Tax Notches in the Lab: Disentangling Real and Evasion Responses

Several empirical studies find large behavioural responses to the incentives created by tax notches, highlighting the challenge of distinguishing between responses driven by real effects and by tax reporting. In a lab experiment, we find strong evidence of excessive bunching in a tax notch system, both with and without evasion possibilities. The effort adjustments are mainly from above the threshold, while the evasion adjustments are mainly from below the threshold. Both adjustments contribute to a reduction in the underreporting rate. We also confirm evidence of optimisation frictions. They are stronger when evasion is possible than when it is not. A gender breakdown highlights both the robustness of the effects found and the impact that heterogeneous preferences can have on the overall responses to tax notches.