Banks have magic powers: they can conjure money out of thin air, send it across the street or around the world instantly, make your startup dreams come true, and these days, even make you a cappuccino. They are also fragile and toxic: liable to fail suddenly and bring people, firms, other banks, and entire economies down with them. Generations of reformers have tried to make bank failure more orderly, less destructive, and less costly to the public. Judging by the crop of papers inspired by the crop of bank failures in 2023, they have failed again.
Why do we keep failing at bank failure? This article by Jeremy Kress suggests a piece of the puzzle: we mismeasure success.
The U.S. regime for dealing with bank failure promotes a single-minded focus on minimizing costs to the industry-funded federal Deposit Insurance Fund (DIF) managed by the Federal Deposit Insurance Corporation (FDIC). Kress makes a compelling case that protecting the DIF increases the overall social cost of banking and bank failure: it contributes to systemic risk, reduces competition, and distorts incentives. His alternative to the prevailing “least-cost” test is a more holistic social cost assessment, which would require more rigorous oversight of the FDIC, and would be a hard sell in any political climate—but worth the fight in the long run.
To appreciate the broader implications of Kress’s contribution, consider the bank failure process and the FDIC’s role in it.
The FDIC wears many hats: it insures bank deposits up to a generous $250k,1 regulates and supervises insured banks, and when one fails, the agency serves as receiver to manage its resolution. The FDIC as receiver decides how to dispose of the assets and liabilities that make up the bank receivership. It could sell the bank, wholesale or piecemeal, or liquidate its assets and pay off the liabilities. Unless the FDIC as receiver finds a healthy bank to assume the failed bank’s insured deposits immediately, the FDIC as insurer pays insured depositors out of the DIF, and takes their place in line for the receivership proceeds.2
The DIF is the principal pot of money available “to carry out [the FDIC’s] insurance purposes,” meeting its obligations as insurer and receiver. It is funded primarily from regular assessments paid by FDIC member banks. If the FDIC dips into the DIF to resolve a bank, it replenishes the fund with a special assessment on some or all of the remaining member banks. If the DIF runs out of money (as has happened twice since its establishment in 1935), the FDIC has standing authority to borrow from the U.S. Treasury and other sources, backed by the full faith and credit of the United States. This should be comforting to the public, but as Kress and others have pointed out, the FDIC has shunned this authority. Here Kress’s familiarity with the ecosystem pays off: from the agency’s perspective, tapping the Treasury reads as a taxpayer bailout, signals supervisory and actuarial failure, and counts towards the federal public debt limit, all of which invites unwelcome scrutiny.
All else equal, unloading a failed bank as a whole to a single buyer is faster and easier for the FDIC than sifting through, managing, and marketing bank bits and pieces. This is especially so when multiple banks fail at the same time, as they did in the Savings and Loan crisis of the 1980s and 1990s, the financial crisis of 2008-2009, and the would-be crisis of 2023.
Selling the whole bank rescues all its creditors—not just insured depositors—thereby avoiding tweets from furious financiers, front-page stories of shopkeepers struggling to meet payroll, and quite possibly a deeper crisis. It comes at the cost of entrenching expectations of future rescues, more industry concentration, and the associated distortions. Meanwhile, whole-bank sales often call on the FDIC to share in the risk of loss with the buyer. Empirical studies show that whole-bank sales tend to be costly for the DIF.3
In 1991, Congress introduced the “least-cost” test to limit the FDIC’s discretion to use whole-bank sales and rescue uninsured creditors. At least in theory, the test should lead the FDIC to choose piecemeal liquidation if it would cost the DIF less than selling the entire bank franchise. For another example, if the failed bank’s closest competitor or a global conglomerate bids a smidgeon more than a community bank two towns over, the FDIC would have to take the competitor’s bid.
What is not to like about this cost-saving approach? Recent scholarship suggests plenty. Kress highlights dramatic cost estimate fluctuations, which make it hard to police compliance with the least-cost test. A valuable new study by Michael Ohlrogge suggests that the FDIC’s post-crisis resolution practice does not come close to fulfilling the statute’s cost-saving mandate. Ohlrogge proposes to reinvigorate the 1991 test by solving the underlying agency and time inconsistency problems.
Kress builds on these and other scholars’ insights to reach the opposite conclusion: the least-cost test must go because it targets the wrong costs. Prioritizing costs to the industry-funded DIF does not account for the benefits of competition, reducing systemic risk, and access to financial services, among others. There is plenty of evidence that consolidation and systemic risk have grown dramatically since 1991. Kress’s case studies illustrate how the least-cost test may help exacerbate these trends.
His intuition feels right to me because it reflects the political economy of banking. Deposit insurance should deter runs and protect small(ish) depositors. Bank resolution should allocate losses and limit spillovers from bank failure. Both are fraught with distortions, and entail costs and benefits to the public that are hard to quantify as a snapshot in time.
Like bankruptcy, the bank resolution regime is a complex political bargain. Managing bank failure entails intensely political tradeoffs that must reflect the changing institutional structure of finance, align incentives, and deliver broadly legitimate distribution. There is no obvious reason to think that costs to the DIF in a given case of bank failure are a proxy for any of that, and many reasons to worry that a focus on these costs fosters policy myopia.
This is not an argument against any cost test. The author argues for a far more expansive replacement of the current test, which would call for far more robust oversight of the FDIC’s decisions—and would surely set off a political battle. At least we would be battling over the right things.
- The limit applies per person, per institution, per account category. An FDIC brochure shows how family of five could get $3,500,000 in deposits insured in 2024.
- FDIC as insurer is subrogated to the rights of the insured depositor against the failed bank up to the amount paid or deposit liability assumed.
- Michael Ohlrogge cites many of the studies and adds evidence of his own here.
