The Value of Applying Control Fraud Research to Microfinance

By William K. Black

My blog urging readers to look at a column by David Roodman and an article by Milford Bateman and his colleagues about the microfinance meltdown in Bosnia prompted strong exchanges between David and Milford.

Their debates caused me to consider seriously the application of accounting control fraud to microfinance. I began to draft a letter to David and Milford, but it morphed into a substantive piece that I believe may be of more general interest to readers and of particular importance to the microfinance literature. Here is the hybrid letter/article.

David, Milford,

I recognized from the beginning that (as Bosnia has taught) trying to be a peace maker among disputants is a dubious mission. Nevertheless, I think Milford and his colleagues’ article and your piece in the Washington Post combine nicely to show the nature and importance of the overly sanguine views about microfinance’s success.

The application of control fraud theory to microfinance is an interesting concept because microfinance typically does not provide one of the key indicia of fraudulent intent that helps identify accounting control fraud by conventional lenders. Banks have understood for centuries that making large loans where the proceeds are quickly disbursed without adequate underwriting produces adverse selection and produces a negative expected value for the lender. It is a suicidal strategy for a bank, so honest banks do not engage in such practices. Banks engaged in accounting control fraud find making huge amounts of bad loans optimal. The “recipe” for optimizing fictional accounting income (and real losses) for a lender has four ingredients.

Grow like crazy
By making really crappy loans at a premium yield
While employing extreme leverage, and
Providing only trivial allowances against the inevitable massive losses

George Akerlof & Paul Romer, in their 1993 article (“Looting: the Economic Underworld of Bankruptcy for Profit”) famously described the fraud strategy as a “sure thing.” As a mathematician, you will recognize that the recipe is simple mathematics. It requires no particular skill (it is easy to make bad loans) and it does not require “risk” as we conventionally define risk in finance (hence the “sure thing”). The same formula maximizes fictional income and real CEO compensation under modern executive compensation schemes that are one of America’s most destructive exports. If you reexamine the formula you will also see why it maximizes real losses and the misallocation of capital and resources.

Because the accounting fraud recipe is simple to mimic, because frauds will tend to cluster in fields most conducive to fraud (weak regulation, minimal risk of prosecution, ease of entry, and assets that lack a readily verifiable market value are key factors in producing a superior criminogenic environment), and because the fraud recipe calls for extreme growth in lending even into the teeth of a glut (e.g., record commercial real estate (CRE) vacancy rates) epidemics of accounting control fraud are superb devices for hyper-inflating financial bubbles. Financial bubbles are themselves criminogenic for they make it simple to refinance bad loans and delay loss recognition.

For the sake of brevity, I will simply assert here that accounting control fraud drove the second phase of the savings and loan debacle (the first phase was interest rate risk), the Enron era frauds, and the current U.S. crisis. I believe there is strong evidence that it played a dominant role in Iceland and Ireland as well. I invite you to read the extensive materials I have prepared supporting my assertions.

The reasons we were able to convict over 1,000 defendants of felonies in “major” cases in the S&L debacle had a great deal to do with the ability of jurors to understand the importance and irrationality (for an honest lender) of underwriting actions taken by fraudulent lenders and the desirability of those actions for frauds. To implement the recipe an accounting control fraud must gut its underwriting standards and suborn its internal and external controls. (The art is to suborn, not defeat, controls – to turn them into valuable allies.) Banks, of course, are designed to have layers of underwriting and controls designed to prevent them from making large numbers of bad loans. That means that the CEO must transform the bank into an entity that primarily makes bad loans. That transformation is stark and it makes no sense for an honest lender. Jurors “get it.” For example, we would present evidence that lenders failed to conduct any appraisal on a (purportedly) $50 million office building or inflated the appraisal. Lenders would inflate the borrower’s income. Similarly, a lender’s loan file would not contain any evidence of a credit check on a borrower receiving a $50 million CRE loan. Our credit check would reveal that the borrower had filed for multiple bankruptcies. These anti-underwriting actions are only sensible for a bank engaged in accounting control fraud.

Microfinance, of course, typically involves minimal or nonexistent underwriting. The fascinating question for us as criminologists and financial regulators is how to identify, deter, and prosecute accounting control fraud in a microfinance environment that is spectacularly criminogenic. Microfinance theorists and practitioners, of course, have known from the beginning that the inability to underwrite poses one of the central challenges to microfinance. Some microfinance strategies seek partial substitutes for conventional underwriting. The three “c’s” in conventional lending are creditworthiness (the borrower has the capacity to repay the loan), collateral (the borrower has good title to a pledged asset with sufficient value to prevent the lender from suffering a loss even if the borrower defaults on the loan), and character (the borrower will repay the loan if he has the capacity to repay). Some micro finance schemes rely on trying to evaluate character through informal community evaluations and reinforce character through peer pressure on borrowers. Some schemes emphasize collateral and other pledges (co-signers on the debt).

