The standard model of taxpayer evasive behavior was given by Allingham and Sandmo in 1972, theorizing that taxpayers chose how much to evade based on their perceived cost of evasion i.e., the amount of penalty that will be levied if evasion is detected via tax audits. However, it remains an open question in the current literature as to whether tax audits permanently alter evasion and non-compliance behavior in the absence of penalties and whether that behavior varies between heterogeneous agents i.e., randomly selected, evasive, and non-compliant taxpayers.  To answer this question, I leverage natural experiments in Pakistan to identify and estimate the effects of audit selection on firm tax evasion and tax non-compliance. Pakistan used a unique audit policy that varied audit selection between completely random, parametric, and risk-based methods focusing on representative taxpayers, evasive taxpayers, and non-compliant taxpayers, respectively.

The standard model of taxpayer evasive behavior was given by Allingham and Sandmo in 1972, theorizing that taxpayers chose how much to evade based on their perceived cost of evasion i.e., the amount of penalty that will be levied if evasion is detected via tax audits. However, it remains an open question in the current literature as to whether tax audits permanently alter evasion and non-compliance behavior in the absence of penalties and whether that behavior varies between heterogeneous agents i.e., randomly selected, evasive, and non-compliant taxpayers.  To answer this question, I leverage natural experiments in Pakistan to identify and estimate the effects of audit selection on firm tax evasion and tax non-compliance. Pakistan used a unique audit policy that varied audit selection between completely random, parametric, and risk-based methods focusing on representative taxpayers, evasive taxpayers, and non-compliant taxpayers, respectively.