Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/120651 
Year of Publication: 
2012
Series/Report no.: 
52nd Congress of the European Regional Science Association: "Regions in Motion - Breaking the Path", 21-25 August 2012, Bratislava, Slovakia
Publisher: 
European Regional Science Association (ERSA), Louvain-la-Neuve
Abstract: 
It has often been observed that there is substantial spatial variation in criminality, i.e. criminality clusters in neighborhoods. Differences in neighborhood characteristics are one possible reason, social interactions another. In this paper we use detailed data on the residential location of criminals to disentangle the effects of individual characteristics, neighborhood characteristics and social interaction on criminality. Our basic model is an individual binomial logit model for the probability of being a criminal which we use to extract neighborhood effects. In a second stage, we model neighborhood effects, where we use as explanatory variables physical and social neighborhood characteristics such as characteristics of the housing stock and the demographic composition. We also include the share of criminals to be able to measure a social interaction effect. Note that this approach takes into account unobserved neighborhood characteristics. Since these may affect the criminality rate we instrument for this variable in a two stage estimation procedure.
Document Type: 
Conference Paper

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.