The Distortionary Effect of Evidence on Primary Behavior
In this Essay, we analyze how evidentiary concerns dominate actors’ behavior. Our findings offer an important refinement to the conventional wisdom in law and economics literature, which assumes that legal rules can always be fashioned to achieve socially optimal outcomes. We show that evidentiary motivations will often lead actors to engage in socially suboptimal behavior when doing so is likely to increase their likelihood of prevailing in court. Because adjudicators must base decisions on observable and verifiable information—or, in short, evidence—rational actors will always strive to generate evidence that can later be presented in court and increase their chances of winning the case regardless of the cost they impose on third parties and society at large. Accordingly, doctors and medical institutions will often refer patients to undertake unnecessary and even harmful examinations just to create a record that they went beyond the call of duty in treating them. Owners of land and intellectual property may let harmful activities continue much longer than necessary just to gather stronger evidence concerning the harm they suffer. And even the police will often choose to allow offenders to carry out crimes in order to improve the chance of a conviction. The effect we identify is pervasive. It can be found in virtually all areas of the law. Furthermore, there is no easy way to eliminate or correct it. It should be noted, however, that the evidentiary phenomenon we discuss also has a positive side effect: it reduces adjudication costs for judges and juries and improves the accuracy of court processes. In some cases, this improvement will exceed the social cost stemming from actors’ suboptimal behavior. In other contexts, however, the social cost will far outweigh the benefit.
The Distortionary Effect of Evidence on Primary Behavior, 124 Harvard Law Review 518-548 (2010) (with Gideon Parchomovsky) download
The Flawed Probabilistic Foundation of Law & Economics
This Article challenges the mathematical probability system that underlies law and economics and behavioral analysis and argues that many of the core insights of both approaches are irremediably flawed. The Article demonstrates that mathematical probability is only suitable for pure gambles and hence does not provide a useful epistemic tool for analyzing individual decisionmaking. As a result, mathematical probability cannot serve as a useful tool for lawmakers. Mathematical probability, the Article proposes, ought to be replaced with causative probability—a system of reasoning compatible with the causal structure of people’s physical, social and legal environments. Originating from the writings of John Stuart Mill and Francis Bacon, causative probability differs from its mathematical cousin both conceptually and substantively. By contrast to the mathematical system that bases probability estimates on abstract averages, the causative system bases probability estimates upon case-specific evidential variety. Under the causative system, the probability that a person’s action will bring about a particular consequence—gain or loss—is determined by the number and scope of the consequence’s evidential confirmations in the individual case, and not by general averages that are usually irrelevant to the individual determination at hand. Causative probability allows a person to develop a better epistemic grasp of her individual case relative to what she could achieve under the mathematical system. This epistemological advantage turns causative probability into a superior tool for understanding how legal mechanisms work, for improving those mechanisms, and for defining the rationality of individuals’ decisions.
The Flawed Probabilistic Foundation of Law & Economics, 105 Northwestern University Law Review 199-260 (2011) download
Strategic Enforcement
Doctrine and scholarship recognize two basic models of enforcing the law: the comprehensive model, under which law-enforcers try to apprehend and punish every violator within the bounds of feasibility; and the randomized model, under which law enforcers economize their efforts by apprehending a small number of violators and heightening their penalties so as to make violations unattractive. This Article supplements this list of options by developing a strategic model of law enforcement. Under this model, law enforcers concentrate their effort on the worst, or most rampant, violators at a given point in time while leaving all others unpunished. This enforcement strategy will force violators into a cascaded retreat: to avoid detection as one of the worst violators, every individual wrongdoer will bring the level of his unlawful activity down to the point of inconspicuousness—a process that will repeat itself several times to society’s benefit. This Article identifies the circumstances that call for the strategic model’s adoption and illustrates the model’s potential as an enforcement tool in diverse areas of the law that include employment discrimination, election districting, and copyright protection.
Strategic Enforcement, 95 Minnesota Law Review 9-58 (2010) (with Margaret H. Lemos) download
ORIGINALITY
In this Essay we introduce a model of copyright law that calibrates authors’ rights and liabilities to the level of originality in their works. We advocate this model as a substitute for the extant regime that unjustly and inefficiently grants equal protection to all works satisfying the “modicum of creativity” standard. Under our model, highly original works will receive enhanced protection and their authors will also be sheltered from suits by owners of preexisting works. Conversely, authors of less original works will receive diminished protection and incur greater exposure to copyright liability. We operationalize this proposal by designing separate rules for highly original works, for works exhibiting average originality, and for works that are minimally original or unoriginal. We illustrate our rules’ application by showing how they could have altered court decisions in classic copyright cases in a socially beneficial way.
