Guilt Used for Lawmaking is Heterogeneous and Dynamically Dependent
In Response to Ian Ayres, Joseph Bankman & Daniel Hemel, The Law and Economics of Guilt and Shame, 93 U. Chi. L. Rev. 247 (2026).
I. Introduction
In The Law and Economics of Guilt and Shame, Professors Ayers, Bankman, and Hemel transcend the typical law and emotions literature that primarily focuses on the efficacy of shaming penalties; they instead analyze the policy implications of guilt and shame on optimal lawmaking and optimal enforcement. They do so by situating guilt and shame in a novel unified model where the deterrence benefit of these negative emotions are examined alongside the reduction in social welfare that arises from emotional disutility.[1] Their findings demonstrate that, depending on context, negative emotions may substitute for legal rules and enforcement, and they may also increase or decrease the optimal degree of lawmaking and law enforcement.[2]
In this response, I offer two critiques on the model that they developed. The first critique concerns the representative agent assumption that simplifies the model by assuming that all parameter values are the same for all individuals. I argue that the inherent heterogeneity in individual preferences over subjective emotional experience renders this assumption uniquely inapposite and urge for it to be relaxed. The second, perhaps more important, critique concerns the authors’ dichotomous distinction between guilt arising from moral transgression and guilt arising from legal transgression as applied in the context of optimal lawmaking. I argue that we cannot draw a clear line between the two because future moral (‘law-independent’) guilt is dynamically dependent on whether or not a legal prohibition is enacted in the present. I conclude by proposing to capture this dynamic dependency by adapting the existing one-period model into a two-period inter-temporal model.
II. The Representative Agent Assumption
The authors simplify their model by assuming that a single rational decision maker is representative of all individuals in a society and share the same parameter values. Although the representative agent assumption is commonplace in economics, particularly in macroeconomics, scholars have always been skeptical of its application in deriving policy recommendations. According to Alan Kirman, even if we accept the assumption that the current choices of the aggregate is accurately represented by a single agent, “[t]he reaction of the representative to some change in a parameter of the original model—a change in government policy for example—may not be the same as the aggregate reaction of the individuals he ‘represents.’”[3] Indeed, more recent empirical research suggests that this assumption is likely to be misleading.[4]
Although the authors point to the law and economics literature on optimal taxation to justify their reliance on the representative agent assumption, preferences in the realm of taxation are intrinsically different from preferences over the emotion of guilt.[5] One would expect greater heterogeneity in how individuals experience guilt vis-à-vis how much disutility they experience when they are taxed more. Intuitively, this greater heterogeneity arises from three sources. First, a particular individual’s subjective experience of guilt is informed by a myriad of idiosyncratic factors such as cultural background, past experiences, and social context.[6] Second, the vast majority of individuals are not in the habit of quantifying their emotions in terms of utils or converting them into to monetary values; by contrast, taxes are already quantified in monetary terms. Third, while being taxed is a near-universal experience, it is safe to assume that most individuals do not know how guilty they will feel when they transgress a particular moral norm or legal rule; of course, it is difficult to estimate the disutility of guilt arising from murder when one has never committed murder. Therefore, it is unclear whether individuals even have well-behaved preferences regarding guilt, let alone the assumption that these preferences are homogeneous in the aggregate. In sum, the representative agent assumption should be relaxed and heterogeneity should be incorporated into the model.
III. Dynamic Dependency of Guilt from Moral Transgression
A. The Problem
Building on the contrast between Shavell’s and Mungan’s conceptions of guilt, the authors draw a dichotomous distinction between law-independent guilt and law-dependent guilt.[7] Specifically, they define law-independent guilt as guilt that “arises when individuals transgress a moral norm whether or not that norm has been codified into a legal prohibition,” and law-dependent guilt as guilt that “results specifically from breaking a legal rule.”[8] However, even within the authors’ model, it is unclear whether or not this rigid dichotomy exists. In order for a line to be cleanly drawn between guilt arising from moral transgression and guilt arising from contravening a legal rule, it must be assumed that the development of a society’s conception of morality is completely detached from the legal regime that governs it; that is, it must be assumed that the evolution of moral norms over a given behavior is not influenced by its legality.
This assumption that morality is static, or at least independent of legality, does not seem to comport with our understanding of how moral norms evolve. Intuitively, legality impacts future moral norms because by making an act illegal, the society is collectively characterizing that behavior as wrongful and expressing disapproval. As Kenworthey Bilz and Janice Nadler argue, “[l]egal regulation can [] transform the social meaning of behavior, changing people’s perceptions regarding the moral acceptability or desirability of the behavior.”[9] Empirical research suggests that “the status of a conduct as legal or illegal has a [statistically] significant, albeit modest, impact on an individual’s moral appraisal of that conduct.”[10] Nigel Walker and Michael Argyle report that survey respondents who did not treat heavy smoking as morally wrong said they would change their moral view if the legislature criminalizes heavy smoking.[11] In sum, both intuition and empirical evidence indicate that guilt arising from moral norms cannot be fully distinguished from guilt arising from legal rules; this is because legality interacts with morality as time progresses.
