TO THINE OWN SELF BE TRUE? INCENTIVE PROBLEMS IN PERSONALIZED LAW.

• Author: Barry, Jordan M.

TABLE OF CONTENTS INTRODUCTION 726 I. PERSONALIZED LAW 732 II. INDIVIDUAL RESPONSES TO PERSONALIZED LAW 739 A. Elicitation 739 B. Signaling 743 1. Counterargument: High Individual Cost of False Signaling 750 2. Counterargument: Impenetrability of Personalized Law 753 C. Moral Hazard 757 III. REGULATORY RESPONSES AND THEIR LIMITS 760 A. Information Problems 1: Muddling 761 B. Information Problems 2: Privacy 765 C. Regulatory Authority 771 D. Transparency 778 E. Regulatory Conservatism 787 CONCLUSION 789 INTRODUCTION

Imagine a new technology that lets us measure each individual's health perfectly, immediately, and at no cost. Such a technology would have tremendous potential value; it would enable society to allocate and distribute medical care in the best possible way. We could perfectly take into account how sick people are, what benefits they stand to gain from treatment, and any other factors society deems relevant when deciding what medical care everyone should receive and when they should receive it. We could update our medical care distribution plan in response to new developments--for example, if someone suddenly takes a turn for the worse, we could move her up the queue and treat her earlier. Our new technology would let us maximize the potential gains from medical care in a way that simply was not possible before its invention. (1)

This vision sounds Utopian. And all is well in Utopia--until Bob, who has a bad liver, realizes that he can move his way up the liver transplant queue by drinking a bottle of champagne and making his liver worse. This self-destructive behavior may help Bob if moving up the queue advantages Bob more than becoming sicker hurts him. Meanwhile, Bob's actions make everyone else in the queue unambiguously worse off. The Bobs of the world can wreak havoc on our carefully designed healthcare system--and make care more expensive to provide, as beneficiaries become sicker than they otherwise would be. Thus, paradoxically, precisely targeting benefits can sometimes produce worse outcomes than allocating them in a simpler, more straightforward way (for example, first come, first served). (2)

The Utopian care queue that drove Bob to drink exemplifies "personalized law." Personalized law is a rapidly growing phenomenon in legal thought, the subject of a bevy of articles (3) and even a symposium in the Chicago Law Review. (4) The basic idea underlying personalized law is that regulators can use sophisticated analytical techniques, large amounts of data, and powerful computers to draw exacting distinctions between similar (but not identical) circumstances and assign them different legal outcomes." Individualized decision rules can be instantaneously distributed to regulated parties via smartphones, smart cars, and other "smart" devices. (6) Thus, the law can weigh all the important facts, instantly analyze what actions should be permitted, and tell us the results in a clear, easy-to-understand way: personalized law "will provide ex ante behavioral prescriptions finely tailored to every possible scenario," thereby achieving "all of the benefits of both rules and standards without the costs of either.'" At the same time, it will "mitigate economic and legal inequality" (8) and provide "ultimate transparency." (9) Scholars have asserted that personalized law "will only increase" over time; (10) indeed, they have repeatedly dubbed it "the wave of the future." (11)

The personalization revolution is already well under way in business. (12) Companies now have unprecedented levels of information about their customers, including where they go, what they view online, what they purchase, and whom they communicate with. (13) They use this information to finely tailor their services and products to customers' individual preferences. (14) Many of the new economy's biggest successes--Google, Amazon, Facebook, Uber, and others--are built on using this kind of information to provide consumers with exactly what they want. (10)

But there are important differences between matching consumers to products and matching facts to legal outcomes. In particular, customers want to be directed to their preferred products, so they generally do not resist such personalization efforts. (16) But in many circumstances, the legal outcome that policymakers prefer will not be the outcome that the regulated party desires. Accordingly, regulated parties may react to personalized laws in ways that--from the perspective of policymakers--are undesirable. These responses by regulated parties can limit the effectiveness of personalization efforts, or even thwart them outright.

