Harvesting a legacy business is a sale you have not priced

Most North American suppliers of combustion-engine parts face the same portfolio question. They own a business that is profitable today, whose new awards are shrinking and whose products will eventually be needed only as replacement parts. They can invest in it, run it for cash, spin it off or sell it. Most choose to run it for cash, at least for a while, because that needs no announcement and no buyer, and because the cash funds the businesses they would rather be in.

I argued against that choice at Tallisford, and I want to set out why, including the part of my argument that turned out to be wrong.

From 2018 to 2021 I ran our Engine Cooling business: radiators, charge-air coolers, oil coolers and exhaust-gas coolers for light vehicles, with US$1.15 billion of revenue, five plants and about 4,300 people. It earned an EBITDA margin of around 12 per cent and needed little capital. Our own award data also showed, year after year, that it was winning less new work as vehicle makers committed their next platforms to hybrids and battery-electric drivetrains.

Why the run-off case flatters itself

The standard harvest model is simple. Revenue declines along an assumed curve. Capital spending falls to maintenance. Costs fall with volume, and the business throws off cash for ten or fifteen years until it closes. In 2023 our finance team's version showed eight years of cash with a present value of about US$850 million, more than any buyer seemed likely to pay.

I had built the first run-off model for that business myself, in 2020, and by 2023 I knew where it was weakest. Costs do not fall with volume in a plant built for twice the output. Depreciation is fixed, maintenance on ageing brazing furnaces rises, and the most capable engineers and plant managers leave for parts of the industry with a future, often before the volume goes. Customers notice a supplier in managed decline and begin dual-sourcing, which speeds the fall. And the business makes claims on management attention that never appear in the model: the quarterly argument over whether to fund a repair, the labour negotiation at a plant with eight years to run, the customer seeking a price reduction because it knows we are no longer investing.

When we adjusted the harvest case for those effects, using what had actually happened to fixed costs in two plants we had already consolidated, its value fell to between US$640 million and US$720 million. A sale at US$790 million, which is what Corrander Capital Partners eventually paid, beat the realistic harvest and came close to the optimistic one.

Who should own the last decade of a product

The deeper argument was about ownership. A business in long, slow decline needs an owner whose whole strategy is the decline: consolidating capacity across several companies, serving the replacement market carefully, and pricing the last years of a programme with the confidence of a supplier that knows it is one of the few left. A private-equity firm assembling a group of combustion-engine businesses can do that well. A company like Tallisford, whose strategy and capital point towards battery thermal systems and heavy-duty cooling, does it reluctantly and therefore badly.

Timing argues the same way. A buyer pays for the life a business has left, so every year a seller waits removes a year from what it can sell, while the run-off model quietly assumes the seller's own costs will shrink in step. On our adjusted numbers, two more years of harvesting would have cut the price a buyer could pay by more than the cash those two years produced.

I also think a harvest damages the people inside it. When we sold Engine Cooling, its plant managers joined an owner that wanted them to run those plants as well as possible for as long as possible. Under us they would have been running plants we were waiting to close. I felt this sharply, because several of them had worked for me. Some were angry about the sale, and I understood why. I still believe they were better served by an owner that valued the work in front of them.

Keep a share of the outcome you think unlikely

Here is the part that went against me. Our models, and the buyer's price, assumed that combustion-engine light-vehicle production in North America would decline steadily from 2024. Instead, hybrid demand grew faster than we expected, and several vehicle makers extended combustion and hybrid programmes that had been due to end. Hybrids still need engine cooling. The business we sold earned about US$141 million of EBITDA in 2025, against the US$110 million our own plan had expected for that year.

By that measure we sold too early, or too cheaply. I do not think the decision itself was wrong, because a slower decline improves the buyer's run-off without changing the argument about ownership. But I made a structural mistake. I was confident the decline would come quickly, and I let that confidence decide the shape of the deal. We took the whole price in cash at closing. With an earn-out tied to 2025 and 2026 volumes, we would have kept part of the outcome I thought unlikely, and the buyer, who believed in that outcome more than I did, would probably have agreed to it.

That is the lesson I would offer any board weighing a legacy business. Once the harvest case is made honest, the choice between running a business for cash and selling it is rarely as close as the models first suggest. But the timing of the decline is genuinely uncertain, and the seller usually knows less about it than it believes. Structure the sale so that you are paid something if you are wrong.

Three questions are worth asking whenever management proposes to run a business for cash. What has actually happened to fixed costs in plants we have already shrunk? Who would own this business if we did not, and what would they do differently? And if we sell, what part of the price depends on a forecast we do not trust? A run-off plan that cannot answer those questions is a sale postponed, usually to a weaker moment, and usually at a lower price than the one available now.