Convexity of Mortgage-Backed Securities

What Is Convexity of Mortgage-Backed Securities?

The convexity of mortgage-backed securities (MBS) refers to the non-linear price-yield relationship that arises because homeowners hold an embedded prepayment option. When interest rates fall, homeowners refinance, returning principal early and forcing investors to reinvest at lower yields, a phenomenon called negative convexity. Unlike standard Treasury bonds, whose prices accelerate upward as yields drop (positive convexity), MBS prices are capped by prepayment risk, creating an asymmetric payoff profile that penalizes investors in both rising and falling rate environments.

The key metric governing this is option-adjusted convexity (OAC), which strips out the embedded option value using prepayment models built around benchmarks like PSA speeds and more sophisticated econometric frameworks that account for loan-level characteristics. A typical current-coupon agency MBS might carry an OAC of roughly -1.0 to -3.0, compared to a plain-vanilla 10-year Treasury at approximately +0.7. That negative sign is the mathematical signature of the homeowner's refinancing option working against the bondholder.

The aggregate agency MBS market exceeds $9 trillion outstanding, making it the largest single fixed income sector in the United States after Treasuries. When rates move sharply, duration extension or compression forces large holders, the Federal Reserve, commercial banks, insurance companies, pension funds, and mortgage REITs, to rebalance their hedges, generating enormous mechanical flows in Treasuries, interest rate swaps, and swaptions markets that can dwarf normal supply-demand dynamics.

Why It Matters for Traders

MBS convexity hedging is one of the most powerful and systematic mechanical flows in all of fixed income. The core feedback loop works as follows: when rates rise rapidly, prepayment speeds slow because fewer homeowners have an economic incentive to refinance. As cash flows extend further into the future, the effective duration of the MBS portfolio lengthens, sometimes by 1.5 to 2.5 years across the entire index during a significant selloff. Holders managing duration-neutral books must then sell Treasuries or pay fixed in interest rate swaps to compensate, adding further upward pressure to yields and creating a self-reinforcing feedback loop that amplifies the initial rate move.

The reverse occurs during rapid rallies. Sharp yield declines accelerate prepayments, compressing MBS duration and forcing duration-neutral holders to buy Treasuries or receive fixed in swaps. This mechanic was visible in late 2020 when the Fed's QE purchases had effectively absorbed much of the convexity-sensitive float, damping what would otherwise have been significant hedging flows.

Traders monitoring the mortgage basis, the spread between current-coupon MBS yields and equivalent-duration Treasuries, gain early warning of when hedging pressure is building. A widening basis often precedes or coincides with Treasury selling from MBS holders. Swaption implied volatility, particularly in the 1-year into 10-year sector (the so-called 1y10y swaption), typically spikes during intense convexity hedging episodes as dealers hedge their exposure to this non-linear risk.

How to Read and Interpret It

• Duration extension threshold: When the 10-year Treasury yield rises more than 50 basis points within a four-to-six week window, watch for systematic duration extension selling. A move from 3.50% to 4.20% can add 1.5–2 years of duration to the aggregate MBS index, implying hundreds of billions in notional hedging activity.
• Refinancing incentive monitoring: Track the percentage of the outstanding MBS universe with coupons at least 75 basis points above current mortgage rates, a common threshold for refinancing economics. When this share collapses from, say, 60% to under 20%, extension risk is acute.
• MBA Refinance Index: A sustained drop in this weekly series below its 4-week moving average signals reduced prepayment expectations and rising extension risk. During 2022, this index fell roughly 80% from its 2021 peaks, reflecting the near-total elimination of refinancing incentives.
• Swaption volatility: A spike in 1y10y swaption implied volatility above 120–130 basis points (normalized) often signals active convexity hedging. Dealers who sell optionality into this demand face their own negative-gamma exposure, compounding volatility.
• Mortgage basis widening: Current coupon MBS spreads widening beyond +150 basis points over comparable Treasuries typically reflects forced selling or elevated hedging demand. In late 2022, the current coupon spread briefly touched +175 bps, levels not seen since the 2009 crisis period.

Historical Context

The most dramatic MBS convexity unwind in modern history occurred across 1994, when the Fed raised the federal funds rate from 3.00% to 5.50%. The speed and magnitude caught the market off-guard, causing MBS durations to extend sharply. Mortgage holders were forced to sell an estimated $50–75 billion in Treasuries and pay fixed in swaps, contributing materially to a 250 basis point backup in 10-year yields, one of the worst bond bear markets of the 20th century. The episode established MBS convexity as a systemic amplifier that policymakers could no longer ignore.

A structurally similar, though more contained, episode unfolded during the 2013 Taper Tantrum. From early May to early September 2013, 10-year Treasury yields surged approximately 140 basis points. MBS effective durations extended materially, with Fannie Mae 3.5% pools seeing duration jump from roughly 4.5 years to over 7 years. The resulting hedging flows contributed to the velocity of the selloff and caused the mortgage basis to widen sharply.

Most recently, 2022 offered a real-time laboratory. The Fed's 425 basis points of tightening within a single calendar year drove the largest extension of aggregate MBS duration on record. With the Fed simultaneously running off its $2.7 trillion MBS portfolio rather than reinvesting, private holders absorbed duration at an unprecedented pace, intensifying the convexity feedback across swaps and Treasuries.

Limitations and Caveats

MBS convexity models depend critically on prepayment speed assumptions that are notoriously difficult to forecast with precision. Burnout effects, the tendency of refinancing-eligible borrowers who have not already refinanced to become progressively less responsive to rate incentives, can cause realized speeds to undershoot model predictions, understating extension risk. Conversely, cash-out refinancing demand driven by home price appreciation can keep prepayments elevated even when rate incentives have diminished, overstating extension.

