When this workfile was prepared, the app had read a half-finished 2026 year as Lennar’s latest full year, and took book value from total equity, which counts minority interests. Corrected, at the same September 11 inputs ($79.60 and a 4.83% AAA yield):
The three-year declines in earnings, revenue and book value, the negative owner earnings and the section 12 tests are unchanged. The rest of this workfile is left as it was published.
America’s second-largest homebuilder trades below the value of what it owns, passes seven of Graham’s eight tests, and fires five of eight warning signs. Berkshire Hathaway has been buying it for two years and is underwater. So are we.
Thirteen instruments, pointed at one company, and they do not agree. The spread is the finding.That disagreement is not noise, and it is not a flaw in the tools. It is what a land-heavy cyclical looks like when you measure it from several directions at once. The methods built on assets and reported earnings say Lennar is materially cheap. The methods built on cash say the opposite — owner earnings are negative and cash return on capital is essentially nil. Both readings are correct. A homebuilder consumes cash to grow, because it buys land years before it sells a house.
A financially conservative survivor of the last housing collapse, trading below book value in the middle of a rate-driven earnings trough, being accumulated by the most patient buyer in the world. Cheap on assets, weak on cash, and entirely dependent on a rate cycle nobody can time. Sized accordingly, bought in slices, and held to a written test.
Lennar builds houses and sells them, which makes it about as easy a business to understand as exists on a public exchange. Founded in 1954. It delivered roughly 11.7% of all new single-family homes sold in America in 2024, second only to D.R. Horton at 13.6%.
| Builder | Share of 2024 closings | Note |
|---|---|---|
| D.R. Horton | 13.6% | 93,311 closings |
| Lennar | 11.7% | The subject |
| PulteGroup | 4.6% | |
| NVR | 3.3% | |
| Meritage | 2.3% | |
| Taylor Morrison | 1.9% | Acquired outright by Berkshire, July 2026 |
The top three are 29.9% of the market and the top ten are 43.6%. Everyone below second place is a rounding error by comparison, and that concentration is not an accident — it is the product of the last downturn, which is where this file has to start.
Anyone who lived through 2008 flinches at the word homebuilder, and they are right to. Here is what it did to this company.
The share price told the same story more violently. Lennar closed at $67.81 in July 2005 and $6.08 in March 2009 — a fall of 91%. Its land joint venture, LandSource, filed for bankruptcy protection in June 2008. In February 2009 an analyst published a piece putting 72% odds on Lennar itself going under.
It didn’t. And then it ate.Lennar came through, and bought LandSource’s assets out of bankruptcy court. These did not come through: TOUSA (January 2008, $2.24 billion of debt, the largest builder bankruptcy of the cycle), WCI Communities, Levitt & Sons, Kimball Hill, Neumann Homes, Kara Homes.
Downturns are the mechanism by which large builders get larger. The levered ones fail, their land is sold cheaply, and whoever is still standing buys it. Lennar is a top-two builder today because of what 2008 did to its competitors.
So the question is not whether Lennar can survive a bad housing market. It is whether it is financed to be a buyer in one. On that, the balance sheet has been rebuilt into the opposite of what it was: current ratio 4.66, debt-to-equity 0.29, and long-term debt of $3.9 billion against net current assets of $11.6 billion.
No, and the numbers are not close. This is the objection every reader will raise, so it gets answered with data rather than reassurance.
| 2008 | Today | |
|---|---|---|
| Mortgages worth more than the house | 23–24% | 1.6% |
| Seriously behind on payments | ~25% of subprime ARMs | 1.2% of all mortgages |
| Homes in foreclosure | 1.5M starts in 2007 alone | 0.4% of mortgages |
| Homeowner equity | Collapsing | Record high |
They are opposite problems. 2008 was a glut — too many houses, built on too much bad credit, and it took years to clear. Today prices are still rising, roughly 2% over the year, and the trouble is that almost nobody holding a 3% mortgage will list their home. That starves the resale market and pushes buyers toward new construction, which is a builder-favourable dynamic hiding inside a housing-negative headline.
The level is fine. The direction is not. Foreclosure inventory has reached a six-year high, serious delinquency ticked up from 1.0% to 1.2% over the year, and the strain is concentrated in FHA and VA loans among buyers from 2022 onward. Nothing resembling 2008 — but worsening at the edges, not improving, and worth watching every quarter.
This is the whole toolkit, run top to bottom, at the September 11 close — the end of a week in which the shares touched $76.63, the lowest they have traded in a year, before recovering to close at $79.60. Nothing is averaged into a single score, because a score cannot tell you what it is hiding.
| Method | Reading | What it says |
|---|---|---|
| Graham Number | $126.92 | 37% below fair value |
| Graham 1962 formula | $165.32 | Deeply undervalued |
| Graham 1974 formula | $150.61 | Deeply undervalued |
| Earning Power Value (B) | $225.26 | Extremely undervalued |
| Earning Power Value (A) | — | No discount rate available |
| Net-Net / NCAV | $19.55 | Nowhere near a net-net |
| Discounted cash flow | — | No discount rate available |
| Reverse DCF | — | No discount rate available |
| Lynch PEG | 2.00 | Expensive against growth |
| Acquirer’s Multiple (EV/EBIT) | 8.10× | Moderately cheap |
| Buffett owner earnings | −$1.46/sh | Negative — no ceiling can be set |
| Return on capital | ROIC 8.4% · ROCE 9.7% | CROIC 0.11% — effectively nil |
| Piotroski F-Score | 4 of 9 | Weak financial momentum |
| Altman Z-Score | 3.53 | Safe zone, no distress |
| Beneish M-Score | −2.36 | No sign of manipulated earnings |
Three groups emerge, and the reason they split is the single most instructive thing about this company.
