The LAND Framework: How I Find Assets The Market Can't Properly Value

In 1888, the Texas and Pacific Railway ran out of money.

Its bondholders wanted to be paid. There was nothing to pay them with. So they were handed the only thing the railroad had left in quantity, which was dirt. Millions of acres in West Texas. It was dry, empty, and close to worthless to anyone who had to make a living on it.

A trust was set up with instructions to sell the acreage off over time and hand the proceeds back. Getting land instead of cash was widely understood to be the bad outcome.

What remains of that trust is now worth roughly $26 billion.

And here is the part that matters for this article.

Around 800,000 of those acres still sit on the balance sheet today at a carrying value of zero.

Not "understated." Not "conservatively marked." Zero. Nobody paid anything for them in a transaction the accounting system recognizes, and under U.S. GAAP a company cannot write land up. Ever. Land goes on the books at cost, it is never depreciated, and it is never marked to market.

So the financial statements will describe one of the great land positions in the country as worth nothing for as long as the company exists.

That is how the rules work.

Now here's the thing.

Once you understand that the rules produce this outcome by design, you stop treating it as a curiosity, and you start hunting for it. Because irreplaceable land attached to a business that gets paid for it is about as close to a permanent wealth-building asset as public markets offer. Nobody can print more of it. Nobody can compete the returns away with capital. And the accounting will keep telling you it's worth less than it is, year after year, for decades.

I have a framework for finding these. I call it LAND, and after a lot of requests to write it up properly, here it is.

Spoiler for those who want the scorecard first: Texas Pacific Land (TPL) scores highest at 92 out of 100. LandBridge (LB) is second at 81. Rayonier (RYN) comes in at 76. And Whirlpool (WHR), which I included on purpose, scores 11.

Let me explain why any of that matters.

What LAND Is, And What It Is Not

As many of you know, my primary framework is TOLL+M. Tangible assets, Oligopoly power, Low incremental capital intensity, Long-duration cash flows, and Macro alignment. That's the one I use to decide whether I want to own a business.

LAND is a different animal, and I want to be honest about its place in my process.

TOLL+M scores a business. It asks whether the economics are durable and whether I want to own them for a decade.

LAND scores an asset that the accounting system is describing badly. It asks a much narrower question: is there something on this balance sheet carried at a value that bears no relationship to what it's actually worth, and does that thing throw off cash?

That makes LAND a mispricing detector. It says nothing at all about whether the business is good.

This distinction is the whole ballgame, and it's where most people go wrong with asset-value investing. Every few years someone discovers that a struggling department store chain owns its real estate and the stock is "trading below the value of the buildings." That analysis is usually correct on the arithmetic. And it's usually a disaster, because a company burning cash will burn through the asset value long before anyone monetizes it.

Remember J.C. Penney? Or when everyone said that Macy's (M) real estate was a game-changer?

So LAND is a screen. TOLL+M is the filter that comes after.

I'll show you how the two work together shortly. First, the four tests.

L: Locked Supply

Can a competitor create more of this?

That's the entire question. And the answer has to be mechanical, not rhetorical.

If I had unlimited capital and ten years, could I replicate this asset? If yes, the test fails no matter how attractive the current economics look.

How I verify it. I refuse to accept "management says the land is irreplaceable." Especially in the REIT space, I often see comments from people who massively overstate the importance of the land and/or buildings owned by a company. I look for the specific barrier. Geology. A permitting timeline. A zoning regime. A state licensing framework. A 19th-century right of way that would be politically impossible to assemble today.

Aggregates is my favorite example because the barrier is so boring and so total. Getting a new greenfield quarry permitted in a core U.S. state is a multi-year process involving zoning, environmental review, water impact studies, and a local population that would rather not live next to a rock crusher. Capital does not solve this. Time barely solves it. Think of Vulcan Materials and Martin Marietta Materials (MLM).

Same logic on rail. You cannot build a competing mainline across the eastern United States in 2026. The right of way does not exist, and the environmental review would outlive everyone reading this.

A: Accounting Gap

Is the carrying value materially below what the asset would fetch?

This is the test that produces the actual opportunity, and it's the one people skip because it requires opening the 10-K. And as much as I like finance and data, working my way through a 10-K isn't my favorite thing in the world, either.

How I verify it. I find the specific line item in the property note. Not total assets. Not book value per share. The line that says "Land," or "Timber and timberlands," or "Land and quarries." Then I check when the acreage was acquired.

And then I apply the rule that most people miss.

The accounting gap has a half-life.

Every acquisition kills it. When a company buys land, purchase accounting requires that land to be recorded at fair value on the acquisition date. The gap resets to zero on everything acquired. If you think about it, this essentially means a serial acquirer of land, however excellent the business, structurally cannot have a large accounting gap. The accountants keep closing it for them.

