How much of a problem is crime for firms in Latin America?

Editor's Note: Mohammad Amin is a PSD specialist in the Enterprise Analysis Unit of the World Bank.

Could crime be a worse problem for businesses in Latin America than either bribery or power outages? Data from Enterprise Surveys (ES) – a routine World Bank survey of firms around the globe on various issues related to the business climate and firm performance – show that a third of the firms in 14 Latin American countries experienced at least one incident of crime during 2005. 72.8% of all firms lost money either due to crime or expenses on security, which together average 2.7% of annual sales for a typical firm. As the figure below shows, these losses are more than the reported amounts paid in bribery or losses due to power outages, and about 40% of what firms spend on infrastructure (electricity, water, telecom and transport). In short, crime prevention can lead to significant cost savings for firms.

Three important findings emerge from the ES data. First, firms are as likely to be victims of crime as individuals and households (33% vs. 38% for households are reported in a study by Gaviara and Pages, 2002). Second, large firms are more likely to be victims of crime than small firms (42.4% vs. 31.4%). However, losses due to crime as a percentage of annual sales are much higher for the small than the large firms (1.4% vs. 0.65%). The burden remains heavier on the smaller firms even if we include security expenses (2.9% vs. 2.2%). This casts doubt on existing studies (Gaviara and Pages, 2002; Glaeser and Sacerdote, 1999) that suggest that the relatively better off (richer individuals or larger firms in our case) suffer more from crime than the rest - a result based solely on the incidence of crime rather than the burden it imposes (as a percentage of income).

Third, the incidence of crime is higher in the bigger than the smaller cities (population wise). Criminals prefer larger cities where it is easier for them to remain anonymous and there is more wealth to steal. However, the ES data show that this result holds only across cities within a country - not across countries. Hence, what matters is how big or small a city is relative to other cities in the country, with the absolute size of the city being completely irrelevant. A doubling of all city populations in a country should therefore have little effect (if any) on the level of crime in the country. The good news in this is that the natural population growth in cities does not call for more resources towards crime prevention – only its reallocation from the slower to faster growing cities.