Restitution Archives - Columbia Law Review

Introduction Suppose you work at the bank. As part of your job, you process payment orders from top clients. One day, you make a tiny mistake. You check the wrong box. But a small mistake can have dire consequences: Instead of paying the client’s interest payment on a loan, a payment of one million dollars, you transferred the principal amount of the loan, thereby making a payment of nearly one billion dollars from the bank’s money. The next day, when the mistake is detected, is not a happy day at the bank. But no worries. You send an email to the recipients of the money, alerting them of the mistake and asking them to wire it all back. Nobody is too stressed or angry: Everybody at the bank knows that, despite heavy investments in security, such mistakes sometimes happen. But here is the twist: After you contact the recipients and explain the situation, they refuse to return the money. This is irregular. Tension at work starts to rise. People are yelling. You fear for your job. There is still hope, of course. The bank sues the recipients; surely, the court will make them return the money. Some months later, when the court decides that the recipients of the mistaken payment have a right to keep it and that your mistake is irreversible, you finally lose your faith in humanity. Unfortunately, this rather bizarre scenario is not fictional. It is based on a recent New York District Court decision.[footnote]In re Citibank August 11, 2020 Wire Transfers, 520 F. Supp. 3d 390 (S.D.N.Y. 2021).[/footnote] In this high-profile case, Citibank transferred almost a billion dollars by mistake, but restitution was denied under the discharge for value doctrine.[footnote]Id. at 431–52; Banque Worms v. BankAmerica Int’l, 570 N.E.2d 189, 196 (N.Y. 1991) (“When a beneficiary receives money to which it is entitled and has no knowledge that the money was erroneously wired, the beneficiary . . . should be able to consider the transfer of funds as a final and complete transaction, not subject to revocation.”); Restatement (Third) of Restitution and Unjust Enrichment § 67(1) (Am. L. Inst. 2011) (“A payee without notice takes payment free of a restitution claim to which it would otherwise be subject . . . .”).[/footnote] Citibank’s appeal is now pending.[footnote]See Notice of Appeal, Citibank , 520 F. Supp. 3d 390, appeal filed, No. 21-487 (2d Cir. 2021).[/footnote] The Citibank decision was widely criticized in the general media[footnote]See Andrew Scurria, Alexander Gladstone & Becky Yerak, Judge Lets Revlon Lenders Keep Citi’s Botched $500 Million Payment, Wall St. J., https://www.wsj.com/articles/judge-lets-revlon-lenders-keep-citis-botched-500-million-payment-11613490508 (on file with the Columbia Law Review ) (last updated Feb. 16, 2021); Andrew Ross Sorkin, Jason Karaian, Michael J. de la Mered & Ephrat Livni, Ouch, That Hurts: Citi Committed “One of the Biggest Blunders in Banking History,” a Federal Judge Says., N.Y. Times (Feb. 17, 2021), https://www.nytimes.com/2021/02/17/business/dealbook/citigroup-900-million.html (on file with the Columbia Law Review ); Ramishah Maruf, Citibank Can’t Get Back $500 Million It Wired by Mistake, Judge Rules, CNN Bus. (Feb. 17, 2021), https://www.cnn.com/2021/02/16/business/citibank-revlon-lawsuit-ruling/index.html [https://perma.cc/BQQ4-SAQ7]; Jonathan Ponciano, Citi Can’t Get Back $500 Million It Accidentally Wired to Revlon Lenders, Federal Judge Rules, Forbes (Feb. 16, 2021), https://www.forbes.com/sites/jonathanponciano/2021/02/16/citi-cant-get-back-500-million-it-accidentally-wired-to-revlon-lenders-federal-judge-rules/?sh=181381185de2 (on file with the Columbia Law Review ).[/footnote] and in business circles.[footnote]See Matt Levine, Matt Levine’s Money Stuff: Citi Won’t Misplace $500M Again, Bloomberg L. (Mar. 3, 2021), https://news.bloomberglaw.com/banking-law/matt-levines-money-stuff-citi-wont-misplace-500-million-agai (on file with the Columbia Law Review ).[/footnote] After all, the bank transferred the money by mistake and attempted to fix this mistake the very next day. What justification can there be for the recipients to retain the money? And why should someone be made to lose hundreds of millions of dollars due to a technical error that did not harm anyone? Matt Levine, a former mergers and acquisitions lawyer and Goldman Sachs investment banker and current author of the popular Wall Street newsletter Money Stuff , had the following to say of the decision: There is no earthly reason that those funds should be able to keep the money, except that there happens to be a weird doctrine of New York law that lets them keep it. As a former lawyer I am tempted to say, sure, fine, whatever, counterintuitive old doctrines are what make law school fun and keep lawyers employed. “Citi just sent us money by mistake, do we have to give it back,” the hedge fund analyst asks, and instead of saying “of course duh we live in society,” the portfolio manager replies “hang on, let me consult with a lawyer,” and the lawyer says “hang on, let me consult with my firm’s specialist in Finders Keepers Law,” and the Finders Keepers specialist consults some dusty old tomes of arcane lore and says “lemme tell you about the doctrine of discharge for value.” And she bills the hedge fund $2,000 an hour and is absolutely worth it .[footnote]Id.[/footnote] Despite the dismissive and comical tone, Levine’s critique is spot on. The outcome of the Citibank case seems nonsensical—legal “transcendental nonsense”[footnote]Felix S. Cohen, Transcendental Nonsense and the Functional Approach, 35 Colum. L. Rev. 809, 809–10 (1935) (famously describing doctrinal elements as “transcendental nonsense”).[/footnote] originating, as Levine so aptly puts it, from the “old tomes of arcane lore” of “Finders Keepers Law.”[footnote]evine, supra note 5.[/footnote] Jokes aside, this is indeed a gloomy picture. In fact, we, as avid “Finders Keepers specialists,” find the impression left by the Citibank decision deeply disturbing. “Finders Keepers Law”—or as it should be called, restitution law[footnote]The law of restitution is also referred to as the law of unjust enrichment. See Douglas Laycock, The Scope and Significance of Restitution, 67 Tex. L. Rev. 1277, 1277 (1989); see also Ward Farnsworth, Restitution: Civil Liability for Unjust Enrichment 1–2 (2014).[/footnote]—should make sense. If the law is nonsensical, we are down a dangerous road. Unfortunately, the Citibank decision is symptomatic of a larger problem. The law of restitution typically does make sense, yet the Citibank decision reflects a prevalent contemporary assumption, according to which the law of restitution is somehow detached from real-life pragmatic considerations and has no broad policy implications. Andrew Kull, the reporter for the 2011 ALI Restatement of Restitution and Unjust Enrichment,[footnote]Restatement (Third) of Restitution and Unjust Enrichment (Am. L. Inst. 2011).[/footnote] writes the following: “[I]n the law of restitution it is difficult to identify rules that have even the ordinary instrumental dimension that we tend to look for in private law, let alone any broader social implications.”[footnote]Andrew Kull, Restitution and Reform, 32 S. Ill. U. L.J. 83, 87 (2007).[/footnote] With such an insular attitude to the law of restitution, it is not surprising to see restitution cases commonly decided as matter of arcane law, in a highly technical manner, and without any reference to broader policy goals and real-life implications. The Citibank decision is a stark example of this. We argue that this approach to the law of restitution is counterproductive. It clearly misses an important aspect of the law. Different restitutionary rules have immediate policy implications. This is manifestly clear in the Citibank decision: After Citibank was denied restitution and lost hundreds of millions of dollars, banks and other financial institutions will make efforts to avoid a similar fate. They will increase their investment in precautions to prevent mistakes, insure against mistakes, or try to contract around the Citibank ruling.[footnote]Levine, supra note 5 (describing Citibank’s new policy of adding a “Revlon clawback” to credit agreements requiring the return of money received by mistake).[/footnote] These are all costs that will make the operation of the payment system more expensive and eventually be borne by all of us, as the users of this system. More broadly, the law—and restitution law is no exception—should make pragmatic sense.[footnote]See Hanoch Dagan, Restitution’s Realism, in Philosophical Foundations of the Law of Unjust Enrichment 54, 60–62 (Robert Chambers, Charles Mitchell & James Penner eds., 2009) (advocating for a contextualist approach to the law of restitution and arguing it should be applied and understood in relation to social goals and a rich normative environment).[/footnote] The law is a social instrument, and it should reflect social and practical goals and policies.[footnote]For the origins of this approach to adjudication, see generally Oliver Wendell Holmes, The Path of the Law (1897); Hanoch Dagan, The Realist Conception of Law, 57 U. Toronto L.J. 607 (2007); Joseph William Singer, Legal Realism Now, 76 Calif. L. Rev. 465 (1988); see also Richard A. Posner, The Economics of Justice 74 (1983); Duncan Kennedy, Form and Substance in Private Law Adjudication, 89 Harv. L. Rev. 1685 (1976) (“There seems no basis for disputing . . . the idea that . . . choice [of law] will have wide-ranging practical consequences.”).[/footnote] If the law of restitution is developed without reference to such goals and policies, it will quickly and surely lose its touch with reality. We therefore offer a reorientation of current trends in the law of restitution by suggesting an approach that emphasizes the broader practical implications of restitution decisions, instead of ignoring them. We propose a novel normative framework of analysis that explores and delineates the relevant policy considerations that guide restitution law and explain how the main features of restitution doctrine reflect these considerations. This is an effort to revive the law of restitution and assure that its application by courts tracks the first-order rationales, policies, and goals that animate this area of law. As can be seen from the Citibank decision, such intervention is desperately needed. To establish our framework for the analysis of cases of mistake, we draw on Guido Calabresi’s seminal work, The Costs of Accidents: A Legal and Economic Analysis .