The Empirical Study Of Combination Forecasting Model To Tax Revenue In Guangdong Province

Tax collection is the very foundation on which the state exists and society develops. It also has made a vital role in the functioning of state machine, as well as in the relationship to the common people. The scale of the tax is an important index to measure the state financial ability and the range of government functions in respect of economic and social activities. Tax forecasting is so rich in significance for tax plan and forecasting management. In order to take the initiative in tax collection, the present economic situation requires a scientific forecasting system based on the forecasting of the tax revenue to be put in practice as soon as possible.This thesis analyzes the tax revenue forecasting method and its modifying factors by the way of quantitative analysis from different angles based on the review of great quantity of tax revenue forecasting method, then Stepwise Regression, Cointegration Analysis and Neural Networks Model are adopted to construct three tax revenue forecasting models of Guangdong Province respectively, moreover, error analysis is used in these models. Combination forecasting method is put forward to forecast the tax revenue of Guangdong Province based on taking advantage of all kinds of forecasting method and avoiding deficiencies of them, historical data is use to simulate tax revenue ,and acts well.Through this thesis, combination forecasting model of tax revenue for Guangdong Province is put forward to forecast the tax revenue of Guangdong Province in the coming years in order to provide the reference for the scientific tax plan accord with the actual state of Guangdong Province and regulating government's budget.