REDWOOD TRUST INC (RWT) stock intrinsic value fair value estimate & owner earnings analysis

Redwood Trust is a mortgage lender and investor organized as a real estate investment trust. It runs three loan-buying and origination businesses and keeps a portfolio of the investments those businesses create. Sequoia buys jumbo home loans (too large for Fannie Mae and Freddie Mac) from banks and mortgage companies and sells them on through Sequoia securitizations and whole-loan sales. Aspire does the same for expanded-credit and non-QM home loans. CoreVest lends to landlords and small developers through rental-property term loans and bridge loans. Redwood Investments holds the securities and loans the company keeps. A fifth segment, Legacy Investments, holds older assets being wound down, mainly bridge loans to residential investors made in 2021 and 2022. The business is a leveraged spread-and-fee operation. At June 30, 2026 it had $4.5 billion of recourse debt, about 5.0 times its invested capital, most of it warehouse lines that carry loans for a few weeks until they are sold or securitized. Above that sits what we estimate at close to $1 billion of corporate debt: senior notes due 2029 and 2030 at 9.00% to 9.50%, $125 million of 9.75% notes due June 2031 issued in May 2026, $185 million of 7.00% convertible notes due September 2030 issued in September 2026, and the unrepurchased remainder of the 7.75% convertible notes due June 15, 2027, which we estimate at about $177 million. There is also $67 million of 10% preferred stock. Borrowing is the raw material of a lender, so we value Redwood on its earnings after interest and preferred dividends and do not subtract its debt a second time. The business value in each case below is therefore already what is left for common shareholders after its lenders, and the leverage is charged in the bear case instead. Earnings. Redwood's operating cash flow is dominated by billions of dollars of loan purchases and sales, so it says nothing about profitability. We measure owner earnings as net income available to common shareholders, with stock compensation treated as a real cost and fair-value markdowns left in, because for a leveraged lender they are mostly real credit losses. Per-share earnings over the last ten years were: 2016 $1.54, 2017 $1.60, 2018 $1.34, 2019 $1.46, 2020 -$5.12, 2021 $2.25, 2022 -$1.39, 2023 -$0.08, 2024 $0.36 and 2025 -$0.59. The earlier years were earned on a different business and a much larger equity base per share, so we build the normal level from the business as it now stands. In the first half of 2026 the three lending businesses earned $76.8 million after preferred dividends, Redwood Investments lost $7.3 million, Legacy lost $36.4 million and corporate overhead cost $43.3 million. The core businesses, after their share of corporate cost, earned about $33 million in the half, or about $66 million a year. From that we take the overhead that will stay behind as Legacy disappears and about $10 million a year of extra net interest on the May and September borrowings. That leaves a normal level of about $48 million, or about $0.40 a share on roughly 120 million shares. We give no credit for production running at about $8 billion a quarter in the first half, which was close to a record and was reached before a sharp rise in interest rates in the third quarter. The company's chief executive described that quarter as the sharpest selloff in the 10-year Treasury in 32 years, and CoreVest was already reporting a weaker term-loan pipeline from higher rates in the second quarter. The dividend. Redwood pays $0.72 a share a year. That is well above the roughly $0.40 we think the business can earn, and above anything it has earned since 2021. The company expects 59% of its 2025 common dividends to be treated for tax purposes as a return of capital: most of the payout was shareholders' own book value being handed back, which is a large part of why book value per share keeps falling; it was $6.90 at June 30, 2026, down from $7.12 at March 31. Its annual report warns that dividends may be discontinued if they would represent a return of capital. Our base case assumes the dividend is reduced to roughly what the business earns. This matters less to value than it might seem, because a dividend paid above earnings is mostly capital coming back to the owner rather than income. Recent results. Owner earnings fell from $0.36 to -$0.59 a share in 2025, and the first half of 2026 was a further loss of $0.10. The base case assumes about -$0.15 for the rest of 2026, which includes the roughly 2% hit to book value the company expects from selling up to $190 million of Legacy bridge loans in the third or fourth quarter. It then assumes $0.25 a share in 2027. That