Response to: Adam J. Kolber, The Subjective Experience of Punishment, 109 Colum. L. Rev. 182 (2009).
Adam Kolber's thoughtful essay contends that variations in the subjective experience of punishment warrant a more nuanced manner by which courts mete out sentences for violations of criminal law.1 Failure to take subjective experience into consideration, Kolber argues, violates not only notions of retributive justice, but also consequentialist arguments grounded in theories about how we can best deter crime.2
This Response considers Kolber's consequentialist arguments in light of the neoclassical economic understanding of deterrence. According to the standard economic understanding of criminal law, individuals are deterred from committing crimes when they are made to feel "costs" equivalent to the harm they cause society, modified by the probability that they will be punished.3 When the "costs" of engaging in criminal conduct outweigh the net benefits of such conduct, the criminal refrains from committing that crime. Apart from reputation costs and moral qualms, the government-mandated punishment that the criminal defendant receives upon conviction is a significant portion of the costs that she must weigh against the perceived benefits of completing a given crime.
According to Kolber, society loses out when sentencing regimes fail to consider varying subjective experiences of imprisonment because they fail to take into account the defendant's true costs of punishment. As a result, if we set punishment at some specified amount for everyone (say, a ten-year imprisonment for all car thefts), we may excessively deter the person who is terrified of prison ("Sensitive," in Kolber's essay), and underdeter the person who is either unafraid or even looking forward to spending time in prison ("Insensitive"). Since underdeterrence and overdeterrence are both costly to society, Kolber argues, we should desire a system that either avoids or corrects these results. Thus, calibration of sentencing is desirable because it decreases the overall costs of deterring crime.4
As Kolber himself recognizes,5 most sentencing systems are not quite as crude as he fears. For example, in numerous states and under the Federal Sentencing Guidelines, recidivists receive substantially higher sentences than first-time offenders for the same crimes.6 These increased penalties (which ordinary deterrence theory sometimes struggles to explain7) make sense if one assumes that recidivists are less sensitive to prison than "ordinary" first-time criminals and therefore require harsher punishments. Moreover, in numerous systems, courts may take into account particularly heinous aspects of a crime in order to increase a defendant's sentence above some prescribed baseline.8 Although retributive accounts may well explain this increase, deterrence theory also provides an explanation: Through particular conduct, the defendant has indicated that he will be less deterred by some baseline sentence than the "ordinary" violator of the same statute. By the same token, in very limited circumstances, courts have reduced sentences to reflect a defendant's particularized vulnerability to prison conditions.9
Despite these safety valves, Kolber is correct: As a general rule, we do not finely evaluate a given defendant's subjective experience of punishment. Nevertheless, there are good reasons to reject a more calibrated sentencing regime. To better understand why, the remainder of this Response considers two counterarguments: First, subjective reactions to punishment (most often imprisonment) may benefit law enforcers by causing overdeterred individuals to "self-regulate" while the police focus scarce resources on more recalcitrant individuals; and second, excess sensitivity to punishment deters group-oriented crimes by spurring individuals to take extra steps to report or prevent others from engaging in criminal conduct.
Although these arguments provide different glosses on the problem that Kolber has identified, they are tied together by a common theme: Any call for calibrating punishment must also take into account the way in which we distribute law enforcement resources for monitoring and detection.10 Sentencing is just one variable of deterrence. However desirable it may be to take greater account of subjective experience from a retributivist perspective (the primary thrust of Kolber's essay), it is far less clear that Kolber's proposed regime delivers the consequentialist benefits that he promises.
I. Self-Regulation and the Probability of Detection
The first reason we might prefer the current sentencing system to Kolber's is that we believe it is more effective to focus calibration efforts on the criminal's probability of detection and not on the sanction that he ultimately receives. For a number of reasons discussed below, calibrated policing may be more efficient (and therefore more beneficial) than calibrated sentencing.
