What Happens on Quarter-Ends in the Repo Market, Accessible Data

Figure 1. Spread of SOFR over the ON RRP Rate

The x-axis of this chart starts in late June 2022 and ends shortly after the beginning of 2025. The y-axis denotes the spread in basis points between the Secured Overnight Financing Rate (SOFR) and the Overnight Reverse Repurchase Agreement (ON RRP) rate. The solid blue line shows the daily SOFR-ON RRP spread over time. The dashed grey line represents a spread of 0 basis points. There are eight vertical grey bars, which represent two-day windows around quarter-ends starting in 2023. Prior to November 2022, the blue line is consistently negative or zero, indicating that SOFR was less than or equal to the ON RRP rate over this period. Starting in late November 2022 through the end of 2023 Q3, the blue line occasionally rises to levels of one to two basis points outside of quarter-ends and rises to levels between three and seven basis points during quarter-ends, but is otherwise is at zero. Starting in 2023 Q4 through the end of 2024 Q2, the SOFR-ON RRP spread begins to exhibit more volatility, ranging from zero to nine basis points outside of quarter-ends and ranging from three to ten basis points during quarter-ends; outside of quarter-ends, the spread is 1.56 basis points on average. In the last two quarters of 2024, the spread continues to fluctuate as upward pressrue on repo rates continues, with non-quarter-end values at 4.23 basis points on average. On quarter-ends, the spread reaches peaks of 25 and 24 basis points; the spread reaches a peak of 28 basis points on December 26, 2024.

Notes: Quarter-end periods are +/- 2 business days from quarter-end.

Sources Federal Reserve Bank of New York

Return to text

Figure 2. Month-End Changes in SOFR-ON RRP Spread

The x-axis of this chart starts at the beginning of 2017 and ends at the end of 2024. The light blue bars represent the month-end change in the SOFR-ON RRP spread and the dark blue line represents the average SOFR-ON RRP spread over the month (excluding the spread on month-end). The month-end change in the SOFR-ON RRP spread tends to move in line with the average spread over the rest of the month. The average SOFR-ON RRP spread trended up from 2017-2019, then declined to near-zero starting in 2020, as the Federal Reserve expanded the size of the balance sheet. In 2023 and 2024, as the Fed has reduced the size of its balance sheet, average spreads and month-end changes in spreads have increased.

Notes: Average SOFR-ON RRP spread is the average daily SOFR-ON RRP spread over the month, excluding the last business day of the month. Month-end change in SOFR-ON RRP spread equals the SOFR-ON RRP spread on the last business day of the month minus the average SOFR-ON RRP spread.

Sources Federal Reserve Bank of New York

Return to text

Figure 3. Behavior of Repo Markets around Recent Quarter-Ends

A: Triparty Borrowing. The x-axis of this chart spans from -4 to 3, representing four days prior to three days after quarter-end. The dashed blue line represents total triparty borrowing and the solid orange line represents foreign triparty borrowing; all values are relative to four days prior to quarter-end. The y-axis units are in billions of dollars and span from -50 to 20. The chart shows that on quarter-ends, total and foreign triparty borrowing volumes declined by an average of approximately $45 billion relative to four days prior to quarter-end. Three days after quarter-end, total triparty borrowing volumes increased by an average of $10 billion relative to four days prior to quarter-end, and foreign triparty borrowing retraced to volumes similar to four days prior to quarter-end.

B: FICC Activity. The x-axis of this chart spans from -4 to 3, representing four days prior to three days after quarter-end. The dashed blue line represents sponsored repo volumes in FICC DVP, the solid orange line represents sponsored reverse repo volumes in FICC DVP, and the green dot-dashed line represents interdealer trading in FICC DVP; all values are relative to four days prior to quarter-end. The y-axis units are both in billions of dollars; the left scale spans -20 to 180 and the right scale spans -10 to 70. Sponsored reverse repo is plotted on the left scale; sponsored repo and interdealer trading are plotted on the right scale. The chart shows that all three lines increase over the quarter-end period and have values higher on three days after quarter-end relative to three-days prior to quarter end. On quarter-end, sponsored reverse repo volumes are on average $140 billion more than they were four days prior to quarter-end, and retrace to $20 billion by three days after quarter-end. On quarter-end, sponsored repo volumes were on average $50 billion higher than they were four days prior; on three days after quarter-end, they fell to $14 billion higher on average than volumes four days prior to quarter-end. Average interdealer trading relative to four days prior to quarter-end reached a peak of $63 billion one day after quarter-end, then fell to $12 billion by three days after quarter end; on quarter-end, average interdealer volumes were $22 billion more than they were four days prior.

C: Use of ON RRP Facility. The x-axis of this chart spans from -4 to 3, representing four days prior to three days after quarter-end. The solid blue line represents total takeup at the Federal Reserve's Overnight Reverse Repurchase Facility relative to four days prior to quarter-end. The y-axis units are in billions of dollars and span from -75 to 150. The chart shows that leading up to quarter-ends, takeup at the facility increases, reaching an average of $150 billion dollars more than takeup on four days prior to quarter end. On the day after quarter-end, takeup at the facility quickly retraces to $25 billion lower than the takeup four days prior to quarter end. Three days after quarter end, average takeup stands at $50 billion less than takeup on four days prior to quarter end.

