A Dearth of Data in the De-Risking Debate

As readers of this blog are likely well aware, the fight against grand corruption is closely linked to the fight against money laundering. After all, kleptocrats and others involved in grand corruption need to hide the origins of their ill-gotten wealth. While the criminals who seek to launder their illicit cash are sometimes prosecuted for money laundering, much of the burden of the anti-money laundering (AML) regime falls on banks and other financial institutions. These institutions have obligations to perform due diligence on prospective clients—especially those clients with attributes suggesting high risk—and to report suspicious transaction to the government. Financial institutions can be held liable for failing to fulfill these obligations, and in some cases for their complicity in money laundering schemes. Yet many advocates believe that the current AML framework is not stringent enough, and have called for reforms that would impose additional obligations, and potential liabilities, on the financial institutions that handle clients and transactions that pose a high money laundering risk.

Banks and other skeptics often resist these reforms, arguing not only that the various proposals will do little to reduce money laundering, but also that more stringent AML regulations will lead to a phenomenon known as “de-risking.” This piece of industry jargon refers to the practice of ending or avoiding relationships with individuals or businesses perceived as “high risk” for money laundering. Of course, we want banks to eschew an individual client or transaction with characteristics that suggest a high probability of money laundering. But when banks and others warn about de-risking, they are referring to a phenomenon in which banks refuse to do business with broad categories of clients – for instance, those from particular countries or regions, or in specific lines of business – despite the fact that most of the individuals or firms in that category do not actually present a serious money laundering risk. If the monitoring costs and legal risks associated with certain kinds of accounts are too high relative to the value of those accounts, the argument goes, it’s easier for banks to simply close all of the accounts in the “de-risked” category. But this indiscriminate closure of allegedly risky accounts cuts off many deserving people, firms, and organizations from much-needed financial services.

Is de-risking really a significant problem? Skeptics might observe that the financial industry has incentives to resist more stringent AML regulation, and their warnings of de-risking may be, if not deliberately pretextual, then at least self-serving. That said, other actors, including non-profit groups, have alleged that they have experienced account closures due to de-risking. So the concern is likely a real one. Still, to set rational AML policy, we would want to know not just whether de-risking is a potential problem (it is) or whether it occurs sometimes (it probably does); we would want to know whether it is a systematic and serious problem, one that would likely be exacerbated by a significant enhancement of banks’ AML obligations.

So, what do we know about the extent and magnitude of de-risking in response to AML regulations? The short answer is: not much.

Continue reading

Small Town Corruption: The Cautionary Tale of Jasiel Correia

Elected at the age of 23 to serve as mayor of Fall River, Massachusetts, Jasiel Correia looked like a wunderkind. A tech entrepreneur who founded his own startup, Correia was the youngest-ever mayor of his hometown, the golden boy who promised to use his technological prowess and puckish energy to bring his aging town into the 21st century

Then it all came crashing down. In 2018, Correia was charged with various personal misdeeds, including tax and wire fraud, related to his tech company. A defiant Correia maintained his innocence and rejected calls for his resignation. Then, a second round of charges hit, this time alleging public corruption. Correia purportedly took over $600,000 in bribes from marijuana business license applicants—including one marijuana business owner who paid the Mayor $100,000 and promised him 2% of his future sales revenue in exchange for a lucrative operating permit. By the time Mayor Correia went to trial, he faced 24 separate criminal charges, and on May 14, 2021, the jury found him guilty of 21 of those 24 counts.

Mayor Correia’s downfall might seem like a relatively minor matter involving local corruption in one small city. (Such stories are, alas, all too common.) But this incident usefully highlights the corruption risks associated with devolving regulatory authority to local governments. While there are certainly virtues of giving local governments power over local affairs, we need to be clear-eyed about the dangers that local control can pose, particularly in the context of regulating lucrative industries like legal marijuana. The Fall River example highlights several such risks:

Continue reading

ML for AML: Is Artificial Intelligence Up to the Task of Anti-Money Laundering Compliance?

Fighting corruption—especially grand corruption—requires effective anti-money laundering (AML) systems capable of efficiently and correctly flagging suspicious transactions. The financial institutions responsible for identifying and reporting suspicious transactions employ automated systems that identify transactions that involve certain red flags—characteristics like transaction amount, location, or deviation from a customer’s typical activity; when the automated system flags a transaction, this triggers further review. But—given the ever-increasing volume and complexity of financial transactions that occur each day, as well as the increasing sophistication of kleptocrats, criminal groups, and others in disguising their illicit activities to avoid the usual red flags—picking out the genuinely suspicious transactions can be extraordinarily difficult. Even the cleverest compliance system designer couldn’t hope to incorporate every potential red flag into the automated system.

The need to stay one step ahead of the bad actors has fueled greater interest in how new advances in data processing technology may help make automated suspicious transaction detection systems more effective. Techno-enthusiasts are particularly interested in deploying deep learning artificial intelligence (AI), as well as classic algorithms that fall under the machine learning (ML) umbrella, in the AML context. ML and AI systems extract patterns from training datasets, and “learn” (by induction) what data patterns are associated with particular identifiable categorizations. Email spam filters provide a simple example. A spam filter, which can be created to conduct a process known as classification, sorts input variables into two categories: “spam” and “not spam.” It makes its categorization based on individual characteristics of the emails (such as the sender, body text, etc.). In the AML context, the idea would be to train an algorithm with data on financial transactions, so that the system “learns” to identify suspicious transactions even in cases that might lack the usual red flags that a human designer would program into an automated system. Advocates hope that ML/AI systems could be used both to filter out the false positives (transactions which are flagged as suspicious but turn out, on review, not to raise any concerns—an estimated 99% of all flagged transactions), while also identifying unusual, potentially fraudulent behavior that may be overlooked by human regulators (false negatives). Indeed, industry experts are understandably enthusiastic about AI systems that will cut costs while improving accuracy, and proponents claim that “AI holds the keys to a more efficient and transparent AML stance[,]” urging that “[b]anks must take hold of this new [AML] weapon[.]”

To the extent that AI tools can improve upon the admittedly-clunky automated systems currently in use, it could be a step forward. But ML/AI systems have a less than stellar track record in other contexts, and a model targeted at AML compliance presents some unique challenges.

Continue reading