The Fed Funds Market During the Quantitative Tightening of 2017-19

September 19, 2025

The Fed Funds Market During the Quantitative Tightening of 2017-19

David Bowman

The effective federal funds rate (EFFR) declined by about 10 basis points on every month end from 2016 to February 2018. Then, in March 2018, this pattern suddenly stopped (Figure 1). This Note discusses the dynamics behind the federal funds market, including its relationship with repo markets, to explain this change and to better understand the behavior of money markets during the period of the Federal Reserve's quantitative tightening between October 2017 and September 2019. The market dynamics of the last period of quantitative tightening have potential implications for the current period. Then, as is the case now, quantitative easing was proceeding at a modest pace, while demand for liquidity in repo financing was growing rapidly, reflecting the rising level of federal debt and associated cash-futures basis trades to finance hedge-fund holding of Treasury debt.

Part of the reason that EFFR would have declined on month ends is that at least some portion of the foreign banks that are the dominant borrowers in the fed funds market have incentives to pull back from fed funds borrowing in order to reduce their regulatory exposures on month-end reporting dates. Figure 2 shows the change in borrowing activity on month ends by the foreign banks that I have identified as regularly reducing their borrowing, along with the change in borrowing by other banks.¹ As can be seen, borrowing volumes for the identified banks (the blue line) regularly declined by $5-$15 billion on month ends, declining more on quarter ends than on other month ends. Borrowing by other banks did not show a particular pattern around month ends, sometimes rising and sometimes falling. Importantly, however, these patterns continued even past March 2018 (and have in fact continued to the present), although EFFR stopped declining at month ends as of March 2018. Thus, there wasn't a change in borrowing behavior that would explain the shift in the dynamics of EFFR.^{2 3}

What did change between February and March 2018 was the relative level of EFFR and repo rates, which I measure with the Triparty General Collateral Rate (TGCR), relative to the rate of interest on reserve balances (IORB) offered by the Federal Reserve. As shown in Figure 3, the TGCR had been routinely below EFFR in the period up to February 2018; however, that abruptly changed in March 2018, when the TGCR abruptly moved higher than EFFR and remained close to or above it throughout the rest of the period of quantitative tightening.

The sudden rise in repo rates was related to another change over the same period: beginning on December 8, 2017, the U.S. federal government had reached the debt limit set by Congress and the Secretary of the Treasury had declared a debt issuance suspension period that continued until February 9, 2018, when Congress suspended the debt limit. Following the resolution of this debt limit episode, bill supply increased sharply, by $320 billion, between mid-February and the end of March 2018.^{4 5} As shown in Figure 4, bill supply rose notably relative to assets under management (AUM) by money market mutual funds (MMFs) and remained at those higher levels thereafter. In the seven weeks following the resolution of the debt limit, bill supply relative to AUM rose 12 percentage points, from 63 percent to 75 percent (it remained near that share for the rest of 2018 before declining back toward 60 percent as investors moved funds into MMFs). The scale of bill supply relative to MMF AUM is important because MMFs - particularly government MMFs - are key buyers of Treasury bills and also key cash investors in the Treasury repo market. An increase in bill supply relative to MMF AUM puts upward pressure on repo rates because it tends to draw funds away from the repo market and toward bills.^{6 7}

Just as the increase in bill supply helped to push repo rates higher, the rise in repo rates helped to push EFFR higher. For reasons that are discussed further in the next subsection, bargaining dynamics in the federal funds market tend to change when GC repo rates move above EFFR.

The Relationship Between EFFR and Repo Rates

The fed funds market is tied to repo markets in large part because the Federal Home Loan Banks (FHLBs), the main lenders in fed funds, also lend in repo. The FHLBs use both markets as liquid investments intended to support their ability to issue advances to their members at short notice.⁸ Because the FHLBs invest in these markets in order to meet their primary objective of supporting their members' potential needs for advances, the total amount of liquid resources they invest in depends more on the potential demand for advances than the rates earned on these investments. While they value some amount of diversification by investing in both markets, they tend to prefer to lend more in the fed funds market because of the flexibility of when money is returned, which is an advantage compared to triparty repo where money is returned late in the day. However, even with this preference toward fed funds, the rates that can be earned on money lent in the repo market serve as reservation point when the FHLBs determine the rates they are willing to lend at in fed funds. And when repo rates are high enough compared to EFFR, the FHLBs will begin to move more of their liquid investments into repo. Figure 5 shows that FHLB volumes lent into the triparty repo market began to increase in March 2018, when repo rates moved up. While the amounts moved into repo would have had no or only negligible impact on repo rates given the size of that market, any movement out of fed funds or move to raise FHLB lending rates in fed funds would have a large impact on the fed funds market given the size of the FHLBs in that market.

