Association for Public Policy Analysis & Management

We investigate the degree to which criminal offenses committed by youth and young adults are influenced by their neighborhood surroundings, especially the demographic, socioeconomic and criminal dimensions of their social housing development.  This is one of the first studies to employ a natural experiment to invetigate teh effect of neighborhood context on criminality, and the first one to do so using European data.  We identify causal relationships by using a natural experiment wherein the Copenhagen, DK municipality assigns households with urgent housing needs to every third social housing unit that becomes vacant, a process we show produces quasi-random assignment.  We obtain from administrative sources information on the 1,078 individuals aged 15-29 who were thus assigned during 2007-2010 and on their social housing developments in the week before assignment; their violent, property and drug charges and convictions during the subsequent three years are modelled.  We find that the proportion of neighborhood residents aged 15-29 receiving social assistance increases drug crimes and the proportion aged 30-59 with only basic education increases property crimes.  Unspecified fixed effects measured at the school district level also predict all sorts of criminality.  Unexpectedly, the proportion of residents with prior criminal charges or convictions was not predictive.  These neighborhood effects at both the housing development level and the school district level are large in magnitude and are robust to a variety of specifications.  We draw a variety of implicatins from our work for social housing and anti-crime policies.