A failed savings and loan bought 4,800 acres for $19.5M — then spent 20 years on the paperwork

That is a twenty-three-fold difference in price per acre, and it is the cleanest illustration in American real estate of what an entitlement is worth. But it is not the same land sold twice — the 2015 parcel is a different, adjacent piece that arrived already permitted for about 2,640 single-family lots.⁶ What the pair actually shows is what raw and entitled ground in one ranch traded for, seventeen years apart, which is a better piece of evidence than a resale would be.

And the buyer was not a vulture fund. It was a dead bank.

The buyer was a dead bank with a tax asset

On 6 July 1992 regulators seized HomeFed Bank, then the eighth-largest thrift in the United States, with $13.5 billion of assets and $1.05 billion of losses over the preceding two years.² Its chairman, Kim Fletcher: “I’m very sorry for what has happened. This is certainly not the way we planned it.”²

The bank was gone. The holding company was not. HomeFed Corporation filed for bankruptcy the same year and emerged in 1995 under a plan principally funded by Leucadia, which took an initial 41.2 per cent and later went to 89.6.¹ What that shell carried was not property. It was $266,245,000 of federal net operating losses — and in 1997 the IRS granted a private letter ruling permitting $219,324,000 of the failed bank’s own losses to be reattributed to the parent.¹

Read the 1998 purchase with that in hand and the structure explains itself. A company holding a quarter of a billion dollars of expiring tax losses needs future taxable income to put against them, and it needs that income to arrive before the losses expire — which, per the same filing, they began doing in 1999 and finished doing in 2018. Land that takes two decades to entitle, and then produces large gains, is close to a purpose-built answer.

No source found anywhere describes this buyer as a vulture fund. That framing is a fair description of the behaviour and a poor description of the entity, and the real one is more interesting: the vehicle was the corpse of a savings and loan, and the fuel was its losses.

What four thousand dollars an acre buys

Nothing about the ground changed. Everything about the paperwork did.

Chaparral in the Otay Mountain Wilderness, near but not inside Otay Ranch. This is the kind of terrain the 1998 purchase covered.Photo: BLM California, public domain, via Wikimedia Commons

The permit is the product

A useful thing to understand about this business is that the seller never builds anything. HomeFed’s output is a legal condition, not a structure. The land arrives as scrub with a general plan attached — Chula Vista and the county had approved one for the whole 22,900-acre ranch in 1993, five years before HomeFed bought in⁶ — and leaves as a recorded map with a specified number of permitted dwellings on it, graded, with sewers and roads.

Two things paid for that work, and neither was HomeFed’s equity. The infrastructure was funded with $125,000,000 raised from 250 foreign investors at $500,000 each through the EB-5 visa programme.⁸ EB-5 money is patient and cheap because the return the investor is buying is a green card rather than a yield, which makes it close to ideal for roads and sewers — assets that produce no income and that no ordinary lender wants to fund. HomeFed has kept going back: its own later statements describe $171 million raised for the next village and approval to seek up to $249.6 million for the one after.

Then the houses. In January 2017 three builders — Brookfield, Lennar and Shea — put $50,000,000 of cash into the Village of Escaya joint venture, against HomeFed’s contribution of land valued at $20,000,000 and $13,200,000 of completed infrastructure.⁸ The builders take the construction risk. HomeFed takes reimbursement of its infrastructure costs up to a $78,600,000 cap, plus what its filings call contract service revenue — a percentage of the retail sale price of every home sold.⁸

That is worth stating precisely, because the reel calls it a share of the profit. It is not. It is a share of the price, which is a materially better deal: it pays whether or not the builder makes money.

What the record shows, and what it stops short of

The record has a stain

HomeFed found contamination on about thirty acres of its own undeveloped land, remediated it by February 2013, sued to recover the cost — and lost, in February 2015. The defendants then sought roughly $13,500,000 in legal fees against it, which the court denied that August.⁶ Separately, while building Escaya’s access road, the company found groundwater carrying petroleum byproducts and soil vapour carrying methane and volatile organic compounds in a section where homes were to go.⁸

The sharper record is the exit. When Jefferies — which already held about seventy per cent — bought out the rest in 2019, minority shareholders approved it overwhelmingly. Two sued anyway, and in July 2020 the Delaware Court of Chancery refused to dismiss the case, finding that Jefferies “did not commit itself to the dual protections” that make a controller buyout presumptively fair “before engaging in substantive economic discussions concerning the transaction that anchored later negotiations and undermined the ability of the special committee to bargain effectively”.¹² The largest minority holder had told the committee’s own banker it thought the price “inadequate” while judging it better than nothing.¹² The case settled for $15 million rather than going to judgment.

And the wider ranch has been fought over hard — the Center for Biological Diversity, the Sierra Club and the California Attorney General among those litigating over wildfire risk and habitat. Those suits were aimed at other landowners’ outlying villages, not HomeFed’s. That is not luck. HomeFed’s parcels sat in the already-planned core of a ranch whose general development plan had been approved in 1993; the fights were over the wildland edges other developers were still trying to entitle.⁶ Buying inside the line somebody else had already drawn is part of the strategy, not an accident of it.

What transfers

The mechanism is simple to state and nearly impossible to copy. Buy ground that cannot be built on, at a price that reflects that. Spend two decades converting a legal condition rather than a physical one. Sell the finished condition to somebody whose business is building, and keep a percentage of what they sell.

What makes it hard is not the idea. It is the balance sheet required to hold unproductive land for twenty years through two recessions while paying property tax on it, and the political tolerance to spend that long in hearings. HomeFed could do it because it was not really a property company at the start — it was a tax-loss carryforward with a corporate wrapper and a patient controlling shareholder, which is a very unusual thing to be, and the reason the strategy was available to it and not to a normal developer.

Two smaller lessons travel further. Fund the part that earns nothing with the cheapest money you can find, and be honest about why it is cheap — EB-5 capital is patient because the investor is buying immigration, not yield, which is exactly why it suits sewers. And when you take a participation, take it on the sale price rather than the profit: the builder’s margin is the builder’s problem, and a percentage of revenue pays in a year when a percentage of profit does not.

Then the caution. Twenty-one years after this began, the company stopped filing, and with it went any public answer to the only question that matters: what the work cost and what the participation eventually paid. A strategy this long-dated is judged on a number that, in this case, nobody outside Jefferies has ever seen.