[footnote]See Guido Calabresi, The Costs of Accidents: A Legal and Economic Analysis (1970) [hereinafter Calabresi, The Costs of Accidents].[/footnote] In this classic, Calabresi famously provides a statement of the social goals of accident law.[footnote]On the influence of Calabresi’s The Costs of Accidents , see Keith N. Hylton, Calabresi and the Intellectual History of Law and Economics, 64 Md. L. Rev. 85, 85 (2005) (“The book has had an enormous influence on the field. It would not be an exaggeration to say that modern law and economics, as we see it practiced today, had its start with Gary Becker’s article on crime and Guido Calabresi’s book, both products of the late 1960s.”(footnote omitted)); see also Yotam Kaplan, Economic Theory of Tort Law, in Research Handbook on Private Law Theory 270, 273–78 (Hanoch Dagan & Benjamin C. Zipursky eds., 2020).[/footnote] Calabresi explains that accidents involve two types of costs: direct costs in the form of the harm resulting from the accident ex post and indirect costs in the form of precautions designated to prevent accidents ex ante. Accident law acts to minimize both types of costs by assigning harm to the cheapest cost avoider.[footnote]Calabresi, The Costs of Accidents, supra note 15, at 143–44 (identifying the cheapest cost avoider as the party who is able to minimize the negative externalities of an accident most efficiently); see also Guido Calabresi, Concerning Cause and the Law of Torts: An Essay for Harry Kalven, Jr., 43 U. Chi. L. Rev. 69, 84–85 (1975) (defining the “cheapest cost avoider” as the person “who can best decide whether avoidance is cheaper than bearing th[e] costs” of injury). The concept of best decisionmakers was later articulated in a celebrated article by Guido Calabresi and Jon Hirschoff. See Guido Calabresi & Jon T. Hirschoff, Toward a Test for Strict Liability in Torts, 81 Yale L.J. 1055, 1060 (1972) (defining the best decisionmaker as the party with better access to information regarding the risks of harm and the costs of preventing it).[/footnote] Calabresi shows that if the party that was better positioned to prevent the accident is made to bear any resulting harms, then the overall costs of accidents are minimized.[footnote]Calabresi, The Costs of Accidents, supra note 15, at 26–31.[/footnote] This is because the cheapest cost avoider is incentivized to minimize the direct costs resulting from the accident and can do so at minimal investment in ex ante precautions.[footnote]Id.[/footnote] We suggest a parallel framework to guide adjudication in mistake cases: a “Costs of Mistakes” framework. The crux of our analysis is that mistakes are somewhat similar to accidents but involve a slightly different cost structure.[footnote]Compare Hanoch Dagan, Mistakes, 79 Tex. L. Rev. 1795, 1810 (2001) [hereinafter Dagan, Mistakes] (highlighting the similarities between mistakes in the law of restitution and accidents in the law of torts), with Maytal Gilboa & Yotam Kaplan, The Mistake About Mistakes: Rethinking Partial and Full Restitution, 26 Geo. Mason L. Rev. 427, 430–31 (2018) [hereinafter Gilboa & Kaplan, The Mistake About Mistakes] (explaining the structural differences between the law of restitution and the law of torts); see also Maytal Gilboa & Yotam Kaplan, Loser Takes All: Multiple Claimants & Probabilistic Restitution, 10 U.C. Irvine L. Rev. 907, 911 (2020) (explaining that since restitution law is the law of gains, “restitution doctrines are often structured as a mirror image of their tort law, harm-based counterparts”).[/footnote] Consider a mistaken payment such as the one in the Citibank case.[footnote]See In re Citibank August 11, 2020 Wire Transfers, 520 F. Supp. 3d 390 (S.D.N.Y. 2021).[/footnote] Like an accident, such a mistake involves indirect costs in the sense of precautionary measures. The bank, like any other payer, will invest ex ante to avoid making mistakes and transferring its money to strangers.[footnote]See id. at 401 (describing Citibank’s “six-eye” approval procedure of review and approval before a transaction was executed).[/footnote] Yet the mistake, unlike an accident, does not necessarily involve a direct cost or injury.[footnote]This point is often neglected, as scholars ignore the difference between mistakes and accidents. See, e.g., J. Beatson & W. Bishop, Mistaken Payments in the Law of Restitution, 36 U. Toronto L.J. 149, 153 (1986) (“[A]nalytically the economics of precaution against mistakes is virtually identical to the economics of accident avoidance.”).[/footnote] That is, if a mistake occurred, the payer may indeed lose a sum of money, but this loss is offset by a parallel gain to the recipient. The mistake is therefore not harmful in the sense that it leads to no net reduction in the overall amount of resources. Of course, some mistakes can be harmful, and in this sense more closely resemble accidents, but mistakes are not necessarily harmful in the sense of causing direct costs. When mistakes cause no direct costs all the law should do is minimize the indirect costs of mistakes, that is, the costs of precautions designed to prevent them. The suggested Costs of Mistakes framework offers normative guidelines for the adjudication of mistake cases. Thus, in the Citibank case, the bank can invest in ex ante precautions to prevent mistakes. But how are those investments affected by the availability of restitution in cases of mistake? Consider, first, the possibility that in case of a mistake, restitution is not available, and the recipient is allowed to retain the mistakenly transferred sum. Under such a regime, investment by the bank to prevent the mistake will be high, as the bank knows that this sum is irrevocably lost in case of a mistake. Second, consider the possibility that restitution is available following a mistake. In this case, the bank will reduce its investment in preventing mistakes, since it does not stand to lose much if a mistake occurs. This analysis leads to a somewhat counterintuitive result. In the traditional framework of The Costs of Accidents , the goal is to find the party who caused the accident and make sure that that party bears the harm of the accident.[footnote]Calabresi, The Costs of Accidents, supra note 15, at 143–44.[/footnote] This will minimize both direct and indirect costs. Conversely, when dealing with a mistake that causes no direct harm (such as a typical mistaken money payment), the goal should again be to find the party that caused the mistake, but this time protect that party rather than deter it from making the mistake. That is, to minimize costs, the law of restitution should make sure the mistake generates no costs for the party responsible for creating it. This will give the mistaken party less incentive to prevent the mistake ex ante, thereby minimizing the indirect costs of mistakes or the investments designed to prevent them.[footnote]Gilboa & Kaplan, The Mistake About Mistakes, supra note 20, at 437 (showing that the availability of restitution for mistakes allows payers to lower investments in wasteful precautions). Other scholars have suggested a different rationale for the law of restitution, that of encouraging positive externalities. See Robert Cooter & Ariel Porat, Torts and Restitution: Legal Divergence and Economic Convergence, 92 S. Cal. L. Rev. 897, 900 (2019); Ariel Porat, Private Production of Public Goods: Liability for Unrequested Benefits, 108 Mich. L. Rev. 189, 191 (2009).[/footnote] Since the mistake causes no direct harms, all the law should do is minimize the indirect costs of the mistake, by making sure the mistaken party has no reason to invest in preventing it. For cases like Citibank , the upshot of our suggested analysis is that as long as the mistake is harmless, restitution should be available to protect the payer. Conversely, if mistakes are harmful, then restitution should be limited to incentivize the payer to optimally invest in precautions that prevent the mistake from occurring. This limited restitution helps minimize the overall social costs associated with mistakes. In what follows, we use this generally applicable framework to offer a more detailed analysis of the guidelines that should shape court decisions in cases like Citibank as well as other mistake cases.[footnote]Cases of mistakes include, inter alia, different types of unintended money transfers as well as mistaken improvements of chattels or land. See, e.g., Estate of Hatch ex rel. Ruzow v. NYCO Minerals Inc., 704 N.Y.S.2d 340, 341 (N.Y. App. Div. 2000) (mistaken payment); Somerville v. Jacobs, 170 S.E.2d 805, 813–14 (W. Va. 1969) (recognizing the contingent nature of restitution in cases of mistaken improvement); P.S. Davies, Making Mistakes, 24 Nat’l L. Sch. India Rev., 2012–2013, at 97, 98) (“The vast array of possible mistakes, and the very different contexts and consequences of such mistakes, have made it difficult for the courts to establish clear guidelines about what the effect, if any, of a mistake should be.”); Saul Levmore, Explaining Restitution, 71 Va. L. Rev. 65, 82–120 (1985) (exploring a variety of claims for restitution).[/footnote] The Piece proceeds as follows. Part I explains the basic rationale for restitution, as outlined above. That is, it shows that when mistakes produce no direct harms, restitution should be allowed in order to minimize the payer’s investment in precautions against mistakes. Part II adds another element to the analytical picture by considering the case in which mistakes do produce direct harms. In such cases, restitution should be denied or limited in order to incentivize the payer to invest in preventing harmful mistakes. Part III studies an intermediate case: mistakes that might be harmful . This Part analyzes the discharge for value doctrine as a proxy rule designed to help courts distinguish harmful and harmless mistakes. Based on this analysis, this Part also provides a critique of the Citibank decision and the manner in which the discharge for value doctrine was applied and argues that the Citibank case should have been decided differently and should be reversed on appeal. A short conclusion follows. I. Rationalizing Restitution This Part highlights the underlying rationale for restitution in cases of mistakes, using the mistaken payment scenario as a paradigmatic example.[footnote]See Peter Birks, Unjust Enrichment 3 (2d ed. 2005).