Agency mortgage REIT scenario analysis: carry is priced; the next proof is book value through the next spread shock

A real housing photograph is the right anchor here because agency mortgage REIT cash flows begin with ordinary homes financed through conforming mortgages, then get transformed into levered spread products on institutional balance sheets.[6]

Agency mortgage REITs still trade like carry instruments, but yield is only half the instrument. Freddie Mac says the average 30-year fixed mortgage rate was 6.30% on April 30, 2026, while the Federal Reserve's same-day H.15 release showed the 10-year Treasury at 4.34% and the 30-year Treasury at 4.92%.[1][2] That backdrop is good enough to make the group look comfortable again. Priced is the carry. New is whether book value can stay intact when volatility and mortgage spreads lurch wider again.

The first-quarter reports from AGNC and Annaly showed the split clearly. AGNC generated $0.42 of net spread and dollar roll income per common share against $0.36 of common dividends, yet still posted a -1.6% economic return on tangible common equity because tangible book value fell by $0.50 per share.[3] Annaly generated $0.76 of earnings available for distribution per average common share against a $0.70 dividend and still produced a positive 1.5% economic return, but its book value also slipped from $20.21 to $19.82.[4] That is the real 2026 setup. Income can look healthy while marks still decide total return.

Image context: the cover uses Carol M. Highsmith's 2010 photograph of row houses on East Capitol Street in Washington, D.C. That choice fits the article because the underlying collateral here is still plain housing stock. The market complexity arrives later, when those mortgages become agency-backed pools, repo-funded portfolios, and levered book-value trades.[6]

Priced vs new

The priced part is straightforward. Funding got cheaper than it was late last year, mortgage coupons remain usable, and both companies still show dividend-carry support. AGNC's first-quarter release shows a 4.98% average asset yield including TBA positions against a 2.92% combined weighted-average cost of funds, producing a 2.06% annualized net interest spread.[3] Annaly's release shows an average GAAP cost of interest-bearing liabilities of 4.29%, a GAAP net interest margin of 1.41%, a net interest spread of 1.07%, and EAD of $0.76 per average common share.[4]

That is why the group still screens well to income-focused buyers. Policy is no longer pressing directly against the top of the rate cycle, and the public mortgage benchmark remains high enough to keep new mortgage production and agency cash flows from collapsing into a refi wave. Freddie Mac's survey says the 30-year fixed rate was 6.30%, up from 6.23% a week earlier but below 6.76% a year earlier.[2] High enough to slow prepayments, low enough to keep housing demand alive, is still a workable carry regime.

The new part sits in the mark-to-market layer. AGNC is still a concentrated agency machine. As of March 31, 2026, 94% of its investment portfolio sat in 30-year fixed-rate Agency MBS and TBA securities, with tangible "at risk" leverage at 7.4x and actual portfolio CPR at 13.2%.[3] That structure can earn attractive spread income when hedges behave and mortgage basis stays orderly. It also means that spread widening can hit tangible book value quickly even if repo costs move in the right direction.

Annaly shows the other version of the trade. It still had a $106.7 billion total portfolio with a $92.2 billion agency book, but it also carried larger residential-credit and mortgage-servicing-rights sleeves than a pure-play agency REIT.[4] Its leverage profile was lower on the economic measure at 5.7x, and the company still managed a positive economic return in the quarter.[4] That does not remove book-value risk. It does show that portfolio mix can change how much of the quarter's pain arrives through spread marks versus how much can be offset through diversification and dividend coverage.

Six numeric anchors

The public mortgage-rate backdrop is still elevated: Freddie Mac's 30-year fixed rate was 6.30% on April 30, 2026, versus 6.76% a year earlier.[2]
Treasury duration is still carrying a real hurdle rate: the Fed's April 30 H.15 release showed the 10-year Treasury at 4.34% and the 30-year Treasury at 4.92%.[1]
AGNC still earned the carry: it produced $0.42 of net spread and dollar roll income per share, a 2.06% annualized net interest spread, and a 4.98% average asset yield including TBA positions.[3]
AGNC still lost book value: tangible net book value fell from $8.88 to $8.38, and economic return on tangible common equity was -1.6%.[3]
Annaly covered the dividend and kept economic return positive: EAD was $0.76 per share, the common dividend was $0.70, and economic return was 1.5%.[4]
Annaly still carried meaningful but lower balance-sheet pressure than AGNC: GAAP leverage was 7.3x, economic leverage was 5.7x, and the agency portfolio alone was $92.2 billion.[4]

Put together, those anchors say the trade is no longer "buy the yield and stop thinking." The better framing is this: the coupon-and-funding machine is working, but the equity outcome still depends on whether book value can avoid another abrupt markdown.

