On Friday, we posted a client alert to a new Federal Reserve study that, to put it succinctly, overturns received wisdom about what makes banks fail. It is a paradigm-busting analysis based on solid, validated, empirical evidence, not on the models notoriously replete with assumptions that suit a researcher’s fancy or whomever backs the work. This study’s main finding is that, even before the advent of federal deposit insurance, bank failure is due to and reliably predicted by growing bank insolvency – not illiquidity – at otherwise-solvent banks and generally not even by runs at very weak banks. Depositors and, worse, supervisors are demonstrably slow to catch on to emerging risk, with depositors understandably subject to information asymmetries and supervisors inexcusably distracted, confused, or even captive. Policy should not be based on one study, but this one study warrants immediate attention backed as it is by many others and replete with damning data analyzed with a straightforward methodology using records going back to 1986. Now would be a very good time to take heed – banking agencies in 2024 are building yet another regulatory edifice to compensate for yet another round of critical supervisory lapses. This may well prove as doomed as its predecessors unless regulators stop blaming banks after failure for bad behavior well within supervisory sight and reach long before indisposition turned terminal.
Importantly, I am not saying that this study proves there is no need for capital or liquidity regulation just as our new merger-policy study does not conclude that regulators should let any bank merge any way it wants. Banks have important – albeit no longer unique – access to public benefits such as deposit insurance and Fed backstops. Reminding them to take care with the taxpayer’s purse with clear guardrails is essential given the rampages wreaked all too often in free-wheeling markets by high-flying executives. There is always a risk that supervisors will catch on too late, be obstructed by internal politics or external litigation, or even do their best yet be overtaken by macroeconomic events wholly outside their control.
But, as well demonstrated by this careful study, the three causes of failure are all causes all the rules are designed to prevent: rising asset losses following booms in asset growth evident in high volumes of unrealized losses, increasing insolvency no matter what the capital rules may say, and increased reliance on expensive non-core funding no matter what the liquidity rules may hope. Deposit runs – seemingly the cause of failures up to and including last year’s—are found to be an implausible cause of most failures in the “history of the U.S.” and, even where a bank run kills off a bank, the bank is usually in imminent danger of failing all on its own.
The finding that simple balance-sheet fundamentals are highly predictive failure indicators is startling enough, but the paper’s findings about the importance of deposit insurance are stunning. Because it has data and, in some cases, examination records going back to 1865, the paper asserts it can determine what killed off banks before federal deposit insurance in 1934 was supposed to stop deadly runs. Failures in which banks are strong enough in receivership to repay depositors and creditors make up less than 0.5 percent of pre-1934 failures. That is, depositors were slow to react to growing bank fragility, fleeing from the teller window only when the bank was already doomed. Depositors may have precipitated failure, but supervisors could have stopped it.
Like most studies, this one has some data gaps and occasional leaps of inferential reasoning, but – unlike all too many studies – it is rigorous and unbound by prevailing wisdom at the regulatory agencies that employ several of the study’s authors. Neither the paper nor I say that prudential regulation is useless, resolution planning is a waste of time, or banks should merge wherever they may. What the study says tactfully and I say emphatically is that none of these standards—now or to come – will work if bank supervision is as hapless as it’s shown to be by the debris of each bank failure and systemic crisis.
I’ve outlined several key and easily-realized ways to improve bank supervision and there are lots of other sound recommendations to be had. Some of them are even in the autopsies provided by the FRB after SVB and the FDIC in the wake of Signature and First Republic. Are any of these being implemented?
We’ll never know because bank supervisors cling to opacity as tightly as those who could lose a few pounds wrap themselves in black to look as good as possible in the dimmest light they can find. None of the lessons about fundamental bank risk have yet to be drawn from the most recent failures in which all of these fundamental, terminal weaknesses are blindingly obvious. Maybe next time new rules will compensate for supervisory delays, but systemic crises are coming more frequently with ever-greater macroeconomic cost. Good supervision can’t stop all crises, but next time won’t be nearly as bad as the last times if supervisors did a lot better a lot sooner.