In the wake of the financial crisis, new regulations on bank capital and stress tests strengthened the ability of the major U.S. financial institutions to withstand shocks. The Trump administration’s plans to weaken capital requirements threaten to undo these gains. If they go forward, the financial system could again be dangerously vulnerable.
The right amount of capital is a question of insurance: how much output to sacrifice in return for reducing the expected loss from a crisis. Equity capital is the buffer that enables banks to absorb losses without being forced into bankruptcy and damaging the wider economy. But extra capital, as my examination of the evidence shows, comes at a cost, and involves an output penalty. The challenge is to get this balance right.
Over the past 40 years, the probability that a banking crisis would occur in an industrialized country has been 2.6% per year. The long-term damage if a banking crisis did occur amounted to two-thirds of one year’s gross domestic product. So each year the expected loss from a banking crisis was 1.7% of GDP.
I calculate the optimal amount of capital in the form of common equity to be between 12-14% of risk-weighted assets (which put a zero risk weight on government bonds, for example) and 7-8% of total assets. This amount of capital would reduce the annual probability of a banking crisis from 2.6% to between 0.2-0.6%, based on a 2010 Basel Committee survey of leading government and academic studies. My optimal range for capital is about one-third higher than the international “Basel III” requirement for “global systemically important banks,” or G-SIBs.
The eight G-SIBs in the United States increased their combined ratio of capital to risk-weighted assets from 8.1% in 2007 to 12.9% in 2016. Against total assets, their capital (or leverage) ratio has risen from 4.1% to 7.9%. So the large U.S. banks are now holding about the right amount of capital from the standpoint of the economy. This amount substantially exceeds the Basel III requirement — partly because the U.S. banks voluntarily maintain a cushion, but mainly because the stress tests mandated by the Dodd-Frank Act have been the binding constraint on their capital in recent years.
Not long ago the climate seemed to favor requiring even more capital. For example, Minneapolis Federal Reserve Bank President Neel Kashkari and prominent Stanford scholar Anat Admati agreed that capital requirements remained far too low. Advocates of far higher capital tend to cite the finance theory whereby requiring more high-cost equity and less low-cost debt will reduce risk and therefore the unit cost of equity by enough to leave the average cost of capital unchanged.
My research on banks shows this is wrong. The unit cost of equity does fall as the amount of equity goes up — but only about half as much as this theory predicts. As a result, higher requirements for equity capital do raise the average cost of capital for banks. As they pass along the additional cost to lending rates, corporations borrow less for investment and long-term production capacity of the economy declines. I calculate that each additional percentage point of total assets in required bank capital reduces long-term output by 0.15%.
As things stand then, the balance is roughly correct. Unfortunately, the political pendulum now seems to be swinging the other way — toward significantly lower capital requirements. The CEOs of JPMorgan Chase and Goldman Sachs say they're too high, and the Trump administration apparently agrees. In its June report with recommendations for the regulatory reform, the Treasury Department said the stress test hurdles should be lowered. It also suggested technical changes that would substantially reduce capital required for large banks.
Under present rules, by 2018 the large banks will be required to hold capital that meets a supplementary leverage ratio (SLR) of 5-6% of Basel III-defined exposure. That exposure adds about 40% to balance-sheet assets by incorporating off-balance-sheet exposure, including for derivatives. So the SLR would require equity capital of about 7% of balance-sheet assets — the bottom of the range I calculated as optimal.
The Treasury proposal would reduce the SLR capital requirement by removing deposits at the Federal Reserve, holdings of Treasury obligations and initial margin on derivatives at clearing houses from the exposure base. I calculate that at present levels of holdings, the eight G-SIBs would be able to meet this modified SLR while reducing their ratio of equity capital to total assets from 7.1% to 5.7%. With a new incentive to replace investment-grade corporate and other assets with assets in the exempted categories, the simple leverage ratio of capital to total assets could fall considerably further.
If it fell to, say, only 5%, the annual probability of a banking crisis would rise from 0.3% to 1.1% point in the base case, and to 2% in an adverse case. After taking account of annual savings from lower capital costs, the resulting net loss to the economy over a decade would range from $400 billion to $2.7 trillion, respectively.
Several regulators have said it would be dangerous to go back to weaker capital requirements for the large banks. They're right. Easing the rules could be a very costly error. Just look at the math.