FORECASTING QUARTERLY AGGREGATE CRIME SERIES

In this paper we assess the forecasting performance of quarterly economic models of aggregate property and personal crime. We show that models that include long-run relationships between crime and its economic determinants tend to generate inaccurate forecasts, and attribute this to structural change. The forecast performance of the economic models is compared with that of time-series models, and forecast encompassing tests are reported. Copyright Blackwell Publishing Ltd and The University of Manchester, 2005.