Encyclopedia of Law and Economics
Show Less

Encyclopedia of Law and Economics

Edited by Gerrit De Geest

The second, expanded edition of the acclaimed Encyclopedia represents a major update of the most authoritative reference work in the field of law and economics and the nine print volumes are now released online as a single integrated product.The Encyclopedia provides balanced and comprehensive coverage of the major domain in law and economics, including: criminal law, regulation, property law, contract law, tort law, labor and employment law, antitrust law, procedural law, and the production of legal rules. Each theme or volume is overseen by a leading scholar and each of the 166 entries is prepared by an expert in the field, providing an in-depth and authoritative overview of the individual topic, combined with an exhaustive bibliography, allowing users to access and filter the entire corpus of literature in law and economics.As with the print edition, the Encyclopedia is unique in serving both as an entry point and a platform for advanced research. The online edition is enhanced with Elgaronline’s powerful search tools, facilitating the search for key terms across the entire Encyclopedia, whilst the browse function allows users to move seamlessly between the volumes. These elements combine to create a powerful research tool for any researcher or scholar in the field of law and economics.
Show Summary Details
This content is available to you

Chapter 14: Cyber crime

Mark A. Cohen

[In: Volume 3, Nuno Garoupa (ed) Criminal Law and Economics]

1. Introduction

The purpose of this chapter is to examine the economic theory and empirical evidence on crime and punishment for economic crimes (for example, fraud and tax evasion) and new technology offenses (for example, computer hacking and viruses), with reference to the role of the United States Sentencing Commission (‘Commission’) in affecting individual behavior.1 This chapter relies upon the economic theory of individual behavior known as the rational actor paradigm. Under this theory, it is assumed that white-collar criminals are no different from most law-abiding citizens. They are rational decision makers acting in what they perceive to be their own best interest.2 Although one might question whether offenders who engage in crimes of passion are rational, there should be little controversy in this view of offenders who commit economic crimes such as fraud and tax evasion. As rational decision makers, white-collar criminals base their decision to commit an economic crime upon their subjective evaluation of the expected costs and benefits of committing the crime. Even if actions appear irrational, they may be based on the offender’s perception of risk and opportunities. Of course, potential white-collar criminals are not the only decision makers that affect the crime rate. Others who play a role are p. 347law enforcement officials, judges, victims, and society as a whole (that is, potential victims).

Section 2 of this chapter examines the incentives that each faces and identifies the interactions between them. This first part provides a ‘systems’ approach to thinking about crime control that takes into account the actions of each decision maker. Section 3 outlines the implications for this ‘systems’ approach to crime control. Finally, Section 4 considers some of the major issues that the Sentencing Commission dealt with in revising its Guidelines for sentencing white-collar offenders for economic crime and new technology offenses.

2. Optimal level of crime and crime control: a systems approach

Although crime itself is socially harmful, the costs of devoting resources to catching criminals and punishing them also creates social harm, in that the resources could have been put to a different, socially beneficial use.3 Thus, the economic approach to crime control is concerned with finding the optimal combination of preventing, detecting, enforcing, and offending behavior. Put differently, although we would all like there to be zero crime, economic theory recognizes that this is an unattainable goal – primarily because of the costs involved in preventing, detecting, and punishing criminal activity.4

What role does sentencing have in this system? Certainly, we cannot expect the Commission – and sentencing policy – to solve all of the problems of the criminal justice system. Moreover, it is important to understand the interrelationship between the various actors in the criminal justice system, since the actions of one might affect the behavior of the others. Thus, for example, if the Commission were to dramatically increase the penalty for computer-based consumer fraud relative to other forms of consumer fraud, one effect of this policy might be to shift some of the ‘offender talent’ away from computer crime into other forms of consumer fraud. Although the goal of the stricter sentencing policy might p. 348clearly have been to reduce computer crime, its net effect might be only a minimal reduction in the overall consumer fraud rate. Alternatively, it might simply induce offenders to use more sophisticated and costly forms of evasion. Ultimately, how much (if any) the overall consumer fraud rate will decrease will depend in part on the cost of committing a crime using a computer versus other fraud vehicles and the cost to consumers of avoiding victimization on the computer.

The concept of an ‘optimal’ level of crime may appear to be morally repugnant to people who believe the government should attempt to eradicate all social evils.5 In particular, it might be difficult to accept the fact that there will always be some crime and that public policymakers will in effect tolerate some crime. Yet, both private citizens and policymakers make such difficult decisions every day.

For example, storeowners must choose how best to cope with the problem of shoplifting. By employing elaborate security measures, they can reduce the level of this crime or if security is sufficiently onerous, perhaps eliminate it completely. Consider the example in a retail store of locked chains used to protect expensive leather jackets from being stolen. Although the use of chains may virtually eliminate shoplifting of these leather jackets, it also imposes very real costs on storeowners, including the cost of the security device and the lost revenues from customer dissatisfaction that results from having to wait for a salesperson to unlock each coat and supervise the entire purchase process. These costs could exceed the store’s losses due to shoplifting, in which case the storeowner might forego this form of crime prevention. Ultimately, the storeowner selects a level of security, and hence implicitly, a positive level of crime, that she believes will maximize net profits.

Of course, the storeowner is not the only actor in this drama. Shoplifting is a crime, and one that results in punishment for the offender who is convicted. Part of the decision calculus of the storeowner and the potential shoplifter is knowledge about what the criminal justice system will do. What is the likelihood that the shoplifter will be caught? If caught, how much time will it take the storeowner to deal with the criminal justice system? How expensive is it to prosecute and convict the shoplifter? What sanction will the judge impose for this offense? All of these questions are important and interrelated. For example, if judges are reluctant to impose any significant sanctions on shoplifters, prosecutors might decline those cases or accept minimal plea bargains or pre-trial diversions. However, p. 349that does not necessarily mean shoplifting will increase dramatically. If the risk of being penalized by the criminal justice system decreases, storeowners may increase their surveillance and prevention measures, as these forms of preventive expenditures become more cost effective relative to punishment. However, if the sanction is very severe – so that shoplifters are guaranteed a year in prison, for example – store owners will shift some of the cost of prevention onto the criminal justice system. Thus, fewer storeowners will purchase costly security systems that help reduce crime, and instead they will rely on the deterrent effect of strict sentencing to reduce shoplifting.

The lessons learned from the shoplifting example can be applied to economic crimes as well. As individual consumers or business people, we take all kinds of precautions against fraud. We hang up on telephone sales persons even if we might otherwise be interested in the product being offered, since we are concerned that the party on the other end – whom we do not know – is attempting to swindle us out of our hard-earned money. Similarly, on the Internet, we are reluctant to deal with merchants we are not familiar with since it is much easier to ‘move’ on the Internet than to close up a store and open in another town where nobody knows the storeowner’s reputation. We are reluctant to give out credit card information over the Internet because unauthorized eyes can easily intercept our transmissions. We are reluctant to bid in online auctions because we do not know if the seller is trustworthy and will deliver what he promises. Of course, purchasing products over the Internet and bidding in online auctions are both relatively safe activities with low frequencies of fraud. Why? Both individuals and businesses take costly actions to prevent and/or raise the costs to potential offenders from committing these frauds. Thus, many consumers will not give out their credit card information on the Internet unless they know that the company uses appropriate encryption technology. Similarly, many consumers will not bid in an online auction unless they know that the auction company has a feedback mechanism to check on the reputation of the seller, and/or they offer insurance against fraud. Finally, many consumers will also look for a well-known, established firm that has an existing reputation. All of these prevention activities are costly – to businesses and ultimately to consumers. In some cases, where these protections are not available, consumers will decide that taking these actions are too costly relative to the benefits, and they will decide either to take a chance or to forego the transaction altogether – perhaps taking a costly trip to a ‘bricks and mortar’ retail store instead.

