Changes in financial regulation are having a profound impact on the demand for safe assets—assets with a fixed nominal value that may be converted at all times without loss into the means of payment. Not only is demand for safe assets on the rise, but the ability of the private sector to produce them is being constrained by new rules that limit the extent and nature of things like securitizations.
Now, the relevant changes in regulation are varied, but their common goal is to make the financial system more resilient. For example, banks are now required to hold more liquid assets to back liabilities that can quickly take flight, while people who engage in derivative contracts are required to post cash or securities guarantees. These changes are a feature of the new system, not a bug. The increased demand for safe assets is largely an intended consequence of the reforms.
So far, the fallout from increased demand and constrained supply looks reasonably benign. But for several years now, broad financial conditions have been very calm, with measures of financial volatility and stress at or near long-term lows. What will happen when the financial system comes under stress again? What if there is a drop in risk tolerance (or a surge in risk awareness) and a flight to safety that causes a jump in the demand for safe assets or a plunge in the supply? Or, as in 2008, what will happen if both materialize at the same time? We need to be ready.
As we will explain in more detail, central banks in advanced economies can satisfy the heightened need for safe assets under stress (as well as the precautionary demand in normal times) by offering commercial banks committed lines of credit for a fee against collateral, as the central banks in Australia and South Africa currently do. In our view, this mechanism for ensuring sufficient supply of safe assets in a crisis has important advantages compared to one in which the central bank operates perpetually—in good times and bad—with a very large balance sheet.
To see how this would work, we start with an explanation of post-crisis liquidity regulation; specifically, we need to explain the liquidity coverage ratio (LCR), which has prompted much of the increase in the demand for safe assets in recent years. In the words of the Basel Committee on Banking Supervision (BCBS), the LCR “aims to ensure that a bank has an adequate stock of unencumbered high quality liquid assets (HQLA) … that can be converted into cash at little or no loss of value in private markets to meet its liquidity needs for a 30 calendar day liquidity stress scenario.” The exact definition of HQLA in the vast majority of jurisdictions—what counts and by how much—is complex, including certain foreign and corporate bonds, for example. But the focus is primarily on central bank reserves and securities that are issued or fully guaranteed by the domestic government.
Looking at the United States, as the following chart shows, in the aftermath of the 2007-09 financial crisis commercial banks dramatically increased their holding of liquid U.S. government instruments. In 2006, for example, the sum of cash, reserves, Treasury and agency instruments constituted $1½ trillion of banks’ $9 trillion in assets. Today, the number is $4.8 trillion out of $16.3 trillion. That is, these top-quality liquid assets now account for 30 percent of total assets, nearly double what they were a decade ago.
Commercial bank holdings of liquid U.S. government instruments (Billions of dollars), 2006-June 2017
Source: Board of Governors of the Federal Reserve System, H.8.
How much HQLA do banks hold to meet the LCR? First, banks likely hold somewhat more than required both to provide a usable buffer and as a precaution to avoid any penalties that might result from a shortfall. While there is no aggregate HQLA data (that we know of) for the U.S. banking system, recent public disclosures from the largest banks allow us to make a rough estimate. Together, JPMorgan Chase, Bank of America, Citigroup, and Well Fargo, which account for about one-half of all U.S. commercial bank assets, report holding $1.9 trillion in HQLA as of the end of March 2017, So, provided they are representative, total HQLA appears to be a bit under $4 trillion. (This estimate is consistent with others, like the one in Nelson.)
Is $4 trillion a big or a small number? We view it as modest. For example, one benchmark might be how large bank holdings of Treasurys and agencies are relative to the domestic supply to the private sector. Commercial banks report holding $2.4 trillion as of March 2017 (see H.8). This reported figure exceeds the estimated HQLA component, which is closer to $1.7 trillion (that is, roughly, $4 trillion minus the $2.3 trillion held in the form of reserves at the Fed). Banks’ total holdings constitute less than 15 percent of the available supply, which includes $7.5 trillion of U.S. Treasury debt and $5.8 trillion of agency and government-sponsored enterprise issuance held by domestic entities outside of the Federal Reserve. (For a global perspective on safe asset demand and supply, see the IMF’s analysis in 2012 here.)
In theory, banks could reduce their need for HQLA by further lowering their reliance on wholesale funding. However, they’ve already shifted substantially in this direction: since the crisis, deposit liabilities—the sort that tend be stable and have favorable treatment under the LCR rules—have risen by more than 80 percent, while other liabilities are steady or down slightly. So, in practice, the banking system’s demand for HQLA is unlikely to fall.
