A structural forecasting model for the Chinese macroeconomy

In this thesis, we construct a structural vector autoregression model for the Chinese macroeconomy (CMSVAR) in neoclassical representative-agent framework. By taking on widely-used techniques of identification for structural VAR, we estimate a 9-variable and a 10-variable CMSVAR to predict the economic growth trend. Root Mean Squared Error (RMSE) and Mean absolute forecast percent error (MAPE) both demonstrate that CMSVAR performs superiorly to the Litterman BVAR and the Gibbs BVAR in short or medium run. By the vector error-correcting model (VECM), we also employ the Chinese quarterly real output, price and the money supply over 1992-2008 and find out superneutrality of money in short run: the money growth does not affect the output even in short run. In order to explore the policy implications, we investigate the fixed asset investment model and the nominal interest rate model. Finally, a theoretical model is presented to integrate money into the neoclassical growth model to show the interrelationship between liquid asset and physical capital in the dynamic economic growth. The dynamic equilibrium and policy implications in this economy are also discussed