Moreover, the availability of criminal remedies might have an effect on the amount of prevention expenditures undertaken by businesses and consumers. If I know that the penalty for computer fraud, for example, is p. 350very severe and that the chance of getting caught is very high, I might be less reluctant to use the Internet to make purchases. However, I might also take less care in avoiding a potentially fraudulent vendor. The question then becomes whether it is more cost effective to rely upon prevention or punishment to deter fraud. At the same time, if enforcement is very stringent and penalties are severe, in a climate of rapidly changing technology and intellectual property, some marginal companies might decide that the risk of being labeled a criminal is too great a risk to go into that line of business altogether.6 Thus, some ‘innocent’ individuals or companies might not engage in the socially beneficial activity at all.7

More formally, there are several types of costs associated with economic crimes: (1) social harm8 which is often equated with the term ‘loss’ that is used in the fraud guidelines of the Commission,9 (2) the cost of apprehending and convicting offenders – law enforcement and the criminal justice system, (3) the cost of imposing sanctions – collecting fines, constructing and maintaining prisons, probation officers, etc., (4) the cost to potential victims who take preventive measures to avoid being a victim, (5) the cost to innocent individuals who are or may be charged with crimes they did not commit, and (6) the resources devoted to committing the crime itself (for example, telephones, computers, mailing list rentals). Under an economic approach to crime, the goal of society is to minimize the sum of all six of these costs.

Although I am framing the question of writing sentencing guidelines in p. 351terms of an ‘optimal’ level of crime, this goal is not necessarily incompatible with the traditional goals of sentencing. Instead, it helps to operationalize the often conflicting goals of deterrence, incapacitation, rehabilitation, and retribution.10 For example, although the optimal crime control model is similar to a deterrence model, it recognizes that deterrence comes at a price – and that its benefits should be balanced against its costs. Similarly, incapacitation makes sense when the cost of imprisonment is less than the expected harm caused by recidivism, and rehabilitation is good when its costs are outweighed by its crime control (or other social) benefits.

Complicating the picture, there are at least five distinct decision makers: (1) potential offenders, (2) potential victims, (3) law enforcement, (4) the courts, and possibly (5) individuals potentially accused of committing a crime they did not commit. It is important to understand the interrelationships between these parties. To do that, we must first ask what motivates their actions, and how we can predict the effect of changing sentencing policy on the actions of each of these five categories of decision maker.

2.1 Potential offenders

We are all potential offenders.11 Ultimately, while this is often an ‘implicit’ decision, we each weigh the benefits of evading or breaking any particular law against the cost of being caught. The expected benefits from committing an economic crime are generally monetary. However, it is also possible that offenders receive psychic benefits from economic crimes – such as peer approval, a feeling of importance, or ‘getting back at the system’. Many computer crimes appear to be motivated more by these psychic benefits than by pure monetary benefits. How else do you explain most computer viruses and hackers? Although some of this activity is motivated by money, such as extortion, it seems that some computer crime is perpetrated for the psychic rewards that the offender receives from crashing other computers.

p. 352The expected costs to the offender include the expected sanction – which depends on the risk of being caught, indicted, and convicted, as well as the expected severity of punishment if convicted. It might also depend on the extent to which the potential offender believes his standing or reputation in the community will be affected. Will this reduce his long-term opportunities for employment? Will this reduce his stature among family and friends? These are all expected costs that the potential offender will weigh against the potential gain from committing the crime.

Notice that I have used the terms ‘expected’ costs and ‘expected’ benefits. The expected cost from criminal action includes both the negative value (utility) to the criminal of the expected sanction from committing the crime and the foregone opportunities that the criminal gives up in order to pursue criminal activity. The expected sanction is calculated as the probability of apprehension and conviction multiplied by the sanction that the offender expects to be imposed. Thus, in order to fully understand the potential offender’s decision calculus, one should look at the offender’s personal perception of variables such as the likelihood of apprehension, conviction, and imprisonment, even if that differs from actual rates. For example, if all potential embezzlers believe there is a 100 percent chance of being apprehended and convicted, even if the truth were that only one in ten embezzlers were caught – this crime would probably be non-existent.

Although all potential criminals may face the same expected sanction for the same crime, different people may value that expected sanction differently. For example, an individual who is risk averse (that is, prefers certain outcomes to gambles) will react differently to a 50 percent chance of being convicted than an individual who prefers risk.12 All else equal, the risk-loving individual will be more likely to commit the crime, since he or she is less responsive to the chance of being convicted than the risk-averse individual. Not only do people differ in their preferences toward risk, but they may also differ on such dimensions as moral standards, family responsibilities, concern for the social stigma of having a criminal record, and tolerance toward living in a prison environment. Differences in these and other individual dimensions will yield differences in the value the individual places on being sanctioned and thus on the degree of responsiveness to those sanctions.

In addition to the negative value of the expected sanction, the potential criminal must consider the foregone legitimate and criminal opportunities p. 353as a cost of criminal activity.13 Foregone legitimate opportunities from engaging in criminal activity include: current wages that the criminal could be earning if he or she did not engage in criminal activity, the value of leisure time the criminal gives up while engaging in criminal activity, and the value of future earning opportunities that may be lost if the individual develops a criminal record. Foregone criminal activity includes the other crimes that the criminal could commit instead of the current one. That is, the potential offender not only considers whether or not to commit a crime, but which crime to commit. This decision will again be based on the (perceived) expected costs and benefits of each type of crime.

Not all potential criminals have the same opportunities. People differ in their ability, training, and access to criminal opportunities. For example, increased computer literacy will inevitably increase the rate of crime committed through the use of a computer. In addition, increased opportunities to earn money in legitimate occupations will reduce the crime rate.

Given our understanding of the costs and benefits of criminal activity, in order to reduce the propensity of potential offenders to commit crimes, we need to raise the expected cost of the criminal activity and/or reduce the expected benefits. There are many different actions that society can undertake to affect the crime rate, including: making it more technologically difficult and thus more expensive to commit a crime,14 providing more non-criminal opportunities to earn a legitimate income,15 increasing the psychic costs of crime through public information and education campaigns, encouraging negative peer pressure and other forms of social stigmas,16 and increasing the expected punishment through increasing the probability of detection or increasing the severity of punishment.17

2.2 p. 354Potential victims

Each potential victim decides what level of resources to devote to avoidance or prevention activities by examining the perceived expected costs and expected benefits of prevention. The expected benefits of prevention consist of the harm caused by the criminal act that is avoided, multiplied by the reduced probability of becoming a victim. The costs of prevention include the physical costs such as security systems, software programs, and auditors. But the costs of prevention also include more subtle costs and foregone opportunities such as the cost of not engaging in an online auction due to fear of being defrauded.18 The cost of prevention is likely to differ by considerable amounts by the type of potential victim. More sophisticated computer users are likely to be able to avoid becoming fraud victims at a much lower cost than others. More vulnerable victims, such as the elderly, might have a higher cost of prevention.

Potential victims will react to changes in the expected costs and benefits of prevention. Elderly individuals have extremely low victimization rates19 among both street crimes and white-collar crimes (with some apparent important exceptions such as telemarketing fraud). The reason is that they generally realize they are especially vulnerable and take added precautions to avoid situations that put them at risk. A sudden change in the crime rate will also affect the costs and benefits of preventive measures by potential victims. For example, increased access to private information on the Internet that lowers the cost of committing identity theft or credit card fraud also increases the expected benefit to potential victims from taking preventive or avoidance measures. Thus, if the government now takes actions that otherwise decrease the crime rate – such as significantly increasing the sanction for identity theft or credit card fraud – this will decrease the expected benefit of taking private preventive actions. Hence, potential victims will respond to the increased probability of detection or increased severity of punishment by reducing their own preventive or avoidance activities. Similarly, increased restitution will decrease the incentive for potential victims to take preventive measures. p. 355That does not mean restitution is bad or that the government should not increase the sanction for identity theft. Simply, we need to recognize the interrelationships between the actions of each actor in this system. By doing so, we can determine how to prioritize crimes in terms of government enforcement and sentencing policy. Finally, victims can affect the probability of conviction by the amount of cooperation they give to government officials.20 This will depend on their treatment in the criminal justice system, including the amount of time they must spend resolving legal controversies.