Originality, 95 Virginia Law Review 1505-1550 (2009) (with Gideon Parchomovsky) download
RECONCEPTUALIZING TRESPASS
This Essay addresses an anomaly in trespass law. Trespass law is generally understood as the paradigmatic example of property-rule protection: an owner can obtain an injunction against the trespasser and have him removed from her land. The property-rule protection enjoyed by the owner protects her right to exclude others and to set the price for the use of her property. However, the property-rule protection only exists ex ante: it avails only against imminent or ongoing trespasses. Ex post, after a trespass ends, the owner can only recover compensation measured by the market value of the unauthorized use, i.e., the going rent. This liability-rule compensation dilutes the ex ante property-rule protection of ownership. Effectively, it grants trespassers a call option on others’ property, creating a mismatch between rights and remedies.
To remedy this mismatch, we introduce the concept of “propertized compensation”—a damage measure that sets compensation equal to the owner’s pre-trespass asking price. We contend that propertized compensation should become the primary remedial option in trespass cases. The use of this measure will reinstate the owner’s position as a price maker, entitling her to recover the amount that she would have agreed to accept ex ante in a voluntary exchange. We further argue that owners who cannot produce evidence regarding their pre-trespass asking price (as well as owners who prefer not to seek propertized compensation) should be entitled to seek disgorgement of the trespasser’s profits. Finally, we claim, contra the extant regime, that market-value compensation should only be used in the exceptional cases of trespass by necessity, media trespass, and good faith encroachments. In all other cases, it should only be awarded if the owners specifically ask for it.
Reconceptualizing Trespass, 103 Northwestern University Law Review 1823-1863 (2009) (with Gideon Parchomovsky) download
TORTS AND INNOVATION
This Essay exposes and analyzes a hitherto overlooked cost of the current design of tort law: its adverse effect on innovation. Tort liability for negligence, defective products, and medical malpractice is determined by reference to custom. We demonstrate that courts’ reliance on custom and conventional technologies as the benchmark of liability chills innovation and distorts its path. Specifically, the recourse to custom taxes innovators and subsidizes replicators of conventional technologies. We explore the causes and consequences of this phenomenon and propose two possible ways to modify tort law in order to make it more welcoming of innovation.
Torts and Innovation, 107 Michigan Law Review 285-315 (2008) (with Gideon Parchomovsky) download
THE ANTI-POOLING JUSTIFICATION OF THE FIFTH AMENDMENT PRIVILEGE AGAINST SELF-INCRIMINATION (THE RIGHT TO SILENCE)
The right to silence has a solid consequentialist justification. The conventional perception of this right -- that it impedes the search for truth and thus helps only criminals -- is mistaken. The right to silence helps triers of fact to distinguish between innocent and guilty defendants. A guilty suspect's self-interested response to questioning can impose externalities, in the form of wrongful conviction, on innocent suspects and defendants who tell the truth but cannot corroborate their stories. Absent the right to silence, guilty suspects and defendants would make false exculpatory statements if they believed that their lies were unlikely to be exposed. Aware of these incentives, triers of fact would rationally discount the probative value of uncorroborated exculpatory statements at the expense of innocent defendants who could not corroborate their true exculpatory statements. Because the right to silence is available, innocent defendants tell the truth while guilty defendants rationally exercise the right when they fear that lying is exceedingly risky. Thus, guilty defendants do not pool with innocents by lying; and as a result, triers of fact do not wrongfully convict innocent defendants.
The Right to Silence Helps the Innocent: A Game-Theoretic Analysis of the Fifth Amendment Privilege, 114 Harvard Law Review 430 (2000) (with Daniel Seidmann) download
For my response to the critics of this article, click HERE
THE OVERENFORCEMENT PARADIGM
Overenforcement of the law occurs when the total sanction suffered by the violator of a legal rule exceeds the amount optimal for deterrence. Overenforcement sometimes generates overdeterrence that cannot be remedied through the adjustment of substantive liability standards or penalties in light of operational and expressive constraints. When that happens, the legal system can counteract the effects of overenforcement by adjusting evidentiary or procedural rules to make liability less likely (ex ante). This framework -- the overenforcement paradigm -- illuminates previously unnoticed features of various evidentiary and procedural arrangements. It also provides a useful analytical and prescriptive tool for creating balanced incentives in cases in which overenforcement is present.
Overenforcement, 93 Georgetown Law Journal 1743 (2005) (with Richard A. Bierschbach) download
MEDIATING RULES IN CRIMINAL LAW
This article challenges the conventional divide between substantive criminal law theory, on the one hand, and evidence law, on the other, by exposing an important and unrecognized function of evidence rules in criminal law. Throughout the criminal law, special rules of evidence work to mediate conflicts between criminal law’s deterrence and retributivist goals. They do this by skewing errors in the actual application of the substantive criminal law to favor whichever theory has been disfavored by the substantive rule itself. The mediating potential of evidentiary rules is particularly strong in criminal law because the substantive law’s dominant animating theories – deterrence and retributivism – respond asymmetrically to the workings of those rules. We analyze the features of “mediating rules,” explore their effects across a range of substantive areas, and offer a tentative normative assessment of their role in the criminal law system.
