A. Introduction
Behavioral economics (BE) and/or Behavioral Law and Economics (BLE) have been subjected to various methodological critiques. One reason why BLE may be stuck in a limbo of methodological intermediacy—percolating around rather than piercing through law and legal systems—is an obscure but nevertheless significant shortcoming: a recurring thorn in BE’s/BLE’s side is that it lacks a theoretical foundation and is merely the “residual” of rational choice.Footnote 1 This critique has several variants: BLE lacks coherence;Footnote 2 BLE must surmount the hurdle of generalizability;Footnote 3 BLE is too descriptive;Footnote 4 BLE is not “unified”;Footnote 5 or BLE is just a set of examples of rational choice theory (RCT) contradictions without proffering a holistic narrative.Footnote 6 Even BLE proponents themselves concede this point and have acknowledged that much work is to be done.Footnote 7 As Owen D. Jones states, “both behavioral law and economics scholarship, and the underlying cognitive literature on which it relies, are simply far better at explaining that people often behave in ways inconsistent with traditional economic theory than they are at explaining why they do so.”Footnote 8 The consequences for BE’s potential penetration of law and legal policy should be clear, though scholars have rarely articulated why a coherent generalizable theory is necessary. Without knowing how, why, and/or when behavioral anomalies appear, it will be “difficult to anticipate in whom they will appear, in what contexts, and with what vigor.”Footnote 9 And because law is a method for governing and ordering behavior generally,Footnote 10 BLE’s generalization deficit would seem to make it especially difficult to predict how society will respond to a particular legal rule or regulation.
This Article unfolds as follows. Section B demonstrates the law’s demand for clairvoyance by describing several futurity issues it often grapples with. Law and Economics (L&E) is offered as an example of a theory supplying predictive capabilities, but L&E’s source of predictive power is questioned. Section C—the Article’s core—discusses sources of predictive power and draws a distinction between what can usefully be termed endogenous and exogenous predictive power. As will become clear, this distinction may have significant implications for BE’s/BLE’s theoretical journey towards generalizability and its limitations therein. In Section D, it is argued specifically that the best BE/BLE can hope for in its theorization attempts is a set of necessary and sufficient conditions about when irrationality, rather than rationality, may govern decision-making. In other words, BE/BLE—unlike RCT—does not, and may never, possess the power to predict endogenously, only exogenously. BE/BLE lacks any normative foundation and in fact might be more accurately viewed as a body of knowledge showcasing anti-normative behavior. Drawing on the concept of fundamental scarcity, the normative foundation of RCT is also explained. Section E provides some reflections for law and legal policy on several levels and identifies the potential incompatibility of an anti-normative, idiosyncratic field like BE with a system of governance like law that is premised on generality and uniformity.
B. Futurity Issues in Law – The Demand for and Supply of Clairvoyance
I. Demand for Clairvoyance
Foresight and futurity are phenomena that have often times been obsessed over and have sometimes been characteristic of diverse fields like poetry, history, and the arts.Footnote 11 From astrologers like NostradamusFootnote 12 to ancient Greek oracles like PythiaFootnote 13 and science fiction protagonists like Paul Atreides’ dreams in Dune,Footnote 14 seeing into the future has often been synonymous with philosophers, mythological figures or science fiction narratives. Predictions have often been central for-real world events too, like the few financial traders in The Big Short who predicted through financial modeling the collapse of global financial markets.Footnote 15 Predictions are necessary across all facets of life. When crossing the road at a crosswalk as a car approaches, a pedestrian may predict that the car will slow down, yield to the pedestrian, and allow them to cross the road. Whether in our professional or personal lives, making predictions—or at the very least being mindful about the future—would thus seem to be necessary in order to navigate our day-to-day, or even year-to-year, endeavors.
Clairvoyance’s relevance to law can most obviously be seen when law is viewed as a tool for governing and ordering societal behavior.Footnote 16 Law has been conceived of as a “lever” for governing and ordering behavior in directions “it would not go on its own.”Footnote 17 Law in this sense has also been conceptualized as “a giant pricing machine,” which can affect incentives and abilities and, consequently, direct “behavior and actions.”Footnote 18 And although not everyone agrees, the well-known metaphor of law as a “command backed by threats” to serve a sovereign’s “wish” would also seem to conceive of law as a method for directing and re-directing societal behavior in desired ways.Footnote 19
Whatever the value of conceiving of law like this, what is clear is that such a conception generates a futurity problem because it requires a prediction about how actors will respond to the lever. In this sense law is “incomplete” because it cannot perfectly stipulate for “all future contingencies”; invariably, there will be some circumstances and phenomena not provided for.Footnote 20 Predictive capabilities in the legal realm have therefore been cited as significant for law-making. Knowing ex-ante the consequences of legal rules and actions is of obvious value for law-makers and legal decision-makers like judges in the sense that they can be empowered to choose and decide optimal rules. Indeed, “just as a lever’s efficiency depends on the solidity of its fulcrum, law’s efficiency in moving behavior depends on the accuracy of its behavioral model.”Footnote 21
Lawmakers enacting legislation, for instance, do not know ex-ante how society will react to the new rules. Regulators face similar problems when enacting new regulations. And judges must be mindful of their decision’s capacity to bind future judges because of precedent. Here we elucidate these demands for clairvoyance in more detail, illustrating several futurity issues that the law can be braced with.
1. Legislation and Legal Policy
Even “well-intended” legislation can sometimes produce undesirable ex-post consequences—or at least fail to produce what was desired ex-ante.Footnote 22 Indeed, recent and emerging calls for “future-proofing” legislation accentuate this issue of futurity and ex-post concerns—particularly in light of developments in the digital sector.Footnote 23 For instance, the European Economic and Social Committee in response to calls for better legislation has recognized via an Opinion the value of rule-making that accounts for future consequences.Footnote 24 And European Union member states have also demonstrated interests in “future-proofing” legislation.Footnote 25 These perspectives would seem to suggest that there exists a legislative demand for clairvoyance.
Consider for example consumer legislation, which “[o]n the one hand … seeks to intervene in the market to advance consumers’ interests and protect them. On the other hand, it aspires to avoid inefficient interventions that may expose excess hardship on businesses.”Footnote 26 Sometimes, however, mistakes are made,Footnote 27 but these mistakes are a function of the degree to which the lawmaker can know ex-ante the ex-post consequences of their actions. In short, “legal interventions—even well-intended—sometimes do not turn out as planned.”Footnote 28 The Dodd-Frank Wall Street Reform and Consumer Protection Act is illustrative, which was passed as pro-consumer legislation in response to the financial crisis.Footnote 29 One study found, for instance, that the legislation’s introduction may have led to ex-post wealth inequality because as a consequence of Dodd-Frank’s effects on mortgage origination costs, lenders focused more on wealthy loanees. This resulted in banks possessing stronger incentives to “reduce their origination of small and medium-sized loans ….”Footnote 30
Another illuminating example is government-substituted loans when the government identifies that the private sector is catering inadequately to poorer consumers. One study found that the Small Business Administration disaster program actually prevented poorer individuals from accessing credit more than the private sector did. Borrowers who would have received a loan at a higher rate of interest instead faced a situation where they were denied credit entirely.Footnote 31
A final example relates to the regulation of food and other substances. Studies demonstrate how the regulation of one product can induce an undesirable—or at least unanticipated—substitution-effect. For instance, the US Food and Drug Administration established extensive tobacco and nicotine product regulations. In response firms started producing synthetic nicotine products.Footnote 32 And altering the minimum legal alcohol drinking age was found to induce a substitution-effect amongst the youth towards marijuana.Footnote 33
These examples can serve to illustrate the potential temporal considerations and effects underpinning and resulting from law-making, whereby ex-ante legislative and regulatory measures are confronted with ex-post unknown futurity issues. They thus highlight the potential demand and need for clairvoyance if efficacious measures are to be taken ex-ante.
2. Legal Doctrine
Futurity issues may also arise at a doctrinal level. A judicial decision today can bind a judge tomorrow.Footnote 34 Portia, for example, cautioned against the setting aside of Shylock’s bond in Shakespeare’s The Merchant of Venice: “It must not be. There is no power in Venice/Can alter a decree established: /Twill be recorded for a precedent,/And many an error by the same example/Will rush into the state: It cannot be.”Footnote 35 Thus, knowing how legal decision-makers ex-post will react to an ex-ante legal decision would seem to be an important consideration for judges.
We can see the significance of a judicial decision on future decision-makers because of the doctrine of stare decisis. As Schauer maintained:
An argument from precedent seems at first to look backward. The traditional perspective on precedent, both inside and outside of law, has therefore focused on the use of yesterday’s precedents in today’s decisions. But in an equally if not more important way, an argument from precedent looks forward as well, asking us to view today’s decision as a precedent for tomorrow’s decisionmakers. Today is not only yesterday’s tomorrow; it is also tomorrow’s yesterday.Footnote 36
Judges today, therefore, would seemingly need to be mindful of their decision’s impact on tomorrow. Otherwise put, they perhaps would like to be able to predict how today’s decisions can govern tomorrow’s decisions and the practical effects of such outcomes.
