Research On Some Problems In Applied Studies Of Econometrics In China

Econometric modeling methodology has been widely used in China's applied studies.However,due to the lack of deep understanding of relevant econometric theories,many problems have arised in the application of econometrics in China.This paper follows the logical structure of pointing out problems,analyzing problems and solving problems,adopts a combination of theoretical derivation and experimental analysis.Combined with the common mistakes in the applied studies of econometrics in China,this paper primarily analyzes three problems about model establishment,multicollinearity and Granger causality test.It aims to deepen the understanding of relevant econometric theories,clarify perplexities of relevant theories in application,and make the application of econometric modeling methodology more mature in empirical analysis.Firstly,it's about model establishment problem.This paper mainly discusses two aspects on variable selection and model judgment.The paper points out the problems about model establishment in applied studies of econometrics in China by examples.On the basis of theoretical derivation,combined with the Monte Carlo simulation experiment,the paper analyzes the consequences of losing important variables or controlling too many variables in model establishment,and gives the application suggestions.Then this paper explains R-squared and adjusted R-squared theoretically.On the basis of theoretical derivation,combined with the Monte Carlo simulation experiment,the paper analyzes the appied problems about R-squared and adjusted R-squared,illustrates that it's not wise to judge the model by only comparing the R-squared or adjusted R-squared.Secondly,it's about multicollinearity problem.This paper points out the appied problems about multicollinearity,explains multicollinearity theoretically,and analyzes the applied problems about Variance Inflation Factor.On the basis of theoretical derivation,combined with the Monte Carlo simulation experiment,the paper analyzes the possible consequences caused by multicollinearity problem,and confirms the truth by drawing in MATLAB.Then the strategies for dealing with multicollinearity problem are given:expanding sample size;substituting parameter relationships;adding explanatory variables;transforming model forms.Thirdly,it's about Granger causality test problem.This paper points out the applied problems about Granger causality test in studies,explains Granger causality and Granger causality test theoretically.The paper also analyzes the differences between causality and correlation,Granger causality and general causality.Finally,this paper gives the application suggestions of Granger causality test.This paper concludes with a summary of the full text and points out the direction for further research.