Dr. Darrell Duffie on Liquidity Strains at Year-End/Quarter-End and When Fed Reserves Will No Longer Be Ample

Episode Summary

Executive Summary: Daryl Duffie explains that the widening SOFR-IOR spread at quarter- and year-end is driven mainly by bank capital constraints, especially foreign-bank window dressing and U.S. GSIB year-end rules, not yet by reserve scarcity. He says QT will likely end next year as reserves approach a more uncomfortable level, and discusses how Treasury bill/coupon issuance, the reverse repo facility, and bank credit-line dynamics all affect repo market stress and monetary policy transmission.

Main Topics: Quarter-end SOFR-IOR dislocations (Priority: 5/5): Duffie says the SOFR-IOR spread widens at quarter- and year-end because banks, especially foreign banks, shrink balance sheets to satisfy capital rules and window dress, pushing repo funding costs higher. Reserves versus capital requirements (Priority: 5/5): He distinguishes reserve scarcity from capital constraints: current stress is mostly about capital and GSIB balance-sheet management, while reserve scarcity is a later and separate risk. Why reserves are structurally more important post-GFC (Priority: 4/5): Post-crisis regulation, interest on reserves, and the reduced use of daylight overdrafts mean banks now need far larger reserve buffers than before 2008. Signals of approaching reserve scarcity (Priority: 5/5): Duffie says SOFR/IOR spreads are useful but can jump suddenly; he favors a timing metric based on when major repo dealers receive incoming payments as a better canary in the coal mine. Treasury issuance, bill supply, and repo pressure (Priority: 4/5): He explains how Treasury bill scarcity, coupon issuance, and large settlement flows affect repo markets by changing demand for financing and draining reserves from the banking system. Quantitative tightening and Fed balance-sheet buffers (Priority: 4/5): QT is expected to slow or stop in 2025 as the overnight reverse repo facility nears zero and reserve balances become the main remaining buffer against balance-sheet runoff. Reference rates and bank funding risk (Priority: 3/5): Duffie summarizes his recent work showing SOFR-based credit lines can create funding stress for banks in crises because borrowers draw heavily when risk-free rates fall.

Key Arguments: Quarter-end spreads reflect capital regulation more than a collapse in market functioning; foreign banks window-dress and U.S. GSIBs also constrain balance sheets at year-end. Reserve balances are still abundant relative to the 2019 shortage, but the system is closer to a point where QT will need to pause. Post-GFC rules and paid reserves changed banks’ incentives: they now hold more reserves and are less willing to use the Fed intraday, making liquidity management more reserve-intensive. SOFR-IOR spreads are informative but not sufficient as an early-warning indicator because the spread can worsen abruptly overnight. A better warning signal is the lateness of interbank payments in the repo market; if payments are consistently received much later in the day, liquidity stress is building. Treasury bill supply, coupon issuance, and Treasury cash management all interact with repo rates and reserve balances, affecting monetary policy transmission and market rates. The reverse repo facility has largely done its job as a QT buffer and is now close to exhausted, so additional balance-sheet runoff will increasingly come out of reserves. SOFR was a good replacement for LIBOR because it is robust and hard to manipulate, but SOFR-based lending shifts crisis risk toward large borrowing draws on bank credit lines.

Data Points: Normal quarter-end SOFR-IOR spread increase: 7 to 9 basis points - Duffie cites his research on typical quarter-end widening Current average SOFR-IOR spread: 20 basis points - Based on NY Fed website data at the time of the interview SOFR-IOR 99th percentile spread: 45 basis points - Current tail of repo funding stress September quarter-end overnight financings above SRF: about $600 billion - Repo transactions done above the Fed’s standing repo facility rate 75th percentile funding rate above SRF in September: about 15 basis points higher - Shows meaningful stress even away from the tail 99th percentile funding rate above SRF in September: about 40 to 50 basis points higher - Tail-rate transactions for highly stressed dealers Reserves before GFC: $40 billion to $50 billion - Tiny reserve balances in the pre-2008 system Current reserve balances: about $3.2 trillion - Reserve balances in the post-QE/QT era Peak reserve balances: over $4 trillion - Peak reached in fall 2021 2019 reserve shortfall threshold: around $1.5 trillion - September 2019 episode where reserves proved insufficient Current reverse repo facility balance: a little over $100 billion - Near exhaustion of the Fed’s QT buffer Reverse repo facility peak: more than $2.5 trillion - Buffer was much larger over a year earlier TGA/debt ceiling volatility: few months ahead - Treasury General Account may add volatility after the debt ceiling extension Treasury debt outstanding: about $28 trillion - Current stock of Treasury securities CBO projected Treasury debt in 10 years: $40.3 trillion - Mid-trajectory projection cited in the discussion Debt outstanding pre-GFC: $7 trillion - Used to illustrate the long-run rise in U.S. Treasury debt Debt-to-GDP: around 100% now; projected about 150% by mid-century - Illustrates fiscal pressure and crowding out concerns Treasury issuance size affecting repo: $60 to $75 billion - Large issuance days are cited as a major repo-market drain 2023 Treasury bill rates: over 6% to 7% - Referenced in the debt-ceiling stress episode Fed funds target range: 4.25% to 4.50% - Used to compare with SOFR and market rates SOFR example during interview: 4.70% - Compared to effective fed funds and IOR in the conversation Effective fed funds example: 4.33% - Illustrates SOFR trading above the policy midpoint Payment-timing signal shift: more than 100 minutes later than normal - Key sign in Duffie’s canary metric when reserves get tight 2025 QT expectation: around end of Q1 or between Q1 and Q2 - Market speculation on when the Fed may slow QT Corporate funding source mix: credit lines larger than term loans - Result from Duffie’s bank-funding-risk paper March 2020 line draws: huge precautionary draws - Crisis behavior after COVID shock, when firms drew on credit lines

Pivotal Quotes: "The sky is not falling, but this isn't a nothing burger." — Daryl Duffie: He characterizes quarter-end repo stress as real but not systemically alarming "Once the Fed expands its balance sheet and provides an enormous quantity of reserves to the banks ... the banks get somewhat addicted to having a lot of reserve balances." — Daryl Duffie: Explaining the ratchet effect in reserve demand after QE "It's a very complicated interplay between all these forces." — Daryl Duffie: Summarizing how Treasury issuance, bills, coupons, reserves, and repo market dynamics interact

Implications: Listeners should expect quarter-end repo stress to persist, but the bigger medium-term risk is QT running too far as reserves and the reverse repo buffer shrink. Bank funding, Treasury issuance, and Fed policy are tightly linked, so market frictions can reappear quickly.