Ada JoAnn Taylor was sentenced to 40 years in prison for playing a role in the 1985 murder and rape of Helen Wilson. The victim was stabbed and suffocated in her Beatrice, Neb. apartment.
Brian Banks was sentenced to seven years in prison for raping a younger woman at the Los Angeles high school where he was attending summer school in 2002. After his release, he was required to register as a sex offender and to wear an electronic monitoring bracelet for a time.
Rodney Roberts was sentenced to seven years in prison for the 1996 kidnapping of a 17-year-old woman in Newark, N.J. After completing his sentence, he was classified as a sexually violent predator and required to spend time in a secure treatment facility.
Each of those convictions share two commonalities besides the horrific nature of the crimes. First, each of the individuals confessed and pleaded guilty to the crimes. Second, the individuals were subsequently exonerated on the basis of evidence or testimony that unequivocally demonstrated they had not committed the crimes in question.
In these cases, there was significant collateral damage. Ochoa, in his plea arrangement, agreed to testify against his roommate, Richard Danziger, for also being involved in the murder. Ochoa, who ultimately spent 13 years in prison, indicated he took his plea deal and testified against Danziger to avoid the risk of a death sentence. Danziger was convicted and was also later found to have not been involved in DePriest’s killing. He had an alibi for the day of the murder, but was convicted on the basis of Ochoa’s testimony and some questionable lab evidence. Danziger, while serving his 12-year prison term, was attacked by another inmate, leaving him with permanent brain damage and confined to a mental institution.
Taylor, who spent 19 years in prison, accepted a plea deal in which she agreed to testify that Joseph White had raped Wilson. White served 19 years of a life sentence. White testified that he had never been in Wilson’s apartment and fingerprints found at the scene were not matches for White, Taylor, or the victim. After White obtained access to DNA testing of semen found at the crime scene, his defense team discovered that the sample matched another individual who had been the leading suspect early in the investigation. Prior to this, White’s requests to have the DNA evidence examined were repeatedly turned down. Taylor claims that police and prosecutors threatened her with the death penalty if she did not confess to the crime and provide testimony against White.
Banks, who before his conviction had received interest from the University of Southern California and the University of California, Los Angeles as a high school football recruit, insisted that the sexual contact between him and his accuser was consensual. Banks’ mother sold her condo and car to fund his defense. Faced with the prospect of a life sentence, Banks pleaded no contest to forcible rape and spent more than five years in prison.
Roberts pleaded guilty to kidnapping to avoid a sexual assault charge that was based on a claim by the police that the victim had identified Roberts in a photographic lineup. Years later, the victim disputed the claim that she had identified anyone. Seventeen years later, DNA evidence excluded Roberts as the rapist.
Why Do the Innocent Plead Guilty?
As sad as these individual cases are, if they were isolated incidents, they could be seen as the inevitable mistakes that arise in any system. Humans and human institutions are fallible and the optimal number of such mistakes is surely not zero. However, there is some evidence that these kinds of problems are not exceedingly rare. According to data from the Innocence Project, in at least 360 cases where a convicted individual was later exonerated, the person had pleaded guilty to the crime.
Federal Judge Jed Rakoff of the Southern District of New York suggested in a New York Review of Books essay (“Why Innocent People Plead Guilty,” Nov. 20, 2014) that such guilty pleas by innocent defendants are a systematic part of the U.S. plea-bargaining system. In his view, prosecutors have significant leverage over defendants because they can use their discretion regarding what charges are filed. If the charges have severe mandatory minimum sentences, plea bargaining can look like a good deal to risk-averse defendants, even innocent ones.
The most attractive plea deals are offered early in the case, often when the defendant and his lawyer (usually an over-worked, under-funded public defender) have very little information regarding the evidence the state has. Turning down an early plea deal generally ensures that worse ones will be offered as time goes on, or at least that’s the threat communicated by prosecutors. Putting all of this together, it is arguably not surprising that plea bargains constitute around 95% of all resolutions of criminal cases (at least of those where charges are not dismissed) at both the state and federal level. Rakoff suggests that this plea-bargaining assembly line significantly undercuts what the nation’s founding fathers viewed as a necessary backstop against tyranny: the jury trial.
Fighting False Convictions
Beyond guilty pleas from innocent defendants, the U.S. criminal justice system generates a non-trivial number of mistaken convictions. According to the National Registry of Exonerations, since 1989 there have been more than 2,000 individuals exonerated through late-coming evidence, such as DNA tests, recanted testimony, and admissions of guilt by others. Based on the registry’s data, these falsely convicted individuals have served a cumulative total of more than 17,000 years in prison—a great loss to both those individuals and society as a whole. Those numbers are likely just the tip of the iceberg.
Although public interest groups such as those affiliated with the Innocence Network work tirelessly to identify and provide legal resources to people who may have been wrongfully convicted, the groups’ resources are limited. Out of practical necessity, they focus their attention on the cases that are most likely to succeed. For cases not taken up by these groups, there is little hope because convicts typically have few financial resources and little human capital to pursue exonerations on their own.