This need to be mindful of the future can be seen in such cases that raise “floodgate” worries. As Levy explains: “[A] floodgates argument is an argument against a particular decision on the ground that it will lead to a large number of new claims.”Footnote 37 Various United States (U.S.) Supreme Court cases have expressly wrestled with this futurity concern.Footnote 38 Consider in this respect Wilkie v Robbins.Footnote 39 The plaintiff (Robbins) owned and operated a guest ranch in Wyoming that spanned 40 miles.Footnote 40 Robbins assumed ownership of this land but was unaware of an easement that the previous owner had granted the United States government. The U.S. government had failed to register the easement, which meant Robbins took title to the land “free of the easement, by operation of Wyoming law.”Footnote 41 The U.S. government reacted zealously despite Robbins saying he would consider the grant of an easement.Footnote 42 Various kinds of alleged intimidation and harassment ensued and Robbins sued for damages under a Bivens cause of action, which permits in limited scenarios individuals to claim damages because of improper interference with Fourth Amendment rights by a federal employee.Footnote 43 The U.S. Supreme Court’s hesitance to expand the Bivens claim illustrates its concern with futurity. The Court ultimately rejected Robbins’ claim, grounding its decision on consequentialist “floodgate” worries. Specifically, “at this high level of generality, a Bivens action to redress retaliation against those who resist Government impositions on their property rights would invite claims in every sphere of legitimate governmental action affecting property interests,” which could include anything from tax claim settlements to regulations governing occupational health and safety.Footnote 44 This could generate an “enormous swath of potential litigation”Footnote 45 in the form of an “onslaught of Bivens actions.”Footnote 46 Congress was thus better positioned to deal with the issue, which would reduce the “risk of raising a tide of suits threatening legitimate initiative on the part of the Government’s employees.”Footnote 47 The Court’s reasoning exemplifies the potential role of futurity in judicial decision-making and being mindful of ex-post consequences. As Justice Ginsburg fittingly stated in her partial dissent: “[T]he Court rejects his claim, for it fears the consequences.”Footnote 48
Another example of futurity concerns are the infamous nervous shock tort cases in the United Kingdom—a further demonstration of floodgate concerns and, as such, futurity issues. Indeed, this concern with future floodgate litigation has led courts to narrow claims for nervous shock significantly. Victorian Railways Commissioners v Coultas seminally espoused the “floodgate” concern:
Damages arising from mere sudden terror unaccompanied by any actual physical injury, but occasioning a nervous or mental shock, cannot under such circumstances…be considered a consequence which, in the ordinary course of things, would flow from the negligence of the gate-keeper. If it were held that they can, it appears to their Lordships that it would be extending the liability for negligence much beyond what that liability has hitherto been held to be.Footnote 49
The Court was thus reluctant to extend liability for fear of a deluge of future litigation.Footnote 50 On the basis of these futurity concerns, the subsequent seminal line of cases starting with McLoughlin v. O’Brien Footnote 51 has limited nervous shock doctrine according to particular parameters. In McLoughlin, the claimant argued that the sights and sounds of her injured family at a hospital subsequent to an accident caused her severe psychiatric issues and that these issues should be attributed to the defendant’s negligence. The Court awarded her damages and laid down the narrow parameters within which future cases should be decided—that is, claims of this kind should be limited by class of persons, only those with close family ties, by time and space,Footnote 52 and by the means of shock (third party communication would be insufficient for example).Footnote 53 These limitations were apparently justified because of the concept of “shock’s” inherent capacities for “affecting so wide a range of people.”Footnote 54 Although the effect of this would limit actions only to genuine claims and prevent the future abuse of the nervous shock doctrine—a point that the Court of Appeal had strongly emphasizedFootnote 55 —the House of Lords did not think—in other words, it predicted—that any floodgate worries would manifest.Footnote 56 And yet in Alcock,Footnote 57 for example, the Court declined to broaden the doctrine of nervous shock partly because of floodgate anxieties. In holding that some categories of claimants could claim while others could not, and also clarifying that viewing generalized scenes through the television would be insufficient, the Court may have been, at least partly, mindful of the effects of their doctrinal conclusions on future cases.Footnote 58
In sum, both legal policy and legal doctrine would seem to have the capacity for generating futurity issues for lawmakers and judges. If ex-ante these decision-makers knew with perfect certainty the ex-post outcomes of their decisions, they could construct, alter, limit, and delimit their decisions accordingly. Of course, this type of perfect clairvoyance does not exist in reality; decisions in the present are always to some extent plagued by fundamental uncertainty.Footnote 59 As such, decision-makers need a substitute—some sort of proxy for future outcomes—that they can rely upon to arrive at reasonable conclusions about the future consequences of their decisions in the present. In short, there would seem to exist a demand for a clairvoyance heuristic.
II. Supply of Clairvoyance
It is open to debate about what the most important parameter—explanatory or predictive—is for judging a theory’s value. Explanatory power means the ability of a theory to give observed phenomena meaning by reference to some abstract benchmark. The greater this benchmark’s ability to “decrease the degree to which we find the explanandum surprising,” then the stronger its explanatory power.Footnote 60 Law and Economics’ explanation that common law rules are efficient is illustrative.Footnote 61 For Popper this was the best judge of a theory’s value.Footnote 62 Predictive power, in contrast, does not confer a theory with the ability to give observed phenomena “meaning”Footnote 63 but rather supplies some abstract benchmark with the capacity to make futuristic statements about how phenomena will manifest in response to some change in circumstances. For FriedmanFootnote 64 this was the best judge of a theory’s value: “[T]he only relevant test of the validity of a hypothesis is comparison of its predictions with experience.”Footnote 65 Elster also opined that a theory is in “serious trouble if the event or state of affairs that actually materializes [ex-post] is among those excluded by the theory.”Footnote 66
Law and Economics (L&E) is demonstrative. L&E is premised to varying degrees on the assumption of rationality and based on RCT,Footnote 67 which has value to legal policymakers and analysts because of its tractability and coherence. As Issacharoff explains:
At the heart of law and economics stands the idea that comparisons of utility, the cost and benefit of engaging in a particular course of conduct for a particular actor, can discipline the analysis of how legal rules should be structured. In turn, the disciplined analysis provided by law and economics can direct the regulatory mission of law so as to avoid regulatory missteps.Footnote 68
In short, L&E’s rationality assumption supplies clairvoyance and empowers lawmakers with a basis for “channel[ling] citizen behavior.”Footnote 69 Good laws, consequently, can be chosen ex-ante so that they produce the intended effects.Footnote 70
Consider for example Calabresi’s vignette of car injury costs and the implications of a legal regime that either (1) makes the state bear all car injury costs or (2) levies car injury costs on the person who induced the injury. In the latter regime the cost of the car is $400 (costs of ownership plus insurance) as compared with $250 for alternative modes of transport. A rational profit-maximizer would forego the car because it is the choice that maximizes benefits over costs. In the former regime, there is a mandatory tax of $200 to cover car accident costs, so regardless of whether a car is purchased or not this must be paid. Thus, the total cost of the car—$200—compared with $250 for alternative modes of transport means that a rational profit-maximizer would purchase a car because it is the choice that maximizes benefits over costs.Footnote 71 By assuming rationality ex-ante, lawmakers can anticipate how actors will react and, as such, they can select the legal rule to induce the desired outcome—whatever that outcome might be according to some normative welfare standard.
But how do Calabresi and others know what the actor will do? In other words, what is the source of RCT’s predictive power? Neither economics nor L&E literature seem to have explicitly answered this question—though no doubt the source of a theory’s predictive powers will have been implied. It is to this question that the next Section turns.
C. Endogenous and Exogenous Sources of Predictive Power – The Limitations of Behavioral Economic Theory
BE/BLE has at times been criticized for not effectively identifying “boundary conditions,” in other words, not sufficiently coalescing into a unified field of study that systematically categorizes conditions and contexts that are either more or less likely to give rise to irrational decision-making.Footnote 72 This in turn generates a criticism that BE/BLE proponents too often generalize their field and the BE anomalies that define it and thereby speak “at too general a level,” resulting in BE anomalies becoming characterized as “tendencies” but hiding the ceteris paribus pre-requisites for predicting the incidence of such anomalies.Footnote 73
At first glance this criticism of BE/BLE may seem unwarranted because RCT/L&E also makes ceteris paribus claims in its own predictions and hypotheses. This has been cited as necessary in order to deal with the volume and complexity of reality.Footnote 74 As Daniel Hausman explains: “Explicit or implicit ceteris paribus clauses are pervasive in economics.”Footnote 75 For example, market actors do not always switch between substitute products if price increases; they may switch ceteris paribus. Therefore, both BE/BLE and RCT would seem to be equally dependent on ceteris paribus qualifications for making their predictions. To put the point another way, one might say that neither BE/BLE nor RCT can predict endogenously; only exogenously.Footnote 76
Yet this raises a question: if both these methodologies are as equally incompetentFootnote 77 as the other, why is RCT the default and BE/BLE secondary; or, why, as Posner observes, is BE/BLE the “residual” of RCT and not vice-versa? Why is it that when people act rationally this is not identified as a contradiction of BE?Footnote 78 In answering these questions, Section C seeks to dispel the observation that RCT can only predict exogenously. First, the co-dependent nature of BE/BLE phenomena is illustrated, demonstrating how predictions about such phenomena seemingly depend on idiosyncratic knowledge and conditions. Second, and contrary to what the ceteris paribus clauses “pervading” economics may suggest, a normative interpretation of RCT is advanced and, in doing so, demonstrates how RCT predictions can be made independently. As such, it is argued that RCT can predict endogenously. These analyses provide the foundation for answering the question posed at the end of Section B: What is the source of RCT’s predictive power?Footnote 79 Thus, in Section D it will be argued that it is BE’s/BLE’s lack of any normative foundation that limits it to predicting exogenously—that boundary conditions would need to be known ex-ante if clairvoyant claims about irrational behavior are to be made. It therefore seeks to highlight the limits of BE/BLE in its “quest for predictive performance.”Footnote 80 It is concluded that it is for this reason BE is the “residual” of RCT.Footnote 81
I. BE/BLE’s Idiosyncratic Co-dependence
In the vignettes that follow, the exogenous nature of BE’s predictive power is showcased. It is argued that this exogeneity derives from the fact that ex-ante, much would need to have been known in order for the prediction to have been made. In other words, BE’s predictive power is exogenously sourced—often co-dependent on specific events and knowledge.
