In response to a comment on my previous post on this topic, the table below the jump shows how the incidence and burden of crime, security and bribery (as points of comparison) vary by selected firm characteristics. (Differences that are significant at the 5% level are marked with an asterisk.)
Expenses on security do not show any significant variation along the listed dimensions. Losses due to crime as a percentage of annual sales (averaged over all firms, victims and non-victims) are higher for non-exporters relative to exporters, those who maintain a below-median vs. an above-median level of inventory and for firms operating in the local vs. national market. At first glance, it is difficult to weave together these findings into a coherent and sensible prediction of the underlying forces that drive them.