Not all police misconduct is the same, and different institutional regimes might manage different sorts of misconduct most effectively. This Article surveys the universe of police malfeasance from the perspective of an important but underappreciated regulatory regime: liability insurance. Nearly all but the very largest municipalities buy insurance that covers claims alleging police misconduct. In assuming the financial risk of bad police behavior, the insurers become motivated to prevent it. Criminal procedure scholarship almost entirely overlooks the salutary regulatory influence these insurers may have on police activity. Yet insurance is no panacea. Indeed, a principal aim of this Article is to probe the limits of the insurance mechanism—the places where the effects of insurance on policing are likely weak or even perverse. This exercise points us toward a typology of misconduct, along with a corresponding set of plausible approaches for reducing the occurrence of each of the types identified. In particular, the Article distinguishes varieties of police misconduct based on (1) the dollar-value of the legal claims to which they give rise and (2) the length of the delay between when the misconduct occurs and when a legal claim is typically filed. The typology suggests, among other things, that the insurance regime is a plausible surrogate for some governmental regulation of police violence but not, at present, of the sorts of misconduct that lead to wrongful convictions.

TABLE OF CONTENTS

I. Introduction

Not all police misconduct is the same, and different institutional regimes might combat different sorts of misconduct most effectively. This Article surveys the universe of police misconduct from the perspective of an important but underappreciated regulatory regime: liability insurance. Most small and mid-sized municipalities in the United States purchase insurance that covers a range of police misconduct claims, from improper service of process to outright assault and battery, discrimination, and other civil rights violations.1 In assuming the financial risk of bad police behavior, the insurers become motivated to marshal their substantial resources to prevent it. Criminal procedure scholarship almost entirely overlooks the salutary “regulatory” influence these insurers may have on police activity, often reasoning, mistakenly, as though little or nothing stands between judicial opinions or statutes and officer discretion.2 Understanding how insurers regulate the police to reduce liability and defense costs is crucial in grasping how civil rights lawsuits deter police misconduct.

Yet insurance is no panacea. Indeed, a principal aim of this Article is to probe the limits of the insurance mechanism—the places where the effects of insurance on policing are likely weak or even perverse, suggesting a need for insurance reform or other, more familiar regulatory interventions. These include domains in which moral hazard—that is, the propensity of insurance to reduce the insured’s incentives to prevent harm—seems most likely to predominate. By laying these cases alongside those in which regulation-by-insurance appears to function well, I show how thinking about police misconduct through the lens of liability insurance points us toward a typology of misconduct, along with a corresponding set of plausible approaches for reducing the incidence of each of the types identified. My typology suggests, among other things, that the insurance regime is a plausible surrogate for some governmental regulation of police violence but not—at least not yet—of the sorts of misconduct that lead to wrongful convictions.

To construct my typology, I distinguish types of police misconduct along two dimensions. First, I consider the dollar value of the legal claims to which each type of misconduct typically gives rise.3 Low-dollar-value claims are poor fodder for civil damages suits and, indeed, rarely arise in that posture. Instead, criminal defendants assert these claims defensively, to resist prosecution. As a class, these claims pose little threat to insurers. Insurers therefore have little incentive to regulate the underlying conduct. High-dollar claims, in contrast, create sufficient financial incentive to induce potential plaintiffs—and plaintiffs’ attorneys—to sue. They give insurers reason to pay attention. This does not mean, however, that civil damages suits work equally well to deter all types of high-dollar misconduct. A second crucial consideration is the length of time between the occurrence of misconduct and the filing of a lawsuit. In insurance parlance, a claim of police misconduct may have either a “short tail” or a “long tail.”4 Long-tail claims, I will explain, pose great difficulties for insurers, and there are reasons to doubt that insurers are effectively managing the risks that give rise to these claims.

These two criteria—the size of a claim’s expected payout and the length of its tail—generate four categories of police misconduct. The Article’s structure is accordingly straightforward: In each of the four sections that follow, I describe one category of misconduct along with a corresponding set of potential solutions. I begin in Part II with high-dollar, short-tail claims. The archetype here is a claim that alleges the use of excessive force. When a police officer employs unlawful force, the harm he inflicts is often substantial (high-dollar)—especially where death results—and also immediately apparent and actionable (short-tail). Drawing on original qualitative research of the police liability insurance industry, I explain how insurers work to manage the problem of police violence and suggest why they may fare better in some regards than traditional sources of regulation.5 The potentially provocative implication is that, despite present political pressure to develop public-law solutions to police violence, limited government resources may be better spent fighting other kinds of law enforcement misconduct. To be clear, I do not suggest that private regulation by insurers can or should substitute for all public regulation of police use of force—only that, if public regulatory resources are scarce, we might direct them elsewhere and let insurers, where they operate, shoulder some of the load in this domain.

In Part III I turn to high-dollar, long-tail claims. Here I have in mind “wrongful conviction” claims filed by individuals who have been imprisoned and later exonerated of criminal wrongdoing. The average exoneree spends over thirteen years wrongfully incarcerated (long-tail). Damages for this sort of harm can be immense (high-dollar). Not all wrongful conviction claims allege police misconduct, but many do. And when they do, they typically fall within the standard police liability policy. Insurers, therefore, have a financial incentive to prevent the kinds of misconduct that lead to wrongful convictions. But the long tail creates complications that make the risk difficult for insurers to price and manage. The relative infrequency of the claims exacerbates these challenges. In this context, insurers may actually make things worse rather than better by creating (but not controlling) moral hazard. And because the rate of exonerations is rising,6 it is reasonable to fear that wrongful conviction cases will, before too long, disrupt this corner of the liability insurance market. With an eye on this possibility, I suggest two sets of reforms—one focused on improving the insurability of the underlying risk and the other on alternative, non-insurance-based regulatory mechanisms for reducing police malfeasance that leads to wrongful convictions.

The third and fourth categories of police misconduct in my typology are largely—though not entirely—outside insurers’ purview. I therefore deal with them more quickly. The third category, addressed in Part IV, contains low-dollar, short-tail claims. This includes many run-of-the-mill violations of law, like an investigatory stop unjustified by reasonable suspicion or the failure to give Miranda warnings before custodial interrogation.7 Although the harm from these interactions manifests immediately (short-tail), it is largely non-compensable in the eyes of the law (low-dollar). To police these violations, we rely heavily on criminal defendants to act as private attorneys general, and on the occasional suit for injunctive relief. These tools undoubtedly help, but more is needed. Drawing on prior work,8 I suggest how we might reorient some of our constitutional doctrine to focus defendants’ claims on systemic rather than individualized concerns.

The last category, the topic of Part V, involves low-dollar, long-tail claims. It is not immediately obvious what, if anything, belongs in this box. But I will make the case that one unique type of claim plausibly does, at least some of the time: a claim alleging racial profiling by the police. Under extant doctrine, absent a smoking gun, a colorable profiling claim must marshal statistical evidence from a broad sample of other cases. This creates, in many cases, a delay (i.e., a long tail) between the profiling and the potential lawsuit brought to challenge it—a delay that distinguishes these claims from mine-run (short-tail) constitutional violations. And here too, even a successful plaintiff is unlikely to recover much in damages (low-dollar). The unfortunate upshot, I think, is that neither civil-damages actions nor criminal litigation will work very well to combat racial profiling. Recourse to more traditional forms of regulation, like legislation and administrative rules, is therefore necessary. Criminal procedure doctrine might be re-fashioned to encourage this political regulation, I will suggest, but I do not think that doctrine alone can save the day.

 

 

Figure 1 pulls together the strands of the typology:

  High-dollar Low-dollar
Short-tail Use of force Terry; Miranda
Long-tail Wrongful convictions Racial profiling

Figure 1

 

This model, of course, is only one of myriad schemes we could sketch to organize the universe of police misconduct. Those with different frameworks or objectives might demarcate categories along different dimensions. My approach is useful because it tells us something about what solutions are likely to gain traction on each type of misconduct. But it is not the only way to think about the topic.

Two caveats are necessary before proceeding. First, in my descriptive case—that is, in determining how to classify different types of police misconduct—I take legal doctrine on liability and damages as given. This is why I refer to claims as “high-dollar” and “low-dollar” rather than “high-value” and “low-value”: I mean to convey nothing about the normative value of different claims, only their dollar value under legal precedent. Second, I largely obscure the distinction among different types of insurance providers. As I explain elsewhere, although most municipalities purchase police liability insurance on the market, that market is segmented between commercial insurers and intergovernmental risk pools.9 And the largest municipalities tend to “self-insure.” In large municipalities that take self-insurance seriously, and replicate in house much of what market insurers do, the distinction between self-insurance and market insurance may be inconsequential. But where “self-insurance” really just means “going bare”—which may be the typical case10 —what I say about the influence of insurance on policing likely does not apply, at least not in full. Unfortunately, too little is known about municipal self-insurance for police liability to permit any confident generalizations, which is why I do not separately analyze self-insured and market-insured municipalities.

My primary aim is to help organize and systematize our thinking about how to police the police. Even those who disagree with the specifics of my typology, I hope, will take away two thematic points: First, “police misconduct” is a capacious and variegated concept, and strategies that are necessary or effective to combat one kind of misconduct may not be necessary or effective to fight another. Second, one cannot fully understand how our legal system does and can deter police misconduct without understanding the role that liability insurers play in that endeavor, both for good and ill.

II. High-Dollar, Short-Tail

Fourth Amendment doctrine prohibits the police from employing excessive force in effecting a search or seizure.11 Police who violate this stricture may be ordered to pay compensation for personal injuries they inflict. Damages, especially in cases involving debilitating injury or death, can be quite high; lawsuits stemming from recent high-profile officer-involved deaths have settled for around five or six million dollars each.12 And the types of injuries these cases involve typically manifest, and are actionable, immediately after the misconduct occurs. Those two characteristics make use-of-force claims high-dollar, short-tail claims.

Because the financial stakes can be so high—and because the use of excessive force is typically not causally related to the discovery of incriminating evidence—use-of-force claims are usually litigated in civil suits rather than defensively through suppression motions in criminal cases.13 Scholars have raised serious questions about how well civil rights suits work to deter police misconduct, and to restrain the use of force in particular.14 Some point out, for example, that “excessive force doctrine is extraordinarily abstract.”15 “This uncertainty in legal authority,” one commentator argues, “results in a lack of institutional guidance and leaves police officers to exercise their own discretion.”16

I take no issue with the premise that excessive force doctrine is abstract and uncertain. But it does not follow that police officers have free rein. That conclusion follows only if we ignore other sources of police regulation, including (though certainly not limited to) regulation-by-insurance. The standard police liability policy, however, covers excessive-force claims,17 so insurers, not municipalities, typically bear the financial risk of police violence. For reasons that I (and others) have explained,18 this financial arrangement gives insurers the incentive to reduce the frequency of these claims. Indeed, when a prominent risk management expert compiled a list of twelve “high risk/critical tasks” in policing that warrant the attention of insurers and risk managers, use of force topped the list.19

In Section A below, I describe the regulatory techniques insurers use to manage the risk of excessive force. In Section B, I discuss the interaction between regulation-by-insurance and more traditional modes of public regulation.

