Making Finance Safe

Walter Wriston, Citicorp’s chief for nearly two decades until 1984, used to argue that banks’ didn’t need much, if any, capital. The global financial crisis put that view to rest. Today, we know that if banks are going to be able to absorb large unforeseen losses that would otherwise threaten financial stability, they need to finance themselves with equity, not just debt.

But how much capital do banks need to have to ensure the financial system is safe? Even after the financial crisis, answers to this question range widely, making it the single most contentious source of debate among bankers, regulators, and academics.

At one extreme, narrow banking advocates call for depository institutions to finance anything but riskless assets with 100% equity capital. Others, like Nobel Prize winner Eugene Fama, have called for equity capital of up to 50%. Admati and Hellwig, in their forceful book and other influential writings, argue that banks should operate with equity capital of 20% to 30% of their total assets, unadjusted for risk. The Basel Committee on Banking Supervision, which sets standards for internationally active banks, endorsed a sharp rise in capital requirements (Basel III) compared to abysmal pre-crisis standards (Basel II). Basel III requires capital of 8% to 10% of risk-weighted assets for the largest systemic banks. Using the tighter Basel III definition of capital, we estimate that the effective pre-crisis Basel II requirement was less than ¾% of risk-weighted assets (see table below)!


Comparing Basel III and Basel II Capital Requirements (Share of Risk-weighted Assets) for the Largest Systemic Banks: Impact of Basel III Capital Definition
Basel III range8% to 10%
Basel II Baseline4%
Adjustment for hybrid capital-2%
Adjustment for goodwill, intangibles, deferred tax assets, etc.-1%
Adjustment for changes in resk weights-0.25%
Effective Basel II converted to a Basel III basis< 0.75%
Source: Basel Committee on Banking Supervision (2010) and authors' calculations.


We cannot directly compare the Basel risk-weighted standard (with its reliance on a broad definition of capital) to a pure leverage requirement based on the ratio of common equity to unweighted total assets. However, many big non-U.S. banks applying the Basel III standards are likely to end up with a requirement that common equity be in the range of only 3% of total assets. Interestingly, the Federal Reserve already has indicated plans to hike capital requirements on systemic U.S. banks beyond the Basel III standard.

So, what is the correct answer? What share of their assets should banks be required to finance with equity or equity-like instruments? No one really knows for sure. That said, we share the view that existing standards are inadequate and should be raised significantly beyond current U.S. levels (and, therefore, even further above those in place for most European and Japanese banks). A large capital buffer is both the principal mechanism to limit systemic risk and the most effective response to the “too big to fail” problem.

At the same time, we have argued in a previous post that narrow banking (100% capital requirements) would not make the financial system safer: instead, it would shift risk-taking to nonbanks that, depending on their design, would be prone to runs and panics that trigger government intervention. We also argued previously that some measure of risk sensitivity – however imperfect – is needed if regulators are to prevent banks from concentrating in the riskiest assets that they are permitted to hold. [For large systemic banks, this means relying on models that are designed to compute the riskiness of asset portfolios, not the “Basel standardized approach” that sets risk weights on a fixed asset-by-asset basis.] In this respect, a leverage ratio alone is insufficient to offset the bad incentives created by the explicit and implicit government guarantees for big intermediaries.

To summarize our view thus far, we believe that the right level of capital is neither 0% nor 100%, but somewhere in between; and we support a risk-sensitive measure of assets. Saying more than that requires that we delve into the role that societies expect financial systems to play in their economies and the details of how those systems operate. This means balancing the costs of imposing higher capital requirements against the benefits of reducing the otherwise devastating economic losses associated with financial crises.

Nevertheless, we can do a bit of revealing analysis. Let’s start with what is one of the fundamental ideas of corporate finance – the Modigliani-Miller (MM) theorem. Subject to several key assumptions, MM elegantly showed more than 50 years ago that the value of a firm is independent of how it is financed. Put differently, how a firm divides its stream of future profits into payments to equity and payments to debt doesn’t alter the total value of those profits. Miller once made the analogy to cutting a pizza into four or eight slices: the size of the pie doesn’t change. That makes the firm indifferent between equity and debt finance. [For a parable that mirrors the classic MM “neutrality result,” see here.]

To the extent that the MM assumptions hold, raising bank capital requirements would be costless for everyone. So why, in practice, do banks view equity capital as so costly? Why do bank managers – like Walter Wriston – aim to minimize it? 

The reason is that some of the MM assumptions – which include no taxes or subsidies, no bankruptcy costs, and no principal-agent problems – are violated in practice. On this, everyone agrees. The key questions are precisely why and to what extent the MM assumptions are wrong.

Admati and Hellwig argue that observed MM violations arise almost exclusively from government-induced subsidies for bank debt – for example, through the differential tax treatment of debt and equity, and through the subsidy for those banks viewed as “too big to fail.” These factors can easily explain why banks view equity finance as more costly than debt finance. Yet, if these were the only drivers of MM deviations, then regulators could raise capital requirements without social cost. Higher capital requirements would simply offset market distortions created by other government practices. That view makes Admati and Hellwig comfortable with their simple remedy: much higher capital requirements.

What about banks’ “special” role in finance? Would higher capital requirements limit banks’ provision of key services? Perhaps, but we are as yet unaware of any effort to quantify these specific costs. Various theories distinguish banks from other firms by focusing on the synergies of deposit-taking with other banking activities. A prominent one treats banks as a device to buffer illiquid firms from liquidity-seeking investors: In this view, bank capital limits financial instability at the cost of reducing the supply of liquidity. In another model, banks are portrayed as efficient managers of liquidity risk – including the uncertainty about withdrawals (on the liability side of the balance sheet) and about takedowns of off-balance sheet loan commitments. Estimating the cost of capital using such banking theories requires calibrating complex dynamic models – work that remains in its infancy. (An example is here.)

Recent academic studies use rougher, less cumbersome, methods to estimate the impact of higher capital requirements on bank funding costs. In these analyses, researchers conclude that equity capital is modestly to moderately more costly for banks than debt. Estimates are in the range of a 3- to 9-basis-point increase in funding costs for each percentage point increase in capital requirements. (See the studies here and here.)

If this is correct, the cost of higher bank capital requirements is far from prohibitive. Even if banks were to pass on these costs fully to borrowers, their impact could be easily offset by monetary policy that is somewhat more accommodative on average. That is, the tightening of financial conditions implied by higher bank capital requirements could be neutralized by a lower target policy interest rate.

At the same time, these additions to bank funding costs could reduce financial safety. With funding so critical for competitiveness, increased costs may drive even more intermediation from banks to risky shadow banks; something regulators are having a difficult time addressing effectively (see, for example, this earlier post on U.S. money market mutual funds).

What to do? Being pragmatists, we think regulators should continue to ratchet up bank capital requirements until the tradeoff between banking efficiency and financial safety shifts appreciably in favor of the latter. Importantly, as capital levels rise, we will be able to measure the costs, in terms of increased lending spreads, reduced loan volumes, and shifts of activity to less-regulated intermediaries. Over time, these responses will give us the information we need to determine the desirable level of capital requirements. And, during this transition, the safety of the financial system will be on the rise. 

In the case of bank capital, it is surely better to be safe than sorry.