1. Judgment Under Uncertainty
Seventy-two percent of respondents violated the “conjunction rule” of probability when asked to make predictions about Björn Borg winning the Wimbledon 1981 finals.Footnote 82 The conjunction rule in basic terms states that the probability of two events happening together cannot have a higher probability of manifesting than one of those constituent events. To understand why this must be the case, reflect on whether two events happening simultaneously is more likely than one of those events happening by itself. Is the likelihood of (a) your long-lost Aunt showing up one random Saturday more likely than (b) your long-lost Aunt showing up one random Saturday and you winning the lottery? Two events happening or manifesting together surely must be less likely than one of those events happening by itself.
Taking this latter example of Bjorn Borg, Tversky and Kahneman hypothesized—predicted—what respondents would do. They stated: “After winning his fifth Wimbledon title in 1980, Borg seemed extremely strong. Consequently, we hypothesized that Outcome C would be judged more probable than Outcome B, contrary to the conjunction rule ….”Footnote 83 Outcome C stated that Borg will lose the first set but win the match. Outcome B stated that Borg will lose the first set.Footnote 84 Tversky and Kahneman’s predictions transpired to be true – that is, subjects rated as more probable two events happening together—Outcome C—than one of those events—Outcome B— manifesting.Footnote 85
Consider, however, the source of Tversky and Kahneman’s clairvoyance; how did they know ex-ante what respondents would choose? It was seemingly the fact that Borg had won Wimbledon in each of the last five years. This “representativeness” rendered the “conjunction” in this scenario—Borg will lose the first set but win the match—more probable than its constituent part —Borg will lose the first set—“violating the conjunction rule in a direct test.”Footnote 86 It could be argued that without this information—that is, without the knowledge of Borg’s previous wins—Tversky and Kahneman may not have been able to predict respondent’s choices accurately. More bluntly, it could be argued that their prediction of representativeness-based decision-making depended on knowledge of the surrounding context. Moreover, it seemed to depend on idiosyncratic knowledge concerning the situation at hand.
This co-dependence in prediction may be seen in various other BE representativeness heuristic experiments. Consider the Linda problem espoused by Tversky and Kahneman to further illustrate conjunction rule contradictions, in other words, some respondents based their judgment on representativeness rather than probability. The problem states: “Linda is 31 years old, single, outspoken and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.”Footnote 87 And here are the answers—to be ranked by a group of 88 subjects according to probability—likelihood:
Linda is a teacher in elementary school; Linda works in a bookstore and takes Yoga classes; Linda is active in the feminist movement (F); Linda is a psychiatric social worker; Linda is a member of the League of Women voters; Linda is a bank teller (T); Linda is an insurance salesperson; Linda is a bank teller and is active in the feminist movement (T&F).Footnote 88
Contrary to what standard RCT/probability theory would predict—because every feminist bank teller is also a bank teller, the probability of T must be greater than T&FFootnote 89 —the results “confirmed” the predictions of Tversky and Kahneman.Footnote 90 Specifically, eighty-five percent of respondents produced a rank F>T&F>T, thereby illustrating the “conflict between the intuition of representativeness and the logic of probability.”Footnote 91 Further, so as to demonstrate that the result was not simply the product of undergraduate statistical naivety, the experiment was given also to PhD students at the Stanford Graduate School of Business, where eighty-five percent fell foul to the error.Footnote 92 Perhaps more starkly, when the test was made even more transparent so as to try and induce the correct result—“which two alternatives is more probable: Linda is a bank teller (T); or Linda is a bank teller and is active in the feminist movement (T&F)”—85 percent of undergraduates ranked T&F>T, “in a flagrant violation of the conjunction rule.”Footnote 93 These results are violations of the conjunction rule because, in a similar vein to the Bjorn Borg illustration above, uniqueness and specificity are the anti-thesis of probability. To understand this, simply consider which is more probable: your child’s high school teacher having (a) an undergraduate degree or (b) an undergraduate degree specializing in Pedagogy, a master’s degree in Education, and a PhD in Social Studies.
Now let us examine again what the source of predicting these representativeness heuristic outcomes might be? Just like in the Borg example, we can see context-specificity playing an influential role. Because research demonstrates that stimuli and information are stored and categorized along lines and sub-lines of mental models and schemata,Footnote 94 “it is therefore natural and economical for the probability of an event to be evaluated by the degree to which that event is representative of an appropriate mental model.”Footnote 95 However, knowing what is representative is clearly required before any predictions about outcomes may be made. Otherwise put, the “essential” (necessary) and “salient” (sufficient) features of the larger group to which the event “belongs” would need to be known ex-ante if judging outcomes under uncertainty are to be made according to this mental shortcut.Footnote 96 Moreover, the requisite features are seemingly idiosyncratically-linked to the scenario in issue, just like in the Borg example. Indeed, predicting the outcomes of the above student experiments would require some knowledge of feminism, the underlying schemata of this concept, and in particular its specific essential and salient features.Footnote 97
Another Tversky and Kahneman study can help illustrate this point about idiosyncratically co-dependent clairvoyance in predicting BE phenomena like the representativeness heuristic. Tversky and Kahneman “expect,” for instance, that a random sample will be subjected to the representativeness heuristic where the sample “preserves the majority-minority relation in the population.”Footnote 98 So for example, where two high school programs, A and B, contain a majority of boys—65 percent—and a majority of girls—55 percent—respectively, it can be predicted that subjects will fall foul to the representativeness heuristic when identifying a random class as either A or B based on whether that class has more boys or girls.Footnote 99 This prediction, however, would presumably be dependent upon a context-specific benchmark—knowing the majority and minority distribution in a class—the absence of which would render it more difficult to ex-ante predict subjects falling prey to the representativeness heuristic.
To some extent Tversky and Kahneman allude to this point in their definition of representativeness—that is, they demonstrate that some independent benchmark must first be known before conclusions about representativeness can be made. They explain: “Representativeness is an assessment of the degree of correspondence between a sample and a population, an instance and a category, an act and an actor or, more generally, between an outcome and a model.”Footnote 100 For example, Kahneman and Tversky would perhaps say that afternoon tea—the instance—is representative of British culture—the category—i.e. afternoon tea is typical or representative of what society might understand British culture to be. Thus, the essential and salient features would also seemingly need to be known ex-ante, just like when a specific person represents a broader “social group if his or her personality resembles the stereotypical member of that group.”Footnote 101
Other judgment heuristics demonstrate these idiosyncratically co-dependent predictive capabilities. The availability heuristic may be an unsurprising candidate as it is a close cousin of the representativeness heuristic. This heuristic predicts that individuals will make probability judgments according to the ease with which a similar event springs to mind.Footnote 102 For instance, divorce rates may be predicted by easily recalling divorces of friends.Footnote 103
Consider Tversky and Kahneman’s study of respondents tasked with judging word letter frequency. Respondents were asked whether the letters K, L, N, R, V are more likely to appear in the first or third position of English language words. Tversky and Kahneman hypothesized—predicted—that subjects would judge probability in a way that was “mediated by assessed availability,” in other words, the ease with which recalling words starting with the respective letter would be the driving force behind ranking that letter as one contained very frequently at the beginning of English language words.Footnote 104 The results contradicted reality—the letters under scrutiny were in fact contained more often in the third rather than first position—and confirmed Tversky and Kahneman’s prediction. Out of 152 subjects, “105 judged the first position to be more likely for a majority of the letters … despite the fact that all letters were more frequent in the third position.”Footnote 105 This idiosyncratic-contextuality can be seen in the source of Tversky and Kahneman’s prediction. In order to hypothesize in the manner that they did, Tversky and Kahneman would have needed to know that the letters K, L, N, R, V spring very easily to mind and, thus, would likely mean that subjects would fall foul to the availability heuristic. Note, however, that the availability heuristic may be more generalizable than the representativeness heuristic in the sense that it may be assumed that all individuals will rely on this shortcut in making judgments, and the shortcut’s content requires simply knowing that easy recall may often be used as a substitute for frequency. However, it still would require knowing ex-ante whether a specific scenario—for example tornadoes as a cause of death—springs to subjects’ minds more easily than other, more statistically significant causes of death.Footnote 106
A final illuminating example of this idiosyncratic crutch that BE seemingly needs to lean on in its clairvoyant endeavors is the concept of availability cascades and how salience can influence judgments about probability.Footnote 107 What this illustration demonstrates is that in order to know if the availability heuristic will be relied upon, knowledge of idiosyncratic features of events will be needed ex-ante. Indeed, if there was any broad-based proposition with predictive policy value arising out of Kuran and Sunstein’s paper, it would be that widespread public reporting can generate artificial demands for risk regulation by inducing “mass delusions” which may “last indefinitely,” resulting in an erroneous regulatory response comprising “wasteful or even detrimental laws and policies.”Footnote 108 Otherwise put, because we are all boundedly rational, our ability to perfectly verify all claims made by the media is very low—we can only assume they are true but we cannot observe and verify in reality the events being reported because doing so would take a lot of time and resources, perhaps prohibitively so. Thus, if we see reporting of certain incidents frequently and vividly enough, our minds cannot help but think that what is being reported is true—when in fact it may not be.
Take for example the Love Canal illustration of an availability cascade “that [has] resulted in socially harmful regulatory responses.”Footnote 109 After a chemical company dumped waste into a waterway and covered it with dirt, it sold it to the Niagara Falls Board of Education. Local government developed this area and built residential houses. Fast forward twenty years and a raft of public reporting on how Love Canal had harmed fish, burnt children, rendered water undrinkable, caused nervous breakdowns, induced miscarriages, and resulted in urinary system diseases—all of which contributed to a global sense of fear and anxiety. True, some of the latter effects were based on studies but ultimately these were discredited. Television documentaries entitled The Killing Ground and newspaper reports all seemed to be the driver behind political capitulation, despite statistical science demonstrating that Love Canal was of very low risk. Kuran and Sunstein’s main point is that the wave of attention accorded to Love Canal was a function of the artificial demand for political action which in the end resulted in wasteful regulatory responses that could have put resources to better use on real issues like smoking.Footnote 110 This is an amplified version of the availability heuristic—no individual member of society could in fact check and verify that children were being burnt, water was undrinkable, or that miscarriages were induced; they could only assume that this was in general the case because it seemed as if these events were widespread due to persistent and vivid media reporting.
