When COVID-19 struck the United States in early 2020, the Federal Reserve responded with a force that dwarfed everything that had come before it. Within weeks, it slashed its benchmark interest rate to near zero and embarked on the most aggressive program of large-scale asset purchases, commonly known as quantitative easing, or QE, in its history. The result was a balance sheet that swelled from roughly $4 trillion on the eve of the pandemic to nearly $9 trillion by early 2022, a figure that would have been unimaginable to central bankers of a prior era.
Over the two and a half years that followed, the Fed reversed course, embarking on what economists call quantitative tightening (QT): a gradual, passive runoff of maturing securities designed to drain the balance sheet and withdraw the extraordinary monetary accommodation that had been put in place. By December 2025, when the Fed formally ended its second QT program, total assets had declined to approximately $6.6 trillion, still more than seven times the pre-2008 level, but representing a meaningful step back from the pandemic-era peak.
Understanding what happened to the balance sheet over this period, why it grew so fast, how it shrank, and what it all means for the economy, requires grappling with some of the most consequential and contested monetary policy decisions of the modern era.
The Long History of an Expanding Balance Sheet
To appreciate just how much the balance sheet has changed, it helps to look back at what it looked like before the 2008 financial crisis. As Chaitri Gulati and A. Lee Smith of the Federal Reserve Bank of Kansas City document in their authoritative 2022 review, the Fed’s asset holdings before the global financial crisis were modest and almost entirely composed of short-term Treasury securities. At the end of 2006, the Fed’s Treasury holdings amounted to less than 9 percent of the entire Treasury market, and roughly half of those holdings matured within a year. The balance sheet hovered around $800–900 billion, a figure primarily determined by the need to supply currency to a growing economy.
Policy in this era operated through a narrow channel: the Fed would target the federal funds rate, the overnight rate at which banks lend reserves to each other, by conducting small open-market operations to keep reserves near a desired level. The balance sheet itself was almost an afterthought—a plumbing mechanism to support the real instrument of policy.
The global financial crisis of 2007–09 shattered this quiet equilibrium. As the economy collapsed and the Fed cut rates to zero in December 2008, policymakers found themselves at what economists call the zero lower bound, unable to cut further without venturing into negative territory, a step U.S. policymakers were unwilling to take. The response was a radical reimagining of the balance sheet as a direct policy tool.
Through three rounds of asset purchases—QE1, QE2, and QE3—the Fed expanded its balance sheet from around $900 billion to approximately $4.5 trillion between 2008 and 2014. Crucially, the composition shifted as well. QE1 marked the first time the Fed entered the mortgage-backed securities (MBS) market, purchasing debt guaranteed by Fannie Mae, Freddie Mac, and Ginnie Mae. By the end of QE1, the Fed owned more than 20 percent of the agency MBS market—an extraordinary intervention in a specific credit market.
A further structural change came with the 2011–12 Maturity Extension Program (MEP), which sold shorter-dated Treasuries and used the proceeds to buy longer-dated ones, pushing the average maturity of the Fed’s Treasury portfolio from roughly three years to more than ten years without increasing the overall size of the balance sheet. This maturity extension was significant: it reflected the core theory behind how QE works, the so-called portfolio balance or duration channel.
The theoretical case for QE has always been contested. As the Richmond Fed’s Tim Sablik notes, former Fed Chair Ben Bernanke famously quipped that “the problem with QE is it works in practice, but it doesn’t work in theory”, a wry acknowledgment that standard models in which Treasuries and bank reserves are perfect substitutes would predict no effect from swapping one for the other.
In practice, however, economists have identified several channels through which QE operates. The portfolio balance channel argues that by removing long-duration assets from the market, the Fed reduces the supply of ‘duration’ available to private investors, bidding up the price of remaining long-dated bonds and pushing down their yields. Since borrowing costs for households and businesses are anchored to long-term rates, lower yields on 10-year Treasuries and MBS translate into cheaper mortgages, corporate loans, and other credit.
