The Federal Reserve, having spent the better part of three years drawing down a securities portfolio that had reached eight point nine trillion dollars at its peak in 2022, has begun, very quietly and with rather less fanfare than the original expansion, to buy Treasury bills again. The Open Market Trading Desk in New York announced on the tenth of December that it would purchase approximately forty billion dollars in bills in the first month of operations,¹ and has continued, each subsequent month, to publish a calendar of intended purchases and the operations to deliver them. By the eighth of April, the Federal Reserve's total assets stood at six point seven trillion dollars,² slightly above the trough of late 2025 and on a trajectory the System's own staff have indicated will continue. The Federal Reserve, in short, is once again expanding its balance sheet.

It is important to be precise about what this is, and what it is not. It is not quantitative easing. It is not, in the ordinary sense, a stimulus measure. It does not represent a change in the stance of monetary policy, a point Federal Reserve officials have made with such frequency that one suspects they themselves are slightly worried about the parallel. The reserve management purchases, or RMPs, of 2026 are the lineal descendant of the bill purchases the Desk conducted between October 2019 and the spring of 2020, after the repo market had displayed, in September 2019, that disorderly tendency which Chairman Jerome Powell would describe at the time as a "technical" rather than a monetary problem. The technical problem then, as now, was an insufficient supply of reserves to satisfy the demands of a banking system that requires considerably more of them than it once did, and the technical solution then, as now, was for the central bank to supply more.

The accompanying chart traces the trajectory. The contour is, in its way, a portrait of the post-2008 settlement: a small balance sheet expanded violently in response to crisis, drawn down with deliberation, and then stabilised at a level that bears no relation to its pre-crisis size. The Federal Reserve's balance sheet, before September 2008, was approximately nine hundred billion dollars. It will not return to that figure within the working lifetime of any economist now writing about it. The reasons are not mysterious. The reserve requirements that once forced banks to demand reserves only narrowly above zero have been replaced by a regulatory architecture, principally the Liquidity Coverage Ratio of Basel III, which compels them to hold high-quality liquid assets in volume; the largest banks have, on top of this, been habituated by supervisory practice to hold reserves at levels comfortably above the regulatory minimum; and the Treasury's General Account, the Reverse Repo Facility, and the foreign official sector all draw on the System's liabilities in ways that did not exist in 2007.

The result is a banking system whose ample reserves, as Federal Reserve officials describe them, are the foundation of monetary policy implementation. The Federal Reserve Bank of St. Louis, in a recent explainer, observed that the framework "supplies enough reserves to control short-term interest rates primarily through administered rates, rather than through active management of the supply of reserves."³ The administered rates are the interest paid on reserve balances and the rate offered at the Overnight Reverse Repo Facility. Together they form the floor and the soft floor of the federal funds market, and so long as reserves are abundant, the federal funds rate trades within their corridor without the need for active intervention. The architecture, in this respect, resembles the British system established by the Bank of England in 2006, which abandoned the previous reserves-averaging regime in favour of remunerated reserve balances and standing facilities. It is also, more distantly, an inheritance from the European Central Bank, which has operated a corridor system since 1999. The American adoption of the model is recent and was contested. It is now, for all practical purposes, settled.

Settled, however, is not the same as costless. Bank reserves, as of last month, stood at approximately three point four trillion dollars, accounting for roughly eleven per cent of nominal GDP.² The Federal Reserve pays interest on these reserves at the prevailing administered rate. At the present interest on reserve balances rate, the System remits to its member banks somewhere in the order of one hundred and fifty billion dollars annually in interest, a figure which, in the years before 2022, was offset by the carry on the System's longer-dated holdings, but which, since the rate-tightening cycle, has produced operating losses recorded as a deferred asset against future remittances to the Treasury. The deferred asset, as of the most recent H.4.1, exceeds two hundred billion dollars. It will be retired only when the System resumes its ordinary remittance pattern, and it represents, in substance, a transfer of seignorage from the Treasury to the banks, mediated by the central bank's policy implementation framework. There is nothing improper in this. It is, however, a fact about which the public is largely uninformed and about which the political class has been largely incurious.

