Improving Calibration With Non-parametric Methods

China’s economy keeps a relatively high-speed growth from the reform and opening up, even in the context of the global financial crisis. However, concomitant with the stylized facts of China’s rapid economic growth, the fact is that the volatility characteristics of the process of economic growth. we use the basic real business cycle RBC model to analyze the cyclical phenomenon in the development process of China. This model exploits the random technology which makes economic fluctuate.Kydland and Prinscotte used a new method:"calibration" for the simulation and analysis of the model. The traditional method to estimate the parameter values have many limitations, so this article focuses on the development of calibration methods. In order to estimate parameter more accurate, the author choses to use non-parametric methods-with variable bandwidth local linear estimation method. First, we introduce the concept of the calibration, as well as domestic and foreign research production. Then the nonparametric methods-variable bandwidth local linear estimation, which is suitable for the parameter estimation. Finally we establish a real business cycle model to take advantage of China’s actual data and variable bandwidth local linear estimation method to estimate model parameters and calibration, using the calibrated model to simulate Chinese economic fluctuations.We found that using of non-parametric method to calibration parameters of the simulation results are more accurate than the using parameter method, and found that the level of the fit of model is up to76.05%, so we specify that should use the non-parameter estimation method to improve the fit of the model calibration and give relevant measures from energy reserve, energy serve, energy cooperation and energy up-grade.