What Quantitative Tightening Actually Does to Markets

A balance sheet the size of a small economy

Start with a number that is hard to picture. Between December 2005 and December 2025, the Federal Reserve's balance sheet grew from roughly $800 billion to about $6.5 trillion — from around 6 percent of GDP to roughly 21 percent. That is not a rounding error; it is a second policy toolkit sitting alongside interest rates.

Why did it get so big? Two decisions, according to Fed staff. First, the central bank bought large quantities of securities after the 2008 financial crisis and again during the pandemic — the programmes we call quantitative easing, or QE. Second, in 2019 it formally shifted to an "ample reserves" system, meaning it deliberately keeps a generous cushion of reserves (cash that banks hold on deposit at the central bank) floating around the system.

When a central bank buys a bond, it pays for it by creating reserves. So asset purchases and reserve levels are two views of the same thing. Shrinking the balance sheet, then, is really a story about draining reserves — and that is where things get interesting.

Runoff, explained without the fog

Quantitative tightening, or QT, is the reverse journey. The simplest version is "runoff": when a bond the central bank owns matures, it does not reinvest the proceeds. The portfolio shrinks, and reserves disappear from the banking system. No dramatic announcement, no market order — just a quiet decision not to roll something over. Economists sometimes call this passive tightening, and it is the Fed's standard tool.

The European Central Bank took a similar road in stages. It stopped net purchases under its asset purchase programme in July 2022, kept reinvesting in full for a while, moved to partial reinvestment from March 2023, and discontinued reinvestments entirely as of July 2023. Notice the pattern: central banks tend to slow-dance their way out, not sprint.

How much of the shrinkage is actually a deliberate policy choice? Fed researchers decomposed the decline in the Fed's securities holdings relative to GDP across two episodes. In 2014–2019, more than half of the reduction came from passive forces — inflation and real economic growth making the balance sheet look smaller relative to a bigger economy. In 2022–2025, by contrast, active reductions accounted for about 59 percent of the decline, inflation about 26 percent and real GDP growth roughly 15 percent. The recent episode also saw a 14 percentage point decline in that ratio, double the 7 points of the earlier period. In plain terms: this time, the central bank meant it.

Why reserves are the quiet protagonist

Here is the mechanic that trips people up. A central bank's balance sheet must balance. If the government's cash balance at the Fed — the Treasury General Account, or TGA — rises because taxes are due, something else must fall, and that something is usually reserves. Reserves drop one-for-one unless the central bank offsets the move.

When reserves are plentiful, banks barely notice such swings. When reserves get thin, the same small shock produces a much bigger move in short-term interest rates. Banks needing to settle payments must pay up for cash, and repo rates (the interest on short, secured loans collateralised by securities like Treasuries) can jump. Fed staff describe this as a trilemma: a central bank can have a small balance sheet, low short-term rate volatility, and limited market intervention — but only two of the three at once.

September 2019 is the cautionary tale. After a period of runoff, a corporate tax date and a heavy Treasury settlement drained reserves. Repo rates spiked as high as 10 percent intraday, and the effective federal funds rate pushed above its target range. The Fed responded with large short-term purchases and repo operations. Its staff were quick to say this was liquidity plumbing, not a new round of QE — a distinction markets found debatable. Calling something technical does not stop it from moving prices.

How it travels to rates, currencies and liquidity

The transmission runs through several channels, and they do not all point the same way.

Term premiums. When a central bank sells or stops buying longer-dated securities, the public must absorb more of them. Fed research notes that this increases the supply of longer-term securities and puts upward pressure on longer-term interest rates. That is the textbook QT effect, and it is generally modest.

Liquidity and funding stress. As reserves shrink, rate sensitivity rises. Academic work from LSE found that the Fed's balance sheet expansion coincided with more bank deposits and more corporate credit lines — claims on liquidity that did not shrink proportionally when the balance sheet did. That left the system more exposed to liquidity shocks, requiring Fed support in September 2019 and again in March 2020. QT, in other words, is not a perfectly benign process.

Currencies. Higher longer-term yields and tighter liquidity can support a currency, but the effect is rarely clean. Global funding markets are intertwined, and a jurisdiction's exchange rate reflects growth, inflation and policy expectations all at once. Anyone claiming a mechanical, one-way FX link is overselling certainty.

Market functioning. The Treasury market has grown while dealer balance sheets have not kept pace, so leveraged arbitrage — such as the cash-futures basis trade — now intermediates a lot of risk. In March 2020, margin calls forced rapid unwinds, and the Fed bought roughly $1.6 trillion of Treasuries in a few weeks to restore order. Economists Anil Kashyap and Jeremy Stein argue for narrower, duration-neutral interventions in future: buy cash Treasuries while selling futures, so market support does not get confused with monetary easing.

What it means for an ordinary reader

You will not see QT on a receipt, but you can feel it in borrowing costs, mortgage rates, and the general availability of credit. It also shapes how much room a central bank has before money markets start to creak. Current frameworks suggest the answer is "less than surveys once predicted" — a lesson learned the hard way in 2019.

And size itself remains unresolved. The Fed concluded runoff on December 1, 2025, then announced reserve management purchases days later to keep reserves ample. There is no consensus among economists or policymakers on the right steady-state size. Different central banks have chosen different compromises, and rules differ by jurisdiction, so conclusions from one market do not automatically travel.

One last thing worth internalising: QE and QT are not mirror images. Institutions expand their demand for liquidity in a crisis but do not fully reverse it afterwards, which means runoff shrinks the balance sheet mechanically without necessarily shrinking the system's appetite for safety. It is a bit like tidying a room by opening the window — the clutter reduces, but the furniture stays.

Watch reserves, repo spreads and quarter-end quirks. They are the plumbing, and plumbing rarely makes headlines until it leaks.