As noted above, the cost of prevention is likely to differ among potential victims. Thus, the elderly or mentally handicapped might be more vulnerable to telemarketing fraud, for example, which might justify more punitive sanctions where vulnerable victims are included. In addition, the government might have a higher cost of prevention than the private sector, as individual decision makers within the government do not have the same private incentives that their private sector counterparts have to help prevent the government from becoming a victim. This might justify more punitive sanctions for economic crimes against the government relative to crimes against private businesses.

2.3 Law enforcement

Law enforcement agencies – both investigative and prosecutorial – are obviously key actors in the criminal justice system. To the extent that devoting more attention to one type of crime increases the probability of detection, conviction, and punishment of criminal wrongdoing, this will result in a deterrent effect for that crime.21

Although we would like to think that all alleged crimes are given equal treatment in both investigation and prosecution, the reality is that government resources are limited and each agency must establish enforcement priorities.22 Those priorities might depend on the perceived harm caused by certain types of crimes, the difficulty of establishing criminal liability, community standards, and even the political aspirations of individual p. 356prosecutors.23 Given scarce resources and the need to prioritize workloads, one factor that inevitably affects investigators and prosecutors is the ultimate punishment the alleged criminal offender can expect to receive. Sentencing guidelines can send strong signals to prosecutors about which crimes should be the highest priority and which should be placed lower on the priority list. Thus, when the Commission increased the probability and severity of punishment for white-collar offenders in 1987, it should have had two effects. First, it should have deterred and incapacitated offenders, and second, it should have shifted the priorities of some law enforcement officials towards white-collar crime. Both of these effects tend to reduce white-collar crime.24

2.4 Courts

As should be clear by now, sentencing guidelines can have an important effect on the incentives of the other actors – and not just the criminals – in this ‘system’. Sentences issued by courts provide signals to all parties about the ‘price’ that one will have to pay for committing a crime that is detected and successfully prosecuted. Such potential offenders may be deterred if the sanction is severe enough. However, as indicated above, there are other effects from raising sanctions. Potential victims might take less care to prevent or avoid crimes as they benefit from the increased deterrent effect of more severe sanctions. Potential offenders might shift to other crimes that have relatively less onerous sanctions attached to them. Prosecutors might move certain crimes up or down their priority list as they adjust to these signals about the ‘payoff’ from their prosecutorial efforts.

Although most of these effects are intended, one must also be careful about unintended consequences. For example, courts setting very high penalties on crimes where innocent individuals are easily mistaken might deter socially beneficial activity as well as criminal activity. This is of particular concern for offenses that might be committed by legitimate businesses as opposed to businesses largely established for the purposes of committing criminal activities.

2.5 Potential mistaken offenders

Unfortunately, there is always the possibility that an innocent individual will be charged with a crime. There are many safeguards in our legal system p. 357to prevent such mistakes from happening – and in reality, it is probably a rare event. However, in some cases – especially in the area of economic crimes – the fact that there is always a possibility that an innocent person will be charged with a crime can have important behavioral consequences on the innocent.25

The concern over mistaken offenders is a particular problem in the area of economic or regulatory offenses involving commerce.26 For example, although stringent copyright protection laws are important, suppose copyright violations were a strict liability offense with severe punishments for offenders. This would certainly deter copyright violations and have a positive effect on innovation and creativity. Yet, there is also the possibility that well-meaning, law-abiding individuals and businesses who might otherwise use copyrighted material legally will decide that the risk of making a mistake and violating the law outweighs the potential gain from engaging in that line of business. Thus, there will be some negative effect on innovation and creativity. This is just one example, and even if this particular example is stretched, the concern is real. Anytime an illegal activity may come about as a result of a normal business activity, stringent penalties run the risk of deterring both the illegal and legitimate business activity. This is the ‘overdeterrence’ problem that economists often talk about in the context of corporate crime.27 It is a particular problem in emerging areas of criminal law where the scope of the law and definitions of illegal activity are largely untested.28 Thus, it is of particular concern in the area of new technology crimes. It is also why we often resort to civil sanctions for these types of offenses, where the remedy is injunctive relief and perhaps some monetary compensation. Thus, for example, the Federal Trade Commission’s (FTC) mandate to control misleading and deceptive advertising generally involves this form of sanction instead of more punitive approaches.29 Legitimate firms that want to try new advertising messages know that they will generally have the first bite of the apple, so that p. 358if a new practice is deemed to be misleading, the repercussions to the firm acting in good faith are not devastating.

Recently, this concern about overdeterrence has come to the forefront in efforts to control money laundering through wire transfers and prepaid credit cards. Reportedly, several companies left the wire transfer business following new rules enacted after 9/11, and there is concern that the same might happen with prepaid card providers if money-laundering rules and penalties are further tightened.30

3. Implications of the economic model of criminal justice

There are many ways which society can reduce crime and its effects. To the extent that we can educate, exert social pressures, and increase the stigma associated with certain activities, we might be able to reduce the propensity of potential offenders to commit crime. Potential victims might be able to do a better job of protecting themselves by taking more precautions, buying encryption software, asking more questions and otherwise taking private actions to reduce their probability of being victimized. We might spend more on law enforcement by hiring new Federal Bureau of Investigation (FBI) agents, more United States Attorneys, or better equipment to increase the productivity of our existing officers. This will raise the probability of detection and hence increase the deterrent effect of our law enforcement efforts. Finally, we might increase the severity of sanctions to deter potential criminals and/or incapacitate known criminals from committing repeat offenses. This chapter focuses on the implications of this economic model for setting rational sentencing policy. That means I will assume all of the other actors’ actions remain unchanged, except to the extent that sentencing behavior influences them.

3.1 Offenders should be held accountable for all losses and costs associated with their actions

Since the goal of the economic model is to optimally deter illegal behavior, and rational potential criminals take into account the expected punishment they will receive from their actions, we want to set penalties based on the losses caused by their criminal actions. This result follows from the ‘optimal penalty’ model of Gary Becker, which calls for a sanction equal to the harm divided by the probability of detection.31 Thus, all else equal, p. 359the severity of punishment should increase with the size of the loss.32 The difficulty in implementing this simple notion will become evident in a later section of this chapter, when I discuss the ‘foreseeability’ standard in defining loss.33 In addition, the severity of punishment should increase with the cost of apprehension, conviction, and the cost of imposing a penalty.34

3.2 Crimes that are difficult to detect or prosecute should be punished more severely

Becker’s model supports the theory that crimes which are difficult to detect or prosecute should be punished more severely.35 Potential offenders weigh the expected punishment against the expected gain they will receive from committing the crime. Since expected punishment is smaller when the risk of detection is small, potential offenders will tend to commit crimes that are relatively more difficult to detect and/or prosecute. Thus, the optimal penalty should increase in order to take this into account.

3.3 Certain punishment has a deterrent effect even if average punishment is unchanged

One of the most important insights of the economics of crime model is that if sanctions are set sufficiently high, only people who prefer risk can be expected to commit crimes.36 Empirical evidence supports this finding.37 Given this fact, one of the best ways to deter risk-preferring individuals from committing crimes is to impose a certain punishment that those individuals will foresee as being likely. Sentencing guidelines are a step in the right direction provided that they do not contribute to uncertainty. p. 360Thus, sentencing guidelines should not be based on random events that are beyond the control of the individual offender.

3.4 Reputation loss from conviction might be more important than prison sentence

The criminal sanction is only one of a myriad of punishments that might be imposed on criminal offenders.38 In addition to civil or administrative penalties, criminals might suffer the wrath of their community in terms of future employability. Doctors or lawyers might lose their licenses to practice their professions.39 Others will be kept away from positions of importance that rely upon trust or otherwise involve some type of fiduciary duty. Indeed, there is empirical evidence that criminal offenders with higher legitimate pre-conviction earnings pay a much more serious price in terms of post-conviction earnings.40 There have also been calls for increasing the use of ‘shaming’ as a form of punishment itself.41 From a policy perspective, criminal sentencing should take into account all of the other penalties – both government and market imposed – on the offender. Prison time is a last resort and is called for if the others do not adequately deter or there is a need for incapacitation to decrease the risk of recidivism.