While Kuran and Sunstein have provided a colorful and convincing synthesis of the availability heuristic on steroids, it would not seem to supply theoretical predictive power in and of itself. The contextual contours of Love Canal could provide third-party benchmarks for regulating, for instance, the modality, velocity, and intensity of reporting that the media may be permitted to engage in—especially in categories that share the idiosyncratic features of Love Canal—for example, public health, toxic waste sites, and locality populous with children. To be sure, this is a valuable insight; but it is an insight, nonetheless, that seems to depend on certain conditions—thereby exemplifying its co-dependence from the perspective of predictive power. Indeed, Kuran and Sunstein would seem to attest to this co-dependence when they enumerate the conditions necessary for availability cascades to arise: an incidence of a phenomenon may be subjected to a strong version of the availability heuristic when its disclosure is salient, vivid, and possibly alarmist. Further, the medium through which the incidences are reported would probably need to be sufficiently ubiquitous, like newspapers or television.Footnote 111 The idiosyncratic risk preferences, which may be conditioned by various social factors, would similarly need to be known ex-ante if availability cascades are to be predicted.Footnote 112 This idiosyncratic contextuality—that is, the conditions necessary for predicting confidently its manifestation—may be made more acute when one realizes that the magnitude of an availability cascade can also be shaped by other heuristics and biases like loss aversion.Footnote 113 And Kuran and Sunstein perhaps most pointedly reveal the idiosyncratic contextuality and variability of availability cascades when they state:
In contexts subject to availability cascades, the distribution of beliefs across the relevant population may entail multiple equilibria. In other words, the same objective information may be capable of sustaining different, even extremely different, belief patterns, depending on whether a cascade occurs and, if so, which of many possible cascades is initiated. One risk may gain salience, receive an enormous amount of attention, and become the object of tight regulation, while another risk, which experts deem equivalent, is treated as “part of normal life.”Footnote 114
Thus, “dramatically different” risk responses across “culturally and economically similar” countries may not be surprising.Footnote 115
2. Decision-Making
Idiosyncratic co-dependent predictions may also be seen in some of the decision-making biases and heuristics, which demonstrate examples of individuals contradicting RCT. The endowment effect, loss aversion, and status quo bias can be illustrativeFootnote 116 because they share commonality in the sense that initial reference points play a pivotal role in their apparent manifestations. As Kahneman explains, “[r]eference points are important because other outcomes are compared to them, and are coded and evaluated in terms of this comparison.”Footnote 117 Switching jobs, for instance, would thus be assessed by reference to “pluses or minuses relative to where you were.”Footnote 118 In other words, starting points—relative to a subsequent decision—matter.
Take as demonstrative the famous Cornell coffee mugs experiment, which showed willingness-to-accept (WTA) was greater than willingness-to-pay (WTP) and thereby contradicted standard RCT’s assumption that this gap is minute in the presence of small income differences.Footnote 119 This is the endowment effect, which posits that an individual will value a good more once they are endowed with it and, consequently, their WTA > WTP. Mugs were distributed randomly and seller information about the mugs was perfect.Footnote 120 Subjects—sellers and buyers respectively—were prompted with the assumption: “You now own the object in your possession. [You do not own the object that you see in the possession of some of your neighbors].”Footnote 121 The results were stark. Selling prices were more than double buying prices and the actual to predicted volume ratio was .20—with this low ratio reflecting the endowment effect hypothesis that individuals are less willing to part with goods they own than they are willing to purchase goods they do not initially own.Footnote 122 To understand why this result was so significant, you only have to realize the following: both buyers and sellers were ascribing different values to the same mug. Ownership, then, can play seemingly powerful roles in our value perceptions and decision-making.
Similar to the judgment examples above, let us now identify the source of predicting the endowment effect. How would one know ex-ante whether the endowment effect would govern an actor’s decision-making? A good starting point might be to leverage the loss aversion bias—that individuals feel losses more than they feel gains—and one could predict WTA > WTP. However, consider the idiosyncratic pre-requisites that would need to be known ex-ante in order to make such a prediction. Both the type of good in issue and a seller’s initial reference point would be required. As to the type of good, some goods are “expected” to not be subject to the endowment effect, like goods exclusively “purchased for resale rather than for utilization.”Footnote 123 Both Plott and Smith demonstrated this “induced-value” experimental methodFootnote 124 —where because nominal tokens are only good for cash redemption and have no outside option, nor are they likely to have any immediate short-term outside option, an endowment effect should not be expected.Footnote 125 Knowing the type of good idiosyncratic to the situation at hand would thus seem to be necessary in order for successful ex-ante endowment effect predictions. Regarding reference points, the idiosyncratic pre-requisites necessary for making ex-ante predictions may be seen as particularly pronounced when one considers a labor wage negotiation scenario. Neale and Bazerman, for example, identify five reference points that would seemingly be required to be known if one were to ex-ante predict a union’s response to a wage offer. These include the previous year’s wage, union’s estimation of management’s reservation point, the initial offer management puts forward, the union’s reservation point, and the union’s publicly announced bargaining position.Footnote 126 Circumstances specific to the negotiating parties at hand would thus seem to be significant for successful ex-ante predictions.
Reference points become especially weighty in decision-making when they represent the status quo.Footnote 127 The status quo can be understood as the option that is perhaps presented first and/or framed and presented as the most legitimate option—like the default out-of-the-box web browser installed on your Apple iPhone, which has recently given rise to significant antitrust litigation in both the EU and US.Footnote 128 Samuelson and Zeckhauser examine sets of scenarios to test whether an option will be selected based on its designation as being the “status quo” and find that “individuals disproportionately stick with the status quo.”Footnote 129 This of course represents a departure from RCT because RCT says that “neither the order in which the alternatives are presented nor any labels they carry should affect the individual’s choice”; only “preference-relevant” features should govern.Footnote 130 Yet in offering example after example where labels do affect individual choice—specifically, status quo labels—Samuelson and Zeckhauser in one sense confirm the idiosyncratic co-dependency of BE in its predictive endeavors. The researchers used questionnaire experiments where subjects were braced with a decision facing an individual, for example, a manager or a governmental policymaker, and there were “mutually exclusive” options from which to choose. Two versions of questions existed: a “neutral” version—where alternatives were presented “on an equal footing”—and a “status quo” version – where one alternative “occupies the position of the status quo.”Footnote 131 For example: you inherit a large sum of money from your uncle and are now considering four different portfolios to invest in–a moderate-risk company, a high-risk company, treasury bills, and municipal bonds. The “neutral” version of this question did not mention status quo; the “status quo” version labeled the moderate-risk company as “[a] significant portion of [your uncle’s] portfolio is [already] invested in moderate-risk Company A.”Footnote 132 Aggregating across many experimental designs and questions, Samuelson and Zeckhauser found that “individuals disproportionately stick with the status quo.”Footnote 133
These illustrations, however, again seem to highlight the idiosyncratic co-dependent nature of BE’s predictive power. One would need to know the ex-ante reference point—the status quo option, in particular—before hypothesizing the outcome of individual choices. Otherwise put, knowing the idiosyncratic contextual frame of the options would seem to matter.Footnote 134
3. Unification Attempts
One potential limitation in the above line of argument might be that focusing just on isolated BE biases and heuristics will of course generate critical propositions about BE’s predictive power. Instead, one should examine BE unification attempts in assessing BE’s predictive power. After all, the biases and heuristics themselves are not unitary but rather more like “a series of particular counter-stories, formed largely in parasitic reaction to the unduly self-confident predictions of rational choice theorists.”Footnote 135
There have been several attempts at unifying BE phenomena,Footnote 136 ranging from biological to neuro-scientific—sometimes known as “dual-process” theories—to methods that simply abstract away from BE phenomena in more classificatory ways.Footnote 137 There is also of course prospect theory, which advances “[a]n alternative theory of choice … in which value is assigned to gains and losses rather than to final assets and in which probabilities are replaced by decision weights.”Footnote 138 Some have even made efforts to improve upon prospect theory in the form of modeling decisions and choice based on assuming that individuals anticipate “feelings of regret and rejoicing”Footnote 139 which is simpler and more intuitive than prospect theory.Footnote 140
The value of these unifications, however, will in part be dictated by their capacity to generate generalizable predictions.Footnote 141 To what extent these unifications can do so without idiosyncratically depending on particularized information and conditions will now be examined.
3.1. Prospect Theory
As stated, prospect theory represents a description of decision-making under uncertainty and differs from RCT in that it states that determinants of welfare are evaluated not on the basis of final outcomes but rather changes in wealth—losses or gains—relative to some initial reference point.Footnote 142 One fundamental proposition of prospect theory is derived from loss aversionFootnote 143 and posits that an individual’s value function is “generally concave for gains and commonly convex for losses” and additionally that the value function is “steeper for losses than for gains.”Footnote 144 Informally, gaining $100 may be accorded less weight and value than losing $100. More intuitively, the pain of losses is felt more than the pleasure of gains.Footnote 145
For example, suppose you choose to pursue a master’s degree in a foreign country: your experience of this country’s cold weather may differ to some of your peers because they hail from countries with colder climates than yours. The reverse may also be true–your experience of the foreign country’s warm season may be felt less by you compared to your peers, who are not used to such humid climates. Kahneman and Tversky point out that “[t]he same principle applies to non-sensory attributes such as health, prestige, and wealth.”Footnote 146 Indeed, a specific level of wealth “may imply abject poverty for one person and great riches for another—depending on their current assets.”Footnote 147 However the wealth change—plus or minus—may be accorded less or more value respectively.
