The carve-out: insuring the past of an entity that has none

Everything above assumed the purchase of a company, that is a legal person that existed before the deal and continues after it, with its history, its accounts and its commitments. A carve-out, where the buyer takes a division detached from a larger group, changes that base assumption and therefore the nature of the product. It is not the sale of a business, it is the assembly of one: assets, contracts, employees, brands and sites are drawn out of a whole and brought together for the occasion. The perimeter does not exist in law, it is defined by the contract, and it will exist only when the contract says so.

The first consequence concerns the accounts, and it disorients buyers and underwriters alike. What does it mean to warrant that the accounts are true when those accounts were manufactured for the deal, out of group bookkeeping, on allocation assumptions chosen by the seller: what share of head office costs, what internal transfer price, what proportion of a team split across three divisions. Those accounts are not false, they are conventional. The insurer therefore warrants a statement bearing on an estimate, which is why many policies exclude or heavily qualify warranties on carve-out financials, tying them to the stated basis of preparation rather than to a notion of truth.

Then come risks that exist only in this configuration and that the buyer discovers after signing. Customer or supplier contracts require the counterparty's consent to transfer, consent that may not come or may come at a price. IT systems, brands and support functions stay with the seller and stop at completion, absent a transitional services agreement. Employees transfer automatically in some legal systems and not in others. And some regulatory permits do not transfer at all: they must be applied for afresh, with a processing period and the possibility of refusal.

The boundary that decides everything must then be drawn, and it fits in one sentence: what is insurable is the past, what is executional is not. A customer's refusal to consent to a transfer is a third party's future decision, not an inaccurate statement about a past fact: it is no breach of warranty and no policy takes it. The same holds for the adequacy of transitional services or the time taken to obtain a permit. What remains insurable is the past of the activity: do the assets exist, is title good, were the contracts as they stood valid, were there undisclosed liabilities inside the defined perimeter.

From this follows a product consequence few buyers articulate, and which should be articulated before placing the policy. In a share sale, most of the risk is historic and therefore insurable. In a carve-out, a considerable share of the risk is executional and forward-looking, therefore outside the scope. A buyer expecting its policy to cover the carve-out going badly has misread the product: it covers the past of an activity that precisely has no past as an entity, which is narrower than the phrase suggests. The practical conclusion is that the uninsurable share must be handled in the price, in an adjustment mechanism or in a seller covenant, and settled before signing rather than hoped for afterward.

One case deserves to be isolated because it looks exactly like a covered loss. It happens that a liability which should have stayed with the seller passes to the buyer, or the reverse, because the perimeter definition allocated it badly. The buyer then discovers a charge it did not expect, exactly as though it were a concealed liability. It is not one: it was not hidden, it was allocated, and badly. That is a drafting error in the sale agreement rather than an inaccurate warranty, and the policy follows the agreement. Notifying it as a breach consumes a deadline and mostly reveals that the buyer has not understood what it signed.

The method follows from all of the above and comes down to three steps. Map, before placing the policy, which risks are historic and which are executional, then handle the second group in the deal terms without hoping for cover. Expect the financial warranties to be qualified, and negotiate the basis of preparation of the accounts rather than their truth, which means nothing here. And read the perimeter definition as the deal's primary risk document, because it, and not the warranties, decides what was bought. In practice it is reviewed by the most junior person on the contract team, while everyone else argues about warranties that will never say anything about what the perimeter left outside.