Making Banking Safe

Professor Mervyn King, our friend, NYU Stern colleague and the former Governor of the Bank of England, has written a wonderfully insightful and thought-provoking new book, The End of Alchemy. His goal is not just to explain the sources of the 2007-09 crisis, but to provide a template for financial reform that would reduce the frequency and severity of future crises. In the end, Professor King proposes a radical structural change intended to make banking safe while preserving the intermediation function that is critical to modern economies.

The alchemy to which Professor King refers in his book’s title is banks’ traditional function of transforming high-risk, illiquid and long-maturity assets into low-risk, liquid and short-term liabilities. But, in the presence of limited liability for the banks’ owners and the government safety net (in the form of deposit insurance and the lender of last resort that remove both solvency and liquidity risk for the depositors), banks’ incentive is to transform too much. Holding assets that are overly risky, insufficiently liquid and too long-term makes banks fragile and run-prone, providing fodder for systemic crises.

Through a combination of capital and liquidity requirements, financial regulation today seeks to reduce the riskiness of both individual institutions and the system as a whole. Capital serves as a buffer against losses banks face, reducing solvency risk, while sufficient liquid assets reassure depositors that they can withdraw their funds at will. Taking account of externalities, regulatory minimums are to be set higher than individual bank solvency risk would imply.

Unfortunately, this approach hasn’t worked very well. Professor King views the problem as arising from economists and authorities having the wrong mindset. His concern is “radical uncertainty” (known by economists as Knightian uncertainty) or what former Secretary of Defense Donald Rumsfeld referred to as “unknown unknowns.” In the presence of radical uncertainty, one cannot list all possible future outcomes and assign them probabilities. As a consequence, the standard tools of probability and statistics are insufficient to quantify risk, including the risk of a bank’s balance sheet. And, importantly, there is no way to ensure that the bank can make good on its promises to its depositors in all possible circumstances. After all, there are surely some states of the world where the bank collapses, even if we are not able to conceive of them.

In response, Professor King proposes a profound and imaginative reform of the financial system: any fixed-value liability (say, redeemable at par) with less than one-year’s maturity issued by any financial institution must be 100%-backed by collateral posted with the central bank. Because the collateral will face a (potentially substantial) haircut, its face value will exceed the level of deposits and any other fixed-value liabilities of less than one year. The haircuts, set in normal times, will change infrequently. King labels the central bank in this scheme as the pawnbroker for all seasons (PFAS). Critically, the PFAS can and will provide access to liquidity through its discount window lending facility in both good times and bad. Even in those unfortunate circumstances that people can’t envision today, the central bank will always be there to ensure the liquidity of the financial system’s short-term fixed-value liabilities.

Deceptively simple in its construction, we interpret this PFAS regime as a hybrid between our current financial system and one of narrow banking. Put differently, Professor King realizes that narrow banking is not all or nothing. To understand this insight, consider a simplified commercial bank that has three types of assets: reserves (deposits at the central bank), securities and loans; and two types of liabilities: demandable deposits and equity (a more complex version would add long-term bonds as a liability without altering the implications). In a narrow bank, reserves must be greater than or equal to the level of demandable deposits, so all risky assets must be backed by equity. Under current bank regulation, required reserves are typically far smaller than deposits. The PFAS scheme is somewhere in between these two. In it, the sum of reserves plus post-haircut loans and securities must exceed the level of demandable deposits. (For those of you who like equations, there is a simple algebraic version in the appendix to this post.)

This scheme is both breathtakingly simple and amazingly powerful. As Professor King points out, if adopted, the PFAS would eliminate the need for deposit insurance, as well as substitute for risk-based capital requirements and liquidity regulation. It also can eliminate runs on bank-like intermediaries with short-term fixed liabilities.

Replacing the current complex prudential regulatory apparatus with something simple is clearly a desirable objective. To list just a few reasons: risk-weighting is difficult and doesn’t work well (see here); the costs of regulatory compliance and enforcement are high; and deposit insurance fosters systemic risk-taking and is very difficult to price properly (see here).

