
Florian Baumann takes an important step forward in the economics of deterrence and group crime in “Should Leniency for Collective Acts Be Extended to Individual Crimes?” The answer to the question is a simple yes, but the answer requires a careful treatment of deterrence in a forward-looking model of both group and individual crime. To provide such a treatment, Baumann adds individual crime to the analysis of ordered leniency for group crime in Landeo and Spier (2018), who find deterrence is maximized when authorities offer leniency for self-reported crime on a first-come, first-served basis. However, if the leniency scheme is not extended to individual crime, deterrence may be weakened or reversed. Intuitively, members of a group can use individual crime as a commitment device to avoid self-reporting. If authorities tend to detect both group and individual crime simultaneously, the additional penalty for the latter reduces the efficacy of an ordered leniency scheme for the former. The deterrent effect of the leniency scheme is reinstated, and even enhanced, through its extension to individual crime. In this comment, I argue for why the implications of this work are deeper than the formal setting of the model may suggest. First, sociology and criminology widely treat crime as a group phenomenon. In fact, one of the few commonly accepted “criminological facts” acknowledges that themajority of criminal acts are committed bymultiple offenders (McGloin et al. 2008). In its preoccupationwith deterrence,
TimFriehe raises a fundamental issue. Should themisconduct ofmoral authorities be punished differently, perhaps more harshly, than the misconduct of followers? It would seemnatural to bemore severewith rolemodels because of the spillover effects of their conduct. The article shows that this reasoning is not necessarily correct. Individuals can engage in an action providing a private benefit b and causing an external harm of amount h. In the standard law enforcement model, deterrence derives from the threat of a legal sanction s. Here behavior also depends onmoral and conformity concerns. There are two types of individuals: A (authority) and F (follower). An individual of either type (i = A, F) suffers an “abstract moral cost” li from committing the act and amoral costrih as a fraction of the harm the individual causes. In addition, a type-F individual suffers a “conformity cost” hb̂A, where b̂A is the benefit threshold for a type-A individual to engage in the activity. The higher b̂A is, the less likely it is
Previous articleNext article FreeComment on “Six Dimensions of Criminal Procedure”Siona ListokinSiona Listokin Search for more articles by this author Full TextPDF Add to favoritesDownload CitationTrack CitationsPermissionsReprints Share onFacebookTwitterLinked InRedditEmailQR Code SectionsMoreGeorgakopoulos and Sullivan’s article “Six Dimensions of Criminal Procedure” contributes to an oft-asked question about courts and judges: Why do justices rule the way they do? Their novel work seeks to describe courts through the judicial tendencies of justice coalitions, eschewing the simplistic unidimensional liberal-conservative model for dynamic multidimensional scaling. This article, along with its predecessors, “Illustrating Swing Votes I: Indiana Supreme Court” and “Illustrating Swing Votes II: United States Supreme Court” (Georgakopoulos and Sullivan 2020a, 2020b), provides ample proof of concept, yet it leaves a number of unanswered questions about the applicability and usefulness of the method.1. How Many Dimensions?Much of the academic and media analysis of court decisions (and in particular, Supreme Court decisions) model the court using the median voter theorem: justices exist on a spectrum from the political left to right, and the median justice has an elevated swing vote characterization.1 Prior work has shown this framing to be inaccurate: courts do not necessarily have a predominate swing voter, and justices at the extremes of the unidimensional spectrum sometimes serve as swing voters.Clearly, then, the median voter theorem does not suffice to comprehensively describe judicial tendencies. Scholars have responded in recent years by suggesting multidimensional scaling to model justice preferences; rather than the left-right divide, justices’ decisions can be mapped onto preferences over civil rights, judicial power, police discretion, and (almost) countless other categories (Clark 2019; Fischman 2015). When done carefully, this mapping can illuminate previously unnoticed coalitional tendencies. For example, the authors note that in their mapping of Indiana Supreme Court split decisions on criminal procedure, the issue of police discretion appears surprisingly rarely despite its assumed centrality in cases with disagreements over intrusive police conduct. This pattern (or lack of pattern) may be hard to discern if one justice is marked as the political center and all decisions are surveyed in that lens.If one dimension misses important coalitions and tendencies, though, how many dimensions are tractable and useful for describing courts? Georgakopoulos and Sullivan are themselves heavily invested in this question. Prior work showed that the extreme case of mapping justices completely, by every rule, produces unusable, noisy results and is not viable. Thus their focus in this article is on the reasonable middle ground, an intermediate multidimensional scaling that uses the authors’ discretion for a particular court and area of law. Here, they focus on rulings on criminal procedure and offer six tendency dimensions, including