Key Highlights

• Reserves are now below $3tn – potentially testing the border between abundant and ample
• Repo rates appear to be rising as reserves fall, suggesting the border is getting closer
• Funding market volatility persists post-September, but this development was not unexpected

View chart pack

EXHIBIT #1: FALLING RESERVES, RISING TGA, TRIVIAL RRP

_{Source: BNY Markets, Federal Reserve Board of Governors, U.S. Treasury}

U.S. banking system reserves have fallen below $3tn for the first time since early 2023. From their recent peak on April 9 this year, systemwide reserves have fallen by nearly $500bn, some 14.5% lower. While this development was not unexpected, it suggests that reserves are close to becoming merely ample – rather than abundant. This, in our opinion, is a large driver of higher funding rates and volatility in money markets. 
 
Such a notable decline in reserves was not unexpected (see here, for example) when the budget deal was approved by Congress in early July, allowing the Treasury Department to resume T-bill issuance; since July 8, net bill issuance has exceeded $600bn. Take-up of the Fed’s reverse repo facility (RRP) over that same period has fallen to trivial levels, outside of month- and quarter-end dates. Finally, the increase in the Treasury General Account (TGA) has proceeded apace, with the government’s checking account growing from under $300bn in early July to well over $800bn today – roughly the same increase as the decline in reserves. Exhibit #1 shows the evolution of the liability side of the Fed’s balance sheet over the past 12 months. 
 
Last week (see here), we reviewed three resent speeches from Federal Reserve officials – Dallas Fed President Logan, Fed Vice Chair for Supervision Bowman and New York Fed Deputy Manager of the System Open Market Account Remache – that collectively discussed topics from Fed operating policy, the Standing Repo Facility (SRF) and volatility in funding markets. We argued that although they deal with different topics, taken as a whole the three speeches demonstrate the nexus among these topics: the responsiveness of funding market rates to changes in reserves as the latter decline and the implementation of monetary policy – particularly given the presence of the SRF. 

EXHIBIT #2: RESERVES NOT WORRYINGLY LOW VERSUS KEY AGGREGATE VARIABLES

_{Source: BNY Markets, Federal Reserve Board of Governors, Bureau of Economic Research} 

In absolute levels, reserves at sub-$3tn may seem low but in relationship to the size of the U.S. economy (using nominal GDP) or the aggregate size of the banking system (total banking sector assets), they are still (at 21.6% and 9.7%, respectively) well above where they were when we last experienced severe repo stress in September 2019. These are crude measures, admittedly, and we don’t really know at which level reserves will transition from abundant to ample. 
 
In an abundant reserve regime – where we have been and presumably still are, the quantity of reserves in the system is so high that there is little impact from periodic fluctuations in their levels on the effective federal funds rate. In this regime, the Fed uses its administered rates (IORB and RRP) to keep the funds rate in its targeted range. In an ample regime, the supply of reserves does influence short term rates. In supply and demand analysis, the price of liquidity (interest rates) and the supply of liquidity (reserves) have a relationship which traces out a demand curve. In an abundant reserve regime, the demand curve is flat. Changes in the supply of liquidity don’t lead to changes in its price. In an ample regime, the demand curve is downward sloping, with changes in liquidity supply leading to changes in its price: fewer reserves result in higher rates. 
 
The current challenge facing the Fed is assessing where that change in the demand curve for liquidity will occur. The New York Fed’s new “Reserve Demand Elasticity” gauge tries to help in that assessment (see here). It is essentially the slope of the demand curve for liquidity, determined by regressing changes in rates versus changes in daily liquidity. Currently – as has been the case since the pandemic – it is close to zero, indicating the demand curve is flat. If the estimated slope were to turn negative, it would be a strong indicator that reserves were merely ample. 
 

EXHIBIT #3:  DOWNWARD SLOPING DEMAND CURVE?

_{Source: BNY Markets, Federal Reserve Board of Governors, Bloomberg}

However, in light of President Logan’s argument that the federal funds market’s “connections [to the money market] are fragile and could break suddenly” and the federal funds target is “outdated,” what about the relationship of reserves to other funding markets. Is it also true that changes in reserves have no impact on these rates? We performed a relatively simple exercise. As daily reserve levels aren’t publicly available, we used weekly changes in reserve levels from the Fed’s H.4.1 table and regressed the tri-party general collateral rate (TGCR) spread over IORB against weekly changes in reserves. The results are shown in Exhibit #3, where we plot the regression coefficient. Recall that a negative value indicates the price of tri-party repo rises (falls) as reserves decline (increase). This is the case most recently. However, unlike the pre-2019 period, when this slope was sharply negative, it’s only slightly below zero currently. Nevertheless, this bears watching.  

EXHIBIT #4: PERSISTENT FUNDING MARKET VOLATILITY

_{Source: BNY Markets,Bloomberg, Federal Reserve Board of Governors}

However, in light of President Logan’s argument that the federal funds market’s “connections [to the money market] are fragile and could break suddenly” and the federal funds target is “outdated,” what about the relationship of reserves to other funding markets. Is it also true that changes in reserves have no impact on these rates? We performed a relatively simple exercise. As daily reserve levels aren’t publicly available, we used weekly changes in reserve levels from the Fed’s H.4.1 table and regressed the tri-party general collateral rate (TGCR) spread over IORB against weekly changes in reserves. The results are shown in Exhibit #3, where we plot the regression coefficient. Recall that a negative value indicates the price of tri-party repo rises (falls) as reserves decline (increase). This is the case most recently. However, unlike the pre-2019 period, when this slope was sharply negative, it’s only slightly below zero currently. Nevertheless, this bears watching.  

Forward look

If federal funds are not truly reflective of liquidity conditions in the market (and, incidentally, the federal funds rate has nudged a basis point higher over the last two weeks) but repo is, this would support Logan’s take. In addition, it could suggest that liquidity is indeed getting tighter and its price rising – suggesting the recent drop in reserves is meaningful.

The volatility we saw in funding markets in September was also not entirely unexpected given lower reserve levels and month-specific factors like the September 15 tax date and quarter-end. In Remache’s words, “this phenomenon is – within limits – normal, not concerning and exactly what we expect as we continue to normalize our balance sheet.” Indeed, post-September money market rates remain volatile and elevated, as Exhibit #4 shows. We plot the spread of TGCR over IORB, as well as that for SOFR versus EFFR. Since end-September, these spreads are still elevated. Combined with our – admittedly simple – exercise described above we should hear more about the possibility that reserves are indeed transitioning from abundant to merely ample.

Chart pack