While prospect theory can be said to have produced “remarkable insights,”Footnote 148 it would not be unreasonable to say that its penetration of applied economics has been slow.Footnote 149 Save for some moderate successes in finance and insurance (which is unsurprising given they are classic cases of decision-making under risk), prospect theory has been sluggish in making inroads into other fields.Footnote 150 Some argue that the reason for this languid progress is because “it is hard to know exactly how to apply [prospect theory].”Footnote 151 This is particularly relevant for present purposes because it implies a lack of generality in its application and, consequently, contextual co-dependence. As Barberis explains:
The central idea in prospect theory is that people derive utility from “gains” and “losses” measured relative to a reference point. But in any given context, it is often unclear how to define precisely what a gain or loss is, not least because Kahneman and Tversky offered relatively little guidance on how the reference point is determined.Footnote 152
An example from finance may demonstrate this idiosyncratic co-dependence for the purposes of predictive power. How might we predict the portfolio “an investor with prospect theory preferences will hold[?]”Footnote 153 If prospect theory is based on changes in welfare rather than ultimate welfare states, then immediately we would need to know about the kinds of gains and losses the investor is considering. For instance, they could be gains or losses in relation to “overall wealth”, total stock market holdings, or the value of specific stocks. In the case that it is stock market holdings, is a gain represented by positive stock market returns, returns that exceeded “the risk-free rate, or the return the investor expected to earn?”Footnote 154 Besides specifying the content of these gains or losses, an additional temporal consideration would be whether the investor is concerned with gains or losses annually, monthly, or even weekly.Footnote 155
Some researchers have made attempts to surmount this shortcoming and make prospect theory more durable across contexts.Footnote 156 Otherwise put, they have attempted to move towards generality and away from idiosyncrasy. However, in what might be viewed as borrowing from RCT, these efforts may only serve to further highlight prospect theory’s co-dependence. Take for example Kószegi and Rabin, who proffer a model of “reference-dependent preferences,” building on Kahneman and Tversky’s prospect theory in an effort to “develop a more generally applicable theory.”Footnote 157 However, their model is dependent on standard RCT in the form of standard consumption utility. Specifically, “[a] person’s utility depends not only on her K-dimensional consumption bundle c but also on a reference bundle r. She has an intrinsic ‘consumption utility’ m(c) that corresponds to the outcome-based utility classically studied in economics.”Footnote 158 More intuitively, in making the reference point depend on beliefs about outcomes, the authors are able to proffer a more generalizable model because the reference point becomes less idiosyncratic.Footnote 159
For example, in a gamble between $100 and $0, “an outcome of $50 feels like a gain relative to $0, and like a loss relative to $100, and the overall sensation is a mixture of these two feelings.”Footnote 160 Otherwise put, the reference point can be determined endogenously because the “model makes the extreme assumption that the reference point is fully determined by the expectations a person held in the recent past.”Footnote 161 Specifically, “a person’s reference point is her probabilistic beliefs about the relevant consumption outcome held between the time she first focused on the decision determining the outcome and shortly before consumption occurs.”Footnote 162 A more intuitive example is illustrative: suppose a consumer desires a new pair of football boots. If they expected to buy the boots ex-ante, a loss would be felt if they did not buy and, moreover, this loss aversion will induce them to purchase at higher prices. In contrast, if the consumer did not expect to buy, a purchase would result in lost money, and this loss aversion will lead them to not purchase.Footnote 163
However, “[b]y directly constructing reference-dependent utility from consumption utility and assuming that the reference point is endogenously determined as rational expectations about outcomes,” Koszegi and Rabin borrow from RCT in their attempts at rendering prospect theory more “portable and generally applicable.”Footnote 164 In doing so, they would seem to only exacerbate the notion that even a more unified understanding of BE like prospect theory may not be enough to save BE from its co-dependent predictions. Indeed, this proposition—BE’s idiosyncratic co-dependence—may be one reason why scholars more often suggest co-mingling BE and RCT models rather than touting BE as a standalone replacement.Footnote 165
3.2. Dual-Process Theory
Judgment and decision-making anomalies have also been subjected to a theorization attempt called dual-process theory (DPT), commonly known also as System 1 and System 2.Footnote 166 DPT posits that two separate mental processes control our “human cognition,” one of which is deliberate, controlled, slow, analytical, conscious, and effortful while the other is automatic, unconscious, associative, and fast.Footnote 167 Heuristics and biases research generally agrees on the existence of these two mental processes.Footnote 168
One criticism of DPT—one which is particularly pertinent for this Article’s purposes— is that it lacks the capability to predict when individuals will make judgments or decisions under irrational, rather than rational, processes. As Grayot explains:
[T]he standard view of DPT does not actually provide an account of how reasoning tasks are accomplished, and decisions made; what it provides is a generic theory about the potential origins of reasoning and decision errors. This, it would seem, is a major deficiency for the theory: if it cannot explain how the mind inhibits or overrides bad judgments that are generated by rapid or automatic mental processes, then what is the point of making the distinction to begin with? After all, we don’t always submit to our biases—we are often able to restrain gut-reactions and to recognize hasty errors for what they are.Footnote 169
An example of this criticism may be seen in Kahneman’s exposition about System 1 accessibility–that “[h]ighly accessible impressions produced by System 1 control judgment and preferences, unless modified or overridden by the deliberate operations of System 2.”Footnote 170 Kahneman’s point is that BE heuristics and biases—decision-making shortcuts that are used instead of more intentional and considered cognitive processes—are cognitively easy to access and process and, therefore, may easily influence judgment and decision-making.Footnote 171 BE phenomena that have been shown to be easily accessible include: losses and gains—which are more easily accessible and influential than absolute outcomes on decision-making and judgment;Footnote 172 framing effects—because rendering aspects of a decision more salient and vivid can lead to greater influence of those attributes;Footnote 173 and representativeness—where similar phenomena that arise “more readily to mind” are substituted for judgments about probability.Footnote 174 An example of the latter is The Tom W experiment,Footnote 175 which is a similar study to the Linda problem and demonstrates how stereotypes may be “highly accessible natural assessment[s], whereas judgments of probability are difficult.”Footnote 176 This is somewhat unsurprising because the very nature of a stereotype is its default status as the mental model first and automatically relied upon to interpret an incoming stimulus like a situation or person.Footnote 177
Thus, Kahneman has offered an admirably descriptive theory that unifies some of the most significant BE phenomena. However, while accessibility assumptions can clearly do “genuinely explanatory work,”Footnote 178 their predictive efforts would seem to suffer from the same co-dependence as prospect theory. This is perhaps unsurprising, because “descriptive accuracy is purchased at a price, the price being loss of predictive power.”Footnote 179 This co-dependence may be seen in the ex-ante pre-requisite of knowing what determines accessibility, a point that Kahneman himself identifies.Footnote 180 Accessibility can, for example, oscillate according to factors like salience, the specific characteristics of the objects or decisions being assessed, and “[m]otivationally relevant and emotionally arousing stimuli.”Footnote 181 Consequently, if cognitive accessibility is the basis for irrationality and System 1 thinking governing decision-making and judgments,Footnote 182 DPT’s predictive powers would seemingly be co-dependent if specific knowledge about accessibility and its variance is a pre-requisite.
Moreover, DPT’s co-dependence may be somewhat exacerbated by the fact that System 2 as a remedial mechanism for System 1 thinking may also change according to context. Kahneman and Frederick showcase this point by proffering conditions under which System 2 may fail to override System 1. The point is made more explicit by Kahneman himself when he states “[i]n the context of an analysis of accessibility, the question when intuitive judgments will be corrected is naturally rephrased: When will corrective thoughts be sufficiently accessible to intervene in the judgment?”Footnote 183 Indeed, the highly variable conditions that can determine System 2 remedial efficacy may only amount to additional evidence of the idiosyncratically co-dependent nature of DPT in predicting irrational conduct. Some illustrations of this variance in System 2’s effectiveness include time-pressured decisions,Footnote 184 multi-tasking, decisions at a time of day that reflects the decision-maker’s idiosyncratic preferences,Footnote 185 and a decisionmaker’s emotional state—a determinant with high levels of variance.Footnote 186 Contrastingly, a stronger System 2 is more often associated with intelligence,Footnote 187 and, unsurprisingly, education in statistical reasoning.Footnote 188
One might conclude, then, that if the “comprehensive list” of factors that may influence accessibility—and hence the likelihood of (ir)rationality—is “long,”Footnote 189 then DPT would seem to be co-dependent in its predictive power capacities and should be judged as such in its value as a theory.
II. RCT’s (In)dependence
Our discussion so far has centered on the potentially handicapped nature of BE as a theory in generating accurate predictions. The foregoing sub-section highlighted how predicting BE phenomena ex-ante depended on idiosyncratic specifications. It also demonstrated how even at the level of unitary conceptions of BE idiosyncratic information would still likely be required if successful ex-ante predictions are to be achieved. Otherwise put, BE’s predictive power is seemingly exogenously sourced.
