A professional in this field spends part of their time saying no, and the quality of their work is measured as much by how they refuse as by how they place. But one must refuse for the right reason, because there are two species of refusal with different consequences. A commercial refusal means that this market, on this day, has no appetite: it reverses by changing insurer, by waiting for a renewal, by supplying information that was missing. A structural refusal means that the object presented is not insurable in principle, and no amount of insistence will make it so. Confusing the two wastes time in one direction and credibility in the other.
The first structural boundary is fortuity, and it is the easiest to state and the most often crossed. A contract of insurance bears on an event whose occurrence or amount is uncertain at the moment of commitment. What is certain is not a risk, it is a debt, and transferring a debt against a premium amounts to asking the insurer to pay it in the debtor's place at a discount. No drafting works around that, because certainty is not a feature of the contract but of the world. The practical difficulty is never understanding the rule, it is recognising that one is on the wrong side of it: between a contestable tax position and an assessment already notified with its amount settled, there lie only a few months and one letter, and the second is no longer insurable.
The second boundary concerns the behaviour of the buyer, and it explains the greatest number of apparently arbitrary exclusions. Cover that protects someone against the consequences of their own decision changes the decision, and if it changes it enough, the loss stops being an accident and becomes a choice. That is why the insured's own fraud is irremediably outside cover, including in the most buyer friendly wordings, whereas the seller's fraud is the textbook case the product exists to handle. The line does not run between the honest and the dishonest, it runs between the one who suffers and the one who could bring about.
The third boundary is one of public policy, and it calls for explanation rather than assertion. A financial penalty imposed by an authority is meant to deter a determined person; if that person can shift the burden onto an insurer, the penalty loses its effect and the authority has worked for nothing. National laws draw consequences from this that do not exactly coincide, some prohibiting only criminal fines, others refusing administrative fines as well, a few leaving an uncertain zone. What must be retained is not a list but a test: what punishes is not insurable, what compensates is. One file may carry both, and the first task is to separate them before discussing price.
Alongside those three boundaries a fourth limit exists and it is honest not to file it with them. Some exposures are insurable in principle and refused for want of any way to bound them: a loss whose tail runs over thirty years, a risk for which no data supports a distribution, an exposure whose correlation with the rest of the portfolio is unknown. That refusal is not one of nature, it is one of measurement, and it moves: subjects refused twenty years ago are placed today because they can now be quantified. A professional presenting an unusual exposure must therefore know which of the two registers they are in, because one is worked on with data and the other is not worked on at all.
From that distinction follows the way a refusal is delivered, and it matters as much as the refusal itself. Telling a client that a request is structurally uninsurable, and giving the reason, does them a lasting service: they stop searching, they redirect their money towards what actually addresses their problem, a holdback, a specific indemnity, a price reduction, walking away. Merely saying the market has no appetite, when the reason is structural, sends them to three more intermediaries and costs six weeks. The symmetrical mistake exists and costs as much: presenting as impossible what was only a badly presented submission.
One frequent confusion must finally be named, between what is uninsurable and what is merely expensive. A bounded, measurable and lawful exposure almost always finds a carrier; the question then becomes whether the cost of transfer is less than the value of the certainty acquired, and that question does not belong to the insurer, it belongs to the buyer. An adviser who answers uninsurable where the honest answer would be too expensive for what it is worth closes a discussion that ought to have happened. The two answers look alike in their immediate effect and do not have the same standing at all.
This module closes the certification and stands as its summary, because every exclusion met until now attaches to one of the boundaries described. The known falls out because it has no fortuity left. The insured's fraud falls out because it is brought about. Fines fall out because they punish. Historical contamination falls out because it is known and unbounded. These are not eight rules to memorise but four reasons to understand, and a professional holding them can rule on a request they have never seen before, which no list of exclusions allows.
A broker receives from a buyer, three weeks before signing a 210 million euro acquisition, a request covering three exposures it wishes to transfer together. One, a proposed tax assessment notified to the target in March 2026 by the tax authority, for 4.1 million euros in principal, contested but with the amount settled and the authority's position in writing. Two, a 2.8 million euro fine imposed in 2025 by a competition authority on the target for price fixing, with an appeal pending. Three, a commercial dispute brought by a former distributor claiming 15 million euros, whose outcome turns on the construction of a termination clause, and on which two counsel opinions give a probability of adverse judgment of between 20 and 35 per cent, for a likely amount of 4 to 7 million. The buyer asks for one answer covering all three.
The three requests call for three different answers and they must be delivered separately, failing which a blanket refusal will be read as a matter of market appetite. The third is insurable and it is where the work should go: the outcome is uncertain, the amount is distributed, the subject matter is lawful, and two conflicting opinions are precisely the input an underwriter wants, since they allow a distribution to be built rather than a conviction asserted. The first is no longer insurable, and its closeness to what the market handles routinely is the trap in this file: an uncertain tax position transfers, a notified assessment with its amount settled does not, because what remains bears not on the existence of the risk but on the outcome of an appeal against an already quantified debt. It must be put that way and not as an absence of market appetite, or the buyer will shop elsewhere through the three weeks it has left. The second is refused for a reason of another order, and it is the refusal to defend most carefully because it looks formalistic: a competition fine is a penalty, it aims to deter the target itself, and an insurer discharging it would undo the effect the authority intended. The pending appeal changes nothing, it bears on the existence of the penalty and not on its nature. What remains to be done for the two refused requests belongs to the sale agreement and not to insurance: they are handled by a price reduction, an earmarked holdback or a specific seller indemnity, and the useful advice is to say so now rather than in ten days, because those three instruments are negotiated before signing and not after.
- 01A commercial refusal reverses by changing market; a structural refusal does not, and confusing them wastes time in one direction and credibility in the other.
- 02What is certain is not a risk but a debt: between a contestable position and a notified assessment with its amount settled lie a few months and one letter.
- 03The line of inducement runs not between the honest and the dishonest, but between the one who suffers the loss and the one who could bring it about.
- 04The public policy test fits in one sentence: what punishes is not insurable, what compensates is, and one file often carries both.
- 05A fourth limit exists and is not of the same nature: an unbounded exposure is refused for want of measurement, and that refusal moves with the data available.