Why Agencies Punish

Max Minzner

Administrative agencies, the fourth branch of government, famously blend the functions of the other three.  Agencies write rules, adjudicate their meaning, and penalize violations. Several recent high-profile agency penalties have attracted attention to this understudied area of administrative law.  In 2010 alone, the National Highway Traffic Safety Administration imposed a statutory maximum $16.4 million penalty on Toyota and the Securities and Exchange Commission recovered $535 million from Goldman Sachs, the largest civil penalty ever paid by a financial services firm.

Agencies are commonly assumed to have a straightforward goal when they punish – agencies penalize to induce compliance with their rules.  Penalties aim to curb violations and prevent their reoccurrence.  In the language of the criminal law, agencies are seen as consequentialists.  Penalties seek to achieve positive social outcomes.  For example, penalties might deter misconduct by raising the expected cost of violations above the cost of compliance, or might attempt to reinforce norms of compliance through punishment.

Criminal law, of course, recognizes a second primary aim of punishment: retribution.  The goal of retributivism is punishment itself – wrongdoers are punished because they deserve it, not to achieve some broader social end.  As a general matter, though, scholars have assumed that agencies are not trying to achieve retributive ends through their punishment schemes.1 There are good reasons for this assumption.  Agencies generally disavow retribution as a goal of administrative enforcement and claim to only seek consequentialist results.

There is reason to be skeptical of this rejection of retribution.  A number of recent experimental studies have looked at punishment decisions to try to determine whether people primarily care about deterrence, retribution, or something else.  When asked to impose punishment, test subjects tend to focus on facts that are relevant from a retributive framework rather than a consequentialist perspective. These results hold even though test subjects identified deterrence as an important reason for punishment.  In short, people talk like consequentialists but act like retributivists.

Testing the Deterrence Claim

How can we tell whether agencies care about deterrence or retribution?  These different theories require punishers to emphasize different facets of the violation.  For example, theories of deterrence focus on the expected cost of the violation.  Penalties are set such that the expected value of the fine ensures that violators either bear the costs they impose or are deprived of any gains from misconduct.  As a result, penalties should be largely shaped by gain, harm or risk of harm, and the probability the penalty will be imposed.

Very different concerns shape retributive penalties.  Unlike the forward-looking approach of deterrence, retribution is backward-looking, focusing on the violator and his conduct.  Punishment is imposed because a wrongdoer deserves it, not to achieve some broader external goal.  For instance, while the probability of detection is a central concern for theories of deterrence, detection probability is irrelevant to retribution. Desert-oriented punishment gives no reason to incorporate the likelihood of a punishment into the calculation of its size.  In contrast to theories of deterrence, the mental state of the wrongdoer with respect to the wrongful conduct dominates the field for retributivists.  A more culpable mens rea increases the wrongfulness of the act and, as a result, the amount of punishment that is deserved.  Individuals that knowingly cause harm deserve greater punishment than those inflicting harm merely recklessly.

This Article looks at the civil penalty systems of four agencies with missions spanning the federal administrative state: the Mine Safety Health Administration (MSHA); the Office of Foreign Assets Control; the Federal Communications Commission; and the Nuclear Regulatory Commission (NRC).  It examines the factors that play a role in these agencies’ penalty schemes and attempts to classify them according to the theory of punishment – deterrence or retribution – that they primarily serve. After analyzing both the theory behind the aforementioned punishment concerns and each agency’s practice, the Article concludes that, for each of these agencies, retribution, not consequentialism, provides the best explanation of their civil penalty policies. Consistent with a retributivist approach, all of the agencies emphasize harm and the risk of harm, place culpability at the center of the penalty process, and exclude enforcement probability when calculating penalties.Similarly, gain is largely irrelevant to the penalty determination for all four agencies.  All other things being equal, extremely lucrative and difficult-to-detect rule violations are punished no more severely than those where compliance with the rule is cheap and violations are always punished.

In selecting these points of emphasis in their punishment decisions, agencies look like the rest of us.  Numerous experimental studies have demonstrated that individuals place little or no importance on the probability that the defendant’s conduct will be detected and punished.2 Study participants do not adjust penalty amounts based on the probability of the misconduct being detected. These experimental results have also repeatedly demonstrated that the mental state behind a violation plays a key role in punishment determinations.  Additionally, harm matters for the public at large. Experimental studies find that when asked to punish, test subjects are strongly motivated by the outcome of the misconduct.  Behavior that causes harm is punished more severely than conduct that does not, and punishment severity increases as the magnitude of the harm increases.

Agency Retribution: A Reason for Concern?

Does it matter if agencies (like rest of us) talk about deterrence but engage in retribution? I think it does. As an initial matter, agency retribution raises questions of legitimacy.  Legitimate retribution requires transparency. By its very nature, in order to make retributive punishment legitimate, it must communicate condemnation.  As a result, retributive punishment has to be labeled as such.  Because penalties are not received as punishment when they are cloaked in the language of deterrence, hidden retribution does not communicate the appropriate message.

There is a second significant problem with agency retribution.  Agencies might not be very good at it.  We often assume that agencies are experts in their area of enforcement.  The NRC knows a great deal about nuclear power plants, so scholars and the courts generally assume that the NRC also knows what types of penalties are appropriate for which violations of nuclear safety regulations. If, in fact, agencies do primarily impose penalties for purposes of retribution, this assumption is dubious.  Agencies are unlikely to be able to claim any particular expertise in the allocation of desert-based punishment.  Compare agencies to the two natural alternatives: civil juries and district judges. Juries are well placed to mete out retributive justice.  If the goal of punishment is to reflect community desires, lay jurors have a far better claim of expertise on that subject than administrative agencies.  Similarly, federal district judges routinely impose retributive penalties in the course of criminal sentencing.  Both of these institutional actors seem more likely to impose accurate retribution than agencies.

