matlab对称误差修正方法,一般门限非对称误差修正模型的 估计与检验

Abstract:This paper extended the common threshold asymmetric vector error correction models, and developed a conditional least square estimation (CLSE) strategy in the joint estimation of cointegration vector and threshold variables. This paper also proposed a SupWald test for threshold asymmetric effect, and presented a bootstrap approximation of asymptotic distribution. Results of Monte Carlo simulation showed that the CLSE estimators of cointegration vector and threshold variable display reasonably small bias to their true magnitudes in large samples, and that the finite sample properties of the proposed test were satisfactory with good test size and power. Applying these methods to study the dynamic relationship of HS300 index futures and spot prices, the results showed that relationship between futures and spot prices was stable in long term, but there was threshold asymmetric effect in the error correction terms in short term.