Focus
March 06, 2026 | 13:33
Fed’s Balance Sheet: Big but Just Enough
Economics
Economics
Fixed Income Strategy
Economics
Economics
Economics
Economics
Economics
Economics
Economics
Fixed Income Strategy
Economics
Economics
Economics
Economics
Economics
Fed’s Balance Sheet: Big but Just EnoughShort of banks’ liquidity requirements being relaxed and/or interest paid on reserves being eliminated, the Fed’s ‘big’ balance sheet is already as ‘small’ as it’s going to get. $6.6 trillion and counting |
|
The Fed’s balance sheet is currently $6.6 trillion (as of March 4), as measured by total assets (Chart 1). This is down $2.3 trillion or 26% from the record high of $9.0 trillion during April 2022. Both figures are well above the pre-COVID peak of $4.5 trillion. If the balance sheet had continued growing at the (exponential) trend in place before the Global Financial Crisis (GFC… with its alphabet soup of asset-boosting liquidity and credit measures along with three rounds of quantitative easing (QE)), total assets would currently be about $3 trillion. (If the ratio to GDP was the same as it was in early 2008, the balance sheet would currently be pushing $2 trillion.) These figures beg the question: Is the Fed’s balance sheet still too big? President Trump’s nominee for Fed Chair, Kevin Warsh, believes so. As a Fed governor (February 2006 to March 2011), he was considered a policy ‘hawk’ and was highly critical of QE. Indeed, it is reported that the disagreement over QE was a reason Warsh left the Fed. According to the Wall Street Journal (and Greg Ip), Warsh “blames Fed bond-buying for encouraging the federal government to run steep deficits, suppressing market signals and ultimately unleashing inflation.” Warsh and other critics of the balance sheet’s size may seize upon the fact that core PCE inflation peaked at four-decade highs around the time the Fed’s balance was peaking, with total assets roughly doubling from their pre-COVID level (Chart 2). However, tempering a causality conclusion is the fact that the Fed struggled for more than a dozen years to push core inflation up to 2% and keep it there while total assets roughly quintupled from their pre-GFC level. |
|
The major drivers of the asset- or left-side of the ledger are the Fed’s holdings of Treasuries and mortgage-backed securities (MBS) (Chart 3). The latest amounts are $4.3 trillion for Treasuries and $2.0 trillion for MBS, for a combined $6.3 trillion. The other (occasional) major drivers are the various liquidity and credit facilities designed to address market dislocations and crises. They surge and then they subside. Note that MBS holdings are being allowed to shrink, up to $35 billion per month (the QT program’s peak run-off cap which tends not to be binding), reflecting the prepayments generated by refinancings and sales. To shrink these more quickly would require outright sales, which could cause dislocations in the mortgage market at a time when housing affordability is a major issue. |
Treasury holdings, in contrast, are no longer shrinking (as of December 1, 2025). In fact, net purchases have begun again (as of December 10), to offset the reduction in total assets caused by continued MBS run-off and to put total assets on a modest growth path over time. (Since December 10, the Fed has purchased a net $148 billion of Treasuries, all T-bills, with the balance sheet increasing by $90 billion.) While critics complain about the balance sheet’s size, the Fed has already begun growing it (modestly) again. To understand why, one must turn to the liabilities- or right-side of the ledger (and recall that changes in assets are mirrored in changes in liabilities, and vice versa). The major Fed liabilities are, in order (with their current levels), reserves ($3.0 trillion), Federal Reserve notes ($2.4 trillion), Treasury’s general account balance ($847 billion), and reverse repos ($320 billion) (Chart 4). The latter was the second-largest liability item at one point, approaching $2.7 trillion, after the reverse repo facility was expanded to include money market funds to help absorb excess market liquidity outside the banking system. Indeed, reverse repos with money funds became a partial substitute for bank reserves. These entities could do reverse repos with banks, which would add amounts to banks’ reserve accounts as the transactions settled, or the money funds could deal directly with the Fed. The Fed began tracking the combination of reserves and reverse repos as a broader measure of liquidity. Currently, the reverse repo amount is only transactions with the Fed’s (regular) foreign official and international accounts. |
|
Elsewhere, Treasury targets a general account balance in the $850-to-$900 billion range, but the amount can bounce around a lot. And Federal Reserve notes are supplied at a pace that accommodates the growth in the domestic and international demand for U.S. dollars in physical form (currently 3.3% y/y). Finally, reserves increase when assets increase (e.g., the Fed’s securities purchases being ‘paid for’ by crediting banks’ reserve accounts) or when other liabilities decrease (e.g., tax refunds being debited from Treasury’s account and credited to banks’ reserve accounts as the amounts settle, or reverse repos with money market funds plummeting from $2.4 trillion to near zero). Reserves decrease when assets decrease (e.g., during QT) or when other liabilities increase (e.g., new Federal Reserve notes being distributed by banks and debited from their reserve accounts). Reserves peaked at $4.3 trillion during December 2021 and have since fallen by $1.3 trillion, or 29%. The key question for the Fed is whether reserves (and thus total assets and the balance sheet) can be reduced further without causing dislocations in the overnight money market. The risk of this happening escalates as the supply of reserves approaches and potentially slips below the demand for reserves. Amid the GFC and in