The Fusion On VAR Model And Application

Vector Autoregression Model (VAR) is one of the most successful, flexible andeasiest models for the analysis of multivariate time series. When the VAR model isintroduced into the economy, it promotes the wide application of the dynamic analysisof the economic system. It is used to analysing of dynamic interaction betweeninterrelated time series so that explain the impact of different various economic shockson economic variables.The thesis combines the theories of Econometrics, Factor Analysis to overcome thedefect that the VAR model variables cannot be too many. It also deals with the problemof multicollinearity between variables by Factor Analysis. It is successful to extractrelevant information from a mass of macroeconomic variables that bases on the actualdata. Then the thesis researches the stochastic disturbance of VAR model, analyses ofthe constitution of the different types on stochastic disturbance, discusses the effects ofdifferent types on stochastic disturbance to explained variable. It researches the dynamicreaction between the explanatory variable and explained variable by impulse responsefunction. The model is verified by employment data in Shaanxi Province. The thesisdefines the employment effects indicators to measure the impact about the effects thatthe employment policies on employment of Shaanxi Province. Then this paper bases onabove research work, combines BP neural network to achieve the better forecastingresult. The model’s rationality is proved by the employment forecasting of ShaanxiProvince.Finally, the thesis discusses the mathematical expressions of the stochasticdisturbance, the problem of causality theories and BP neural network model which needto be improved and optimize for the further studies. This research is supported by the fund of Xi’an University of Architecture andTechnology.