3.5 Monetary sanctions are preferable to prison

Another important implication of the Becker model is that because prisons are costly, we should resort to less costly means of punishment to the extent that they offer adequate deterrence and otherwise serve the goals of punishment.42 Thus, if monetary penalties adequately deter and punish offenders, that is the preferred method. The difficulty in implementing this as policy is that the fine for many such offenders would easily exceed their assets; thus prison is necessary.43 To take a simple example, if there is a one in five chance of detecting and convicting a person who fraudulently obtains $1 million in a credit scam, the optimal penalty would be $5 million ($1 million of which might be used for restitution). In reality, p. 361in most cases, these scam artists have few assets remaining, and prison may be the only form of punishment available.

3.6 Need to take into account interdependencies of crimes and actors

For example, as discussed above, although potential victims can take precautions against victimization, the cost of avoidance might vary across individuals. In some cases, it might cost less for potential victims to avoid victimization than it does for law enforcement agencies to prevent the crime from occurring. However, in other cases, the cost of avoidance might be prohibitive. In those instances, where crimes impose higher avoidance costs or where it is more difficult for potential victims to avoid victimization, offenders should be punished more severely relative to crimes where potential victims can cheaply avoid the crime.

The fact that the criminal justice arena is an interdependent system also suggests that before a policy is developed, the behavior of other actors is considered. Indeed, there is some evidence that while the Commission increased corporate criminal sanctions in 1990, this increase was partly offset by reducing other monetary sanctions – hence, the overall effect was not as punitive as envisioned.44

4. Application of economic crime control model to sentencing of economic crimes

The US Sentencing Commission published their first set of Guidelines in 1987 following passage of The Crime Control Act of 1984 which established the Commission as an independent regulatory body to write guidelines for sentencing all federal criminal offenders.45 In addition to the overall goal of reducing disparity through restricting judicial discretion, Congress specifically called for increased sanctions for white-collar offenders.46 The Commission followed this mandate by increasing both the probability of imprisonment and the length of the sentence for most white-collar offenses. It also increased monetary penalties significantly.

Generally, the Guidelines are based on ‘offense levels’, where each crime is given a base level and increases or decreases from that level are based on various elements of the offense (for example, whether a weapon p. 362was present, if there was a particularly vulnerable victim, or the dollar magnitude of the loss or gain). These offense levels, coupled with information on an offender’s prior criminal history, determine whether or not any prison time is warranted, and if so, the appropriate length (expressed in a range where the judge has some discretion). Offenders are virtually always required to pay restitution if there is a readily identifiable victim. Monetary penalties for corporations convicted of federal crimes are based on a ‘multiple’ of up to four times the maximum of ‘loss’ or ‘gain’.

The remainder of this section applies the economic crime control model to several specific proposals that were under consideration in 2000 by the Commission when revising their guidelines for economic crimes. Section 4.1 considers the overriding question of whether the overall punishment for economic crimes should be increased. A variation on this proposal was to increase the punishment only for frauds involving very large losses. Section 4.2 examines whether the offense level should be based on ‘actual’ versus ‘intended’ loss and the extent to which foreseeability should be used in defining loss. Finally, Section 4.3 considers whether or not interest should be included in the loss definition used to calculate the monetary fine.

4.1 Increased offense levels for fraud

One proposal considered by the Commission was to increase the relationship between monetary losses and the offense level – which would ultimately increase the length of prison sentences for larger frauds. One commentator, the Judicial Conference’s Committee on Criminal Law (‘Judicial Conference Report’), indicated that the fraud and theft guidelines were too lenient relative to street crime, and that the slope of the loss table was not steep enough.47 In support of that position, they noted that there was no ‘proportionality’ between dollar losses and offense levels. For example, while there is a three-level difference between a $10,000 fraud and a $70,000 fraud, there is only a one-level difference between and a $1.5 million and a $2.5 million fraud.48

Although the jury is still out on the need to increase the loss tables, there is no compelling evidence that such changes are necessary or even prudent. First, sanctions have increased dramatically for economic crime offenders relative to pre-Guideline sanctions.49 Second, the effect of inflation on p. 363the loss tables has increased (and will continue to increase) the severity of punishment.50 Third, there is no empirical evidence that economic crimes have been increasing in number or severity. Fourth, the notion of proportionality is not one that should be operationalized by simple statistics. In addition to considering these arguments below, this section proposes reforms that would address some of the concerns of proportionality in a manner that is more consistent with economic theory and the concept of ‘marginal deterrence’.51

4.1.1 Sanctions have increased in severity compared to pre-guideline sanctions

When the Commission first promulgated the Guidelines in 1987, it made a conscious effort to increase both the percentage of white-collar offenders who spend time in prison and the average length of prison terms for those who are sentenced to prison.52 In addition, just two years later the Commission changed the fraud loss table and increased the severity level for all frauds greater than $40,000.53

According to Commission data, sanctions have become increasingly severe for economic crime offenders (defined as including fraud, tax evasion and embezzlement).54 In fact, as shown in Table 14.1, the rate of imprisonment has increased from 39 percent in 1984 to 61 percent in 1999 – a 50 percent increase. Of those who received prison sentences, the average length for economic crime offenders increased slightly from about 14 months to 18 months. The reason the average sentence length for economic crimes did not increase very much was primarily due to the fact that many offenders who previously would have received probation now spend some time in prison. However, the short prison sentences that these offenders received offset the increased sentence length that the Guidelines imposed on other economic crime offenders. Thus, the Guidelines did increase the severity of punishment for economic crimes. More importantly, the ‘expected’ prison sentence (probability of serving time multiplied by the average time served) has doubled, from 5.5 months pre-Guidelines to 11.0 months today.55

Table 14.1

p. 364Prison Time for Economic Crime Offenders Sentenced under Federal Sentencing Guidelines, 1984—1999*

Percent prisonAverage time served (for those serving time)‘Expected prison’
198439%14 months5.5 months
199051%14 months7.1 months
199961%18 months11.0 months

Note: * See Report on the Operation of the Guidelines System, supra note 54, at 377, 379. See also US Sentencing Commission, Sourcebook of Federal Sentencing Statistics (1999), pp. 20, 28.

4.1.2 Inflation has increased the severity of punishment

The Guidelines first went into effect in 1987. Although the inflation rate in the US has been relatively low since then, time does take its toll. For example, $100 in 1987 had the same buying power as $150 in the year 2000. Thus, if we compare the ‘seriousness’ of two frauds, each involving $6,000, for example, but one being committed in 1987 and another in the year 2000, the 1987 fraud was much more serious in terms of victim and social harm. Yet, if we look at the loss tables, both frauds of $6,000 earn base offense levels of eight (six plus two for the $6,000 loss).56 In fact, if we convert the 1987 fraud into year 2000 dollars, it would be the equivalent of a $4,000 fraud in 2000. However, a $4,000 fraud only receives an offense level of seven.

The effect of inflation has therefore been to increase the severity of the fraud, theft and tax loss tables by approximately one level between 1987 and 2000. Furthermore, this inflationary creep in the Guidelines will continue every year into the future, eventually increasing severity by another level and onward. This inflationary creep affects the entire loss table. However, it does not affect the monetary fine calculation for organizational defendants, since inflation does not affect the multiple that is applied to the loss.

4.1.3 No empirical evidence that economic crimes are on the rise

Although there is anecdotal evidence that economic and white-collar crime is on the p. 365rise, there is no empirical evidence to support this claim. As one who has studied white-collar and corporate crime for years, I know the difficulty of measuring crime rates. Unlike street crimes, there are no comprehensive government surveys or systematic collection agencies that report on the incidence of white-collar or corporate crime. Often, victims are unaware or misinformed about their victimization, making any such estimates suspect.

