Should You Split the Company? How to Test a Spin-Off or Divestiture

What a separation actually changes

A spin-off, a demerger or a carve-out is argued on value: the market discounts the group, and the pieces would be worth more on their own. That may be true. But on the day of the split, nothing operational changes. The plants, the customers and the people are the same. What changes is how the market prices them, and what it costs to run two companies instead of one.

That makes a separation unusually testable. The prize is a single number, the re-rating, and the costs can be listed line by line. Most separation papers do the opposite: the prize is argued at length and the costs are a footnote.

The prize no one can check, and the costs everyone can

The sum of the parts is the only number in a separation paper that cannot be audited. It is built from peer multiples chosen by the people making the case, for companies that do not exist yet. Everything else can be checked: the profit of each piece, the group's own multiple, and the costs.

An illustration, from the fictional group used in the Separation Strategy Pack's worked files: the pieces are argued to be worth $1,665 million more apart. The duplicated center capitalizes at $1,097 million against a $480 million one-time bill, so the separation clears its own costs by $88 million. The entire case turns on the difference between 0.90 turns of re-rating and 0.85. That is a case to bring to a board only when someone has said so out loud.

Spin-off, sale, or something smaller

A full spin-off is one of several ways to capture a discount, and the cases show companies choosing between them deliberately.

Separate reporting is the cheapest of these and is often enough to make the discount visible. Ford's segment reporting showed Model e losing $2.1 billion in 2022, $4.7 billion in 2023 and $5.1 billion in 2024 while the gas and commercial businesses earned money. It made the cost of the electric business public without paying for a second company.

What nine separations show

Each case below is drawn from the Stratrix case library, checked against the filings and reporting listed under Sources. No disclosure separates the effect of a re-rating from everything else moving in the same period, so the last column says what each one does not prove.

Testing a separation in a week

What this method cannot tell you

It prices the financial decision, not the operating consequences of separating. Those are real and usually cut the other way, which is why they deserve their own section rather than a line in the model.

It cannot tell you what the pieces will trade at. No one knows that before they trade, which is why the re-rating goes across a range rather than into a single cell. And the cases are illustrations, not base rates: separations get written about when they are large or when they go wrong.

Questions