The discussion now takes a corollary turn by asking to what extent RCT—the opposite of BE—as a theory of choice and decision-making can generate useful predictions without having to rely on the presence of idiosyncratic boundary conditions. In other words, to what extent can RCT predict endogenously? Some scholars argue that RCT cannot achieve this. Their doubts are grounded in arguments about both the structure and content of preferences. For example, Brennan contends that because RCT neither specifies (1) whether an agent will utility maximize nor (2) the content of their preferences, it will lack “any predictive bite at all.”Footnote 190 This lack of predictive power due to insufficient knowledge of preference structure and content may perhaps be exacerbated by the notion that we are “multiple selves” with a variance in preferences across time.Footnote 191
This sub-section seeks to dispel this doubt. An argument is advanced that RCT possesses normative faculties—faculties that we will soon see BE does not possess—upon which preference structure and content can be inferred “independently.”Footnote 192 Building on these discussions, Section D below advances the argument that RCT rests on a ubiquitous premise such that all –most—agents’ preferences—both structurally and substantively—can be endogenously predicted ex-ante.
The fact that the concept of rationality—in all its theoretical forms—has been subjected to much debate may be seen as a signal of its nebulousness. As Etzioni notes, “over seventy different definitions have been advanced.”Footnote 193 These range from the pursuit of self-interest, in several forms, and utility maximization, again susceptible to several forms, to acting with internal consistency.Footnote 194 As Korobkin and Ulen surmise, “there are probably nearly as many different conceptions of rational choice theory as there are scholars who implicitly employ it in their work.”Footnote 195
One relatively agreed upon categorization, however, is that RCT is a normative theory–that is, it is underpinned with assumptions and prescriptions for how individuals ought to act.Footnote 196 As Hausman and McPherson note: “Utility theory lays down formal conditions that choices and preferences ought to satisfy … t]o define what rational preference and choice are is ipso facto to say how one ought rationally to prefer and choose.”Footnote 197 Even BE proponents recognize these normative foundations of RCT.Footnote 198 This sub-section elucidates RCT’s normativity at three levels: axioms; models; and values. It describes the unifying goal as the maximization of benefits over costs.
1. Axioms
The axioms of rationality are all examples of the normative underpinnings of RCT.Footnote 199 They assume, for instance, with the axiom of commensurability that actors should be enabled to compare the “utility consequences of all alternatives ….”Footnote 200 This is also known as completeness.Footnote 201 The axiom of dominance further demonstrates RCT’s normative underpinnings.Footnote 202 This axiom states that an individual should “never choose an option in which every feature is only as good as the features of a competing option, and at least one feature is not as good.”Footnote 203
Another axiom—transitivity—perhaps better serves to demonstrate RCT’s normative foundations. This axiom assumes that if X is preferred to Y and Y is preferred to Z, then X should also be preferred to Z.Footnote 204 An illustration of why your preferences should be transitive is the money-pump argument, which identifies a “guaranteed disadvantage” if you were to violate the axiom of transitivity.Footnote 205 Suppose you prefer Y to X, Z to Y, and X to Z–thus contradicting the transitivity axiom. It can therefore be assumed you would be willing to pay some money to replace X with Y. It can similarly be assumed you would be willing to pay some money to replace Y with Z. And it can be inferred you would pay to replace Z with X. However, you are now in the position of possessing the option you began with “but with less money.”Footnote 206 In other words, you have acted irrationally.Footnote 207
Invariance is one final demonstration of RCT’s normativity. This axiom states that “the preference between two or more choices should not depend on how the choice is presented or structured, so long as the outcome possibilities are constant.”Footnote 208 One study, for example, examined the effect of price frames on consumer decision-making. It found that despite decision problems remaining unchanged, consumer surplus was reduced by 25 percent in the event that the price was dripped, that is, revealed at elongated intervals.Footnote 209 The framing effect has been demonstrated in many other studies also, illustrating the idea that how a choice is framed does matter irrespective of outcomes remaining the same.Footnote 210
2. Models
In addition to RCT’s underlying axioms, the most popular versions of RCT all share the common trait of normativity. These include means-ends RCT, expected utility theory, and wealth maximization.
Consider for example RCT as “suiting means to ends” in the least costly way.Footnote 211 Suppose an individual is sitting in a room with glass windows and a door. The windows cannot be opened from the inside. The door is slightly ajar. The individual announces that they are leaving the room. The least costly method is simply to walk through the door, so it should be chosen; no rational person would jump through a glass window and risk serious bodily injury.
Expected utility theory also exemplifies RCT’s normative foundations because it is fundamentally an “explicit or implicit cost-benefit analysis.”Footnote 212 Suppose you want to leave the house for a walk but are unsure whether to take your umbrella.Footnote 213 The decision—which as we will see is a normative decision—can be analyzed in light of outcomes, states, and acts. There are three possible outcomes: dry and unencumbered; dry and encumbered; wet. There are two states: raining; not raining. And there are two acts: bring umbrella; leave umbrella at home. The decision is a decision under uncertainty because of the risk of rain. Consequently, the choice will be based on the expected utility of each outcome, with the highest utility decision dominating the choice. Expected utility theory thus “provides a way of ranking the acts according to how choice worthy they are: the higher the expected utility, the better it is to choose the act. (It is therefore better to choose the act with the highest expected utility…).”Footnote 214 In this sense, it is “a theory of how people should make decisions.”Footnote 215
3. Values
Wealth-maximization—another version of RCT—similarly exhibits normative foundations because it dictates how actors should choose. This RCT version posits that actors will make decisions that maximize their financial well-being. It has even been proposed as “a normative theory of law.”Footnote 216 Its normativity can be seen in the implicit assumption that an individual would prefer more to less wealth and, consequently, decisions that maximize the former will be chosen. Take for example an actor who has two job offers–one of which offers a higher salary. Assuming wealth is the exclusive goal of the actor, the job with the higher salary will—should—be chosen.
Common across all of these components and conceptions of RCT is the maximization of benefits over costs. They all dictate in one form or another that an agent should maximize benefits over costs. But in what sense might this preference structure generalize across all individuals? And in what sense can this structure be populated with generalizable content? Such questions would seemingly need to be answered if the proposition that RCT can predict endogenously is to hold any weight. Section D therefore turns to answer these questions.
D. Generalizing RCT’s Cost-Benefit Normativity
At this Article’s outset, it was acknowledged that the significance of generalizability for a theory can be seen in legislative and regulatory contexts. Legal regimes are “designed to serve a large number of addressees for long periods of time and to cover a great variance of cases.”Footnote 217 Laws are also meant to apply equally to everybody.Footnote 218 Thus, having a general theory about how all, or, at least most, humans behave and will respond to governance measures is of obvious value when governing and ordering the behavior of society in general.
Building from the analysis in Section C(II), might individuals’ preference structures exhibit cost-benefit analyses most of the time? And can those cost-benefit preferences be populated with generalized content? If these questions can be answered affirmatively, we move some way towards conceiving of RCT as a theory with endogenous predictive power.
I. Generalized structures and content in RCT
As to the first question, it could be argued that individuals always—or mostly—engage in cost-benefit analysis due to the familiar concept of fundamental scarcity. Mullainathan and Shafir have vividly demonstrated, for instance, that we live in a world of ubiquitous scarcity.Footnote 219 Scarcity, as the authors highlight, is literally everywhere, forming “a common chord across so many of society’s problems.”Footnote 220 Everything from finite food, time, space, and money to “a lack of companionship” pervades our lives.Footnote 221 For example, losing a job generates financial scarcity, which contributes to household budget scarcity, which in turn results in “too little income to cover the mortgage, car-payments, and day-to-day expenses.”Footnote 222 Obesity, perhaps “counterintuitively,” is also a scarcity problem: sticking to your diet “requires having less to eat than you feel accustomed to,” thereby illustrating the need to ration your calories.Footnote 223 And feeling lonely is a kind of social scarcity, which highlights both our limited capacities for isolation and the limited supply of “social bonds.”Footnote 224
Furthermore, scarcity in many forms can apparently dominate our thinking. One study found, for instance, that starvation—a consequence of food scarcity—induced subjects’ minds to obsessively focus on food and food-related phenomena like cookbooks. As one participant described in the starvation study:
I don’t know many other things in my life that I looked forward to being over with any more than this experiment. And it wasn’t so much … because of the physical discomfort, but because it made food the most important thing in one’s life … food became the one central and only thing in one’s life.Footnote 225
Another study demonstrated the domineering nature of scarcity on our mindsets when subjects—some who had been told to eat and others who were left hungry—were asked to identify words that would intermittently appear on a screen. Both groups performed equally well; however, the hungry subjects “had greater awareness of food-related words than satiated participants did.”Footnote 226 Moreover, these scarcity-induced fixations can happen at a subconscious level. One study found that subjects recognize the word “water” in milliseconds when they are thirsty.Footnote 227
A countercharge might be that the primeval nature of these illustrations renders them less persuasive because it is unsurprising that subjects’ minds myopically focus on items that can satisfy visceral cravings. However, other studies demonstrate that scarcity capture is not limited to these categories. A study found for instance that children’s estimation of coin sizes was directly correlated to their level of indigence, that is, poorer children—meaning children experiencing financial scarcity—over-estimated coin size. As Bruner and Goodman hypothesized, “the greater the subjective need for a socially valued object, the greater will be the role of behavioral determinants of perception.”Footnote 228 More pertinently, a “loneliness” study further showed the powerful effects of scarcity on our thinking. Lonely subjects possessed enhanced levels of social acuity.Footnote 229
One may argue, then, that because scarcity permeates much of our lives and our thinking, individuals may consistently subject their preferences to cost-benefit analyses–endeavoring to maximize resources in least costly ways to achieve desired outcomes. Stated otherwise, RCT through cost-benefit analysis is likely to be a general—and generalizable—decision-making structure. And if we know this ex-ante, we have taken the first step towards identifying why RCT may have the capacity to predict endogenously.
Preference content, in contrast, may be a little more difficult to identify ex-ante in the sense that the subjective costs and benefits of the decision-maker will be unknown. However, this is not necessarily the case because one could argue that preference content can be reasonably ascertained ex-ante by reference to generally accepted social norms. Consequently, preferences on this view would be (in)dependent of context rather than idiosyncratically (co)dependent.