the years that followed, banks’ demand for reserves increased immensely, reflecting a crop of acronyms (such as IORB, LCR, and HQLA). Unfortunately, demand cannot be measured precisely. IORB: The Fed pays interest on reserve balances (IORB) as of October 2008. However, IORB and its predecessors IORR (interest on required reserves) and IOER (interest in excess reserves) were not children of the GFC. Congress passed legislation authorizing the Fed to pay interest in October 2006, to begin in 2011. The GFC expedited the timeline. (Note: The Fed eliminated reserve requirements in March 2020.) A key rationale for the 2006 legislation was that when banks received no interest, they kept reserve balances that covered required amounts and those needed for settlement of payments at the smallest level prudentially possible. As payments expanded over time, this contributed to volatility in the federal funds rate (the Fed’s targeted policy rate). The volatility, in turn, forced the Fed to intervene often in the money market to add or withdraw reserves and keep the fed funds rate at its point target (at the time). Paying interest would lessen the volatility and interventions as banks held higher levels of reserves than before. And beginning in 2008, this became a critical tool for the Fed in ensuring that the massive amount of reserves being created by QE and other liquidity/credit facilities did not undermine the central bank’s ability to control the policy rate, with the establishment of the overnight reverse repo facility further helping. Then, post-GFC regulatory changes designed to ensure that banks held sufficient liquidity to brave another financial crisis (the lack of liquidity being a key catalyst of the GFC) caused banks’ demand for reserves to increase immensely. LFC and HQLA: The Liquidity Coverage Ratio (LCR) requires banks (mostly large ones) to have enough liquidity on board to cover at least 100% of their funding requirements in the coming 30 days (to survive a 30-day stress scenario). The liquidity must be parked in High-Quality Liquid Assets (HQLA). There are limitations (and up to 50% haircuts) on certain assets being included. However, there are no limits or haircuts on (Level 1) assets such as cash, reserves, or Treasuries. An individual bank’s demand for reserves will reflect a myriad of drivers: expected settlement of payments plus a cushion for settlement surprises, the desired reserves share of Level 1 assets, the desired Level 1 assets share of HQLA, and the desired LCR (how much above the 100% minimum to maintain). Nearly all these factors are influenced by idiosyncratic risk management practices. The total demand for reserves is thus the aggregation of these (private and unpublished) individual demands, making the all-bank figure a murky metric. This became problematic for the Fed during the first episode of QT and the steady reduction of reserves. At what point would the consequent lower supply of reserves butt against the elevated (and still likely rising) murky demand for reserves? By the late summer of 2019, the question was answered (Chart 5). Banks’ reserve balances had already fallen to levels close to their apparent aggregate demand, if not below. Banks started becoming reluctant to engage in overnight lending because it further sapped their liquidity. This caused the complex of overnight rates to become more volatile and then spike. In reaction, the Fed ended QT earlier than planned and ramped up its repos with banks and outright purchases of T-bills to rebuild reserves (and start growing the balance sheet, moderately, again). The bill purchases continued until the pandemic shifted the reserves building gear into overdrive. QT2 ends at year-end |
The Fed took the lessons learned from QT1 into QT2. The gameplan was to stop QT and reserves reduction well before the point where 2019-type dislocations could occur. The latest episode started in June 2022 and was ramped up in September 2022 to monthly roll-off caps of $60 billion for Treasures and $35 billion for MBS. The former was tapered to $25 billion in April 2024 and to just $5 billion in April 2025 as pressures in the overnight money market started emerging. They became more pronounced (and sustained) as the autumn approached. The Fed announced the December 1 end of QT at its October 29 meeting. It is noteworthy that the same day the Fed announced the end of QT, bank reserves were reported at $2.85 trillion, the lowest level in more than five years (apart from one print owing to technical factors on January 4, 2023). Is the $2.8 trillion-range where demand resides? The emerging pressure in the overnight money market suggests it is close. Chair Powell and other Fed officials, such as Governor Waller, recently began referring to reserves as a share of GDP, to shed light on demand. At the height of 2019’s liquidity crisis, reserves (and reverse repos with money funds) dropped to 6.8% of GDP in Q3 (Chart 6). This was obviously too low. In an often-cited speech by Governor Waller on “Demystifying the Federal Reserve’s Balance Sheet” in July 2025, he argued that in 2019, liquidity pressures began to emerge when the reserves-to-GDP ratio slipped below 8% and assumes that “9 percent is the threshold below which reserves would not be ample.” That translates to $2.8 trillion based on 2025 Q4 GDP, with the actual figure (a $2.9 trillion average) coming in at 9.2%. There may have been a bit more room to reduce reserves further but, out of an abundance of caution, the Fed would rather allow the growth in Federal Reserve notes to absorb any remaining above-ample liquidity. |
|
Bottom Line: While relaxing liquidity requirements could shrink banks’ demand for reserves and the Fed’s balance sheet further, relaxing rules that could potentially lift bank profits would seem to be a stretch in an election year. The size of the Fed’s balance sheet looks to have hit its low watermark for the foreseeable future. |