NCJ Number
86177
Date Published
1978
Length
66 pages
Annotation
This study reexamines Ehrlich's data from his 1973 article to determine, within the context of Ehrlich's theoretical model, whether the deterrent effects of punishment are real or an artifact of a particular model specification.
Abstract
The data available for the present investigation are for crimes in 1960 in 47 of the United States (New Jersey, Alaska, and Hawaii were excluded). For each State, the reported crime rate was studied for each of the seven FBI Index crimes. Eleven variables of socioeconomic status have also been used, and per capita police expenditure in each State for 1960 and for 1959 have been used to describe the resources available to combat crime. The reexamination of the data indicated inaccuracies in the data as well as in the reported results. Ehrlich's model was reanalyzed to correct for the data inaccuracies. The results of the analyses of different model specifications generally indicated negative point estimates for the elasticities of the probability of imprisonment and the time served. The magnitudes of these elasticities were similar across the different specifications. The only large changes in the point estimates occurred in the murder and assault equation when these equations were only identified with lagged police expenditures and the lagged crime rate and when certain States were omitted; however, in this situation the estimates were very unstable, possibly due to excessive multicollinearity. It appears, therefore, that with the available data and within the present model, the negative relationship between the crime rate and the probability of imprisonment and between the crime rate and the time served is not spurious. The appendixes contain the classification of crime Index offenses, symbols and sources of the variables, and tables. Thirteen references are listed. (Author summary modified)