closure of process, need for defendant’s consents and warnings, governmental and trial bias, police discretion versus warrants, trust in juries, and retroactivity of defenses.The resulting map is copied for reference (see fig. 1). The visual representation is quite clever and well suited to this framing method, and it is a contribution in its own right. Georgakopoulos and Sullivan note that mapping the five-justice court along six dimensions of criminal procedure produces a signal-to-noise ratio of three to one, meaning a quarter of split decision criminal procedure cases in the sample do not fit. These cases have a different coalition of justices/swing voter along the predetermined dimensions, or do not fit into the dimensions well.Figure 1. The circle of all possible dissenting teams and their tendencies about criminal procedure.View Large ImageDownload PowerPointIt is unclear how to evaluate these results. Georgakopoulos and Sullivan provide the following assessments within the article: “[Result] gives cautious optimism about describing the work of courts with locational models of this level of generality,” and “[result] revealed some order, perhaps more than expected. The 3:1 signal-to-noise ratio is not unacceptably low—neither is it strikingly high.” This is not a case of inflated language from overenthusiastic authors, and Georgakopoulos and Sullivan should be commended for their levelheaded appraisal.And yet, the original issue remains over the choice of dimensions. Multidimensional scaling can describe court tendencies more accurately than unidimensional models, but the description will be more complex and will be accompanied by a healthy dose of unexplained variation. In this example, the complexity is high. The model uses six dimensions to describe just one area of the law (criminal procedure), and it is easy to imagine framing courts along 20 or more facets if the author is interested in a broader range of subjects.The trade-off between tractability and completeness is a feature of all models, and future work in this vein will ultimately be judged by the insight that results from authors’ dimension choices. For this proof-of-concept study, though, it would have been useful to offer a different set of dimensions for comparison. This would provide the opportunity to evaluate competing models’ perceptiveness and signal-to-noise ratios.The above discussion highlights that the pivotal choice of dimensions is more art than science. Georgakopoulos and Sullivan are exacting in describing how each of the six dimensions fit the cases that fall into their respective categories. However, the article provides very little insight into the process of dimension choice. Did the authors independently evaluate the cases and come to this decision? Were there competing taxonomies? How might future authors design the classification process? Georgakopoulos and Sullivan note that future analysis could use text analysis to determine dimensionality but do not provide additional guidance as to how.2. ApplicabilitySome of these issues will be resolved as scholars increasingly frame court decisions with multiple dimensions. Concerns over applicability, however, are not easily resolved with familiarity. Georgakopoulos and Sullivan use their deep knowledge of the Indiana Supreme Court to their advantage in this study, and the court is particularly well suited to multidimensional scaling. The state supreme court is composed of five justices, thereby limiting the number of potential coalitions within minority pairs and swing voter (though note that there are multiple possibilities even with the smaller court). In addition, the Indiana Supreme Court was exceptionally stable for a period, leading to a sizable sample of 65 split decisions on criminal procedure between 1999 and 2010.The author’s impressive sample highlights a weakness of the dynamic scaling. Although compositional stability is not required for these models, it simplifies the process greatly. Multidimensionality requires computational depth to be meaningful, by definition. Georgakopoulos and Sullivan note this constraint within their article and conclude with a hope for more stable court compositions to aid this type of analysis.The promise of multidimensional scaling is thus tied in part to its applicability to a range of courts and across time. This inward focus on the courts illuminates a potential folly, however. No analysis of court decisions should be completed in isolation, without consideration of history, culture, and politics. In analyzing how justices make decisions, modeling along more than one dimension should illuminate connections with other institutions rather than disregard them. Other authors within this literature, like Clark (2019) and Fischman (2015), use their dimensional mapping to evaluate the courts’ interactions with other branches of government, special interests, and cultural shifts. Georgakopoulos and Sullivan’s series of articles offers an in-depth look at the potential and pitfalls of multidimensional models for court decisions. Ultimately, its usefulness will depend on our understanding of the world outside the court as well.Notes*George Mason University. Email: [email protected].1 See, e.g., Blasecki (1990); Enns and Wohlfarth (2013); Schmidt and Yalof (2004).ReferencesBlasecki, Janet L. 1990. “Justice Lewis F. Powell: Swing Voter or Staunch Conservative?” Journal of Politics 52 (2): 530–47.First citation in articleLinkGoogle ScholarClark, Tom. 2019. The Supreme Court: An Analytic History of Constitutional Decision Making. Cambridge: Cambridge University Press.First citation in articleGoogle ScholarEnns, Peter K., and Patrick C. Wohlfarth. 