Study Of Assessment System For Tax Credit Rating About Enterprise Taxpayer

With the gradual development of Chinese tax system and social credit system,the tax credit is playing a more and more important role.In 2014,the State Administration of Taxation promulgated the tax credit management measure(trial).The tax credit rating evaluation system started across the country,which is a new beginning of tax credit evaluation and also an attempt to constantly improve.After sorting out the experience of tax credit system in developed countries such as the United States and Australia and carefully studying the regulations of tax credit rating evaluation system in China,this paper aims at the problems of index setting,unscientific classification and single deduction method of the current tax credit rating evaluation system,and proceeds from a statistical point of view to 95 indicators in the current tax credit rating evaluation.This paper makes a new classification and selection,redefines the existing taxation credit rating mechanism,and studies its scientificity.Firstly,through cluster analysis,this paper combines "tax assessment and tax audit" and "refusal,obstruction of tax authorities' law enforcement","amount of tax arrears" and "tax audit information of large enterprises" in the secondary indicators,and obtains the classification results with 80% similarity to the original classification,which is more practical and operable than the original classification of indicators.? Secondly,this paper uses principal component analysis to select 10 principal components with a cumulative contribution rate of 75%.After getting the principal component score of each enterprise,a new rule definition of tax credit rating is given,which makes the coincidence between principal component inference and the original score close to 75%.Finally,this paper uses discriminant analysis to replace the whole sample data and test the discriminant effect of the discriminant function.The results show that the misjudgement rate of category A is very low,only 0.7%,and that of category B is 12.2%.The misjudgement rate of category C and D is obviously high,78.8% and 41.2% respectively.With the continuous development of tax credit system and the accumulation of data,the effect of discrimination will gradually improve,which will provide a good help for tax bureau to carry out tax early warning management.In order to further optimize the results,on the basis of principal component analysis,this paper uses Bayesian method,using the evaluation situation of enterprises in previous years as a prior probability,to modify the current results of principal component analysis and form the final posterior judgment results.The correctness of Bayesian model is verified by numerical examples.There are also correction errors in the test results.Considering that the data span of the case in this paper is only two years,the effectiveness of Bayesian correction model has not been fully developed.The model also needs to be optimized and perfected with the development of the system.