Risk assessment of VAT invoice crime levels of companies based on DFPSVM: a case study in China

Abstract

In recent years, with the implementation of the policy of “Replacing Business Tax with Value-Added Tax” and “Streamlining Administration, Delegating Powers and Improving Regulation and Services” in China, criminals have been issuing false invoices, and such cases have shown a trend of high frequency in the category of economic crimes. Tax departments and public security departments are facing increasingly a serious crime situation that has created a new challenge. By studying the current trend of false invoice crime, the difficulties of investigation in such cases are analyzed. Using the tax information of enterprises that have conducted false invoice as the breakthrough point, the machine learning method is introduced to build a risk pre-warning assessment model based on the Support Vector Machine (SVM) method to detect enterprises issuing false invoices. Three steps were designed in this paper. First, a risk pre-warning assessment model was established to detect enterprises issuing false invoices. Second, enterprises were classified into three groups according to the risk levels: A, B, and C. Third, collected data were used to make an empirical analysis, and the results show that the accuracy rate of the model is 97%. In China, due to the high crime rate of tax fraud cases, it is important to obtain data from tax and public security departments to establish a model that can detect such crimes as early as possible. The police and tax authorities can use this model to jointly combat such crimes.