Art as an Asset: Title, Provenance, Storage and Lending

When a family buys a block of shares, the legal side of the transaction is almost invisible: there is a custodian, an entry in an account, and certainty about who owns what. A painting works differently: it has no register of title, and there is nothing to point to in a dispute. Ownership is evidenced by a chain of documents — invoices, exhibition catalogues, expert opinions, letters from galleries — and the quality of that chain determines not only the price but the title itself.

Hence the paradox of the art market: market value and strength of title are related but not the same. A work can carry an eight-figure price and at the same time carry the risk that fifteen years from now an heir of a pre-war owner will bring a claim and a court will uphold it. The investment side of the question — liquidity, costs, returns — is covered in the piece on collectibles as an asset class. This article deals with a different layer: title, provenance, lending against a collection, and export restrictions.

That layer is worth understanding before the deal. A mistake in how shares are held is corrected by a transfer between accounts; a mistake in title to a painting is often uncorrectable, because it has to be resolved in the jurisdiction where the object physically sits, under a law the buyer never chose.

Title: nemo dat versus the good faith purchaser

The fork runs between two legal traditions. Common law starts from nemo dat quod non habet — no one gives what he does not have. A buyer of stolen goods does not become owner even if he acted in good faith and paid market price: he acquires possession only, while title stays with the original owner, who recovers the object through an action in conversion. The good faith purchaser in England and the United States is weakly protected: his remedy is a claim against his own seller, not a right to the thing itself.

Civil law protects the flow of commerce and therefore gives a good faith purchaser a chance to become owner once a period has run. In Switzerland the owner of a stolen movable can recover it within five years, while for cultural property the period is extended to thirty years from the loss, or one year from the moment the location and the possessor became known. The Swiss Cultural Property Transfer Act also imposes a duty of care whose scope depends on the market segment: an antiquities dealer investigates provenance from the outset, a buyer of modern art only once suspicion arises.

The practical conclusion: where the deal is done and where the object sits matter no less than the name of the artist — the same purchase in London and in Geneva produces a different level of protection.

The key parameters of the regime, to which everything that follows reduces:

Provenance is not title, and a certificate is not clean title

Four things are routinely merged into one and need to be kept apart, because a dispute usually turns on only one of them. Provenance is the evidenced history of an object — who held it, when, where it was shown. Legal title is the question of who owns it now. Possession is who physically holds it. Authenticity is whether it is what it is said to be. They move independently.

A gap in provenance is not the same as a defect in title. Legitimately owned works routinely have holes — a private collection kept off the record, a dealer who did not name a client, papers lost in a move or a war that was not a confiscation. The gap raises a question; it does not answer it. The converse is the more dangerous trap: an unbroken provenance and a fresh authenticity certificate can still sit on top of void title if there was a single theft anywhere in the chain, because in a nemo dat system title never passed at that link. A certificate speaks to whether the object is genuine and, at most, to the buyer's good faith; it does not speak to whether the seller could convey ownership. A negative result from the Art Loss Register or the INTERPOL database is a due-diligence step, not a warranty of clean title.

Which law decides, the good-faith divide, and the clock

Because a painting has no register, the first question in any title dispute is not "who is right" but "whose law applies". The settled conflict rule for movables is lex situs at the time of each transfer: whether title passed is judged by the law of the country where the object physically sat when the transaction happened. In Winkworth v Christie, Manson & Woods [1980] works stolen in England were sold to a good-faith buyer in Italy and sent back to London for sale; the English court applied Italian law to the Italian sale, the buyer had acquired title under it, and the original English owner lost. The situs of a movable therefore decides the outcome as much as the facts do — and it can be changed deliberately by moving the object.

On top of that rule sit two divergent traditions. Common law starts from nemo dat and protects the original owner; civil law protects commerce and lets a good-faith buyer become owner once a period runs. France is the sharpest illustration: article 2276 of the Code civil states that en fait de meubles, la possession vaut titre — for movables, possession stands as title — so a good-faith possessor is presumptively owner. The dispossessed owner has three years from the loss or theft to reclaim the object (art. 2276 al. 2), and if the current holder bought it at a fair, a market, a public auction or from a merchant selling like goods, the owner can recover it only by reimbursing the price paid (art. 2277). England and New York give the true owner no such three-year cut-off, but they measure time very differently, which is the second half of the question.