This is why vintage matters more than acreage. Eight hundred thousand acres acquired in the 1880s and eight hundred thousand acres acquired in 2024 are the same land and a completely different investment case.

N: Nominal Linkage

Does the revenue from this land reprice with inflation, without requiring new CapEx?

Land is only interesting to me if the cash it produces moves up with the price level. Otherwise I own a fixed nominal stream sitting on top of a hard asset, which is a bond with extra steps. There's a huge difference between owning land that's just land and land that actually produces valuable cash flows with inflation protection.

How I verify it. I look for royalties expressed as a percentage of gross revenue, market-rate rent resets, commodity linkage, or fee structures tied to volumes. A royalty on gross is the purest form of this. When the commodity price doubles, the royalty doubles, and the owner spent nothing.

There's just one problem with the way people apply this test.

A long lease is not automatically good here. A twenty-year triple-net lease in the REIT space with a fixed 2% annual escalator will underperform inflation in exactly the environment where you wanted land exposure in the first place. Great real estate, weak nominal linkage. Those two things coexist all the time.

Realty Income (O), for example, doesn't disclose where these rent escalator ceilings are (I know they are somewhere around 3.5% to 4.0% from most peers), but this is what it said in its 10-K:

Leases generally provide for limited increases in rent as a result of fixed increases, increases in the consumer price index, retail price index in the case of certain leases in the U.K. (typically subject to ceilings), or increases in clients’ sales volumes. We expect that inflation will cause these lease provisions to result in rent increases over time. During times when inflation is greater than increases in rent, as provided for in the leases, rent increases may not keep up with the rate of inflation and other costs. - O 2025 10-K

NNN REIT (NNN) is making the same case without revealing where its upper limits are:

NNN’s leases typically contain provisions to mitigate the adverse impact of inflation on NNN’s results of operations. Tenant leases typically provide for limited increases in rent as a result of fixed increases and/or capped increases in the Consumer Price Index. As a result of limitations on rent increases, during times when inflation is high, rent increases may not meet or exceed the rate of inflation. - NNN 2025 10-K

I am not saying that Realty Income or NNN are bad, especially because it benefits from a net lease structure, but elevated inflation can substantially hurt one's real income growth from these companies on a prolonged basis.

D: Dual Use

Can the same acre earn from more than one payer?

This is the optionality test, and it's the one that has changed the most in the last three years.

How I verify it. I count the revenue streams in the segment disclosure, and I ask whether they stack on the same physical ground. Timber and solar. Grazing and water. Oil and gas royalties and produced water handling and pipeline easements. One acre, multiple checks. That's when a lot more value gets unlocked.

Dual use is what converts a static asset into a compounding one. And it's where the power and data center buildout has quietly rewritten the math on rural land across the U.S.

• Rayonier gave us a clean number on this during its 2Q call. The company sold a 460-acre bolt-on parcel to a solar developer for $4.6 million. That is roughly $10,000 per acre.

• Now compare that to what timberland actually costs. In the same period, Rayonier completed a like-kind exchange that included acquiring about 57,000 acres in Alabama and Texas for $146 million. Call it $2,561 per acre.

That's the same company, same rough asset class. Roughly four times the price, because a different buyer wanted the dirt for a different reason. It's why LandBridge had its focus on "powered land" since day one. They knew oil and gas is a big deal in the Permian (obviously), but they knew the added value for non-oil operations was even bigger. Think of water, data centers, and related infrastructure. The same goes for Texas Pacific and others.

On a side note, I need to mention that some get carried away rapidly when they sense a dual-use opportunity. I see it a lot in the mining space. Peabody Energy (BTU), which is one of the biggest coal miners on the planet, is working on rare earth discovery in its mines. That is happening. However, we shouldn't let that result in a higher valuation multiple unless there is clear evidence of a strong growth improvement over time. My point is that there's another cash flow opportunity, but it's still mostly a coal miner that should be valued as one.

Scoring And The Grid

Each test is worth 25 points. Total of 100. Symmetrical with TOLL+M in output, except we're dealing with two very different tests.

I want to be explicit about how a LAND score should be used, because on its own it's dangerous.

Run every LAND result through TOLL+M, and you get four outcomes.

• High LAND, high TOLL+M. Asset is mispriced, and the business is durable enough to survive until someone notices. This is the quadrant I actually buy from.

• High LAND, low TOLL+M. The value trap quadrant. The asset really is worth more than the balance sheet says, and management will (likely) destroy it before you get paid. This is where the department store analysis lives.

