BankThink

Regulators' first step to avert bank failures is to close the information gap

AdobeStock_29952127.jpeg
Through the power of digital technology and artificial intelligence, regulators around the world could revolutionize the way banks' call reports are filed, writes Jo Ann Barefoot, CEO and co-founder of the Alliance for Innovative Regulation.

On the surface, the spate of bank failures that embroiled the financial sector last spring had little in common with the global financial crisis from a decade and a half ago.

The 2008-era meltdown stemmed from toxic mortgage debt on the balance sheets of globally entwined behemoths. In contrast, the Federal Deposit Insurance Corp. seized Silicon Valley Bank and Signature Bank in March and First Republic Bank in May — with even uninsured depositors being protected in all three failures — largely because they were ill-prepared for the Federal Reserve's rate hikes.

But look closer and both crises highlight a common problem: the regulators' lack of complete, real-time information about the condition of the banks they oversee. During the 2008 crisis, it was the lack of off-balance sheet reporting about systemic risk from derivatives and other exotic instruments. More recently, updated data about banks' interest rate risk, uninsured deposits and deposit outflows did not reach regulators fast enough for them to right the ship.

The epicenter of this problem is the quarterly call reports that FDIC-insured institutions submit to their regulators. These reports, which are expensive for banks to complete and contain thousands of data fields, typically don't make it to the regulators' hands until the very end of the quarter, or later. This means that supervisors are often reviewing information that could point to an institution's troubled condition four or five months after the fact.

This information gap is most evident at community banks. The regulators can collect more detailed and fresher data points about key indicators from the largest institutions between each quarterly submission. For thousands of smaller institutions, other than their regular exams, the call report is perhaps the most substantial glimpse that regulators get of their balance sheets — and that snapshot is often 120 days old.

It doesn't have to be this way. Through the power of digital technology and artificial intelligence, regulators around the world could revolutionize the call reporting process, and in some cases, they are already doing so. A rapidly developing market of regulatory technology (Regtech) providers offer data formatting and aggregation solutions that allow banks and regulators to exchange vast amounts of up-to-the-minute information.

Regulatory agencies in several countries have begun to establish Digital Regulatory Reporting (DRR) platforms, which can provide them with more up-to-the-minute information. Those efforts must be accelerated in the age of social media, AI and a rapidly innovating financial sector.

The release of two malicious language models — WormGPT and FraudGPT — demonstrate attackers' evolving capability to harness language models for criminal activities.

August 8
Data And Text On Computer Screens

If such mechanisms had been in place earlier this year, the Federal Reserve, FDIC and other agencies may have had better insight into the contagion risk that grew at the banks that ultimately failed. That risk evidently can escalate, literally overnight, with social media-driven bank runs and instant electronic withdrawal of funds. More information and timelier information could have given regulators more options to respond. Looking forward, a digitized reporting regime could be crucial in preventing the next crisis.

The failures this past spring also called into question whether regulators had sufficiently timely data about the institutions' uninsured deposits, which were high in all three of the banks that collapsed. Working over the weekend after closing SVB, they didn't know whether and how uninsured funds might be moving through the banking system as depositors sought institutions they perceived to be safe, especially those considered "too big to fail." If the agencies had declined to cover uninsured accounts, they might have triggered domino effects that they could not fully anticipate and address. Several U.S. banks recently revised their fourth-quarter 2022 reporting of uninsured deposits. In a sign that the FDIC was concerned about banks underreporting such amounts, the agency issued a "Financial Institution Letter" on July 24, clarifying expectations for accurately estimating uninsured funds.

The aim of DRR systems is to digitize financial information so that it can be reported with less cost and effort by the industry, and can be obtained by regulators — with necessary controls and limitations — at a moment's notice and in real time.

In the U.S., the FDIC began a project in June 2020 to modernize bank financial reporting, inviting 20 technology firms to participate in a "rapid prototyping competition" — similar to a tech sprint or hackathon — to design digital reporting solutions. Organizations such as the Financial Industry Regulatory Authority have also pursued the creation of DRR systems with machine-readable information about rules and regulations to track compliance through the reporting process. FINRA and the Securities and Exchange Commission have also used AI to monitor "big data" for signs of market abuses. Other entities developing DRR initiatives include the U.K.'s Financial Conduct Authority, Bank for International Settlements, the G-20, the Monetary Authority of Singapore (MAS) and the New York State Department of Financial Services.

Speaking in June at the Point Zero summit in Zurich, MAS Managing Director Ravi Menon likened regulators' current risk reports to driving a car with dials that show how fast you were driving in the past, but not your actual speed. Digital financial reporting regimes would give bank regulators the ability to respond to risks with the same speed and accuracy of a physician conducting an emergency room X-ray.

The emergence of generative AI is also creating a chance to simplify implementation of DRR, since newer systems can increasingly ingest unstructured data. Regulators should assess the feasibility of getting better information, without having to require financial companies to revamp how they manage their own systems.

With more innovative reporting tools, regulators would know about dire trouble at a financial institution before it hit the airwaves of social media and before depositors ran for the doors. In an age when digital technology can move money and propel risks at lightning speed, revamping the financial reporting process would allow regulators to be first on the scene with a solution to save an ailing bank from failure.

For reprint and licensing requests for this article, click here.
Technology Regulation and compliance Artificial intelligence
MORE FROM AMERICAN BANKER