Why Bills, and Why Now

The Desk's choice to conduct the present reserve management purchases entirely in Treasury bills,⁴ rather than in coupon-bearing securities, is a small but consequential design decision. Bills mature within a year and impose, by virtue of their short duration, almost no interest-rate risk on the System's portfolio. They are also, importantly, the security in which the Treasury can issue most flexibly without disrupting the term structure of the public debt. The Treasury Borrowing Advisory Committee, in its February charge, examined precisely this question and concluded that bill purchases by the Federal Reserve, in conjunction with continued bill issuance by Treasury, were the most prudent path for restoring an ample reserves position.⁵ The decision avoids the appearance of a stimulus measure, preserves the System's balance sheet duration in a defensible range, and leaves the Federal Reserve free, should monetary conditions warrant a more contractionary stance, to allow its longer-dated holdings to roll off without difficulty.

The volumes contemplated are not trivial. The New York Fed has indicated that, for calendar year 2026, reserve management purchases to maintain ample reserves will total approximately three hundred and sixty billion dollars, against agency mortgage-backed security paydowns of approximately one hundred and eighty billion, for a combined Treasury bill demand from the System Open Market Account of roughly five hundred and forty billion dollars.⁴ This is, in the aggregate, the largest source of marginal demand for Treasury bills in the present market, and it arrives at a moment when the Treasury itself has signalled an intention to continue tilting issuance toward the bill sector, in part to take advantage of the comparatively flat short-end of the curve and in part to manage the cash balance more flexibly than would be possible with longer issuance. Whether this combination will, over time, produce material distortions in the bill market is a question on which reasonable observers may differ; the early evidence, helpfully summarised by the Desk in its March update,⁶ is that the operations have proceeded without difficulty.

The Strong Parallel

The historical parallel that suggests itself is the Federal Reserve's experience between 1923 and 1929, when Benjamin Strong, as Governor of the New York bank, used open-market operations in government securities to manage liquidity in the call-money market and, less successfully, to coordinate with the European central banks then attempting to restore the gold standard. Strong's purchases, like the present operations, were generally framed as technical adjustments to reserve positions rather than as instruments of policy stance. The framing was, in the strict sense, accurate. It was also, as the events of 1928 to 1930 would demonstrate, incomplete. Open-market operations in any volume and any direction influence the financial conditions facing the real economy; the question is whether the influence is large or small, intentional or incidental, and whether the institution conducting them has the analytical apparatus to monitor what it has set in motion.

The Federal Reserve of 2026 has analytical apparatus that Strong could not have imagined. Its staff produce, with considerable regularity, estimates of the demand for reserves, the supply of reserves, the resulting elasticity of the federal funds rate to changes in either, and the implications for money market spreads. This apparatus is the principal defence against the suggestion that the present bill purchases are quantitative easing in a more politically palatable wrapper. The defence is technically sound. It is also, however, the kind of technically sound defence that institutions tend to advance most insistently in the period before circumstances force them to admit that the line between technique and policy is not always as clear as the policy makers would prefer.

The complications, when they arrive, will arrive at the seams. There are, in my view, three to watch. The first is the interaction between Treasury bill issuance and System bill purchases. If the Treasury increases bill issuance more rapidly than the System absorbs it, the supply will weigh on bill yields and complicate the Federal Reserve's effort to keep the federal funds rate within its target range; if the System absorbs bills more rapidly than the Treasury issues them, the System will be forced into purchases at the longer end of the curve, with the duration consequences that implies. The second is the political reception of the System's accumulating losses, which, while ledger entries of the kind that economists rightly insist do not constrain a central bank's ability to act, are politically combustible items that the Federal Reserve has historically preferred to avoid. The third is the question, never quite settled, of what the Federal Reserve does if the next recession requires another round of asset purchases on a scale comparable to 2020. A balance sheet that is again expanding, however gently, leaves less room to expand further.

A century ago, in the spring of 1924, the Federal Reserve System was a relatively new institution still discovering what its open-market operations could and could not accomplish. The New York Bank's purchases of government securities that year were small by present standards and large by the standards of the time, and they were undertaken in part to ease conditions in the United States and in part, more controversially, to assist the Bank of England's return to gold. The historical record suggests that the operations achieved their immediate objectives and contributed, indirectly, to the development of asset bubbles whose deflation would consume the following decade. The lesson is not that bill purchases are dangerous. The lesson is that institutions which conduct open-market operations of any size acquire, by virtue of their conduct, responsibilities they did not anticipate and effects they cannot fully foresee.

The Federal Reserve has resumed buying Treasury bills. The operations are technically well designed, transparently announced, and consistent with the framework the System has chosen for the implementation of monetary policy. They are also, when one considers the volumes involved, the second largest expansion of the System's balance sheet in living memory after the pandemic. To call the present moment a return to normality is to use the word "normality" in a sense that would have surprised any monetary economist before 2008. We are in a new regime. The bill desk is busy again.