PROACTIVE APPROACH IN TAX RISK MANAGEMENT: DATA ANALYSIS TECHNIQUES TO IDENTIFY HIGH-RISK TAXPAYERS

Abstract

There is a transition from labour-intensive traditional methods to the use of technology-intensive systems to identify risky taxpayers in tax audits. With this process, which has been accelerating for decades, statistical models that will reduce tax loss and evasion of states and increase efficiency in collection have been diversified in the field of machine learning. While innovations in software programs with digitalisation transformation reduce the manual workload of the tax administration, machine learning algorithms are used with experts employed in the field in continuously developing risk analysis studies. For tax administrations that fall behind these developments, the identification of high-risk taxpayers, the effectiveness of collection, and tax compliance are becoming more difficult to audit in the context of rapidly increasing global data.

The influence of a country's distinctive historical, economic, and sociological characteristics is a significant determinant of taxpayer behaviour. Establishing relationships across taxpayer characteristics and classifying behaviours in the tax risk management process affects the success of tax scenarios. Although, the work of experts in traditional methods and data analysis in auditing is limited, the effectiveness of data mining methods and the combination of the results obtained with tax scenarios approach the desired accuracy rate in risk detection. For this reason, it is observed that different methods are tested in experimental studies since a single technique that audits all business records cannot be sufficient and reliable. This study elucidates the deployment of information technology and digital tools in tax risk management, with consideration of the nuances inherent to the diverse national contexts.