Microfinance theorists, however, have overwhelmingly focused on the risk of borrowers defrauding the lenders. The far more dangerous risk is accounting control fraud by the persons controlling the lender. Milford and his colleagues deserve great credit for raising this risk. We can hope that their work will force all serious papers on microfinance and proposals for microfinance programs to consider explicitly that risk.

So what do we know about accounting control fraud by lenders before we consider the specifics of the Bosnian meltdown? First, we know that it causes by far the largest losses among conventional lenders in the U.S. Second, we know that it can hyper-inflate financial bubbles. Third, we know that it is a “sure thing” – it will make the persons controlling fraudulent lenders wealthy and it will do so quickly. Fourth, we know that the recipe for accounting control fraud requires a conventional bank to pervert its underwriting and controls in a manner no honest lender would allow. Fifth, we know that the accounting control fraud recipe for a lender is freely available to most microfinance lenders. There are no meaningful rules or regulatory barriers to following the recipe and there is no meaningful risk of the controlling officer being sanctioned for following the recipe. The “sure thing” is available to most microfinance leaders. If they follow the recipe they will become wealthy, and they will do so quickly. Sixth, we know that many microfinance lenders follow the recipe. (Look at the four ingredients again and compare them to microfinance lenders with which you are personally familiar. Do they grow rapidly by making loans to large number of uncreditworthy borrowers at a premium yield? Do they employ extreme leverage – remember that their reported capital numbers are often inflated, which understates leverage. Do they provide remotely adequate allowances for loan losses? We may vary in the descriptor we would use, but my hope is that we recognize that we are observing the same phenomenon. My descriptor is that the recipe is a “common” practice among microfinance lenders. I am interested in what descriptor you two would employ to describe the frequency.) Sixth, a single very large accounting control fraud can cause catastrophic losses. Seventh, we know that epidemics of accounting control fraud by lenders can occur and that they can hyper-inflate bubbles and drive financial crises. Such epidemics are weapons of mass financial destruction.

The implication of these points, unless your descriptor is that it is “rare” or “unheard of” for microfinance lenders to follow the recipe, is that it would be a miracle if accounting control fraud were not the bane of microfinance. Microfinance lenders have major presences in nations with exceptionally weak rule of law, high corruption, and tremendous ease of entry. Why would we believe that they would consistently pass up a “sure thing” of being made wealthy very quickly? What priors would lead us to such a prediction? The senior executives would have to be saints and they would have to be saints uniformly. If that is the prior that underlies microfinance policies and evaluations then it is time to state it explicitly in every major paper on microfinance rather than to implicitly assume it – the most dangerous practice because one does not even realize that one is assuming a “fact” that is exceptionally unlikely to be true. The next step would be to test the implicit assumption rather than take it on faith. The first rule of investigating elite frauds is that if you don’t look, you don’t find. That is overwhelmingly what we have done to date in microfinance. I see no evidence in the microfinance literature, prior to Milford and his colleagues’ piece, that the field has even read the criminology literature or Akerlof & Romer’s classic piece on looting. Akerlof is a Nobel Laureate in Economics (2001).

David, you demand proof of accounting control fraud by Bosnian microfinance lenders from Milford and his colleagues. Fair enough, but recognize that they were operating with no meaningful investigation by the government. How exactly are Milford and his colleagues supposed to prove the fraud? They drew the absolutely vital analytical conclusion prompted by reading the criminology and economics literature on accounting control fraud. They correctly observed that the Bosnian microfinance lenders operated in a fashion predicted by accounting control fraud theory. Here is what they wrote about the predictions.