Neoclassical criminal law and economics dictates that policymakers should set criminal sanctions to approximate the value of the harm caused by the defendant, modified by the probability of his punishment.11 When sanctions and harm are not strictly monetary in nature, there is a translation problem: The sanction that the government sets might not accurately reflect the harm the defendant causes, regardless of the probability of detection and punishment. Accordingly, some people are overdeterred because they are overly sensitive to prison, while others are underdeterred because they are unafraid of prison. While some might argue that these mistakes counteract and average out over time, Kolber sees overdeterrence and underdeterrence as imposing different, but added, costs.12
Overdeterrence fuels risk aversion and imposes excessive prison costs, causing prisoners to miss opportunities for contributing more productively to society. Underdeterrence, by contrast, permits criminals to commit more crimes (assuming they know in advance that jail is no big impediment) thereby causing more harm.
So overdeterrence and underdeterrence are costly. The issue, however, is not whether we should find ways to reduce these costs, but whether Kolber's scheme offers the best method for doing so. As I explain in this Response, calibrated sentencing is less optimal than its logical alternative: calibrating the probability of punishment, which I call "calibrated policing."13
Two factors explain the preference for calibrated policing over calibrated sentencing. First, public law enforcement agencies have limited resources. They can investigate, indict, and litigate only so many cases. Second, public law enforcement is largely a probabilistic enterprise. Through trial and error, and with the benefit of some sophisticated analysis if they are lucky, policymakers and high-level police personnel distribute resources based on estimations either that a particular type of crime will occur, or that a particular geographic area can be made less crimogenic through additional policing or sanctions. These decisions may be further clouded by predictions as to how a particular type of policing will affect a given community.
Given the fact that law enforcers have to make decisions on the basis of imperfect information, it is not surprising that they often find shortfalls in deterrence. Some people will be overdeterred by a given combination of sanctions and policing, whereas others will be underdeterred. Policymakers then need to adjust the formula. They can do so by increasing overall sanctions or overall enforcement, or by calibrating sanctions or calibrating policing.
The problem with increasing overall sanctions—however attractive this tactic may seem—is that it increases the overall pool of overdeterred individuals without necessarily improving the underdeterred side of the equation. This is so because the underdeterred may be overly myopic, overly optimistic about their ability to avoid detection, or thrill seekers who like to violate laws. Moreover, increased sanctions will usually translate into more (and more expensive) prison expenditures. Finally, if sanctions for lesser crimes are pushed too high, we lose the benefit of "marginal" deterrence. If stealing $100 results in the same punishment as stealing $1,000, criminals have no reason to stop at $100.
Increasing overall enforcement is also problematic insofar as it requires society to devote more resources to policing. More important, it is unnecessary because the police can instead calibrate their efforts. If, for a given pool of putative defendants, Group A is initially overdeterred (due to its excessive fear of a given sanction) and Group B is initially underdeterred (because its members are substantially less sensitive to that sanction), law enforcement agents can transfer enforcement focus from Group A to Group B without requesting additional resources or increasing the overall crime rate. Even though the probability of detection decreases for Group A, its members remain compliant with the law because they were already overdeterred. Accordingly, Group A "self-regulates."14
So the question is not whether we should increase overall sanctions or overall enforcement, but rather, whether it is better to calibrate punishment or calibrate policing.15 Apart from the administrative concerns that Kolber addresses in his essay,16 a number of conditions might favor calibrated policing over calibrated sanctions. They include the fact that "insensitivity" may be a local or temporary phenomenon and therefore better solved more flexibly through increased law enforcement focus. The transaction costs involved with calibrated sentencing (likely to require some sort of legislative action, at least at the outset) will likely exceed the costs of calibrated policing, which is primarily the domain of the police.
In addition, calibrated sentencing may veer too far toward helping "sensitive" individuals reduce their sentences, without appropriately increasing the sentences of "insensitive" individuals. This is so because sensitive individuals have far more incentive to publicly raise their subjective fears to courts and legislators than their underdeterred counterparts.