2013. “The Swing Justice.” Journal of Politics 75 (4): 1089–107.First citation in articleLinkGoogle ScholarFischman, Joshua Β. 2015. “Do the Justices Vote Like Policymakers? Evidence from Scaling the Supreme Court with Interest Groups.” Journal of Legal Studies 44:269–93.First citation in articleLinkGoogle ScholarGeorgakopoulos, Nicholas L., and Frank Sullivan Jr. 2020a. “Illustrating Swing Votes I: Indiana Supreme Court.” Indiana Law Review 53:95–134.First citation in articleGoogle ScholarGeorgakopoulos, Nicholas L. and Frank Sullivan Jr. 2020b. “Illustrating Swing Votes II: United States Supreme Court.” Indiana Law Review 53:135–62.First citation in articleGoogle ScholarSchmidt, Patrick D., and David A. Yalof. 2004. “The ‘Swing Voter’ Revisited: Justice Anthony Kennedy and the First Amendment Right of Free Speech.” Political Research Quarterly 57 (2): 209–17.First citation in articleCrossrefGoogle Scholar Previous articleNext article DetailsFiguresReferencesCited by Supreme Court Economic Review Volume 282020 Sponsored by the Antonin Scalia Law School, George Mason University Article DOIhttps://doi.org/10.1086/709739 Views: 247 HistoryPublished online September 23, 2020 © 2020 by the University of Chicago. All rights reserved.Crossref reports no articles citing this article.
This review of the criminal deterrence literature focuses on the questions that are largely missing from many recent excellent and comprehensive reviews of that literature and even from the literature itself. By "missing" I mean, first, questions that criminal deterrence scholars have ignored either completely or to a large extent. These questions range from fundamental (the distributional analysis of the criminal justice system) to those hidden in plain sight (economic analysis of misdemeanors), to those that are well-known yet mostly overlooked (the role of positive incentives, offender's mental state, and celerity of punishment). Second, I use "missing" to refer to the areas where substantial relevant knowledge exists but is largely disregarded within the criminal deterrence research program. The empirical analysis of environmental and tax compliance is a stark example. Finally, I stretch "missing" to describe topics that have been both studied and reviewed but where substantial challenges remain. These include the theoretical explanation for the role of offense history, the proper accounting for the offender's gains, the estimation of the costs of various crimes, and the cost-benefit analysis of crime-reduction policies. Among the literature's missing pieces, several stand out both on their own and because they combine to produce a highly unfortunate result. First, although the empirical side of the literature focuses almost exclusively on street crime, the literature makes only a minor effort to estimate the cost of crime and essentially no effort to estimate the cost of white-collar offenses. This adds to the impression—not supported by the available evidence—that street crime is a great social problem while white-collar crime is a minor one. Second, the literature fails to treat misdemeanors and misdemeanor enforcement as an independent subject of study. This failure contributes to the notion—also unjustified—that 13 million or so misdemeanor charges a year and countless millions of stops, frisks, and interrogations that lead to no charges—all heavily skewed by race and class—are also not a major social problem. Third, the literature is only starting to develop a benefit-cost analysis of various crime-reducing strategies. The analysis considers almost exclusively measures reflected in the optimal deterrence model. This creates an impression—almost surely false—that deterrence is the only means of reducing future crime. Finally, the literature ignores distributional analysis altogether, even though the burdens of crime and the criminal justice system vary dramatically, predictably, and disturbingly by race and income. By disregarding this variation, the literature may be reinforcing it. For all these reasons, the criminal deterrence literature may well be contributing to the overwhelming, singular focus of American society and law enforcement on the forceful deterrence of street crime. Addressing the missing pieces would enrich the literature, expand its appeal and policy relevance, and enable academics to contribute to the effort of setting the US criminal justice system on the path of long overdue structural reforms.
In this study, I statistically analyze the deterrent effect of capital punishment in Japan. I first introduce three relatively lesser-known prior studies. The studies have several common limitations: first, in sample size; second, in the econometric methodology (serial correlation, multicollinearity, simultaneity bias, and nonstationary data); and, third, in variable selection. I conduct new analyses to address these problems. In this study, the vector error correction model is adopted to analyze data from 1948 to 2018 with a sample size twice as large as the one used in prior studies. I include both the death sentence rate and the execution rate, analyze homicide and robbery-homicide separately, and use the conviction rate and the life sentence rate to measure the marginal effect of capital punishment. I find that neither the death sentence rate nor the execution rate has a statistically significant effect on the homicide and robbery-homicide rates, whereas the life sentence rate has a significant negative effect on the robbery-homicide rate.