Mediating Rules in Criminal Law, 93 Virginia Law Review 1197 (2007) (with Richard A. Bierschbach) download
ASYMMETRIC AMBIGUITY AVERSION FOILS CRIMINAL JUSTICE
Ambiguity aversion is a person's rational attitude towards probability's indeterminacy. When a person is averse towards such ambiguities, he increases the probability of the unfavorable outcome to reflect that fear. This observation is particularly true about a criminal defendant who faces a jury trial. Neither the defendant nor the prosecution knows whether the jury will convict the defendant. Their best estimation relies on a highly generalized probability that attaches to a broad category of similar cases. The prosecution, as a repeat player, is predominantly interested in the conviction rate that it achieves over a long series of cases. It therefore can depend on this general probability as an adequate predictor of this rate. The defendant only cares about his individual case and cannot depend on this general probability. From the defendant's perspective, his individual probability of conviction is ambiguous. The defendant consequently increases this probability to reflect his fear of that ambiguity. Because most defendants are ambiguity-averse, while the prosecution is not, the criminal process systematically involves and is thoroughly affected by asymmetric ambiguity-aversion.
Asymmetric ambiguity-aversion foils criminal justice. The prosecution can exploit it by forcing defendants into plea bargains that are both inefficient and unfair. Because plea bargain is a predominant method of case-disposition across the United States, this exploitation opportunity is particularly pernicious. The legal system ought to eliminate it.
Two Fifth Amendment doctrines -- the rule against double jeopardy and the grand jury review of indictments -- have the effect of mitigating this problem. The rule against double jeopardy sets a pro-defendant system of asymmetric rights to appeal. This system reduces the probability of conviction for all defendants, regardless of the merits. This probability reduction offsets -- but not eliminates -- the upward adjustment that an ambiguity-averse defendant introduces into his probability of conviction. Grand jury review disambiguates the defendant's probability of conviction when he is informed about the grand jurors' voting score. This disambiguation is only partial, though, because grand jurors are authorized to indict upon mere showing of a "probable cause."
The prevalent constitutional doctrine should therefore be modified by giving a defendant the right to choose between a bench trial and a trial by jury. Judges are repeat institutional players that credibly commit themselves to reasons for decisions that are evenhanded, known and institutionally approved. This commitment is induced not only by the judges' fear of reversal and other career-related repercussions, but also by the defendant's constitutional entitlement to a trial by jury. For judges, jury trial is a time-consuming and effort-intensive process with virtually no career-enhancing returns. Judges therefore strongly prefer a bench trial over a trial by jury. To actualize this preference, judges need systematically to deliver evenhanded decisions that follow the institutionally approved reasons. This makes judges' decisions predictable. The defendant's probability of being convicted by a judge in a bench trial thus becomes unambiguous, which neutralizes the prosecution's ambiguity-exploiting pressure in plea bargaining.
Empirical data confirm these findings. Specifically, they identify three major trends. First, bench trials are prevalent in jurisdictions featuring high trial rates, generated by a non-meticulous selection of cases for prosecution. Second, the rate of acquittals in bench trials is much higher than in trials by jury. The defendants' ambiguity-aversion is the most plausible explanation of these trends. Defendants with real prospects for acquittal have much to lose and are therefore unwilling to depend upon unpredictable juries. Finally, there is a demand for jury-consulting services and no discernible market for judge-consulting services. Litigants are willing to pay for information predicting the outcomes of jury trials and are generally unwilling to pay for information predicting judges' decisions in bench trials. This leads to the conclusion that ambiguity aversion is particularly problematic in trials by jury.
Ambiguity Aversion and the Criminal Process, 81 Notre Dame Law Review 1495 (2006) (with Uzi Segal) download
LIABILITY FOR LOST CHANCES: WHAT COMPENSATION IS DUE?
DISTINGUISHING BETWEEN THE (EX ANTE) RISK OF INJURY AND THE (EX POST) PROBABILITY OF CAUSATION
The following example is paradigmatic:
The claimant required urgent surgery, which -- if performed properly and on time -- would have given her a 75% chance of recovery. The doctors negligently delayed the surgery. The delayed surgery was performed impeccably, but it promised the claimant only a 25% chance of recovery. Ultimately, the claimant did not recover.
The claimant cannot prove causation and attempts to recover compensation under the lost-chances doctrine. Under this doctrine, courts uniformly award the claimant 50% of her total damage.
This approach is wrong: the claimant should recover 2/3 of her damage, not just 50%.