Even if we could detect one's responsiveness to punishment with a low error rate, we still should prefer calibrated policing to calibrated sentencing when underdeterred individuals maintain an overly optimistic low probability of apprehension and punishment.17 If a defendant believes there is only a one percent chance he will be punished, the expected value of the "extra" sentence for the insensitive individual is also reduced to one percent of that amount. Even worse, punishment is a remote event in the distant future. Discounts therefore figure into the criminal's calculation of the "disutility" of punishment.18 Just as a dollar today is worth more than a dollar tomorrow, the disutility of prison today is worth far more than the disutility of prison four years from now. According to some accounts, criminals discount future events more than law-abiding individuals.19 If the psychological factors that cause someone to be insensitive to the reputational and physical stresses of prison also cause them to severely underestimate their likelihood of punishment, heightened policing provides far more benefit than harsher or lengthier terms of confinement.
Some may question a state's ability to identify overdeterred and underdeterred persons at the policing stage.20 Signaling and data collection can greatly assist the police in figuring out who needs more policing and who needs less. That is, those who wish to receive less police oversight may take certain steps to signal their adherence to law.21 By the same token, some individuals will signal their unwillingness to comply with the law, primarily by breaking it repeatedly. Calibrated policing is thus a story about how the state learns from its prior enforcement mistakes. The aggregation and analysis of statistical data permits police departments to assess risks and shift resources over time. The use of data to calibrate policing is in fact the basis of "COMPSTAT," the widely praised program that the New York Police Department and other departments across the country adopted in the 1990s and thereafter.22 That is not to say the system is perfect; the data can be flawed and the response can be either premature or overdue.23 But the system carries with it the ability to adjust relatively quickly to new information and unforeseen circumstances.
To sum up, societies must make decisions on how to distribute policing resources and sanctions. When they do so, they will inevitably make mistakes because their decisions are based on probabilistic predictions of how people will react to policing and sanctions as a whole. To the extent calibrated policing allows police to focus more on the underdeterred, yet also benefits from the self-regulation of the overdeterred, society might well embrace it as the most sensible and cost-effective option for adjusting to changing conditions in law enforcement.
II. Regulation of Others
Another reason we may prefer objective sentencing to Kolber's proposal is that sensitive individuals might assist the police not only by regulating themselves, but also by regulating others. Following most economic models of deterrence, Kolber's model presumes that defendants commit crimes individually.24 In real life, however, much criminal conduct is executed by groups. For crimes as diverse as corporate accounting fraud to complex drug dealing and gang violence, groups of individuals plan, execute, and cover up their conduct. Group crime poses two particular problems. First, because it leverages the abilities of multiple perpetrators, it often causes greater harm than a singular offender.25 Second, because it spreads responsibility out among different perpetrators and often shields the most dangerous of those perpetrators from public view, it is difficult to detect and punish.26
Even legitimate group settings (a publicly owned company, for example) can create conditions for criminal conduct to thrive over surprisingly long periods of time. Antisnitching27 and loyalty norms may cause employees who suspect wrongdoing to keep quiet rather than report wrongdoing to superiors within the company or outside investigators.28 Cognitive biases may cause witnesses to interpret questionable behavior as legitimate business conduct rather than a violation of law.29 Finally, collective action dynamics may result in an atmosphere in which everyone assumes that someone else should (or will) take care of the problem, and yet no one does.30
In sum, groups can promote, hide, or acquiesce in bad behavior. Because they can do such harm, much of substantive and procedural criminal law is aimed at enabling law enforcement agents to penetrate group conduct, either by criminalizing the group's existence, infiltrating it with undercover officers, or offering incentives to group members to cooperate with law enforcement agents and report on the rest of the group.31
Just as excess sensitivities to punishment may benefit society by creating "self-restraint" when the probability of detection is low, they may be of even greater value in restraining criminal conduct in group settings. The putative criminal who is overdeterred ex ante by a given sentence may, in addition to regulating herself, also regulate others to reduce the risk that she will be wrongfully associated with (or incidentally harmed by) members of a criminal group.