D: Lending and Borrowing Rates. The x-axis of this chart spans from -4 to 3, representing four days prior to three days after quarter-end. The dashed blue line represents sponsored repo rates in FICC DVP and the solid orange line represents sponsored reverse repo rates in FICC DVP; all values are relative to four days prior to quarter-end. The y-axis units are in basis points and span from zero to ten. On quarter-ends, sponsored repo and sponsored reverse repo rates are on average 1.7 and 6.4 basis points higher than rates four days prior to quarter-end, respectively. Both reach relative peaks one day after quarter-end, with averages of 2.1 and 8.6 basis points. By three days after quarter-end, sponsored repo has returned to levels seen on four days prior to quarter-end, and sponsored reverse repo is on average 1.9 basis points higher.

E: GSIB NCCB Repo. The x-axis of this chart spans from -4 to 3, representing four days prior to three days after quarter-end. The dashed blue line represents non-centrally cleared bilateral repo by GSIBs and the orange line represents non-centrally cleared bilateral reverse repo by GSIBs. The y-axis units are both in billions of dollars; the left scale spans 680 to 730 and the right scale spans 400 to 450. Reverse repo is plotted on the left scale and repo is plotted on the right scale. The chart shows that NCCB reverse repo by GSIBs declined leading into quarter-ends and returned to pre-quarter-end volumes three days after quarter-end. On average, NCCB reverse repo volumes were $718 billion on four days prior to quarter-end, $688 billion on quarter-end, and $720 billion three days after quarter-end. The chart also shows that NCCB repo by GSIBs increased leading into quarter-ends and remained elevated in the days immediately following. On average, NCCB repo volumes were $412 billion on four days prior to quarter-end, $442 to $443 billion on quarter-end and the following two days, and $431 billion three days after quarter-end.

F: GSIB Net Cash Lending. The x-axis of this chart spans from -4 to 3, representing four days prior to three days after quarter-end. The solid blue line represents GSIB net cash lending (reverse repo - repo) relative to four days prior to quarter-end. The y-axis units are in billions of dollars and the scale spans from -5 to 40. GSIB net cash lending increased leading into quarter-end, with a peak average of $34 billion more on quarter-end than four days prior to quarter-end. Following quarter-ends, relative volumes remained slightly elevated, with an average of $6 billion three days after quarter-end.

Notes: Plotted data reflect averages of values on and around quarter-ends from 2023 Q1 to 2024 Q3. Values are relative to average levels 4 days prior to quarter-end (except for Panel E).

Sources Bank of New York Mellon; Office of Financial Research; Federal Reserve Board, Form FR 2052a; Federal Reserve Bank of New York.

Return to text

Figure 4. Change in Foreign Triparty Borrowing on Recent Quarter-Ends

The figure is a single-line chart depicting the change in borrowing volume by foreign entities on the Triparty platform on quarter ends with respect to four days prior. The x-axis of this chart starts at the beginning of 2021 and ends at the end of 2024, with intervals of quarters. The y-axis is in billions of dollars. The blue line denotes the change in borrowing at every quarter-end. The dashed grey line is at -$30 billion and represents the change in 2021 Q1, the beginning of the figure. The blue line starts at -$30 billion and stays close to -$30 billion until 2021Q4, where the change increases to -$10 billion. The blue line then returns to -$30 billion for most of 2022, except for a dip to -$45 billion on 2022 Q3. Starting 2023, there average borrowing decreases to around -$45, reaching -$60 billion in 2023 Q3. Aside from a spike back to -$30 billion on 2023 Q4, the change stays around -$50 billion until 2024 Q3, where the blue line goes back and stays at -$30 billion.

Notes: Values shown reflect the difference between foreign triparty borrowing volumes on quarter−end and foreign triparty borrowing volumes four days prior to quarter−end.

Sources Bank of New York Mellon; Author Calculations.

Return to text

Figure 5. Changes in Sponsored Reverse Repo on Recent Quarter−Ends

The figure is a bar chart depicting the change in sponsored reverse repo volumes in FICC DVP on quarter-ends with respect to volumes four days prior. The x-axis starts at the beginning of 2021 and ends at the end of 2024, with a bar for every quarter. The y-axis is in billions of dollars, with 0 at the minimum and 300 at the maximum. The blue bars denote the change in sponsored reverse repo volumes at every quarter end. Between 2021 to 2023, the change in reverse repo volumes increases on around $5 to $10 billion every quarter end on average, starting at $30 billion on 2021 Q1 and ending around $100 billion on 2023 Q4. Starting 2024 Q1, the increase in volume dramatical increases, with around $130 billion on 2024 Q1, $160 billion on 2024 Q2, $300 billion on 2024 Q3, and $250 billion on 2024 Q4.

