Tax Audit in Turkiye: Simulation and Estimations Based on Kernel and Weight Functions

Abstract

This research examines the use of kernel estimation and $FindDistribution$ methods in $Mathematica$ software to analyze the ratio of taxpayer audits to total taxpayers, focusing on two large populations: one with approximately 80,000 audits per 100,000 taxpayers and the other with 4.5 million audits per 6 million taxpayers. Comparing the maximum statistics, the study shows that a larger number of taxpayers leads to more audits. The dataset also includes a weighted average for audits and taxpayers with a maximum of around 75,000 and 4 million respectively. These numerical values have been determined using the simulation carried out after modeling the real data sets of the total number of taxpayers and their audits from the years 2012 to 2023. These results show that different taxpayer populations require the targeted audit strategies and highlight the importance of the statistical models with corresponding estimation method to better understand complex distributions and improve tax audit processes.