• Low LAND, high TOLL+M. A perfectly good business that LAND simply cannot see. Nothing wrong here. It just means this particular tool is the wrong tool.

• Low on both. Move on (this one is obvious)

And, obviously, there are good non-TOLL businesses. I cannot mention that often enough. TOLL is a great framework that I use a lot, but there are obviously exceptions.

Here's the scorecard of a few select businesses that I like and/or believe make for great examples:

• Texas Pacific Land

• LandBridge

• Ryonier

• CSX Corp. (CSX)

• Martin Marietta

• EagleRock Land (EROK) - similar to LB. I will very soon give this one more attention.

• Whirlpool - I knew it would score poorly. It's included for educational purposes.

TPL: The Purest Accounting Gap Available

Back to the company from the introduction, because it anchors the entire framework.

Texas Pacific Land owns roughly 882,000 surface acres in West Texas, of which approximately 800,000 came from the 1888 Declaration of Trust at zero basis.

You can see the split cleanly in the filings. The company separately reports "real estate acquired" of 74,493 acres at $143.2 million net book value, which is the land it has gone out and bought with cash in the modern era. Everything else is the original acreage, carried at nothing.

Against that, the market values TPL at roughly $26.5 billion with no debt and a net cash position.

Revenue was $798.2 million in FY25, split between Land and Resource Management, which is oil and gas royalties plus easements plus land sales, and Water Services, which is water sales and produced water royalties. Roughly half the revenue now comes from water, on land where the oil royalty was supposed to be the whole story.

That's a 92, and I struggle to see how anything scores higher on this framework.

The obvious critique is valuation. TPL trades around 30x forward EBITDA and about 47x trailing earnings. LAND identifies the asset. It does not tell you the price is reasonable. Those are separate conversations, and anyone using a framework score as a purchase trigger is misusing it. For the thesis to work (I'm bullish), it needs to execute on the booming water business and its data center plans.

RYN: Where The Gap Goes To Die

Rayonier completed its merger of equals with PotlatchDeltic on January 30, 2026, creating a combined company with roughly 4.2 million acres across the U.S. South and Northwest, plus six sawmills and a plywood mill.

Legacy Rayonier acreage dates largely to the 1990s. That's a real accounting gap.

And then purchase accounting stepped up the acquired PotlatchDeltic acreage to fair value at close.

So roughly half of the combined land base now carries almost no gap at all, while the legacy half still does. Same company, same fence lines, two entirely different accounting stories depending on which side of the merger an acre came from.

This is the half-life problem in its clearest form, and it's the single most useful thing in this article for anyone applying LAND to their own names. M&A closes accounting gaps. Serial acquirers do not accumulate them.

RYN's saving grace is the D test. About 77,000 acres sit under solar option or lease. Another 154,000 acres are under carbon storage lease. Plus rural land sales, plus higher and better use conversions. That is a lot of ways for one acre to get paid, and it's why RYN still scores 76 despite a gutted A column.

Worth noting the operational picture is messy right now. 2Q adjusted EBITDA came in at $123.7 million, and the trailing multiples in most screeners are distorted by merger costs and the New Zealand disposal. Cash available for distribution through the first six months was $177 million versus $47 million a year earlier, which is the cleaner number.

LB And EROK: Two Companies, One Basin, One Very Different Balance Sheet

This is the comparison I want people to take away from this piece.

LandBridge and EagleRock Land are both Permian surface land companies. Both monetize water, surface use, easements, and royalties. Both are structured as Up-C entities with Class A shares and Class B units. Both pitch the same story about data centers and power on Permian dirt.

Read the investor decks side by side, and you would struggle to tell them apart.

Now open the balance sheets.

• LandBridge owns more than 315,000 surface acres in the Delaware Basin and carries "land and land improvements" at $1,032.3 million as of year-end 2025. That land is not depreciated.

• EagleRock controls roughly 286,000 acres across the Delaware and Midland sub-basins. Its June 30, 2026 balance sheet shows land of $225.8 million.

But that's not the interesting part.

EagleRock also carries $674.2 million of intangible assets and $643.3 million of goodwill, against total assets of $1.80 billion. So about 73% of the company's assets are goodwill and intangibles, and land is roughly 13%.

Here's what I think that actually means, and I want to be careful because it would be easy to read this as a knock on the company.

It isn't.

EagleRock's control over that acreage is completely real. A substantial portion of the position is held through New Mexico state and federal leases rather than fee title, and my honest view is that the economics of controlling Permian surface access do not care much about the deed. If an operator needs your water, your caliche, and your road, you are the counterparty either way.

The accountants agree with me. That's exactly why the value shows up as intangibles and goodwill. When EagleRock assembled this business through 2025 and 2026 acquisitions, purchase accounting valued the surface use agreements, water contracts, and leasehold interests at fair value and put them on the balance sheet.

