NCJ Number
134810
Journal
Journal of Research in Crime and Delinquency Volume: 28 Issue: 4 Dated: (November 1991) Pages: 418-425
Date Published
1991
Length
8 pages
Annotation
The statistical properties of the U.S. time series is examined in order to determine whether the criticism of Hale and Sabbagh regarding the crime and unemployment rate time-series data of Cantor and Land are applicable to the analysis.
Abstract
The analysis strategy employed in the 1985 analysis of crime unemployment rates was based on the assumption that the unemployment rate is a coincident indicator of the business cycle. For the first differences, crime-rate-dependent series of the model, the conclusion is similar to that of Hale and Sabbagh for England and Wales; these first difference series appear to be stationary or I (0) by whatever statistical test applied. The basis of Hale and Sabbagh's critique of the model specifications depends on whether one accepts an annual unemployment series that contains a single historical spike as suitable, represented either by the combination of a stationary series with a step function or an I (1) series. Taking the first representation, the unemployment rate series is nonstationary in mean, but not in variance and thus is not fundamentally flawed. The empirical findings of Hale and Sabbagh for England and Wales are not applicable to the analysis of U.S. data. Under a plausible specification of the U.S. annual unemployment rate, the models used in the 1985 analysis are not fundamentally flawed nor misspecified. 2 notes, 6 tables, and 19 references