[/footnote] Cases of mistaken payments are a useful category for our framework, as they are considered the core case of the law of restitution.[footnote]Id.; Hanoch Dagan, The Law and Ethics of Restitution 19 (2004) [hereinafter Dagan, Law and Ethics of Restitution].[/footnote] Such cases are common,[footnote]See Andrew Burrows, Restitution of Mistaken Enrichments, 92 B.U. L. Rev. 767, 767 (2012) [hereinafter Burrows, Restitution of Mistaken Enrichments] (“The restitution of a mistaken payment is generally regarded as the paradigm example of the restitution of an unjust enrichment. The central issues are clear cut, the case law is voluminous, and mistaken payments are commonplace in everyday life.”).[/footnote] regularly arrive at court,[footnote]See Davies, supra note 26, at 98 (“Given the frequency with which mistakes occur, it is unsurprising to find the courts regularly grappling with cases involving mistakes.”).[/footnote] and their resolution involves significant social costs.[footnote]Gilboa & Kaplan, The Mistake About Mistakes, supra note 20, at 430.[/footnote] We first describe the basic features of the doctrine,[footnote]See infra section I.A.[/footnote] and then move on to explain its underlying rationales by analyzing the effects of the doctrine on the parties’ incentives using a stylized example.[footnote]See infra section I.B.[/footnote] We then generalize from this example to offer a general framework justifying restitution in cases of mistake.[footnote]See infra section I.C.[/footnote] A. The Doctrine in Detail In mistaken payment cases, a payer unintentionally transfers money to an unintended recipient.[footnote]Mistaken payments may occur for different reasons, such as clerical errors, see, e.g., Gen. Elec. Cap. Corp. v. Cent. Bank, 49 F.3d 280, 281–83 (7th Cir. 1995) (involving a mistaken payment due to the erroneous omission of an account number from a payment order); Credit Lyonnais N.Y. Branch v. Koval, 745 So. 2d 837, 839–41 (Miss. 1999) (detailing examples of misunderstanding of payment orders); Banque Worms v. BankAmerica Int’l, 570 N.E.2d 189, 191–92 (N.Y. 1991) (detailing examples of mistaken interpretation of the legal validity of a debt); Estate of Hatch ex rel. Ruzow v. NYCO Minerals Inc., 704 N.Y.S.2d 340, 341 (N.Y. App. Div. 2000) (detailing example of mistaken royalty payments).[/footnote] The general rule in such cases is that, subject to some defenses,[footnote]The change of position doctrine is a central defense in such cases. This doctrine is used to limit restitution when the recipient of a mistaken payment relied on the mistake in good faith, so that returning the paid sums to the payer would cause a loss to the recipient. Restatement (Third) of Restitution and Unjust Enrichment § 65 cmts. a, d (Am. L. Inst. 2011). See infra Part II for an elaboration on the change of position doctrine.[/footnote] the mistaken payer holds a claim in restitution against the recipient of the payment,[footnote]Id. § 6 (“Payment by mistake gives the payor a claim in restitution against the recipient to the extent payment was not due.”).[/footnote] who is typically considered to be unjustly enriched[footnote]Id. § 6 cmt. a (“Mistaken payment of money not due presents one of the core cases of restitution, whether liability is explained by reference to the transferee’s unjustified enrichment or to the transferor’s unintended dispossession.”).[/footnote] at the payer’s expense.[footnote]Id. § 57 cmt. h, illus. 26 (explaining the element “at the expense” by noting the existence of a causal link between a claimant’s mistake and the defendant’s enrichment).[/footnote] In such cases, the recipient must make restitution of the mistakenly transferred sum despite the fact that the recipient is the passive party and did nothing to invite this obligation.[footnote]Hanoch Dagan, Unjust Enrichment: A Study of Private Law and Public Values 4 (1997) (explaining the origins of liability in the law of restitution when the defendant is a passive and innocent party and recognizing such cases as a central category within the law of restitution).[/footnote] Recipients, the defendants in such cases, are typically innocent parties: They committed no tort,[footnote]Lionel D. Smith, The Province of the Law of Restitution, 71 Canadian Bar Rev. 672, 674 (1992) (“[The defendant] ha[s] done nothing wrong . . . . Nor is there any need to prove the breach of any duty imposed by the legal system.”); Robert Stevens, The Unjust Enrichment Disaster, 134 Law Q. Rev. 574, 577 (2018) (“[T]here is no contractual entitlement to repayment; nor has the defendant committed any wrong.”).[/footnote] and there is no contract between them and the mistaken payer,[footnote]Smith, supra note 41, at 675.[/footnote] seeing as the payment is a unilateral legal action,[footnote]Id.[/footnote] not a contract.[footnote]Id.[/footnote] Part I focuses on cases of harmless mistakes, or, in the language of Costs of Mistakes analysis, mistakes that do not lead to direct costs. As we show, in such cases, the law allows for restitution, and the benefit that was unintentionally transferred to the recipient is returned to the mistaken transferor. This demonstration provides the first element of our normative framework, explaining the basic rationale motivating the rule of restitution for mistakes. Part II discusses cases of harmful mistakes to demonstrate the rationale for limiting restitution.[footnote]The change of position doctrine is the primary defense used in such cases. See infra Part II. [/footnote] B. Analyzing Incentives To explain the basic logic that guides the rule of restitution for mistakes, consider the following stylized example, illustrating a simple mistaken payment case. Example 1: Bank A intends to transfer a sum of $100,000,000 to Bank B. However, due to a clerical error, Bank A mistakenly transfers the money to Bank C instead. Bank C and Bank A had no prior engagement, and Bank A immediately notifies Bank C of the mistake. Bank A makes numerous transfers daily, so eliminating mistakes is costly; for simplicity, assume Bank A could have prevented the mistake by investing an additional $1,000 in ex ante precautions. These precautions might involve, for example, the employment of additional clerks to review each transfer, or the purchase of more sophisticated software to identify and prevent mistakes. Example 1 illustrates an important point: preventing mistakes ex ante is possible but costly.[footnote]Maytal Gilboa & Yotam Kaplan, The Other Hand Formula, 26 Lewis & Clark L. Rev. (forthcoming 2022) (manuscript at 8) (on file with the Columbia Law Review ).[/footnote] The reason preventing mistakes is costly is that the payer cannot know in advance what type of mistake will occur and in which particular payment. Considering the frequency of money transfers, payers have to invest heavily to prevent mistakes in each and every transfer they make.[footnote]Burrows, Restitution of Mistaken Enrichments, supra note 29, at 767. Large transfers are typically monitored with added precautions. See, e.g., In re Citibank August 11, 2020 Wire Transfers, 520 F. Supp. 3d 390, 401–03 (S.D.N.Y. 2021). Yet, as demonstrated by the Citibank case, even the most highly monitored money transactions are not error-proof. Id. at 404–05.[/footnote] Thus, even if the price of preventing a mistake per transfer is low, the aggregate cost of preventing all mistakes can be high, as the cost of precautions for each individual transfer is multiplied by the (very large) number of transfers executed.[footnote]This observation is comparable to the advantage of litigation over regulation described by Professor Steven Shavell. Steven Shavell, A Fundamental Enforcement Cost Advantage of the Negligence Rule Over Regulation, 42 J. Legal Stud. 275, 275–76 (2013).[/footnote] Naturally, payers have a strong incentive to avoid mistakes, as they risk losing significant sums if mistakes occur.[footnote]Payers have a strong incentive to prevent mistaken payments since, even under a rule of full restitution, they stand to lose transferred money if, for example, the mistake is not detected, the recipient is never located, or the recipient is judgment proof. See Gilboa & Kaplan, The Mistake About Mistakes, supra note 20, at 430–31; Andrew Kull, Defenses to Restitution: The Bona Fide Creditor, 81 B.U. L. Rev. 919, 926 (2001) [hereinafter Kull, Defenses to Restitution]; see also Dhammika Dharmapala & Nuno Garoupa, The Law of Restitution for Mistaken Payments: An Economic Analysis (Univ. Chi. Coase-Sandor Inst. L. & Econ., Working Paper No. 931, 2021), https://ssrn.com/abstract=3902607 [https:// perma.cc/C3P4-2NH2] (modeling restitution for mistaken payments based on an assumption of perfect enforcement).[/footnote] But compared to the high costs of detecting and preventing mistakes ex ante, the cost of correcting mistakes after they are discovered can be trivial.[footnote]Gilboa & Kaplan, The Mistake About Mistakes, supra note 20, at 437 (showing that even if preventing unintended transfers is cheaper per transfer than reversing unintended transfers using the litigation system, the cost of ex ante precautions is borne for every transfer, while the cost of litigation is probabilistic, thus borne only for those rare transfers where a mistake occurred).[/footnote] The reason for this is that the cost of reversing mistakes does not have to be borne for every transfer made, but only in those very rare cases in which a mistake actually took place. Thus, the cost of correcting a mistake, once discovered, will typically amount only to the minor cost of making another transfer to reverse the mistaken one. Considering the costliness of preventing mistakes, should Bank A be entitled to restitution of the mistakenly transferred sum of $100,000,000? In answering this question, we focus on the practical implications of this decision. That is, if restitution is allowed or denied, how will this affect the behavior of the parties in Example 1 and the ex ante incentives of Bank A to invest in costly precautions? We argue that the decision regarding the availability (or unavailability) of restitution should be made in reference to the goal of lowering the overall costs of the payment system and, in particular, the overall costs of mistakes. Consider therefore the ex ante incentive of Bank A before the mistake occurred, under both relevant legal regimes: a regime of full restitution and a regime of no restitution.[footnote]Id.