Before insurance entered these transactions, the buyer's protection against a breach of warranty was cash. A fraction of the price, five to ten percent, stayed in a third party's hands for the survival period of the warranties, and the seller's covenant covered the rest. That arrangement has virtues people forget when replacing it: it is simple, it carries no exclusions, and it pays on a demand mechanism without anyone needing to be persuaded. Its drawback lies elsewhere, and it does not show up in an income statement.
What an escrow really costs is not interest, it is time and immobilization. For a fund at the end of its life, it prevents the final distribution and keeps the vehicle open, with its management costs and reporting obligations, sometimes years after the point at which it should have closed. For a corporate seller, it is cash that does not go into the next acquisition. That cost is one of opportunity and timing, not a line of expenditure, and that is precisely why it is systematically underestimated in negotiation, including by those who bear it.
Insurance acts exactly there, and it should be framed as a balance sheet effect before being framed as a transfer of risk. It substitutes an insurer's covenant for the seller's cash. The escrow shrinks to a nominal amount or disappears, the price is paid in full at completion, the fund distributes and closes. That effect is ordinarily worth far more to the seller than the cover costs, and it explains a fact that puzzles newcomers: interest in this product often comes from the seller, while the policy eventually bought protects the buyer.
The buyer, for its part, makes an exchange that deserves to be looked at squarely. It gives up a certain fund, immediately available, for a conditional claim against an insurer, subject to exclusions, to a retention and to a burden of proof. Its protection becomes better in size, since limits far exceed what any reasonable escrow would have held, and worse in certainty and in speed. Whether that exchange is a good one cannot be settled in the abstract: it depends entirely on the kind of breach this target is liable to produce. A single catastrophic breach is covered only by insurance; small, frequent, easily proved breaches were better served by the escrow.
From this comes the structure well built deals actually use, and it is neither one nor the other. A holdback is retained, often set at the amount of the policy's retention, and the insurance takes everything above it. That combination repairs the one place where insurance is weak, the layer beneath the retention, which is precisely where frequent breaches live. It costs the seller a fraction of what a full escrow would have immobilized, and it delays almost nothing of its distribution.
One consequence the commercial presentation masks must then be named, and it is this module's principal contribution. Where the sale agreement caps the seller's liability at one euro and no holdback is provided, the buyer bears the retention alone. In other words, it has agreed to self-insure the first tranche of every breach, a tranche often counted in hundreds of thousands. That ought to be a quantified and provisioned decision. It is almost always presented otherwise, in the reassuring form of a seller with no remaining liability, which sounds like a problem solved for the buyer when it is a risk that has just been transferred to it.
Good method therefore reverses the usual order of questions. One does not ask which protection is larger, nor how long the seller is willing to wait, but which losses this particular target will plausibly produce. A business with many contracts, year-end rebates and inventory produces multiple modest breaches; a business with a few very large contracts produces the opposite. The retention is then compared against that expected frequency, rather than against a customary percentage. And the escrow discussion is never settled on the seller's distribution timetable alone, because that timetable is the seller's cost and has nothing to do with the buyer's protection.
A fund at the end of its life sells a distributor at an enterprise value of 160 million euros; it wants to distribute and wind up its vehicle. The structure first contemplated provided for a 12 million escrow, that is 7.5% of the price, for twenty-four months. It is replaced by a warranty and indemnity policy with a 16 million limit and a 1.6 million retention, the sale agreement capping the seller's liability at one euro and providing no holdback. Nine months after completion, the buyer establishes four separate breaches: an unprovisioned employment claim for 180,000 euros, an unrecorded year-end rebate liability for 340,000, an operating permit non-compliance for 120,000, and an inventory overstatement for 610,000. The total comes to 1.25 million. What does it recover?
Nothing. Taken individually, each breach falls far below the retention; even aggregated, the 1.25 million stays under the 1.6 million retention. And the seller is capped at one euro. The buyer therefore bears the whole sum, which is exactly the loss profile a 12 million escrow would have covered in full and immediately, with no argument and no burden of proof. One must be precise about the error, because it is not that the buyer bought badly: the protection obtained is not inferior, it is shaped for a different loss. The reasoning at signing was that a 16 million limit is obviously better than a 12 million escrow, which is true for a catastrophic breach and false here. The question never asked was which losses this target would produce: a distributor, with its many contracts, its year-end rebates and its inventory, produces multiple modest breaches, never a single ten million one. The right structure was hybrid, a holdback set at the 1.6 million retention alongside the policy, which would have cost the seller a fraction of the 12 million originally immobilized and would have delayed its distribution hardly at all. That leaves the framing that allowed nobody to see the problem: the deal was presented as a seller with no liability at all, which sounds like a difficulty removed, when what it meant was that the buyer was self-insuring the first 1.6 million of every breach.
- 01An escrow immobilizes the seller's cash: its real cost is time and a deferred distribution, not a line of expenditure.
- 02Insurance substitutes an insurer's covenant for that cash: a balance sheet effect before a transfer of risk, which is why the interest often comes from the seller.
- 03The buyer exchanges a certain, immediate fund for a conditional claim: better in size, worse in certainty and in speed.
- 04A one euro seller cap with no holdback means the buyer self-insures the first tranche of every breach, and it is rarely presented that way.
- 05The right question is not which protection is larger but which losses this target will produce: frequent small breaches never reach the retention.