What evidence do we have that fraud is on the rise? We are commonly told that health-care fraud is an increasing problem. A poll by the AARP in 1999 found that 90 percent of respondents believed health-care fraud was steady or increasing.57 Yet, according to government estimates, Medicare ‘actually lost about seven cents of every dollar to fraud, waste and mistakes in 1998. . . [which is] only half of what was lost by the government’s health insurance program for the elderly and disabled just two years ago’.58 This reduction in fraud was attributed to stepped-up enforcement.59 This is just one example of how there is misinformation about the growth in fraud.

What about fraud convictions? According to IRS data, criminal pros-ecutions were recommended in 3,526 cases in 1987,60 compared to an almost identical number (3,427) in 1998. However, the number of convictions per year appears to have increased from about 2,500 to 3,000 per year.61 Similarly, the number of convictions for financial fraud reported by the United States Department of Justice (DOJ) Criminal Division increased slightly from 2,309 in 1987 to 2,613 in 1998.62 According to Commission data, the number of fraud offenders sentenced under the Guidelines has remained relatively stable, with 5,905 offenders sentenced in 1995 and 6,199 in 1999.63 Embezzlement increased from 809 to 949, while tax offenses have decreased from 737 to 722 over the same time p. 366period.64 None of these figures indicates a significant increase in cases tried in Federal courts between 1987 and 2000.

As noted above, the Guidelines significantly increased the likelihood that white-collar offenders would spend time in prison – and increased the average time served. Recall from the model that an increase in expected punishment is likely to have several effects. Although it might deter potential offenders from committing the crime, it might also spur law enforcement agencies and prosecutors to pursue more cases against those offenders since the ‘payoff’ is now greater. Therefore, even if there is a deterrent effect from the higher punishment, the ‘detection’ effect might offset, and even overwhelm, any reduction in crime. That is, even if the crime rate is going down, we might observe more convictions. Thus, it is generally not appropriate to draw inferences about the rate of crime from sentencing data alone. Dramatic changes in sentencing policy are only justifiable for addressing to dramatic increases in fraud or the rate of crime in general. Currently no persuasive evidence of such increases exists.

4.1.4 Proportionality requires more than simple statistics

As noted above, the loss tables that determine the offense level were criticized for being ‘flat’. For example, while there is a three-level difference between a $10,000 fraud and a $70,000 fraud, there is only a one-level difference between a $1.5 million and a $2.5 million fraud. The Commission originally set the loss table for fraud by empirically estimating the relationship between loss and prison time from past practice and using that as a guide.65

At a minimum, we should note that the ‘flat’ structure was consistent with judicial preferences. Is there any reason why doubling the dollar value of a fraud at low levels should result in the same increase in offense level as it does at high levels of loss? Is there any reason why a doubling of the dollar value of a fraud should double the length of a prison sentence? The answer to both these questions is ‘no’.

In theory, there are several reasons why the loss table should be ‘flat’. First, larger frauds are generally easier to detect since they involve more victims, larger losses per victim, and/or victims who are more eager to obtain restitution and retribution. Second, the fact that criminals tend to be risk preferring suggests that a more certain punishment has more of a p. 367deterrent effect than a more severe punishment. Thus, we should be willing to trade off reduced sentences at the upper end of the distribution for more certain – but shorter-sentences for most white-collar offenders. Third, individuals (especially criminal offenders) are likely to have very large discount rates (that is, they put a large premium on current consumption and less weight on their future).66 This reduces the deterrent value of long prison sentences.

In addition to proportionality across the fraud loss table, the Judicial Conference noted that there is no proportionality between street crimes and economic crimes.67 However, it is difficult to compare the severity of punishment across crime types unless factors such as harm and detectability are compared as well. Presumably, someone who sells $10,000 worth of cocaine or takes $10,000 by threat of physical harm causes more social harm than someone who embezzles $10,000 from a bank. Accordingly, a small fraud starts with a base offense level of six, while the small-time cocaine dealer receives a base level of 12 and a robbery has a base offense of 20 (or 22 for a bank robbery).68 Even burglary has a higher base level (17 for burglary of a residence or 18 for a burglary of $10,000).69 Note that the commentary to the Guidelines indicates that residential burglary has a higher base level than other burglaries because of the increased risk of physical or psychological harm.70 To the extent that fraud involves victims who suffer serious psychological trauma or financial hardships, one might want to increase the base offense level. The current Guidelines do this with a provision for a two-point increase for vulnerable victims, or a four-point increase for a large number of vulnerable victims.71

At lower levels of sales, it appears there is actually some comparability across drugs and fraud values. According to the 1995 Commission Report on Cocaine and Sentencing Policy, a marijuana defendant selling $42,000 worth of drugs would receive an offense level of 14.72 This is only one offense level higher than a fraud defendant with a loss of $42,000 p. 368and more than minimal planning would receive.73 The difference in base sentences is only about three months. However, the larger dollar values of between $500,000 and $8 million (depending on the drug) would have a base offense level of 32.74 The highest offense level for fraud with more than minimal planning is 26.75 Of course, the difference between an offense level of 26 and 32 is five years – the former being slightly higher than five years, while the latter is greater than ten years.

Therefore, it does appear that both the offense level and prison sentence increase faster for drug offenses than they do for comparable dollar values of fraud. But there are other reasons why the large fraud offender might be dealt with less severely than the large drug dealer. First, it is much more difficult to detect and convict a large drug dealer. They have sophisticated networks of workers who shield them from legal culpability. More important, perhaps, is the fact that there are generally few readily identifiable victims who would complain about the drug dealer. On the other hand, large frauds are likely to be exposed, as victims or class action lawyers will no doubt seek out legal authorities to remedy their concerns. Thus, I would expect frauds to be much more easily detected and a perpetrator convicted than in the case of drug offenses. Recall, in the model, that more detectible crimes require less severe sanctions for optimal deterrence.

4.1.5 Proposal for differentiating between temporary and permanent losses

A fraud that results in victim losses and ultimate restitution paid of $1 million is different from a fraud that results in $1 million where no restitution is ever paid. Yet, the current Guidelines treat these offenses as being identical. Restitution is generally required where victims and losses are identifiable.76 However, the judge is free to reduce restitution below victim loss based on financial considerations and the assets of the offender.77 Further, there is no penalty to the offender if the amount of restitution ordered does not fully cover victim losses. The result of this anomaly is that there is an incentive built into the Guidelines to hide assets or squander them on a high style of living. An alternative rule would differentiate between offenders that can afford to pay restitution and those that cannot.

Although there are many ways that one could implement such a proposal, one method might be to add the uncompensated amount to the p. 369loss calculation for purposes of calculating an appropriate offense level. Alternatively, the Commission might spell out specific offense characteristics leading to an increase in the offense level depending on how much of the victim loss is not recoverable.

This proposal would partly address one of the concerns over the severity of the loss tables that was mentioned in the Judicial Conference Report. The report took note of United States v. McDowell, where all but $8,000 of a $300,000 fraud had been spent and was not recoverable for victim restitution.78 The concern expressed by the Judicial Conference centered on what it considered a low offense level of 15 and a resulting guideline range of 18 to 24 months. This range was based on an offense level calculation of six for the base fraud, eight for the dollar loss, and one additional net point for offense and offender adjustments. The trial judge ignored the Guidelines and imposed an upward departure to a level 19 and a sentence of 37 months. The trial judge’s reasoning was overturned, but the sentencing departure was upheld on other grounds. The proposal by the Judicial Conference was to raise the severity of the loss tables so that a loss of this magnitude would yield a level 18, which provides for 27 to 33 months.79 However, that approach would also increase the sanction for the fraud offender who did not spend the money and was able to pay restitution. My proposal would treat these two offenders differently. For example, if the Commission were to simply add uncompensated losses to the existing loss estimate, the offender who paid back the $300,000 in restitution would still receive a level 15 with a range of 18 to 24 months. However, the offender that squandered all but $8,000 would be charged with a loss of $592,000 ($300,000 fraud plus $292,000 unpaid restitution), for an offense level of 17 and a sentence of 24 to 30 months. Alternatively, specific upward adjustment could be built into the Guidelines to increase the offense level for such cases accordingly.