The asset and earnings methods are bullish because Lennar has real book value and, over a full cycle, real earnings. The cash methods are bearish because a homebuilder buys land years before it sells a house, so growth consumes cash by design — owner earnings of −$1.46 and a CROIC of 0.11% are what that looks like on a spreadsheet. The safety checks are clean: Altman says no distress, Beneish says nobody is cooking anything, and the balance sheet supports both.
Cheap on what it owns. Poor on what it generates. Safe on whether it survives.| Test | Result | Detail |
|---|---|---|
| Adequate size | Pass | ~$19.8B market capitalisation |
| Current ratio ≥ 2 | Pass | 4.66 |
| Long-term debt < net current assets | Pass | $3.9B against $11.6B |
| Positive earnings, ten years | Pass | Every year in the window |
| Twenty years of dividends | Pass | Paid in all 28 years on record |
| Earnings growth over ten years | Fail | EPS down roughly a third, not up a third |
| P/E ≤ 15 | Pass | 12.5 |
| P/B ≤ 1.5 | Pass | 0.89 |
Seven of eight. The one failure is growth, and it is failing for a reason we can name and date rather than a mystery.
The app fires five of its eight flags on Lennar. All five are printed here, because a workfile that hides them is a brochure.
| Flag | Status | Our reading |
|---|---|---|
| Earnings declined over three years | Firing | True, and the central issue. See section 7. |
| Revenue declined over three years | Firing | True but mild — roughly 5% year over year. |
| Book value per share declined | Firing | Substantially explained. See below. |
| Dividend cut in the history | Firing | True — cut to $0.16 through the 2009–2019 period. |
| Negative owner earnings | Firing | True, and structural to homebuilding. |
| Negative free cash flow | Clear | |
| Unprofitable | Clear | |
| Debt exceeds book equity | Clear |
Book value per share fell from $103.01 in fiscal 2024 to $85.90 in 2025. That looks like erosion. It is mostly a corporate action.
On February 7, 2025 Lennar spun off Millrose Properties, contributing approximately $5.5 billion of land — some 87,000 homesites — and about $1.0 billion of cash, distributed to shareholders at one Millrose share for every two Lennar shares. The stated purpose was to turn Lennar into an asset-light, pure-play home manufacturer that options finished lots just in time rather than warehousing raw land.
So shareholders did not lose that value; they were handed it in a separate security. An automated flag cannot know that. An appraiser has to.
Our first draft said the thesis breaks on a further land write-down. After the Millrose spin-off that is a weaker test than it was, because far less raw land now sits on Lennar’s balance sheet to be written down. The right test for an asset-light builder is deliveries and gross margin, not land impairment. The revised condition is in section 9.
| Third quarter | Earnings per share |
|---|---|
| 2024 actual | $4.26 |
| 2025 actual | $2.00 |
| 2026 estimate | $1.29 |
Revenue is down roughly 5% year over year. Earnings are down about 36%. Those two numbers together are the entire diagnosis: houses are still selling, but Lennar is paying to sell them. Builders buy down their customers’ mortgage rates to keep volume moving, and that subsidy comes straight out of gross margin.
This is not a demand collapse. It is a margin transfer from the builder to the buyer, funded by the builder.It also means the usual cyclical trap does not apply here in the usual direction. The classic error is a low multiple on peak earnings. Lennar’s earnings are already about 70% below their 2022 peak, so 12.5 times is a multiple on depressed earnings. The risk runs the other way: if the trough deepens, the multiple rises while the price falls.
No filing states an investor’s cost. A 13F lists holdings and quarter-end values; Berkshire’s own reports disclose cost by broad category, never by name. But every quarter’s share count is public, so the accumulation can be traced exactly.
Their answer to being wrong on the price has been to buy more, because the thesis was never about this year. That is either conviction or stubbornness and it is too early to say which — but it tells you exactly what kind of position this is meant to be.
Greg Abel became chief executive on January 1, 2026. On May 31 Berkshire agreed to acquire Taylor Morrison, America’s sixth-largest builder, outright at $72.50 a share in cash — about $6.8 billion of equity value and a 24% premium. It closed July 24. In the same quarter Berkshire lifted Lennar by 29.8% and reopened a position in D.R. Horton. It already owned Clayton Homes and one of the largest residential brokerages in the country.
Nobody at Berkshire has explained the Lennar purchase specifically. The closest thing to a stated reason came at the 2026 annual meeting, where Abel described high mortgage rates squeezing buyers at Clayton Homes and framed Clayton’s purpose as delivering an affordable home to the American consumer.