Moreover, by establishing safety rules for bank-like (rather than just bank) intermediation, the PFAS approach could stop runs on shadow banks with fixed-value liabilities. Professor King’s proposal addresses this problem by imposing the same rules on anyone engaging in the same economic function. Critically, the collateral posting requirement applies to all financial intermediaries offering fixed-value liabilities that are redeemable in less than one year. This means, for example, that money market funds would have to finance themselves with equity equal to the haircut on their posted collateral. It also implies that insurance companies (or anyone else) offering lines of credit that can be drawn at will by their policyholders (or customers) would have to post collateral to back up contractual arrangements that are functionally equivalent to demand deposits. Put differently, anyone who offers to convert a liability at a fixed value into cash on short notice must obtain the backing of the central bank (see our earlier discussions here and here).

The PFAS approach also appears to have an important advantage over pure narrow banking schemes. The latter typically rely on floating-value mutual funds to provide the credit to households and businesses that narrow banks cannot. However, as they are currently constructed, such funds offer redemption without notice. Consequently, if these open-end funds hold a substantial volume of illiquid assets (remember they have taken over the traditional function of bank lending), they also would be subject to “first-mover advantage” on the part of claimants who can exit while the funds still have a few liquid assets available for a sale. As a result, even modest declines in confidence could readily turn into panics that undermine the supply of credit (see our earlier post).

In our view, the PFAS mechanism allows some leeway in managing this problem. First, so long as the haircuts on loans are not 100% (as they would be in a narrow banking world), banks can still play some of their traditional role as lenders. In effect, banks are “narrower” but not “narrow.” Moreover, to the extent that the haircuts are deep enough to shift a major portion of traditional lending to mutual funds, a regulator might consider extending the collateral posting requirement to the highly illiquid assets of these funds. That is, funds that are purely open-ended, allowing immediacy, would face a liquidity requirement. Alternatively, the funds themselves could choose to impose a notice period for withdrawal sufficient to limit the lingering first-mover advantage, become a bit more closed, and thereby avoid posting collateral.

In practice, the implementation of the PFAS approach would be far easier in jurisdictions where the regulatory framework is streamlined and shadow banking limited. For example, in the United Kingdom, where the Bank of England regulates and supervises the entire financial system (with the exception of policing conduct and protecting consumers), this looks straightforward. More broadly, we suspect that the EU and Japan also could introduce this approach in a reasonably uncomplicated manner. However, in the United States, with the overlapping jurisdictions of multiple federal and state regulators, and a regulatory framework filled with loopholes, implementation probably would require a radical regulatory modernization.

Returning to the meat of Professor King’s proposal, the impact of this reform on the structure of the financial system would almost certainly be profound. But the benefits of a safer banking system will not come without cost. The most important is that the central bank, through its haircut policy, will influence the composition of banks’ balance sheets. Since securities and loans with low haircuts will be cheaper to fund, the central bank’s haircut policy will determine the relative desirability of funding different borrowers. Depending on how the haircuts are set, they could also determine a bank’s leverage ratio.

Professor King notes that this role for central banks involves decisions that are not dissimilar from what they already do. Consider, for example, the collateral framework of the Eurosystem. The ECB’s website posts daily updates its list of eligible collateral, together with applicable haircuts. On June 6, 2016, for example, the list included 30,127 instruments, classified into one of 122 categories, with haircuts ranging from a minimum 0.5% to a maximum 65% (see here). (And, for a table with the Fed’s haircuts, see here.)

To be sure, these collateral rules create obvious complexities. What is the optimal number of categories? How often should securities be reclassified? How frequently can banks substitute one instrument for another? What about new issues and new types of instruments? Should the haircuts be used as a macro-prudential tool in the same way some people currently view changes in risk weights? Will political pressures induce the central bank to set haircuts unduly low, or to favor some politically preferred assets? Will haircuts impede useful financial innovation?

Put differently, it’s easy to think both of costs and benefits associated with the PFAS scheme, in which the central bank becomes the backer of all fixed-value liabilities with less than one-year maturity. At first blush, the benefits in terms of the improved resilience of the system, combined with the simplification or regulation, seem enormous. Are the costs even remotely as large? Probably not. Can the proposal be made to work without central banks becoming a credit allocation agency for the entire economy? We don’t really know, but we hope so.

The bottom line: Professor King’s promising proposal merits serious consideration by policymakers and scholars alike. Importantly, it has the potential to be significantly more effective than narrow banking at preventing financial crises.

APPENDIX