[/footnote] First, under a regime that does not allow for restitution in case of mistakes, the mistake will cost Bank A $100,000,000 ex post, as the bank will lose that sum if a mistake occurs. This means that Bank A will want to invest $1,000 and prevent the mistake ex ante. Intuitively, this may seem like a desirable outcome, as, after all, the bank can invest in precautions to prevent the mistake. Yet, in fact, such an investment is socially wasteful and undesirable since the mistake in Example 1 is harmless.[footnote]The only loss the mistake in Example 1 generates is the cost of reversing the transfer. As explained below, this cost can be assumed to be trivial. See infra notes 90–91 and accompanying text.[/footnote] To explain the notion of harmless mistakes, we distinguish between two types of costs in cases of mistakes: direct costs and indirect costs. In Example 1, the mistake resulted in no direct cost as it generated no net harm: Bank A indeed lost the sum of $100,000,000, but this loss is offset by Bank C’s gain of the same amount. The mistake caused no direct harm as nothing was destroyed, and there was no reduction in overall social welfare following the mistake. Yet, if restitution is unavailable and Bank A invests $1,000 in preventing the mistake (to avoid losing the sum of $100,000,000), this constitutes a real indirect cost. It is an indirect cost as it is not a harm that follows the mistake; and yet, it is a very real cost, as it constitutes a reduction in overall social resources. Thus, if restitution is unavailable, the overall cost of the mistake is $1,000. Conversely, if restitution is available, the mistake is not only socially harmless but also privately harmless for Bank A. The reason for this, of course, is that under a regime allowing restitution, Bank A knows it is entitled to receive the money back in case of mistake. Therefore, as the mistake causes it no harm, Bank A will not invest $1,000 in preventing it. If the mistake occurs, it will be reversed at some trivial cost, and the overall social cost will be close to zero. This is an improvement compared to the overall cost of $1,000 under a regime that denies restitution. This simple illustration therefore explains the basic rationale for the prevailing rule allowing restitution in cases of mistakes: When mistakes are harmless and generate no direct costs, restitution is beneficial in lowering the indirect costs of mistakes (the investments designed to prevent mistakes).[footnote]See supra notes 35–37 and accompanying text.[/footnote] C. Harmless Mistakes As the analysis of Example 1 shows, full restitution for mistakes is preferable to no restitution, at least in simple cases.[footnote]For a model supporting this conclusion, see Gilboa & Kaplan, The Mistake About Mistakes, supra note 20, at 447.[/footnote] This outcome is somewhat counterintuitive. Intuitively, one might think that the law should act to incentivize Bank A to invest in preventing the mistake ex ante. Such an intuition fits with our habitual thinking regarding accidents and tort doctrine.[footnote]Beatson & Bishop, supra note 23, at 153.[/footnote] After all, in the familiar Costs of Accidents framework, we are used to assigning harms to those who caused the accident.[footnote]Calabresi, The Costs of Accidents, supra note 15, at 31–36.[/footnote] Yet mistakes are different from accidents and, as can be seen from Example 1, do not necessarily result in direct costs. In such cases, the goal of the legal regime should only be to minimize the indirect costs of mistake or, in other words, lower the investment in precautions designed to prevent mistakes. This analysis explains existing doctrine and reveals the policy considerations shaping the law of mistakes. Thus, according to prevailing law, in the circumstances of Example 1, Bank A should be granted full restitution of the $100,000,000 it mistakenly transferred to Bank C.[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 6 (Am. L. Inst. 2011).[/footnote] Subject to some exceptions, on which we elaborate below,[footnote]See infra Part II.[/footnote] this outcome represents the law in all jurisdictions.[footnote]See Farnsworth, supra note 9, at 3–4 (explaining that the law of restitution does not rely on precise definitions of intent). But see Restatement (Third) of Restitution and Unjust Enrichment § 65 (holding that there is an equitable exception to full restitution if receipt of a benefit has led the recipient to change their position in a manner that would make full restitution detrimental).[/footnote] As we explain, it also makes pragmatic sense. More generally, this outcome helps to establish our proposed analytical framework for analyzing mistake cases, under the headline of Costs of Mistakes. Thus, costs of accidents include both direct and indirect costs, and the law should minimize both by assigning the harms of accidents to those who caused them. Conversely, costs of mistakes include indirect costs and may not include direct costs depending on the mistake. When a mistake causes no direct harm, the law should minimize its indirect costs by assigning any harms away from the party that caused the mistake. The comparison between the Costs of Accidents and Costs of Mistakes frameworks can also be explained in terms of deterrence versus protection. When a mistake is harmful (i.e., equivalent to an accident), the law seeks to burden whoever caused the mistake or was able to prevent it, in order to create deterrence. Thus, under the familiar framework of the “cheapest cost avoider,” the party most able to prevent the accident must be made to bear all resulting losses.[footnote]Calabresi, The Costs of Accidents, supra note 15, at 143–44.[/footnote] This will deter them—that is, incentivize them—to invest in preventing the accident (or the harmful mistake). Conversely, when the mistake is generating no direct costs, there is no need for deterrence. On the contrary, it would be useful to protect the party who caused the mistake, so that this party will not wastefully invest in protecting themselves (by attempting to prevent the socially neutral mistake). This is an important end because, in the absence of restitution, the party who made the mistake has ample incentive to invest in preventing the mistake ex ante, even if the mistake is socially harmless. The reason for this, of course, is that the mistake can be socially costless but privately costly for the mistaken party.[footnote]See Gilboa & Kaplan, The Mistake About Mistakes, supra note 20, at 431 (“Mistakenly transferring a sum of money to another is harmful to you , but it is not socially harmful. The reason for this is that your mistake benefited the recipient, so that any harm to you is (at least partly) offset by a gain by the recipient.”).[/footnote] II. Limiting Restitution The analysis in Part I explains the rationale for granting restitution for mistakes. In Part II, we complete our proposed analytical framework by explaining the rationale for limiting restitution. We show that when mistakes are harmful, restitution to the party who caused the mistake ought to be limited in order to incentivize this party to prevent the mistake and the direct costs that it entails. A. The Doctrine in Detail In this section, we illustrate the notion of harmful mistakes by adding the possibility of detrimental reliance to the basic mistaken payment scenario. Thus, mistaken money transfers can be socially harmful when the recipient relies on the payment to their detriment[footnote]A change of position is considered detrimental only if reversing the unintended transfer entails significant and unavoidable losses for the recipient. See, e.g., First Nat’l City Bank v. McManus, 223 S.E.2d 554, 558–59 (N.C. Ct. App. 1976).[/footnote] in good faith.[footnote]Detrimental reliance is considered the harm resulting from mistaken transfers. See Dagan, Law and Ethics of Restitution, supra note 28, at 47–49 (arguing that it is only the detrimental change of position of a recipient who relied on a mistakenly conferred benefit that constitutes actual harm and should reduce the mistaken party’s restitutionary award); Dagan, Mistakes, supra note 20, at 1806 (“[F]ollowing the conventional wisdom, this Article also uses the recipient’s reliance as the measure for the harm the recipient may potentially incur from an award of restitution to the mistaken party.”).[/footnote] In such cases, the law states that the recipient “changed their position,”[footnote]The change of position doctrine is widely applied in common law jurisdictions. In England, the House of Lords embraced the doctrine in the seminal case Lipkin Gorman v. Karpnale Ltd. [1991] 2 AC 548 (HL) 560, 579–80 (appeal taken from Eng.) (“The principle is widely recognised throughout the common law world. . . . The time for its recognition in this country is, in my opinion, long overdue.”). In Canada, the doctrine was first accepted in Rural Mun. of Storthoaks v. Mobil Oil Can., Ltd., [1976] 2 S.C.R. 147 (Can.) (affirming that “it should be open to the Municipality to seek to avoid the obligation to repay the moneys it received if it can be established that it had materially changed its circumstances as a result of the receipt of the money”). For a review of the change of position doctrine in England, see generally Andrew Burrows, Change of Position: The View From England, 36 Loy. L.A. L. Rev. 803 (2003) [hereinafter Burrows, Change of Position].[/footnote] and restitution is usually limited or, in some cases, denied outright.[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 65 cmts. a, d (Am. L. Inst. 2011) (“If receipt of a benefit has led a recipient without notice to change position in such manner that an obligation to make restitution of the original benefit would be inequitable to the recipient, the recipient’s liability in restitution is to that extent reduced.”).[/footnote] As we explain below, the change of position doctrine makes pragmatic sense, and fits our proposed Costs of Mistakes analysis. The change of position doctrine represents settled law in virtually all jurisdictions.[footnote]Dagan, Mistakes, supra note 20, at 1814–15.[/footnote] Although the manners by which restitution is limited and the precise measures of recovery do vary between jurisdictions, the general principle of the defense is universally applied.