A.    How Insurers Regulate the Use of Force

As I detail in other work, insurers use a variety of tools to try to tame police violence, including operational policy development and education, training, auditing, and risk-responsive underwriting and rating. Insurers invest substantially, for example, in improving covered agencies’ policies on the use of force. At the outset of the insurance relationship, the insurer typically requests copies of the agency’s policies.20 The insurer, sometimes contracting with outside attorneys or expert consultants, then works with the agency to ensure the policy incorporates industry best practices.21 In particular, insurers encourage agencies to adopt a “use of force continuum” that specifies the degree of force appropriate in various scenarios, and to focus on de-escalating tense citizen encounters.22 At least some empirical evidence suggests that these policies matter—that a good use-of-force policy helps reduce the use of excessive force.23 Throughout the insurance relationship, insurers also disseminate written materials and videos designed to educate officers (and agency leadership) about the appropriate use of force, and some even engage in classroom instruction.24 Many insurers also provide financial incentives for agencies to seek accreditation from a recognized authority, such as the Commission on Accreditation of Law Enforcement Agencies.25 Accreditation requires, among other things, continuous compliance with a thirteen-part set of standards on the use of force.26

Importantly, insurers’ efforts extend well beyond the Fourth Amendment’s ambit into an array of extra-legal considerations that likely affect the frequency and severity of use-of-force events. Consider four brief examples. First, insurers educate officers on managing the significant stresses of the job.27 Officers who handle stress poorly, the evidence suggests, are more likely to act out.28 Second, insurers encourage psychological testing of each job applicant “to ensure that the applicant is free of mental illness or other defect that would render him or her incapable of self-control or appropriate behavior in positions of authority and/or responsibility.”29 Third, insurers increasingly tout the harm-reducing potential of body-worn cameras. At a recent conference for the Association of Governmental Risk Pools, for example, one speaker, addressing a group of municipal insurers, discussed studies finding that body-worn cameras had reduced both the use of force and citizen complaints about the use of force.30 And fourth, insurers have begun efforts to reduce the risk of harm from police departments’ use of military equipment. Although I have not seen insurers purport to forbid agencies from employing military equipment, “police departments with access to military equipment,” one insurance newsletter admonishes, must take care to have “appropriate training and deployment standards in place.”31

Insurers also help police departments train their officers by supplying materials, funding training programs, or even purchasing costly training equipment. Especially relevant here, some insurers subsidize expensive “virtual reality” training on use-of-force simulators.32 These “[m]odern computerized use of force simulators expose police officers to highly realistic and interactive scenarios whereby they can learn appropriate responses using the full range of use of force options available.”33 Rather than simply honing marksmanship, these tools teach officers to apply their skills appropriately under “field-compatible” conditions.34 In empirical research, such “simulation training has been demonstrated to increase the number of preventative actions taken by police officers, enhance shooting accuracy, reduce the number of shots fired to achieve an objective, increase the degree to which police officers use cover, and decrease the number of unjustified shootings.”35

Insurers also facilitate training about how not to use force. Another presentation at the Association of Governmental Risk Pools conference focused on using mediation and alternative dispute resolution techniques on the policing beat.36 Many insurers offer courses on verbal de-escalation techniques, subsidized for policyholders.37 And in a newsletter recommending measures to reduce the use of force, one insurer advises that “a comprehensive training program should be conducted by outside personnel and focus on defusing incidents.”38

One ubiquitous challenge in policing the police is ensuring continued compliance with departmental policies. Adopting good policies and procedures, that is, and training on those policies when officers join the force, may get a department off to the right start, but proper maintenance is essential.39 Accredited agencies must demonstrate their continued compliance with accreditation standards, though the extent of those obligations is unclear.40 Insurers address the compliance problem by auditing agencies on a regular basis—anywhere from semi-annually to once every three years, according to the experts I interviewed.41 Insurers send auditors or consultants to visit insured agencies, sometimes for two- or three-day stints. The auditors review police reports, internal affairs files, and other liability-related documentation.42 They may go out in the field with the chief or other officers.43 Insurers also use data on claims involving the use of force to identify troubled agencies and problem officers.44 Agencies that make an insurer’s “watch list” are audited more frequently and intensely. These auditing practices seem to capture at least the spirit of the “early warning systems” many policing experts have praised.45

None of what I’ve said so far, importantly, addresses whether municipalities listen to their insurers—whether they actually do what insurers say they ought to do. The answer, generally speaking, seems to be yes.46 Insurers have two principal ways to induce agencies to cooperate with loss-prevention initiatives: they can raise rates and threaten to terminate coverage. Insurers charge more to agencies that frequently tender claims, just like your auto insurer does to you; and they lower prices for agencies that demonstrate commitment to loss prevention, such as by obtaining accreditation. Some also adjust rates based on the existence and quality of various departmental policies. And when rate adjustments are not enough to make municipalities listen, insurers can terminate coverage, or credibly threaten to do so. At least partly in response to these incentives, police agencies adopt or amend departmental policies on important subjects like the use of force and firearms. They change the way they train their officers. And they even fire problem officers, all the way up to the chief.47 In extreme cases, municipalities have shuttered police departments after their insurers pulled coverage.48

According to one expert with several decades of experience in the industry, municipal liability insurers are more heavily focused on policing now than at any time since the early 1990s. The Rodney King beating in 1991, this expert said, had “ripple effects” throughout the industry. Insurers sought to ensure that police agencies had adequate policies and procedures on the use of force and related subjects. After a while, though, attention waned as other sources of municipal liability captured insurers’ interest. Now, after the recent wave of highly publicized officer-involved deaths, insurers find themselves “back in the soup.” Many insurers, moreover, now recognize that the problems with police go “beyond policies and procedures”; in order to reduce misconduct, insurers—and we, as a society—“need to find the root cause.”49

B.    Private Regulation, Public Regulation

Meanwhile, politicians, too, have considered how to respond to recent police violence. According to the National Conference of State Legislatures, in the year 2015 alone, twenty-four state legislatures considered bills to address investigatory procedures for officer-involved deaths.50 Measures in thirteen of those states concern the appointment of special prosecutors. Seven states are evaluating bills about the collection of data and tracking of statistics in officer-involved deaths. Other related bills deal with chokeholds, body-worn cameras, the public’s ability to film law enforcement, and military equipment.

These are positive developments that I do not wish to disparage in any way. They may be crucial for regulating self-insured departments that do not take loss prevention seriously, which likely describes many major metropolitan agencies, including some that are (rightly) under the microscope today. And they may serve retributive and expressive purposes necessary to rebuild community trust in the police. Yet, from a regulatory (i.e., deterrence) perspective, I am skeptical about their capacity to improve meaningfully upon what insurers are already doing.51 In fact, insurers’ private regulation may well be more nuanced, nimble, and data-driven than what state legislatures can accomplish.52 To the extent that legislative resources (or the political support necessary to marshal those resources) are scarce, one can make a case that they should be spent where the private regulatory system is absent or fails to function, rather than where it seems to function best.

Instead, we might devote public efforts to supporting the regulatory function insurers provide. We could consider, for example, mandating market insurance for municipalities that cannot demonstrate a good-faith commitment to loss prevention. We might also examine whether there are legal threats that cause insurers to shy away from more intensive regulation. Although, as I mentioned above, some insurers have successfully pressured agencies to terminate problem officers, others I interviewed expressed fear that doing so might subject them to liability, or at least to legal action, under employment, labor, or contract law.53 If this fear is well founded, it may be worth creating narrow safe harbors from liability for insurers to remove a disincentive to socially beneficial risk regulation.54

III. High-Dollar, Long-Tail

An individual who is wrongfully convicted, incarcerated, and later exonerated and freed can generally sue for damages for the time spent unjustly imprisoned.55 Successful plaintiffs—by one estimate, twenty-eight percent of all those exonerated by DNA who sue56 —have in some cases obtained judgments and settlements upwards of one million dollars for each year of incarceration.57 Multiply that by the thirteen years the average exoneree spends in prison,58 and it’s not hard to see how enormous the liability risk can be. Police misconduct, my focus here, contributes to many, though not all, of these wrongful convictions.59 The leading police-related causes of wrongful conviction include erroneous eyewitness identification, faulty forensic evidence, false informant testimony, and false or coerced confessions.60 Where police are implicated, police liability policies generally cover the claims.61

One might think, therefore, that insurers would be highly attuned to the risk of wrongful convictions and, as with the use of force, would closely regulate the agencies they insure in an effort to reduce that risk. But the long tail of most wrongful conviction claims—the delay between when the wrongful conduct occurred and when the claim is filed—makes the claims an insurance nightmare.62 Section A briefly explains why. Section B discusses potential responses to the difficulties insurers face.

A.    The Challenges of Insuring Long-Tail Risks

Insurance theory suggests several reasons that long-tail risks may strain insurers’ regulatory capabilities. The problem worsens when, as here, the risks are also low in probability but highly consequential. Sections 1 and 2 address these respective points. Section 3 reports qualitative empirical findings that tend to substantiate the challenges theory predicts.

1. The basic theory.

A long tail of liability creates at least three distinct problems for insurers. First, the long tail heightens the degree of correlation (i.e., statistical dependence) among the covered risks. All liability insurance, at least in theory, involves problems of correlated risk “if the rules under which liability is determined and damages are assessed change over the life of the insurance contracts.”63 In many insurance contexts, the law is unlikely to change much, but the delay entailed by long-tail risks exposes insurers to “judicial, legislative, and economic changes that commonly influence the ultimate determination of awards.”64 “For example, unexpected inflation over the runoff period could commonly increase the ultimate payoff on all outstanding claims beyond the amount reserved by the insurer. Similarly, a new judicial precedent or legislation can expand the area of liability, ease the burden of proof for future plaintiffs, or cause a common upward shift in the value of awards on all outstanding claims.”65 Even changes in prevailing attitudes toward the police and the criminal justice system can have a substantial (and correlated) effect.66 All this matters because, the more that risks are correlated, the less insurers can reduce the risks by aggregating them, which is one of their defining functions.67 This is why losses from risks that affect many policyholders simultaneously, like earthquakes, are commonly excluded from homeowners and renters policies.68

Second, the long tail can create incentives that lead insurers to regulate too little. “Because of turnover, risk managers may have a much shorter time horizon than the firm. Current decisionmakers may reap no reward within the organization for reducing remote risks and may even be penalized for expending current funds for doing so.”69 As a result, insurance managers may “externalize to the future,” maximizing their short-term results at the expense of long-term interests.70 This “coordination-across-time problem” is another reason that “latent harms . . . can put insurers in a poor regulatory position.”71

Third, and maybe most important, the long tail creates uncertainty about the number and magnitude of wrongful conviction claims insurers should expect. Here I mean to reference the distinction insurance theorists draw between risk and uncertainty. Risk refers to “a probability that can be estimated, whether on the basis of observed frequency or of theory.”72 Uncertainty refers to “a probability that cannot be estimated.”73 Risk, put differently, “is something that you can put a price on,” whereas uncertainty “is risk that is hard to measure.”74 The long delay between the collection of premiums and the processing of claims complicates measurement of a long-tail risk. “Even if the frequency and severity of future claims were predictable in current dollars, economic and legal inflation over such a long period would make assessment of ultimate financial exposure extremely speculative.”75

2. The additional difficulty of low-probability, high-consequence risks.

It seems, moreover, that the frequency of future wrongful conviction claims is not particularly amenable to prediction, largely because it is too low.76 A wrongful conviction claim is what is called a low-probability, high-consequence event, a phenomenon that consistently eludes human predictive capacity: “Much of the time, human beings ignore low-probability, high-consequence events, giving them far less attention than they deserve. But when people experience or see a relevant bad outcome, their concern frequently becomes exaggerated.”77 This may be why one insurer asserts that “[t]here is no way for [an insurer] to predict the likelihood or volume of exonerations and civil claims.”78

All this uncertainty gums up the insurance machine, for insurance deals far better with risk than uncertainty. Insurers faced with uncertainty cannot accurately price coverage. “[D]espite high levels of expertise and strong incentives to make logical decisions, [insurance managers] make errors with respect to situations where there is uncertainty or ambiguous information regarding the low probability risks they face. When insurers have limited data and limited past experience with extreme events, there is a tendency for them to engage in intuitive thinking when determining what coverage to offer against specific risks and how much to charge.”79 For this reason, “[i]t would not be unusual for a governmental risk pool to have little or no remaining [incurred but not reported] reserves in the accident year in which the trigger date falls,” i.e., the year in which a wrongful arrest or conviction occurred.80 “This means a single significant claim could put significant pressure on the pool’s surplus.”81

Nor can insurers be confident about what loss-prevention measures they can reasonably insist upon. As Kenneth Abraham explains, “the threat of uncertain liability can promote optimal safety levels only by mere chance, because risk-optimizing behavior requires cost-benefit calculations that are necessarily impossible in the face of great uncertainty.”82 And any attempt by insurers to adjust to new understandings of the scope of liability—through, say, raising premiums or crafting an exclusion—cannot have impact for years, when claims stemming from today’s conduct will arise. It can only cut future losses.83 As for the conduct covered long ago, which may give rise to a claim tomorrow, insurers cannot but concede the familiar truism, that “the past is the past.”84

3. The empirics.

My own empirical research, conducted for this Article, largely confirms what theory predicts. Exploiting contacts made while conducting a larger, related empirical project, I interviewed eight insurers—including both commercial firms and intergovernmental risk pools—about how they price and manage the risk of wrongful convictions.85 For the most part, they don’t. Some had not even given the issue any thought.86 The insurers confirmed that wrongful conviction suits are covered when law enforcement is implicated, absent some particular exclusion. But none of them attempts to determine the share of the premium attributable to wrongful conviction coverage. Nor, for the most part, do they engage in any loss prevention specifically designed to reduce the incidence of wrongful convictions. These insurers seemed simply to ignore the risk.87

A few insurers emphasized that, although they take no steps to address wrongful conviction risks in particular, their general underwriting and loss-prevention practices should sufficiently control the exposure. One stressed the importance of accreditation by agencies that impose continuing education standards and check for well-maintained policies and procedures.88 Another said his pool educates the police about their obligation to disclose exculpatory evidence and audits agencies’ evidence rooms for conformance with best practices.89 That pool also scrutinizes the integrity of agency personnel, reviewing the veracity of employment applications and performance during probationary employment periods.90 The same expert added that underwriters will notice a department that’s generally sloppy, based on responses to questionnaires about best practices.91 But, he added, tellingly, as for how such measures actually affect (and reflect) the risk of wrongful conviction, “There’s no science behind it.”92

My research therefore suggests that, to the extent insurers attempt to price the risk of wrongful convictions, their efforts are crude, at best. This means that, in this context—in contrast to what I said in Part II about the use of force—insurers may be making matters worse rather than better. “Failure to risk-rate premiums,” it is well known, can “create[ ] moral hazard.”93 Depressed prices, assuming that’s what we have, weaken incentives for loss prevention and send inaccurate signals to both insured municipalities and the public about the risk and cost of wrongful convictions.94

What is more, if experience in analogous domains is any guide, insurers’ ostrich-like approach to covering wrongful conviction claims may mask fragility and volatility in this corner of the market, which a run of wrongful convictions may lay bare. “After a severe loss, insurers may withdraw from covering this risk because they focus on the losses from a worst-case scenario without adequately reflecting on the [low] likelihood of this event occurring in the future.”95 This is what happened with coverage against terrorism risks. “[P]rior to 9/11, insurance losses from terrorism were viewed as so improbable that the risk was not explicitly mentioned or priced in any standard policy.”96 But “[f]ollowing the 9/11 attacks, most insurance swung to the other extreme.”97 The same thing occurred with pollution insurance.98 The few insurers I spoke to who do have an eye on the issue agreed that wrongful conviction coverage could meet a similar fate, especially if the present “soft” market firms up.99

B.    Responding to Insurance Shortfalls

If what I have said is correct, a two-headed reform agenda is in order. First, we should seek ways to shore up the insurance function, both to improve insurers’ capacity to regulate the risk of wrongful convictions and to avert the type of insurance crises we experienced with terrorism and pollution coverage. If changes are required, better to implement them in a proactive, orderly fashion than to walk headlong into a disruptive, destabilizing felt emergency. Second, cognizant that insurers do not appear to control this risk especially well, we should bolster other forms of regulation that do not rely on the threat of civil liability to create the incentives necessary for harm reduction.