improvement rests on dated facts: Legacy, which alone cost about $0.29 a share in the first half, had shrunk to under 10% of capital after third-quarter activity, from 15% at March 31, and the bridge-loan sales take out much of what remains. Base case: about $5.79 a share. Earnings of $0.25 in 2027 rise to $0.40 in 2028 and stay there, with one weaker year of $0.25 in 2030 for an ordinary mortgage-market slowdown, because Redwood's own record shows earnings rarely run smoothly. After year ten we assume about $0.35 a year continues, resting on the Sequoia and Aspire conduits' established place with bond investors: they priced nine Sequoia securitizations in the second quarter of 2026 alone, and the company completed more than 20 securitizations in the first half. We credit no growth from market share, Aspire's expansion or new joint ventures, because none is yet a demonstrated, repeatable process. Bear case: about $0.60 a share. Higher rates keep earnings weak into 2027, a year in which the remaining 2027 convertibles of about $177 million must be repaid against roughly $190 million of cash at the end of September 2026. Then, around 2029, the business suffers a mortgage credit and funding shock like 2022-2023, when it lost $1.39 a share and then $0.08: here a loss of $1.50 a share and a further $0.60 the following year, from credit marks on retained securities and investor loans and funding stress on warehouse lines. This is deeper than 2022-2023 itself because close to $1 billion of corporate debt at 7% to 9.75% does not shrink when the business does. Losses over the case total about $2.60 a share, roughly 39% of today's book value, and recovery comes on a diluted base at about $0.25 a share, with about $0.20 continuing after year ten. A 2020-style seizure, when the company lost $5.12 a share, is the unmodelled outcome that would leave shareholders with little or nothing. Bull case: about $11.01 a share. Legacy is resolved by the end of 2026, production holds near the first-half 2026 pace with the lending businesses earning the 20%-plus returns on capital they reported for eight straight quarters, and investment marks turn positive. Earnings reach $0.50 in 2027 and $0.70 a year after that, with about $0.65 continuing after year ten. What changed since our previous estimate. Our previous estimate did not fully reflect the company's 2026 financing. The company added $125 million of 9.75% notes in May and, in September, $185 million of 7% convertibles, using part of the proceeds to repurchase about $124 million of its 2027 notes and 5.5 million shares for about $20 million. That pushed most of the 2027 maturity out to 2030 and bought shares well below book value, both helpful. But it also left the company with corporate debt that costs about as much as, or more than, its core businesses currently earn on their capital, and the third-quarter rate shock has made the production pace of early 2026 less dependable. Together these lower our normal earnings level from about $0.55 to about $0.40 a share and deepen the bear case. Separately, a significant shareholder, Bradley Radoff, publicly urged the board in September to sell the company and said he would nominate directors at the 2027 annual meeting if it does not. He cited $21.3 million of 2025 pay for the named executive officers. We give no value to a possible sale. Margin-of-safety price: $2.50. We are not very sure of the earnings level: the current business has not produced a profitable year with the Legacy book gone, book value has fallen for several quarters, and third-quarter results due November 3, 2026 will show how the rate shock landed. The worst plausible outcome is another sudden mortgage-funding seizure. With $4.5 billion of recourse debt and close to $1 billion of corporate debt, that could force sales at bad prices and dilutive capital raising, and common shareholders lost a large share of book value in both 2020 and 2022. If we are wrong about the two things we are most likely to be wrong about, so that normal earnings are only $0.25 a share and the first-half production pace does not return, the shares are worth about $3.88. From $2.50 over a three-year holding period, that outcome implies about 23% a year including a $0.25 dividend. If the base case plays out with the dividend cut to about $0.40, it implies about 41% a year. In the bear case the shares lose money from almost any price, which is why the price must sit well below the base value. At $2.50 the price is about 37% of an estimated book value near $6.70 a share (June 30 book adjusted for the loan sales, buyback and third-quarter dividend, before third-quarter earnings). The price assumes about a three-year holding period with a reduced but still meaningful cash dividend.