Many empirical studies of crime assume that victims andperpetrators live in a single geographical unit, the implication being thatthe socio-economic characteristics of victims' places of residence canbe treated as determinants of crime. This study offers an alternativeapproach which consists in measuring crime by the proportion of allegedoffenders in the whole population and treating the characteristics of theirhome communes as socio-economic causes of criminal behaviour. Theconclusion is that those charged with crimes present a high degree ofgeographic mobility. In the case of economically motivated crimes, theevidence partly supports Becker's propositions. Lastly, we show that thenumber of people charged with crimes tends to be greater in communesthat have low incomes, a larger police presence, a predominance of urbanareas with higher levels of education and a geographical location in thenorth of the country, which to some degree bears out the findings of otherstudies on Chile.