1. Improving insurance.

This section considers two commonplace insurance devices that might allow insurers to gauge the risk of wrongful convictions more accurately: feature rating (rather than experience rating) and claims-made coverage (rather than occurrence-based coverage).

a. Experience rating vs. feature rating

Insurers use two principal techniques to tailor rates for their policyholders: experience rating and feature rating.100 Experience rating uses the insured’s history of past losses—its “loss experience”—during some designated period (say, five years) to calculate current premiums. Feature rating instead relies on the insured’s individual risk characteristics. (Experience rating might raise Richard’s auto premium because Richard had an accident last year; feature rating because Richard is only nineteen years old and young drivers are accident-prone.) Because wrongful convictions are a low-probability, long-tail risk, insurers lack the experience base necessary to do experience rating well.

As far as I can tell, however, insurers are not feature rating based on the risk characteristics known to affect the likelihood of wrongful convictions. For example, insurers could—but, to my knowledge, do not—raise rates for agencies that fail to videotape interrogations (to help avoid false or coerced confessions) or use double-blind lineup procedures (to reduce the danger of erroneous eyewitness identifications).101 To facilitate reform, insurance regulators could work with state attorneys general or other law enforcement experts to devise a list of risk-related features that underwriters should (or must) consider when setting rates. Insurers might combine a feature-rating approach with “sub-limits” capping the amount of coverage available for wrongful conviction claims, which alerts the insured municipalities to the gravity of the perceived risk.

b. Occurrence vs. claims-made coverage

Even feature-rated premiums, however, may not be accurate enough to encourage optimal loss prevention and ensure insurer stability. There is an additional, somewhat more drastic step insurers might take. I have assumed, so far, that police liability coverage is offered, as has been the general (though not exclusive) tradition, on an occurrence basis.102 “Occurrence policies cover liability for activities that take place during the policy period, regardless of when a suit that seeks to impose liability for these activities is filed.”103 Policyholders tend to like this feature, but it makes pricing very difficult, especially for long-tail risks. Pricing occurrence coverage requires insurers “to charge in the present for all the eventual results of present activities.”104 The principal alternative form of liability coverage is the claims-made policy. “[C]laims-made policies insure against liability for claims that are filed during the policy year” that arise out of activities occurring after a specified retroactive date.105 That is, a “claims-made policy provides coverage during the policy year for injuries caused by activities occurring in the past.”106 And “[b]ecause the insurer need not predict long-term claim exposure, claims-made policies can be priced more confidently than occurrence policies.”107 In fact, claims-made policies were first introduced when “concern over the difficulty of predicting the scope of long-tail liabilities became pronounced.”108

More accurate pricing, while generally beneficial, may not translate straightforwardly into better loss prevention. “Although the shift from occurrence to claims-made coverage solves many of the insurance industry’s prediction problems,” Kenneth Abraham explains, “it does little to remove the obstacles to thorough cost internalization. If anything, such a shift may be a step in the opposite direction.”109 This is because a “claims-made premium increase reflects only the additional costs anticipated this year as a result of past activities,” which “can send incomplete and imprecise messages to insureds.”110 Abraham continues: “In contrast, an increase in occurrence premiums is a message about the future costs of this year’s activities.”111 The fear is that “a claims-made pricing system may induce an enterprise to underestimate the cost of prospective liability as compared to the cost of an investment in loss prevention that would avoid some of that liability.”112

That is not to say, of course, that a claims-made insured is without any incentive for careful behavior—it is “always at risk that its coverage will not be renewed because of unsafe operations, and that it will be exposed thereafter to claims that have not yet been reported.”113 In any event, and despite the drawback just mentioned, the gains from more accurate pricing may justify a shift to claims-made coverage. In particular, insurers could segregate coverage for wrongful conviction claims and write that portion alone on a claims-made basis, continuing to write the rest of the police liability policy on an occurrence basis if the municipality prefers.114

2. Complementing insurance.

Even were insurers to adopt all of my proposals, I would remain skeptical about just how well they could regulate the risk of wrongful convictions. There is a need here for a more active government presence—a vessel into which we might funnel some of the reform efforts presently focused on the use of force.115 Fortunately, as I mentioned in passing above, we know a fair amount about a set of policies with promise to reduce the risk of wrongful convictions.116 There are numerous ways to encourage adoption of these policies. Old-fashioned political lobbying has more promise than one might think; states have gradually adopted meaningful reforms concerning interrogation, eyewitness lineups, and other critical stages of the criminal process.117 Reorienting some of our constitutional criminal procedure doctrine from individual to systemic issues—that is, making a criminal defendant’s constitutional claim turn in part on the systemic measures the prosecuting jurisdiction has taken to safeguard the underlying constitutional principle—is one way to support and foster these political reforms.118 Another might be Congress’ spending power. Congress could reduce federal law enforcement funding to jurisdictions that have not yet implemented best practices to prevent wrongful convictions;119 in theory, the moneys not disbursed could even pour into a fund to help compensate exonerees. Nor should we stop with the basic set of established reform proposals. Why not take a page from the financial-crisis literature and try to develop an early warning system to anticipate wrongful convictions and mitigate the damage from them?120 If insurers substantially improve their ability to reduce the risk of wrongful convictions, these public regulatory mechanisms may recede in importance. But until that time, the case for intensified government intervention is strong.

IV. Low-Dollar, Short-Tail

Low-dollar, short-tail claims are the bread and butter of constitutional criminal procedure litigation. That is, many constitutional violations in the criminal process cause some immediate harm, but not one the legal system deems compensable to any significant extent.121 This is because deprivation of a constitutional right, in itself, is compensable only by nominal damages.122 To exceed nominal damages, a plaintiff must demonstrate some separate, compensable harm, like a physical injury or loss of wages.123 Many plaintiffs cannot make this showing.

Consider some familiar examples. Even a brief investigatory stop, especially if accompanied by a frisk, may intrude significantly on privacy and dignity interests the Fourth Amendment is said to protect.124 Where the stop is not justified by reasonable suspicion, the Constitution forbids it.125 Yet no damages are due for an unjustified stop, without more. Similarly, a “bare” Miranda violation—the failure to give Miranda warnings before conducting a custodial interrogation—is not compensable at all.126

What this means is that insurers have little reason to fear paying out on claims stemming from such mine-run violations, and thus little incentive to expend resources to prevent them. Three additional facts bolster this conclusion. First, although attorney fees provisions of the federal civil rights statutes were enacted partly to overcome plaintiffs’ financial disincentives to suit,127 fees are likely to be paltry—or even waived in settlement—when a plaintiff recovers only nominal damages.128 Second, although, in theory, defense costs in suits for injunctive relief could be substantial enough to justify insurer loss-prevention efforts, standing doctrine makes it extremely difficult for plaintiffs to seek injunctions.129 Moreover, many police liability policies cover only damages claims.130 And third, some—maybe many—policies contain a deductible or self-insured retention that effectively allocates to the municipality losses for claims below a specified attachment point.131 An occasional Terry claim resulting in a modest damages payment may never even touch the insurer.

Insurers do not disregard these claims entirely. Some insurers—especially municipal risk pools, which are owned by their policy-holding member municipalities—see one of their roles as promoting police professionalism. Reducing the number of Terry or Miranda violations could support this goal.132 And where a plaintiff is able to make a lawsuit financially viable—such as by obtaining pro bono representation by an interested organization—defense costs might be substantial even where damages are not. Perhaps most important, “everyday” violations, even if not themselves compensable, may ground claims of more egregious conduct that are compensable. Although a brief (unlawful) Terry stop, for example, likely causes no compensable harm, a prolonged one, or one accompanied by abusive language or conduct, might. Insurers thus have some incentive to make sure their agencies are generally following Terry. Similarly, although a bare Miranda violation is non-compensable, it may help support a compensable claim of outrageous conduct that violates due process.133

Still, at the end of the day, insurers will be hard-pressed to regulate closely an aspect of police conduct that does not, itself, threaten substantial civil liability. This describes a good deal of plain vanilla unlawful behavior. And if insurers do not regulate, someone else has to. For the most part, we rely on criminal defendants to act as private attorneys general by raising the claims in a defensive posture, in motions to suppress evidence.134 According to one study, ninety-five percent of Terry claims are litigated in this fashion.135 We supplement this with a tiny bit of injunctive litigation136 and a handful of “pattern or practice” suits by the U.S. Department of Justice under 42 U.S.C. § 14141.137 These are useful tools but they are probably insufficient, as they seem to leave us with underdeterrence, judging by the number of legal violations we presently experience.138

As I have argued elsewhere, and alluded to above, in my view the biggest advance that can be made here, within the limits of practicality, is to reorient some of our constitutional doctrine to focus defendants’ claims on systemic rather than (or in addition to) individual issues.139 The courts in this model would not be first-order regulators, announcing conduct rules for cops on the street to follow, but second-order regulators, articulating constitutional principles for political policymakers to operationalize as they craft the conduct rules that govern the police. Suppression hearings would then focus not only on the isolated conduct of the individual officers alleged to have acted illegally, but also on the steps that policymakers took—both before and after the challenged conduct—to channel discretion and encourage law compliance. The basic theory is to address the systemic underdeterrence of police wrongdoing by having the courts pressure political actors to prevent and punish misconduct rather than try to catch it all themselves. I would not expect this systemic turn to eradicate the targeted misconduct, but I do think it could improve upon the status quo.

V. Low-Dollar, Long-Tail

As a doctrinal matter, the Constitution forbids racial profiling—targeting an individual for suspicion of crime because of his race.140 Payouts on profiling claims, however, while more than nominal, are typically insubstantial. In one recent case, for example, a Kashmiri man stopped at subway checkpoints twenty-one times by the New York City police settled for $10,001,141 a fraction of the deductible many insured municipalities carry.142 Other, similar examples are not hard to find.143 Although the media occasionally report high-dollar judgments and settlements in racial profiling cases, those cases always involve, as far as I can tell, some separate compensable harm from police violence or other aggravating conduct.144 As a result, despite substantial evidence of discrimination on the roadways and elsewhere,145 insurers have little incentive to expend resources combatting “mere” racial profiling.146

As before, this is not to say that insurers pay no attention to racial profiling. Discrimination can factor into broader, more costly occurrences like the racially motivated use of force. In an effort to manage this risk, some insurers have begun to consider the racial and ethnic diversity of the police force at underwriting or rating.147 Others encourage departments to train officers on confronting implicit racial bias.148 Still, racial profiling does not make the risk manager’s top-twelve list of “high risk/critical tasks.”149

One commonsense way to focus additional attention on racial profiling would be to lobby for loftier damages. Higher payouts would better reflect the dignitary harms racial profiling inflicts and create stronger financial incentives for municipalities—and their insurers—to beef up efforts to prevent profiling. This strategy would likely improve upon the status quo, yet substantial impediments to effective regulation-by-insurance would remain. One such impediment is that racial profiling claims frequently have longer tails than one might expect.