[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 51 cmt. a; Beatson & Bishop, supra note 23, at 154–55; Dagan, Mistakes, supra note 20, at 1814–15. While scholars are in agreement that restitution should be limited, they support different rules in order to optimally incentivize both the payer and the recipient. Such rules differ in the cost of their application and in their ability to achieve optimal results under different factual assumptions regarding the information parties have and the possibility of error in adjudication. For example, Professor Hanoch Dagan considers the rule of comparative fault to be superior, see Dagan, Mistakes, supra note 20, at 1814–15, 1817, while Professor Peter Huber is of the opinion that the rule of contributory fault is generally superior, see Peter K. Huber, Mistaken Transfers and Profitable Infringement on Property Rights: An Economic Analysis, 49 La. L. Rev. 71, 86–87 (1988).[/footnote] This fact will prove significant as we turn to discuss the Citibank decision and the more controversial discharge for value doctrine in Part III. B. Analyzing Incentives To demonstrate the operation and rationale of the change of position doctrine, consider the example below: Example 2: A bank transfers $2,000 by mistake to Emma, an unintended recipient. Emma, believing the sum is a present from her aunt Jill, uses it to buy a new laptop computer. By the time Emma learns of the mistake, she cannot return the computer and can only sell it used for $1,500. For simplicity, assume that the bank makes similar mistakes, on average, once every 100 transfers (i.e., in 1% of the cases). The bank can invest in preventing mistakes; in particular, the bank can invest $1 per transfer in order to reduce the likelihood of mistakes by half (regular precautions), or it can invest $6 per transfer to eliminate mistakes completely (high precautions). Example 2 reflects several assumptions. First, the marginal cost of preventing mistakes rises with the level of protection. That is, the bank can invest moderately to reduce the likelihood of mistakes, but eliminating them completely is very expensive. This reflects the idea that mistakes, at some level, are nearly unavoidable. There is always a chance something will go wrong. Second, it is assumed Emma changed her position in reliance on the payment, meaning she would not have bought the computer otherwise.[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 57 (explaining the causation requirement in this context).[/footnote] To evaluate the appropriate legal response under the circumstances of Example 2, consider the ex ante incentives of the bank under three different restitutionary regimes: no restitution, full restitution, and partial restitution. First, under a regime of no restitution, if the bank makes a mistake, it loses the entire transferred sum. This means that, ex ante, the mistake involves a prospective cost of $20 per transfer for the bank (1% chance of losing the full sum of $2,000). If the bank invests only in regular precautions, the cost for the bank would be $11 per transfer (reflecting $1 investment in precautions and the resulting 0.5% chance of losing $2,000). The total prospective cost is greater than $6 per transfer. Therefore, the bank will choose to invest in high precautions to eliminate mistakes completely, even though this is a socially inefficient choice. This will set the overall cost of the mistake (the harm caused by the mistake and the cost of precautions) at $6 per transfer. Conversely, under a regime of full restitution, the bank will invest nothing in preventing the mistake, as it is entitled to fully recover for its mistake. Yet, in case of a mistake, Emma will suffer a loss of $500. She will have to return $2,000 to the bank but can only sell the laptop for $1,500, meaning she will have to spend $500 of her own money. In other words, the mistake produces a direct cost, by the fact that it made Emma rely on the payment to her detriment and caused her to make a purchase she would not have made otherwise, had she known her true wealth. This sets the expected cost of a mistake under a regime of full restitution at $5 per transfer (1% chance of the mistake occurring and thereby resulting in a harm of $500 to Emma). Finally, under a regime of partial restitution, namely under the change of position doctrine, the recipient’s liability is reduced to reflect the losses it suffered in relying on the mistaken payment.[footnote]Id. § 65 (stating that the recipient’s liability is reduced to the extent of their change of position in reliance on the payer’s mistake).[/footnote] Under this rule, Emma will have to make restitution to the bank but only of the sum of $1,500, which is the value still held by her of the original payment she received.[footnote]Id.[/footnote] In this case, Emma is not harmed. She is not left worse off but arrives at the same position she was in before the mistake: She simply gives up the laptop and does not need to add any of her own money to make restitution to the bank. At the same time, the mistake costs the bank a sum of $500, as it paid $2,000 to Emma and only received $1,500 in restitution. Ex ante, the mistake therefore costs the bank $5 per transfer (1% chance of losing $500). This means the bank will not invest $6 per transfer in preventing mistakes completely, as this cost is higher than the cost reflecting the risk of actually suffering the mistake. The bank will, however, invest $1 per transfer in regular precautions to lower the chance of mistakes. This option is preferable for the bank, as it entails a cost of $3.5 per transfer for the bank, instead of $5 per transfer. This will also be the overall social cost of the mistake: $1 in precautions per transfer, plus a 0.5% chance per transfer of a loss of $500. The analysis of Example 2 demonstrates that the rule of partial restitution is superior to both the regime of no restitution and the regime of full restitution in terms of overall social cost and is, in fact, optimal. Under a rule of no restitution, the overall social cost is $6 per transfer; under a regime of full restitution, the overall social cost is $5 per transfer; and under a regime of partial restitution, the overall social cost is only $3.5 per transfer. Considering the immense number of transfers taking place, this advantage of the limited restitution regime can easily translate into a significant efficiency advantage and lower the operational costs of the payment system.[footnote]Davies, supra note 26, at 98 (“Given the frequency with which mistakes occur, it is unsurprising to find the courts regularly grappling with cases involving mistakes.”).[/footnote] C. Harmful Mistakes The reason for the superiority of the limited restitution rule is simple. Limited restitution offers adequate protection to the party that caused the mistake while also assigning to it all social costs related to the mistake. This balance of burdens allows the payer to lower its investment in precautions without harming others. That is, the payer can reclaim the money paid by mistake, minus those sums already spent by the recipient in reliance on the payment. This assures optimal incentives by the payer, who will invest to prevent the direct social cost of the mistake that it would bear should the mistake occur. In this way, the payer will optimally invest in precautions, thus minimizing both the direct and indirect costs of the mistake. In Example 2, the mistake causes some direct harm, and the bank is the party causing this harm. In this sense, the mistake in Example 2 is similar to an accident in Calabresi’s framework:[footnote]Calabresi, The Costs of Accidents, supra note 15.[/footnote] It is an unintended harmful event. By limiting restitution, the law makes the bank bear the direct cost caused by the mistake, therefore providing the bank with optimal incentives to invest in preventing that harm ex ante.[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 65 cmt. a (describing the requirement of causation as part of the change of position defense).[/footnote] Notice that limiting restitution makes sense here only because the mistake is somewhat harmful. Conversely, if the mistake is harmless, as in Example 1, the only goal of legal intervention should be to reduce the indirect costs of the mistake—that is, the costs of the investment designed to prevent mistakes. This goal, as we explained above, is achieved by granting full restitution to the payer.[footnote]See supra section I.B.[/footnote] III. Discharge for Value The analysis of Example 1 and Example 2 establishes a normative and analytical framework for deciding cases of mistakes. Example 1 establishes the rationale for restitution in cases of harmless mistakes; Example 2 establishes the rationale for limiting restitution in cases of harmful mistakes. These examples explain the existing doctrines of restitution for mistakes and the change of position defense. After presenting the basic elements of our proposed framework, we now turn to study a more controversial case: a mistaken payment of an existing debt. We first offer a general analysis of these issues and then apply this analysis to the Citibank decision.[footnote]In re Citibank August 11, 2020 Wire Transfers, 520 F. Supp. 3d 390 (S.D.N.Y. 2021).[/footnote] A. The Doctrine in Detail In some cases, a payer mistakenly pays off an existing debt. For instance, in Banque Worms v. BankAmerica International , a debtor sent a payment order to its bank, intending to pay up a loan owed to its creditor.[footnote]570 N.E.2d 189, 190 (N.Y. 1991).[/footnote] Before the payment was processed at the bank, the debtor changed its mind and sent the bank a message canceling the payment order.[footnote]Id.[/footnote] Yet due to some confusion at the bank, the payment order was processed anyway, and the creditor received the transferred money despite the canceled payment order.[footnote]Id.[/footnote] In this case, the payment was made by mistake, but the recipient had a pre-existing right in the transferred sum. In Banque Worms , the court decided that the recipient of the mistaken payment is not obligated to make restitution of the mistakenly transferred sum, as it enjoys a defense under the discharge for value doctrine: When a beneficiary receives money to which it is entitled and has no knowledge that the money was erroneously wired, the beneficiary should not have to wonder whether it may retain the funds; rather, such a beneficiary should be able to consider the transfer of funds as a final and complete transaction, not subject to revocation.[footnote]Id. at 196.[/footnote] This is a standard restatement of the discharge for value doctrine,[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 67 (Am. L. Inst. 2011).