Evaluation Of Monetary Policy Effect Based On HMC-Bayes Logistic Regression Model In Dynamic Potential Outcome Framework

Quantitative evaluation of the effect of macroeconomic policy is one of the core issues in many applied fields of economic research and policy research.At present,there are many model methods used to analyze the effects of macroeconomic policies,such as the mainstream structural vector autoregressive model(SVAR)and dynamic stochastic general equilibrium model(DSGE),which rely on rigorous theories and become powerful tools for macroeconomic policy research.However,these methods largely depend on some structural norms of the whole economic system,and the validity of the research results obtained in this way depends on the accuracy of the assumed economic model.Only the accuracy of the model can achieve good results.Thus,the use of new macroeconomic policy assessments can yield some accurate estimates,providing better insights for the further development of relevant structural models for policy simulations.Based on this,this paper applies the dynamic potential outcome framework to the evaluation of China’s monetary policy effects,and creatively combines Hamilton Monte Carlo algorithm(HMC)with Bayesian logistic regression model to propose an estimation method of monetary policy decision mechanism based on HMC-Bayes multiple classification logistic regression model.The reason for adopting dynamic potential outcome framework is that it does not depend on the structural paradigm of the entire economic system,and does not need to model the entire economic system,but only needs to model the policy decision mechanism.Moreover,the dynamic causal effect under this framework is exactly corresponding to the pulse function of the focus content of mainstream DSGE model,SVAR model and other modern macroeconomic models.It can be seen that,The dynamic potential outcome framework can be used to evaluate the effect of macroeconomic policies on the premise of reducing the risk of model missetting.It is a better method to evaluate the effect of macroeconomic policies.The research content is mainly divided into two parts: In the first part,based on the collected data of monetary policy index and high-dimensional macroeconomic operation index that reflect the dynamic changes of Chinese monetary policy,a Bayesian multiple classification logistic regression model is established to fit the decision mechanism of Chinese monetary policy--monetary policy tendency score,and the HMC algorithm is introduced through model transformation and the matrix representation of likelihood function.Parameter estimation results were obtained through a posterior sampling.Model comparison and analysis showed that the HMC-Bayes logistic regression model proposed in this paper not only showed the advantage of efficient operation,but also obtained more accurate prediction results of monetary policy tendency score compared with other algorithms.In the second part,the dynamic potential outcome framework is used to conduct dynamic evaluation of monetary policy effects,and the inverse probability weighting method(IPW)and the improved inverse probability weighting method(AIPW)are used to quantify the impact of our monetary policy on price stability,economic growth and financial stability.The empirical results show that in terms of price stability,monetary policy easing can reduce CPI and PPI index and increase the price index,while monetary policy tightening can reduce the price index.In terms of economic growth,loose monetary policy can increase the growth of industrial added value and promote economic growth,while tight monetary policy can inhibit economic growth.In terms of financial stability,monetary policy easing can increase real estate investment,on the contrary,monetary policy tightening can reduce real estate investment.Further,our monetary policy tools can effectively manage the economy and maintain economic and financial stability as expected.Compared with the impulse response analysis of VAR model,it is found that the application of dynamic potential outcome framework can more accurately identify the direction and degree of influence of loose and tight monetary policies on the three outcome variables.