The New York rule repays attention because it is counter-intuitive. In Guggenheim v Lubell the Court of Appeals refused a "discovery rule" that would have started the clock when the owner could have found the work; instead the three-year period runs only once the owner demands the object and the holder refuses. An owner who sat on his rights is not time-barred for that reason alone — his delay is weighed under the equitable defence of laches, a separate and fact-specific inquiry. The lesson across the table is that the same stolen work can be irretrievable in Geneva, recoverable on reimbursement in Paris, and litigable for decades in New York.

Two courts, one painting, and the limits of a holding. Cassirer v Thyssen-Bornemisza Collection Foundation is the clearest modern illustration of how little a restitution ruling decides. A Pissarro sold under duress in 1939 was bought by a Spanish state foundation in 1993 and sued for in California in 2005.

The Supreme Court (No. 20-1566, 21 April 2022, Kagan J., unanimous) answered one question and no more: in a suit against a foreign state under the Foreign Sovereign Immunities Act raising non-federal claims, the court applies the same choice-of-law rule it would apply in a similar suit against a private party — the forum State's rule, not a federal common-law rule. It expressly declined to decide whether Californian or Spanish law governed, whether a good-faith purchaser of stolen art may keep it, or who owned the painting.

On remand the Ninth Circuit (No. 19-55616, 9 January 2024) ran California's governmental-interest analysis, held that Spanish law governed, and held that the foundation had acquired title by acquisitive prescription under article 1955 of the Código Civil — three years of uninterrupted good-faith possession — because it had been found not to know the work was stolen when it bought in 1993. Had it been an encubridor, the six-year period of article 1956 would have applied and no title would have run before the 2005 suit.

Read that pair for what it decides and not a step further. The Supreme Court settled a method for choosing law; the Ninth Circuit settled a title question on the facts of one purchase, in one year, by one buyer, under one country's prescription rule. Neither holds that a museum in possession wins, that Spanish law reaches other collections, or that a 1930s loss is spent elsewhere: move the buyer's knowledge, the year of acquisition or the situs, and the same reasoning delivers the opposite answer. Which forum hears the claim in the first place is a question in its own right, mapped in the piece on where a cross-border claim is brought and enforced.

Title, possession and security: a matrix

The axes above — who owns, who may hold, who has enforceable security — do not line up. The matrix sets out the common positions and what each one actually turns on; the row that decides a dispute is usually not the one the parties argue about.

The through-line is that title, possession and security answer different questions and are settled by different rules — perfection ranks creditors, nemo dat protects owners, and situs law decides which of them the object's history has produced.

Three conventions and what they actually deliver

The international layer consists of three instruments with different functions. The 1954 Hague Convention and its First Protocol address armed conflict: the Protocol requires the occupying party to prevent export of cultural property from occupied territory, to return it at the close of hostilities, and prohibits retaining it as war reparations; a good faith holder is entitled to an indemnity.

The 1970 UNESCO Convention is a state-to-state instrument: it obliges parties to prevent the import of objects stolen from museums and monuments of another party and listed in an inventory, and to return them on request with compensation to the good faith possessor. It is not retroactive, it operates only from the point at which both states have implemented it, and it gives a private owner no direct rights.

The 1995 UNIDROIT Convention is the only one of the three that operates at the level of private law. Article 3 provides that the possessor of a stolen cultural object shall return it. The period is three years from the moment the claimant knew the location of the object and the identity of the possessor, and in any event no more than fifty years from the theft; for objects from public collections and monuments the fifty-year cap does not apply, and states may declare a seventy-five-year period. Article 4 gives a good faith possessor the right to "fair and reasonable compensation", but only if he exercised due diligence — and the criteria are spelled out: the character of the parties, the price paid, and whether he consulted any reasonably accessible register of stolen cultural objects. That provision makes a database search a legally significant act.