This phenomenon manifests itself in several ways. An overdeterred individual may be so risk averse that she declines to deal with people who she perceives as "shady." She may be so wary of criminal conduct that she feels the need to report any suspect activity to authorities in order to demonstrate her bona fides. She may even attempt to convince others not to engage in a given criminal act because she fears the consequences of such conduct for herself and her colleagues. If our ordinary inclination is to mind our own business and leave others to their own devices, then oversensitivity to punishment may be the force that causes us to abandon the mind-your-own-business norm and our natural unwillingness to confront others with evidence of wrongdoing. In sum, "excessive" fear of punishment may transform otherwise self-interested individuals into monitors and whistleblowers, thereby enhancing and extending the police powers of the state.
Whether overdeterred individuals make themselves obvious to others, or whether they serve as a hypothetical risk, they impose a cost on criminal groups insofar as they force those groups to invest time and effort in avoiding detection. Instead of spending time harming others, the criminal group must devote limited resources to avoiding getting caught.
This is not to say that the above phenomena are all normatively desirable or dependent solely on variations in deterrence. Social norms would likely promote a desire to avoid unethical behavior, regardless of whether individuals were particularly fearful of imprisonment.32 More important, a system that overly relies on individuals to police each other, or encourages them to do so excessively, may create problems far greater than the original harm the government seeks to avoid.33 This is particularly the case when the distrust that permeates criminal matters spills over into ordinary and desirable group conduct, such as community action or legitimate business activity. Nevertheless, the overall benefits of these mechanisms may well outweigh the costs of eliminating them outright. Proposals such as Kolber's therefore must be evaluated with group dynamics (and group-based crimes) in mind. If calibrated sentences on the back-end reduce private "policing" on the front-end, we may find ourselves with more group-oriented (and perhaps more dangerous) crimes.
Conclusion
For matters of deterrence, sentencing and policing are interdependent. Viewed piecemeal, the argument for calibrating sentences makes sense. A holistic analysis of law enforcement, however, yields different conclusions. Without proper attention to policing, calibrated sentencing can produce less deterrence and more crime, an outcome society ordinarily would choose to avoid.34
One of the reasons deterrence is such a difficult notion is that we must translate the often unquantifiable harm that someone has caused into a punishment that, in many respects, is also unquantifiable. At the same time, we must make numerous probabilistic calculations about how people will respond to the manner in which we define, police, prosecute, and punish a wide swath of criminal conduct. Kolber is indeed correct that one person's subjective experience of a given prison sentence can vary wildly from another person's experience. But it is not at all clear that uncalibrated punishment leads to "massively inaccurate"35 deterrence, particularly if one takes into account the monitoring and detection that precedes the imposition of punishment. Whatever the plausibility of implementing calibrated sentencing, it is impossible to assess its value without also taking into account the manner in which we police criminal conduct.
*Assistant Professor of Law, Brooklyn Law School. The author gratefully thanks Adam Kolber for his comments on this Response.
1 Adam J. Kolber, The Subjective Experience of Punishment, 109 Colum. L. Rev. 182 (2009).
2 Id. at 183–84.
3 See generally Gary S. Becker, Crime and Punishment: An Economic Approach, 76 J. Pol. Econ. 169 (1968); Richard A. Posner, An Economic Theory of the Criminal Law, 85 Colum. L. Rev. 1193 (1985).
4 Kolber, supra note 1, at 217–18.
5 See id. at 188–89 (noting "courts sometimes accommodate objective differences in prisoners' likely conditions of confinement" and providing United States v. Blarek, 7 F. Supp. 2d 192 (E.D.N.Y. 1998), as an example).
6 See, e.g., U.S. Sentencing Guidelines Manual § 4A1.1 (2008) (calculating defendant's "Criminal History" level on basis of prior convictions); id. ch. 5, pt. A (setting up sentencing table that dramatically increases defendant's sentence depending on Criminal History category); see also Carissa Byrne Hessick, Why Are Only Bad Acts Good Sentencing Factors?, 88 B.U. L. Rev. 1109, 1112 (2008) (observing that "all American jurisdictions . . . in some form or another" take defendant's "bad acts," including his prior convictions, into account at sentencing).