On a look-through basis Berkshire owns roughly 2.6% of new single-family closings — Taylor Morrison’s 1.9% owned outright, plus about 0.7% through its 6.2% of Lennar. They are not consolidating the industry. They are taking a concentrated position in it.
And the sizing is instructive. Berkshire’s stock portfolio was $299.3 billion across 29 positions at the end of June. Lennar is 0.41% of it. Apple is 22%. A firm that just paid $6.8 billion cash for a whole homebuilder holds this one at four-tenths of one percent.
Three named risks, in the order we weight them.
The 30-year fixed mortgage sat at 6.76% on September 10, up from 6.65% three weeks earlier — it rose in each of those three weeks, into the week the shares made their low. If yields keep climbing, mortgage rates follow, buyers keep being priced out, and no figure in this file survives it. This is the dominant variable and it is entirely outside the company’s control.
Every argument for housing rests on new households forming, and that flow has thinned sharply.
| Measure | Reading |
|---|---|
| Net international migration, 2023–24 | 2.7 million |
| Net international migration, 2024–25 | 1.3 million |
| Census projection, 2026 | ~321,000 |
| Total population growth, 2024–25 | 0.5% — slowest since 2021 |
No political point is being made here; it is simply the demand input. And it lands hardest on this company, because of where it builds: roughly 25% of deliveries in Florida, 20% in Texas, 18% in California — about 63% in three states, with Houston, Dallas, San Antonio, Tampa and Phoenix its largest markets. Those are precisely the places that absorbed the last decade of migration.
A housing shortage closes on its own if the people stop coming.Analysts expect $1.29 a share on about $8.3 billion of revenue, after the close on Wednesday. A weak number alone does not break the thesis; everyone already expects a weak number.
What the report will do mechanically, however, is worth setting out before it happens rather than after. Section 4 values this company on fiscal 2025 earnings of $7.98 a share — the last completed, audited year. Trailing twelve-month earnings are already lower, at $6.39. On September 16 the $2.29 Lennar earned in the third quarter of fiscal 2025 leaves the twelve-month window and the $1.29 expected for this one replaces it. Meet the estimate exactly and trailing earnings fall to about $5.39.
Every earnings-based figure in this file moves with that. A Graham Number computed on trailing earnings rather than the audited year is about $104 instead of $126.92, and the price-to-earnings ratio of 12.5 becomes roughly 14.8 — at Graham’s ceiling rather than under it. None of that is new information about the business; it is the calendar moving a strong quarter out of a twelve-month window. It is recorded here because a reader who recalculates after the report should find the arithmetic already accounted for rather than discover it as a surprise. It is also a concrete instance of the risk section 7 names: on a cyclical company past its peak, the multiple rises as earnings fall, so the shares can get cheaper and look dearer in the same week.
What would tell us we are wrong: a fall in homes actually delivered, or further compression in gross margin on those deliveries. Softer prices per house we can live with; that is the cycle. Fewer houses sold, at worse margins, is the thesis breaking.
What we are no longer testing: land write-downs. After the Millrose spin-off, Lennar carries far less raw land, so impairment is a weaker signal than it would have been two years ago. Recorded here because the first version of this file got that wrong.
| Considered | Rejected because |
|---|---|
| KB Home instead of Lennar | Cheaper on the screen — 47% margin of safety, 0.81× book — but a third the size, less recognisable to a general reader, and without the Berkshire confirmation. |
| Meritage Homes | Also passes the screen. Same objection as KB Home, with less brand recognition again. |
| Leading on “below book value” | True and striking, but Berkshire’s year is the stronger hook and it carries the reader into the same conclusion. |
| Calling this an arbitrage by Berkshire | Owning Taylor Morrison outright while holding 6.2% of a competitor invites the theory. It does not survive the arithmetic: acquiring Lennar would mean paying a premium on 93.8% of the shares to mark up the 6.2% already held. |
| A hostile takeover of Lennar | Structurally impossible. Dual-class shares give Class B ten votes; the founding family controls roughly 41.9% of the combined vote. Berkshire holds the low-vote Class A. |
| Quoting the app’s blended $138 fair value as the headline | It is a genuine output, but it is pulled upward by growth formulas built for a different kind of business. The conservative Graham Number is the more defensible number to lead with. |
| A full-position recommendation | Earnings land three days after publication. Sizing and staging are the responsible answer, not conviction. |
The author owns Lennar at $82, and is down roughly 3% at the date of value. No shares were bought or sold in the ten days before publication. If the price falls a further 10% or so and the September 16 report leaves the test in section 9 intact, the intention is to add.
Buying more because you like the price is not the same as buying more because the business is sound. The only thing separating them is whether you wrote the test down before the fall.Berkshire itself demonstrated both halves of that inside one quarter in 2009 — buying Goldman Sachs, Harley-Davidson and Swiss Re hand over fist while selling ConocoPhillips at roughly $50 against an $82 cost, having written to shareholders that buying oil at its peak was his own major mistake. Same man, same quarter, opposite actions, because in one case the reason held and in the other it had gone.