The restitution layer: why 1933–1945 is checked separately

Provenance covering the Nazi confiscations is treated as a separate exercise. The 1998 Washington Principles, adopted at a conference attended by 44 states, set a framework for resolving claims in good faith — but they are not legally binding. National commissions exist in only five countries: Austria, France, Germany, the Netherlands and the United Kingdom, and their recommendations are generally not binding either.

In the United States the logic is harder. The HEAR Act of 2016 (Public Law 114-308) introduced a single federal limitation period — six years from the claimant's actual discovery of the location of the work and of his possessory interest in it — for losses in the 1933–1945 period. The Act was originally due to expire on 1 January 2027. The revised version, signed on 13 April 2026 (Public Law 119-82, from bill S. 1884), removed the sunset and expressly precluded time-based defences — laches, adverse possession and acquisitive prescription — as well as the act of state doctrine, forum non conveniens and international comity: cases are to be decided on the merits.

For an owner, 1933–1945 provenance is a concrete legal risk which, in the United States, no longer expires with time.

Authenticity: who is left to attribute

The second risk after title is forgery, and the infrastructure for confirming authenticity has weakened markedly over fifteen years. Artist foundations have closed their authentication boards en masse: the Andy Warhol Foundation after protracted litigation that cost it more than $7 million in legal fees; the Keith Haring Foundation in 2012; the Roy Lichtenstein Foundation and the Noguchi Museum in the same line. The reason is uniform: a negative opinion destroys value, the owner sues, and the defence costs more than the board is worth; works without the blessing of a now-defunct institution are hard to sell.

What is left: the catalogue raisonné, independent experts, technical analysis and contractual warranties. The auction house authenticity guarantee is the most practical instrument, but its boundaries are narrow. The standard term is five years from the date of the auction (for gemstones, wine and books, a matter of days). The guarantee is non-transferable and runs only for the benefit of the original buyer; the sole remedy is rescission of the sale and refund of the purchase price. The exclusions are material: the guarantee does not bite where the catalogue description reflected the generally accepted scholarly view at the time of sale, or where the forgery can be established only by methods that were then unavailable, impractical or damaging to the work. A shift in scholarly consensus is not the auction house's risk.

Four disputes and how they resolve

The rules above only become concrete in conflict. Each of the four common disputes starts with the same governing-law question and is won or lost on a specific piece of evidence, not on who behaved better.

Art lending: why the painting is pledged in New York

Lending against art comes in two shapes: a pledge with delivery of the work to the lender or to agreed storage — safe for the bank but inconvenient for the owner, since the collection leaves the walls — and security without dispossession, where the work stays with the client. The choice between borrowing and selling is a calculation in its own right, weighing interest on one side against capital gains tax and the auction commission on the other: it is worked through in the piece on buy-borrow-die on illiquid assets.

The second structure is not available everywhere, and the reason is purely technical. In the United States Article 9 UCC allows a security interest to be perfected by public filing — a UCC-1 — which puts third parties on notice of the encumbrance and gives the lender priority without physical control of the object. The civil law tradition historically required dispossession for a pledge of movables; hence US leadership in this market.

Some jurisdictions have solved the problem: France introduced a registered non-possessory pledge in 2006, and Belgium by its 2013 Act, which only took effect together with the national pledge register on 1 January 2018; Spain allows both models; England, Switzerland, Italy and Germany remain within possessory logic.

The scale and terms of the market, on Deloitte Private and ArtTactic estimates:

The gap between the last two rows is what separates the bank and non-bank halves of this market.

The lender's risks are the buyer's three risks in money terms: valuation is subjective and lags the market, liquidity is low, and a defect in title wipes out the collateral entirely. Credit due diligence is therefore no shallower than pre-sale diligence.

Export, storage and AML

A purchased object cannot always leave: national export regimes for cultural property operate independently of ownership.

Four regimes show what triggers control and how the state keeps a work at home.

The UK deferral is recommended by the Reviewing Committee on the Export of Works of Art, and only a UK buyer paying the fair price can lift it.

