California Sea Grant College Program

The deterrent effect of capital punishment has been debated in scholarly and pol­icy circles for at least two centuries. Much more recently, considerable efforts have been expended to characterize efficient operation of the entire criminal justice system, including its penal function.

In the first chapter, data on daily U.S. homicides are analyzed to test whether severe punishments act as a deterrent to murder. Previous linear regression analyses are discussed, after which the Poission regression model is argued, then demonstrated, to pro­vide a superior fit to the data. A specification test for the mean-variance equality implied by the Poisson model is derived, and negative binomial models utilized when these tests reject the Poisson. Both parametric and non-parametric methods are used to test the deterrence hypothesis: previous findings of a deterrent effect are shown to be quite fragile.

In the second paper, similar techniques are used to analyze a superior set of daily data from California over the period 1960-67. Specification tests for the negative bino­mial model are developed and a technique is employed to account for the stochastic dependence among the estimated regression coefficients, thereby providing sharper tests of the deterrence hypothesis.

In the third paper, the efficiency of criminal sanctioning policy is addversed. An illustrative model is posited, and optimality conditions derived and interpreted. Data from California counties are used to estimate standard economic models of crime for several categories of homicide and to test for efficiency in sanctioning.