Diagnosing Institutional Corruption

by D. James Greiner

According to petwave.com, “Dogs infested with sarcoptic mites will present to the veterinarian with a history of the sudden onset of intense itchiness, and probably also with red, raw skin sores and thick crusted areas caused by self-trauma from the dog’s effort to alleviate the itchiness.”[1]

The problem is, of course, that lots of things can cause dogs to scratch too much, including allergies, hormonal imbalances, fleas, ticks, and perhaps most intriguingly, boredom.[2]  So how would a veterinarian know that sarcoptic mites are the problem, meaning that Fido needs a topical cream to kill mites as opposed to, say, an extra walk every day or a new tennis ball?  There are apparently a few intriguing[3] or disgusting options,[4] in addition to the more straightforward idea of examining skin scrapings from the dog under a microscope to see if mites are visible.  But in practice, vets actually use few of these methods to detect sarcoptic mites.  Instead, vets diagnose the affliction by instructing the dog’s owner to apply a topical cream that kills mites, then observing if the dog’s scratching problem resolves in the next two weeks or so.  If it does, the vet concludes that mites were at issue.  If it does not, then the vet searches for another cause.

I confess that when I reviewed the materials for Professor Christopher Roberson’s recent presentation to the Lab on “Blinding as a Solution to Institutional Corruption,” I did not immediately think of the mange.  Nevertheless, the point of this blog post is that those of us who care about institutional corruption might learn something from vets, and that Professor Robertson’s lecture shows us how this is so.

Stepping back for a moment, how do we know whether an institution is corrupt?  Begin by returning to the definition of institutional corruption that Professor Lessig offered at the Lab’s creation, namely, an economy of influence that weakens the ability of the institution to further the institution’s purpose.[5]  A classic form of corruption occurs when an institution created to serve one purpose alters its procedures or its output so as to serve, at least partially, a second purpose; the second purpose might be money or funding.  So, in an example that Professor Robertson uses in one of his papers, the institutions of biostatistics and pharmacological investigation are in part designed to figure out whether medical treatments, particularly drugs, improve health.  But under the current system, pharma companies fund such research, and they want studies of drug effectiveness to produce positive results.  As biostatistical and pharmacological researchers compete for funding, they may produce results that are more positive for pharma companies than the underlying science would support.

So, we (think we) know what institutional corruption is, or at least we have a rough idea.  Now, the hard part:  For any institution or set of institutions, how do we know whether corruption (so defined) is present?

I can think of three ways to diagnose institutional corruption; there are probably others.  The first is one that Professor Lessig employed in his 2009 inaugural lecture for the Lab: examine closely what an institution says or does (i.e., the institution’s actions or outputs).  If those actions or outputs are “incorrect,” then we might suspect that institutional corruption is to blame.  In the Lab’s inaugural lecture, Professor Lessig provided us with at least two governmental decisions, the level of added sugar in a balanced diet and the responses to global warming, as examples of “easy public policy question[s] that the government gets wrong.”

This method of diagnosis is unquestionably helpful, but it has drawbacks.  The primary drawback is that it assumes that we know what the correct answer is.  Thus, it tends to be helpful in settings in which the right output for an institution is, as Professor Lessig suggests, “easy.”  On the level of added sugar example, it is obvious that a balanced diet is not one in which 25% of caloric content comes from added sugar, and if the Food Nutrition Board says that 25% is fine, we might suspect institutional corruption.

When Fido scratches to the point of creating “red, raw skin sores and thick crusted areas,” Fido might have mites.

In many other settings, however, the right answer can be less clear or perhaps even unknowable.  Process theorists, for example, believe that good public policy cannot or should not be defined by measuring governmental outputs against a correctness scale, but rather by whether the processes that gave rise to those outputs are in some sense just.  One need not go this far to believe that looking only at an institution’s outputs could make it difficult to distinguish corruption from insufficient technical expertise, bad judgment, bad management, insufficient enforcement power, alternative theories of justice, or any other reason that an institution might not produce the outputs that an observer sniffing for corruption believes to be correct.  Further, why should we trust the observer’s sense of what is correct over the institution’s?  The observer itself may be responding to inappropriate (corrupting) incentives, or might itself suffer from any of the other pathologies mentioned above.  The point here is not that examining institutional outputs is unhelpful, but rather that this method of discerning whether corruption is present has limits.

A second method of diagnosing institutional corruption is to examine the institution’s innards closely.  Professor Lessig’s 2009 inaugural lecture for the Lab again provided us with an example of this method.  In the sugar example above, Professor Lessig documented how the ultimate decision by the Food Nutrition Board to adopt a definition of a balanced diet that included 25% added sugar came after an additional sugar industry insider became a member of the voting body.

If one looks closely at Fido’s skin samples, one might actually see sarcoptic mites.

