Bayesian Estimation And Application Of FAVAR Model Considering Sign Constraints

Since Lucas’ criticism,the Vector AutoRegression(VAR)model proposed by Sims has gradually replaced the macroeconomic simultaneous equation model and become a mainstream analysis model in the macroeconomics community.In particular,the VAR model is very effective in macroeconomic analysis and structural inference,but it is undeniable that there are also many shortcomings,such as excessive parameterization and failure to consider the time-varying characteristics of economic structure.Over-parameterization will make the estimated parameters unstable,and then the results will be unstable,and the ordinary VAR model can not capture the nonlinear characteristics of macroeconomic data,which may cause the model to be wrongly set.Therefore,it is very important to develop another model that can inherit the advantages of the VAR model and effectively solve these problems.Based on the above considerations,this article analyzes from three aspects to make up for the existing defects of the VAR model:(1)The factor extended vector autoregressive model(Factor Augemented VAR,FAVAR)is introduced.Extract effective information,and then enter the extracted information into the VAR model,which effectively solves the problem of information loss and excessive parameterization;(2)The Bayesian VAR modeling process is an important feature of Bayesian estimation It is possible to combine the prior information with the sample information,obtain the posterior distribution of the parameters through Bayes’ theorem and further determine the parameter estimates,and the accuracy and credibility of the parameter estimates obtained by this method are greatly improved;(3)Sign constraints are imposed in the impulse response analysis.The traditional VAR model recognition method is mainly based on equality constraints,but the sign constraints are to convert existing economic theory into a priori information,and then use the inequality to constrain the relationship between variables,Effectively avoiding the problem that the model lacks theoretical basis.To sum up the three aspects,this paper constructs the realization process of Bayesian estimation of FAVAR model considering symbol constraints.Finally,the newly constructed model is used to conduct an empirical study on the impact of monetary policy on the quality of economic growth under the new normal to prove the effectiveness of the symbolically constrained FAVAR model from the Bayesian perspective.The application research of this paper takes the annual data from 2000 to 2018 as an example,selects the 7-day interbank interest rate and the money supply M2 as the price and quantity monetary policies,and selects more than 60 macroeconomic variables as the dynamics of monetary policy.According to the effect transmission variables,according to the constructed economic growth quality evaluation index system,16 of the economic growth quality characterization scalars are selected,and the more representative latent factors are extracted through principal component analysis to construct the model.In particular,the interest rate factor,output factor,economic prosperity factor and inflation factor are extracted to represent the macroeconomic dynamic effects under the impact of monetary policy.Under the above four macroeconomic factor variables,price and quantitative monetary policies have an impact on the economy.The effect of growth quality control.By comparing the effects of two monetary policies on economic goals and comparing actual economic theory,it is found that the FAVAR model,which also considers symbolic constraints and uses Bayesian estimation methods,can more effectively reflect the actual economic operation status and laws,and more comprehensively.Reveal all aspects of information in the macroeconomic operation.