The Application Of VAR-GC-SSAEPD Model On The Phillips Curve

Zhu(2009)proposed an asymmetric exponential power distribution that is different from the normal distribution.A large number of scholars believed that this distribution could better present the characteristics of data's distribution,as well as the characteristics of the tail.This VAR-GS-SSAEPD model is a new vector autoregressive model which is standardized by the Gaussian connection function.This paper introduces this model into the study of the Chinese Phillips curve.It is concluded that VAR-GS-SSAEPD model better shows the negative relationship between inflation and the unemployment rate in China.The data are the unemployment rates and inflation rates from 1999 to 2017 in the China and maximum likelihood estimation method(MLE)is used to estimate the parameters of VAR-GC-SSAEPD model.Comparing the Akaike Information Criterion(AIC),Schwarz Criterion(SC)and residuals,we can see that the new model better fits the actual data than the classic VAR.The priority of the model is shown by,firstly,the smaller SC value and AIC value generated by VARGC-SSAEPD model.Usually the smaller the SC value and AIC value,the better the model.Secondly,the impulse response function generated by the VAR-GC-SSAEPD model,which shows the tradeoff between inflation and the unemployment rate in China lasts longer and is more significant than those of the VAR model with normal distribution.Furthermore,we apply VAR-GC-SSAEPD to the study of Philips curve in the United States,Germany,Japan and Korea.By comparing with vector autoregressive model(VAR),we finally find that the fitting effect of VAR-GCSSAEPD model is better than that of VAR model in four countries,which verifies that the fitting effect of VAR-GC-SSAEPD model on Philips curve in four countries is better than that of VAR model.Impulse response function can better reflect the substitution relationship between inflation rate and unemployment rate.In a word,we proved the superiority of AR-GC-SSAEPD model over VAR model.