4.2 Intended versus actual loss and the issue of causation

One of the most difficult issues in drafting Guidelines is whether to use intended or actual loss as the basis for computing an offense level. Related to this question is what standard of ‘causation’ should be used in determining loss. To answer these questions, recall the purpose of sanctions – to provide foreseeable consequences for illegal behavior that will optimally p. 370deter potential offenders from committing crimes. In other words, potential offenders should be able to anticipate the punishment they will receive if they commit an act.

From a pure deterrence standpoint, if an individual offender is risk neutral, it does not matter whether the penalty is based on intended or actual loss. Before deciding on whether or not to commit a crime, the ‘expected’ loss is all that can be estimated, and thus the potential offender relies upon that measure in determining whether or not to commit a crime. The fact that will be punished on the realization of actual loss does not change that decision calculus if he is risk neutral. However, if the potential offender is risk preferring (as, I argued earlier, criminal offenders tend to be), he prefers the gamble to the expected value of the outcome. This means that a penalty based on the expected value of the outcome will have a greater deterrent effect than one that is based on the actual outcome.

Although ‘expected loss’ is the appropriate criterion for determining offense levels, note that if a loss is not foreseeable, it cannot be part of the expected loss calculation. That is, if the potential offender does not foresee the possibility of an outcome occurring, it is not taken into account in the decision calculus about whether or not to offend. Thus, a ‘foreseeability’ standard that ignores such losses would be appropriate. The difficulty is in determining which losses are foreseeable.

Consider two different offenders, both of whom commit insurance frauds that take $100,000 from customers with no intention of paying claims. Total claims by customers of the first fraud are $50,000, while customers of the second fraud incur $10 million in losses following a disastrous hurricane. Under most definitions of ‘foreseeable’, one would argue that the first offender’s customer claims of $50,000 were foreseeable. However, even in the second case, courts have ruled that such losses were indeed caused by the fact that there was no insurance and hence the full value of the victim losses is the proper measure of loss.80 I believe these courts are largely correct in considering the loss foreseeable as long as there is some estimable chance that a hurricane would hit the area being insured. The problem is in the current rule that calls for calculation of actual instead of expected loss for purposes of determining the offense level – irrespective of whether the loss was foreseeable.

Under the current loss rules, the first offender is likely to be held liable for $150,000 p. 371in losses ($100,000 in fraudulent premium plus $50,000 in claims not paid), while the second will be held accountable for $10,100,000 in losses ($100,000 in fraudulent premium plus $10 million in unpaid claims). This effectively means that offenders are entering a lottery, where they know that if they get caught, they will be held accountable for lost premiums plus any actual losses that occur. As noted above, the ideal way to handle these two cases would be to sanction them the same – based on the ‘expected loss’. In this case, expected loss would equal the premiums ($100,000) plus the average payout of insurance policies of this type. Thus, if 90 percent of premiums normally go to pay claims, the expected loss in this case would be $190,000. This is higher than the loss for the first offender, but lower than the loss for the second one.

Note that I have not argued for $190,000 in restitution. Restitution has a completely different goal than efficient crime control – making the victim whole. In this example, the restitution amount for the first offender should be $150,000, while the second offender would be entitled to $10,100,000. Even if an ‘intended loss’ standard is adopted for purposes of computing an offense level, actual loss plays an important role in sentencing the fraud offender. Moreover, note that in the previous section, I proposed that the unpaid restitution be used to increase the offense level of the crime in the case wherein the second offender does not pay full restitution, thus preserving marginal deterrence.

My analysis of these two insurance frauds is contrary to both the current approach and the proposed changes to loss rules. Furthermore, I realize that there are many potential objections to my proposal. First, it might be unworkable in practice, as it is often more difficult to estimate the expected loss in a case than to observe actual loss. A careful analysis of recent cases and perhaps some boilerplate provisions to reduce the burden on the sentencing court could overcome some of these objections. Second, because of the manner in which we communicate crimes and punishment, concern might be expressed that it would send a bad signal to potential criminals if the sanction for imposing a loss on victims of over $10,000,000 appears to be a slap on the wrist. Third, by imposing higher sanctions when actual loss is greater, one might argue that we provide an incentive for offenders to avoid these more costly outcomes. Thus, for example, if potential offenders were contemplating setting up a fraudulent insurance scheme, they might think twice about locating the operation in Florida, and instead move to an area where extreme outcomes are much less likely. Thus, to the extent that offenders can reduce the likelihood of some of these ‘controllable events’, using an actual loss standard is likely to lower actual losses. My response to these latter two concerns, however, is that the requirement that restitution be imposed, coupled with my proposal to increase loss by any unpaid restitution, largely overcomes these concerns and preserves marginal deterrence.

4.3 Interest

p. 372Currently, the loss rules for fraud exclude ‘interest the victim could have earned on such funds had the offense not occurred’.81 Similarly, in a fraudulent loan case, the rules appear to exclude interest payments a lender might have received and instead focus on ‘the amount of loan not repaid at the time the offense is discovered, reduced by the amount the lending institution has recovered. . .’.82 That is, although lending institutions might incur significant foregone interest income on a fraudulent loan, this amount is not included as a loss.

Apparently, there is disagreement among the courts over how to interpret the current Guidelines, as many circuits have decisions that explicitly include interest that had been bargained for, but exclude other interest calculations based on the opportunity cost of the fraudulent losses. For example, courts have included late fees and interest in cases involving unauthorized use of credit cards or specific investment returns that were promised but not delivered.83 However, they have refused to permit similar lost profits in fraudulent investments when interest payments were not specified.84 The distinction is supposedly between contractually promised interest and the opportunity cost or time value of money.

Economic theory would tell us that interest is an important component of loss. The time value of money is well established in economics as being worthy of consideration in all areas of cost-benefit analysis and policy analysis in general.85 From the perspective of the victim, there is no doubt that foregone interest is a loss. Moreover, from the perspective of the potential offender who is deciding whether or not engage in a fraudulent activity, the time value of money will certainly play a role.

For example, an offender who defrauds victims out of $1 million and is able to hold that amount for several years before being sentenced will have p. 373earned (or been able to earn) hundreds of thousands of dollars in interest. The fact that this is not considered a loss for purposes of sentencing means that the offender has an incentive not to plead guilty and settle the case, since he has use of those resources for a longer period of time at no real cost to himself. Of course, interest may be charged in restitution orders at the discretion of the judge.86 Therefore, to the extent that restitution with interest is ultimately ordered, some of the incentive for delay is reduced.

Interestingly, the Staff Report summarizing the comments of interested parties on this matter found near agreement that interest should generally be excluded from the loss calculations.87 The reasons appear to be more on practical than theoretical grounds. Essentially, the arguments are: (a) interest would only be a small part of loss and hence not have a significant impact on sentencing outcomes, and (b) estimating interest would be a burden. To the extent that these concerns are valid, one possibility would be to allow the sentencing court to consider interest if it believes that doing so would more adequately reflect the severity of the offense and where estimating interest is not deemed by the court to be too burdensome.

5. Conclusion

Economic theory, empirical evidence, and social science in general, can play an important role in developing guidelines for sentencing offenders convicted of economic crimes. The Guidelines were originally written with the goal of reducing disparity while imitating past practice. However, punishment for economic crimes was made more punitive by increasing the likelihood of a prison sentence as well as the average length of prison for those who are incarcerated. Aside from notions of fairness or justice, one of the benefits of reduced disparity is the fact that ‘certainty of punishment’ has crime control benefits. That observation is largely based on the fact that criminal offenders tend to be risk-loving and would prefer the gamble of a prison sentence to a certain punishment. As a result, the original Guidelines – coupled with upward revisions in 1989 and the built-in inflationary impact of the loss tables – resulted in a significant increase in the severity of criminal sanctions for economic crime offenders.

