This thesis consists of three chapters on post-reform Chinese business cycles and alternative methods for solving non-linear rational expectations models. Using quarterly data for the period 1980-2009, Chapter 1 examines the effects of aggregate demand and supply shocks on aggregate fluctuations in China. It further decomposes demand shocks into money supply, money demand and fiscal shocks as in the IS-LM-PC model by applying both long- and short-run restrictions in the context of the structural VAR proposed by Galí (1992). The results show that the estimated impulse responses, in terms of the supply and the three demand shocks, match well with the predictions of the theory. However, as the forecast error variance decompositions show, supply shocks are the main source of fluctuations, accounting for about 89% of output variations in the short-run. Given the nature of this transition economy, this may indicate that there are still institutional obstacles due to incomplete economic reform which prevents the market mechanism from working fully. Despite the overall dominance of supply shocks, the historical decomposition of the five cycles in output between 1983 and 2009 detects important roles played by various demand shocks in some sub-periods. The above results are robust to alternative choices of data for money and interest rate. In Chapter 2, an RBC model with utility generating government consumption and productive public capital is calibrated to annual Chinese data for the post-reform period 1978-2006. The main findings are: (i) the model generates a reasonable overall account of the business cycles in the Chinese economy; (ii) TFP shocks mainly contribute to the good fit of the model, whilst the two fiscal policy shocks help to further improve the model's performance; (iii) our results are robust to alternative calibrations such as high and low capital shares, weights of components in utility and constant return to scale aggregate production function in public capital; and (iv) the shock to the ratio of government consumption to output delivers a dominant negative wealth effect, whilst the shock to the ratio of government investment to output can generate significant positive wealth effects in both the short- and long-run. The third chapter solves the benchmark New Keynesian model using the log-linearization method, second order approximation and the parameterized expectations algorithm (PEA). The results show that the three solution methods display varying degrees of quantitative differences in simulated population moments, distributions, policy functions and impulse response functions. In particular, the generated price dispersions are significantly different across solution methods. The accuracy evaluations in terms of Judd's criteria and Marcet's statistical test show that the PEA performs better than the other two methods, particularly when solving the price-adjustment equation. This result is robust to a number of alternative calibrations.
展开▼
