posted on 2009-03-30, 11:45authored byDerek Deadman
There is an extensive literature on the modelling of property crime but little of this literature has attempted to use the estimated models for forecasting. A recent exception is the published Home Office predictions of recorded burglary and theft for 1999 to 2001. These Home Office predictions (which are for a reversal of the recent downward trend in recorded offences) are compared here with alternative econometric and multivariate time series predictions. The crucial role of the error correction structure of the econometric estimates compared with the forecast profiles which come from time series approaches is highlighted. Diaggregation of theft into vehicle and non vehicle crime is seen as potentially a fruitful area for further research.