Pitfalls of a Reserves-only Narrow Bank

“TNB’s sole business will be to accept deposits only from the most financially secure institutions, and to place those deposits into TNB’s Master Account at the FRBNY, thus permitting depositors to earn higher rates of interest than are currently available to nonfinancial companies and consumers for such a safe, liquid form of deposit.”
TNB USA Inc. vs Federal Reserve Bank of New York, Case 1:18-cv-07978, August 31, 2018.

“TNB’s business model poses no threat to the financial system or to depositors that we can see.”
The Editorial Board, The Wall Street Journal, September 12, 2018.

Since retiring from the Federal Reserve in mid-2016, our friend Jamie McAndrews has been very busy. Unlike most of us, he is putting his ideas into action: in 2015, he and a number of his colleagues, proposed the creation of segregated balance accounts (SBAs). As they write, “SBAs are accounts that a bank or depository institution (DI) could establish at its Federal Reserve Bank using funds borrowed from a lender.” Their proposal is that a bank would offer a special account that is fully collateralized by a deposit at the Federal Reserve. Furthermore, the SBA deposits would be remunerated at the interest rate the Fed pays on excess reserves (the IOER), minus a small fee for the bank.

We have no expertise whatsoever in determining whether the Fed has legal grounds for denying TNB a Master Account—the subject of the court case in the opening quote. But we do have concerns about SBAs and narrow banks: we worry that they would shrink the supply of credit to the private sector and aggravate financial instability during periods of banking stress. Compared to what may be large costs, we suspect that the benefits would be small.

To understand our conclusion, start with the obvious: as far as we know, no bank has adopted the SBA model. That is, commercial banks do not offer customers the option of opening such an account. Here are two possible reasons. First, SBAs are unlikely to be attractive unless reserves are remunerated at a meaningful rate. While the Fed began to pay interest on excess reserves (IOER) in 2008, the rate was just 0.25 percent until the end of 2015. And, banks would have had to subtract from that the cost of deposit insurance. So, until recently, such a business would have had poor prospects.

Second, while emphasizing that we are not lawyers, we wonder whether holders of SBAs provided by a traditional commercial bank might be concerned about their property rights in a resolution proceeding. Would they face the same risks as other liability holders—incurring losses (depending on their seniority) along with other bank creditors as if their assets were not fully segregated in bankruptcy? If so, SBA-like accounts inside of a traditional commercial bank—even if they appear segregated inside a going concern—might be of limited interest to customers. Put differently, they may not be quite so “risk-free” as the SBA term suggests.

Given that traditional commercial banks have not offered SBAs and seem unlikely to do so, and that with the IOER rate near 2 percent the SBA business appears more promising, it is natural that fully segregated SBA banks would now try enter the competitive fray. Hence, TNB, which is short for “The Narrow Bank.” Importantly, an SBA-only bank faces no liquidity, maturity or credit risk. Consequently, the rise of such narrow banks would intensify the funding competition for commercial banks: the latter currently do not need to offer risk-adjusted returns that are comparable to an interest-bearing deposit at the Fed. (A relatively small issue is that, in the absence of some regulatory change, it seems likely that SBAs would face a deposit insurance premium related to the overall riskiness of the issuing bank’s overall balance sheet, thereby increasing the relative cost to traditional commercial banks of providing SBAs.)

We view the economics of an SBA-only bank as virtually identical to that of a range of past proposals for altering the financial system, including Tobin’s deposit currency accounts; Greenwood, Hanson and Stein’s suggestion that the Fed offer large quantities of reverse repurchase agreements (RRPs); Cochrane’s advocacy of narrow banking, as well as his proposal that the U.S. Treasury issue variable rate perpetuities in single dollar amounts; and Barrdear and Kumhof’s suggestion that central banks should offer universal, unlimited access to deposit accounts. Through one mechanism or another all of these structures allow individuals to place their funds directly or indirectly in an interest-bearing deposit at the central bank (or, in one of Cochrane’s schemes, with the U.S. Treasury).

To be clear, we see no functional difference between a central bank providing retail deposits directly to all comers and a commercial bank or money market mutual fund (MMMF) offering liabilities that are guaranteed to be deposited at the central bank and always demandable—rain or shine. (Of course, some MMMFs already are Treasury-only funds.) Consequently, an SBA bank that initially offers accounts only to large commercial customers—picking off the low-hanging fruit of uninsured depositors—should be viewed as the first step to offering deposits to retail customers, either directly or through brokers and aggregators.

We have made our view of these proposals to provide universal access to the central bank’s balance sheet clear: they risk disintermediation of the commercial banking system, without necessarily stemming the runs that their proponents wish to stop.

So, suppose that uninsured depositors find an SBA-only bank attractive. How large could the shift of funds from commercial banks be? And what would that imply for the supply of bank credit? To get a sense of the magnitude of the risk, we can look at a bit of data. The following chart, using data from the FDIC’s Quarterly Banking Profile, shows the evolution of uninsured deposits over the past 30 years, as well as the ratio of uninsured deposits to bank loans. Uninsured deposits have increased roughly six-fold since 1990, and have come to be roughly 45 percent of total deposits, or more than $6 trillion.

Uninsured deposits: Value (trillions of dollars) and ratio to total deposits (percent), 1990-2Q 2018

Note: The abrupt shifts during and after the financial crisis appear to reflect the changes in the scope and scale of the guarantees extended in the course of the FDIC’s Temporary Liquidity Guarantee Program (TLGP) that began in 2008 and expired at the end of 2012. Source: FDIC, Quarterly Banking Profile, and authors’ calculations.