Freeports defer duties and VAT, but regulatory pressure on them is rising; the mechanics are set out in the piece on vaults and free ports. At the same time the AML perimeter is tightening: art dealers, intermediaries, auction houses and storage operators are obliged entities for transactions from €10,000, single or linked; in the UK art market participants register with HMRC. EU Regulation AMLR 2024/1624 applies from 10 July 2027 and introduces a ban on cash payments above €10,000, with counterparty identification in the €3,000–10,000 range.

The other side of the border: import control and criminal exposure

Export control asks whether a work may leave. Import control asks whether it may arrive, and it is the harder half, because the destination state applies its own test to an object whose entire history happened somewhere else. A clean invoice and a good title under the situs law do not get a work across a border that asks a different question.

The European Union built that question into Regulation (EU) 2019/880 on the introduction and the import of cultural goods. Article 3(1) prohibits the introduction of Part A cultural goods removed from the territory of the country where they were created or discovered in breach of that country's laws — a prohibition that borrows the source state's law wholesale — and it has applied since 28 December 2020. Above it sit two documentary regimes that became applicable with the centralised electronic system on 28 June 2025 (art. 16(2)): an import licence under Article 4 for the Annex Part B categories, meaning archaeological objects and elements dismembered from monuments, where the object is more than 250 years old, whatever its value; and an importer statement under Article 5 for the Annex Part C categories, where the object is more than 200 years old and worth €18,000 or more per item.

The United States runs two independent layers. The Convention on Cultural Property Implementation Act restricts designated archaeological and ethnological material either by bilateral agreement with a source state — five years at a time, renewable for further periods of not more than five (19 U.S.C. §2602) — or by emergency action, capped at five years and extendable by three (§2603).

An importer of designated material must produce an export certificate from the source state or satisfactory evidence under §2606; without it the material is held in customs custody and, after ninety days, exposed to seizure and forfeiture under §2609. Section 2607 bans outright the import of property documented as belonging to the inventory of a museum or a religious or secular public monument of a State Party and stolen after the Convention entered into force for it.

The second layer is criminal: the National Stolen Property Act, 18 U.S.C. §§2314–2315, reaches receiving, possessing, storing, selling or disposing of stolen goods worth $5,000 or more that have crossed a State or United States boundary, with up to ten years' imprisonment — and the threshold drops to $500 where the goods are pledged as security for a loan, which puts an art-secured lender inside the statute at a tenth of the ordinary figure.

A patrimony law can turn an export into a theft — but only a real one. In United States v Schultz (2d Cir., 333 F.3d 393, 2003) a New York dealer was convicted of conspiring to receive Egyptian antiquities. The court held that Egypt's Law 117 of 1983, which vests ownership of antiquities found in Egypt after 1983 in the Egyptian state, is a genuine national ownership law, so objects taken in breach of it are "stolen" for the purposes of the NSPA.

The limit sits in the same reasoning: following United States v McClain, the court separated a statute that actually asserts ownership — Egypt seized what was found, policed internal dealing as well as export, and enforced both — from one that merely restricts export, which does not make the object stolen. The case does not stand for the proposition that breaching any foreign export rule is a crime in the United States; it stands for the proposition that the foreign statute has to be read to see whether it took ownership or only forbade a departure. The court also declined to hold that the CPIA is the exclusive route for imported antiquities, so the two layers run in parallel rather than in sequence.

The United Kingdom uses a narrower criminal instrument. Under the Dealing in Cultural Objects (Offences) Act 2003, dishonestly dealing in a tainted cultural object knowing or believing it to be tainted carries up to seven years on indictment, or six months summarily (s. 1). Section 2 draws the taint tightly: the object must have been removed from a building, structure or monument of historical, architectural or archaeological interest, or excavated, that removal or excavation must have been an offence — under any country's law, wherever it happened — and it must have occurred after the Act commenced. Material looted before that date falls outside the Act, whatever else it may offend.

Read the table as a sequence rather than a menu: a work can clear the EU prohibition, still need an Article 4 licence, and still be stolen property in an American court, because each row asks its own question and none of them answers another. Whether the counterparty, the source country or the object is separately caught by a restrictive-measures regime is a fourth check again, and the route map for it is the piece on sanctions and how they reach an asset.

