Banks’ reserve demand and the limits of QT
The paper by David Lopez-Salido, Annette Vissing-Jorgensen develops a framework for understanding banks’ reserve demand and uses it to guide Fed policy on interest rate control and quantitative tightening. The authors model a “convenience yield” i.e. the extra value of reserves beyond interest earnings, that declines as reserves increase but rises with liquid deposits. The spread between the short market rate (EFFR) and the interest rate on reserves (IOR) spread naturally measures reserve scarcity: positive spreads indicate banks value liquidity highly, while negative spreads show reserves are abundant and balance sheet costs dominate.
Estimation methodology
- Data: Monthly U.S. data from January 2009 to March 2025
- Key innovation: Models EFFR-IOR spread as depending on log reserves and log liquid deposits, with ON RRP facility creating a lower bound on the spread
Findings
- Stability: Reserve demand is remarkably stable (R²=0.97) once liquid deposits are controlled for. This finding also explains why the Fed and market participants have struggled to predict when reserve scarcity would emerge during balance sheet normalization.
- Elasticities: The coefficient on log reserves of negative 0.082 means that a 10% reduction in reserve supply increases the EFFR-IOR spread by approximately 0.8 basis points, holding deposits constant. To put this in practical terms: if reserves fall from 3 trillion to 2.7 trillion (a 10% decline), banks would bid up the federal funds rate by about 0.8bp relative to the IOR. Moving from very abundant to moderately abundant reserves has limited impact, but as reserves approach scarcity, the convex shape of the demand curve means further reductions have increasingly large effects on rates.
- Deposit-adjusted reserves: The tight relationship between deposit-adjusted reserves and the EFFR-IOR spread holds consistently throughout the entire sample period, whereas the relationship between raw reserves and the spread breaks down repeatedly. For instance, in 2020-2021, reserves increased substantially but so did deposits—meaning deposit-adjusted reserves actually didn’t change as much as raw reserves would suggest.
QT guidance
- September 2019 serves as a critical reference point because it represents the only modern episode where reserve scarcity led to observable market dysfunction. The authors’ model predicts that the EFFR-IOR spread in September 2019 was approximately 4 basis points. The paper uses this 4bp predicted spread as a “warning threshold,” not necessarily the point where problems definitely occur, but rather where the system becomes vulnerable to volatility if faced with typical demand or supply shocks.
- Using the estimated reserve demand function and March 2025 liquid deposit levels, the authors calculate that reserves plus ON RRP balances would need to fall to approximately 2.29 trillion to generate the same 4bp predicted spread observed in September 2019. However, the authors emphasize this should be viewed as a zone to approach cautiously rather than a cliff edge.
- The authors note that the Standing Repo Facility introduced in July 2021 fundamentally improves the framework relative to September 2019, potentially allowing more QT for a given level of volatility risk
Conclusion: Reserve scarcity depends on both supply AND deposits—same reserve level means different scarcity as deposits change.
Reserve Demand, Interest Rate Control, and Quantitative Tightening
Authors: David Lopez-Salido, Annette Vissing-Jorgensen
From: Federal Reserve Board
Managing TGA volatility
The paper by Annette Vissing-Jorgensen proposes that the Fed should back the Treasury General Account (TGA) with Treasury bills rather than use the current ample reserves approach to manage volatile government deposits. Since 2015, the TGA has been large ($850B target) and volatile due to Treasury cash management and debt ceiling dynamics. The Fed’s current ample reserves policy, letting reserves fluctuate passively with TGA changes, creates three problems: (1) uncertainty about reserve adequacy risks September 2019-style rate spikes, (2) debt ceiling episodes complicate quantitative tightening by causing reserves to rise temporarily then “snapback,” potentially overshooting QT targets, and (3) TGA drawdowns during debt ceilings create Treasury bill scarcity, distorting yields by 100+ basis points in 2023.
Vissing-Jorgensen evaluates four Fed approaches to TGA fluctuations using three criteria: interest rate control, maintaining consistent policy stance, and communication clarity. She uses daily Treasury data (2007-2025) to document TGA drivers and analyzes the March 2025 FOMC decision to slow QT using reserve demand estimates showing 2.29T reserves+ON RRP would replicate 2019 scarcity.
Solution: Back TGA with bills
The Fed would hold Treasury bills equal to the TGA amount, adjusting bill holdings when the TGA changes by letting bills roll off during drawdowns or purchasing bills during rebuilds. This would require increasing Fed bill holdings from around $200B to $850B.
Key advantages
- Stable reserves: Reserves remain constant regardless of TGA movements, reducing volatility risk or enabling smaller average balance sheet
- Smooth QT: Balance sheet reduction of notes/bonds/MBS can proceed unaffected by debt ceiling episodes, avoiding unintended reserve scarcity
- Reduced bill scarcity: Fed bill sales during TGA drawdowns free up safe bills for private holders during debt ceilings
The approach effectively separates the Fed’s government banking function from monetary policy implementation.
Fluctuations in the Treasury General Account and their effect on the Fed’s balance sheet
Authors: Annette Vissing-Jorgensen
From: Federal Reserve Board
Quantitative easing and tightening
The Federal Reserve’s Balance Sheet
Author: Marc Labonte
From: Congressional Research Service
