Rebuilding Trust in Banking With Artificial Intelligence

In the Spring of 2008, Stephen Green, the then-Chairman of HSBC, penned his early reflections on the unfolding cataclysm later called the Financial Crisis.  “There has been a massive breakdown of trust:  trust in the financial system, trust in bankers, trust in business, trust in business leaders, trust in politicians, trust in the media, trust in the whole process of globalization – all have been severely damaged, in rich countries and in poor countries alike.”

Nearly a decade later, these words have a regrettable continued resonance.

Indeed, such views featured prominently in Senate Banking Committee hearings late last year regarding the opening of false accounts at Wells Fargo.  John Stumpf, the bank’s then-CEO, argued that such behavior was an aberration:  “That’s not our culture,” Stumpf said.  Senators took a different view, summed up by Committee Chairman Richard Shelby:  with over 5,000 employees involved in the opening of fraudulent accounts over a period of years, “there's something wrong with the culture and there's something wrong with the bank.”

A pervasive erosion of trust in our core institutions has been chronicled by pollsters such as Gallup, FiveThirtyEight, and Pew, among others.  This is of central interest to regulators charged with safeguarding the integrity of the industries they oversee, and particularly so for regulators of the banking and financial services sector, given its out-sized impact on whole economies.

The loss of public trust in banks comes, principally, as a consequence of widespread misconduct scandals.  It has been argued that such misconduct reflects a “toxic culture” polluting the financial services sector broadly – a culture that actually encourages dishonesty.  We believe this to be an overstatement:  most bankers are honorable and honest.  And, yet, arguments for a connection between culture and conduct are not to be dismissed.

The Financial Industry Regulatory Authority (FINRA) announced last year that it would begin to monitor “indicators” of firm culture in the course of its supervisory activity, “given the significant role culture plays in how a firm conducts its business.”  Other regulators, in the US and abroad, are engaged in, or contemplating, similar supervisory efforts.

But, regardless of such regulatory activity, it is clearly in the self-interest of banks and bank leaders to consider the nexus between culture and conduct closely.

Consider:  a recent study from BCG reports that banks, globally, have paid $321 billion in fines since 2008 for regulatory failings driven by misconduct; the Bank of England estimates that misconduct costs reduced UK banks’ pre-tax profits by an average of 40 percent between 2011 and 2015; and Bain estimates that governance, risk and compliance (GRC) costs now account for 15-20 percent of the “run the bank” cost base of most major banks.  It is this, we believe, more so than regulatory pressure, that will drive change.

Bank boards and shareholders have grown intolerant of such costs, and regulators are no longer prepared to view enormous GRC budgets as a proxy for efficacy.  The need to cut costs while still demonstrating a commitment to improved culture and conduct is, today, driving industry-wide adoption of “regtech,” defined by the Institute of International Finance, as “the use of new technologies to solve regulatory and compliance requirements more effectively and efficiently.”

Many regtech firms bring artificial intelligence to typically “manual” and staff-intensive GRC tasks, automating such processes and providing for real-time updates when thing go wrong.  Some even promise to forecast risk events, moving from merely descriptive analytics to predictive tools.  Several do so in the context of managing culture and conduct risk.

While approaches vary, regtech startups capitalize on the ever-increasing troves of data possessed by banks, and machine learning algorithms capable of distilling relevant patterns in such data as they evolve in real time.  Transaction data, for instance, can be mined for signals that indicate anomalous trading behavior.  And internal communications data provides a reliably rich source of insight into the relational dynamics among employees that shape risk and performance.

By bringing quantitative metrics to the qualitative challenge of human behavior, regtech firms can help banks to assess culture and conduct risk meaningfully, driving a reduction in costs through increased efficiency and effectiveness of risk management, and helping to avert the catastrophic fines and reputational damage that follow in the wake of misconduct scandals.  They may even help to restore trust in the financial services sector – something that is in all of our best interest.

Richard Ketchum is the former Chairman and CEO of the Financial Industry Regulatory Authority (FINRA) and an advisor to Starling Trust Sciences, an applied behavioral sciences technology company in the “regtech” space. 

Stephen Scott is the founder and CEO of Starling.