In the end, perhaps the best argument for agency retribution is norm reinforcement.  An important body of criminal law scholarship argues that retribution not only can serve utilitarian ends, but is in fact the best way to do so.3 Retributive punishment can reinforce norms of compliance by punishing only those who truly deserve it.  In this sense, retribution may be better able to achieve compliance than classic deterrence theories.

In the administrative context, though, using desert to reinforce norms faces serious implementation problems.  The arguments that justify desert as a utilitarian approach in criminal law are much weaker when applied to administrative civil penalties.  Two important characteristics of the criminal law are largely absent in the administrative context – norms are weaker and the goals of punishment are less transparent.  As an initial matter, arguments about the consequentialist value of desert generally assume a preexisting set of norms.  These norms classify conduct as either acceptable or unacceptable regardless of the existence of punishment.  Criminal punishment can then respond to these preexisting norms in several important ways.  By focusing criminal punishment on conduct that already violated these norms, law enforcement can leverage the preexisting stigmatizing effect of norm violation to more cheaply obtain compliance,  and authorities can reinforce these norms by serving as the legitimate institutions that punish norm violation. Criminal law can also extend the boundaries of these norms by establishing its legitimacy in situations in which norms are clear. By gaining a reputation for fairness in the easy cases through reinforcing preexisting norms, the criminal law can set norms in the borderline cases.

If administrative civil penalties are aimed at norm reinforcement, the task is far more difficult because the norms do not necessarily exist prior to agency action.  The problem is clearest in the context of safety regulatory agencies.  Both MSHA and the NRC regulate highly dangerous conduct by drawing lines based both on the riskiness of the conduct and the cost of further safety measures.  The line dividing acceptable behavior (conduct that is sufficiently safe given the cost of additional preventative measure) and unacceptable conduct (behavior in which additional safety measures should be mandated given their cost) is hard to draw and is, in many ways, inherently arbitrary.  Entities that are regulated by these agencies learn the acceptability of conduct by looking at the rules themselves rather than drawing on a preexisting body of societal norms.  Agencies are thus forced to take on the difficult task of norm setting rather than the comparatively easier task of norm reinforcement.

Is Retribution Inevitable?

The presence of retribution in agency punishment provides an important data point on the plausibility of placing deterrence at the center of any punishment scheme.  Administrative agencies are perhaps better situated than any other actor to impose penalties in a manner consistent with deterrence theories.  Deterrence-based systems of punishment have struggled to achieve traction for a variety of reasons, but a key component is the real-world difficulties of the approach. Traditional deterrence theories make strong assumptions about how both enforcers and violators behave, assumptions that are frequently not true in practice.  For instance, deterrence theory assumes that the decisions about the level of punishment and the probability of detection are decided jointly and simultaneously.  In traditional criminal enforcement, law enforcement sets the probability of detection for violations, and a sentencing judge sets the level of punishment.  Administrative agencies are different.  Because of the blend between executive, legislative, and judicial roles in the administrative context, agencies oversee their inspection and enforcement functions (providing control over the probability of detection), issue rules establishing systems of penalties, and decide penalties in individual cases (providing control over the eventual punishment).  Because agencies can control all aspects of enforcement, they have a much greater opportunity to establish a deterrence-based system if they so choose.

Similarly, the information problems in administrative enforcement are far smaller.  In the traditional criminal context, the assumption of perfect information by defendants and enforcers is a dubious one.  Violators are unlikely to be aware of the situations under which fines might be imposed and what those fines are.   In contrast, in the civil regulatory environment, regulated entities not only have counsel, they are repeat players in the regulation game.  They have the opportunity to observe the rules as they develop.

Of course, an argument that administrative agencies are more able to implement a deterrence-based approach to punishment than other enforcement entities does not demonstrate that it is possible.  Even with the advantages agencies have over traditional law enforcement entities, punishment based on theories of deterrence might be out of reach.  If that is true, though, we should be skeptical of any system of punishment which claims to be designed to achieve deterrence rather than retribution.


Max Minzner is an Associate Professor at University of New Mexico School of Law.

A version of this article appeared in the February 2012 issue of the William and Mary Law Review: Max Minzner, Why Agencies Publish, 53 WM. & MARY L. REV. 853 (2012).

Copyright © 2012 William and Mary Law Review.

  1. See Ian Ayres & John Braithwaite, Responsive Regulation 20 (1992); Timothy F. Malloy, Regulation, Compliance, and the Firm, 76 Temp. L. Rev. 451, 453 & n.9, 454 (2003); David B. Spence, The Shadow of the Rational Polluter: Rethinking the Role of the Rational Actor Models in Environmental Law, 89 Calif. L. Rev. 917, 919-20 (2001).
  2. See, e.g., Jonathan Baron & Ilana Ritov, The Rule of Probability of Detection in Judgments of Punishment, 1 J. Legal Analysis 553, 580-82 (2009); Cass R. Sunstein et al., Do People Want Optimal Deterrence?, 29 J. Legal Stud. 237, 243 (2000).
  3. See Tom Tyler, Why People Obey the Law 161 (1990); Paul H. Robinson & John M. Darley, The Utility of Desert, 91 Nw. U. L. Rev. 453, 498 (1997).

Post a Comment (all fields are required)

You must be logged in to post a comment.