Tax Revenue Forecast Model Selection And Empirical Analysis

Tax revenue is an important part of the state's fiscal revenue.The fiscal and taxation of a country is related to the normal operation of the state's various mechanisms.It is the foundation of a country's development.The importance of taxation for the development of the country also determines the prediction of the scale of the tax.Taxation is closely related to the economy,and taxation affects all aspects of economics through the impact on total social demand and supply.Similarly,the various elements of the economy also determine the size of tax revenue.Establishing a good tax revenue forecasting model is of great significance to tax source management,tax planning,tax policy formulation,and financial budget.This paper starts with the description of the significance of tax revenue model prediction,and deeply analyzes the research status at home and abroad.Then it discusses the related basic theories of tax revenue forecasting,including the definition of tax revenue forecast and the choice of types.On this basis,the relationship between tax revenue and macroeconomics is deeply analyzed,including the relationship between tax revenue and national income,tax and price and so on.In order to use the model for analysis,it is important to compare and classify the methods and models of tax revenue forecasting.The application conditions of the model affect the choice of forecasting model.The classification of the model includes the one-dimensional or multiple linear regression,the time series model,the gray correlation,and the data mining algorithm BP neural network,support vector machine and the system dynamics model.In order to select the model and variables,this paper also analyzes the economic development and tax revenue of Jilin Province,including the scale of tax revenue,the growth rate and the structure of tax source structure,and a comparative analysis of taxation and economic growth in Jilin Province.Then,an empirical analysis of the tax revenue forecasting model was carried out.The principal component analysis model,time series ARIMA model,neural network model,and system dynamics model are established in turn,and then the effect of forecasting tax revenue is compared.The forecasting model shows that the most important factors affecting tax revenue forecasting include the number of loss-making enterprises and industrial added value.And a better tax revenue predictive models is the combination of the time series model and the system dynamics model.