Greystone Housing Impact Investors LP (GHI) stock intrinsic value fair value estimate & owner earnings analysis

Greystone Housing Impact Investors is a partnership that lends to affordable multifamily housing. Most of its roughly $1.39 billion of assets is about $891 million of tax-exempt mortgage revenue bonds and $37 million of governmental issuer loans, which fund the construction and preservation of apartments with rent limits for lower-income tenants. It also holds about $156 million of equity in joint ventures that build market-rate and seniors apartment properties, and $111 million of apartment properties it owns directly, four of which it took back from defaulted bond borrowers in early 2026. It funds these assets with leverage of about 74% of cost, mostly securitized term financing and credit lines from banks such as Mizuho and Barclays, hedged with interest-rate swaps. It also issues redeemable preferred units, about $102 million, which rank ahead of the common units valued here. Roughly half of the bond and loan principal is on properties in California. How we measured earnings. Because this is a lender, interest is a cost of doing business, so we value the units' cash earnings after interest and do not subtract the borrowings a second time. We started from operating cash flow rather than reported net income, which swings with non-cash marks on bonds and swaps. From operating cash flow we removed timing swings in receivables and payables, including cash that came in or went out of the joint ventures' accrued preferred returns. We also subtracted unit-based pay and the distributions owed to preferred unitholders. On that basis, earnings per common unit over the past ten years ranged from $0.67 to $1.30, with a middle value of about $0.80. In 2024 and 2025 the figures were $0.68 and $0.72. Why we no longer anchor on the ten-year middle. That ten-year history included steady gains from selling joint-venture properties. Those gains show up as investing cash, not operating cash, but they paid for much of the distribution, which ran around $1.54 to $1.85 a unit in cash paid from 2023 to 2025. The business has changed. In 2025 the joint ventures produced $12.5 million of losses and a $13.3 million reversal of previously accrued preferred returns. The partnership booked a $9.8 million credit-loss provision. It took four properties back through deed-in-lieu of foreclosure, and the quarterly distribution fell from $0.30 a unit in mid-2025 to $0.14 a unit by mid-2026. Management is now selling the market-rate joint-venture stakes and putting the money into bonds. Cash available for distribution in the first half of 2026 was $0.24 a unit, against the $0.28 of distributions declared, and operating cash flow was only $1.1 million after a large paydown of payables. Credit recoveries of about $2.0 million and $2.2 million of deed-in-lieu gains helped that half-year. We therefore set the earning power of the business as it exists now at about $0.55 to $0.60 a unit. That figure is a portfolio of about $930 million of bonds and loans earning a spread over hedged financing, plus some income from owned properties, minus about $1.1 million of preferred distributions a quarter. It is below the ten-year middle and above the recent half-year run rate. Last year's owner earnings grew about 6% to $0.72. We do not carry that forward, because the 2026 first half fell well below it. Base case, about $10.30 a unit. The partnership collects about $0.14 for the rest of 2026 and about $0.58 in 2027 as redeployment from joint-venture sales offsets the income lost from them. Earnings then hold flat at about $0.60 a unit. We credit no growth. Growing the bond book would require new common units or more preferred units. The partnership's history shows that brings in capital but not higher earnings per unit, and its leverage is already near the board's 80% ceiling. Bonds that mature or prepay have historically been replaced with new affordable-housing bonds from the Greystone origination platform. That decades-long record is what keeps the earnings going past year ten, so we do not fade them. Bear case, about $5.30 a unit. We include one full credit cycle like the multifamily stress of 2009 to 2011, when tax-exempt affordable bonds on lease-up properties defaulted. This partnership went through that cycle itself, and the deed-in-lieu actions already taken in 2025 and 2026 show the same stress mechanism working now. We assume about 6% of the $930 million bond and loan book ends in losses, roughly $58 million or $2.50 a unit, spread over 2028 to 2030. Interest on the troubled assets also stops, costing about $0.22 a unit a year. Earnings are negative in those three years. Distributions would be suspended, but the business would not need to raise equity, because the covenants test leverage on cost and the term financing is matched to the bonds. Earnings then recover to about $0.46 a unit on a smaller book. This case is still above zero because most of the debt is term securitized financing matched to specific bonds rather than short-term funding that must be rolled. A total loss on the junior joint-venture equity of about $6.70 a unit is not in this case. It would take a regional apartment price collapse far deeper than the bond losses assumed here. Bull case, about $12.89 a unit. The remaining eight completed market-rate joint-venture properties sell over 2027 to 2029 at gains similar to the partnership's past joint-venture sales, continuing the program begun with the Vantage at Loveland sale in 2026, which adds about $0.22 a unit a year of realized gains. The redeployed capital earns full bond spreads, taking steady earnings to about $0.72 a unit, close to 2024 and 2025 levels. We credit no growth beyond that. How sure are we of the earnings level? Only moderately sure. The bond book itself has been dependable: every bond and loan was current at June 30, 2026, and credit losses have historically been small. The uncertainty is whether the joint-venture equity, now being wound down, produces gains or more losses. The 2025 losses and the falling book value, $11.77 a unit at the end of 2025 and $11.20 at June 30, 2026, show it has recently been a source of losses. The CFO resigned in June 2026, and that adds some uncertainty to the reporting. The worst plausible outcome is a California-focused downturn in affordable rents and property values that hits the bonds and the joint-venture equity together, while lenders tighten advance rates. With 74% leverage, a covenant tied to leverage and to remaining listed, and preferred units ranking ahead, the common could be crushed in that event, though the matched term funding makes outright failure unlikely. If we are wrong about the earnings level, say a steady $0.45 rather than $0.60, and if the joint-venture equity is written down by half, the units would be worth roughly $5 to $6. The price assumes a three-to-four-year holding period. At the current distribution of $0.56 a year the business pays about 14% a year on the price below, if the distribution is maintained. Margin-of-safety price, $4.00 a unit. At that price the distribution is about 14% a year while waiting, if maintained. If the base case plays out, the units' worth would roughly double within a few years. The bear case implies roughly the price back plus several years of distributions before the credit cycle. Even in that case, the business's value after the cycle is above $4.00. The bear outcome must still work out acceptably over a three- to four-year holding period, and $4.00 is about the highest price at which it does. Changes since the prior analysis. Our previous estimates assumed the ten-year middle level of cash earnings of about $0.79 a unit would continue. The 2026 interim results changed our view. The joint ventures are being sold rather than held for gains, and four bonds became owned properties. Half-year cash available for distribution was $0.24 a unit, and book value fell further. All of these lower the level we can defend for the business as it exists now. Two things are better than we expected: there were credit recoveries in 2026, and the bond and loan book was fully current at mid-year.