Consider the following vignette. Imagine packing a suitcase for a business trip. If you were to pack the suitcase rationally—that is, in a way that maximizes benefits over costs—you would begin with packing the essentials. You would include toiletries and business attire for sure. You would also surely include fresh underwear and socks. Less obviously, you may include gym clothes and runners—in the event that you find time to go to the gym.Footnote 230 You might also include some light entertainment—perhaps a novel or your handheld gaming console—in case you get bored. Even less obviously—perhaps because it is the summertime with record temperatures—you decide to pack several scarves; these have some form of hedonic, sentimental value to you. But why are the most latter items not top of the list? The answer would seem to be that benefits over costs would not be maximized. To see why this is the case, imagine the reverse, along with the consequences: you manage to pack everything except your business suit trousers, business shirt, and spare underwear. You embark on your business trip and check-in at your hotel. The next morning, you have an important business meeting with several of your company’s most important clients and executive directors. In fact, the clients comprise 40 percent of your company’s turnover. With no business suit trousers, business shirt, or fresh underwear, you proceed to the meeting wearing only your business suit jacket, gym shorts, runners, underwear that has now been worn for 36 hours after a long flight, and two of the many scarves that you derive hedonism from. Too polite to say anything, and perhaps content with remaining at a distance lest they encounter the waft emanating from your close vicinity, your colleagues and clients proceed with the meeting. Two days later, after arriving home from your business trip, you receive a firm but professional email informing you that your employment contract has been terminated. Your spouse, who struggles to comprehend why you decided to pack your suitcase in the way you did, decides to move out “temporarily.” Your children move with your spouse. News of your escapade percolates through your industry, and all the while you find it surprising that other employers continue to rebuff your job applications. Your mortgage payments have become so overdue that the bank forecloses on your house. You now reside on the side street of an otherwise bustling city road, but nonetheless, smiling, as you rub your cheek up and down those scarves that you love so much.
What can this vignette tell us about preferences and preference content more specifically? Besides the fact that it represents an additional exemplification of scarcity’s ubiquity—physical space, employers’ levels of leniency, jobs, spousal love, and income streams—it can serve to demonstrate how in making choices to maximize benefits over costs—which as the foregoing paragraphs have shown, we probably do most of the time—the content of the benefits and the content of the costs possess normative underpinnings. Otherwise put, most—all—people would agree on what the content should be. There can scarcely be anybody who would think that the outcome in the above vignette was a welfare maximizing outcome. Very few people would possess such a preference.
Korobkin and Ulen may disagree on this latter point because they seem to be amenable to entertaining the idea that the preference for packing the suitcase in the reverse order above exists not as an extreme case. In criticizing what they term “thin” versions of RCT, they rely on the somewhat idiosyncratic example of the Buridan Ass.Footnote 231 They contend that the Expected Utility version of RCT would be unable to predict ex-ante how the ass will choose between two haystacks located away at exactly the same distance. The ass would face three outcomes: haystack 1, haystack 2, or death by starvation. The inability to predict is because “[expected utility] provides no theory or account of the content of the ass’s utility function.”Footnote 232
Yet this example powerfully illuminates just why it is that RCT can predict (in)dependently and BE can only predict (co)dependently. If the content of the ass’s preferences were specified—for example, by stipulating that the ass valued avoiding making difficult decisions more than it valued survival—then the theory making this stipulation would be rendered idiosyncratically co-dependent in its predictions. One could hardly fathom a realistic scenario, which the Buridan Ass hypothetical is not, where preferences would be populated as such. If such preferences did exist, they would exist as an anomaly—not as a majority or a social norm—and would only serve to illustrate how one should not act. We can therefore see how the normative conditions of RCT can be generalized at both a procedurally structural level—choosing in a manner that maximizes benefits over costs—and a preference content level—populating the benefits and costs with normative content like survival over death.
An even more compelling illustration of how RCT’s normativity can be generalized at both the process and content levels is a further example from Korobkin and Ulen. Asserting that their example is of “direct relevance to legal policy,” the authors assess a scenario where a society desires to reduce crime levels. Should the length of prison sentences, therefore, be increased or decreased?Footnote 233 Korobkin and Ulen argue that while an increase may seem to be intuitively correct, this would only be so “if we assume that most people prefer to live outside of prison than to be incarcerated.”Footnote 234 Consequently, thin versions of RCT are unable to generate predictions because they do not “include some [specification] about the content of preferences.”Footnote 235 Brennan argues similarly that “some restriction on the agent’s ends … is required to give rationality any predictive bite at all.”Footnote 236 But these arguments fail to account for the normativity of preferences. Just like survival in the Buridan Ass example, almost all people would prefer liberty to incarceration. Thus, content specification is unnecessary to make predictions ex-ante when a normative benchmark already exists by which to identify the content of those preferences.
Thus, we would now seem to have taken the second and final step towards identifying why RCT may be able to predict endogenously. Fundamental scarcity, coupled with generally accepted social norms, represent a counterargument to the notion that RCT may require a “crutch” to generate ex-ante predictions. Of course, it must be noted that while decision-making structures may generally be subjected to cost-benefit analysis, the generality of the benefits and costs, that is, the preference content, is a closer question. This Article only attempts to highlight the potential for further research in classifying norms according to general levels of acceptance and how the accuracy of ex-ante predictions can be, and are, a function of these levels. A preference for liberty, as stated, may be a very general norm, as may a preference for life. A preference for wealth could also fall into this category. But other phenomena—for example a preference for sobriety versus intoxication or for living in a city versus the country—may be less accepted and would therefore be “quasi-normative.”
While these are questions that will need to be answered in advancing the concept of endogenous predictive power, we now take a final turn towards examining the potential for BE to match RCT in this respect. Can BE ever hope to predict endogenously?
II. BE is Anti-Normative
Section C of this Article exposed us to the idiosyncratic co-dependency of BE in its predictive powers. Across judgment biases, decision-making biases, and perhaps more worryingly even at the unified theoretical level, BE’s contextual dependence was illustrated. Unfortunately for BE, this may be the best it can ever hope to achieve. In other words, its predictive power is exogenously sourced and, therefore, endogenously limited.
To illustrate this point, consider the contention that BE is fundamentally anti-normative. More specifically, it is a self-proclaimed body of anomaliesFootnote 237 —which on one view is just a synonym for “abnormal,” “peculiar,” and “difficult to classify”—that showcase how individuals should not act. More informally, individuals make “mistakes.” For example, it has already been shown above in Section C(II) how in suffering from anomalous framing effects, a violation of the invariance axiom,Footnote 238 consumers make decisions that decrease their surplus when prices are revealed in distorted ways.Footnote 239 Note also anomalous violations of preference transitivity and the “money pump” demonstration,Footnote 240 where decisions result in less welfare for individuals. And we saw how the availability heuristic can generate wider social costs in the form of availability cascades.Footnote 241
Let us take further the most latter example. Aside from Love Canal, the more illuminating Alar vignette showcases how an “irresistible demand for regulation” may generate negative social costs.Footnote 242 Alar was a pesticide for apples.Footnote 243 A carcinogen, UDMH, comprised approximately one percent of Alar. Alar’s manufacturer undertook a study about carcinogenic effects and found that they were negligible, although tumors manifested in higher instances when exposure increased.Footnote 244 The Natural Resources Defence Council (NRDC) concluded in its own analysis that approximately 1/4200 school children would develop cancer by age six if exposed to Alar.Footnote 245 In a similar vein to Love Canal, exaggerated and vivid claims about these results began to permeate society’s consciousness. An NRDC lawyer stated: “If EPA doesn’t think the most potent cancer causing chemical in our food supply is grounds enough to declare it an imminent hazard and remove it from food, well, I don’t know what kind of risk it takes then to declare an imminent chemical hazard.”Footnote 246 The 60 Minutes television show presented the story with an apple converged with a skull and bones, while the show’s host declared: “The most potent cancer-causing chemical in our food supply is a substance sprayed on apples to keep them on the trees longer and make them look better.”Footnote 247 Newspapers, television shows, and the popular media all zealously covered the story. Public protests emerged alongside celebrity protests, with the actress Meryl Streep founding a group called Mothers and Others for Pesticide Limits.Footnote 248 The EPA announced that the risk was in fact exaggerated (not 1/4200 but rather 1/111,000) but by this time demand for apples had drastically dropped: they were no longer to be found in schools and stores and poison control services were receiving such questions as whether apple juice should be poured down the drain or destroyed in a toxic waste dump.Footnote 249 Alar sales were ultimately discontinued. The social costs were huge in the form of defamation suits and an estimated $125 million loss to the apple industry. The U.S. Department of Agriculture used $15 million of taxpayer money to purchase surplus apples. Also, the removal of Alar made apples more expensive, which resulted in a drop in consumption, a possible adverse effect for human health greater than the risk of Alar itself. As Kuran and Sunstein conclude: “[T]here are socially cheaper ways to bring about a substitution effect of industrial inputs. It should not be necessary to cause widespread panic or to disrupt production and consumption patterns for the sake of preventing a dubious or minor hazard.”Footnote 250
All of these examples share a common theme, which can serve to demonstrate the limitations of BE as a theory with predictive power. They are all instances where decisions led to outcomes generating costs > benefits which, as we have already seen, is antithetical to a world of fundamental scarcity where individuals are most likely to default towards decision-making structures that generate outcomes where benefits > costs. And nobody could really contend that a preference for the Alar outcome represents normative content; it was more plausibly “a serious mistake.”Footnote 251
In this important respect, RCT can be seen as the starting point. It is how individuals should act. It is therefore arguable that BE may never be the starting point because in general people do not seek to maximize costs > benefits, unless BE finds another fulcrum with at least the equivalent ubiquity as the concept of fundamental scarcity. Thus, if BE has any predictive power, it will likely be exogenously sourced via identifying various boundary conditions and idiosyncratic information under which individuals are more likely to depart from rationality, or rather, act anti-normatively. For this reason, it is not the default. It is, in other words, the “residual” of rational choice.Footnote 252
