Abstract: | This experiment evaluated a procedure whereby tax evasions can be more effectively deterred by dividing the taxpayers' population into two, high- and low-audit probability, categories. The procedure employs a rule that assigns tax evaders to the high-audit probability category and assigns those who abide by the law to a low-audit probability category. A simple empirical simulation showed that the “variable probabilities” procedure prevented tax evasions more effectively than an alternative, “fixed probability” procedure, in which the same audit probability applied to all individuals, regardless of their past record. |