Although a number of academic papers model the fed funds market as a search market, in reality the FHLBs tend to lend the bulk of their funds to the same fairly small set of banks each day. As shown in Figure 6, about a third of FHLB fed funds lending volumes were directed to banks that borrowed every day between January 2016 and February 2018, more than two-thirds were directed to banks that borrowed 98 percent or more of the time. More than 90 percent of lending was directed to banks that borrowed 70 percent or more of the days in this period. In other words, this is a relationship-driven market rather than a search market, and the nature of trading makes rate setting in the fed funds market a repeated bargaining game.

For the FHLBs, lending in fed funds is attractive so long as the rate earned is at least as high as the rates that could be earned by lending in repo. For the most frequent borrowers, which tend to be foreign banks seeking to earn an arbitrage spread to IORB, borrowing from the FHLBs is attractive so long as the rate offered is sufficiently below IORB to cover balance sheet and other costs involved in unsecured borrowing.

When repo rates are well below the effective federal funds rate as they were in the period prior to March 2018, there will be a range between the repo rate and IORB at which the FHLBs can lend in the fed funds market that allows (a) those foreign banks able and willing to arbitrage to earn some spread to IORB that covers their balance sheet costs and allows for some amount of profit, and (b) the FHLB to earn a higher rate than the repo rate. In these circumstances, the rate that FHLBs lend at in the fed funds market seems to be negotiated so as to split the respective gains from trade. On month ends, when banks' incentives to borrow in fed funds declined, the fed funds rate fell closer to (but not typically below) the repo rate. Apart from month ends, these negotiated rates tend to be somewhat sticky since there are a limited set of lenders and borrowers involved, the parties transact regularly so the negotiated rates are part of a long-term relationship, and the amount of return to either side in negotiating a slightly higher or lower rate on any given day is fairly small.

However, when repo rates are persistently higher than the effective federal funds rate, the negating dynamics change. As long as the federal funds rate remains sufficiently below IORB, the banks engaged in arbitrage are still willing to borrow, but the FHLBs will no longer earn a better return than they could earn by investing in repo instead. As a result, the FHLBs are likely to insist on a higher rate earned in the fed funds market. Thus, the rise in repo rates in March 2018 can explain both the puzzle that we started with - with repo rates close to or above EFFR, the FHLBs would no longer be willing to allow fed funds rates to fall on month ends - and the general rise in EFFR over the next several months.

As shown in Table 1, empirically, we can see that EFFR's dynamics seem to change when repo rates rise above it. Excluding the changes around month ends, EFFR tends to move slowly and on most days it does not change from the previous day's value: as seen in column 1, on roughly 91 percent of the days between 2014-2025, EFFR did not move up or down. On the remaining 9 percent of days when EFFR did move, it was almost equally likely to move up or down. These probabilities are essentially unchanged conditioning on days when the TGCR was below or equal to EFFR (column 2) - while the fraction of days when EFFR declined was slightly higher than the fraction of days in which it increased, the null hypothesis that moves in either direction remained equally likely cannot be rejected. In contrast, conditioning on days when the TGCR was above EFFR (column 3), the probabilities clearly change - EFFR is somewhat more likely to move and roughly 4 times as likely to move up than it is to move down, and the null hypothesis that moves in either direction are equally likely can be rejected at the 1 percent significance level.

Table 1. The Empirical Relationship between Daily Movements in EFFR and the TGCR

To summarize the points made thus far, the dynamics involved between the FHLBs and their borrowers in the fed funds market explains the puzzle that we started with as to why EFFR suddenly stopped declining on month ends. The rise in repo rates following the resolution of the federal debt limit in 2018 eliminated the wedge between EFFR and the TGCR, thus closing off the bargaining room that was available to allow EFFR to fall on month-end reporting dates. More importantly, the evidence presented argues that the rise in repo rates drove EFFR higher after February 2018.

The Impact of the Rise in EFFR on Fed Funds Borrowing

In this final subsection, I examine the effects of the resulting rise in EFFR on the demand for borrowing in the fed funds market. To the extent that the rise in EFFR was caused by the increase in bill supply following the resolution of the federal debt limit and the resulting rise in repo rates, it would reflect a shift in FHLB's supply curve, and borrowing demand should decline if it is elastic.