To be sure, an individual who is profiled by the police might personally suspect (or even know) as much immediately. (The same is true, of course, of a wrongfully convicted defendant who knows he is innocent.) But from a risk management perspective, the more important question is when the individual can prove that he has been profiled, i.e., when he is likely to sue. And it turns out that racial profiling claims frequently lie dormant for years. Why? In all but the rarest cases that involve direct evidence of discriminatory intent,150 a colorable racial profiling claim must, to trigger Fourteenth Amendment protections, marshal statistical evidence from a broad sample of other incidents.151 This can make it impossible to prove that racial profiling is happening right when it first occurs—only situating the challenged conduct within the context of future (or future-disclosed) police activities will reveal its discriminatory nature. To reflect this reality, at least some courts have applied the “discovery rule” and held that a racial profiling cause of action does not accrue until the injured party discovers, or reasonably should have discovered, the basis for an actionable claim. The Third Circuit, for example, delayed accrual in one case for eleven years, until the State of New Jersey released documents revealing a statewide practice of racially selective law enforcement, “information vital to [the plaintiff ’s] selective-enforcement claim.”152

More generally, there is an incentive for racial profiling claimants to delay pursuing their claims; many cases will gain strength as time passes and more data roll in. The incentive may well be the opposite in the use-of-force context: sue quickly while witnesses’ memories are fresh. In addition, both plaintiffs and courts routinely rely on data from outside the limitations period to inform the legality of more recent activity, which means that expiration of the statute of limitations does not signal the same degree of repose for insurers that it might in, say, the use-of-force setting.153 To put the point slightly differently, the factual and evidentiary basis for a racial profiling lawsuit may ripen for years without alerting insurers to the need to collect premiums and build reserves to cover any eventual payout.

To give one example, the plaintiffs in a major racial profiling lawsuit in California—a class of individuals whom the police had stopped beginning in 1998—alleged that the California Highway Patrol (CHP) had “long relied upon race and ethnicity in conducting stops, detentions, interrogations and searches of motorists” as part of an “unabated, continuing pattern and practice of discrimination” that had intensified in recent years.154 The challenged conduct dated back to at least the late 1980s, when the CHP became involved in “Operation Pipeline,” a federally funded drug interdiction program the plaintiffs described as a “roving program of discrimination.”155 Yet the plaintiffs did not sue until 1999, the year California’s Joint Legislative Task Force on Government Oversight released a report, based on a review of thousands of CHP records, finding that Operation Pipeline discriminated against motorists of color.156

Two implications follow. First, even if racial profiling triggered heavier damages, insurers would face substantial challenges in pricing and managing the risk. This is true for the reasons discussed in Part III—namely, that the long tail of liability heightens the degree of correlation among the covered risks, tempts insurers to externalize to the future, and creates uncertainty about the number and magnitude of claims insurers should anticipate.157 Consider, for example, the effects of a statute easing the burden of proof for racial profiling plaintiffs, which would buoy a large number of outstanding claims simultaneously, upending insurers’ financial planning.

Second, criminal defendants are poorly situated to serve as private attorneys general, at least relative to how well they can play that role for claims with shorter tails. In many cases, present proof of discriminatory intent—through statistical analysis of prior incidents—will simply be unavailable when the defendant is charged, even though proof may bubble up in patterns of future police activity. But even where present proof is theoretically available—and where defense resources exist to analyze and present it—pretrial detention creates strong incentives for defendants to resolve their cases as quickly as possible, sooner than will allow for the development of a relatively complex racial profiling claim.158 Stingy discovery standards further hamstring criminal defendants’ efforts to prove profiling.159 Indeed, I could find no criminal case in which a defendant has prevailed on a racial profiling defense under federal law.160

It appears, therefore, that neither civil plaintiffs—with liability filtered through insurance—nor criminal defendants, acting as private attorneys general, show much promise in the fight against racial profiling. A relatively small amount of injunctive-relief litigation has helped call attention to the problem, and may reduce the frequency of profiling in the targeted jurisdictions, but it seems unlikely to be a general solution.161 In light of these realities, it would be wise to shift our attention from the judicial forum to political ones. This is not a novel suggestion; the work is already underway. According to one report, thirty states have enacted some form of ban on racial profiling, and seventeen also forbid pretextual traffic stops. Eighteen require mandatory data collection for all stops and searches.162 The most recent efforts pin hope on cutting-edge reforms like implicit bias training (“debiasing”) and body-worn cameras.163 Increasing the diversity of police forces may help too.164 At the national level, federal policy prohibits U.S. agents from profiling, although the policy has some loopholes, and efforts to enact broader, federal legislation have faltered.165 A recent report by the President’s Task Force on 21st Century Policing urges local agencies to adopt their own policies banning profiling.166 There is a supportive role for doctrine to play as well: The Court’s current, hands-off approach to stamping out discriminatory motive could become a sort of doctrinal carrot—a “safe harbor” available only to jurisdictions that have implemented adequate safeguards to prevent discrimination.167

There is one major sticking point I have obscured, however: We have little idea which, if any, of these leading reform proposals actually work.168 While mustering political will to enact reforms is an important step toward change, there will be no change if the reforms enacted turn out to be ineffective. This field is ripe with opportunity for researchers who can figure out how to measure the effects of our “leading solutions” to the problem of racial profiling.

 

VI. Conclusion

To tame police misconduct, we must first understand the nature of the beast. Careful attention to the incentives of, and constraints on, some of the major players in policing reveals not one but many species of misconduct. These players include the constitutional rights-holders—i.e., the victims of misconduct, on whom we rely to serve as private attorneys general—as well as the intermediary institutions, like insurers, that help operationalize the deterrent ambitions of our civil liability regime. Some misconduct inflicts harms the legal system compensates meaningfully; some inflicts harms that, even if normatively serious, the system leaves largely unremedied. Some legal injuries manifest immediately; others manifest only after significant delay. These distinctions make it unlikely that any one solution, or any single remedial regime, will work best to reduce police misconduct across the board. Given these truths, police reformers ought to start thinking like foxes rather than hedgehogs.169

 