7 See Manuel A. Utset, Hyperbolic Criminals and Repeated Time-Inconsistent Misconduct, 44 Hous. L. Rev. 609, 621 (2007) ("Since for the purpose of setting optimal sanctions all instances of the same crime are identical, the neoclassical approach does not draw a distinction between the one-time and serial offender.").
8 If these facts place the case beyond a prescribed statutory maximum, they must be proved to a jury. Apprendi v. New Jersey, 530 U.S. 466, 490 (2000) ("Other than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt."). Otherwise, courts may rely on these facts to increase the defendant's sentence but are not necessarily bound to do so. See United States v. Booker, 543 U.S. 220, 250 (2005) (finding provision making sentencing guidelines mandatory "along with those inextricably connected to it" to be constitutionally invalid as violating Sixth Amendment).
10 Although he does not specifically refer to the term, Kolber dismisses calibrated policing on the grounds that police cannot predict subjective reactions to punishment. See Kolber, supra note 1, at 218 n.101.
11 See supra text accompanying note 3.
12 Kolber, supra note 1, at 217–18.
13 The probability of punishment is the multiple of the likelihood that one will be detected, prosecuted, and punished. Miriam H. Baer, Linkage and the Deterrence of Corporate Fraud, 94 Va. L. Rev. 1295, 1306–07 (2008).
14 See Robert Innes, Violator Avoidance Activities and Self-Reporting in Optimal Law Enforcement, 17 J.L. Econ. & Org. 239, 239–40 (2001). Innes describes the benefits of "self-reporting" which is a somewhat different concept from self-regulation. Self-regulators refrain from criminal conduct, whereas self-reporters admit what they have done. Nevertheless, self-regulators and self-reporters produce similar effects in that they require less monitoring.
15 See Louis Kaplow & Steven Shavell, Accuracy in the Determination of Liability, 37 J.L. & Econ. 1, 3 (1994) (opining that when it is expensive to invest in improving accuracy of sanctions, "it may be more efficient to raise the level of enforcement effort").
16 Kolber, supra note 1, at 219–28 (discussing administrative challenges, including predicting and monitoring subjective experiences of punishment).
17 For the notion that probability should be increased to reduce criminals' mistakes in calculating the value of expected sanctions, see Lucian Ayre Bebchuk & Louis Kaplow, Optimal Sanctions When Individuals Are Imperfectly Informed About the Probability of Apprehension, 21 J. Legal Stud. 365, 366–67 (1992) (demonstrating that errors will be greater when probabilities of punishment are low because even small changes in probability will be multiplied against very large sanctions). For the argument that enforcement agents should focus on increasing probability of punishment when defendants discount or experience decreased disutility in regard to marginal increases in sanctions, see A. Mitchell Polinsky & Steven Shavell, On the Disutility and Discounting of Imprisonment and the Theory of Deterrence, 28 J. Legal Stud. 1, 11 (1999).
18 Polinsky & Shavell, supra note 17, at 3–4.
19 See, e.g., Yair Listokin, Crime and (with a Lag) Punishment: The Implications of Discounting for Equitable Sentencing, 44 Am. Crim. L. Rev. 115, 124 (2007) (collecting sources indicating that criminals are particularly myopic).
20 Kolber dismisses the ability of a state to do this. See Kolber, supra note 1, at 218 n.101.
21 Cf. Jodi L. Short & Michael W. Toffel, Coerced Confessions: Self-Policing in the Shadow of the Regulator, 24 J.L. Econ. & Org. 45, 62 (2008) (theorizing in regulatory context that firms might self-report violations as means of signaling pro-compliance stance to regulators).
22 COMPSTAT is a management tool that police commissioners use to gather statistical information on a regular basis and hold precinct commanders accountable for policing deficiencies. "Combining cutting-edge crime analysis and geographic information systems with state-of-the-art management principles, COMPSTAT burst onto the scene when it was first implemented in 1994 by then-Commissioner William Bratton of the New York City Police Department (NYPD)." James J. Willis, Stephen Di Mastrofski & David Weisburd, Making Sense of COMPSTAT: A Theory-Based Analysis of Organizational Change in Three Police Departments, 41 Law & Soc'y Rev. 147, 148 (2007).
