (Hong Kong) [presenting]
Philip Yu - The Education University of Hong Kong (Hong Kong)
Abstract: Buffered autoregressive models are extended to buffered vector error-correction model (VECM). We propose a SupWald test for the presence of buffer-type threshold effect. We derive the null asymptotic distribution, and show how to make use of a bootstrap method to obtain the $p$-value. Least squares estimation is discussed, and the consistency of the corresponding estimators is derived. Furthermore, we discuss three different methods to determine the number of cointegrating vector for each regime. We investigate the effectiveness of the test and estimation by simulation studies. We apply our model to two data sets: the monthly bond rates of United States and high frequency data of futures and spot of S\&P 500. We find the evidences that there exists buffering regime, and buffered VECM is more reasonable than the traditional two- and three-regime threshold VECM.