  • 1“Municipality” and “police,” as used here, include both city- and county-level entities and law enforcement officers.
  • 2I describe the phenomenon of police “regulation-by-insurance” in detail in other work. See John Rappaport, How Private Insurers Regulate Public Police, 130 Harv. L. Rev. (forthcoming 2017), http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2733783 (https://perma.cc/2EC3-SBHE). For two of the very few other legal academic papers to touch on insurance, see Joanna C. Schwartz, How Governments Pay: Lawsuits, Budgets, and Police Reform, 63 UCLA L. Rev. 1144 (2016) [hereinafter Schwartz, How Governments Pay]; Joanna C. Schwartz, Who Can Police the Police?, 2016 U. Chi. Legal F. 437; see also Charles R. Epp, Making Rights Real 115–37 (2009) (studying insurance among numerous factors that might influence “legalized accountability” by police agencies); Candace McCoy, How Civil Rights Lawsuits Improve American Policing, in Holding Police Accountable 111, 112 (Candace McCoy ed., 2010) (asserting that “insurance companies [have] demanded that police improve their policies and practices in adherence to constitutional requirements and thus avoid monetary payouts to injured citizens”).
  • 3For a different typology that also sorts civil rights claims according to their dollar value, see Paul D. Reingold, Requiem for Section 1983, 3 Duke J. Const. L. & Pub. Pol’y 1, 13 (2008) (organizing claims by dollar value and substantive strength).
  • 4“Long-tail claims are those which are not brought (or if brought, not resolved) for some years after the action by the defendant that gives rise to the claim.” Kenneth S. Abraham, Environmental Liability and the Limits of Insurance, 88 Colum. L. Rev. 942, 964 n.69 (1988). In practice, legal claims lie on a spectrum, ordered by the length of their tail; the trait is continuous, not binary.
  • 5In my initial research for Rappaport, supra note 2, I conducted thirty-three semistructured telephone interviews, mostly with members of the industry who were high-ranking officials within their respective firms. I located my subjects using a “snowball sampling” technique. See, e.g., John Lofland et al., Analyzing Social Settings: A Guide to Qualitative Observation and Analysis 43 (4th ed. 2006) (discussing “snowball” or “chain-referral” sampling: “a method for generating a field sample of individuals possessing the characteristics of interest by asking initial contacts if they could name a few individuals with similar characteristics who might agree to be interviewed”). My interview subjects were geographically diverse, including representatives of firms in every time zone and consultants who travel the country. I followed up on some of these interviews with targeted questions via email or a quick additional telephone call, which I did not count toward the total thirty-three. I also attended the 2015 annual conferences of the Association of Governmental Risk Pools and the International Association of Chiefs of Police. For this Article—in particular for the research in Part III—I requested additional telephone interviews with the relevant experts who had been most helpful during the first round of interviews; I ended up speaking with eight of them. I ceased interviewing new subjects when responses became repetitious.
  • 6See sources cited infra note 76.
  • 7See Terry v. Ohio, 392 U.S. 1 (1968); Miranda v. Arizona, 384 U.S. 436 (1966).
  • 8See John Rappaport, Second-Order Regulation of Law Enforcement, 103 Calif. L. Rev. 205 (2015).
  • 9A risk pool is a nonprofit, mission-driven organization formed by a group of local government entities, usually within one state, to finance a risk, typically by pooling or sharing that risk. The entities themselves ordinarily own and govern the pool. See Rappaport, supra note 2 (manuscript at 21–22); see generally Jason E. Doucette, Note, Wading in the Pool: Interlocal Cooperation in Municipal Insurance and the State Regulation of Public Entity Risk Sharing Pools—A Survey, 8 Conn. Ins. L.J. 533, 541–42 (2002).
  • 10See, e.g., Carol A. Archbold, Police Accountability, Risk Management, and Legal Advising 25 (2004) (concluding, based on survey results, that “risk management programs are still in the infancy stage of being embraced by police agencies”); Joanna C. Schwartz, Introspection Through Litigation, 90 Notre Dame L. Rev. 1055, 1095–1101 (2015) (reviewing evidence that few police departments have risk managers); Joanna C. Schwartz, Myths and Mechanics of Deterrence: The Role of Lawsuits in Law Enforcement Decisionmaking, 57 UCLA L. Rev. 1023 (2010) (finding that the largest police agencies only rarely learn from lawsuits filed against them or their officers).
  • 11See, e.g., Graham v. Connor, 490 U.S. 386 (1989); Tennessee v. Garner, 471 U.S. 1 (1985).
  • 12See, e.g., Monica Davey, Chicago Pays $5 Million over Killing of Teenager, N.Y. Times, Apr. 16, 2015, at A15 (reporting $5 million settlement in death of Laquan McDonald); Richard Fausset, Settlement Reached in Shooting by Officer, N.Y. Times, Oct. 9, 2015, at A24 (reporting $6.5 million settlement in death of Walter Scott); Sheryl Gay Stolberg, University of Cincinnati to Pay $5 Million to Family in Killing by Police, N.Y. Times, Jan. 19, 2016, at A16 (reporting $4.85 million payout in death of Samuel DuBose, which “appears in line with other recent settlements of cases involving police officers”); Sheryl Gay Stolberg, Baltimore Announces $6.4 Million Settlement in the Death of Freddie Gray, N.Y. Times, Sept. 9, 2015, at A20 (reporting $6.4 million settlement in death of Freddie Gray, which “[l]egal specialists said . . . was in line with settlements for recent racially charged police misconduct cases,” including Eric Garner, whose estate settled for $5.9 million).
  • 13See Nancy Leong, Making Rights, 92 B.U. L. Rev. 405, 425 (2012) (finding that “98% of excessive force claims are litigated in the civil context”); see also id. at 441 (asserting that criminal “courts routinely ignore the use of force in analyzing the legality of an investigative stop”).
  • 14See, e.g., Nat’l Research Council of the Nat’l Acads., Fairness and Effectiveness in Policing 278–80 (Wesley Skogan & Kathleen Frydl eds., 2004); Oren Bar-Gill & Barry Friedman, Taking Warrants Seriously, 106 Nw. U. L. Rev. 1609, 1626–34 (2012); Richard Emery & Ilann Margalit Maazel, Why Civil Rights Lawsuits Do Not Deter Police Misconduct: The Conundrum of Indemnification and a Proposed Solution, 28 Fordham Urb. L.J. 587 (2000); Diana Hassel, Living a Lie: The Cost of Qualified Immunity, 64 Mo. L. Rev. 123, 145 n.106 (1999); Daryl J. Levinson, Making Government Pay: Markets, Politics, and the Allocation of Constitutional Costs, 67 U. Chi. L. Rev. 345 (2000); Daniel J. Meltzer, Deterring Constitutional Violations by Law Enforcement Officials: Plaintiffs and Defendants as Private Attorneys General, 88 Colum. L. Rev. 247, 283–86 (1988); Joanna C. Schwartz, Police Indemnification, 89 N.Y.U. L. Rev. 885 (2014). But see, e.g., Myriam E. Gilles, In Defense of Making Government Pay: The Deterrent Effect of Constitutional Tort Remedies, 35 Ga. L. Rev. 845 (2001).
  • 15Leong, supra note 13, at 446; see also Rachel Harmon, When Is Police Violence Justified?, 102 Nw. U. L. Rev. 1119, 1127 (2008) (calling the doctrine “indeterminate” as well as “unprincipled”).
  • 16Leong, supra note 13, at 447.
  • 17See, e.g., Law Enforcement Liability Coverage Form, Nat’l Cas. Co. 5–6, http://euclidps.com/2/wp-content /uploads/2014/05/Law_Enforcement_II_PE_PL_2_0802.pdf (https://perma.cc/E538-LB3C]; see also Kenneth S. Abraham, Four Conceptions of Insurance, 161 U. Penn. L. Rev. 653, 656 (2013) (stating that “virtually all property-casualty insurance policies . . . are standard-forms used by most insurers”).
  • 18See Rappaport, supra note 2 (manuscript at 16–17); see also Omri Ben-Shahar & Kyle D. Logue, Outsourcing Regulation: How Insurance Reduces Moral Hazard, 111 Mich. L. Rev. 197, 203–05 (2012).
  • 19G. Patrick Gallagher, Successful Police Risk Management 53 (2014); see also Robert J. Girod, Police Liability and Risk Management 163 (2014) (placing “[u]se of force and deadly force” atop a list of the “most common ‘actionable conduct’ involving civil rights liability” for police).
  • 20See, e.g., Police Professional Liability Insurance Application, Prof ’l Gov’tal Underwriters, Inc. (June 2014), http://www.pgui.com/App_Content /media/2014%20apps/DRWN%20PPL%201010%2006_14%20PGU%20Police%20New%20Bus%20Application.pdf (https://perma.cc/XLA2-36Z3).
  • 21See, e.g., Trident Risk Points: Operational Policies & Procedures: Use of Force Policy Elements, Trident Ins. Servs. (July 2012), https://www.argolimited.com/media/03C10U7X865H/docs/en_US/9a5e44de9e07465726bcb14894240b67ba9c4565/5PLYTFCP9D5H/Trident-LEL-Use-of-Force-Policy-Elements-2012.pdf (https://perma.cc/2VCJ-QJAW]; see also Brandon L. Garrett, Innocence, Harmless Error, and Federal Wrongful Conviction Law, 2005 Wis. L. Rev. 35, 107 (explaining that municipal liability “can be premised on a police department’s deviation from national police practices”).
  • 22See Focus On: Police—Excessive Use of Force, Munich Re (May 2015), https://www.munichre.com/site/mram-mobile/get /documents_E876514504/mram/assetpool.mr_america/PDFs/3_Publications/Research_Spotlight /FOCUS%20ON_Excess-Force.pdf (https://perma.cc/4LW9-T4S6]; Trident Risk Points: Operational Policies & Procedures: Suggested Controls for Electronic Stun Weapons, Trident Ins. Servs. (July 2012), https://www.argolimited.com/media/03C10U7X865H/docs/en_US/dad1345ec51a64376baf01f33f257328aa66bfd5/G5V9I05T5956/Trident-LEL-Electronic-Stun-Weapon-Policy-2012.pdf (https://perma.cc/SA9W-DU3M]; Trident Ins. Servs., supra note 21; see also The Use-of-Force Continuum, Nat’l Inst. of Justice, Office of Justice Programs (Aug. 4, 2009), http://www.nij.gov/topics/law-enforcement /officer-safety/use-of-force/Pages/continuum.aspx [perma.cc/L2M3-S8LK). A recent report by the Police Executive Research Forum challenges the continuum’s status as a “best practice,” arguing that the continuum can encourage officers to escalate encounters and that “there are more effective ways to respond to many threats than through a use-of-force continuum.” Police Exec. Research Forum, Guiding Principles on Use of Force 19–20 (2016).
  • 23See, e.g., Stephen A. Bishopp et al., An Examination of the Effect of a Policy Change on Police Use of TASERs, 26 Crim. J. Pol’y Rev. 727 (2015) (electronic stun weapons); James J. Fyfe, Police Use of Deadly Force: Research and Reform, 5 Just. Q. 165 (1988) (firearms).
  • 24See, e.g., Munich Re, supra note 22; Law Enforcement Training Videos, Ind. Mun. Ins. Program, http://www.indianamip.com/law_enforcement.html (https://perma.cc/5JKX-XUEG] (last visited Aug. 31, 2015) (listing dozens of training videos covering topics such as “Straight Baton Techniques,” “Line Officer Tactical Shotgun,” and “Basic and Power Handcuffing Techniques”); Telephone Interview with Consultant A (Aug. 16, 2014) (describing having conducted, on behalf of insurers, “hands-on” training on use of force, transportation of prisoners, and other related topics).
  • 25Risk Management, Liability Insurance, and CALEA Accreditation, CALEA, http://www.calea.org /content /risk-management-liability-insurance-and-calea-accreditation (https://perma.cc/UW7A-3VRU] (last visited Aug. 31, 2015) (maintaining list of “liability insurance providers known to CALEA to offer some type of financial incentive to CALEA accredited agencies”); Ileana Garcia, Slidell Police Accreditation Keeps the Department’s Insurance Rate Low, Slidell Sentry-News, reprinted in CALEA Update Mag., Feb. 2001, http://www.calea.org /calea-update-magazine/issue-75/accreditation-works/slidell-police-accreditation-keeps-departments-in (https://perma.cc/H43T-YXMS).
  • 26See Standards Titles, CALEA, http://www.calea.org /content /standards-titles [perma.cc/D2VM-9ZT8] (last visited Nov. 30, 2015) (listing standards); Law Enforcement Program: The Standards, CALEA, http://www.calea.org /content /law-enforcement-program-standards (http://perma.cc/JR8C-3JZM] (last visited Nov. 30, 2015) (requiring compliance with standards); Law Enforcement Program: Process, CALEA, http://www.calea.org /content /law-enforcement-program-process (http://perma.cc/QJ84-ARP5] (last visited Nov. 30, 2015) (requiring continued compliance and reaccreditation).
  • 27E.g., Telephone Interview with Risk Pool D (July 6, 2015).
  • 28See, e.g., Daniel Cruse & Jesse Rubin, Determinants of Police Behavior 5 (1973) (reporting, based on field study, that “the amount of stress seems to have a good deal of effect on the behavior of the officer”); Gail A. Goolkasian et al., Coping With Police Stress 10 (1986) (reporting findings that stress can negatively affect work performance, though noting studies’ limitations); Ronald J. Burke & Aslaug Mikkelsen, Burnout, Job Stress and Attitudes Towards the Use of Force by Norwegian Police Officers, 28 Policing Int’l J. Police Strats. & Mgmt. 269, 269–72 (2005) (summarizing studies finding that chronic work stress causes burnout, which is positively and significantly related to the use of force); Nicolien Kop & Martin C. Euwema, Occupational Stress and the Use of Force by Dutch Police Officers, 28 Crim. Just. & Behav. 631 (2001) (similar); Manny Fernandez, Officer Was Under Stress When He Arrived at Texas Pool Party, Lawyer Says, N.Y. Times, June 11, 2015, at A15 (describing lawyer’s assertion that McKinney, Texas, police officer who was videotaped tackling a black teenager in a bikini outside a pool party was under stress after responding to two earlier calls involving a suicide and attempted suicide); Mark Bond, The Impact of Stress and Fatigue on Law Enforcement Officers and Steps to Control It, In Public Safety (Feb. 24, 2014), http://inpublicsafety.com/2014/02/the-impact-of-stress-and-fatigue-on-law-enforcement-officers-and-steps-to-control-it (https://perma.cc/L8ZS-VBLN] (asserting that officer stress can lead to fatigue, which in turn can lead to misconduct and “inappropriate reactions to a situation”).