A second channel involves signaling: by committing to holding long-duration assets, the Fed signals its intention to keep short-term rates low for an extended period, reinforcing verbal guidance about future policy. A third channel is liquidity: in moments of acute financial stress, as in September 2008 and March 2020, the Fed’s willingness to absorb assets provides a backstop that prevents fire sales and restores market functioning.
The COVID-19 Era
Nothing in the Fed’s prior experience prepared it, or the markets, for the scale of the pandemic response. When COVID-19 hit in March 2020, the FOMC (Federal Open Market Committee) immediately cut rates back to zero and launched a new round of asset purchases. What followed was staggering in its velocity: purchases in April 2020 alone exceeded $1 trillion. By the time the Fed concluded net purchases in March 2022, cumulative pandemic-era purchases had exceeded $4.6 trillion, more than all three previous QE programs combined.
The balance sheet peaked at close to $9 trillion, representing roughly 37 percent of U.S. GDP, a level without precedent in the Federal Reserve’s history. The composition was similarly transformed. By March 2022, the Fed’s MBS holdings stood at more than $2.7 trillion, amounting to nearly 30 percent of all outstanding agency MBS securities in existence.
How much did this expansion of the balance sheet actually push down long-term interest rates? Gulati and Smith attempt to answer this with a careful meta-analysis of eight peer-reviewed studies on the relationship between the Fed’s asset holdings and the 10-year Treasury yield. Their median estimate, drawn from research spanning different time periods and methodologies, finds that every $100 billion in 10-year equivalents purchased by the Federal Reserve reduces the 10-year Treasury yield by approximately 4.5 basis points.
Applying this estimate to the pandemic-era balance sheet, Gulati and Smith conclude that the Fed’s expanded portfolio was depressing the 10-year Treasury yield by roughly 160 basis points—1.6 percentage points—as of early 2022. About half of that accommodation stemmed from the post-2020 purchases, with the remainder reflecting the still-unresolved legacy of the post-2008 QE programs. Wide uncertainty surrounds this estimate: the range across studies suggests downward pressure of anywhere from 130 to 270 basis points.
The Turn Toward Tightening (2022–2025)
With inflation surging to multi-decade highs in 2021–22, peaking at 8 percent in 2022, the highest rate since 1991, the Federal Reserve faced a dual challenge. It needed to raise short-term interest rates aggressively while simultaneously unwinding the enormous monetary stimulus baked into its balance sheet. The FOMC ended net asset purchases in March 2022 and commenced formal QT in June 2022, just three months later, a far more rapid pivot than the nearly three-year gap between ending QE3 in 2014 and beginning QT in 2017.
The mechanics of the new QT program closely followed the framework first used from 2017 to 2019. Rather than actively selling assets into the market, which could cause price disruptions and potentially undermine confidence in future QE programs, the FOMC chose to allow maturing securities to ‘roll off’ the balance sheet without being replaced, up to specified monthly caps. The fully phased-in caps were set at $60 billion per month in Treasuries and $35 billion per month in MBS, roughly double the pace of the 2017–19 episode.
The decision to rely on passive runoff rather than outright asset sales reflects a fundamental tension in balance sheet policy. As Gulati and Smith explain, for QE to be effective in moving long-term interest rates, market participants must believe the central bank will hold the assets it purchases for a significant period. If the Fed were to signal rapid asset sales shortly after purchases, markets would discount the effect of future QE programs, potentially undermining one of the most powerful tools the Fed has available for future crises.
This asymmetry between expansion and contraction is, in a sense, built into the design. The Fed can expand its balance sheet quickly to provide accommodation; unwinding that accommodation is necessarily slower to preserve the credibility and effectiveness of the tool. As the Richmond Fed’s article on QT notes, Philadelphia Fed President Patrick Harker captured this spirit in 2017 when he assured markets that balance sheet reduction would be ‘like watching paint dry’.
One of the most significant constraints on the pace of QT has been the behavior of mortgage-backed securities. Unlike Treasury bonds, which mature on fixed schedules, MBS prepay at rates that depend heavily on the interest rate environment. When mortgage rates are low, homeowners refinance frequently, accelerating MBS runoff. When mortgage rates rise, as they did sharply from 2022 onward, refinancing activity collapses, dramatically slowing the pace at which MBS leave the Fed’s balance sheet.