The rational actor paradigm helps explain the behavior of potential offenders, potential victims, law enforcement agencies, and courts. Understanding the behavior of these actors provides important insights p. 374into the effect that proposed changes in sentencing policy would likely have on crime rates. Key to this understanding is awareness that sentencing policy should not be viewed in isolation, but instead as part of an interdependent crime and punishment system. Increasing the severity of punishment might deter crime, but it might also have other unintended consequences such as making other crimes more attractive, causing offenders to take more expensive and elaborate activities to evade detection, causing potential victims to take less care in preventing victimization, and encouraging law enforcement officials to prosecute more cases of that nature. In some cases, increasing the severity of punishment might even stifle legitimate behavior. Thus, it is important that the Commission work through the implications of any Guideline revisions on the likely actions of other parties.

This chapter is based largely on a paper entitled, ‘Economic Theories of Crime Control: Implications for Federal Sentencing Policy’, presented at the Symposium on Federal Sentencing Policy for Economic Crimes and New Technology Offenses, sponsored by US Sentencing Commission and George Mason University Law School, October 12–13, 2000. That paper was subsequently published as, ‘The Economics of Crime and Punishment: Implications for Sentencing of Economic Crimes and New Technology Offenses’, George Mason University Law Review, 9 (2), 503–28 (winter 2000). The George Mason University Law Review has granted permission to reprint (and update) this article.

See Mark A. Cohen and Sally S. Simpson, ‘The Origins of Corporate Criminality: Rational Individual and Organizational Actors’, Chapter 2 in Debating Corporate Crime, edited by William S. Lofquist, Mark A. Cohen and Gary Rabe (Cincinnati, OH: Anderson Publishing Co. and Academy of Criminal Justice Sciences, March 1997).

See Gary S. Becker, ‘Crime and Punishment: An Economic Approach’, Journal of Political Economy, 76, 169, 171 (1968) (hereinafter ‘Crime and Punishment’) (‘Public expenditures in 1965 at the federal, state, and local levels on police, criminal courts and counsel, and “corrections” amounted to over $4 billion, while private outlays on burglar alarms, guards, counsel, and some other forms of protection were about $2 billion’); see also David A. Anderson, ‘Aggregate Burden of Crime’, Journal of Law and Economics, 42, 611, 612 (1999) (‘Overt annual expenditures on crime in the United States include $47 billion for police protection, $36 billion for correction, and $19 billion for prosecution (the legal and judicial costs) of state and local criminal cases’).

See ‘Crime and Punishment’, supra note 3, at p. 180.

See Kenneth G. Dau-Schmidt, ‘An Economic Analysis of the Criminal Law as a Preference-Shaping Policy’, Duke Law Journal, I, 11–12 (1990).

See Jeffrey S. Parker, ‘Criminal Sentencing Policy for Organizations’, American Criminal Law Review, 26, 513, 554–63 (1989) (discussing optimal penalties and overdeterrence).

See Daniel L. Rubenfeld, ‘Econometrics in the Courtroom’, Columbia Law Review, 85, 1048, 1051 (1985) (‘Type 1 errors involve the cost of concluding that an activity was illegal – for example, that there was discrimination amounting to a violation of Title V11 – when in fact it was not. Type 2 errors involve the cost of wrongly concluding that an activity was not illegal, when in fact it was’). We are seeing this to some extent in the area of privacy rights on the Internet.

Technically, ‘social cost’ and ‘loss’ are not the same thing. Loss is closer to the concept of an ‘externality’ than to a social cost, since the former may include private losses (for example, transfers from victim to the offender) that are not always considered to be social losses. For a discussion of these concepts in the context of crime, see Mark A. Cohen, The Costs of Crime and Justice (New York, NY: Routledge, 2005).

See US Sentencing Commission Guidelines Manual (hereinafter Guidelines), § 2B1.1 – Larceny, Embezzlement, and Other Forms of Theft, Offenses Involving Stolen Property; Property Damage or Destruction; Fraud and Deceit; Forgery; Offenses Involving Altered or Counterfeit Instruments Other than Counterfeit Bearer Obligations of the United States.

For a description of the goals of sentencing in the criminal law, see Model Penal Code § 1.02(2)(a–d) (1985).

For example, although I do not think I would ever be tempted to commit a fraud or evade taxes, the reason I am not so tempted is not that I am inherently any different from someone who does. Instead, the perceived costs of evading taxes, for example, far outweigh any benefits I can see from doing so. In other words, in choosing what is best for me, I weigh the benefits from tax evasion – saving money – against the costs of tax evasion – which include the expected penalty, possible prison sentence, and moral outrage I would expect to endure from my family, friends, and community. On April 15 every year, I choose to honestly report my taxes. However, for someone who chooses otherwise, they perceive the expected benefits from tax evasion to exceed the expected cost to them.

See generally Kenneth W. Simons, ‘Assumption of Risk and Consent in the Law of Torts: A Theory of Full Preference’, Boston University Law Review, 67, 213 (1987) (providing a detailed explanation of risk preference theory).

See David J. Elbaz, ‘The Troubling Entrapment Defense: How About an Economic Approach?’, American Criminal Law Review, 36, 117, 141–2 (1999) (explaining how criminals consider opportunity costs when contemplating criminal activity).

See Jonathan Todd Laba, ‘If You Can’t Stand the Heat, Get Out of the Drug Business: Thermal Imagers, Emerging Technologies, and the Fourth Amendment’, California Law Review, 84, 1437, 1441 (1996) (providing examples of new crime fighting technologies).

See generally Sylvia G. McCollum, ‘Mock Job Fairs in Prison–Tracking Participants’, Federal Probation 64, 13 (2000) (describing an in-prison job fair program in Texas).

See generally Dan M. Kahan, ‘Social Influence, Social Meaning, and Deterrence’, Virginia Law Review, 83, 349 (1997) (describing how stigmatizing crime reduces criminal activity).

See Laba, supra note 14; see also Miriam A. Cavanaugh, ‘If You do the Time, You Will do the Time: A Look at the New “Truth in Sentencing” Law in Michigan’, University of Detroit Mercy Law Review, 77, 375 (2000) (describing Michigan’s new sentencing regime that sets stricter sentences for criminals).

See Elbaz, supra note 13.

See for example, Mark Warr, ‘Public Perceptions and Reactions to Violent Offending and Victimization’, in Understanding and Preventing Violence: Consequences and Control of Violence, volume 4. edited by Albert J. Reiss, Jr and Jeffrey A. Roth, Committee on Law and Justice, Commission on Behavioral and Social Sciences and Education, National Research Council (Washington, DC: National Academy Press, 1994), pp. 11–15.

See Deborah Kelly, ‘Victim Participation in the Criminal Justice System’, in Victims of Crime: Problems, Policies, Programs, edited by Arthur J. Lurigo, R.C. Davis and W.G. Skogan (1990), pp. 172–3 (noting that lack of victim cooperation results in high dismissal rates for criminal prosecutions).

However, it may also increase the rates of other crimes on the margin.

See G. Robert Blakey, ‘Federal Criminal Law: The Need, not for Revised Constitutional Theory of New Congressional Statutes, but the Exercise of Responsible Prosecutive Discretion’, Hastings Law Journal, 46, 1175, 1246 (1995) (discussing prosecutorial enforcement priorities in the federal system).


See Mark. A. Cohen, ‘Sentencing Guidelines and Corporate Criminal Liability in the United States’, Business and the Contemporary World, 4, 140, 150 (1992) (noting that increasing sanctions for certain crimes provides an incentive for the government to search for and prosecute those types of crimes).

See Michael L. Travers, ‘Mistake of Law in Mala Prohibita Crimes’, University of Chicago Law Review, 62, 1301, 1325 (1995).

See Jeffrey S. Parker, ‘The Economics of Mens Rea’, Virginia Law Review, 79, 741, 757 (1993) (describing the effects of overdeterring crime).

See id.

See Kathyleen A. O’Brien, ‘Strategies for Successfully Defending against Federal Trade Commission Investigations of False and Deceptive Advertising’, 1010 Practising Law Institute PLI Order No. B407202, 269, 312–26 (1997).