See, for example, Mays v. United States, 608 F. Supp. 1476 (D.C. Colo. 1985) (upon finding that medical malpractice reduced the patient's chances of recovery from 40% to 15%, the court reasoned that the damage was 25% of the patient's total loss); Herskovits v. Group Health Cooperative of Puget Sound 664 P.2d 474 (Wash. 1983) (holding a 14% reduction, from 39% to 25%, in the decedent's chance for survival as sufficient evidence to allow the case to go to the jury); Alberts v. Schultz, 975 P.2d 1279, 1287 (N.M. 1999) (holding that if medical malpractice reduced the patient's chance of survival from 50% to 20%, that patient's compensation would be 30% of the value of his or her life); Jorgenson v. Vener, 616 N.W.2d 366, 372 (S.D. 2000) (if instead of completely eliminating the chance of recovery, the physician's negligence merely reduced the chance of recovery from 40% to 20%, then the value of the lost chance would be 20% of the value of a complete recovery).
For reasons provided below, the claimant should have recovered 29% of the damage in Mays; 19% of the damage in Herskovits; 37.5% of the damage in the Alberts example; and 25% of the damage in the Jorgenson example.
Take a person who sustains injury after being wrongfully exposed to a risk of sustaining that injury. Before the wrongdoing, this victim's probability of sustaining the injury equaled 1-p, which is parallel to her probability of remaining uninjured (p). After the wrongdoing, the victim's probability of sustaining the injury became 1-q, which is parallel to her probability of escaping the injury (q). Because the victim actually sustained the injury, her case falls into the 1-q category. This statistical category comprises two jointly exhaustive and mutually exclusive scenarios that reflect the victim's initial position. In the first scenario, the victim sustains the injury irrespective of the wrongdoing. Under this scenario, the victim was doomed to sustain the injury, so that the wrongdoing made no impact on her well-being. As already indicated, the probability of that scenario equals 1-p. In the second scenario, it is the wrongdoing that causes the victim's injury. Under this scenario, the victim would have remained uninjured had she not been exposed to the wrongdoing. The probability of this scenario equals (1-q)-(1-p), that is, p-q. This ex ante probability represents the reduction in the victim's chances of remaining uninjured, as effected by the wrongdoing.
Now consider the ex post probability of the scenario that the wrongdoing was the actual cause of the victim's injury. This probability is represented by the fraction of scenarios featuring a victim who could not sustain her injury without being subjected to a wrongdoing in the more general cluster of cases that feature an injured victim, a wrongdoing, and the exhaustive variety of causal factors that could inflict the same injury on the victim. The above fraction of scenarios equals p-q. The cluster of cases covering all possible scenarios equals 1-q. The ex post probability of the scenario in which the wrongdoing actually inflicts the victim's injury therefore equals (p-q)/(1-q).
As already mentioned, the victim's (ex ante) risk of sustaining injury as a result of the wrongdoing equals p-q. Consequently, in cases in which the victim actually sustains injury, the (ex post) probability of causation -- that is, the probability of the allegation that factually attributes the injury to the defendant's wrongdoing -- would generally be higher than the (ex ante) risk of injury. This would be so because, on numerous occasions, a wrongdoing increases the victim's probability of becoming injured without transforming this prospect into empirical reality. In any such case, since the wrongdoing still leaves the victim with chances of escaping the injury, 0<q<1. Hence, (p-q)/(1-q)>p-q. The two probabilities would be equal only when q=0, that is, when the wrongdoing totally eliminates the victim's chances of escaping the injury. In q=0 cases, the risk of injury and the probability of causation would concur and would equal p.
The courts' approach is detrimental to society.
Using the same notation, let p and q denote, respectively, the victim's chances of remaining uninjured before and after the wrongdoing. Allow D to denote the average amount of damage that the wrongdoing inflicts in the long run of cases, and let T denote the total number of cases in which the risky activity takes place. The ideal compensation that the legal system should exact from the wrongdoer would thus equal (p-q)DT.
In reality, however, only injured victims can successfully sue the wrongdoer.
Therefore, the number of cases in which the wrongdoer would have to pay compensation would equal (1-q)T. The wrongdoer's compensation duty would thus be below the optimal. Using the probability of causation as an award-multiplier would eliminate this shortfall. As already established, the probability of causation equals (p-q)/(1-q). The total amount of the wrongdoer's compensation duty would consequently be [(p-q)/(1-q)]DT(1-q), that is: (p-q)DT.
This compensation duty equals the losses inflicted by the wrongdoer. It would therefore optimally deter prospective wrongdoers (and would also promote corrective justice).
Indeterminate Causation and Apportionment of Damages, 23 Oxford Journal of Legal Studies 667 (2003) (with Ariel Porat) download