Notes: Values shown reflect the difference between sponsored reverse repo volumes in FICC DVP on quarter−end and sponsored reverse repo volumes in FICC DVP four days prior to quarter−end.

Sources Office of Financial Research; Author Calculations.

Return to text

Figure 6. Primary Dealer NCCB and Sponsored Reverse Repo

The figure is a single-line chart depicting the sum of primary dealer non-centrally cleared bilateral (NCCB) reverse repo and sponsored reverse repo volumes through FICC DVP. The x-axis starts at the beginning of 2017 and ends at the end of 2024. Data is on the weekly level. The-y-axis is in billions of dollars with a minimum of 800 and maximum of 2400. The blue line represents the amount of NCCB and sponsored reverse repo performed by primary dealers. Between 2017 and 2021, volumes are relatively stable around $1200 billion. Throughout 2017, volumes stay around $1100 billion, but start dipping toward $900 billion throughout 2018. By end of 2018, volumes recover to $1100 billion and increase steadily to $1400 billion by end of 2019. In the beginning of 2020, there is a sudden spike to $1700 billion and a quick reversal back to $1200 billion. Volumes stay around $1200 billion for the remainder of 2020 and decreases and settles around $1100 billion in 2021. Starting 2022, the trend changes and volumes begin growing rapidly. Through 2022, volumes start at $1100 billion and grow steadily to $1400 billion by the end of 2022. In 2023, volumes continue to grow steadily to $1700 billion. Finally in 2024, volumes grow to $2400 billion by year end.

Notes: The Primary Dealer Statistics report NCCB repo and reverse repo positions beginning in 2022. Prior to 2022, NCCB reverse repo is calculated based on total reverse repo positions for primary dealers (as reported in the Primary Dealer Statistics) less their triparty and centrally−cleared reverse repo positions (which are based on non−public data). Primary Dealer NCCB and Sponsored Reverse Repo both include U.S. Treasury and TIPS collateral.

Sources Primary Dealer Statistics; Bank of New York Mellon; Author Calculations.

Return to text

Figure 7. Measure of Net Supply−Demand and Repo Market Conditions

The figure is a combination of a single-line and two bars. The black line depicts the net supply-demand measure, the light-blue bar depicts the month-end change in SOFR-ON RRP spread, and the dark-blue bar depicts the average SOFR -ON RRP spread. The x-axis starts at the beginning of 2017 and ends in December 2024. All three series are on the weekly level. The black line, the net supply-demand measure, uses a y-axis starting at $500 billion and ending at $4500 billion. The bar charts, the month-end change and average SOFR-ON RRP spreads, use a y-axis starting at -20 basis points to 60 basis points (bps). The black line starts at and is stable around $1500 billion through 2017, but steadily decreases to $700 billion by the tail-end of 2019. Volumes then increase up to $2000 billion over the first two months of 2020, and settle around $1800 billion by the end of 2020. Volumes further increase to $4000 billion over the first half of 2021, where it settles through the middle of 2023. Starting the second half of 2023, volumes steadily decrease back to $1800 billion by end of 2024. The light-blue bar stays within 5 to 10 bps in 2017 and becomes more volatile through 2018 where it bounces around 0 and 20 bps. At the year-end of 2018, the spread spikes up to 50 bps, corresponding with the beginning of the decrease in the net supply-demand measure. The spread returns to an average of 15 bps through the first ten months of 2019, but with greater volatility - a minimum of 5 bps and maximum of 30 bps. In the last two months, the month-end change shrinks to near 0 bps. In the beginning of 2020, the spread temporarily becomes -10, corresponding to the increase in net supply-demand measure over the first two months. The spread then remains near 0 bps through mid-2023, where the spreads has widened to 5 bps on average through mid-2024. Finally, spreads start showing volatility again, bounding between 5 and 15 bps in the latter half of 2024, corresponding with the lower net supply-demand measure. The dark-blue bar mimics the light-blue bar, staying between 5 to 10 bps in 2017, increasing to between 10 and 20 bps in 2018, and further to 20 bps and 40 bps in the first half of 2019. Spreads stabilize to around 10 bps by late 2019, where it remains throughout 2020. Spreads decrease to near 0 and remain throughout middle of 2024, where it increases to between 5 and 10 bps by end of 2024.

Notes: Average SOFR−ON RRP spread is the average daily SOFR−ON RRP spread over the month, excluding the last business day of the month. Month−end change in SOFR−ON RRP spread equals the SOFR−ON RRP spread on the last business day of the month minus the average SOFR−ON RRP spread. Net Supply−Demand measure equals the volume at the ON RRP plus reserves held by large banks plus triparty borrowing by primary dealers, minus primary dealer non−centrally cleared bilateral repo and all sponsored reverse repo.

Sources Federal Reserve Bank of New York; Primary Dealer Statistics; Bank of New York Mellon; Federal Reserve Board H.8; Author Calculations.

Return to text

Last Update: June 06, 2025