Risk-based Tax Audit Selection

Due to the problems such as large and increasing volume of tax returns, lack of pre-assessment, time limit and lack of standards for tax auditing, limited manpower, the arbitrary judgment in this field and failure to provide tax returns by some taxpayers, a new approach is needed to solve these problems. The present research aims to apply the most efficient up-to-date methods and techniques of the world (chaos theory and artificial neural network theory) by which Iranian National Tax Administration (INTA) would be capable of computerized assessment of tax returns on the basis of the difference percentage between declared and forecasted pre-tax incomes and select tax returns with the greatest risks and refer them to the tax auditors for auditing. In this regard, the chaotic time series variable would be forecasted by artificial neural network non-linear model. The required data has been gathered through a library method. Parsportfolio Data Management Software has been used for collecting the data of companies listed in Tehran Stock Exchange. Meanwhile, the research has used the statistics published by Iranian Statistics Centre (ISC), Central Bank of Islamic Republic of Iran (CBI) and (INTA).