June 25, 2007
Adam Levitin on Friedman's Reinventing Consumer Protection
Posted by Christine Hurt

Law and economics has largely ignored (or spared, depending on one’s vantage) consumer protection qua consumer protection.  While there are numerous and excellent economic analyses of particular consumer protection issues, economic analyses of larger themes in consumer protection, such as consumer fraud, much less the overall concept of consumer protection, are largely absent from the literature.  David Adam Friedman’s paper on consumer fraud protection is an important step toward filling that void and opening a conversation about the way we allocate resources for and structure consumer protection. 

Friedman’s paper presents a keen analysis of the incentive structure of consumer fraud and proposes a method for reducing fraud through an innovative alteration of the incentive structure.  He argues cogently that our current consumer fraud protection regime is inadequate because it is diffuse and reactive and incapable of adapting rapidly enough to the ever-evolving world of fraud.  A wide array of federal and state agencies has consumer protection as part of their missions.  As a result, consumers often do not know what agency has jurisdiction over their complaint, and many complaints fall between agency jurisdictions.  Moreover, anti-fraud efforts are usually responding to the last consumer fraud problem, not anticipating the next type of fraud. 

The solution that Friedman prescribes is to leverage existing consumer fraud protection resources by focusing them on particular groups so as to make fraud a more risky venture, and thus reduce the incentive for scammers to engage in fraud.  Friedman proposes doing this by providing hyper-protection to two particular groups of consumers:  a vulnerable minority group and a randomly selected set of consumers.  First, by focusing anti-fraud resources (both education and enforcement) on a discrete, but vulnerable group, the “low-hanging fruit” will no longer be as accessible to scammers.  Therefore, scammers will be forced to attack more sophisticated targets that are more likely to detect the fraud and report it.  Thus, the overall risk of fraud will be raised, which Friedman believes will reduce the marginal fraud rate.  Moreover, to the extent that fraud becomes focused on more sophisticated consumers, they are better able to bear the economic loss. The discrete, but vulnerable group that Friedman identifies as most suitable for hyper-protection that can be leveraged to benefit all consumers is African-American males living below the poverty line because they are a sufficiently sizeable group, reasonably well-dispersed throughout the country, that is frequently targeted for fraud, yet historically with a low rate of reporting fraud. 

>p>Second, Friedman proposes adopting the concept of LoJack auto theft protection to consumer protection.  LoJack is an auto theft prevention system that is invisible to car thieves.  When a car is stolen LoJack allows the car to be tracked and enables prompt recovery of the car and frequent arrest of the thieves.  As Ian Ayres and Steven Levitt have shown, When LoJack is introduced into a metropolitan area, it has the affect of reducing auto theft not just of LoJack equipped cars, but of all cars because auto thieves never know if they are dealing with a LoJack equipped car.  Because LoJack is invisible, it raises risk in a way that cannot be managed by careful target selection.  LoJack adds an unavoidable risk to engaging in auto theft, and all car owners get a free ride. 

Friedman proposes creating a random LoJacked group consumers who will be given special anti-fraud education and protection, such as a “complaint window” which will serve as a clearing house for all enforcement resources.  Because scammers cannot know who is in this invisible group of consumers, they never know if they are approaching a hardened target or a vulnerable one.  This uncertainty increases the risk of engaging in fraud and should, Friedman argues, reduce fraud across the board. 

Friedman’s proposals for leveraging focused protection into benefits for all consumers is intriguing and has much to commend it as a deterrence tactic.  However, it doesn’t really address the two central flaws he identifies in the current consumer protection system:  diffusion of consumer protection responsibility among many agencies and reactive enforcement. 

If diffusion of responsibility, which causes both consumer confusion as to where to file complaints and jurisdictional lapses among agencies, is the key problem, Friedman’s solution of leveraging hyper-protection of discrete populations seems too narrow.  Instead of making a clearing house complaint window just for hyper-protected consumers, why not create such a complaint window for all consumers, like the 311 system that many cities have instituted as the interface for all communications other than 911 emergencies.  Or, for that matter, why not consider creating a unified federal consumer protection agency?  If the problem with consumer protection is an organizational one, it seems that the solution should be an organizational one, rather than one of resource allocation. 

Friedman’s proposal is able to get around the reactive enforcement problem because it is a generic solution that does not depend on the particulars of any fraud; it is a general deterrent.  Beyond whatever deterrent effect Friedman’s proposal would have, however, consumer protection would still remain a reactive proposition; even educational activities aim at teaching consumers to identify known types of fraud.  This raises the question of whether there are generic solutions to consumer fraud. 

I am dubious of any one-size-fits-all approach to consumer fraud.  There is huge variation in scams and scammers and many (although hardly all) are industry specific. The scammer community is not ecumenical.  Scams are often industry- or community-specific.  Immigration scammers often prey upon their own national/ethnic group; other con artists specialize in bilking the elderly.  Crooked auto dealers know tricks in their trade, but not in others.  Because so many things fall under the rubric of consumer protection (or even consumer fraud), it is hard to make generalizations about the field.  While there are large, crosscutting themes like disclosure, it is hard to talk about disclosure without delving into situational details.  We know from the behavioral psychology literature about the general limitations of disclosures, but the ultimate value of disclosures depends on the underlying transaction.  Thus, Friedman’s proposal might have widely varied deterrence effects depending on the particular scam. 