Given the limited resources of groups like the Innocence Project (a founder and affiliate of the Innocence Network) and the limited ability of a convict’s own self-help efforts to reverse mistakes in the U.S. criminal system, a more systematic approach to avoiding errors is necessary. In the context of coercive plea deals, Judge Rakoff notes that reducing the severity of criminal penalties, including getting rid of mandatory minimums and other sentencing guidelines, would help to limit the leverage prosecutors have. That said, he acknowledges that there is little taste for this kind of leniency among the general public. Beyond that, such an approach, while reducing penalties for the innocent, would also reduce penalties for the guilty.
The English jurist William Blackstone famously opined that it is better to let 10 guilty people escape than to let a single innocent person suffer. But there are costs to such a position. Prison terms generate benefits in terms of both incapacitation and deterrence. Reducing penalties likely will increase crime directly as potential criminals fear punishment less and those who have chosen to commit criminal acts spend less time in prison, allowing them to recidivate. It would also increase crime indirectly because it would become more difficult for prosecutors to do their job because they would no longer have as much leverage to secure pleas, and thus they would have to spread their resources more thinly across cases.
Rakoff suggests establishing a sort of nonbinding arbitration where magistrates examine the available evidence and make recommendations about whether a case should proceed or be dropped by the prosecution. If the case is pursued, the magistrate would propose a sensible plea deal. The core of the idea is to provide more balance wherein the defendant is not merely at the mercy of the prosecutor. The process would limit the information advantage held by the prosecution. Also, by providing a neutral viewpoint, the defendant cannot be misled into thinking it is surely the case that if the plea is not taken, the ultimate outcome will certainly be worse for the defendant.
While this neutral broker approach has much to recommend it, it potentially involves a significant resource infusion into the system. The magistrate/arbitrator’s time is obviously not free. Further, this extra step in the plea-bargaining system will require that both the prosecutor and the public defender spend more time and resources on a given case. These costs are not necessarily a reason to reject the proposal, but they may limit its feasibility.
Payments to the Innocent
In a recent Journal of Law and Economics article (“Reducing False Guilty Pleas and Wrongful Convictions through Exoneree Compensation,” 59:1, 173–189 [February 2016]), we take a different approach to the problem of legal mistakes. We propose a system in which innocent parties who are convicted are promised a potentially large payment in the event they are subsequently exonerated. As discussed below, such an approach has benefits both in terms of fairness concerns and in terms of the efficiency of the criminal justice system.
Our proposal uses what economists call a “mechanism design approach” to the plea bargaining “game.” In many strategic situations, the various parties involved have private information: one party knows something relevant that the others do not, and vice versa. In a mechanism design approach, the goal is to provide incentives for the parties to reveal their private information either directly or indirectly by making choices that expose important aspects of the private information.
These mechanism design approaches abound in non-criminal settings. To take a simple example, employers may have a desire to attract individuals who plan to remain at a firm for a long time. Employers would want this because the hiring process entails costs and the firm will also make some investments in training the employee. At the interview stage, although the firm would like to ask candidates if they plan to stick around for the long term, there is no way to tell which individuals are honest when they answer affirmatively. To screen for the individuals whose private information indicates they do plan to stay in the job, the employer may offer a package where the salary is relatively low (as compared to similar employment opportunities), while augmenting the pay with some longer-term bonus (e.g., retirement contributions that only vest after the employee has stayed in the job for a number of years). For those job applicants who believe they will leave after a year or so, this salary-plus-bonus arrangement is a bad deal, while for those planning to stay, the bonus can be set so as to make the job more attractive than other alternatives. Essentially, by offering a package that is only attractive to candidates who have the desired but unobservable characteristic, the firm can induce individuals to reveal their private information about whether they have that characteristic or not.
In the plea-bargaining setting, the defendant has private information regarding his guilt or innocence. Of course, much like the hiring setting, everyone has an incentive to claim he has the good characteristic, in this case innocence. Because everyone claims innocence, the defendant’s assertion has no probative value. In such a case, defendants implicitly do a cost-benefit calculation in which they compare the plea deal (say, five years in prison with 100% certainty) to the trial alternative (say, a 50% chance of being acquitted plus a 50% chance of 20 years in prison). Faced with this choice, many people will take the plea deal.
Our contribution is to suggest that the state implement a system in which individuals who are falsely convicted are paid damages in the event that they are later found innocent through DNA evidence or some other method of exoneration. Using the comparison noted above, the plea deal of a certain five years in prison would be compared with the 50% chance of acquittal plus a 50% chance of 20 years in prison plus p% chance of an exoneration payment multiplied by the amount of the payment. For the guilty, p will essentially equal 0, leaving the decision problem unaffected. For the innocent, however, with a high enough expected exoneration payment, the outcome of the decision could be flipped where more innocent individuals take their chances at trial without affecting the choices of those who know they are, in fact, guilty.