Perhaps the most important structural caveat is Fed ownership. When the Federal Reserve holds a substantial share of the MBS market, as it did at peak QE, owning roughly 30% of the agency float, mechanical hedging flows are suppressed because the Fed does not duration-hedge its portfolio. This makes publicly observable positioning and spread data a less reliable guide to the true convexity overhang in the market.

Finally, in periods of extreme market stress, liquidity in the swaption and Treasury markets themselves can deteriorate, meaning that convexity hedging demand arrives into illiquid conditions, amplifying rather than resolving dislocations.

What to Watch

• Fed's MBS reinvestment pace and portfolio runoff: Changes in Fed ownership shift the convexity-sensitive float to private duration-hedging holders, raising systemic sensitivity.
• Current coupon MBS spread vs. 10-year Treasury: A sustained widening beyond +150 bps warrants close attention as a stress indicator.
• MBA Refinance Index (weekly): The most timely real-world measure of prepayment pressure; track trend breaks rather than absolute levels.
• 1y10y swaption implied volatility: The most sensitive market-based indicator of anticipated convexity hedging demand from dealers and large portfolio managers.
• Aggregate MBS effective duration: Published by Bloomberg and major dealers; a jump of more than 0.5 years month-over-month signals that hedging flows are likely already underway.

How MBS Convexity Plays Out in Practice

Consider a regional bank Treasury desk holding 8 billion USD of FNMA 30-year 5.5% MBS in May 2026, purchased at par in 2023. The current coupon Fannie 30Y trades at roughly 5.75%, so the 5.5s are slightly below the refi threshold. Bloomberg models price the position with an effective duration of 5.8 years and an option-adjusted convexity of minus 1.9. The desk has hedged 90% of the duration via a 10Y UST short of approximately $4.5 billion DV01-equivalent.

Now assume 10Y UST yields drop 60 bps over four weeks, from 4.31% to 3.71%, on a softening labor narrative. The 30Y mortgage rate falls from 6.85% to roughly 6.25%, putting the 5.5% coupon decisively in the money for refinancing. Prepayment speeds, modeled at 8 CPR before the move, jump to 22 CPR within two months as the MBA Refinance Index spikes. The MBS effective duration collapses from 5.8 years to 3.4 years. The bank's hedge, sized for 5.8 years of duration, is now over-hedged by 41%, meaning the 10Y short rallied $260 million ($4.5 billion DV01 times 60 bps times approximately 0.97 hedge ratio), while the MBS itself rallied only $140 million because the negative convexity capped its upside. Net P&L: a loss of $120 million on a position the desk thought was hedged.

The alternative scenario is even more painful. If 10Y yields instead rise 75 bps to 5.06%, prepayments collapse from 8 CPR to 4 CPR. The MBS effective duration extends from 5.8 to 8.1 years, a 40% extension. The bank's 10Y short, sized for 5.8 years, is now under-hedged by 28%. The MBS loses approximately $410 million on the rate move; the hedge gains $325 million. Net loss: $85 million. The asymmetric extension-compression dynamic means the bank loses meaningful capital in both directions, the literal definition of negative convexity.

Desk practice is to dynamically rehedge using swaptions rather than cash Treasuries. A 6,000 million notional 1y10y payer swaption struck 50 bps OTM gives the desk positive gamma to long-dated rates rising, partially offsetting MBS extension risk. Cost: roughly 95-110 bps upfront, or $5.7-6.6 million for the hedge. The desk treats this as insurance: in a 100 bp rate move scenario, the swaption gains 4-6x its premium while MBS loses 5-7% of notional. The breakeven is a rate move of 35-45 bps; anything less and the swaption decays unused, anything more and the hedge pays. Most large MBS holders run continuous swaption rolls at 15-25 bps annualized drag as the systematic cost of being short convexity at scale.

Current Market Context (Q2 2026)

The aggregate agency MBS market is approximately $9.4 trillion outstanding as of Q1 2026, with the Fed's SOMA portfolio holding $2.1 trillion of that, down from the 2022 peak of $2.74 trillion. Fed QT continues to run agency MBS off the balance sheet at the runoff cap, currently $35 billion per month, with no reinvestment. The Fed's exit shifts incremental negative convexity to private holders, primarily commercial banks ($2.6 trillion), GSEs ($420 billion), and money managers including Vanguard, BlackRock, and PIMCO, who collectively manage roughly $1.8 trillion.

The current coupon Fannie 30Y trades at 5.72% in May 2026, a spread of roughly 141 bps to the 10Y UST at 4.31%, near the post-2022 median of 145 bps and well inside the 195 bp peak of October 2023. The 30Y mortgage rate at 6.85% versus the prevailing coupon stack at 5.0-6.0% means roughly 38% of outstanding MBS principal is currently refi-able, but homeowner refi behavior remains sluggish at 7-9 CPR aggregate due to lock-in effects and persistent rate volatility memory from 2022-2024.

Watch the MOVE index, currently 92, against its 12-month range of 78-118. A move above 110 typically triggers dealer convexity hedging flows of $40-80 billion in 10Y UST equivalents per session. The 1y10y swaption implied volatility, currently 98 bps annualized, is the cleanest forward-looking signal; readings above 115 signal the market is pricing imminent convexity hedging supply. MBA Refinance Index (weekly Wednesday) at roughly 320 in May 2026 versus the 2024 peak of 980 shows refi activity is subdued; a sustained breakout above 500 would inject prepayment pressure.

What to monitor: the FN 5.5 OAS versus the FN 5.0 OAS. When the higher-coupon OAS widens by more than 12 bps relative to the lower coupon over two weeks, the market is repricing prepayment risk in real time and 10Y volatility usually follows.