Although Friedman’s proposal is not really a fix for the problems he has identified, it potentially has an important general deterrence effect on fraud.  I think Friedman actually has the makings of two papers—one focusing on adopting LoJack to fraud prevention and the other on the structural problems of consumer protection. 

Most of the paper at hand, however, and hence the remainder of my comments, is about the LoJack proposal.  Like many big ideas, Friedman’s proposal might have trouble in implementation.  The distributional and political concerns involved with extending hyper-protection to discrete population groups are truly substantial, and the details of how a “complaint window” system would actually function not insignificant.  Friedman acknowledges these concerns, but the success of his idea will depend heavily on its implementation. 

Bracketing the political and technical obstacles to implementing Friedman’s proposal, I still hesitate to fully endorse it for a few reasons.  Like many law and economics analyses of how regulatory action can affect behavior by altering incentives, Friedman’s argument is based on several assumptions that are open to questioning.  First, does it properly assume that the group whose behavior it seeks to alter (here, swindlers) is rational?  Second, does it properly assume that the target group has sufficient information to decide to alter its behavior?  Third, can it adequately predict the result of altered incentives on the rational target group (here, whether the deterrence curve is known)? And fourth, does it properly assume that its solution (here, shifted resource allocation) is the efficient one?  I will address these in turn. 

1) Rationality of the Fraud Perpetrator

Unfortunately, this is where I found the paper the least convincing.  Friedman cites the work of James Q. Wilson and Alan Abrahamse for the proposition that criminal are “temperamentally disposed to overvalue the benefits of crime and to undervalue its costs.”  Criminals aren’t so different than the rest of us—they have their own hyperbolic discounting problems, it would seem. 

Friedman argues that fraudsters are the exception to this rule.  His argument lacks support, however, including from Wilson and Abrahamse, other than his own assertion that fraud is different because it is not impulsive, but often cunningly designed to sit just on the edge of legality.  The mere fact that fraud is often a crime of craft, not Kraft, does not mean that fraudsters are rationale.  Indeed, it seems reasonable to posit that many fraudsters are not rationale precisely because they think they are so clever.  Literature and life are filled with characters blinded by their self-confidence from the Sicilian genius Vizzini in The Princess Bride to Grand Moff Tarkin in Star Wars to, as some might argue, Donald Rumsfeld.  Friedman’s paper would benefit greatly from a presentation of stronger evidence that fraud perpetrators are fundamentally rational and remain so throughout their careers. 

2) Epistemic Constraints on Rational Actors

Even if scammers are rational (and somehow different than people as a whole in terms of hyperbolic discounting), they simply lack accurate information about risk, so it is unclear what, if any, effect Friedman’s proposal would have, as criminals might not know of the increased risk, or, even if they did, they might not know how to evaluate the absolute risk level.  Criminals do not accurately know the probability of their fraud being detected, much less the potential consequences.  While Friedman’s proposal might raise the probability of fraud detection, it is would only raise it from X to X+Y.  If the scammer does not know the value of X, then unless Y is very large, the increase probability of detection might not have much meaning to the scammer.

And even if the scammer thinks detection is more likely, the journey from detection to adverse consequences is a long one.  Even if a criminal knows the maximum sentence possible once all enhancements are included, there is still the great unknown variable of prosecutorial discretion.  A rational criminal cannot determine ex-ante the likely consequences of being caught for consumer fraud, except, perhaps, in the most egregious cases.  Even though Friedman’s proposal would alter actual risk, its effectiveness depends on whether it affects perceived risk.  Absent a change in perceived risk, there will be no change in criminal behavior. 

Because of the importance of risk perception, we might ask how criminals get information about the risks of engaging in crime.  For some crimes, accurate enforcement data is available.  Thus, tax cheats can benefit from public information about IRS audit rates, although I doubt many consciously calculate their odds based on such figures.  Most crooks, I imagine, evaluate risk based on anecdotal information, so depending on their connections with other scofflaws, their risk perceptions will change. 

This suggests an important distinction between auto thieves and consumer fraudsters.  Auto theft is an organized industry (joyriding punks aside) that involves chains of criminals, from the actual thieves to chop shops to resellers.  The chop shops and resellers act as hubs for the thieves.  Because of this network of relationships, the arrest of one thief is likely to be widely known within a community.  Consumer fraudsters, in contrast, are more likely to operate independently.  Consumer fraud lacks informational hubs.  Accordingly, consumer fraudsters are less likely to be aware of enforcement trends than auto thieves, which might limit the effectiveness of a LoJack consumer protection system.  Even if consumer fraudsters are rational, they might be making rational decisions on inaccurate perceptions of risk derived from incomplete or distorted information and therefore systemically undervaluing (or overvaluing) the risks of engaging in fraud.   

