The current paper addresses social contribution fraud by estimating accurately the shortfall in tax revenue. This loss in tax revenue is defined as the potential sum of the tax adjustments that could have been imposed on firms that have committed fraud (or made erroneous social security declarations), if they had been effectively controlled by an inspection authority, whereas they were not in reality. We define first the tax shortfall from a theoretical point of view, highlight then the tractability of the theoretical approach via Monte Carlo simulations, and finally validate empirically the methods proposed for estimating such an amount. The empirical application is  performed on real data provided by MSA (Mutualité Sociale Agricole). Therefore, in addition to the methodological contribution, this study allows us to quantify the financial impact of illegal activity (fraud, abuse, optimization or intentional error, ...) for the entire French agricultural system.