The value is recognized. It is simply not hidden.

Which is the whole point. LAND hunts for value that the balance sheet is understating. EagleRock's balance sheet is not understating anything, because a company built entirely from recent acquisitions has already had every asset marked. EROK scores 19 on locked supply and 4 on the accounting gap, and both of those numbers are correct.

So the framework says LB and EROK are different investments, and it says so for a reason that has nothing to do with which company is better run.

One caution on both names. These are Up-C structures, and the enterprise value calculation is a trap. LandBridge has 27.8 million Class A shares and 47.2 million Class B units. EagleRock has roughly 26.4 million Class A and 105.2 million Class B. If you take the Class A market cap and divide it into consolidated EBITDA, you will produce a forward multiple that is wrong by a factor of two to five. Use all units. On that basis, LB trades around 32x forward EBITDA and EROK around 26x, rather than the low-teens and mid-single-digit figures that some data providers will hand you.

If I had a dollar for every time a screener has quietly mangled an Up-C, I could buy a few acres myself.

The Rest Of The List (And Some Additional Picks), Briefly

CSX scores the maximum 25 on locked supply, and I do not think that's arguable. Roughly 19,000 route miles of corridor assembled across the 19th and 20th centuries, through 26 states, Ontario and Quebec. It cannot be rebuilt. The problem is that the accounting gap, while almost certainly enormous, is essentially unmeasurable, because Item 2 of the 10-K describes track and infrastructure rather than acreage. The framework knows something is there and cannot size it.

MLM has the cleanest supply constraint of anything here after CSX. Roughly 500 quarries, mines and distribution yards across 29 states, Canada and The Bahamas, in a business where new permits take the better part of a decade. Its pending $13.5 billion combination with Lhoist North America has received all regulatory approvals and is expected to close in 3Q26, adding 2 billion tons of limestone reserves. Which means, in framework terms, MLM is about to have a large chunk of its reserve base re-marked at fair value. The gap closes again. MLM scores 8 on dual use because a quarry is a quarry.

Rexford Industrial (REXR) is the honest partial failure, and I say that as a holder. It is NOT included in the table. Southern California infill industrial is about as locked as supply gets, with entitlement scarcity and CEQA review as the binding constraints. But Rexford acquired all of its 232 properties essentially between 2015 and 2026, at cost, with fair value allocation on every deal. Great land, no gap. It scores 56, and I still own it, because TOLL+M is what governs that decision. And I think it's way too cheap.

VICI Properties (VICI) is the case where the framework simply cannot see the asset. VICI's real estate does not appear as land. It sits inside $24.58 billion of net sales-type leases and $19.28 billion of financing receivables, because the accounting converted the property into a receivable. The Las Vegas Strip is about as supply-constrained as ground gets on this planet, and LAND scores it a 54.

Viper Energy (VNOM) is a boundary case worth mentioning because people ask. Viper is minerals, not surface. Subsurface royalty interests under the full cost method, with all recent acquisitions at fair value. Excellent nominal linkage, a 3 on dual use, because you cannot put a solar farm on a mineral interest.

WHR scores 11, and I put it on the list to make a point. Whirlpool is not a land company and has no material land position. So, obviously, it scores low. I know it's silly, but I had to test if my models spot this (of course, they did).

Where LAND Breaks

A few honest limitations apply. Without them, this wouldn't be a good research piece.

• The framework rewards disclosure quality. CSX almost certainly has a bigger accounting gap than its score reflects, purely because railroads do not disclose land the way REITs do. LAND is partly measuring what companies choose to tell you.

• It cannot value anything. A LAND score of 92 says nothing about whether TPL at 30x forward EBITDA is a good purchase. I have watched people conflate "this asset is worth more than book" with "this stock is cheap." Those are unrelated statements.

• And it decays. Every score in this table has a shelf life measured in acquisitions. RYN's A score was materially higher fourteen months ago.

Now, onto my final thoughts.

Takeaway

The biggest takeaway from LAND is that great assets and great businesses are not automatically the same thing. LAND is designed to find something very specific: scarce, productive assets whose economic value may be dramatically understated by the balance sheet. That is why Texas Pacific Land scores so well, while other excellent businesses score much lower.

For me, the real opportunity is where LAND and TOLL+M overlap: a hard-to-replicate asset, hidden accounting value, inflation-linked cash flows, multiple monetization paths, and a business capable of compounding that advantage for decades. But valuation still matters. A 92 LAND score does not make a stock cheap. It simply tells me where I should start digging.

And that, ultimately, is exactly what I built this framework to do.