B. The Implications
The most evident implication here concerns applying the author’s model to analyze optimal lawmaking. As the above discussion demonstrates, criminalization of an act in the present will generally make that act more morally condemnable in the future; again, survey respondents expressed that they would start treating heavy smoking as more morally dubious if it is criminalized.[12] Therefore, what the authors define as ‘law-independent’ guilt (guilt that arises from transgression of moral norms) is in fact dynamically dependent on whether a legal prohibition is being enacted in the present. Specifically, the present enactment of a legal prohibition will increase the ‘law-independent’ guilt arising from future transgressions of the corresponding moral norm because the moral opprobrium against that act is strengthened. If we do not take this dynamic effect into consideration, we will inevitably underestimate the guilt-based deterrence effect of enacting a legal prohibition and therefore deviate from optimal lawmaking; that is, we will adopt less laws than is socially optimal.
C. The Proposed Solution
It is admittedly difficult to capture this dynamic effect in the existing one-period model. A potential solution is to adapt the current model into a two-period inter-temporal model and capture the dynamic effect by stipulating that if a law prohibiting the act is enacted in period 1 and if not. Under this stipulation, we expect the range of ‘law-independent’ guilt that will yield inframarginal effect to be narrower in period 2 than in period 1. This reflects the scenario where, although enacting a law does not provide sufficient combined guilt to deter the behavior in period 1 (, the enactment nonetheless alters the relevant moral norm in period 2 and increases guilt arising from its transgression; through the increase in ‘law-independent’ guilt from moral transgression, the combined guilt in period 2 may nevertheless exceeds the private benefit ( and deters the proscribed behavior. Even with temporal discounting of the social welfare benefit of period 2 deterrence, this formulation will certainly make it easier to satisfy the cost-justification constraint of lawmaking; therefore, it will suggest that lawmaking is optimal in a wider set of circumstances. Although this will complicate the model, it will better reflect the fact that moral norms evolve in the context of a society’s prevailing legal regime.
IV. Conclusion
As the forgoing discussion demonstrates, guilt as an emotion used for lawmaking is heterogeneous in the population and dynamically dependent on whether a proscribed behavior is presently criminalized. In contrast to taxation, individuals do not have well-behaved homogeneous preferences over the subjective emotion of guilt because guilt is informed by idiosyncratic life experiences, rarely quantified, and individuals simply cannot be expected to correctly estimate how much guilt-based disutility they will feel from breaking laws or moral norms that they have never broken before. This inherent heterogeneity renders the representative agent assumption uniquely inapposite. Moral norms are not static and their evolution is influenced by the legal regime that currently governs. Intuition and empirical evidence confirm that society’s future moral appraisal of a conduct depends on whether that conduct is criminalized in the present. As morality interacts with legality as time progresses, we cannot draw a dichotomous distinction between ‘law-independent’ moral guilt and law-dependent guilt; the degree of future moral guilt is dynamically dependent on the present enactment of laws. In order to properly capture the guilt-based deterrence effect, a model of optimal lawmaking ought to take such intertemporal interaction between morality and legality into account.
[1] Ian Ayres, Joseph Bankman & Daniel Hemel, The Law and Economics of Guilt and Shame, 93 U. Chi. L. Rev. 247, 247 (2026).
[2] Id., at 296–297.
[3] Alan P. Kirman, Whom or What Does the Representative Individual Represent?, 6 J. Econ. Persp. 117, 118 (1992).
[4] See Terry Barker & Sebastian A. de-Ramon, Testing the Representative Agent Assumption: the Distribution of Parameters in a Large-scale Model of the EU 1972–1998, 13 Appl. Econ. Lett. 395 (2006).
[5] Ayres et al., supra note 1, at 260.
[6] Roy F. Baumeister, Arlene M. Stillwell & Todd F. Heatherton, Guilt: An Interpersonal Approach, 115 Psychol. Bull. 243, 244 (1994).
[7] Ayres et al., supra note 1, at 255–56.
[8] Id., at 259–60.
[9] Kenworthey Bilz & Janice Nadler, Law, Moral Attitudes, and Behavioral Change, in The Oxford Handbook of Behavioral Economics and the Law 241, 252 (Eyal Zamir & Doron Teichman ed., 2014).
[10] Dan M. Kahan, Gentle Nudges vs. Hard Shoves: Solving the Sticky Norms Problem, 67 U. Chi. L. Rev. 607, 612–13 (2000).
[11] Nigel Walker & Michael Argyle, Does the Law Affect Moral Judgments?, 4 Brit. J. Criminol. 570, 574–75 (1963)
[12] Id.