Sending a work abroad on loan: immunity from seizure

The other kind of lending is the museum loan, and it carries a risk no loan agreement can solve: while the work sits in a foreign country, anyone with a claim to it can try to attach it there. Two states answer with a statutory shield, and they cut its edges very differently.

In the United States, 22 U.S.C. §2459 grants immunity from judicial seizure to an object of cultural significance imported for temporary exhibition — but only where the President or his designee has determined that the object is culturally significant and that its display is in the national interest, and a notice to that effect has been published in the Federal Register before the object is imported (s. (a)). Once that is done, no court of the United States, of a State, of the District of Columbia or of a territory may issue or enforce judicial process against the institution or carrier holding it, and the United States attorney is entitled as of right to intervene (s. (b)).

Immunity from seizure, however, is not immunity from suit, and the second half was addressed only in 2016, when 28 U.S.C. §1605(h) was added: activity in the United States associated with an exhibition covered by a §2459 determination is not commercial activity for the purposes of the Foreign Sovereign Immunities Act.

The carve-out is the substance of the provision. Under §1605(h)(2)(A) that protection falls away where the claim alleges that the work was taken in connection with the acts of a covered government during the covered period — 30 January 1933 to 8 May 1945 — and §1605(h)(2)(B) extends the exception to systematic confiscatory takings of works from members of a targeted group after 1900. The United States therefore grants a broad shield and cuts a deliberate hole in it precisely where the restitution claims are.

The United Kingdom buys comparable protection at a different price. Under Part 6 of the Tribunals, Courts and Enforcement Act 2007 a protected object may not be seized or forfeited (s. 135) if, on entering the country, it is usually kept outside the United Kingdom, is not owned by a person resident in the United Kingdom, its import breaches no prohibition, and it is brought for public display in a temporary exhibition at a museum or gallery that meets the publication requirements (s. 134). The protection runs for not more than 12 months from entry, extendable by the appropriate authority for further periods of up to three months each (s. 134(4A)), and the museum must be approved by that authority, which examines its procedures for establishing the provenance and ownership of objects (s. 136). There is no Nazi-era carve-out; instead the United Kingdom conditions immunity on the borrowing institution having done — and published — the provenance work in advance.

Two shields, two prices. The American model protects widely and then reopens the door for 1933–1945 claims by statute; the British model protects narrowly in time and makes the institution's own provenance diligence the entry ticket. A lender choosing between them is really choosing which risk to carry: an American exhibition exposes a work with a wartime gap to a claim decided on the merits, while a British one exposes any work to a claim the moment the twelve months run out with the object still in the country. Where a collection is held through a structure rather than by an individual, who signs the loan, who gives the indemnity and who would be sued are governance questions that sit with the family office.

Passing it to the next generation

A collection survives succession badly without preparation: it is indivisible, illiquid, and demands a valuation that heirs obtain at the worst possible moment. The general framework is in the pieces on succession planning and the inheritance tax map. What is specific to art is that the tax can sometimes be paid with the works themselves: the UK acceptance in lieu scheme settles inheritance tax by transferring pre-eminent objects into public ownership, and the French dation en paiement, under the 1968 statute, covers droits de succession, gift tax and IFI.

The rest is hygiene: all-risks insurance with a clear grasp of the exclusions (wear, inherent vice, confiscation, war, damage in transit), item-by-item valuation, and a deliberate choice of storage location, because the situs of the object determines both applicable law and tax.

What pre-sale diligence covers

Diligence before a purchase runs across every layer described above. Provenance is traced to full depth, and the 1933–1945 period is checked as a separate exercise. The Art Loss Register and the Interpol database are searched, with the caveat that a certificate evidences good faith and does not guarantee clean title.

The law of the country where the object sits is established next: whether it follows nemo dat or protects the good faith purchaser, and which limitation periods apply. Export status is settled at the same stage — whether the work can leave at all, and on what terms. Where the work will cross a border, the destination state's import rules, and for an outbound museum loan the seizure-immunity route, are settled before the object moves rather than after.

The contractual layer closes the exercise: the warranties, with their term, who holds them and what is excluded; and an independent valuation and insurance cover in place before delivery.