In this paper, the authors use a direct method to estimate tax evasion in Italy assuming that tax evaders might consider declaring a closer-to-true income in an anonymous interview. The methodology is applied to work income only, as pension income cannot be hidden to tax authorities and capital income is measured with large error in available survey data sets. The data sets considered are the 1998 and 2000 Survey of Household Income and Wealth (SHIW) by the Bank of Italy and the 1998 and 2000 tax forms table produced by Servizio Consultivo ed Ispettivo Tributario (SeCIT). Posing particular attention to the post-stratification of the data, the authors find that tax evasion is consistently higher for self-employment income than for employment income and it is larger at bottom deciles, although some under-sampling problems need to be considered. The pattern of work income concealment found, shows that personal income tax evasion reduces the average tax rate but it also increases the progressivity of the tax system. This result is, however, driven by the large values of income avoidance found in bottom deciles, which might be due to the under-sampling of income receivers with the poorest income. The results are consistent across the two years considered.

Tax evasion in Italy is a serious issue: Between a quarter and half of the GDP seems to be hidden to the tax authorities (Schneider, 2000a). These aggregate figures are of great importance at a macro level, for instance for the reliability of official statistics and the efficiency of national productions. However, they do not provide insights to policymakers that wish to investigate who are tax evaders and to start understanding the reasons why some taxpayers might consider under-declaring their income.

Measuring tax evasion is all but simple: Schneider (2000b, p. 1) describes tax evasion measurement “as a scientific passion for knowing the unknown”. However, tax evasion analysis is relevant for public policy design and for estimating the bias that tax evasion introduces in some statistics, both at the macro and at the micro level. Section 2 provides a brief review of recent contributions on the estimation of tax evasion in Italy, focussing mainly on microeconomic approaches and in particular on direct methods of tax evasion estimation. Section 3 critically analyzes the representativeness of the data set and describes a grossing-up procedure to correct major deviations from population totals. This procedure is then implemented in the empirical analysis of tax evasion presented in Section 4. Section 5 explains how tax evasion is estimated and introduced in a tax-benefit microsimulation model, Section 6 analyzes redistributive effects of work tax evasion applying a microsimulation model for Italy and Section 7 concludes.