The intuition behind this idea is relatively simple. Our technical paper demonstrates that the proportion of innocent individuals who plead guilty can be decreased through exoneration compensation. This, in turn, would result in a reduction in the number of wrongful convictions because fewer innocent people choose the option of serving time with certainty. Our article incorporates a realistic concern, namely that individuals do not only differ with respect to whether they are guilty or not, but on other dimensions that affect their willingness to plead guilty. This is why only a fraction of all innocent individuals can be incentivized to refuse pleading guilty.
Other benefits / Of course, the real world and the world of economic theory are often orthogonal. Real people may find it difficult to trade the prospect for probabilistic money damages against additional years of freedom lost. These problems most likely get even worse when the sentence avoided through a plea deal involves life in prison or even the potential for the death penalty. Therefore, in some cases, even large exoneration compensations may not have a substantial effect on the choices of innocent individuals.
However, even if the prospect of an exoneration payment only helps improve the plea bargain system on the margin (i.e., makes it slightly more likely that the innocent will forgo a plea deal without changing the incentives of the guilty), the proposal has other benefits that should be considered. First, such a payment accords with many people’s normative views about fairness. Our tort system calls for damages in cases of false imprisonment when the one doing the imprisoning is a private party. Presumably, intuitions about fairness apply in the public context as well.
Additionally, the exoneration payments can improve incentives in other ways beyond the effect on choices in the plea-bargaining situation. On the prosecution side, if the damages are somehow linked to the prosecutor’s budget or—perhaps more directly yet—the prosecutor’s income or career prospects, it could induce prosecutors to make better decisions, be more honest in the evaluation and presentation of evidence, and focus on cases where there is less uncertainty regarding a defendant’s guilt. Moreover, this would enhance deterrence because prosecutors would spend more resources on cases against guilty individuals and, therefore, criminals would expect more severe punishment.
Operationally, this might seem difficult to pull off. But presumably prosecutors’ offices would secure some kind of insurance to cover the exoneration payments. Insurers would be attractive in this regard because they could address a time mismatch problem that could arise. That is, if the prosecutor is making legal decisions in year t, leaves the job in year t + 5, and any exoneration happens on average in year t + 9 (the average indicated by the National Registry of Exonerations’ count of exonerations and years served by those exonerated), it is perhaps unlikely that future payments influence the prosecutor’s decisions. Insurers, however, have a longer time horizon and, therefore, could be incentivized to determine what constitute best practices to optimize exoneration risk and then impose them on the prosecutors’ offices they insure. It might even prove to be sensible for the insurer to be the funder and administrator of any retirement benefits that the prosecutors accrue. In such a set-up, the more exoneration payouts there are, the lower the retirement benefits, providing a back-end incentive to do better work on the front end.
John Rappaport, a law professor at the University of Chicago, has documented a similar insurance-based approach to reining in police misconduct. In that work, he demonstrates how insurance companies, through their decisions to offer or deny coverage to police departments and in the way they set premiums, affect police department policies (such as use-of-force policies), what training officers are provided, and even officer hiring and firing decisions. A benefit in using insurers to drive the risk management policies, whether it be those of police departments or prosecutors’ offices, is that insurers tend to have access to much more data than the police department or prosecutor’s office itself does, and the insurer often has much more of an incentive as well as the capability to use those data effectively.
On the defendant’s side, the prospect of an exoneration payment may provide the incentive and the ability to continue to fight a conviction. The resource constraint most defendants and public interest groups face is a huge impediment to fighting wrongful convictions. The prospect of exoneration payments may allow the innocent to induce more parties to examine their cases in the hopes of earning both their freedom and a share of the payment. Essentially, the payments would allow for a criminal law analogue to a plaintiff’s contingency fee arrangement with his lawyer in tort law. As things stand now, individuals fighting their convictions must rely on the goodwill of public interest lawyers and groups like the Innocence Project, or pursue freedom on their own with no training and few resources.
Conclusion
Currently, 18 states have no compensation statute of any form, leaving little prospect for an exonerated individual to receive a payment when he is proved innocent. In such states, the only path to compensation would involve a private tort claim. In most cases, the claim fails because the relevant defendants (namely prosecutors) enjoy immunity to such suits. Of the other states, many take the approach of providing compensation only through private bills, which are essentially political petitions to a state legislature to provide compensation. Either of these options, tort claims or private bills, is so unlikely that it is doubtful they could affect anyone’s decisionmaking in the plea process. Nor is it likely they could provide the impetus for anyone to seek out the evidence necessary to prove someone’s innocence.
To harness the hypothesized incentive effects, damages for a mistaken conviction would need to be more reliable than low-probability tort awards and idiosyncratic private bills. Even many of the states with more robust statutory compensation frameworks might benefit from higher award amounts to get the benefit of these incentive effects. Compensation for mistaken convictions is an area where efficiency and fairness happily coincide.