  • 29See, e.g., Trident Risk Points: Operational Policies & Procedures: Law Enforcement Employment Hiring Policy, Trident Ins. Servs. (Sept. 2012), https://www.argolimited.com/media/03C10U7X865H/docs/en_US/74ab995451582b436242abc67f5252a8f9e34329/795717IY6X5I/Trident-Employment-Hiring-Program-2012.pdf (https://perma.cc/2P9J-PQAM). The research on the benefits of psychological screening of officer candidates is mixed. See, e.g., Michael G. Aamodt, Predicting Law Enforcement Officer Performance with Personality Inventories, in Personality Assessment in Police Psychology 229, 238–40 (Peter A. Weiss ed., 2010) (reporting, based on meta-analysis of 200 studies, that some aspects of certain personality tests do a reasonable job of predicting officer performance, including disciplinary problems, but many do not); Robert E. Cochrane et al., Psychological Testing and the Selection of Police Officers: A National Survey, 30 Crim. Just. & Behav. 511, 515 (2003) (collecting sources that have “demonstrated the usefulness of personality measures in predicting job performance” of police officers); Candice Bernd, Evaluating Police Psychology: Who Passes the Test?, Truthout (Feb. 20, 2015, 9:25 AM), http://www.truth-out.org /news/item/29191-evaluating-police-psychology-who-passes-the-test (http://perma.cc/YXM7-TFD6] (asserting that, “[w]hen departments forgo psychological screenings, the result is often violence,” but also highlighting the lack of standardization and sophistication of screening exams).
  • 30Ken Wallentine, Vice President & Senior Legal Advisor, Lexipol, Body Worn Cameras: Risk and Reward (Oct. 6, 2015). Wallentine cited studies from Rialto, California, and Denver, Colorado. See Barak Ariel et al., The Effect of Police Body-Worn Cameras on Use of Force and Citizens’ Complaints Against the Police: A Randomized Controlled Trial, 31 J. Quant. Criminology 509 (2015); Cole Zercoe, Body Camera Study: Denver Police See Drop in Arrests, UOF Complaints, PoliceOne (Sept. 4, 2015), http://www.policeone.com/police-products/body-cameras/articles/9485301-Body-camera-study-Denver-police-see-drop-in-arrests-UOF-complaints/ (https://perma.cc /ZD3X-QETH]; see also Munich Re, supra note 22 (“[P]olice departments are encouraged to purchase body cameras, provide training and set standards for their use.”); The Effect of Body Worn Cameras & Police Use of Force, Trident Ins. Servs. (Feb. 2015), https://www.argolimited.com/media/03C10U7X865H/docs/en_US/d94acf91189467168892daca4ba90de76bc8a263/D6L33YO488Y0/Your_Risk_Control_Resource_-_Feb15.pdf (https://perma.cc/33HQ-LVYG] (“[B]ody-worn cameras appear to be a potentially valuable tool in reducing police use of force and related incidents.”).
  • 31Munich Re, supra note 22.
  • 32See, e.g., Law Enforcement Training Solutions, Meggitt Training Sys., http://meggitttrainingsystems.com/‌Law-Enforcement (https://perma.cc/2M6B-4F76] (last visited Aug. 31, 2015); VirTra, http://www.virtra.com/ (https://perma.cc/3DKA-Q4ZK] (last visited Aug. 31, 2015); Milo Range, http://www.milorange.com (https://perma.cc/UQ35-KHNN] (last visited Aug. 31, 2015).
  • 33Craig Bennell & Natalie J. Jones, Can. Police Research Ctr., The Effectiveness of Use of Force Simulation Training 8 (2005), http://www.publicsafety.gc.ca/lbrr/archives/cnmcs-plcng /cn000032136920-eng.pdf (https://perma.cc/P35H-NGY6).
  • 34Id.
  • 35Id. at 6; see also Jennifer Armstrong et al., Monitoring the Impact of Scenario-Based Use-of-Force Simulations on Police Heart Rate: Evaluating the Royal Canadian Mounted Police Skills Refresher Program, 15 W. Criminology Rev. 51 (2014); Evelyn-Rose Saus et al., The Effect of Brief Situational Awareness Training in a Police Shooting Simulator: An Experimental Study, 18 Military Psych. S3 (2006).
  • 36William Micklus, Assoc. Dir., Upper Midwest Cmty. Policing Inst. & Jeffrey Range, Senior Practitioner, Resologics, Dispute Resolution: Skills Training for Law Enforcement (Oct. 7, 2015).
  • 37See, e.g., Training Calendar, Wash. Cities Ins. Auth., http://www.wciapool.org /education-training /calendar (https://perma.cc/6QWJ-F4NE] (last visited Nov. 30, 2015); Featured Training Notice, Cities & Vills. Mut. Ins. Co. (Mar. 26, 2015), http://www.cvmic.com/featured-training-notice/ (https://perma.cc/DQ6E-NEJP).
  • 38Munich Re, supra note 22.
  • 39See, e.g., Barbara E. Armacost, Organizational Culture and Police Misconduct, 72 Geo. Wash. L. Rev. 453 (2004) (discussing the separation between formal departmental policies and informal de facto norms); Samuel Walker, The New Paradigm of Police Accountability: The U.S. Justice Department “Pattern or Practice” Suits in Context, 22 St. Louis U. Pub. L. Rev. 3, 45 (2003) (disaggregating “primary compliance,” which “involves the development of a formal policy on a particular aspect of police operations”; “secondary compliance,” which “involves evidence that the department has incorporated the policy into training and supervision”; and “operational compliance,” which “involves evidence that officers comply with the policy in their routine activities”).
  • 40See supra note 24.
  • 41Telephone Interview with Commercial Insurer A (July 20, 2015) (every three years); Telephone Interview with Risk Pool A (Sept. 2, 2014) (annual); Telephone Interview with Risk Pool B (Sept. 2, 2014) (semi-annual). Some insurers require online updates from agencies in between audits or encourage self-audits in addition. Telephone Interview with Risk Pool B, supra (requires periodic online updates); Trident Risk Points: Operational Policies & Procedures: Monitoring Compliance with Law Enforcement Manual, Trident Ins. Servs. (July 2012), https://www.argolimited.com/media/03C10U7X865H/docs/en_US/1ff185e22af3a5c38781aa76ee68fcccad4ee5e6/IQJ40V3Z1A83/Trident-LEL-Compliance_with-Manual-2012.pdf (https://perma.cc/G5FS-XPR8] (providing guidelines for self-audits).
  • 42Telephone Interview with Risk Pool A, supra note 41; Telephone Interview with Consultant B (Aug. 27, 2014).
  • 43Telephone Interview with Risk Pool A, supra note 41; Telephone Interview with Risk Pool C (June 29, 2015).
  • 44Telephone Interview with Commercial Insurer A, supra note 41; Telephone Interview with Commercial Insurer D (Oct. 13, 2015); Telephone Interview with Risk Pool A, supra note 41; Telephone Interview with Risk Pool D, supra note 27; Telephone Interview with Consultant A, supra note 24; Telephone Interview with Consultant C (Aug. 20, 2014).
  • 45See, e.g., Samuel Walker & Carol A. Archbold, The New World of Police Accountability 137–77 (2d ed. 2014); Armacost, supra note 39, at 459–60; Samuel Walker et al., Responding to the Problem Officer: A National Study of Early Warning Systems 2.4–2.6 (2000), https://www.ncjrs.gov/pdffiles1/nij/grants/184510.pdf (https://perma.cc/EXE6-X4TN).
  • 46For far more detail on this topic, see Rappaport, supra note 2 (manuscript at 50–55, 57–59).
  • 47See, e.g., Alex Green, Niota Officials Tied to Beating Fired; They Say Insurance Company Forced the Action, Times Free Press (Chattanooga), Aug. 24, 2013 (quoting mayor’s report that city’s coverage would have been dropped if two officers involved in misconduct had been allowed back on duty); Rutledge Mayor “Had No Choice” in Firing: Police Chief Refused to Resign; City at Risk of Losing Insurance, Knoxville News-Sentinel, Mar. 23, 2010 (reporting mayor’s assertion he “‘had no choice’” but to fire a police chief accused of misconduct because “the city was at risk of losing its liability insurance” if the chief remained (quoting Mayor Danny Turley)); see also Rob Karwath, Calumet City Will Lose Police Liability Insurance, Chi. Trib., Mar. 29, 1988 (reporting council member’s comment that city’s insurance cancellation was “the final argument for the mayor to pick a new police chief from outside the department” when the interim chief retired).
  • 48See, e.g., Schwartz, How Governments Pay, supra note 2, at 1190–91 & nn.165–71 (collecting four examples of police departments that closed due to premium increases or termination of coverage); see also George J. Church, Sorry, Your Policy Is Canceled, Time, Mar. 24, 1986, at 16, 17, 18 (reporting that police patrols were suspended in two towns and five counties closed their jails due to lack of coverage); Tyler Jett, City of Niota, Tenn., Shutting Down. Again., Times Free Press (Chattanooga), June 19, 2013 (reporting that the city’s “police department is closed” after its insurer pulled coverage); cf. Liability Insurance in Crisis, N.Y. Times, Mar. 4, 1986, at A26 (reporting that “police in West Orange, N.J., had to stop patrolling in cars they could no longer insure”).
  • 49Telephone Interview with Commercial Insurer B (July 24, 2015); see also Telephone Interview with Commercial Broker A (July 22, 2015) (agreeing that underwriters have become more concerned with police liability since Ferguson); Roberto Ceniceros, Scandals Can Influence Police Liability Coverage, Bus. Ins. (June 5, 2000), http://www.businessinsurance.com/article/20000604/ISSUE01/10002637/scandals-can-influence-police-liability-coverage (https://perma.cc/26HY-XAF5] (discussing effect of police scandals on rates and coverage nationwide); Zusha Elinson & Dan Frosch, Cost of Police-Misconduct Cases Soars in Big U.S. Cities; Data Show Rising Payouts for Police-Misconduct Settlements and Court Judgments, Wall St. J., July 15, 2015 (“[I]nsurers and lawyers who defend police say current scrutiny of law enforcement is broadly affecting the resolution of lawsuits.”).
  • 50Law Enforcement Overview, Nat’l Conference of State Legislatures (May 29, 2015), http://www.ncsl.org /research/civil-and-criminal-justice/law-enforcement.aspx (https://perma.cc/EP3X-EKD5).
  • 51The prospect of criminal punishment may discourage some egregious misconduct, but I am inclined to agree with Paul Chevigny that “[c]riminal law is . . . not a system of ‘discipline’ for police misconduct; it defines the outer limits of what is permissible in society” and is at best a “patchy deterrent.” See Paul Chevigny, Edge of the Knife 98–101 (1995).
  • 52See generally Ben-Shahar & Logue, supra note 18.
  • 53See, e.g., Telephone Interview with Risk Pool D, supra note 27.
  • 54Cf. Kyle D. Logue, Encouraging Insurers to Regulate: The Role (If Any) for Tort Law, 5 U.C. Irvine L. Rev. 1355 (2015).
  • 55“Generally” is a meaningful modifier here. See Garrett, supra note 21, at 54; see generally Michael Avery, Obstacles to Litigating Civil Claims for Wrongful Conviction: An Overview, 18 B.U. Pub. Int. L.J. 439 (2009).
  • 56See Innocence Project, Making Up for Lost Time: What the Wrongfully Convicted Endure and How To Provide Fair Compensation 4, http://www.innocenceproject.org /files/imported/innocence_project_compensation_report-6.pdf (https://perma.cc/CS7U-MJMW).
  • 57Garrett, supra note 21, at 43 n.30, 44 n.32 (collecting cases); see also Mark Iris, Your Tax Dollars at Work! Chicago Police Lawsuit Payments: How Much, and for What?, 2 Va. J. Crim. L. 25, 44–45 (2014) (reporting average payouts of $2.4 and 3.2 million per case for Chicago police defendants in wrongful conviction cases between 2006 and 2012 in federal and state courts, respectively). Compensation is all over the map, however, and in some cases is grievously lacking. Compare Martin G. Hacala, Insights: Wrongful Convictions: What Governmental Risk Pools, and the Public Entities They Insure, Need to Know, Genesis, at 5 (May 2012), https://www.genesisinsurance.com/assets/pdfs/In%20the%20News/Insights20125-2.pdf (https://perma.cc/P9KB-GNT4] (“[D]amages awarded in wrongful conviction claims vary significantly. Indeed, it isn’t unheard of for the wrongfully convicted to recover nothing or a trivial amount.”), with id. at 6 (collecting cases sustaining damages in the millions).
  • 58Hacala, supra note 57, at 3. A prisoner may not sue for wrongful conviction until his conviction is reversed, expunged, or otherwise invalidated. See Heck v. Humphrey, 512 U.S. 477 (1994).
  • 59See Garrett, supra note 21, at 42 (“[I]n a surprisingly large number of cases, wrongful convictions were caused by police misconduct.”); see generally Russell Covey, Police Misconduct as a Cause of Wrongful Convictions, 90 Wash. U. L. Rev. 1133 (2013). According to a 1999 study, the figure is fifty percent. Jim Dwyer et al., Actual Innocence 246 (2003). Other common causes of wrongful convictions include prosecutorial misconduct and deficient defense counsel. See, e.g., James R. Acker & Allison D. Redlich, Wrongful Conviction: Law, Science, and Policy (2011); Jon B. Gould et al., Predicting Erroneous Convictions: A Social Science Approach to Miscarriages of Justice (2012).
  • 60Brandon L. Garrett, Judging Innocence, 108 Colum. L. Rev. 55, 75–91 (2008); see generally Brandon L. Garrett, Convicting the Innocent (2011).
  • 61See, e.g., Nat’l Cas. Co., supra note 17, at 5–6 (covering “[f]alse arrest, detention or imprisonment, or malicious prosecution” and “[v]iolation of civil rights protected under 42 USC 1981 et sequential or State Law”); Telephone Interview No. 2 with Risk Pool A (Oct. 5, 2015); Telephone Interview No. 2 with Risk Pool B (Oct. 5, 2015).
  • 62See Hacala, supra note 57, at 2 (“[E]xoneration . . . presents a significant legal and financial challenge for public entities and the [entities] that insure them.”).
  • 63Neil A. Doherty & Georges Dionne, Insurance with Undiversifiable Risk: Contract Structure and Organizational Form of Insurance Firms, 6 J. Risk & Uncertainty 187, 188 (1993).
  • 64Id. at 198.
  • 65Id. For an excellent clarifying discussion that classifies types of “liability developments risk,” see Tom Baker, Insuring Liability Risks, 29 Geneva Papers on Risk & Ins. 128 (2004).
