Nonlinear Dsge Method And Its Applications In Monetary Policy Analysis

Over the last decade, the dynamic stochastic general equilibrium model of the new Keynesian theory of monetary policy to occupy an important position in the mainstream macroeconomics and monetary economics research, many researchers, especially central banks researchers widely adopted this framework for policy analysis and evaluation. Most of the existing research using log-linear method of dynamic stochastic general equilibrium model, this research method in the simplified model analysis of the log-linear approximation for variable higher-order terms ignored due to the bias of the model conclusions.Based on existing research criticized for linear dynamic stochastic general equilibrium model, the tentative two types of nonlinear dynamic stochastic general equilibrium model. First nonlinear moving average method has been suggested extensions to enable them to analyze the existing interconnected exogenous shocks, and contains a unit root case of dynamic stochastic general equilibrium model. Extended nonlinear dynamic stochastic general equilibrium model with a linear model differences are analyzed based on simulation data, this paper combined with the impulse response function of exogenous shocks. The research results show that exogenous shocks have a significant impact for the variables of the second-order terms with third-order item, and the variables of the second-order terms and third order for the change of variables has a significant effect. Therefore, the linear dynamic stochastic general equilibrium model ignored the higher-order terms of the variable caused bias.For clearer comparison of the expansion of the non-linear moving average method and linear model differences, constructed a small scale unemployment new Keynesian theory of monetary policy in a dynamic stochastic general equilibrium model to sustained output, and respectively using log-linear expansion of the non-linear moving average method of model processing. Based on the structural parameters of existing theory and empirical values of the the persistent output model calibration. Exogenous shocks after calibration pulse response function shows the output variable component of the second-order and third-order components of the second component has quite strong, persistent, variable responses to exogenous shocks that error correction term for exogenous shocks The effect corrective action; calculated based on the actual data output persistence nonlinear moving average method processing model is closer to the log-linear model significantly exaggerated the continuity of China’s output. Conclusions based on the analysis of the nuclear density function in the second and third-order nuclear density function of the output variables that the second component of the output variables with the third-order component of past information maintained considerable memory characteristics. Thus, when the exogenous shock occurs, the second component of the output variables with the third-order component of the output variable changes significantly affect. Due to the significant presence of the second-order terms with the impact of the third-order terms of the output variables, then ignore the higher-order terms of the variable log-linear model caused by the bias of the model. On this basis, the paper further but not included in the unemployment model contains physical capital investment as well as investment in physical capital is not included but with unemployment continuing analysis of the model of China’s output. The results found in not included in unemployment, but the model contains physical capital investment, the Chinese output persistence maximum; without including physical capital investment, but contains a minimum sustained output model of unemployment. The conclusions of this study indicate that investment in physical capital to promote the continuity of our output, and the existence of unemployment continued to reduce our output.Based on the analysis of the first class of nonlinear dynamic stochastic general equilibrium model, the paper further proposed a second class of nonlinear dynamic stochastic general equilibrium model:the quadratic autoregressive moving average model. On the the quadratic approximation foundation established dynamic stochastic general equilibrium model, this article will be reduced after the Schorfheide (2012) framework as the data generation process, and the model were log-linear quadratic since regressive moving average processing model analysis. The test data based on the Monte Carlo, we can model structural parameters identification test and stability test methods were compared with the log-linear model with quadratic form auto-regressive moving average model. The results showed that, relative to the quadratic form in terms of self-regressive moving average model, the correlation coefficient between the log-linear model Monte Carlo data with realistic data is far less than the quadratic since ARMA model Monte Carlo data The correlation coefficients between the real data; model structural parameters identifiable inspection also found that poor identifiability of the log-linear model. On this basis, the paper further stability test method to test the stability of the two different models. The test results show that, relative to the quadratic form autoregressive moving average model, the log-linear model with a relatively poor stability.Based on quadratic autoregressive moving average model with log-linear model to compare the conclusions, the paper constructs a multi-sectoral, multi-market of new Keynesian theory of monetary policy in a dynamic stochastic general equilibrium model framework, and using a Bayesian approach model Bayesian estimation. Based on Bayesian estimation of structural parameters, respectively, using the Monte Carlo method for quadratic forms autoregressive moving average model, and the log-linear model simulation. The exogenous shock pulse response results show that under the conditions of existence of the unit root, the log-linear model of Chinese and foreign students in response to the impact of the pulse completely opposite trend with realistic data showed a greater bias function, or even the performance of certain variables. Monte Carlo-based test data further second-order moment matching method examines the log-linear model with quadratic form autoregressive moving average model fitting degree of reality. The correlation coefficient of model data with realistic data to match the results show that for most variables, quadratic autoregressive moving average model data and real data correlation coefficient; volatility match concluded also found quadratic autoregressive The moving average model data with the volatility of the real data is closer. This conclusion is further confirmed by a quadratic autoregressive moving average model than the log-linear model better fit the reality of China’s economic data.