3) Shape of the Deterrence Curve

While Friedman emphasizes that increases in the probability of arrest, conviction, and punishment and increases in the severity of punishment have a significant deterrent effect on crime in general and (because of their supposed hyper-rationality) on fraudsters in particular, the absolute levels might be just as important as the relative levels. Does increasing the risks of engaging in fraud actually deter fraud?  Might it depend on the amount by which we increase the risks? Is there an absolute risk threshold that must be achieved before we will see a drop-off in fraud?  At what point will crime no longer pay? 

Without knowing the slope of the deterrence curve (the law enforcement to crime level relationship), it is hard to know what incentive structure will work.  Does law enforcement relate to crime levels as a straight line or exponentially or otherwise?   Without knowing something about the deterrence curve, any proposal for changes in enforcement is a wild guess that might or might not be efficient.  While something like a 1:1 relationship is useful for basic modeling (and hard to test in the case of fraud), it is an assumption that should be relaxed.

4) Assorted Efficiency Issues

There are several efficiency arguments the paper could better address, but I was glad that paper avoided the fundamental, if unanswerable law and economics question:  what is the efficient level of consumer fraud?  (And as a corollary, might the proposal result in overdeterrence, especially of questionable, if legal activities?)  Friedman assumes that fraud levels are currently higher than they should be; if he attempted to engage with the sophistry about the efficient level of consumer fraud, his paper would never get off the ground.  It is simply impossible to identify such a level outside of a very simple model or without resorting to the tautology of assuming that the market is efficient so we must have the efficient level of fraud.  Nonetheless, I note this issue precisely because it will inevitably be raised at presentations.

An efficiency query that I would like to see the paper address is the problem of identifying the least cost avoider of consumer fraud.  Friedman’s model of consumer fraud is a binary relationship between the fraudster and the consumer.  But often the relationship is more complicated, involving a legitimate intermediary.  Such an intermediary can range from a merchant to an ISP to an auction website to financial institution.  Might there be cases in which these intermediaries are themselves the least cost avoider? 

For example, consider a common advanced fee fraud, the 419 scam (so called after a section of the Nigerian penal code), in which the target is contacted, typically by e-mail, with a plea for help transferring funds out of a third world country, in exchange for which the target is promised a cut.  The target, however, has to lay out his own funds in order to advance the requested transfer, which will never occur, of course.  A good ISP spam filter should be able to catch many 419 scam solicitations before they ever reach consumers.  Might the ISP be the least cost avoider?  I don’t know.  Most likely the least cost avoider of any fraud involving a legitimate intermediary depends on the details of the fraud, but the 419 scam example illustrates the need to explore the least cost avoider principle outside of a direct binary relationship. 

The paper would also improve by explaining why a regulatory solution is necessary to consumer fraud (although this could be a paper unto itself).  Why not a market solution?  The market does not seem particularly responsive to consumer fraud, either through deterrence mechanisms or through insurance products.  Why is this the case?  Why haven’t we seen the market develop a consumer fraud LoJack device?  Friedman’s paper could benefit from probing this question and considering what makes consumer fraud different than auto theft.

The paper would also benefit from a comparison of consumer fraud protection systems to another anti-fraud regime—IRS auditing.  Friedman’s proposal to focus consumer fraud protection resources on vulnerable groups and on an invisible, random group closely resembles IRS auditing strategy.  The IRS focuses its audit resources on two groups of taxpayers:  those who it thinks are likely to engage in tax fraud and on a random group of taxpayers.  The randomness of IRS audits means that a taxpayer cannot structure around the risk of engaging in tax fraud. When audit rates go up, does tax fraud decrease?  By fleshing out this parallel, Friedman should be able to go a long way in answering some of the efficiency questions raised by his proposal.

Finally, isn’t consumer fraud protection currently random?  A fraudster never really knows when a consumer is likely to detect and report a scam or if any enforcement action will be taken.  Isn’t Friedman really calling for vigorous enforcement for consumer protection laws, rather than a reallocation of enforcement resources?  If there was hyper-enforcement of the complaints currently filed, we might see similar results to those Friedman predicts.  What’s more, it might encourage consumers to actually file complaints when they encounter fraud because they will see it has an effect.   

My analytical reservations aside, Friedman has raised an important question of whether we are efficiently using our current allocation of anti-fraud resources and whether our current anti-fraud regime is a sensible design.  Fraud is only one concern of consumer protection, but Friedman’s paper is an invitation for a much-needed debate about the larger structure of our consumer protection system. 

Junior Scholars | Bookmark

TrackBacks (0)

TrackBack URL for this entry:

Links to weblogs that reference Adam Levitin on Friedman's Reinventing Consumer Protection:

Recent Comments
Popular Threads
Search The Glom
The Glom on Twitter
Archives by Topic
Archives by Date
January 2019
Sun Mon Tue Wed Thu Fri Sat
    1 2 3 4 5
6 7 8 9 10 11 12
13 14 15 16 17 18 19
20 21 22 23 24 25 26
27 28 29 30 31    
Miscellaneous Links