There is a loss that leaves the object perfectly intact, its attribution intact, its market value intact, and takes everything: the one where the holder must give the work back because they never validly acquired it. It is the exact counterpart of the module on de-attribution, with a difference that changes the whole reasoning. De-attribution destroys the value and leaves ownership; a title claim leaves the value and destroys ownership. No damage cover responds in either case, but the reasons differ and so do the recourses.
The risk comes from gaps in the chain of provenance, and the most documented of them is Nazi-era spoliation, framed since the Washington Principles of 1998 and still generating restitution litigation. The Cassirer case, concerning the Pissarro painting 'Rue Saint-Honoré, après-midi, effet de pluie', followed the work through the courts for decades after the events. What matters to an underwriter is not the history but its shape: the claim arrives very late, it bears on a holder who is often in good faith, and it is settled on a right that does not expire the way a contractual warranty does.
The mechanism that makes this an insurance subject rather than a purely legal one is insurable interest. It is a condition of the contract's validity, and it is assessed at inception AND at the loss: without it, you are not insuring, you are betting. Yet if the holder never validly acquired ownership, they may never have had an insurable interest in the property. The difficulty is that restitution does not make the interest disappear as of its date: it finds that the interest never existed. The consequence therefore reaches back over everything the policy did in the meantime, and that is what most files discover too late.
An uncomfortable and perfectly logical consequence follows. An insurer that indemnified partial damage to this holder during the years they believed they owned the work paid an indemnity to someone later established to have had no interest in the property. The true owner, for their part, recovers an object restored with the money of an insurer they never contracted with. Three parties, two of whom never met, and one object repaired at the expense of the wrong person's insurer. This is not a curiosity of the file, it is the normal consequence of retroactive ownership, and subrogation does not solve it since there is no third party responsible for the damage.
The product that addresses this risk exists and is not damage insurance: title insurance guarantees the value of the work against a successful claim. It is bought at acquisition, by those who have a doubt, which describes exactly the population that needs it most and least wishes to admit it. Long marginal, it is gaining ground as due diligence requirements harden, notably among institutions that can no longer accept a loan without a provenance file.
The general-law recourse is the seller's warranty of title, and an asymmetry with the authenticity module is worth noting, because it runs the other way. Authenticity warranties are time-bounded and almost always extinguished by the time a de-attribution arrives. A warranty of title is often longer, sometimes without term, for a simple and strong reason: a seller cannot transfer what they did not own, and that impossibility does not become stale in the way an expert opinion does. The recourse that always fails in one case may therefore succeed in the other, and it is the first thing to check.
That leaves disclosure, the only lever before the loss. A holder who knows of a gap in their provenance and insures without saying so places themselves on the wrong side of contractual good faith, and turns a risk into a defect. The useful discipline is therefore not to claim the chain is complete, which it almost never is, but to name the gaps at inception. A named gap is a risk, which can be priced, excluded or covered by another product. A silenced gap is a flaw in the contract. The provenance file is an insurance document, even though nobody calls it that.
In 1998 a collector buys a painting for 1.1 million euros from an established dealer acting in good faith. The provenance file goes back to 1952 and stops there, which the invoice mentions without further comment. The painting is insured at agreed value. In 2015 water damage occurs: the insurer settles 180,000 euros of restoration and depreciation. In 2024 the heirs of a dispossessed family establish a forced sale in 1938 and obtain restitution. The collector turns to their insurer, which replies that nothing in the policy covers a claim to ownership.
The insurer is right about what is asked of it, and the collector loses 1.1 million that no damage policy ever covered: a title claim is neither physical loss nor physical damage, it strikes the right and not the object. That is the obvious answer, and stopping there misses the file's two real questions. The first concerns the 180,000 euros of 2015. If the collector is deemed never to have validly acquired the painting, they never had an insurable interest in it, and interest is assessed at the date of loss as well: the 2015 indemnity was therefore paid to someone later established to have had no interest in the property. Whether the insurer can recover it depends on the applicable law and on how the interest was described in the contract, but the question arises and nobody saw it coming. It has a symmetrical counterpart: the heirs recover a work restored at the expense of an insurer they never contracted with, and subrogation retrieves nothing since there is no third party responsible for the water damage. The second question is the only recourse that could exist, the 1998 dealer's warranty of title. Unlike a warranty of authenticity, it is not necessarily extinguished, because a seller cannot transfer what they did not own; that is where one should have looked first. That leaves the decisive fact, and it dates from 1998: the provenance stopped at 1952, the invoice said so, and that visible gap was insurable at the time by a product that is not damage insurance.
- 01A title claim leaves value intact and destroys ownership: the exact inverse of de-attribution, and no damage cover responds in either case.
- 02Insurable interest is assessed at inception and at the loss, and restitution does not extinguish it as of its date: it finds it never existed.
- 03An insurer that indemnified damage during that period paid someone with no interest, and the true owner recovers an object restored at a stranger's expense.
- 04A seller's warranty of title often survives where a warranty of authenticity is extinguished: a seller cannot transfer what they did not own.
- 05A provenance gap named at inception is a risk that can be handled; a silenced gap is a flaw in the contract.