Benno Torgler What Do We Know about Tax Fraud? An Overview of Recent Developments QUESTIONS ABOUT TAX FRAUD HAVE BEEN AROUND AS LONG AS TAXES themselves and will remain an area of discovery as long as taxes exist. To understand the impact of a tax system, it is important to know who complies with the tax law as well as who does not. Tax fraud is a large and growing problem in almost all countries. Consider the example of the US annual growth rate of unpaid federal individual and corporate income taxes since 1973 (reported by Aim, 1999). Reliable estimates by the Internal Revenue Service (IRS) suggest a tax gap of the federal income tax of some $257 billion to $298 billion in 2001, which equals a noncompliance rate ofabout 15.5 percent to 16.6 percent for the federal income tax (IRS, 2005). The exploration of tax fraud is relevant for many reasons. Tax fraud reduces tax collection and tax performance within a country. It may lead to externalities such as an increase ofalternative taxes, which leads to an increase in the tax burden of compliant taxpayers. Tax fraud can create misallocations in resource use whenever individuals make an effort to cheat on their taxes. It also leads to behavioral changes (for example, choices of hours to work). The presence of tax fraud requires the government to invest in resources to deter noncompliance. There are various ways of engaging in tax fraud, such as claim ing false deductions (overreporting the amount of deductions), knowFor helpful comments and suggestions, thanks are due to Alison Macintyre and Marco Piatti. social research Vol 75 : No 4 : Winter 2008 1239 ingly changing income, concealing or transferring assets or income, and recording personal expenses as business expenses. The traditional approach treats the problem as one of rational decision making under uncertainty. This means that cheating on taxes is a gamble that will either pay off in lower taxes or, with the probability of detection, end in sanctions. This view oftaxpayer behavior was first presented in a formal model by Allingham and Sandmo (1972), influenced by the economics of crime approach (see Becker, 1968). A large num ber of research approaches to tax compliance continue on this framework (see Cowell, 1990). In a departure from this view, I will argue that such a portfolio analysis has its lim ita tions, including an inability to explain why many households comply more fully than predicted by this approach. Thus, it makes sense to go beyond a traditional deterrence approach in order to understand the causes and consequences of tax fraud. I will show that the recent trends in the literature bear out this argument. In particular, this review shows the usefulness of applying a multifaceted and interdis ciplinary approach. METHODOLOGIES A major difficulty in analyzing tax fraud is the fact that individuals have incentives to conceal their cheating. Thus, how can we measure tax fraud? We will discuss several methods, stressing that they are all subject to imprecision and controversy. Surveys One method of measuring tax fraud is the use of surveys that are typi cally designed to elicit taxpayers’ attitudes about their reporting. Such surveys can, however, also be used to estimate tax fraud. A main advan tage is that they include many socioeconomic, demographic, and attitudinal variables. This helps to investigate and test a rich set of (new) theories. Furthermore, surveys help to compare different countries and obtain insights into development over time. Certainly, the accuracy of surveys is not fully clear. Individuals may not remember their reporting 1240 social research decisions, they may not respond truthfully, or the respondents may not be representative, since the sensitive nature of tax fraud might create the incentive not to participate in such a survey. Audit Data Another direct method relies on information generated by the tax author ity as part of its audit process. The IRS conducts line-by-line audits of indi vidual tax returns for its Taxpayer Compliance Measurement Program (TCMP). TCMP provides information about noncompliance based on actual tax return data. These audits yield an estimate of the taxpayer’s “true”income, allowing measures ofindividual and aggregate tax evasion to be calculated. Tax returns...