No bank should be paying dividends right now

In the midst of one of the fastest and deepest economic declines in U.S. history, the largest banks are expected to pay out about $14 billion in dividends in the third quarter, while remaining under a cap imposed by the Federal Reserve in response to the coronavirus pandemic.

This comes after the same banks paid out about $17.5 billion in the second quarter, according to an analysis using Bloomberg data.

While the Fed did somewhat restrict shareholder payouts in late June, it has already missed an opportunity to require banks to conserve more than $30 billion in capital that could have proved invaluable if any of its recent economic stress scenarios pan out. Both the economic scenarios and policy on dividend payouts were included in the results of the 2020 stress test of the 33 largest banks.

The results were not rosy. The Fed forecast $433 billion in loan losses through March 2022, under a scenario designed before the coronavirus lockdown. But in a rough sensitivity analysis that took the coronavirus into account, projected losses ballooned to $560 billion in a V-shaped economic recovery scenario; $680 billion in a W-shaped scenario; and $700 billion in a U-shaped scenario.

Those would be heavy losses compared to the banks’ aggregate $1.2 trillion of capital. In the worst case, a quarter of banks would be near or below their regulatory minimums, although the Fed did not say which banks or how close they would be.

The Fed admitted that its sensitivity analysis was not as robust as its typical stress test. So it took an unusual additional step: it is calling on banks to reassess their capital needs and resubmit capital plans later this year. Those plans could result in further dividend restrictions.

Still, amid the extraordinary uncertainty that the coronavirus lockdown has induced, suspending dividends would seem an easy way to conserve capital right now. This is exactly the kind of situation for which these forward-looking stress tests are designed, in order to prevent another collapse of the financial system.

The Fed’s rule normally prohibits banks from proposed capital distributions while it is reviewing capital plans. But the Fed hasn’t prohibited dividends this time.

Note that banks’ capital ratios were already elevated by an average of about 30 basis points in the first quarter because regulators allowed them to delay the capital impact of the new expected credit loss accounting framework.

Regulators reasonably justified that forbearance by the need to promote lending during a crisis. But allowing banks to deplete capital through dividend payments seems counter to that purpose. Regulators’ reluctance to suspend dividends may elicit some déjà vu.

During the last financial crisis, banks were still paying more than $10 billion in dividends per quarter through late 2008, and didn’t cut dividends by more than half until 2009. Regulators raised no objection.

Those same banks ultimately needed hundreds of billions of dollars in government capital through the Troubled Asset Relief Program. One paper showed that all the banks that received TARP capital paid at least 45% of the amount as dividends in 2007 to 2009. The combination of paying dividends and receiving government capital represented a massive transfer of wealth from taxpayers to shareholders.

The Fed did announce some modest measures this time. For example, a bank’s common stock dividends in the third quarter can’t exceed what it paid in the second quarter, or its average net income over the past four quarters. The Fed also banned share repurchases for the third quarter entirely.

But these limits are unlikely to significantly affect banks’ capital decisions. Net income is falling for most banks anyway, and the largest banks suspended repurchases voluntarily in March; share repurchases similarly fell sharply in the last financial crisis.

So why not suspend dividends immediately? The Fed’s peers in Europe quickly called for a temporary halt to bank dividends and share repurchases in this crisis, and later did the same for insurers and other financial companies.

The Fed’s vice chairman for supervision, Randal Quarles, argued that banks are much better capitalized this time. He also argued that banks performed well under the Fed’s rough sensitivity analysis. But that’s no reason to allow banks to deplete their capital, as Fed Gov. Lael Brainard pointed out in a rare public dissent.

Some argue the Fed has already gone too far, meddling in bank board decisions. But it’s the Fed’s job to set limits in a crisis. This is simple macroprudential policy — look out for the system as a whole, by providing cover to bank boards to conserve their capital in case things get worse.

Without that cover, bank boards hardly have autonomy right now. Supervisors may hope banks would reach the same decision through their own capital planning and governance, which have improved since the last crisis. But no bank wants to be the outlier while everyone else is paying shareholders. So, heading into an extraordinary period of uncertainty, most banks are going to maintain dividends.

However, any dividends paid may look frivolous if the recovery from the coronavirus turns more U-ish than V-ish.

Here’s a paper floated by the European Central Bank worth further study: a cyclical dividend prudential target set by regulators as guidance for banks. Such a policy could allow relatively high payouts in good times but automatically constrain payouts in downturns. Arguably, this would encourage banks to meet regulatory capital ratios during downturns by retaining earnings rather than shrinking assets.

Let’s consider ideas like this for next time.

More immediately however, the Fed should require banks to halt dividends until the coronavirus recession has passed, as other central banks have done. In the very worst case, Fed officials should not want to repeat history by using government funds to recapitalize banks, while justifying in retrospect why they didn’t conserve as much capital as they could when the crisis began.

Devyn Jeffereis provided research support for this article.