QT Did Not Eliminate the Convexity. It Privatized It.

Scope

This note is not about rates going higher. It is not a prediction about where the 10-year yields or when the Fed cuts. Those are surface questions.

This note is about who bears the convexity when yields move — and why the answer to that question changed structurally during the QT cycle in a way that makes the next rates volatility episode categorically different from every episode since 2008.

The Federal Reserve warehoused an enormous amount of mortgage convexity for fourteen years. It no longer does. That convexity did not disappear. It transferred to private balance sheets that behave differently under stress — and that difference is now baked into the structure of the Treasury market whether yields rise or fall.

What Changed

At its peak in April 2022 the Federal Reserve held approximately $2.74 trillion in agency mortgage-backed securities — roughly 30 percent of the entire agency MBS market. That concentration was not incidental. It was the deliberate residue of two quantitative easing programs designed to support housing finance and compress mortgage spreads.

The important characteristic of the Fed as an MBS holder is what it did not do: it did not hedge.

When yields rose and the Fed's MBS portfolio extended duration, the Fed absorbed that extension passively. No Treasury selling. No swap positions. No hedging flows. The convexity sat on the Fed's balance sheet and went nowhere. That passivity was itself a form of market stabilization — a $2.74 trillion position that would never generate hedging flows regardless of how far yields moved.

QT changed that. The Fed reduced its MBS holdings from $2.74 trillion to approximately $2.1 trillion between June 2022 and December 2025, when QT ended. Roughly $600 billion in agency MBS moved from the Fed's passive balance sheet into the hands of private holders — banks, real money managers, mortgage REITs, insurance companies, and hedge funds. Those holders do hedge. When their MBS extends duration they sell Treasuries, pay fixed in swaps, or buy payer swaptions to offset the risk. The hedging flows that the Fed's passivity suppressed for fourteen years are now embedded in private portfolio management across the system.

The current coupon stack compounds the problem. The 2022 through 2024 vintage of mortgage originations carries coupons in the 6 to 7 percent range. At current 30-year mortgage rates, those mortgages are sitting at or near par — precisely the point of maximum convexity. A 50 basis point rise in yields at current coupon levels can extend MBS duration from roughly 5 years to 8 to 10 years. That extension does not happen gradually. It happens in a narrow yield range centered on where the current coupon already sits.

The mechanism that follows is endogenous. Private holders extend. They sell Treasuries or pay fixed. That selling pushes yields higher. Higher yields extend the MBS further. More hedging flows. More selling. The loop does not require a new macro shock to activate. It requires yields to move past the threshold where hedging flows become large relative to dealer absorption capacity — and that threshold is closer than it appears because dealer absorption capacity has been deteriorating for reasons documented in this archive since April.

The April 5 note documented the Japan leg and the leveraged basis leg losing absorption capacity simultaneously. The April 9 note documented the marginal Treasury buyer becoming a leveraged basis trade on repo — an entity with finite balance sheet that can become a seller under stress rather than a buyer. The May 15 note asked who still has balance sheet. The May 21 note showed that the TGA shock absorber is gone.

None of those notes addressed what generates the hedging flows that stress all four of those structures simultaneously. Mortgage convexity is the answer. It is the mechanism that can activate the absorption capacity problems documented in prior notes without requiring a new exogenous shock. The convexity is already loaded. The private holders are already positioned. The hedging infrastructure is already in place. The only missing variable is a sufficient yield move to cross the extension threshold.

Agency MBS remains the largest fixed income market in the United States by outstanding balance. The convexity embedded in that market has not changed in magnitude — it has changed in ownership and therefore in behavioral response to yield moves.

The Fed remains the single largest holder of agency MBS at approximately $2.1 trillion. Its passivity still dampens some convexity transmission. Going forward the Fed has indicated it will roll maturing MBS into Treasury securities, allowing its MBS holdings to decline gradually — continuing the ownership transfer to private holders who hedge.

The convexity hedging mechanism itself is not new. Every rates volatility episode since agency MBS markets developed has involved some degree of convexity-driven amplification. The 1994 bond market massacre, the 2003 summer selloff, the 2013 taper tantrum — all showed some MBS convexity transmission. What is new is the combination of three factors simultaneously: the scale of private ownership after the largest MBS balance sheet reduction in history, the concentration of the current coupon stack at maximum convexity, and the deterioration of the dealer absorption capacity that historically dampened the amplification.

→ MOVE index behavior during yield rises — the critical signal is MOVE accelerating faster than yields rise; that pattern indicates hedging flows are amplifying the move rather than macro repricing alone driving it

→ 5-year to 10-year Treasury sector performance relative to the 2-year and 30-year during yield rise episodes — convexity hedging concentrates in the belly because that is where MBS duration extension maps; belly underperformance during yield rises is the convexity flow signature

→ Agency MBS option-adjusted spread — widening OAS means private holders are demanding additional compensation for bearing convexity they are increasingly unable to hedge efficiently; persistent OAS widening without macro deterioration is the structural signal

→ Current coupon MBS versus off-the-run MBS spread — compression of the spread between current coupon and seasoned lower-coupon MBS indicates hedging pressure concentrating in the current coupon tier

→ 30-year mortgage rate versus 10-year Treasury spread — the primary mortgage spread; widening beyond 250 to 270 basis points indicates hedging cost is being passed through to the mortgage market and that convexity sellers are demanding a wider buffer

→ Federal Reserve MBS SOMA holdings — the pace of residual decline from $2.1 trillion; each additional reduction transfers more passively-held convexity into actively-managed portfolios

→ Treasury futures positioning in CFTC data for asset managers — net duration reduction in asset manager positioning during yield rises, especially concentrated in 5-year to 10-year contracts, is the visible signal of convexity hedging flows

The standard framing of QT is that it removed accommodation — that a smaller Fed balance sheet means less support for asset prices and higher long-term yields. That framing is correct as far as it goes. It does not go far enough.

The more important structural consequence of the QT cycle is not what the Fed removed from the market. It is what the Fed transferred to it. $600 billion in agency MBS moved from a holder that never hedges to holders that do. The convexity those securities embed — the sensitivity of their duration to yield moves — did not change. The behavioral response of their owners to yield moves changed entirely.

The result is a Treasury market that is structurally more reflexive than at any point since the Fed became the dominant MBS holder in 2009. When yields rise past the current coupon threshold, private holders hedge. Their hedging pushes yields higher. Higher yields force more hedging. The loop has no internal dampening mechanism — it terminates only when yield moves exhaust dealer absorption capacity or reverse. The dealer absorption capacity is thinner than it was in every prior episode this mechanism activated, for reasons documented in the notes that preceded this one.

The market is not pricing this. MBS volatility surfaces and Treasury options markets reflect event risk and macro uncertainty. They do not explicitly price the structural amplification embedded in the private ownership of $600 billion in convexity that the Fed used to hold passively.

QT did not eliminate the convexity. It privatized it. And private holders, unlike the Fed, hedge.

Hampson Strategies — Market Note · May 22, 2026

Not investment advice. Personal observations based on publicly available data.

© 2026 Andrew C. Hampson II / Hampson Strategies. All rights reserved.

Full archive: hscai.org/market-notes · Institutional engagement: hscai.org · 865-236-1026