Reverse Repo Risks

Since the collapse of Lehman Brothers in September 2008, the Federal Reserve has stabilized the financial system and put the economy back on a path to sustainable growth.  This task involved creating a colossal balance sheet, which now stands at $4.37 trillion, more than four times the pre-Lehman level ($940 billion).  As textbooks (like ours) teach, along with this increase in Fed assets has come an increase in reserve liabilities (which represent deposits by banks at the Fed). Today, banks’ excess reserves (that is, the extra reserves beyond those that banks must hold at the Fed) are at $2.56 trillion, compared to virtually zero prior to the crisis.

Getting the money and banking system back to normal requires doing something to manage these excess reserves.  In an earlier post, we argued that the Fed will use reverse repos (RRPs) and target the repo rate to restore normality. In a repo operation, the Fed lends overnight by providing cash against the collateral of securities. A reverse repo goes the other way:  the Fed borrows overnight by receiving cash from its lenders while providing them securities as collateral. In this post, we will highlight a pitfall with the reverse repo approach to managing the Fed’s balance sheet.

To do so, we need to start with a few basics.  The motive of the reverse repo approach (and for the introduction of related operational tools such as term deposits at the Fed) is to drain reserves from the banking system, ensuring that the market interest rate climbs to the desired level. Two factors are driving the Fed’s change in monetary operations: (1) the desire to be able to raise overnight market rates without a massive and disruptive sale of assets; and (2) the large scale of the transactions that will be needed.

To raise interest rates, you might think that the Fed could simply raise the rate it pays on excess reserves – the interest on excess reserves or IoER. Surely no bank will lend to a risky counterparty (like another bank) at a lower rate, so the IoER sets a rate floor for bank lending. Yet, the overnight market interest rate (the federal funds rate) has been consistently below the 0.25% IoER since late 2008. Why? The answer is that some overnight lenders – including government-sponsored enterprises (GSEs) – cannot currently make deposits at the Fed. As the marginal suppliers of overnight funds, they appear willing to lend below the IoER.

To address this problem, and to make feasible open-market operations on a trillion-dollar scale, the Fed is broadening the range of its counterparties for reverse repo (and similar tools) well beyond the 22 primary dealers with whom it usually transacts. As of June 2014, the expanded list adds 18 banks, 6 GSEs, and 94 mutual funds, for a total of 139 counterparties.

Since September 2013, the Open Market Operations (OMO) desk of the Federal Reserve Bank of New York (FRBNY) has been running a series of reserve-draining experiments.  These experiments have included engaging in RRPs, with the transaction sizes and the rates fixed by the Fed. The Fed has slowly increased the maximum amount per counterparty per day and moved the rate around. The maximum started at $500 million and currently stands at $10 billion, while the interest rate has ranged between 0.01% and 0.05% (1 and 5 basis points).  The volume peaked recently at more than $200 billion (see chart). The Fed also has introduced term deposits (typically with a maturity of seven days) to absorb reserves.

Sources: Federal Reserve Board H.4.1 and FRED.

Sources: Federal Reserve Board H.4.1 and FRED.


Can the Fed really manage the supply of reserves using RRPs?  Our answer is yes, but there are some worrisome risks. The prime concern is the potential for disintermediating the banking system. Disintermediation occurs when depositors exit the banks and move their funds elsewhere, say to money market mutual funds (MMMFs).  There are two ways this disintermediation could occur: one in normal times, and one when the financial system comes under stress.

The issue in normal times is a technical one – complex, but likely manageable. The problem is that banks face costs with RRPs and term deposits that are different from the costs facing other Fed counterparties, so these cost differences must be reflected in the interest rates on these tools. For example, banks pay a premium for deposit insurance. The spread between the federal funds rate and the IOER – currently 15 basis points – gives us some idea of this insurance cost. If the Fed were to set the RRP rate much above the IOER rate minus 15 basis points, then banks would not be able to compete for the funds with nonbanks. Banks also face capital and liquidity coverage requirements (LCR). As a result, it may be difficult to attract LCR-burdened banks into deposits of longer maturity. More broadly, it could take some time for the Fed to calibrate the interest rate settings on these new tools to avoid bouts of disintermediation and re-intermediation.

Our primary concern with RRPs is not in normal times, but rather when the financial system comes under stress. In such circumstances, large-scale RRPs could trigger a disruptive bout of bank disintermediation. To see the risk, imagine that you are a large institutional investor or the treasurer of a large firm.  Your cash on hand far exceeds the FDIC insurance limit.  (According to Table I-C of the Quarterly Banking Profile for the first quarter of 2014, nearly 40% of domestic deposits are not insured by the FDIC, which also does not insure deposits abroad.) Now, imagine that the banking system starts to look shaky (again). Would you rather have your cash in a bank whose assets are of uncertain value or in an MMMF that lends only to the Fed against good collateral (using the RRP facility)? We expect that uninsured depositors would flee the banks for the MMMFs.

What can the Fed do to limit this risk? One solution would be to cap the pace of expansion of the RRP and term deposit facilities (as discussed recently by FRBNY President William Dudley). If such limits were to bind in normal times, the Fed probably would need yet other policy tools to absorb reserves. In a crisis, the Fed also could scale back or halt reverse repos, but relying solely on such adroit policy discretion to keep the financial system stable seems risky, so it makes sense for the Fed to explore mechanisms that would be more automatic and self-correcting.

During the financial crisis, the Fed had no choice but to use its balance sheet to steady the financial system and the economy. But this policy has left the Fed holding enormous quantities of assets that it cannot sell without massively disrupting financial markets.  That means finding another way to tighten policy. Because the Fed can pay interest on reserves, everyone has believed that the tools for this are there.  But actually putting into practice a reserve-absorption process that has never been used to raise interest rates entails risks. As a result, when the Fed starts to tighten policy, it will again have virtually no choice but to learn by doing. Expect a bumpy ride.