By focusing on discovered frauds, we introduce two selection biases: we do
not observe frauds that are never caught and we do not observe frauds caught
so early that they never enter the public domain. In addition, we cannot say
anything about the importance of specific mechanisms in preventing fraud that
does not occur.