Three scenarios for the rest of 2026

Base case: rates stay high but less chaotic, mortgage spreads stay range-bound, and both companies keep covering dividends. In that world, AGNC's higher leverage helps the carry show up more quickly in quarterly income, but its book value remains more exposed to any renewed basis move.[3] Annaly's mix gives it a slightly wider set of cushions, so the upside is less explosive but the path may be steadier.[4]

Bull case: Treasury volatility cools, agency technicals stay constructive, and mortgage spreads retrace part of the March widening. AGNC management is already arguing that mortgage spreads look compelling and that supply-demand technicals improved because of higher mortgage rates, bond-fund inflows, and proposed regulatory capital changes.[3] If that view proves right, the group does not need another big drop in funding costs. It mainly needs book values to stop leaking.

Bear case: volatility re-accelerates and agency spreads widen faster than falling repo costs can help. That is the quarter AGNC just previewed. Spread income still prints, dividends still look superficially safe, and shareholders still lose through book value.[3] In that scenario, high stated yields stop being the signal that matters most. The mark on the equity base becomes the real story.

Strongest counterweight

The strongest pushback is that the sector may already have passed through the uglier part of the spread shock. AGNC says spreads widened enough to create compelling value and that technical conditions improved.[3] Annaly says its flagship agency business operated with some of the most supportive technical factors it has seen in years, while its diversified housing-finance sleeves kept expanding.[4] If those two management reads are directionally right, then first-quarter book-value pain may end up looking like a reset rather than the start of another bad sequence.

That counterweight deserves respect. The article's point is narrower. Mortgage REITs can still be attractive here, but the reason to own them is not "yield solves everything." The reason is "carry works, and now the question is whether marks stop fighting the carry."

Falsifier

This view is wrong if the next quarter shows that book value can stabilize or rise even with mortgage rates still around current levels, while dividend coverage stays intact and leverage does not have to fall materially.[2][3][4] If that happens, then the market is right to treat the first-quarter drawdown as noise and the trade really is closer to a straightforward carry story than this article assumes.

Watchlist

Every Thursday Freddie Mac PMMS release: the most public version of the mortgage-rate backdrop now arrives on a fixed schedule, and the gap between the 30-year mortgage rate and benchmark Treasuries remains the simplest public check on whether housing finance is easing or tightening at the margin.[1][2]
Around May 20, 2026 for the April 28-29 FOMC minutes: the Board says minutes are released three weeks after a regularly scheduled policy decision, so the next read on policy-path discussion and volatility management should arrive around then.[5]
June 16-17, 2026 FOMC meeting: this is the next Summary of Economic Projections meeting, which matters because mortgage REITs care less about one headline cut than about the full path of front-end funding and rate volatility.[5]
July 28-29, 2026 FOMC meeting: if volatility is still unresolved by midsummer, this meeting becomes another checkpoint for whether book-value pressure can finally give way to a cleaner carry regime.[5]

Takeaway

In 2026, agency mortgage REITs are still paying investors to own the carry. That part is real. The harder part is that shareholders do not collect only the carry. They also absorb the book-value mark. As long as that remains true, the sector's next rerating depends less on headline yield and more on whether the next spread shock arrives softer than the last one.

Sources

Editor’s Pick Review

This article takes today’s merged standard/add-on editor-pick slot because it does the rare finance job cleanly: it lets yield look attractive without letting yield become the whole thesis. The piece is numerically dense, current to the April 30 rate tape, and useful at the portfolio-decision layer because it separates dividend carry, mortgage-basis marks, leverage, and book-value risk into testable scenarios rather than flattening the sector into a high-income slogan.

It also clears the stricter visual-policy gate. The cover is an immersive, topic-grounded real housing photograph, not an analytical chart or symbolic finance graphic; it connects the reader back to the ordinary mortgage collateral beneath the levered agency-MBS trade. The Chinese edition preserves the same causal structure with stable finance terminology, natural paragraph flow, and low translationese, which made it the strongest bilingual candidate in the last-24-hour pool.

cronfeed.work