AMREP CORP. (AXR) stock intrinsic value fair value estimate & owner earnings analysis

AMREP owns about 16,200 acres in Sandoval County, New Mexico, centered on Rio Rancho, the state's third-largest city. It makes money two ways: it entitles and develops residential lots and sells them, almost entirely to a handful of homebuilders, along with commercial sites and raw acreage; and it builds and sells its own homes in the same area. Land sales carry high margins (about 61% gross in fiscal 2026) and homes lower ones (about 24%). The company has essentially no debt (about $17,000 of notes) and held $48.7 million of cash at July 31, 2026, down from $52.3 million at April 30 because the June quarter was weak. It has paid no dividend since 2008. How we measured earnings. We start from cash generated by operations, remove short-term swings in payables, taxes and other assets, and deduct stock compensation as a real cost. We do not remove spending on land inventory, because a land developer must keep developing what it sells; that spending stays in as a cost. Equipment spending is negligible. For each of fiscal 2017 to 2026 the operating cash flow rebuilt line by line from net income matched the reported figure. Owner earnings were $10.7M, $9.0M, $12.2M, $9.4M, $10.1M and $12.6M for fiscal 2021 through 2026. Fiscal 2017 to 2020 are not used: those years included the former magazine-fulfillment business (sold in 2019), pension payments and settlements, and a land business earning roughly nothing. Because we count the cash pile separately, we also strip out interest earned on it ($1.7M in fiscal 2026), leaving business earnings of about $8.4M to $12.1M a year since 2021, with a median of about $9.85M, or roughly $1.83 per share on 5.39 million diluted shares. There is no interest-bearing debt, so there is no interest to add back. Adjustments to that level. The fiscal 2025 and 2026 figures each include one bulk sale of raw land to a single buyer (549 acres for $2.5M and 467 acres for $2.2M), which the company says it does not expect to be indicative of future undeveloped land sales. We treat roughly $1.8M a year of after-tax cash flow from such sales as non-recurring after the next two years. The company also has $20.6M of federal net operating losses with no expiry, which keep cash taxes low for several more years; that benefit is already reflected in recent cash flows and runs out within the forecast, which is part of why we step the level down. Growth path. Owner earnings rose about 25% in fiscal 2026. We assume no growth in any case, well below that rate, because the gain came from lumpy lot timing and a jump in land margin (61% versus 52%) that we cannot count on repeating, and the June 2026 quarter showed revenue of $6.1M versus $17.9M a year earlier and operating cash outflow of $3.6M. We credit nothing for new land purchases, expansion beyond New Mexico, rental properties or the mineral rights under about 55,000 surface acres, none of which has produced evidenced earnings. How long the earnings last. The land bank is finite: about 15,300 acres undeveloped and about 900 acres in subdivisions. Fiscal 2026 land sales were about 590 acres including a bulk sale, so the holdings cover roughly 25 years at that pace, and longer if raw acreage is sold more slowly. We credit 25 years in the base and bear cases, 30 in the bull, and nothing afterwards, because nothing we can name produces earnings once the land is gone. Base case. Business cash flow of about $9.85M a year for two years, then about $8.0M a year once the bulk raw-land sales stop recurring, for 25 years in total. That business is worth about $106M. It owes essentially nothing. Of the $48.7M of cash we hold back $10M as working capital for lot development and homes under construction, so about $38.7M is excess. Total about $144M, or $26.76 per share. Bear case. The bear case uses the downturn this company has actually lived through. After 2008 the Rio Rancho housing market collapsed; by fiscal 2016-2017 single-family starts were still only about 530 a year (versus 1,139 in 2021), developed land sales were around $5-9M a year at gross margins of 14-18%, and the land business earned little after overhead for most of a decade. We model five years of slightly negative cash flow (about -$0.6M a year, roughly corporate overhead not covered by thin land sales and few home closings), two recovery years, and then $4.7M a year, about half the recent median, reflecting the concentration of lot sales in three builders and a home-building operation that would shrink in a slump. That business is worth about $36M; with no debt and $38.7M of excess cash the total is about $74M, or $13.80 per share. Because there are no borrowings, the company can carry this case without lenders; the cash burn would be absorbed by the cash pile. Bull case. Land margins near 61% and home margins near 24% hold and volumes stay at the best year on record (fiscal 2023, about $12.1M) for two years, then about $10.0M a year after bulk raw-land sales stop, over 30 years as the land is sold more slowly at higher values. That business is worth about $142M; with no debt and $38.7M of excess cash, about $180M or $33.41 per share. What changed since our previous analysis. Three things moved values down: cash fell by $3.6M in the June 2026 quarter as sales dropped sharply; the 10-K states the recent bulk raw-land sales are not expected to recur, so we no longer count them in the long-run level; and we now look back to fiscal 2016-2017 to size how deep a Rio Rancho housing slump runs for this company, which lowers the bear case. Earlier figures built on a different basis are not comparable. Margin-of-safety price. How sure are we of the earnings level? Moderately: six years of $8.4M to $12.6M despite Rio Rancho starts falling from 1,139 to 588 and back, but land sales are lumpy, three builders take all developed lot sales, and the latest quarter was very weak. The worst plausible outcome is a repeat of the post-2008 decade, when the land was nearly unsaleable at a profit for years; the structure survives it because there is no debt, the land is a hard asset carried at old cost, and about $7 per share of excess cash sits on the balance sheet. A further risk no case models is management spending that cash on land or markets outside its proven area. If we are wrong about the margin level and the length of a slump, the shares are worth roughly the bear figure, about $14, most of it cash plus slowly sold land. The company pays no dividend, so the return depends entirely on the business value being recognized; the price assumes a three-to-four-year holding period. A $15 price covers roughly $7 of excess cash plus about $8 for a land business that has produced about $1.80 per share of cash in a normal year; if the base case plays out over three to four years, value rises toward the high $20s, which the base case implies is a return near 20% a year, while the bear case implies a value only slightly below that price. Above $15 that cushion disappears, so $15 is the margin-of-safety price. What would make this wrong. A prolonged New Mexico housing slump, loss of one of the three lot buyers, or capital spent on acquisitions outside Rio Rancho would lower the value; faster lot absorption at today's prices, or monetization of the mineral rights, would raise it. We could not fully reconcile why developed residential lot revenue fell from $21.9M to $13.8M in fiscal 2026 while total profit held, beyond higher home closings, a commercial land sale and the bulk raw-land sale making up the difference.