[/footnote] explaining that restitution for a mistaken payment is denied when two conditions are met: First, the transfer was made as payment of an existing debt,[footnote]Id. § 67(1); Gilboa & Kaplan, The Mistake About Mistakes, supra note 20, at 442.[/footnote] and, second, the recipient had no notice of the mistake.[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 67(2).[/footnote] In such cases, the recipient is exempt from the duty to make restitution, even if there is no proof of detrimental reliance.[footnote] Banque Worms , 570 N.E.2d at 196.[/footnote] The discharge for value doctrine is sometimes justified as supporting a notion of finality of payments.[footnote]Id. at 195; see also Andrew Kull, Restitution and Final Payment, 83 Chi.-Kent L. Rev. 677, 677–78 (2008) (discussing “whether some rule of restitution law gives the payee an affirmative defense” where a mistaken payment was made).[/footnote] According to this explanation, the doctrine is meant to free the recipient from the need to check whether a mistake occurred once a payment was made, which would supposedly result in unnecessary social costs as recipients would need to verify every payment even though mistakes account for a small percentage of all payments. This argument, however, is left largely unexplored and has never been supported through an analysis of the parties’ incentives. We now turn to provide such an analysis. B. Analyzing Incentives Our study of the discharge for value doctrine starts with an analysis of its effects on the parties’ incentives. For this purpose, consider the following example: Example 3: Miranda is a designer and manufacturer who owns her own fashion house. Miranda’s company has long-standing arrangements with multiple suppliers. The company buys supplies on credit and makes periodical payments against its outstanding debt. In making one of these payments, Andy, an employee in Miranda’s company, makes a mistake, paying a supplier $1,000,000 (the entire debt amount) instead of $10,000 (the interest amount due at the time). The company learns of the mistake and reports it to the supplier the very next day. The company asks for the money back and will be at serious financial risk if the mistake is not reversed. Yet the recipient refuses to return the money. Assume that the company can invest ex ante in precautions to prevent mistakes; in particular, the mistake would be prevented with absolute certainty if the company invests $5,000 in added precautions ex ante. Example 3 offers a scenario that stands between the cases presented in Example 1 and Example 2 above. Like in Example 1, the mistake in Example 3 did not actually generate any direct costs. The mistake was discovered immediately, and there was therefore no detrimental reliance by the recipient. But Example 3 is also somewhat similar to Example 2 in the sense that it seems there is some potential for detrimental good-faith reliance on the payment. In cases like Example 1, when the payment was made “out of the blue” with no existing debt, it would be harder to believe that the recipient relied on the payment in good faith. Conversely, in cases like Example 3, reliance on the payment in good faith seems a more likely possibility, as the recipient can understandably assume the money was intentionally transferred given the existing debt between the parties.[footnote]As a proxy rule, the discharge for value defense provides greater protection for the recipient compared to the change of position doctrine, since it does not require actual proof of detrimental reliance. For a comparison between the two doctrines, see Charlie Webb, Reason and Restitution: A Theory of Unjust Enrichment 230–32 (2016); Kull, Defenses to Restitution, supra note 49, at 924–25.[/footnote] As we show below, the comparison between the different examples is helpful in explaining the appropriate legal response to cases like Example 3. Consider the ex ante incentives of the mistaken payer in Example 3. First, if restitution is denied, the mistake is hugely costly for the company, entailing a private loss of $1,000,000. Therefore, if restitution is not available, the company will prefer investing $5,000 in ex ante precautions to prevent the mistake.[footnote]See Eric Talley, Discharging the Discharge-for-Value Defense, 18 N.Y.U. J.L. & Bus. 147, 153, 200–03 (2021) (showing that sophisticated parties responded to the Citibank decision by increasing their ex ante investment in contractual precautions designed to save them from the losses associated with a mistaken transfer of a due debt).[/footnote] This is a wasteful investment designed to prevent a socially neutral event, meaning that overall social costs under a regime of no restitution will equal $5,000. Conversely, if restitution is available, the mistake is no longer harmful for the company. In this case, the company will not invest in preventing it. Therefore, under a regime that allows for restitution, the overall cost of the mistake is zero (or close to zero). This means that restitution is socially desirable (since, as explained above, if restitution is denied this causes a social waste of $5,000). This analysis of the parties’ incentives also shows that allowing restitution does not harm the finality of payments. The recipient is not required to make any investigation before it is allowed to use the money paid to it. Rather, the recipient is only required to return the money when and if the payer reports a mistake.[footnote]The recipient is not obligated to return a payment if they are unaware of any mistake. Similarly, if they have no reason to think a mistake occurred, and relied on the payment, they are protected under the change of position doctrine. See Abraham Drassinower, Unrequested Benefits in the Law of Unjust Enrichment, 48 U. Toronto L.J. 459, 488 (1998).[/footnote] Thus, as long as no request from the payer arrives, the recipient is free to use the money. Similarly, if the recipient indeed relied on the payment and spent the money, it is already protected under the change of position doctrine and is exempt from restitution.[footnote]Restatement (Third) of Restitution and Unjust Enrichment § 65 (Am. L. Inst. 2011).[/footnote] The availability of restitution thus only involves the trivial cost of reversing the mistake, in those rare cases in which it occurred, and places no additional burden on the recipient. More broadly, denying restitution in cases like Example 3 seems inefficient. Looking on Example 3 and extrapolating out to the general case, we can see that, as long as the mistake produces no direct costs, any limitation on restitution only increases the investment in precautions, thereby increasing the overall social costs and the cost of operating the payment system. Denying restitution when mistakes are harmless directly contradicts the rationales outlined in Part I and Part II and only increases the overall costliness of mistakes. Therefore, the discharge for value rule, denying restitution in cases of existing debt, even when the recipient shows no detrimental reliance, seems questionable from the get-go and should be treated as a narrow exception at best. C. Proxy for Harm We argue that the analysis of Example 3 above explains the rationale for the discharge for value doctrine as a proxy rule, or a rule designed to identify a second-best decisionmaking mechanism. The core question in mistake cases is the question of harm, as can be seen from the analysis of Examples 1 and 2. If the mistake caused no direct costs (like in Example 1), restitution should be granted to the party who made the mistake in order to lower the indirect costs of the mistake—that is, the investment in precautions designed to prevent it. Conversely, if the payment caused direct costs in the form of detrimental reliance by the recipient (like in Example 2), then restitution should be limited or even denied.[footnote]The limits of restitution in this case should be determined in correspondence with the recipient-defendant’s reliance on the payment. See id.[/footnote] A proxy rule can be useful here since the core question of the degree of the recipient’s reliance can be difficult to determine. That is, in real life scenarios, it can be difficult to know, and even more difficult to prove in court, whether the recipient indeed relied on the payment and whether this reliance was made in good faith.[footnote]See First Nat’l City Bank v. McManus, 223 S.E.2d 554, 558–59 (N.C. Ct. App. 1976); Burrows, Change of Position, supra note 64, at 813–15.[/footnote] Sums paid by mistake can be intermingled with the recipient’s funds, making it difficult to determine whether the mistaken payment is still held by the recipient or has already left the recipient’s accounts.[footnote]Andrew Kull, Ponzi, Property, and Luck, 100 Iowa L. Rev. 291, 304 (2014) (illustrating the problem of intermingled funds).[/footnote] Against this backdrop, the discharge for value doctrine can prove useful as a proxy tool. That is, if the mistaken payment was made against an existing debt, it seems likely that the recipient would have relied on the payment in good faith; since the recipient had a valid claim to the money, it is quite reasonable that they believed the money was theirs to spend. Therefore, in cases of mistaken payment of existing debt, even if a change of position is not proven, it seems reasonably likely (yet not certain) that a change of position took place. For this reason, in such cases it might be useful to deny restitution, even if reliance cannot be shown. This analysis completes our framework: Example 1 presents the case of harmless mistakes, Example 2 illustrates harmful mistakes, and Example 3 deals with the middle case—mistakes that might be harmful . This framing helpfully delineates the reasonable limits of the discharge for value doctrine and its appropriate implementation. That is, the discharge for value doctrine, if applied as a useful proxy rule, should lead to an outcome that restitution is denied when it seems reasonably likely that the recipient indeed relied on the payment in good faith.[footnote]The application of the discharge for value defense is still independent of the application of the change of position rule. That is, if the recipient proves they relied on the payment in good faith, restitution is denied or limited under the change of position rule.