Often these kind of regulation cause more problems than they solve; not because they do not work but for reason of scale and Bayesian proportions. Suppose that

Often these kind of regulation cause more problems than they solve; not because they do not work but for reason of scale and Bayesian proportions. Suppose that 10% of affected bussiness buy catalytic converter illegally and that an audit will determine di 90% accuracy whether a shop does buy catalytic converter illegally or not, then half* of fined business will end up being innocent (In practice including lax bookkeping and edge cases not considered in the law) * I = innocent, F = fined, G = Guilty P(I|F) = P(F|I)P(I)/P(F) = P(F|I)P(I)/(P(F|I)P(I) + P(F|G)P(G)) = (0.1 * 0.9)/(0.10.9 + 0.10.9) = 0.5 I am pro regulation in almost all cases, but it is not easy...