Deterrence and the Death Penalty by Daniel S. Nagin

Author:Daniel S. Nagin
Language: eng
Format: epub
Tags: ebook, book
ISBN: 0309254191
Publisher: The National Academies Press
Published: 2012-11-27T00:00:00+00:00

Homogeneity

Still another assumption of the panel regression model in Equation (4-1) is that any effect that the death penalty has on homicide rates is the same in every state and every year. This assumption of a homogeneous treatment effect is unlikely to hold in practice. This assumption relies on “unit exchangeability,” which requires that if the change in the death penalty measure observed in a particular state and year were instead to be observed in a different state and year, then the effect seen on homicide would be the same. For the legal status of the death penalty, this assumption would mean that the death penalty would have the same effect on homicides in the first year a low-crime state instituted the death penalty by legislative action as it would in the 15th year in Texas, a state in which it is widely used. The assumption would also mean that the effect would be the same in the year before the death penalty was removed as a possible sanction due to the courts’ determining the state’s death penalty law was unconstitutional in a state that had the death penalty but did not implement it. The death-penalty-intensity models also invoke this assumption. These models assume that every possible death-penalty-intensity level would have the same effect on homicide rates in every state and year if it was present in that state and year, regardless of the prior sanction regime, a state’s history with the death penalty, or any other factor.

Although this homogeneity assumption is commonly invoked in regression models, no support is offered for it in studies of the death penalty, and on its face it appears unlikely to hold. In fact, there is some evidence to the contrary. Figure 4-2 displays the distribution of estimates found by Donohue and Wolfers (2005, p. 810, Figure 4) when they estimate state-specific parameters using the same basic specification as in Dezhbakhsh and Shepherd (2006). They find that reinstatement of the death penalty in 1976 is associated with an increased homicide rate in 17 states and a lower rate in 24 states. Similarly, when Shepherd (2005) estimated state-specific deterrence parameters using the same basic specifications as in Dezhbakhsh, Rubin, and Shepherd (2003), she finds that executions deterred murder in 8 states, and increased murders in 13 states. The committee does not endorse these state-specific models and estimates, but the findings do suggest the potential for substantial heterogeneity in the effect of the death penalty across states, which violates a basic assumption of the panel data model in Equation (4-1). Moreover, relaxing this homogeneity assumption can lead to very different inferences on the effect of the death penalty (see Chapter 6).

Finally, we note that the panel regression models also rely on the assumption that the sanction regimes of adjacent states do not have any bearing on the effect the death penalty in a particular state. In other words, the assumption asserts that the effect of the legalization of death penalty (or an increase to a higher

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.