Optimal dynamic tax evasion: a portfolio approach

Most tax evasion models are set in a timeless environment, though this is not suitable for a study of revenues from financial activities where tax evasion occurs in a dynamic environment. This study examines a representative agent aiming to maximize the expected utility from inter-temporal consumption and could invest in both riskless and risky assets, through tax evasion is possible only in the latter case. The investor must pay a fine when his/her evasion is detected (with a given probability). We show that: (i) optimal consumption is higher with tax evasion, (ii) optimal evasion is affected by neither the return nor the volatility of the risky asset, (iii) evasion reduces investment in the risky asset and increases investment in the riskless asset, (iv) evasion can be reduced more efficiently by increasing the fine rather than by increasing the frequency of controls, (v) for a sufficiently high tax rate on the riskless asset, the optimal evasion is zero, and (vi) if the fine is proportional to the amount of taxes, for only ‘sufficiently’ low fines the Yitzhaki (1974) paradox is confirmed