  • 66See Ceniceros, supra note 49.
  • 67See Howard C. Kunreuther et al., Insurance and Behavioral Economics 83–85 (2013); S. Hun Seog, The Economics of Risk and Insurance 47 (2010).
  • 68See Ben-Shahar & Logue, supra note 18, at 215.
  • 69Martin T. Katzman, Pollution Liability Insurance and Catastrophic Environmental Risk, 55 J. Risk & Ins. 75, 83 (1988).
  • 70Kenneth S. Abraham, Distributing Risk 48 (1986); cf. Ben-Shahar & Logue, supra note 18, at 230 (“Some of the risks that insurers regulate materialize into harms far into the future, which means that insurers’ efforts to reduce such risks will largely benefit future insurers.”).
  • 71Ben-Shahar & Logue, supra note 18, at 230.
  • 72Richard A. Posner, Catastrophe 171–72 (2004).
  • 73Id. at 172.
  • 74Nate Silver, The Signal and the Noise 29 (2012). The distinction traces back to Frank H. Knight, Risk, Uncertainty, and Profit 197–232 (1921); see also Cass R. Sunstein, Worst-Case Scenarios 147 (2007) (describing uncertainty as a situation “where outcomes can be identified but no probabilities can be assigned” and risk as “where outcomes can be identified and probabilities assigned to various outcomes”); id. at 159–62 (defending the distinction). But see Milton Friedman, Price Theory 282 (1976) (challenging the distinction’s validity).
  • 75Abraham, supra note 70, at 47.
  • 76Samuel R. Gross & Michael Shaffer, Nat’l Registry of Exonerations, Exonerations in the United States, 1989–2012, at 3 (2012), http://www.law.umich.edu/special/exoneration/Documents/exonerations_us_1989_2012_full_report.pdf (https://perma.cc/97NR-SVK7] (“[E]ven 2,000 exonerations over 23 years is a tiny number in a country with 2.3 million people in prisons and jails.”). The rate of exonerations does seem to be increasing, however. See, e.g., Nat’l Registry of Exonerations, The First 1,600 Exonerations 5 (2015), http://www.law.umich.edu/special/exoneration/Documents/1600_Exonerations.pdf (https://perma.cc/AT36-S4WG]; Sam R. Gross et al., Exonerations in the United States: 1989 through 2003, 95 J. Crim. L. & Criminology 523, 527 (2005). Perhaps this is not surprising, given that the errors that lead to wrongful convictions are typically systemic, not idiosyncratic. See, e.g., Garrett, supra note 60, at 274; Jennifer E. Laurin, Still Convicting the Innocent, 90 Tex. L. Rev. 1473, 1475 (2012) (book review).
  • 77Sunstein, supra note 74, at 21–24. There is a substantial related literature finding that individuals purchase too little insurance against low-probability events. See, e.g., Mark J. Browne, Behavioral Bias and the Demand for Bicycle and Flood Insurance, 50 J. Risk & Uncertainty 141 (2015); Philip T. Ganderton et al., Buying Insurance for Disaster-Type Risks: Experimental Evidence, 20 J. Risk & Uncertainty 271 (2000); Paul Schoemaker & Howard C. Kunreuther, An Experimental Study of Insurance Decisions, 46 J. Risk & Ins. 603 (1979); Paul Slovic et al., Preference for Insuring Against Probable Small Losses: Insurance Implications, 44 J. Risk & Ins. 237 (1977); see also Gary H. McClelland et al., Insurance for Low-Probability Hazards: A Bimodal Response to Unlikely Events, 7 J. Risk & Uncertainty 95 (1993) (finding extreme bimodality in willingness to pay). But see Susan K. Laury et al., Insurance Decisions for Low-Probability Losses, 39 J. Risk & Uncertainty 17 (2009).
  • 78Hacala, supra note 57, at 9.
  • 79Howard C. Kunreuther & Mark V. Pauly, Behavioral Economics and Insurance: Principles and Solutions, in Research Handbook on the Economics of Insurance Law 15, 21–23 (Daniel Schwarcz & Peter Siegelman eds., 2015); see Kunreuther et al., supra note 67, at 7 (“Insurers . . . misunderstand how to predict rare events and therefore sometimes make decisions that appear to ignore risks altogether. Alternatively, they sometimes fixate on the magnitude of recent losses and claims without weighting these figures by an estimate of the likelihood of another catastrophe occurring.”); Katzman, supra note 69, at 85. This is not to say there are no methods to predict extreme events.  See, e.g., Robert Lund, Revenge of the White Swan, 61 Am. Statistician 189, 190 (2007) (discussing extreme value theory, “the statistician’s bible for quantifying rare events”).
  • 80Hacala, supra note 57, at 9.
  • 81Id.
  • 82Abraham, supra note 4, at 944; see also Abraham, supra note 70, at 51 (“[T]he capacity of any system of liability to promote optimal deterrence depends on how predictable liability is.”). Compare Richard Lempert, Low Probability/High Consequence Events: Dilemmas of Damage Compensation, 58 DePaul L. Rev. 357, 385 (2009) (listing reasons, including “psychological denial,” that “make it almost inevitable that insufficient precaution will be taken” against low-probability, high-consequence events), and Katzman, supra note 69, at 83 (“Given a ‘finite reservoir of concern,’ risk managers may pay little attention to low-probability risks, no matter how severe the potential consequences.”), with Eric A. Posner, Probability Errors: Some Positive and Normative Implications for Tort and Contract Law, 11 Sup. Ct. Econ. Rev. 125, 126 (2004) (showing that a “person who discounts remote risks might take too much care, rather than too little”).
  • 83Abraham, supra note 70, at 965.
  • 84Telephone Interview with Commercial Insurer C (Oct. 26, 2015). I do not mean to suggest there is nothing insurers can do to improve their position vis-à-vis claims arising from actions taken years ago. Some insurers have gone back to shore up their reserves for past coverage years. Id. But this does not reduce the likelihood that harm from past actions will manifest.
  • 85For more on methodology, see supra note 5.
  • 86See, e.g., Telephone Interview No. 2 with Risk Pool A, supra note 61 (explaining that the pool had never priced the risk or looked at specific loss-prevention measures because it had never considered the risk, which it believed more relevant in big cities and in the South).
  • 87Id.; Telephone Interview No. 2 with Commercial Insurer B (Oct. 1, 2015) (speculating that wrongful convictions are not a significant issue for pools his company reinsures because the pools’ members are “puny munis”). But see Telephone Interview with Commercial Insurer D, supra note 44 (reinsures small pools that have encountered wrongful conviction claims).
  • 88Telephone Interview with Commercial Insurer D, supra note 44.
  • 89Telephone Interview No. 2 with Risk Pool C (Sept. 28, 2015); see also Telephone Interview with Risk Pool E (Oct. 7, 2015) (reporting that wrongful convictions are “just coming onto the radar,” and that an industry conference, last year for the first time, included a session on wrongful convictions, focusing on the duty to disclose exculpatory evidence).
  • 90Telephone Interview No. 2 with Risk Pool C, supra note 89; see also Telephone Interview No. 2 with Risk Pool A, supra note 61 (stressing “integrity first in personal and professional behavior”).
  • 91Telephone Interview No. 2 with Risk Pool C, supra note 89.
  • 92Id. This expert’s admission calls to mind the terrorism context, in which insurers have been similarly unable to gauge risk levels accurately. See Michelle E. Boardman, Known Unknowns: The Illusion of Terrorism Insurance, 93 Geo. L.J. 783, 815–20 (2005) (arguing that terrorism risk is incalculable largely because the data are too scarce and disparate; that insurers are “guessing” on prices, which vary wildly; and that pricing has “not been very scientific” because “underwriters are relying on their experience and instincts”).
  • 93Kunreuther & Pauly, supra note 79, at 24.
  • 94See Boardman, supra note 92, at 836–42 (discussing the efficiency costs of inaccurate pricing, which sends inaccurate signals about the risk and cost of harm). In theory, insurers could be (unwittingly) charging too much, rather than too little, to cover the risk of wrongful convictions, leading insured municipalities to overinvest in loss prevention. I think this is unlikely, however, and I heard no hint of it during any of my conversations with industry experts.
  • 95Kunreuther & Pauly, supra note 79, at 19; see Cass R. Sunstein, Terrorism and Probability Neglect, 26 J. Risk & Uncertainty 121 (2003).
  • 96Kunreuther et al., supra note 67, at 221; accord Boardman, supra note 92, at 786–87.
  • 97Kunreuther et al., supra note 67, at 221.
  • 98See Benjamin J. Richardson, Mandating Environmental Liability Insurance, 12 Duke Envt’l L. & Pol’y F. 293 (2002).
  • 99Telephone Interview with Commercial Insurer C, supra note 84; Telephone Interview with Commercial Insurer D, supra note 44. Liability insurance tends to follow an “underwriting cycle” in which “premiums and restrictions on coverage . . . rise and fall as insurers tighten their standards in response to the loss of capital”—called a “hard market”—“or, alternately, loosen their standards in order to maintain or grow market share when new capital enters the market”—a “soft market.” See Tom Baker & Sean J. Griffith, Ensuring Corporate Misconduct: How Liability Insurance Undermines Shareholder Litigation 55 (2010).
  • 100For descriptions of the two approaches, see Abraham, supra note 70, at 71–74.
  • 101See Garrett, supra note 21, at 103–04 (discussing these and other reforms to reduce the risk of wrongful convictions).
  • 102A 1991 study reported that 61.6% of law enforcement liability policies were occurrence-based. ICMA & Wyatt Co., Law Enforcement Officials Liability Insurance: Current Status—1991, at 6 (1991). A more recent publication states that claims-made forms are now more common, but the basis for this assertion is not made plain. See Albert P. Amato, Reinsurance Reference Guide 117 (2012). 
  • 103Abraham, supra note 70, at 49–50.
  • 104Id. at 50.
  • 105Id.
  • 106Id.
  • 107Id.
  • 108Kenneth S. Abraham, Insurance Law and Regulation 622 (5th ed. 2010); see also Jaap Spier, Long Tail (Liability) Risks and Claims Made Policies, 23 Geneva Papers on Risk & Ins. 152 (1998). More generally, claims-made policies work well to reduce the problem of insuring correlated risk. See Doherty & Dionne, supra note 63, at 198; see also Neil A. Doherty, The Design of Insurance Contracts When Liability Rules Are Unstable, 58 J. Risk & Ins. 227 (1991). Municipal risk pools, which are essentially small mutual insurers, may have less need to use a claims-made policy, as the mutual form may accomplish similar objectives. See Doherty & Dionne, supra note 63, at 196–97.
  • 109Abraham, supra note 70, at 50.
  • 110Id.; see also Katzman, supra note 69, at 88 (asserting that, under claims-made policies, “current premiums are not easily affected by current risk management practices”).
  • 111Abraham, supra note 70, at 50.
  • 112Id.
  • 113Id. at 51.
  • 114See Telephone Interview with Commercial Insurer D, supra note 44; cf. Katzman, supra note 69, at 87 (describing how environmental impairment liability is excluded from occurrence-based general liability policies and written separately on a claims-made basis).
  • 115See Lempert, supra note 82, at 385 (arguing that government planning and regulation is necessary to control low-probability, high-consequence events).
  • 116See, e.g., Gould et al., supra note 59, at 97–101; Int’l Ass’n of Chiefs of Police & Office of Justice Programs, U.S. Dep’t of Justice, National Summit on Wrongful Convictions: Building a Systemic Approach To Prevent Wrongful Convictions (2013), http://www.innocenceproject.org /files/imported/wrongful_
    convictions_summit_report_2013.pdf (https://perma.cc/Y3NW-HTTF]; Garrett, supra note 21, at 103–06 (detailing reforms and describing them as “inexpensive and easy to adopt”).
  • 117See Rappaport, supra note 8, at 252–54.
  • 118See generally id.
  • 119Cf. Kami Chavis Simmons, Cooperative Federalism and Police Reform: Using Congressional Spending Power To Promote Police Accountability, 62 Ala. L. Rev. 351 (2011).
  • 120See, e.g., Matthieu Bussière & Marcel Fratzscher, Low Probability, High Impact: Policy Making and Extreme Events, 30 J. Pol’y Modeling 111 (2008). To some extent, we see this occasionally already. See, e.g., Jess Bidgood, Massachusetts Justices Clear Way for New Trials in Cases Chemist May Have Tainted, N.Y. Times, May 19, 2015, at A11 (describing decision to allow thousands of defendants whose cases may have been tainted by a state chemist’s work to seek new trials); Steve Mills & Ken Armstrong, Hard Calls Face Ryan in Death Row Review, Chi. Trib., Mar. 5, 2002 (describing the Illinois governor’s pledge to review the cases of all 159 death row inmates due partly to concerns about wrongful convictions stemming from systemic flaws).
  • 121See Ronald Jay Allen et al., Criminal Procedure: Investigation and Right to Counsel 337 (3d ed. 2016) (“The typical Fourth Amendment case—say, a gratuitous frisk or car search—does not involve the kind of physical injury or property damage that would translate into significant money damages . . . .”).
  • 122See Memphis Cmty. Sch. Dist. v. Stachura, 477 U.S. 299, 308 (1986) (holding that “damages based on the abstract ‘value’ or ‘importance’ of constitutional rights are not a permissible element of compensatory damages”).
  • 123Carey v. Piphus, 435 U.S. 247, 248, 263–67 (1978) (holding that, “in the absence of proof of actual injury. . . only nominal damages” are due for a procedural due process violation).
  • 124See, e.g., Skinner v. Ry. Labor Execs.’ Ass’n, 489 U.S. 602, 613–14 (1989) (stating that “the [Fourth] Amendment guarantees the privacy, dignity, and security of persons against certain arbitrary and invasive acts by officers of the Government or those acting at their direction”). Ironically, in the very decision that condoned the stop-and-frisk maneuver, laying the groundwork for controversial “order maintenance” policing programs, the Supreme Court articulated quite sharply the personal harms a frisk inflicts:

