AbstractThe work of Feldstein (1995 and 1999) has stimulated substantial conceptual and empirical advances in economists' approaches to analysing taxpayers' behavioural responses to changes in tax rates. Meanwhile, a largely independent literature proposing and applying alternative measures of tax compliance has also developed in recent years, which has sought to provide tax agencies with tools to identify the extent of tax non‐compliance as a first step to designing policies to improve compliance. In this context, measures of ‘tax gaps’ – the difference between actual tax collected and the potential tax collection under full compliance with the tax code – have become the primary measures of tax non‐compliance via (legal) avoidance and/or (illegal) evasion. In this paper, we argue that the tax gap as conventionally defined is conceptually flawed because it fails to incorporate behavioural responses by taxpayers. We show that conventional tax gap measures, which ignore the presence of behavioural responses, exaggerate the degree of non‐compliance. This potentially applies both to indirect taxes (such as the ‘VAT gap’) and direct (income) taxes. Further, where these conventional tax gap measures motivate reforms designed to increase the tax compliance rate, they will likely have a tax‐base‐reducing effect and hence generate a smaller increase in realised tax revenues than would be anticipated from the tax gap estimate.
