The determinants of tax evasion : experiments with Turkish subjects
2014-12-15T10:36:31Z (GMT) by
This thesis explores the results of a series of tax experiments that were undertaken with various groups of people in Turkey in 1998. The experiments are described and their resulting data reported, following which conclusions and implications are reviewed.;In total 268 individuals successfully completed the experiments. The objective of the experiments was to gather information with which to test the effect of certain variables upon (i) the decision to evade income taxes and (ii) the amount of income tax to evade both in absolute terms and as a proportion of income.;Our results indicate a very strong positive effect of tax rates upon both the decision to evade income tax and upon the degree (the proportion of income evaded) and absolute amount as well, once individuals have decided to evade. However, the expected fine has a significant deterrent effect on both the decision to evade taxes and the amount of taxes evaded. The effect of income was positive upon the decision to evade and upon the absolute amount of evasion but negative upon the proportion of income evaded. A large fine with a small probability of detection was not found to be more effective in deterring the occurrence of evasion than a small fine with a high probability of detection, but it was more effective in reducing the degree of evasion amongst evaders. We found that student subjects are more compliant than non-student (when the same experimental methodology is applied). However, student evaders did not behaviour substantially differently from employed people of the same age. While an increase in fines had the expected deterrent effect, an increase in audit rate increased the probability of evasion; this latter finding may be explained by the 'spite' or the 'crowding out' effects. The effect of tax ethics was positive and significant on tax compliance. Finally, we found that young people in general evade more often and evade a larger amount of income.