E. Reflections for Law and Policy – Frequency, Misunderstood!
The insights advanced in this Article about a theory’s predictive power and the distinction between different kinds of sources of predictive power—endogenous or exogenous—can serve to illuminate an error in theory-making generally, and BE especially, and which may be of value to law and legal policy demands like those discussed in Section B.Footnote 253 An oft-cited proposition, usually leveled by BE proponents, is that the range of anomalies identified by BE happen “systematically,” or in other words, frequently.Footnote 254 This “systematicity” is therefore cited as grounds for arguing that BE possesses stronger predictive power than RCT. As Prentice contends: “[F]or K-T Man to have more descriptive, explanatory, predictive, and prescriptive power than Chicago man, people need only be systematically (not universally and uniformly) subject to various heuristics and biases discussed in the literature. And they are.”Footnote 255
However, what these scholars seem to be doing is equating frequency with a theory’s source of predictive power—that is, they seem to be falling foul to a circular narrative that because biases and heuristics happen frequently they are predictable, and they are predictable because they happen frequently.Footnote 256 As Mitchell explains: “[A]fundamental assumption of the new behavioral law and economics movement is that individuals systematically fall prey to a host of ‘cognitive illusions’ that lead to predictable nonrational behaviors.”Footnote 257 An explicit example of this circularity in equating predictive power with “systematicity” is the following: “There are at least two responses to the argument that behavioral economics is not useful because errors are not systematic and thus behavioral economics cannot predict deviations from rationality. First, some scholars have demonstrated that some errors may be systematic.”Footnote 258
This conception of BE’s predictive power, however, is on this Article’s view erroneous because the proper question to ask on this point is not how often a theory predicts accurately but rather why a theory so frequently makes predictions that transpire to be accurate. Why is it, for example, might there exist a “mountain of experiments in which people: display intransitivity; misunderstand statistical independence; mistake random data for patterned data and vice versa; fail to appreciate law of large number effects; [and] fail to [recognize] statistical dominance?”Footnote 259 Frequency only confirms a theory’s predictive power ex-post; it is not its driver—that is, its source, as discussed above—of predictive strength.
This error—referred to by some as the “fundamental methodological error” because it conflates experimental results with broad-based, hypothetical propositionsFootnote 260 —has implications for law and policy on several levels. The perspectives which follow may help illuminate the potential costs that might be incurred if a field like BE—which struggles to coalesce as a normative, standalone theory—is used to satisfy the epistemological demands of a system of governance like those identified in Section B. Exogenous boundary conditions underpinning BE may only serve to undermine a legal system premised on general and uniform command-and-control structures.
At the doctrinal level, the lack of a coherent theory with predictive power for BE may serve to undermine legal certainty if indeed a court’s decision relies on behavioral economic reasoning and phenomena. The ad-hoc, co-dependent nature of BE means that the ratio of a judicial decision, if based on BE reasoning, would, predictably, govern all future cases with similar facts in a rule-based regime but with no formalized basis for such a generalization. If a judge decides, for example, that a party to a contract was manipulated into it due to succumbing to a behavioral bias or heuristic,Footnote 261 then all future cases with analogically relevant factsFootnote 262 would presumably be bound according to the doctrine of stare decisis.Footnote 263 However, because there is no theoretically formalized basis for generalizing the specific behavioral phenomenon in the first judicial decision, the second judicial decision would be assuming the same effects again, that is, hypothesizing that individuals fall foul to irrationality generally,) but on the basis of contingent and empirical phenomena that yield variable results across and within individuals.Footnote 264 Thus, the legal system, through the doctrine of stare decisis, would seemingly result in the scaling and generalizing of BE but on a basis that is not justified either empirically or theoretically. Ironically, this might be viewed as the legal system itself—which strives towards consistency, generality, accuracy, integrity, and “ruleness”Footnote 265 —advancing rule of law ideals but only because its own structure would seemingly be inherently liable towards falling foul to the fundamental methodological error due to the doctrine of stare decisis. In other words, the legal system, due to its form-based nature, would be generalizing and systematizing BE phenomena but which are very clearly by nature contingent empirical phenomena.Footnote 266
At the legislative and rule-making level, legislators and judges acting under conditions of asymmetric information and uncertainty do not know, absent a theory with predictive power, how society in general will react to a new law or legal rule. Behavioral reactions will only be revealed ex-post, so some sort of proxy—in the form of a theory about human decision-making generally—for predicting these reactions ex-ante is required. Considerations like the standard of liability, the exceptions to a general rule, and indeed the scope of a rule itself can all influence whether the behavior that is sought to be targeted will be optimally enforced.Footnote 267 An optimal law would deter behavior to the point at which its deterrence has a chilling effect on otherwise desirable behavior. This is particularly significant in some fields of business law, for example, where a field like antitrust often regulates through the use of broad-based standards business conduct that can sometimes have both positive and negative effects.Footnote 268 This choice of standards reflects the fact that the conduct is too voluminous to describe in detail ex-ante, can manifest in various forms ex-post, and has high levels of variance in its welfare effects.Footnote 269 This is unsurprising when one understands, as explained in Ball Memorial Hospital, that “injuries to rivals [can be] byproducts of vigorous competition.”Footnote 270 Similarly, in the field of directors’ duties the business judgment rule in the United States encourages risk-taking to the point at which it would be socially undesirable. Described as one of the “most fundamental doctrines in corporate law,”Footnote 271 the business judgment rule distinguishes between fraudulent behavior and errors in judgment and no liability will attach for the latter made in good faith.Footnote 272 In the United Kingdom, the combination of subjective and objective standards adopted for some directors’ duties reflects a similar goal—a balancing of wanting directors to feel relatively free to take business risks, subjectively construed, provided they are made in good faith.Footnote 273
Yet if the selection of both the content and scope of these rules, standards, and exceptions are based on an assumption that due to, for example, loss aversion or framing effects market actors will over- or under-interpret these rules, then they may be fashioned in a way that is suitable for some directors but not for others, because in general it is not the case that all members of society always fall foul to these behavioral phenomena. If irrationality was the basis for creating these rules and standards, the result may be a legal regime that is either over- or under-inclusive because of an erroneous assumption, based on BE, about how actors will or will not react to a particular legal rule. As explained above, BE’s lack of a normative foundation means that it is limited to exogenous predictions that often require specific and idiosyncratic knowledge about the decision-making context and often about the decision-maker themselves. The problem for BE is that legislative and judicial rules are—and must be—selected on the basis that they will apply equally and uniformly across all members of societies and/or a market context and its actors. Thus, given that BE is limited only to exogenous predictions, it would seem somewhat antithetical to a legal regime based on general and uniform governance measures.
Finally, at the policy level, the effects of adopting and inserting BE into law-making and legal decision-making by assuming that it is a theory of human decision-making with generalizable predictive capabilities may have undesirable moral hazard effects and a chilling effect on innovative business conduct. As to moral hazard, by assuming that individuals fall foul to decision-making errors generally—that is, they frequently make errors in choice and judgment and may for example mistakenly enter transactions that they would not have entered under RCT—those individuals do not now have an incentive to become more rational.Footnote 274 Knowing ex-ante that rules, liability standards, and their enforcement will presume that individuals are bounded in their willpower and rationality, the intended regulation’s beneficiaries may become, and feel, more protected and therefore less incentivized to invest in their own rationality. This is undesirable from a societal perspective because by continuing to predict that individuals will act and react irrationally, otherwise socially desirable levels of societal rationality may be foregone due to a legal regime that provides no incentives for society as a whole to take ownership for their own irrational behavior.Footnote 275 Regarding a potential chilling of innovative business conduct, businesses cannot now assume ex-ante that market participants make decisions and choices according to RCT. Due to BE becoming baked in to the legal system at both the law-making and enforcement levels, the starting point for businesses, because it is now the starting point for courts, would no longer be RCT but rather BE—that is, irrationality will have come to replace rationality as a starting point for assuming how individuals make decisions. Business’ incentives to innovate become reduced because of a risk of one of their products triggering any one of the myriads of BE phenomena. Products and services that would have been valuable to society are instead foregone. And the overall rate of innovation may consequently decrease.
F. Conclusion
The limits of BE in its theorization attempts have been significantly underdeveloped in current commentary and debate. This Article has sought to highlight these limitations by illuminating the distinction between endogenous and exogenous predictive power, a criterion for assessing a theory’s capabilities. The Article showed the limitations of BE in this respect–specifically, that perhaps the best it can hope for is exogenous predictions by reference to relevant boundary conditions which may indicate the likelihood of irrationality governing decision-making. But as a body of knowledge lacking a normative foundation—in fact, in several foundational respects it seems to be anti-normative—BE may always be the residual of RCT, which can predict endogenously due to its inherent normativity.Footnote 276 In other words, RCT is the “departure point” for BE but not vice-versa.Footnote 277
That RCT is the starting point has been shown to be justified on several legal and policy levels by showcasing the consequences if the starting point—that is, if individuals were predicted to act irrationally in general and only exceptionally acted rationally—were reversed. Law-making and legal doctrine would do well to be mindful about the exogenous nature of BE’s predictive power and that its capacity for dominating decision-making and choice depends on context and conditions. Making irrationality the generalized starting point is not only erroneous as a matter of methodology and empirics, it may also generate negative implications for policy, particularly as it pertains to business conduct and the incentives for societies to move towards higher levels of rationality.
Acknowledgements
The author declares none.
Competing Interests
The author declares none.
Funding Statement
No specific funding has been declared in relation to this Article.
Open access