See Porter & Dietsch, Inc. v. FTC, 605 F.2d 294, 309 (1979) (citing case law supporting the proposition that the FTC has discretion over what punishment to issue in cases of deceptive advertising, depending on the knowledge of the deception).

See ‘Prepaid Cards: The Cleanup’, Business Week, March 3, 2008, p. 32.

See ‘Crime and Punishment’, supra note 3, at 172–80. A more detailed discussion of how the Becker model applies to corporate crime and punishment in the context of regulatory offenses can be found in Mark A. Cohen, ‘Environmental Crime and Punishment: Legal/Economic Theory and Empirical Evidence on Enforcement of Federal Environmental Statutes’, Journal of Criminal Law and Criminology, 82, 1054–108 (1992).

Id., at pp. 170–71.

See infra Section 4.2.

See Mark A. Cohen, ‘Optimal Criminal Fines for Organizations’, working paper prepared for US Sentencing Commission (January 1988); A. Mitchell Polinsky and Steven Shaven, ‘Enforcement Costs and the Optimal Magnitude and Probability of Fines’, Journal of Law and Economics, 35, 133–48 (1992) (including formal models that argue enforcement costs should be included in the penalty calculation).

See ‘Crime and Punishment’, supra note 3, at p. 174.

See id. at p. 180.

See Michael K. Block and Vernon E. Gerety, ‘Some Experimental Evidence on Differences between Student and Prisoner Reactions to Monetary Penalties and Risk’, Journal of Legal Studies, 24, 123–38 (1995) (providing experimental evidence that prisoners respond more to the certainty of punishment than students, who respond to the severity of punishment).

See ‘Crime and Punishment’, supra note 3, at pp. 179–80.

See John R. Lott, Jr, ‘Do We Punish High Income Criminals Too Heavily?’, Economic Inquiry, 30, 583–608 (1992); John R. Lott, Jr, ‘The Effect of Conviction on the Legitimate Income of Criminals’, Economic Letters, 34, 381–5 (1990).

See id.

Dan M. Kahan and Eric A. Posner, ‘Shaming White-Collar Criminals: A Proposal for Reform of the Federal Sentencing Guidelines’, Journal of Law and Economics, 42, 365–91 (1999); John Braithwaite, Crime, Shame and Reintegration, New York: Cambridge University Press (1989).

See ‘Crime and Punishment’, supra note 3, at p. 180.

Id., at pp. 197–8.

See Cindy R. Alexander, Jennifer Arlen, and Mark A. Cohen, ‘Regulating Corporate Criminal Sanctions: Evidence on the Effect of the U.S. Sentencing Guidelines’, Journal of Law and Economics, 42, 393–42 (1999).

Comprehensive Crime Control Act of 1984, Public Law 98-473, October 12, 1984. The Guidelines, which were first issued in 1987, are updated annually.

See US Sentencing Commission, Supplemental Report on the Initial Sentencing Guidelines and Policy Statements (1987), p. 18.

Memo from Honorable J. Phil Gilbert, Committee on Criminal Law of the Judicial Conference of the United States, to Honorable Richard P. Conaboy, Chairman, US Sentencing Commission, April 3, 1998.

See Guidelines, § 2B1.1(b)(1)(f–i), (o-p).

See Stephen Breyer, ‘The Federal Sentencing Guidelines and the Key Compromises upon which they rest’, Hofstra Law Review, 17, 1, 20–22 (1988).

See infra Section 4.1.1.

The theory of marginal deterrence focuses on the incentives, created by particular schemes of punishment, for similarly situated offenders. See David D. Friedman and William Sjostrom, ‘Hanged for a Sheep: The Economics of Marginal Deterrence’, Journal of Legal Studies, 22, 345 (1993).

See Guidelines, Chapter 1 introduction.

Guidelines amendment 154 (effective November 1, 1989).

See US Sentencing Commission, The Federal Sentencing Guidelines: A Report on the Operation of the Guidelines System and Short-term impacts on Disparity in Sentencing, Use of Incarceration, and Prosecutorial Discretion and Plea Bargaining (1991), p. 368 (hereinafter, Report on Operation of the Guidelines System) (defining economic crimes as including fraud, tax evasion and embezzlement).


See Guidelines, §2B1.1(b)(1) – Fraud Loss Table. Note that this table has still not been updated as of 2008.

AARP, America Speaks out on Health Care Fraud: A Consumer Survey, Washington, DC: AARP (1999) p. 5.

Eileen O’Connor, ‘Federal Campaign Enlists Seniors as Medicare “Fraud Busters”’ (February 24, 1999), available at http://www.cnn.com/ALLPOLITICS/stories/1999/02/24/medicare.oconnor.


US Department of Justice, Bureau of Justice Statistics, Sourcebook of Criminal Justice Statistics (1999), p. 474, available at http://www.albany.edu/sourcebook.


Id., at p. 314.

US Sentencing Commission, Sourcebook of Federal Sentencing Statistics (1996), p. 7 (hereinafter 1996 Sourcebook); US Sentencing Commission, Sourcebook of Federal Sentencing Statistics (1999), p. 12 (hereinafter 1999 Sourcebook).

US Sentencing Commission, Sourcebook of Federal Sentencing Statistics (1995), p. 7 (hereinafter 1995 Sourcebook); 1999 Sourcebook, supra note 63, at p. 6.

This is based on the author’s understanding of conversations with Commission staff at the time the Guidelines were adopted in 1987.

See Richard A. Posner, ‘An Economic Theory of Criminal Law’, Columbia Law Review, 85, 1193, 1213–14 (1985).

Memorandum from Honorable J. Phil Gilbert, Chair, Committee on Criminal Law of the Judicial Conference of the United States to Honorable Richard P. Conaby, Chair, United States Sentencing Commission (April 3, 1998).

See USSG §§ 2D1.1(c)(14), 2F1.1(a), 2B3.1(a).

USSG § 2B2.1(a)(1).

Guidelines, § 2B2.1.

See Guidelines, § 3A1.1(b).

US Sentencing Commission, ‘Special Report to the Congress: Cocaine and Federal Sentencing Policy’, (February, 1995), available at http://www.ussc.gov/crack/exec.htm.

Guidelines, § 2F1.1.

Guidelines, § 2D1.1(a)(1–3).

Guidelines, § 2F1.1(b)(2).

Guidelines, § 5E1.1(a)(1).

Guidelines, § 5E1.1(f).

109 F.3d 214 (5th Cir. 1997).

Memorandum from Honorable J. Phil Gilbert, Chair, Committee on Criminal Law of the Judicial Conference of the United States to Honorable Richard P. Conaboy, Chair, United States Sentencing Commission (April 3, 1998).

See United States v. Neadle, 72 F.3d 1104, 1108–9 (3d Cir. 1996), amended by 79 F.3d 14 (1996).

Guidelines, § 2F1.1, comment note 8.

Guidelines, § 2F1.1 n. 8(b).

See United States v. Lowder, 5 F.3d 467, 471 (10th Cir. 1993).

Id. (distinguishing ‘opportunity cost’ interest losses versus specific interest rates promised but not delivered, as in United States v. Bailey, 975 F.2d 1028 (4th Cir. 1992)). ln Bailey, investors were never promised a specific return on investment. See id. Thus, the court ruled that any lost profits would amount to the opportunity cost or time value of money, which is not to be considered under the Guidelines. However, in Lowder, investors were told they would earn a guaranteed 12 percent rate of return and had been issued statements to that effect. Lowder, 5 F.3d at 471. Thus, the court considered this to be a contractual obligation that was fraudulent.

See, for example, Chicago Board of Realtors Inc. v. City of Chicago, 819 F.2d 732, 741 (7th Cir. 1987) (Posner, J, concurring).

See United States v Johannsen, 36 F. Supp. 2d 1135, 1136 (SD Iowa 1999) (interest may also be mandatory in some cases).

United States Sentencing Commission, ‘Staff Working Paper: Loss Issues’ (May 2000), at p. 10.