[/footnote] Symmetrically, the doctrine should be designed so that if it seems unlikely that the recipient relied on the payment, restitution will be granted. Otherwise, if the discharge for value doctrine limits restitution even when detrimental reliance is unlikely, this will only increase wasteful investment designed to prevent harmless mistakes, thus increasing the overall costs of mistakes.[footnote]For a formal proof of this proposition, see Appendix infra.[/footnote] Equipped with this insight, based on our Costs of Mistakes framework, we can now evaluate the finer details of the discharge for value doctrine and the particulars of its application. Our main focus here is on the harsh interpretation of the discharge for value doctrine as applied in the Citibank decision. Based on our analysis, the application of the discharge for value doctrine by the Citibank court seems overbroad. In Citibank , like in Example 3 above, the mistake was reported immediately. In such cases, there is a near-zero chance that the recipient changed its position in reliance on the payment. As there is very little chance that the mistake was in fact harmful, there is no reason to limit or deny restitution. D. Back to Citibank Based on the analysis thus far, we now turn to evaluate the reasoning of the Citibank decision.[footnote]In re Citibank August 11, 2020 Wire Transfers, 520 F. Supp. 3d 390 (S.D.N.Y. 2021).[/footnote] Consider first the facts of the case, which we describe here in some more detail. In August 2020, a Citibank employee made a mistaken transfer of an imaginary sum of nearly a billion dollars, in what the District Court for the Southern District of New York later dubbed “one of the biggest blunders in banking history.”[footnote]Id. at 396.[/footnote] The bank employees intended to transfer a sum of $7.8 million from a business account held by Revlon to several accounts of Revlon’s creditors; instead, the employee transferred an additional sum of nearly $900 million from the bank’s own funds to those creditors. The bank intended to pay only an interest payment but instead paid the principal amount of the loan in its entirety, plus all interest owed.[footnote]The employee operated under the mistaken belief that the money would be transferred to a temporary clearing account and that only the interest payments would actually be sent to the lenders. Id. at 396, 402.[/footnote] The mistaken payment was made from the bank’s own funds and not from Revlon’s due to the structure of Revlon’s roll-up transaction with its lenders.[footnote]Id. at 404.[/footnote] The mistake was discovered the very next day, and the bank immediately sent notices to Revlon’s creditors informing them of the erroneous transfer of money and requesting them to return the sums they received as soon as possible.[footnote]Id. at 404–05.[/footnote] Some creditors, however, refused to do so, retaining sums paid to them by mistake totaling over $500 million. Following the creditors’ refusal to return the money, Citibank filed a lawsuit. The district court denied the bank’s right to restitution, allowing the creditors to retain the funds.[footnote]Id. at 451.[/footnote] As mentioned above, Citibank’s appeal before the Second Circuit is pending.[footnote]See Notice of Appeal, Citibank , 520 F. Supp. 3d 390, appeal filed, No. 21-487 (2d Cir. 2021). [/footnote] Consider now the court’s justification for its decision. The district court based its decision on the discharge for value doctrine, as formulated by the New York Court of Appeals in Banque Worms .[footnote]570 N.E.2d 189, 191–94 (N.Y. 1991).[/footnote] The court decided that the two requirements of the discharge for value doctrine were met in the Citibank case: the payment was made to discharge a valid debt and the recipients had no notice of the mistake.[footnote] Citibank , 520 F. Supp. 3d at 431.[/footnote] We offer a critique of this implementation of the discharge for value doctrine, based on our Costs of Mistakes analysis. To start with, the court’s implementation of the discharge for value doctrine in this case seems misguided simply because it leads to an absurd result.[footnote]See Levine, supra note 5. Levine explains the nonsensical nature of this discharge for value judgment: "[T]his doctrine is dumb and no one in the world of syndicated lending actually meant to sign up for it; 'if you send us the wrong money we will keep it' is not a rule that anyone wanted built into their loan documents. It did not occur to anyone to opt out of it—it did not occur to anyone, outside of the small fellowship of Finders Keepers lawyers, that this rule even existed—until it cost Citi $500 million." Id.[/footnote] The fact that Citibank was denied restitution defies any reasonable expectation by ordinary business parties;[footnote]Id.[/footnote] not only that, the court itself admits the decision seems forced.[footnote]See Citibank , 520 F. Supp. 3d at 451. In explaining the contrived nature of the decision, the court admits: "Were the court writing on a blank slate, it is far from clear that it would reconcile these principles in a way that allowed the Non-Returning Lenders to keep the money that Citibank indisputably transferred by mistake. After all, Citibank realized its error and notified the Lenders within one day, and there is no evidence or suggestion that, in the intervening hours, the Non-Returning Lenders relied to their detriment on the belief that the transfers were an intentional, albeit unexpected, paydown of the 2016 Term Loan." Id.[/footnote] This alone should serve as a strong indication that the doctrine is inappropriately applied here. We support this intuition based on our systematic analysis of the law of restitution for mistakes. According to our Costs of Mistakes analysis, the discharge for value doctrine can only be justified as a proxy rule, designed to assist the court in recognizing cases in which it is likely that the recipient relied on the payment to its detriment (even when such reliance is difficult to prove), and restitution should therefore be denied or limited. The problem with the application of the discharge for value doctrine in the Citibank case is that it clearly diverges from this rationale. Under the specific facts of this case, a change of position by the recipients is nearly impossible: The payment was highly irregular, and the recipients were notified of the mistake almost immediately. Indeed, the court in Citibank notes there is no reason to believe the recipients relied on the payment to their detriment.[footnote]Id.[/footnote] This should have been reflected in the implementation of the discharge for value doctrine. If the likelihood of detrimental reliance is low (or nonexistent), a correct application of the discharge for value doctrine should lead to the outcome that restitution is granted in full. This result follows from a sensible implementation of the discharge for value doctrine to the facts of the Citibank case, an implementation that reflects the underlying rationales guiding the law of mistakes. In Citibank , the payment made was not due for years; the fact it was paid in full so early in advance was a highly exceptional occurrence.[footnote]See id. at 444 (detailing the peculiarity of and the reaction to the Citibank transfer). It is hard to imagine circumstances whereby the bona fide payee defense will be available to a recipient who could have noticed the mistake and prevented the transfer’s finality, but did not do so.[/footnote] If an early payment of hundreds of millions of dollars suddenly appears in your bank account with no explanation or any written notice, you are at the very least bound to wonder if this payment was indeed intentional.[footnote]Talley, supra note 86, at 13.[/footnote] Hence, the recipients in Citibank had at least constructive notice of the mistake.[footnote]Constructive notice is sufficient to deny a recipient of the discharge for value defense on all accounts. See, e.g., In re Calumet Farm, Inc., 398 F.3d 555, 560 (6th Cir. 2005); Citibank , 520 F. Supp. 3d at 427. Note that under New York law, constructive notice is an objective standard, inquiring whether the defendant was aware of facts that would cause a reasonably prudent person to suspect a mistake was made. That is, the test for constructive notice is not concerned with what a particular creditor actually believed, but rather with what a reasonable creditor in the same position would have believed. See Brief of Professors of Law and Economics as Amici Curiae in Support of Plaintiff-Appellant at 9–10, Citibank, N.A. v. Brigade Cap. Mgmt., LP, No. 21-478-CV (2d Cir. filed July 23, 2021), 2021 WL 3239420 (synthesizing cases detailing New York notice law).[/footnote] They were therefore highly unlikely to rely on the payment in good faith and should not be protected under the discharge for value doctrine. By the same token, allowing restitution in such a case would not harm the finality of payments. A rule allowing restitution in a scenario such as Citibank would not require recipients to generally verify the validity of payments; it only requires them to return funds paid under highly irregular circumstances, when the payer actively reports a mistake. This suggested application of the notice requirement is also consistent with the Banque Worms ruling that the Citibank court was supposedly following.[footnote] Citibank , 520 F. Supp. 3d at 427.[/footnote] In Banque Worms , there was no reason for the recipient to suspect a mistake, as the payment was made in due course, where the payer actually made a payment order but then canceled it later (a cancellation the bank mistakenly ignored).[footnote]Banque Worms v. BankAmerica Int’l, 570 N.E.2d 189, 190 (N.Y. 1991).[/footnote] The recipient in Banque Worms thus had every reason to expect the payment, and therefore it quite likely had no basis to suspect the payment was a mistake. The same cannot be said for the recipients in Citibank , who witnessed a highly irregular payment. A correct resolution of the Citibank case can also be achieved via a more reasonable interpretation of the temporal element of the discharge for value doctrine. The Citibank decision is based on a very specific interpretation of the discharge for value doctrine, according to which notice is evaluated at the moment the payment is made .[footnote]See Citibank , 520 F. Supp. 3d at 430.[/footnote] That is, according to the test employed by the court in Citibank , if, at the moment the payment was made, the recipient could have reasonably thought the payment was owed to it and made correctly, this suffices to deny restitution.