We've argued that a substantial volume of borrowing is related to IORB arbitrage, but apart from that motive, there are also a set of domestic (and some foreign) banks that borrow in the fed funds market for their own internal liquidity needs and even banks engaged in arbitrage may borrow some portion of funds for the same types of liquidity needs. These banks will borrow from the FHLBs (in certain cases, some small domestic banks also may borrow from other domestic brokers or settlement banks) even at rates above IORB. As the fed funds rate moves close to or above IORB, any arbitrage demands for borrowing will diminish and these more fundamental sources of demand will become dominant.

However, there isn't an indicator pointing to which banks are borrowing for arbitrage reasons or for other reasons. Because arbitrage borrowers should have more elastic demand than other borrowers, I look at the banks that regularly borrow at lower rates from the FHLBs. Arbitrage borrowers would naturally borrow on most days, and if their demand is elastic, then the FHLBs would tend to charge them lower rates in order to facilitate their borrowing than they would for other borrowers with less elastic demand. I therefore take as our sample the set of borrowers from the FHLBs that (a) that on average borrowed at rates below the average rate charged by FHLBs between January 2016 and February 2018, and (b) borrowed 70 percent or more of days over that period.

Figure 7 compares the borrowing of these banks to the fed funds rate charged by the FHLBs (the data is averaged over each month so that the underlying trends in borrowing are clearer). This cohort of borrowers does appear to have been borrowing for arbitrage reasons. Tellingly, although rates rose 5 basis points in the first half of 2016, borrowing did not decline - there was no reason to curtail arbitrage borrowing at that time because, although they had risen, the rates borrowed at still remained well below IORB and so the arbitrage still remained profitable. In contrast, borrowing declined sharply when rates rose 5 basis points between February and September 2018. Once the spread to IORB falls below a bank's balance sheet costs of borrowing, arbitrage will no longer be profitable, and the borrowing will stop. While different banks may face different costs, essentially all of the decline in borrowing over this period had finished by the time the spread to IORB had reached -3 basis points.

As shown in Figure 8, this group was the only one to meaningfully reduce its borrowing following the rise in EFFR after February 2018. Banks that borrowed frequently but at higher rates reduced their borrowing modestly April 2018 but otherwise their borrowing was essentially unchanged even as rates continued to move higher, while banks that borrowed infrequently increased their borrowing even as rates rose. There is no evidence that the demand from these banks was at all elastic. In contrast, borrowing by our identified arbitrage banks fell by 50 percent between February and September 2019. After that time, once the rate they paid had moved within about 3 basis points of IORB, their borrowing stopped declining and remained essentially unchanged thereafter, even though the rate charged by the FHLBs continued to increase.

Conclusion

This note has sought to clarify the underlying dynamics driving the fed funds market between 2017-19. The sudden shift in dynamics between February and March 2018 is clearly related to the sharp increase in bill supply after the resolution of the federal debt limit episode that was resolved in early February 2018. Over the month after the resolution of the debt limit, repo rates rose abruptly and moved above EFFR, which I argue drove EFFR higher given the bargaining relationship between FHLBs and their borrowers in the fed funds market. The rise in EFFR subsequently caused borrowing by arbitrage banks in the fed funds market to fall by half over the next six months. This arbitrage demand appears to have been fully exhausted by the time borrowing costs reached 3 basis points below IORB, and the remaining sources of demand for fed funds borrowing do not appear to have been at all elastic. Thus, the demand curve for fed funds borrowing seems to have been no longer flat by the time that EFFR moved within 3 basis points of IORB in September 2018.

While the FOMC has not determined when it will stop balance sheet runoff or resume organic growth, to the extent that an ample-reserves regime is taken to be defined in terms of remaining on the flat portion of a demand curve for reserves borrowed in the fed funds market, the lessons from 2018-19 suggest that ending balance sheet runoff as EFFR begins to move closer to IORB and then resuming organic growth when rates are a few basis points below IORB could be viewed to be consistent with the Committee's intentions. Implementing policy such that EFFR was a few basis points lower than IORB would also be consistent with the underlying logic of the Friedman rule. The logic of that rule, as applied to reserves and IORB, would argue that reserves should be set at a level such that banks are not disincentivized from holding them. Given that there are balance sheet considerations that appear to make IORB arbitrage costly, the rule would call for EFFR to lie below IORB by the amount of those arbitrage costs. In the 2017-19 episode, these costs appeared to have been around 3 basis points, and demand for fed funds borrowing appeared to become substantially less elastic after EFFR moved higher than 3 basis points below IORB. Regardless of the FOMC's decision on when to stop balance sheet runoff, the previous experience suggests that it is reasonable to expect the dynamics in the fed funds market to change once repo rates move persistently above EFFR and when EFFR approaches the IORB.