    [I]t is nothing less than sheer torture of the English language to suggest that a careful exploration of the outer surfaces of a person’s clothing all over his or her body in an attempt to find weapons is not a “search.” Moreover, it is simply fantastic to urge that such a procedure performed in public by a policeman while the citizen stands helpless, perhaps facing a wall with his hands raised, is a “petty indignity.” It is a serious intrusion upon the sanctity of the person, which may inflict great indignity and arouse strong resentment, and it is not to be undertaken lightly.

    Terry v. Ohio, 392 U.S. 1, 16–17 (1968) (footnote omitted).

  • 125Terry, 392 U.S. at 20–22.
  • 126Technically, a Miranda violation is not itself a constitutional violation. See Chavez v. Martinez, 538 U.S. 760 (2003). But even if it were, it is hard to see why damages would be more than nominal.
  • 127See Robert V. Percival & Geoffrey P. Miller, The Role of Attorney Fee Shifting in Public Interest Litigation, 47 Law & Contemp. Probs. 233, 239–41 (1984).
  • 128See Farrar v. Hobby, 506 U.S. 103, 114–15 (1992) (holding that, when recovery of damages is the purpose of a suit, the fee award should depend on “the amount of damages awarded as compared to the amount sought,” and that plaintiffs who seek compensatory damages but receive only nominal damages “often” should receive “no attorney’s fees at all” (quoting City of Riverside v. Rivera, 477 U.S. 561, 585 (Powell, J., concurring in judgment))). Notwithstanding Farrar, “there are circumstances where a judgment of nominal damages will support an award of attorneys’ fees, based on the vindication of important constitutional rights.” Michael Avery et al., Police Misconduct: Law and Litigation 974 (3d ed. 2015). Even so, defendants are permitted to condition settlement of civil rights cases on the waiver or reduction of attorney fees, which, by some accounts, has “destroyed section 1983 as a remedy for civil rights plaintiffs with only modest damages.” Reingold, supra note 3, at 4.
  • 129See, e.g., City of Los Angeles v. Lyons, 461 U.S. 95 (1983); Rizzo v. Goode, 423 U.S. 362 (1976); O’Shea v. Littleton, 414 U.S. 488 (1974).
  • 130See, e.g., Nat’l Cas. Co., supra note 17, at 2 (excluding from coverage “‘claim(s),’ demands, or actions seeking relief or redress in any form other than monetary damages”).
  • 131See, e.g., Telephone Interview with Commercial Insurer A, supra note 41 (stressing importance of deductibles and self-insured retentions for effective risk management).
  • 132See, e.g., Ind. Mun. Ins. Program, supra note 24 (advertising police training videos covering Terry and Miranda).
  • 133See, e.g., Chavez v. Martinez, 538 U.S. 760, 798–99 (2003) (Kennedy, J., concurring in part and dissenting in part).
  • 134See generally Meltzer, supra note 14.
  • 135Leong, supra note 13, at 425.
  • 136See, e.g., Floyd v. City of New York, 959 F. Supp. 2d 540 (S.D.N.Y. 2013).
  • 137For a proposal on how to extract the greatest returns from the limited resources available for § 14141 litigation, see Rachel A. Harmon, Promoting Civil Rights Through Proactive Policing Reform, 62 Stan. L. Rev. 1 (2009).
  • 138See, e.g., Floyd, 959 F. Supp. 2d at 559–60 (finding that, during an eight-year period in New York City, “at least 200,000 stops were made without reasonable suspicion,” and “[t]he actual number of stops lacking reasonable suspicion was likely far higher”).
  • 139See generally Rappaport, supra note 8.
  • 140See, e.g., Whren v. United States, 517 U.S. 806, 813 (1996) (stating that “the Constitution prohibits selective enforcement of the law based on considerations such as race”).
  • 141Stipulation and Order of Settlement and Discontinuance at 2, Sultan v. Kelly, No. 09-CV-00698 (RJD) (RER) (E.D.N.Y. June 30, 2009).
  • 142Telephone Interview with Commercial Insurer D, supra note 44 (asserting that $100,000 is a common self-insured retention among his company’s policyholders).
  • 143See, e.g., Settlement Agreement at 1–2, Md. State Conf. of NAACP Branches v. Md. State Police, No. FPS-98-1098 (D. Md. Apr. 3, 2008) (agreeing to settle the claims of six plaintiffs for $300,000 total); Settlement Agreement at 7, Wilkins v. Md. State Police, No. MJG-93-468 (D. Md. Jan. 5, 1995) (settling the claims of four plaintiffs for $12,500 per plaintiff ); ACLU, ACLU of NJ Wins $775,000 for Victims of Racial Profiling by State Troopers (Jan. 13, 2003), https://www.aclu.org /news/aclu-nj-wins-775000-victims-racial-profiling-state-troopers (https://perma.cc/3ZDZ-URSF] (awarding majority of plaintiffs around $31,000 each).
  • 144See, e.g., Gousse v. City of Los Angeles, No. B174896, 2007 Cal. App. Unpub. LEXIS 2882 (Cal. Ct. App. Apr. 10, 2007) (affirming trial court’s decision to grant a new trial on damages after a jury awarded $33,000,000 to a urological surgeon who claimed lost earning capacity, among other damages); ACLU, supra note 143 (reporting having settled claims of two plaintiffs who were beaten and held at gunpoint for $200,000 each).
  • 145See, e.g., The Leadership Conference, Restoring a National Consensus: The Need to End Racial Profiling in America (Mar. 2011), http://www.civilrights.org /publications/reports/racial-profiling2011/racial_profiling2011.pdf (https://perma.cc/4EUN-JNWN] (collecting evidence); Ian Ayres, Racial Profiling in L.A.: The Numbers Don’t Lie, L.A. Times, Oct. 23, 2008.
  • 146Recall that most policies do not cover defense against suits for declaratory and injunctive relief. See supra note 129.
  • 147Susan Kostro, Police Excessive Force Raises Liability Risk Scrutiny, Ironshore
    (Oct. 1, 2015), http://www.ironshore.com/blog /police-excessive-force-raises-liability-risk-scrutiny (https://perma.cc/RFP7-4PUG).
  • 148Telephone Interview with Risk Pool D, supra note 27; Telephone Interview with Risk Pool E, supra note 89.
  • 149Gallagher, supra note 19, at 52; see also Girod, supra note 19, at 163 (declining to list racial profiling among fourteen “most common [types of] ‘actionable conduct’ involving civil rights liability” for police).
  • 150See, e.g., Marshall v. Columbia Lea Regional Hosp., 345 F.3d 1157, 1168 (10th Cir. 2003).
  • 151See, e.g., id. (“In general, the absence of an overtly discriminatory policy or of direct evidence of police motivation results in most claims being based on statistical comparisons between the number of black or other minority Americans stopped or arrested and their percentage in some measure of the relevant population.”); David Rudovsky, Litigating Civil Rights Cases to Reform Racially Biased Criminal Justice Practices, 39 Colum. Hum. Rts. L. Rev. 97, 109–12 (2007) (“[W]here discrimination is sufficiently ‘clandestine and covert,’ statistical evidence of a discriminatory pattern is the ‘only available avenue of proof.’” (quoting Int’l Bd. of Teamsters v. United States, 431 U.S. 324, 339 n.20 (1977))); Brian L. Withrow & Jeffrey Doug Dailey, Racial Profiling Litigation: Current Status and Emerging Controversies, 28 J. Contemp. Crim. Just. 122, 130 (2012). Fourth Amendment doctrine regards discriminatory intent as irrelevant; as long as the police have sufficient cause to, say, pull over a vehicle, it matters not whether the driver’s race supplies their true motive for the stop. Whren v. United States, 517 U.S. 806, 813 (1996).
  • 152Dique v. N.J. State Police, 603 F.3d 181, 184 (3d Cir. 2010).
  • 153See, e.g., Floyd v. City of New York, 959 F. Supp. 2d 540, 558–60, 572 n.100, 590 (S.D.N.Y. 2013) (relying, in a lawsuit filed in 2008, on data and expert analysis reaching to January 2004, a state attorney general report from 1999, and evidence of “more than a decade” of institutional indifference); Floyd v. City of New York, 283 F.R.D. 153, 160 (S.D.N.Y. 2012) (certifying a class covering individuals stopped by police beginning in 2005); Complaint at 9–13, Floyd v. City of New York, No. 08 Civ. 1034 (SAS) (S.D.N.Y. Jan. 31, 2008) (connecting present-day allegations to history of police activity dating to the 1970s).
  • 154First Amended Complaint at 6, 16, Rodriguez v. Cal. Hwy. Patrol, No. 5:99-CV-20895-JF (N.D. Cal. Nov. 30, 1999).
  • 155Id. at 6–7.
  • 156Id. at 7.
  • 157See supra pp. 386–92.
  • 158See, e.g., Nick Pinto, The Bail Trap, N.Y. Times Mag., Aug. 13, 2015, at MM38.
  • 159See Armstrong v. United States, 517 U.S. 456 (1996) (denying defendants discovery on selective prosecution claims unless they can show that the government declined to prosecute similarly situated suspects of other races). Courts are divided on whether (and how) the Armstrong standard applies in the selective enforcement (i.e., policing) context. See Jody Feder, Cong. Research Serv., RL31130, Racial Profiling: Legal and Constitutional Issues 6–9 (2012).
  • 160My research uncovered no complete victories by criminal defendants based on racial profiling under either state or federal law. New Jersey courts have granted motions to suppress evidence based on state-law equal protection violations, though even these partial victories are rare. See, e.g., State v. Segars, 799 A.2d 541, 552 (N.J. 2002) (“This is a very unusual case. Without Officer Williams’s repudiated testimony, the evidence produced by Segars that Officer Williams saw him prior to the MDT check would have been completely inadequate to support an inference of discriminatory enforcement.”); State v. Soto, 734 A.2d 350 (N.J. Super. Ct. 1996); see also David A. Harris, Racial Profiling Redux, 22 St. Louis U. Pub. L. Rev. 73, 77–79 (2003) (characterizing as unsurprising the “lack of litigation success in suits against racial profiling,” and crediting “unusual circumstances” for successes in Soto and a famous civil case); Lewis R. Katz, “Lonesome Road”: Driving Without the Fourth Amendment, 36 Seattle U. L. Rev. 1413, 1427 (2013) (“Only New Jersey courts have granted motions to suppress in Fourteenth Amendment equal protection claims, based on their interpretation of the New Jersey Constitution.” (footnote omitted)).
  • 161See, e.g., Melendres v. Arpaio, 784 F.3d 1254 (9th Cir. 2015) (upholding part of a permanent injunction); Floyd v. City of New York, 959 F. Supp. 2d 668 (S.D.N.Y. 2013) (entering a permanent injunction); see also Arnold v. Ariz. Dep’t of Pub. Safety, No. CV–01–1463–PHX–LOA, U.S. Dist. LEXIS 53315 (D. Ariz. July 31, 2006) (affirming a settlement agreement containing extensive prospective relief, including modifications to police procedures and training requirements); In re Cincinnati Policing, 209 F.R.D. 395 (S.D. Ohio 2002) (similar). On the general difficulty of litigating racial profiling claims effectively, see David Cole, No Equal Justice: Race and Class in the American Criminal Justice System 40 (1999); Albert W. Alschuler, Racial Profiling and the Constitution, 2002 U. Chi. Legal F. 163, 245, 248; Harris, supra note 160, at 78; Kevin R. Johnson, Racial Profiling in America, 98 Geo. L.J. 1005, 1069, 1069 n.394 (2011); Kami Chavis Simmons, Beginning To End Racial Profiling: Definitive Solutions to an Elusive Problem, 18 Wash. & Lee J. Civil Rts. & Soc. Just. 25, 37 (2011).
  • 162NAACP, Born Suspect: Stop-and-Frisk Abuses and the Continued Fight
    To End Racial Profiling in America 19 (Sept. 2014), http://action.naacp.org /page/-/Criminal%20Justice/‌Born_Suspect_Report_final_web.pdf (https://perma.cc/T54T-KVU2).
  • 163Id. at Appx. II (listing, among essential components of an effective racial profiling law, “funds for periodically retraining officers and installing in-car video cameras, body-worn cameras, and gun cameras”); see L. Song Richardson, Police Racial Violence: Lessons From Social Psychology, 83 Fordham L. Rev. 2961, 2975–76 (2015) (describing the Fair and Impartial Policing program, which educates agencies about implicit bias); Robert J. Smith, Keynote Address, Reducing Racially Disparate Policing Outcomes: Is Implicit Bias Training the Answer?, 37 Haw. L. Rev. 295, 300 (2015) (describing implicit bias training as an “ascendant idea in policing and scholarly circles” and providing citations); Implicit Bias, Nat’l Initiative for Building Cmty. Trust & Just., http://trustandjustice.org /resources/intervention/implicit-bias (https://perma.cc/T8XH-Y2A8] (last visited Dec. 19, 2015) (discussing implicit bias interventions and collecting academic research); Press Release, State of Cal. Dep’t of Justice, Attorney General Kamala D. Harris Kicks Off First-of-its-Kind Law Enforcement Training on Implicit Bias & Procedural Justice (Nov. 17, 2015), https://oag.ca.gov/news/press-releases/attorney-general-kamala-d-harris-kicks-first-its-kind-law-enforcement-training (https://perma.cc/ZJ8T-XBM5).
  • 164See, e.g., John J. Donohue III & Steven D. Levitt, The Impact of Race on Policing and Arrests, 44 J.L. & Econ. 367, 371 & tbl. 2 (2001) (finding that “an increase in the number of white police is associated with more arrests of minorities but little change in white arrests,” while “an increase in minority officers is associated with more white arrests but not more minority arrests”); Joscha Legewi & Jeffrey Fagan, Group Threat, Police Officer Diversity and the Deadly Use of Force (Columbia Law Sch. Pub. Law Research Paper No. 14-512, 2016), http://papers.ssrn.com/‌sol3/‌papers.cfm?abstract_id=2778692 (finding that a diverse police force reduces the influence of group threat, lowering the number of officer-involved killings of African Americans); see also Lorie Fridell et al., Racially Biased Policing: A Principled Response 68–78 (2008) (arguing that hiring a racially diverse police force can help reduce racial bias in policing); Robert J. Friedrich, Police Use of Force: Individuals, Situations, and Organizations, 452 Annals Am. Acad. Pol. & Soc. Sci. 82, 90 (1980) (finding that biracial teams of partners use less force). Overall, however, the empirical evidence regarding the performance effects of having a diverse police force is mixed. See David A. Sklansky, Not Your Father’s Police Department: Making Sense of the New Demographics of Law Enforcement, 96 J. Crim. L. & Criminology 1209, 1224–25, 1229, 1230 (2006) (reviewing studies).
  • 165See U.S. Dep’t of Justice, Guidance For Federal Law Enforcement Agencies Regarding The Use of Race, Ethnicity, Gender, National Origin, Religion, Sexual Orientation, or Gender Identity (2014); see also NAACP, supra note 162, at 18–19 (describing repeated attempts to pass federal legislation); Letter from The Leadership Conference to Barack Obama, President of the U.S. (Feb. 24, 2015), http://civilrightsdocs.info/pdf/Sign-On-Letter-Re-DOJ-Guidance-Revisions.pdf (https://perma.cc/AR6Q-8EEH] (conveying the “serious concerns” of eighty public interest groups about the DOJ’s 2014 guidance).
  • 166President’s Task Force on 21st Century Policing, Final Report of the President’s Task Force on 21st Century Policing 28 (2015).
  • 167See Rappaport, supra note 8, at 269.
  • 168See, e.g., Smith, supra note 163, at 302 (noting that, despite the growing popularity of implicit bias training, when it comes to efficacy, “empirical support is lacking”); Jack Glaser, How to Reduce Racial Profiling, Greater Good (May 28, 2015) http://greatergood.berkeley.edu/article/
    item/how_reduce_racial_profiling (https://perma.cc/7UQW-T5ZA] (maintaining that, “to date, research has yet to uncover a straightforward method that can lastingly mitigate implicit biases” that result in racial profiling).
  • 169See Isaiah Berlin, The Hedgehog and the Fox 1 (Henry Hardy ed., Princeton Univ. Press 2013) (1953) (“The fox knows many things, but the hedgehog knows one big thing.”).