Turning to the asset side of commercial banks’ balance sheets, loans and leases―which include commercial and industrial loans, residential and commercial real estate loans, and consumer loans, among other things―currently stand at $9.4 trillion, or 56 percent of total U.S. banking system assets, roughly midway between the 1990s peak of 62 percent and the 2014 trough of 50 percent (see chart). Most of these credit items, which are at the core of the intermediation services that the banking system is supplying to the real economy, are almost surely highly illiquid. It may be straightforward for other parts of the financial system to step in and hold the $5.5 trillion of securities and cash assets that banks own. But it will require far more innovation for institutions and markets to substitute for the supply of commercial and residential loans. At best, we foresee considerable friction (and cost) associated with such a shift.

Ratio of loans and leases to total bank assets for all commercial banks (percent), 1990 to 2018

Source: FRED.

Perhaps most important, if the financial system were to come under stress, the process of disintermediation could prove sudden and disruptive. While commercial banks will compete by offering a higher (non-risk-adjusted) reward for depositors, the larger the risk premium that they need to pay, the smaller (and more costly) the supply of credit to the private sector. That risk premium also would serve as an index of stress.

Since narrow banks don’t make loans, what could substitute for a dwindling supply of commercial bank credit? Private equity funds—that can avoid fire sales—could further increase their lending in good times, but their willingness to do so in a recession (let alone during a crisis) is untested. We suspect that mutual funds and investment companies would try to fill in—much like those funds that have begun purchasing and packaging the consumer loans of peer-to-peer lenders. As we describe in an earlier post, such mutual funds would be engaged in credit, maturity, and liquidity transformation—that is, their assets will be riskier, longer term, and less liquid than their liabilities. Yet, in the absence of deposit insurance and without access to a lender of last resort, such bank-like activities make them vulnerable to a wave of redemptions by investors motivated to sell first while the fund’s relatively liquid assets are still on hand. Just as in the case of bank runs, such a first-mover advantage can trigger fire sales and undermine the supply of private credit (see, for example, Goldstein, Jiang and Ng). Indeed, such stress would favor further shifts into narrow banks that offer no private credit.

Why aren’t SBAs the same as U.S. Treasury-only MMMFs? Or, as John Cochrane points out in his recent defense of TNB, why should we care whether individuals are holding Treasury securities directly, or whether they make deposits in a bank that holds reserves at the Fed that then holds Treasury securities? This is surely right up to a point, but what happens under stress? First, even accounting for the usual flight to Treasurys in a crisis, we suspect that investors will find an SBA-only bank safer and more attractive than a Treasury-only MMMF. That is, the presence of banks like TNB will only serve to intensify the shift of funds out of conventional banks, draining their reserves and boosting the risk of fire sales of illiquid assets.

Suppose, nevertheless, that the courts legally oblige the Federal Reserve of New York to grant segmented narrow banks accounts that allow them access to interest-bearing central bank liabilities. How might the Fed then avoid the risk of disintermediation and financial instability?

We see two unappealing alternatives: the poorer of the two would focus on the quantity of reserves, the other on the price. The first option is that the Federal Reserve could restrict the quantity of reserves in the system, limiting the size of the deposits that segmented SBA banks can make. Unfortunately, this means shifting Fed operations from targeting an interest rate (like the federal funds rate) to targeting the quantity of reserves. When the Fed did this from 1979 to 1982, interest rate volatility skyrocketed, adding unnecessarily to economic and financial uncertainty. Fortunately, the Fed has not tried this experiment again.

The second possibility is to end the current regime of monetary control by eliminating interest payments on reserves. Aside from episodes of severe financial stress, when investors run to safety, we doubt that the business model of segmented narrow banks would be viable at a zero interest rate. As it stands, the Federal Reserve Board has the authority to set the IOER at any level it wishes, including zero.

Yet, giving up on the IOER and the interest rate floor system that authorities have painstakingly constructed over the past decade would be very unfortunate. With the IOER set at zero, the central bank would need to supply the volume of reserves that matches reserve demand at the target interest rate: that is, the Fed would return to operating on the downward-sloping portion of the banks’ demand curve for reserves, in contrast to the current circumstances where banks demand reserves elastically at the IOER rate. Banks would once again economize on their reserve holdings to the extent possible, perhaps reducing the quantity to pre-crisis levels well below $100 billion. To meet their liquidity coverage ratio (LCR) in this world of scarce reserves, banks would have to hold far more Treasury securities.

Perhaps most important, the Fed would sacrifice what some people see as a key tool for financial stability: the current option to set the IOER and the supply of reserves independently allows the central bank to respond rapidly to spikes in demand for high-quality liquid assets during a period of stress, while continuing to set monetary policy consistent with its inflation and economic objectives.

All in all, we worry that the rise of segmented narrow banks could leave the Federal Reserve facing a risk of disintermediation that would diminish the supply of private credit, potentially in a disruptive way. Returning to a pre-crisis world of scarce reserves would diminish this threat at the cost of a sacrificing a useful tool for financial stability. In theory, Congress might find this outcome sufficiently unappealing to restrict SBA-only banks, but that possibility seems remote.

Acknowledgement: Without implicating them in any way, we thank our friends and colleagues Paul Wachtel and Larry White for their very helpful comments.