Can We Predict Long-Term Future Crime Rates?

Yu-Sang Chang, Changyong Choi
2013 Social Science Research Network  
Can we predict long-term crime rates? In this paper, we offer the use of simple experience curve models as an alternative forecasting method. We use the experience curve models to project total crime and violent crime rates in 2030 for 50 individual states and Washington D.C. in the United States. Results are encouraging in that projection models developed from historical data for respective states show, in general, high values of R 2 over .85. Our projected crime rates show both increasing
more » ... ds as well as declining trends compared to 2010. A large variation among individual states is due to highly variable experience curve slopes we estimated across respective states. 3 5 As relevant many of these factors may be in explaining the past crime patterns, most of these factors will not help in making future projection. Future values for most of these factors cannot be reliably forecasted. Therefore the projection of long-term crime rate must, by necessity, rely on selected few factors for which future values can be projected with some degree of accuracy. The most likely factor with such reasonable degree of accuracy in forecasting is often demographic measures such as population size, density, age and race. This explains why the age standardization methods (Hirschi and Gottfredson, 1983; 1989; Steffensmeier and Harer, 1987; 1999) have often been used. The method combines relatively accurate estimates of the age structure of the American population with age-specific arrest rates for various types of crimes to calculate expected numbers of crime rates. By using this methodology, Steffensmeier and Harer (1987) forecasted that violent crime rates would fall about 13 percent and 20 percent for property crime rates during the 1980 to 2000 period. The basic reason is that proportion of young people (ages 15~24 and 15~35) was estimated to decline sharply into the early 1990s by the Census Bureaus. This projection was updated by Steffensmeier and Harer (1999) . Using the same methodology, they projected future rates for both violent and property crimes through 2010. Specifically, they projected 5 percent increase for violent crime rates and 4 percent increase for property crime rates from 1966 levels to 2010. However, it was pointed out by Land and McCall (2008) that "these projections assumed that age-specific arrest rates for Juveniles continues at the levels observed in 1966 to the year 2010" 2 . "If, in fact, these age-specific rates continue to decline, then the modest increases in violent and property crime rates projected could become even more modest or even turn into decrease" 3 . This logic will be decisive in developing our experience curve model. The alternate approach for long-term projection is the development of time-series regression or structural equations. Works by Fox (1978) and Cohen and Land (1987) are wellknown examples of using regressing models of crime rate time series. Fox's model has added both socioeconomic characters of the population, police activities and expenditures as well as 2 Land and McCall, p.331 3 Ibid, p.332
doi:10.2139/ssrn.2307350 fatcat:lvezsp6hbbblhmjbkpccduwyn4