Skip to content Skip to navigation

The Deterrent Effect of Expansions in Death Penalty Eligibility Criteria

May 2009
Working Paper
08-033
By  Michael Frakes, Matthew Harding
Homicides must possess certain characteristics before they become eligible for capital punishment. Over the last several decades, virtually every state has added to its list of possible eligibility criteria. We draw on this rich set of eligibility-law variation to identify the deterrent effects ensuing from expansions in the reach of capital punishment. Eligibility expansions may deter future homicides through two channels: (1) by paving the way for more death sentences and executions and (2) by providing prosecutors with greater leverage to secure enhanced sentences (capital or non-capital). The former channel is only rarely implicated, confounding the ability to identify deterrent forces. The latter channel, on the other hand, is likely to be triggered on a fairly common basis. We focus on the provision most responsible for the within-state variation in eligibility laws and estimate that the adoption of a law making child murders specifically eligible for capital punishment is associated with an approximately 19% reduction in the rate of homicides of youth victims. In two key falsification exercises, we find no evidence to suggest that this estimate is reflective of a differential trend between treatment and control states that originated in the period prior to the eligibility expansions and we estimate no corresponding association between child-murder eligibility laws and adult homicide rates. We estimate deterrence findings of similar magnitude when we turn to the estimation of an empirical specification that draws on variations in the full set of eligibility criteria and that parameterizes general eligibility statutes using a simulated measure of the propensity of each state to extend capital eligibility to a given murder. However, the findings of this general deterrence investigation are relatively noisy and are not robust to the exclusion of the child-murder factor from the simulation analysis.