III. Bosnia is a clear case of Blackian ‘control fraud’

Finally, we found in the model of ‘control fraud’ developed by William Black to explain the rise and fall of the US Savings and Loans Institutions (S&Ls) in the 1980s, a very useful explanatory framework with regard to the emerging structure, conduct and performance of Bosnia’s microfinance sector. First, Black’s model correctly predicted that senior managers in Bosnia’s MFIs would insist on an unsustainable growth strategy, but would nevertheless be able to convince all external parties (e.g., auditors, regulators, the media, the international development community) that it was a sound strategy. Second, Black’s model correctly predicted the methods variously used by senior managers to privately enrich themselves, such as their attachment to inordinately high salaries, regular bonuses, awarding of share options, use of interest free loans, and so on. Third, Black’s model correctly predicted that senior managers would eventually begin to convert the MFI’s assets (original grant funding plus retained profits) into their own private assets, such as when senior managers take a private stake in one of the MFI’s new investment projects, often using favourable loans obtained from their own MFI to do it. Fourth, Black’s model correctly predicted that senior managers would exhibit a visceral hatred towards, and attempt to frustrate, all Bosnian government regulators and any others who might try to properly regulate or monitor the microfinance industry. Fifth, Black’s model correctly predicted that senior managers would seek the psychological rewards associated with being the most profitable MFI, as well as adulation from elite opinion makers, including regular appearances as invited ‘experts’ at high-profile international donor events.

Milford and his colleagues were the first to draw these observations and deserve great credit for their insight. What are implications of their correct observation for microfinance theory and praxis? They are powerful and disturbing. First, they have exposed the implicit assumption that microfinance has operated under – that there is no meaningful accounting control fraud. That assumption will now have to be supported or abandoned. My prediction is that it will prove unsupportable.

Second, assume that what Milford and his colleagues have described in Bosnia is what we call in criminology a “perfect crime.” Assume that it cannot be effectively prosecuted, or even more radically, assume that it is lawful for microfinance lenders to follow the recipe. That would be the worst possible news for microfinance. If the CEO of a microfinance lender can quickly and lawfully (or with total impunity) become wealthy (“sure thing”) by following the fraud recipe then we have no defense against a recipe that leads recurrently to catastrophic losses to the lenders and the borrowers, the failure of the lenders, hyper-inflated bubbles, and financial crises. The harm is not limited to the economic sphere. The CEOs of microfinance lenders that follow the recipe are guaranteed to become wealthy. That translates to political power in most nations. The CEOs have a powerful incentive to use that political power to weaken the regulators and obstruct prosecutors.

Third, what we need is serious investigations of a sample of the Bosnian failures by fraud experts. If we respond to crises such as Bosnia in which there was no investigation by fraud experts by saying “I see no proof that it was fraud” without demanding an investigation we reinforce a system in which we systematically exclude even the possibility of accounting control fraud.

Ultimately, I hope that we (and others who write about and work in microfinance) can agree that the key is to find the facts. Milford and his colleagues have done us all a great service by exposing our implicit assumption in microfinance that accounting control fraud cannot occur. I doubt that anyone in the field is willing to make such an assumption explicitly and attempt to support it. Milford and his colleagues also deserve great credit for pointing out that many microfinance lenders follow the fraud recipe. Whether or not that is criminal; or criminal but not subject to effective prosecution, it is a triple “sure thing.” It will produce record reported (fictional) income in the near term, which will promptly make the CEO of the lender wealthy, and it will cause terrible losses to borrowers and lenders. Once we recognize this triple sure thing and the ability of most microfinance lenders to follow the recipe with impunity we must fundamentally reevaluate the risks of microfinance. The risks to the borrowers and the national economies of developing nations are far greater than the microfinance literature has previously recognized. As a field, microfinance failed to understand the severity and nature of the wakeup call that the Bosnia and Andhra Pradesh sounded. Microfinance can cause grave damage to millions of borrowers and the general economy. The logical leading candidate for explaining this danger and these meltdowns is accounting control fraud. A “leading candidate” is not proof that control fraud drove these crises. It needs to be investigated intensively by experts in accounting fraud to see if it can be falsified. That is simply the norm with regard to hypothesis testing. The microfinance field has not been conducting the minimum norm in hypothesis testing because the field has been unaware of the relevant criminology and economics literature. Milford and his colleagues have brought criminology and Akerlof & Romer’s classic article to the attention of the microfinance field. I hope that it will become a mark of unacceptable research in microfinance to ignore these highly relevant findings in other disciplines.

David, you are correct that we need to find the facts – something that we have consistently failed to do in microfinance. We have not investigated with fraud experts the causes of microfinance failures. We have not provided any facts to support the implicit assumption that accounting control fraud cannot exist in microfinance. What we can say at this juncture is that the known facts in Bosnia about how the microfinance lenders operated are consistent with an epidemic of accounting control fraud. That does not prove they were frauds. Testing to determine the incidence of accounting control fraud in Bosnian microfinance would require investigations by fraud experts of a sample of the failures.