PennyMac Mortgage Investment Trust (PMT) stock intrinsic value fair value estimate & owner earnings analysis

PennyMac Mortgage Investment Trust is a mortgage real estate investment trust. It earns a leveraged return on US home-mortgage assets: mortgage servicing rights (the right to collect a fee for servicing a loan), first-loss credit-risk-transfer positions on about $19.5 billion of loans it bought and sold between 2015 and 2020, subordinate bonds it keeps from its own non-agency securitizations, and hedged agency and non-agency mortgage securities. PNMAC Capital Management, a subsidiary of PennyMac Financial Services, manages the trust externally. PennyMac Financial also services its loans and supplies loans for its securitizations. What changed since our last look. Two things happened in mid-2026 that the business has not done before. In June it stopped buying agency-eligible conventional loans through the correspondent channel (it still buys all non-agency loans). It also agreed to sell servicing rights on $13 billion of low-coupon agency loans, due to close at the end of August, which management called the first step in shrinking servicing exposure. Capital is moving from servicing rights into retained securitization bonds. Management says those bonds can return in the low-to-mid teens, but there is no earnings record behind that yet. Leverage rose: total debt was 12.4 times equity at June 30, 2026, up from 10.9 times in March. These changes lower our near-term earnings path. They are not reasons to raise it. How we measure earnings. Reported operating cash flow is not usable here: it swung from +$1.8 billion in 2022 to -$7.2 billion in 2025 because loans bought for resale count as operating cash. We exclude that loan inventory. Almost every remaining asset and hedge is carried at market value, so net income is the right measure of what the owners earn. Book value confirms it. Common book value plus dividends paid tracked net income per share closely from 2023 to 2025. The earnings level. Net income before preferred dividends was $153 million (2018), $226 million (2019), $52 million (2020), $57 million (2021), -$73 million (2022), $200 million (2023), $161 million (2024) and $128 million (2025). We use 2018-2025 because it covers the business in its present form, with servicing rights at scale, and a full rate cycle: the 2020-2021 refinancing wave, the 2022 rate shock and the recovery. The median is about $140 million a year, or $1.61 per share before preferred dividends and $1.13 per common share after them. Growth check. Net income fell 21% in 2025, from $161 million to $128 million. In the first half of 2026 it was $55.5 million: $24.6 million in the first quarter and $30.9 million in the second. That is about $111 million a year, with common earnings of $0.16 and $0.23 per share. Our base case starts below the trailing rate rather than above it. The first two years run at the current $111 million pace while the trust sells servicing and redeploys the capital, and only then return to the $140 million median the business earned across the cycle. No growth is credited in any case. The securitization program, any benefit from lower rates and management's return targets are left out because none of them is a demonstrated earnings record yet. What the trust owes. For a mortgage investor, borrowing is the raw material, and its interest is already deducted from net income, so the roughly $23 billion of repurchase agreements, secured notes, securitization-trust debt and $685 million of unsecured notes (8.50% due 2028 and 9.00% due 2030) are not subtracted again. Leverage is charged in the bear case. What does come off is the $560 million of preferred shares (8.125% Series A, 8.00% Series B, 6.75% Series C, $25 liquidation value each), which rank ahead of common shareholders. That is $6.42 per share. A class action (Verthelyi), now on interlocutory appeal at the Ninth Circuit, argues that Series A and B should have switched to a floating SOFR-based rate when LIBOR ended. The company has recorded no accrual. We carry it in the bear case as about $15 million of extra cost. Base case ($14.48 per share). Pre-preferred earnings are $111 million in each of the next two years, then $140 million a year through year ten, fading 3% a year after that. The fade reflects the external-management structure and common book value per share falling from roughly $21.50 in 2019 to $14.83 at June 30, 2026. What carries on is a 17-year record of creating and financing mortgage assets from about $1.85 billion of capital. The business is worth about $1.82 billion, or $20.89 per share. After the $6.42 per share owed to preferred holders, $14.48 is left. Bear case ($0.87). This case uses a full mortgage credit cycle, of the kind that came with the 2008-2010 house-price decline, on top of a rate shock. Year one repeats 2022, with a $73 million loss as servicing and hedge values collapse. Years two and three each lose $150 million. That is about $373 million over three years, or roughly 2% of the $19.5 billion of loans whose first losses the trust guarantees. Those losses wipe out much of the roughly $1 billion of credit-risk-transfer deposits and cut book value far enough that, at 12 times leverage, repurchase lenders would ask for capital. We assume $200 million of new common equity raised at trough prices, about $5 a share, adding 40 million shares. From year four, earnings recover only to $100 million a year, because the trust no longer has its servicing book or conventional correspondent flow and depends on PennyMac Financial for what it buys. After year ten they fade 8% a year. The business plus the new capital is worth about $685 million, or $5.38 per share on 127 million shares. It owes $4.52 per share to preferred holders, including the lawsuit exposure, which leaves $0.87. Common shareholders keep almost nothing in this case. Lenders and preferred holders are still covered, but only because new equity is raised. Bull case ($22.49). Year one runs at the $140 million median. After that the trust returns to its 2023-2024 average of about $180 million a year, a level it has already earned in current conditions. That holds through year ten and then fades 2% a year. The business is worth about $2.52 billion, or $28.90 per share, and $22.49 is left after the preferreds. The margin-of-safety price. We are only moderately sure of the earnings level: the eight-year median sits in a range running from a $73 million loss to $226 million of profit, and the trust is changing what it owns. The worst plausible outcome is a credit downturn that hits the first-loss positions while repurchase lenders make margin calls. With debt of 12 times equity, that could force asset sales or dilution, and the trust depends on an affiliate that controls both its management and its loan flow. The things we are most likely to get wrong: earnings may stay at today's $111 million pace rather than returning to $140 million, which makes the shares worth about $10.60; and that lower pace may also come with a faster fade and a loss on the preferred lawsuit, which makes them worth about $8.90. Today's $1.60 dividend is well above common earnings of about $0.79 per share at the current pace, so we assume it is cut to about $0.80 a year (an estimate). The price assumes a holding period of up to five years. From $6.00, with $0.80 a year of dividends, the base case implies about 29% a year over five years if the shares reach the base value, about 23% if earnings stay at today's pace, and about 20% in the case with a faster fade and the lawsuit lost. That is why the margin-of-safety price is $6.00 and not higher. The bear case implies a loss of most of that price. We accept that because it requires a credit cycle as deep as 2008-2010 to arrive while leverage is at its current peak. Even then, the preferred shares and new equity absorb losses before lenders are at risk. What would make this wrong. Higher: the redeployed securitization capital earns the low-to-mid-teens returns management cites, and servicing sales are completed at or above carrying value. Lower: losses on the 2019-2020 vintage loans that make up most of the credit-risk-transfer book, the Ninth Circuit siding with the preferred holders, or a less favorable arrangement with PennyMac Financial. The terms of the August servicing sale were not disclosed in the filings available to us, so any gain or loss on it is not reflected here.