This methodology, too, has limits.  Corrupting influences, like sarcoptic mites, can sometimes hide well.  Sometimes institutions disclose the fact that they are hiding things, citing reasons, mostly bad but some good, for keeping what they do secret.  As a Department of Justice litigator, I occasionally argued that a court could not and should not compel public disclosure of documents that would reveal an agency’s deliberative processes on the grounds that public disclosure of deliberations would chill the frank exchange needed for good decision making.  Such an argument might hold some water; one might ask, for example, whether an open and honest debate would result if Harvard Law School’s tenure deliberations were available online.  In other instances, institutions become good at hiding what they hide.  And secrecy can make it hard for an observer to see corruption at work.  Meanwhile, in other situations, information about potentially corrupting influences is hard to collect, either because those in the best position to collect it have are the alleged corruptors (would we expect big pharma to make the amount of money it spends on payments to and perquisites for doctors readily available?) or because the information’s disperse nature makes it hard to gather (would we expect that anyone would know how much money particular industries contribute to campaigns in all 50 states, including at the municipal level?).

A third method of diagnosis is what the opening paragraph of this post illustrated, diagnosis by treatment.  How do we find out if an institution is corrupt?  We imagine the nature of that corruption and specify an intervention that would combat it.  We then intervene in the specified manner.  If the institution changes in some way, say, by altering the nature of its outputs, then we might guess that the institution was corrupt (and that, as a bonus, we might be on to a way to counteract the effects of the corruption).  Note that the intervention could take several forms.  It might prevent the corruption from occurring in the first place, meaning the intervention gets at the root cause.  Or the intervention might address some element on the causal pathway from the corrupting influence to an institutional outputs, meaning that the intervention not eliminating the corruption but is rather rendering that corruption ineffective.

When Fido gets better after his owner rubs mite-killing salve on his skin, we conclude that Fido had mites.

Back to Professor Robinson’s recent presentation.  Several of Professor Robinson’s papers explore the concept of blinding as a solution to corrupting influences.  Litigation expert witnesses should receive assignments, produce reports, and communicate with clients through neutral intermediaries.[6]  Industry should conduct research on the effectiveness of medical treatments through the NIH.[7]

But, say the institutions at issue, we are not corrupt.  There is no problem to solve.  Yes, money flows, but everyone needs money to live, and the money does not influence our outputs.  Litigation expert witnesses insist that they are as pure as driven snow, and the litigators that hire them frequently contest the standard narrative that their experts are “whores”[8] who will say anything for money (the other side’s experts are a different matter, of course).  Statisticians who evaluate drug or medical device effectiveness contend that they follow the numbers, nothing more or nothing less (regardless of the fact that the numbers show that industry-funded studies reach results more favorable to industry than do independently funded studies).

It is not enough to give contentions of purity the back of one’s hand.  Everyone does need money to live.  And it is not easy to see how we will stretch tax dollars to fund expert testimony, pharma research, political campaigns (all the way down to the local level), and all of the other areas in which, say, money  ­might exercise a corrupting influence.  Diagnosis by treatment is an additional option here.  If decision makers of a particular type appear vulnerable to the corrupting influence of knowing the source of funds received (note that ordinarily it is the knowledge of who paid, rather than the payment itself, that can corrupt), then use the following method:  blind some such decision makers, leave others unblinded, and compare results.  (Note that one would do best to randomize the blind here, but that touches on a different set of issues regarding implementation.)  If the blind has no effect, then one might question whether knowledge of the source of payments is in fact corrupting this type of decision maker. 

Diagnosis by treatment has its drawbacks.  It is an alternative source of information, not a panacea.  In particular, intervening in the world and collecting results is time-consuming, complicated, and hard.  Who willingly participates in an experiment designed to see if she is corrupt?  At present, almost all of my research time consists of setting up, running, and analyzing the results of randomized control trials in the field.  The hardest part of field experiments is persuading people to engage in them.  And none of my RCTs are explicitly and overtly designed to diagnose institutional corruptions.  With corruption as a motivating factor in an RCT proposal, the persuasion necessary in any rigorous field operation becomes that much harder.

Yet my view is that those of us who care about institutional corruption are, at present, underutilizing the method of diagnosis by treatment.  Few diagnostic methods have a comparable potential to resolve arguments about the existence vel non of institutional corruption.  Field operations may be frustratingly slow and hard to do, but they are necessary if we are to prove when corruption exists in the face of protestations that we are hunting hobgoblins.

A suggestion to Professor Robinson:  Change the title of the presentation to “Blinding as a Diagnosis Method and as a Solution to Institutional Corruption.”

[3] Apparently, one can rub the edge of the dog’s ear while watching for a reflex scratching action in its hind legs.  http://www.petwave.com/Dogs/Dog-Health-Center/Skin-Disorders/Sarcoptic-M... It is hard to believe that this method would have high diagnosticity.

[4] It would be better, perhaps, not to inquire too closely regarding a technique that bears the moniker “fecal flotation.”  http://www.petwave.com/Dogs/Dog-Health-Center/Skin-Disorders/Sarcoptic-M....

[6] Christopher T. Robertson, 85 N.Y.U. L. Rev. 174 (2010).

[7] Christopher T. Robertson, The Money Blind, 37 Am. J. L. & Med. 358 (2011).

[8] J. Morgan Kousser, Are Expert Witnesses Whores? Reflections on Objectivity in Scholarship and Expert Witnessing, Pub. Hist., Winter 1984.