Businesses have already encountered these dynamics when personalization is against customers' interests. For instance, companies have often attempted to use big data to engage in price discrimination. (17) If Uber can tell how much individual consumers are willing to pay for rides, it can offer lower prices to people who are strongly price-sensitive and higher prices to those who are not. (18) This means that consumers have powerful incentives to convince Uber that they are very price-sensitive--even if they are not--so that Uber will offer them lower prices. (19) For example, consumers could signal price sensitivity to Uber's algorithm by checking prices on their usual routes and then declining to request a ride. (20) Amazon and other retailers have also tried similar price discrimination strategies and seen similar responses. (21) These consumer responses have made it harder for companies to tell who is truly price sensitive and thus have limited the efficacy of personalized pricing strategies. (22)

The potential of personalized law is likewise bounded. Regulated parties will react to personalized laws by changing their behavior; they will alter or disguise their circumstances in order to improve their regulatory treatment. (23) They will respond to a given regulatory regime in unanticipated and undesirable ways; in some instances, personalized law may produce affirmatively worse outcomes than a depersonalized legal regime. (24) These behaviors will constrain the law's ability to draw fine distinctions between circumstances--in other words, how personalized the law can actually be in practice--and therefore its ability to achieve optimal outcomes. (25)

More precisely, we should expect rational actors to change their behavior whenever they have an incentive to do so. Thus, a personalized legal regime that perfectly matches circumstances to legal outcomes will only function as designed if none of the parties that it governs has an incentive to change its behavior. In economics terms, the personalized legal regime must be "incentive compatible." (26)

It is exceedingly difficult to make personalized law incentive-compatible. That difficulty is further compounded by several significant constraints on regulators' ability to personalize law and to respond to perceived gamesmanship by regulated parties.

The first such constraint is informational. Yes, new technologies make it easier to gather and analyze data; however, that does not mean that regulators will have all of the information that they might want. (27) For instance, regulators may be prohibited from collecting or considering certain types of information due to privacy or other concerns. (28) Even when regulators have data, regulated parties may intentionally muddle it as a means of thwarting regulators' efforts. (29)

Regulators' behavior is also constrained by the scope of their authority. (30) Certain types of perceived misbehavior may fall outside of a regulator's prescribed jurisdiction. (31) Regulators may have limited ways of penalizing misconduct. (32) And even when regulators have all the legal authority they need, they may lack the political will to use it. (33) These issues feed into each other; for example, an aggressive regulator may trigger a political backlash that reduces its jurisdiction or its ability to impose penalties.

Democratic oversight requires that regulators maintain a certain amount of transparency, which creates additional complications. Transparency can serve as a tonic for concerns about a regulator's power; at the same time, greater transparency may facilitate regulated parties' attempts to manipulate personalized laws. (34)

All of these issues interact and overlap in complicated ways that will fluctuate over time. These dynamics may create substantial uncertainty for regulators regarding which actions will earn them plaudits and which will result in a political backlash. (30) That concern will sit atop the uncertainty that regulators face regarding how regulated parties will react to personalized laws. (36) These uncertainties will likely encourage regulators to be conservative in their efforts to personalize law and respond to gamesmanship. (37)

These constraints on regulators, combined with regulated parties' responses to personalized law, will reduce the law's ability to optimally match circumstances to outcomes. We do not mean to be overly dour; personalized law will likely be a valuable tool in the regulatory tool kit, and it may represent a major step forward in policy making. But proponents of personalized law should curb their enthusiasm: the strictures of incentive compatibility will generally prevent personalized law from producing the Utopian outcomes that some envision. Personalized law may be "the future of law"; (38) that does not make it a panacea.

This Article proceeds as follows. Part I begins with the relevant background on personalized law. Part II explains how personalized law will lead to personalized avoidance responses in the form of false elicitation, false signaling, and moral hazard and illustrates these phenomena through a combination of both stylized and real-world examples. Part III turns to the problems that real-world regulators will face ex ante when shaping the law to deal with these anticipated avoidance problems and when...