This paper reports an investigation of the spillover effects of output shocks between regions in China. We use a six-region classification first suggested about two decades ago which still captures relatively homogeneous regions. We start from a recent paper by Groenewold, Lee, and Chen (2005) which uses the same six regions and a vector autoregressive (VAR) framework. They find that the spillover effects are crucially dependent on the order of the variables in the model due to common national influences. They overcome the "ordering problem" by using a two-step procedure. We implement an alternative solution which proceeds by including national output directly into our model. Moreover, we extend their analysis by investigating Granger-causality between regional and national output measures as well as block exogeneity. Our results confirm important conclusions of the earlier paper but also raise some interesting differences.
|Journal||Australasian Journal of Regional Studies|
|Publication status||Published - 2007|