What does theory say about the role of safe assets in a crisis? One recent strain of analysis highlights their importance for cyclical fluctuations of growth and employment. For example, Caballero, Farhi and Gourinchas show how a shortage of safe assets can trigger a decline in output. In their model, safe assets have a number of uses, including as a store of value (wealth) and a means of payment. Now, when demand for an asset rises, normally its price would rise and its interest rate or yield would fall. But there is a limit to the degree to which interest rates can fall: namely, the effective lower bound. Caballero and Farhi describe this as a “safety trap,” which bears some resemblance to a conventional liquidity trap. And, much as fiscal stimulus can boost an economy out of a liquidity trap, government issuance of additional safe assets can facilitate escape from a safety trap.
From our perspective, financial stability concerns provide a more compelling argument for the central bank provision of credit commitments. The concern, articulated by Greenwood, Hanson and Stein, starts with the observation that the Treasury yield curve is very steep at the short end, pointing to a high demand for safe assets with maturities of less than three months or so. Unless the government meets this demand by issuing more short-term debt, the private sector will have a strong incentive to do so, favoring the creation of vulnerable, runnable liabilities. This is one interpretation of what happened prior to the crisis: institutional investors and financial intermediaries wanted safe assets, so they created AAA asset-backed commercial paper and super-senior AAA CDOs, among other things. And, following this logic, when many of these assets were suddenly viewed as unsafe, there was a shortage and the crisis hit. (For such interpretations, see Holmström and Gorton.)
To the extent that there is a shortage of safe assets induced either by regulations like the LCR, by demand for other purposes such as collateral in derivatives transactions, or by a flight to quality, the central bank can address this by offering a committed liquidity facility (CLF).
What is a CLF? To understand it, a bit of history is useful. When the BCBS was considering the creation of an LCR, Australian officials pointed out that—due to the small size of their government debt—there would be insufficient HQLA to meet the regulation-induced demand. (The Reserve Bank of Australia estimates that banks can meet roughly 40 percent of their LCR requirement with government securities.) As defined in the LCR rule, the CLF solves this problem:
[T]hese facilities are contractual arrangements between the central bank and the commercial bank with a maturity date which, at a minimum, falls outside the 30-day LCR window ... Such facilities are only permissible if there is also a fee for the facility which is charged regardless of the amount, if any, drawn down against that facility. (BCBS 2013, paragraph 58)
To avail themselves of the CLF, banks must have eligible collateral, and must pay a fee. In Australia, the current fee is 15 basis points. And, if the bank draws on the line, they also pay the standard borrowing rate—namely, the RBA’s target rate plus 25 basis points. Now, because there is insufficient government debt, the collateral is not free of default risk, but it is subject to a haircut that can be set to reflect that risk. We think of this facility as a means to convert assets with default risk into safe assets for a fee. (See here for a description.) In that sense, the CLF also resembles former Bank of England Governor Mervyn King’s proposal that the central bank serve as a committed “pawnbroker for all seasons.”
By offering to supply safe assets elastically in this manner, the central bank puts a cap on the price of the government-issued assets that qualify as HQLA. A CLF also makes it much less attractive for the private sector to create “safe assets” that can become runnable liabilities in a crisis. And, as Jeremy Stein emphasized several years ago, this mechanism stabilizes the costs of banks supplying liquidity at times when the price of securities could be very volatile.
As an aside, it is worth noting that the ECB’s fixed-rate full-allotment policy for supplying reserves through refinancing operations functions in a manner that is quite similar to a CLF. That system allows euro-area banks to exchange any of roughly 30,000 securities for reserves at the policy rate in unlimited quantity. The securities face a haircut, and the exchange is done at the main refinancing rate—the ECB’s policy rate. What this means is that, for a price, a bank can always convert non-HQLA assets into reserves, which always count as HQLA.
From our perspective, one notable advantage of a CLF is that, during normal times, it can substitute for a large central bank balance sheet. Like a large balance sheet, a CLF can allow for greater volatility of non-reserve central bank liabilities without triggering interest rate changes (see Logan to understand policy operations with a large balance sheet). Moreover, a CLF obviates the need for a central bank to hold perpetually a large inventory of safe securities so that can it lend them (the Fed can reverse repo them) in times of stress. In a period of severe liquidity shortfall, banks would exercise their CLF options, resulting in a virtually automatic expansion of central bank liquidity supply, much as Bagehot would have prescribed.
In our view, there are strong political economy arguments in favor of allowing central bank balance sheets to shrink over the long run (see, for example, Plosser and our recent post). A CLF can allow the central bank to do so without sacrificing monetary control or the ability to stabilize a distressed financial system.
All of this leads us to the conclusion that central banks should seriously consider creating CLFs. Properly designed and properly priced, it is a useful policy tool that substitutes for other, potentially riskier tools. In normal times, pricing can be set to encourage liquidity discipline, something that was clearly lacking in the runup to the Great Financial Crisis. During times of stress, adjustments in the collateral requirements and the pricing could be used to relieve shortages of safe assets, so that banks can use their government securities for other purposes.