Testing Common Nonlinear Features in Nonlinear Vector Autoregressive Models
2013/03/19
研討會日期 : 2013-03-19
時間 : 14:30
主講人 : Professor Changli He
地點 : Conference Room B110
演講者簡介 : Changli He received his Ph.D. in Economic Statistics from Stockholm School of Economics in 1997. He is currently serving as Professor of Statistics at Dalarna University. His research field is in time series econometrics, financial econometrics and nonlinear econometrics.
演講摘要 : The paper studies statistical properties for a special class of vector smooth-tansition autoregressive (VSTAR) models containing common nonlinear features (CNFs). A proposed testing procedure for testing CNFs in a VSTAR model consists of two steps: the first step is carried out by a Dickey-Fuller (DF) type test, testing the null hypothesis of unit root process against a stable STAR model for each time series, and then examine whether CNFs exist in the VSTAR model by Lagrange Multiplier (LM) test if unit root is rejected in the first step. The asymptotic distribution of the LM test is derived and the finite-sample statistical properties of those tests, DF-type and LM-type tests, are studied by Monte Carlo simulations. The theory in the paper is illustrated by an empirical example using the monthly growth of consumption and income data of United States (from 1985:1 to 2011:11), and it is found that the model for modeling of the system of US consumption and income during the studied period contains a common nonlinear feature.
