Annaly Capital Management (NLY)

Business Overview

Annaly is the agency benchmark at scale, and it is no longer only an agency book. About 62% of its capital funds a levered portfolio of agency mortgage-backed securities, financed with repurchase agreements and hedged with swaps and Treasury futures. The remaining 38% splits roughly evenly between residential credit, where the Onslow Bay subsidiary buys whole loans and securitises them, and mortgage servicing rights, contracts that pay a fee for as long as a borrower keeps a loan alive. That mix matters. Servicing gains value when rates rise and refinancing stops, the opposite direction to a long bond book, and neither servicing nor credit can carry the borrowing that agency paper carries.

What an mREIT earned you is the dividend plus the change in book value, and nothing else. FY2025 was a strong year on that test: book value per common share rose from $19.15 to $20.21 while $2.80 of dividends were paid, a 20.2% economic return. Book then slipped to $19.82 at 31 March 2026. Market capitalisation was roughly $15.9 billion on 10 June 2026. The sector treats a 1.00–1.15× price-to-book as around fair for a quality agency book.

That framing is deliberate. Agency mREITs are not valued on property NAV or FFO. The market reads them through price-to-book on disclosed BVPS, and Annaly is the name most analysts reach for when calibrating that lens. The book value guide uses Annaly's FY2025 path as the worked example.

How Agency Carry Works

Earnings come from the net interest spread: asset yield on the MBS portfolio minus the cost of repo funding and hedge drag, measured after premium amortisation catch-up (PAA) is stripped out. Annaly's quarterly annualised net interest spread excluding PAA ran 1.35%, 1.47%, 1.50%, and 1.49% through FY2025. Annaly files the spread by quarter rather than as a full-year figure, so the average of the four, about 1.45%, is the fair summary of the year, and the fourth quarter's 1.49% is the freshest read of it.

Leverage amplifies a thin spread across a large asset base. Annaly's economic leverage was 5.6× at 31 December 2025 and stayed between 5.6× and 5.8× all year. AGNC prints 7.2× at-risk leverage, and the different labels explain very little of that gap: Annaly divides by total equity including preferred, AGNC by total equity less goodwill, and the two denominators land close together. AGNC really does run the harder book, against bonds that are all government-guaranteed. Nearly two fifths of Annaly's capital sits in credit and servicing assets that will not support the same borrowing. The leverage guide rebuilds both ratios from the footnotes.

Repo is where that leverage bites. Lenders mark the collateral every day, so a fall in MBS prices brings a margin call, and the REIT meets it with cash or by selling bonds into the market that has just fallen. That loop is how a bad fortnight in the bond market turns a paper mark into a permanent hole in book value, and it is the reason liquidity is worth watching alongside the ratio: Annaly held $6.1 billion of cash and unencumbered agency MBS at the end of 2025.

Prepayment matters because the borrower picks the moment, and picks it badly for the lender both ways. When mortgage rates fall, borrowers refinance and a bond bought above par is repaid at par, exactly when it would otherwise have risen in price. When rates rise, refinancing stops and the money stays out at a below-market coupon, precisely when the holder would rather have it back. That asymmetry is negative convexity, and it is why an agency book behaves worse than a Treasury of similar duration. Annaly's experienced CPR was 9.7% at 31 December 2025 against a projected long-term 10.8%; the figure is a quarter-end disclosure, not a full-year average.

Book Value and Economic Return

Annaly grew BVPS from $19.15 to $20.21 in FY2025, up 5.5%. Add the $2.80 dividend and divide by opening book, and economic return lands at 20.2%: total return on book capital, before the market's price-to-book opinion enters.

The first quarter of 2026 shows what the measure catches that a book value alone does not. Book fell $0.39 to $19.82 while $0.70 was paid out, so the quarter still returned 1.5% on opening book. A falling book is not by itself a bad quarter, any more than a discount to book is by itself cheap.

Price-to-book says what the market expects: modest further book growth, or a small premium to liquidation value for scale and execution. Discounts below 0.90× normally price expected book erosion; premiums above 1.15× price franchise or growth optionality. A book that grew 20.2% and then gave a fraction of it back would be expected to sit a little above par.

Dividend yield was 12.93% on the 10 June price against FY2025 dividends of $2.80 per share ($2.80 ÷ $21.66). The payout was $0.70 a quarter throughout, against fourth-quarter earnings available for distribution of $0.74. Coverage for an agency name is not a GAAP earnings multiple: the question is whether spread and leverage can keep funding the book value path that stands behind the dividend.

Valuation Framework

Price-to-book is the first check; economic return tells you whether the multiple is paying for history or expecting more. After a 20.2% year, the argument is not the book maths but whether the spread and prepayment path can produce another year like it.

The hedges will not settle that argument, because they are not aimed at it. Annaly's hedge ratio was 90% at the end of 2025, but swaps and Treasury futures hedge the level of interest rates, not the gap between mortgage yields and Treasury yields. If mortgage bonds cheapen against Treasuries, a fully hedged book still loses book value, and that is what a spread-widening quarter means. The Mortgage REIT Sector Primer runs a 100 bps MBS spread as its planning input for exactly that reason: the input exists to show what the book does when the spread moves, not to predict where it goes.

Compare Annaly with AGNC on what each actually owns. Annaly's spread averaged about 1.45% in FY2025 against AGNC's filed 1.92%, on a portfolio where nearly two fifths of the capital is not in levered agency carry at all. The 5.6× against 7.2× leverage gap points the same way: it is mostly a real difference in what the two companies hold, not a difference in footnotes. The agency vs hybrid guide situates both against Rithm and the CRE lenders, which run entirely different risk books.

What to Watch in the Financials

Quarterly BVPS and economic return. Each earnings release updates the scoreboard. A quarter where book falls by more than the dividend paid is a negative economic return, and it usually shows up in the multiple soon after.

Net interest spread excluding PAA. The quarterly annualised series is the live spread read. Watch for divergence between asset yields and repo costs, and for hedge ineffectiveness showing up as spread compression without a matching rate move in the macro data.

Economic leverage path. Stable at 5.6×–5.8× through FY2025 and 5.7× at 31 March 2026. A deliberate move towards 6.0× would amplify both the carry and the mark-to-market swings around it.

CPR versus projection. Experienced CPR of 9.7% at end-2025 against a long-term projection of 10.8%. Speeds running above projection shorten the portfolio faster than the hedges assume, and the repositioning costs money.

The split of capital. Agency, residential credit and servicing were 62/19/19 at the end of 2025. Shifts in that mix change what the share price is actually tracking, and how much of the book is marked on screen prices rather than on models.

Key Risks

Spread widening, not just rates. A rapid move in long rates marks the MBS book before the carry can catch up, and hedges cover that. What they do not cover is mortgage bonds cheapening against Treasuries, which hits book value with the credit quality untouched.

Prepayment shocks. A refinancing wave when mortgage rates fall accelerates CPR and shortens the portfolio faster than fixed hedges adapt, at the moment the bonds would otherwise have been worth most.

Repo funding. The book is financed short against long assets. A stressed market brings margin calls on the way down, and forced selling into a falling market is what converts a mark into a realised loss.

P/B mean reversion. With little premium cushion, if book value growth stalls the multiple can compress towards discount territory quickly, because the sector prices expected book rather than trailing GAAP earnings.