Quantitative tightening (QT) is the process by which the central bank shrinks its balance sheet, reversing the effects of quantitative easing. Rather than selling bonds outright, the Fed typically lets securities mature and does not reinvest the proceeds.
The effect is the mirror image of QE:
The Fed has conducted QT twice:
QT's impact on the curve is more diffuse than rate hikes. It operates primarily through the term premium rather than the expectations channel, making it harder to calibrate and communicate. The Fed has described QT as "running in the background," a passive tightening that complements active rate policy.
The blog posts "The End of the Hedge" and "The Yield Curve Predicted a Recession That Never Came" discuss how QT contributed to the 2023 bear steepener and the divergence between curve signals and economic outcomes.
QT is usually passive, and the mechanism is worth following step by step because it explains the pace.
The Federal Reserve holds Treasuries and mortgage backed securities. Each month some of them mature. Under normal operations the Fed reinvests the proceeds by buying replacement securities, which keeps the balance sheet flat. Under QT it stops reinvesting some or all of that amount, and the balance sheet shrinks by the difference.
The Fed sets a monthly cap on how much it allows to run off. If maturities in a month exceed the cap, the excess is reinvested. If maturities fall short of the cap, the balance sheet shrinks by less than the cap allows. The cap is a ceiling, not a target, so the realized pace is lumpy and follows the maturity profile of the portfolio.
Mortgage holdings shrink more slowly still. Their principal returns through prepayments rather than on a schedule, and prepayments collapse when mortgage rates rise. QT therefore runs slowest in exactly the conditions that triggered it.
The effect on the market is indirect. The Fed is not selling. It is declining to buy. The Treasury still issues the same amount, so a larger share of each auction has to be absorbed by private investors.
The asymmetry is the most useful thing to understand about QT, and it has two parts.
The announcement does the work in QE, not in QT. Asset purchase programs were launched into stressed markets with explicit signaling about policy intent. Much of their effect arrived on announcement, before a single bond was bought. QT is announced in advance, executed mechanically, and deliberately made boring. There is no comparable signaling shock.
The stock matters more than the flow. Research on purchase programs suggests that yields respond to the total holdings expected over time rather than to the rate of buying in any month. If that is right, then removing holdings slowly has a small and gradual effect, spread over years rather than concentrated at any point.
The consequence is a policy tool with a weak and uncertain transmission. The Federal Reserve has described QT as running in the background for that reason. It also means the effect is hard to separate from everything else moving the term premium at the same time, including fiscal deficits and shifting foreign demand.
QT does not run until the balance sheet is empty. It runs until bank reserves become scarce, and that limit is discovered rather than chosen.
Every dollar of runoff drains liquidity from the banking system. For a while this is harmless, because the system starts with far more reserves than it needs. As reserves fall toward the level banks actually want to hold, funding markets tighten. The first symptom appears in repo rates, which begin to trade above the policy rate and spike at month and quarter ends.
The Federal Reserve cannot observe the sufficient level of reserves in advance. It can only watch for the symptoms. That is why QT programs have historically slowed and then stopped in response to funding market stress rather than on a preannounced date.
The practical implication for a curve watcher is that repo market behavior is a leading indicator for QT policy. Persistent pressure in overnight funding is the signal that the program is close to its limit.
The direction is up, and the size is uncertain and probably modest. QT works through the term premium by returning duration to private investors. Estimates of the effect vary widely across studies, and the impact is easily swamped by fiscal supply, inflation expectations, and foreign demand moving at the same time. Attributing any particular yield move to QT alone is not defensible.
Selling would realize losses on securities bought at higher prices, and it would put the Federal Reserve in the position of actively transacting against the market it regulates. Passive runoff avoids both problems. It also makes the pace predictable, which is the stated aim.
Yes, and it has. The two tools work through different channels. Rate policy sets the price of short term money and drives the front of the curve. QT changes the quantity of duration held by private investors and works on the long end. A central bank can ease at the front while continuing to shrink the balance sheet, though it usually stops QT once easing becomes substantial.