This created an uncomfortable reality for policymakers who had stated a preference for holding primarily Treasuries in the long run. Through 2022–25, MBS runoff consistently fell below the monthly caps, meaning the composition of the balance sheet shifted only slowly away from mortgage securities. The Fed’s $2.7 trillion MBS portfolio proved remarkably sticky even as the overall balance sheet shrank.
Between June 2022 and November 2025, the Federal Reserve allowed more than $2.2 trillion in securities to roll off its balance sheet without replacement: approximately $1.6 trillion in Treasury securities and $600 billion in agency MBS. As a share of GDP, the Fed’s securities holdings fell from 33 percent to 20 percent over this period, a decline of 13 percentage points.
Yet as Gulati and Smith had anticipated in their 2022 analysis, the duration-adjusted impact of this reduction was considerably smaller than the headline dollar figures might suggest. Because passive runoff allows only the shortest-dated Treasuries to mature first, the weighted-average maturity of the remaining Treasury portfolio actually increased during runoff, just as it had during the 2017–19 QT episode. This means the downward pressure on long-term interest rates from the balance sheet has receded only gradually, even as the total dollar value of assets has fallen substantially.
A Federal Reserve Board study referenced by both the Kansas City and Richmond Fed articles estimated that reducing the balance sheet by around $2.5 trillion over several years would be roughly equivalent to raising the policy rate by half a percentage point—a meaningful but relatively modest tightening contribution, especially against the backdrop of the most aggressive rate-hiking cycle since the 1980s.
Where Things Stand in Early 2026
At its October 2025 meeting, the Federal Open Market Committee announced that it would cease the runoff of its securities holdings effective December 1, 2025, formally ending the second QT program in the Fed’s history. The decision reflected both the progress made in reducing the balance sheet and the Fed’s desire to avoid draining reserves below the level consistent with smooth monetary policy implementation.
Going forward, the Fed announced it would reinvest all maturing Treasury securities into new Treasury issuance to maintain a roughly stable balance sheet, while redirecting MBS principal payments into shorter-dated Treasury bills, allowing the mortgage portfolio to decline gradually over time without actively disrupting the MBS market.
As of mid-March 2026, the Fed’s total assets stand at approximately $6.6 trillion, down from the peak of nearly $9 trillion in 2022, but still representing a balance sheet that is more than seven times its pre-2008 size. The composition continues to reflect the legacy of the COVID-era purchases: Treasury securities remain the dominant asset class, but a large MBS portfolio persists, with the Fed still holding a significant share of the agency mortgage market.
One notable development on the liabilities side has been the accumulation of what the Fed calls a ‘deferred asset’, essentially a negative liability representing accumulated losses. From September 2022 through late 2025, the Fed’s net income was negative, because the interest rates it pays on bank reserves and reverse repurchase agreements rose sharply with rate hikes while the yield on its legacy low-rate assets remained largely fixed. As of March 2025, this deferred asset stood at $225 billion. This does not affect the Fed’s ability to operate, unlike a commercial bank, the Fed cannot become insolvent, but it means the Fed will not resume remitting profits to the Treasury until future earnings offset the losses.
Works Cited
Gulati, Chaitri, and A. Lee Smith. “The Evolving Role of the Fed’s Balance Sheet: Effects and Challenges.” Federal Reserve Bank of Kansas City Economic Review, Q4 2022.
Sablik, Tim. “The Fed Is Shrinking Its Balance Sheet. What Does That Mean?” Federal Reserve Bank of Richmond Econ Focus, Q3 2022.
Federal Reserve Board. “Policy Normalization.” Updated October 2025. federalreserve.gov.
Congressional Research Service. “The Federal Reserve’s Balance Sheet.” IF12147, 2025. congress.gov.
Plosser, Charles I. “A Skeptical View of the Impact of the Fed’s Balance Sheet.” NBER Working Paper No. 24687, 2018.
Cleveland Fed. “QT, Ample Reserves, and the Changing Fed Balance Sheet.” Economic Commentary, April 2025.
American Action Forum. “Tracker: The Federal Reserve’s Balance Sheet Assets.” March 2026.