23 Moreover, we must monitor the police to make sure that "overdeterred" and "underdeterred" are not proxies for racial or socioeconomic characteristics. These issues also arise, however, in calibrated sentencing schemes to the extent one views sensitivity as a social construct and not some scientifically proven fact.
24 See Thomas J. Miceli & Kathleen Segerson, Punishing the Innocent Along with the Guilty: The Economics of Individual Versus Group Punishment, 36 J. Legal Stud. 81, 81 (2007) (noting that "[i]n standard economic models of law enforcement, the chief objective is first to identify, and then punish, the offender," but "there are many enforcement contexts in which the identity of the offender is uncertain but he or she is known to be a member of a well-defined group").
25 See, e.g., Neal Kumar Katyal, Conspiracy Theory, 112 Yale L.J. 1307, 1315–28 (2003) (explaining importance of group interactions to success of criminal conspiracies).
26 See Samuel W. Buell, The Upside of Overbreadth, 83 N.Y.U. L. Rev. 1491, 1535 (2008) (discussing benefits of RICO statute in overcoming "incentives of an organization's leaders to delegate acts to underlings and thicken the insulation within criminal groups").
27 Antisnitching campaigns have become prevalent in the context of urban policing. See, e.g., Richard Delgado, Law Enforcement in Subordinated Communities: Innovation and Response, 106 Mich. L. Rev. 1193, 1204–06 (2008) (discussing rise and effectiveness of "stopsnitching" campaigns).
28 See Geoffrey Christopher Rapp, Beyond Protection: Invigorating Incentives for Sarbanes-Oxley Corporate and Securities Fraud Whistleblowers, 87 B.U. L. Rev. 91, 121–22 (2007) (describing "social ostracism" that can impede employees from reporting corporate and securities fraud).
29 See Donald C. Langevoort, Monitoring: The Behavioral Economics of Corporate Compliance with Law, 2002 Colum. Bus. L. Rev. 71, 85–88 (explaining that supervisors in corporate settings will interpret ambiguous conduct as consistent with the law to avoid, among other things, unpleasant confrontations or calling into question previous hiring or promotion decisions).
30 Collective action problems occur when the costs of a particular action outweigh its benefits to an individual, even though the action benefits the group as a whole. See generally Mancur Olson, The Logic of Collective Action: Public Goods and the Theory of Groups (1971) (explaining foundations of collective action problems and arguing that large or latent groups are unlikely to act in their common interest); Dan M. Kahan, The Logic of Reciprocity: Trust, Collective Action and Law, 102 Mich. L. Rev. 71 (2003) (explaining collective action problems, how they impact criminal law, and how they may be overcome by feelings of trust and reciprocity).
31 See Buell, supra note 26, at 1527–36 (examining doctrinal development of racketeering law).
32 On the other hand, other norms might "obligat[e] silence." Richard H. McAdams, The Origin, Development and Regulation of Norms, 96 Mich. L. Rev. 338, 348, 350–51 (1997) (defining norms as nonlegal obligations).
33 See Miriam H. Baer, Corporate Policing and Corporate Governance: What Can We Learn from Hewlett-Packard's Pretexting Scandal?, 77 U. Cin. L. Rev. (forthcoming 2009) (manuscript at 55–57, on file with the Columbia Law Review), available at http://ssrn.com/abstract=1133117 (describing benefits and drawbacks of such policing in corporate context).
34 Of course, society may have other considerations, but this Response deals only with the consequentialist implications of Professor Kolber's argument.
35 Kolber, supra note 1, at 186.
------------------------
Preferred Citation: Miriam H. Baer, Evaluating the Consequences of Calibrated Sentencing: A Response to Professor Kolber, 109 Colum. L. Rev. Sidebar 11 (2009), http://www.columbialawreview.org/Sidebar/volume/109/11_Baer.pdf.
