The Fed’s Reserves Management Problem
There has been a lot written about the Fed’s recent decision to start purchasing T-bills to re-expand its balance sheet, in order to release some of the upward pressure on short-term interest rates in the repo market. Some people have called this a resumption of Quantitative Easing, while others point out that it is merely an adjustment to a technical condition of reserves shortage. The problem is that both perspectives may be right, under different circumstances, and that is the underlying problem.
The triggering issue here was that overnight repo rates had been trading tight, and in fact briefly spiked to around 10%. It isn’t surprising that the Federal Reserve responded to this problem by adding lots of short-term liquidity: that’s how they respond to every issue. Banks in trouble? Add liquidity. Economy slightly weak? Add liquidity. “Stranger Things” episode somewhat disappointing? Add liquidity.
Traditionally, the Fed’s response would have been correct. In the “old days,” the overnight interest rate was how the Open Markets Desk gauged liquidity in the interbank market. If fed funds were trading above the Fed’s desired target (which was not always announced, but which could always be inferred by the Desk’s actions in response to reserves tightness or looseness), the Fed would come in to do “system” repos and add short-term liquidity. If fed funds were trading below the target, then “matched sales” was the prescription. It was fairly straightforward, because the demand for reserves was relatively easy to monitor and the adjustments to the supply of reserves small and regular.
But the problem today goes back to something I wrote about back in March, and that’s that reserves no longer serve just one function. In those aforementioned “old days,” the function of reserves was to support a bank’s lending activities in a straightforward statutory formula that was easy for a bank to calculate: this amount of lending required that amount of reserves, calculated over a two-week period ending on a Wednesday. Under that sort of regime, spikes in funds and repo rates (other than occasionally over the turn of year-end) were very rare and the Desk could easily manage them.
This is no longer the case. Reserves now serve two functions, as both lending support and as “High Quality Liquid Assets” (HQLA) that systemically-important banks can use in calculating its Liquidity Coverage Ratio (LCR). This has two really critical implications that we will only gradually learn the importance of. The first implication is that, because the amount of reserves needed to support lending activities is unlikely to be exactly the amount of reserves needed for a bank to achieve its HQLA, at any given time one of these two effects will dictate the amount of reserves the system needs. For example if banks need more reserves for HQLA reasons, then it means they will have more reserves than needed for their existing loan books – and that means economic stability and inflation control will in those cases take a back seat to bank stability. So, as the Fed has struggled to keep up with HQLA demand, year-over-year M2 growth (which is partly driven by reserves scarcity or plenty) has risen fairly quickly to 2-year highs (see chart, source Bloomberg).
The second implication is that, because the demand for each of these two functions of reserves changes independently in response to changes in interest rates and other market forces, it is not entirely knowable or forecastable by the Desk how many reserves are actually needed…and that number could change a lot. For example, there are other assets that also serve as HQLA; so if, for example, T-bill yields were a bit higher than the interest paid on reserves a bank might choose to hold more Tbills and only as much reserves as needed to support its lending activities. But if Tbill rates then fall, or customers lift those bills away from the bank’s balance sheet, or the denominator of the LCR (the riskiness of the bank’s activities, essentially) changes due to market conditions, the bank may suddenly choose to hold lots more reserves. And so rates might suddenly spike or plummet for reasons that have to do with the demand and supply of reserves for the HQLA function, with the Fed struggling to add or subtract large amounts of reserves over short periods of time.
In such a case, targeting a short-term interest rate as a policy variable is going to be exquisitely more difficult than it used to be, and honestly it isn’t clear to me that this is a solvable problem under the current framework. Either you need to declare that reserves don’t qualify as HQLA (which seems odd), or you need to require that a bank hold a certain amount as HQLA and set that number high enough that reserves are essentially the only HQLA a bank has (which seems punitive), or you need to accept that the central bank is either going to have to surrender control of the money supply (which is scary) or of short-term interest rates (which is also scary).
But simply growing the balance sheet? That’s the right answer today, but it might be the wrong answer tomorrow. It does, though, betray that the central bank has a knee-jerk response to err on the side of too much liquidity…and those of us who remember that inflation is actually a real thing see that as a reason for concern. (To be fair, central banks have been erring on the side of too much liquidity for quite some time. But maybe they’ll keep being lucky!)