[footnote]Id.[/footnote] This interpretation of the discharge for value doctrine is too broad and is detached from the rationales that are supposed to shape the doctrine and guide adjudication. In particular, this formulation of the discharge for value doctrine seems to function as a very bad proxy: It hardly correlates with a high probability of detrimental reliance by the recipient.[footnote]See Talley, supra note 86, at 18.[/footnote] The reason for this, of course, is that detrimental reliance takes time. A better formulation of the discharge for value doctrine would therefore determine that restitution is to be denied if the recipient could have reasonably thought the payment was owed to it—not at the moment of the payment—but for a period of time long enough to actually allow some “change of position” or an action in reliance of the payment. That is, had Citibank reported the mistake after a week, for example, it might have been reasonable to deny restitution based on the discharge for value doctrine. In such a case, even if the recipients cannot prove they changed their position and increased their spending in reliance on the mistake, such a possibility at least seems more likely. After all, the recipients could have reasonably believed they were owed the money paid and held that money for quite some time without this belief being challenged. It is therefore likely to assume they conducted their affairs in reliance on that money. If they are now ordered to make restitution, there is a reasonable chance that they will be harmed because of direct costs incurred by their detrimental reliance, which justifies limiting or outright denying restitution. If, on the other hand, the mistake was discovered immediately, there is no real possibility that the mistake was harmful, and there is therefore no reason to limit or deny restitution.[footnote]Naturally, after returning the payment received by mistake, the recipient will continue to hold a debt against the payer (the same as before the mistaken transfer was made). Kull, Defenses to Restitution, supra note 49, at 930–31.[/footnote] To sum up this point, the primary reason to limit or deny restitution is if the mistake generates direct costs. If no direct costs can be proved, it might still be justified to limit or deny restitution based on a proxy rule that helps to identify those cases in which it seems more likely than not that the mistake was indeed harmful. Importantly, the proxy rule should be constructed in such a way that it efficiently singles out harmful mistakes. The formulation used by the Citibank court to operate the discharge for value doctrine fails in this regard. If restitution is denied when mistakes are clearly harmless (as it was in this case), it simply leads to a wasteful increase in the overall indirect costs of mistakes as payers would invest more in ex ante precautions to prevent the mistake from happening. We suggest that the Citibank decision represents a problematic mode of judicial decisionmaking in restitution cases. The court itself declares that “[w]ere the court writing on a blank slate,” the outcome of the case would probably have been different.[footnote] Citibank , 520 F. Supp. 3d at 451.[/footnote] In other words, the court understands its decision makes little sense but decides its hands are tied.[footnote]Id. at 425 (explaining that prior decisions compel the conclusion that the relevant point in time for evaluating whether the recipient of funds sent by mistake is on notice of that mistake is the moment the payment is received).[/footnote] As we show above, the hands of the courts are not tied, and they have sufficient means to arrive at a sensible decision. Lastly, but surely not less importantly, when judges say their hands are tied, this facilitates a process by which the law slowly (or quickly in some cases) stops making sense.[footnote]See Cohen, supra note 7; Singer, supra note 14.[/footnote] As our Costs of Mistakes analysis shows, the law of restitution for mistakes does make sense, and we hope courts will cease to ignore this reality. Conclusion This Piece offers a normative framework guiding adjudicating in cases of restitution for mistakes. We delineate the policy implications of restitution decisions and highlight the way restitution doctrine reflects these considerations. The analysis we offer distinguishes harmless mistakes, or mistakes that cause no direct costs, from harmful mistakes, namely mistakes that do cause such costs. We show that when mistakes are harmless, the law provides full restitution. Conversely, when mistakes are harmful, restitution is limited or denied under the change of position doctrine. We then use this analysis to study the discharge for value doctrine as a proxy rule designed to help courts adjudicate between harmful and harmless mistakes. Based on this insight, we offer a critique of the recent application of the law of restitution as manifested in the Citibank decision. We argue that this decision represents an unfortunate trend in the law of restitution, namely the preference to adjudicate restitution cases as if divorced from common sense and from any broad pragmatic considerations. Appendix: Formal Model This Appendix offers a simple model generalizing on the examples presented in Parts I–III. The model studies the investment by the payer in preventing mistakes. Mistakes represent a potential loss for the payer (and an equivalent gain for the recipient), but restitution can result in a loss for the recipient if the recipient changed its position in reliance on the payment. The purpose of the model is to state more explicitly the intuitions explored in Parts I–III. However, our main points stand regardless of the specifics of the model and are unambiguously established in the examples analyzed in Parts I–III. In this sense, the aim of the model below is merely to clarify and provide a formal proof for our argument. A. Setting A payer makes a monetary transfer. With probability p , the payer makes some kind of mistake in making the transfer and transfers an unintended sum to the recipient. Further, p decreases with the payer’s investment in precautions, c (assume p' ( c ) p ''( c ) >0 for any c ). If the payer is able to retrieve the sum, the payer will bear an administrative cost of l for doing so. Also, if a mistake occurred and the sum is returned to the payer, the recipient may suffer a loss of h . This loss represents the recipient’s detrimental reliance. For simplicity, assume that the harm h is caused by the payer’s mistake and cannot be prevented by the recipient at a reasonable cost. B. First Best While the payer may fear the loss of the transferred sum t , this does not represent a social loss (as it entails an equivalent benefit to the recipient). The social cost therefore includes the cost of precautions by the payer ( c ), the administrative cost of retrieving the money ex post (l ) , and the harm of detrimental reliance ( h ). Therefore, the social planner minimizes: (1) c + p ( c )( l + h )

The first-order condition is: (2) - p' ( c )=1/( l + h ) This first-order condition defines the first best level of investment, c* . Note that as the cost of the mistake ( l + h ) increases, so does the optimal level of investment in precautions. This makes intuitive sense. C. Change of Position This section studies investment levels under the change of position doctrine, limiting restitution when mistakes caused harm in the form of the recipient’s detrimental reliance. Thus, under a rule of limited restitution, the payer is entitled to recover the mistakenly transferred sum minus the administrative cost of restitution ( l ) and the recipient’s harm of detrimental reliance ( h ). Thus, the payer minimizes: (3) c + p ( c )( l + h ) The first-order condition is: (4) - p' ( c )=1/( l + h ) This first-order condition defines the level of investment under a regime of limited restitution, c l . D. Full Restitution Under a rule of full restitution, the payer is entitled to recover the mistakenly transferred sum minus the administrative cost of restitution ( l ). Therefore, the payer minimizes: (5) c + p ( c ) l The first-order condition is: (6) - p' ( c )=1/ l This first-order condition defines the level of investment under a regime of full restitution, c f . E. Investment Levels By comparing (4) and (6) with (2), one can see that investment under limited restitution is optimal, while investment under full restitution is too low. Comparing (4) and (6) also shows: (7) c f c l These results can be summarized in the following proposition: Proposition 1: The change of position rule leads to optimal levels of investment in precautions, while a rule of full restitution leads to levels of investments that are lower compared to the first best. From (2) and (6), we see that the rule of full restitution generates optimal levels of investment ( c f = c *) only if: (8) h = 0 This can be summarized in the following proposition: Proposition 2: Full restitution leads to optimal results when the mistake is harmless or caused no detrimental reliance to the recipient ( h = 0) . Now consider the possibility that restitution is partial (or denied) when no harm was caused to the recipient in the form of detrimental reliance (that is, when h = 0). From (2) and (8), in such circumstances the optimal level of investment, c *, is defined by the first-order condition: (9) - p' ( c )=1/ l Yet the payer operates as if the change of position rule is applied. Therefore, from (4), the payer minimizes: (10) - p' ( c )=1/( l + h ) By comparing (9) and (10) with (2), one can see that when mistakes cause no harm, then investment under full restitution is optimal, while investment is too high if restitution is nevertheless limited or denied. This conclusion explains the correct application of the discharge for value rule. If there was no detrimental reliance ( h = 0), full restitution is optimal. Therefore, restitution should be denied or limited under the discharge for value rule only when it is reasonable to believe that h = 0 does not hold true. This result can be summarized in the following proposition: Proposition 3: Limiting restitution using the discharge for value rule can only be efficient if there is a real likelihood that the mistake harmed the recipient by causing it to rely on the payment to its detriment. Otherwise, the discharge for value doctrine leads to levels of investment that are too high compared to the first best.