Investing in collectibles pays less than the indices suggest

Is investing in collectibles a good idea?

Investing in collectibles has made money in some categories and periods, but the best-measured returns are modest and no study below deducts everything a buyer pays. In a 2016 Review of Financial Studies paper, Korteweg, Kräussl and Verwijmeren analysed 32,928 paintings that sold repeatedly between 1960 and 2013. Correcting for selection bias cut average annual index returns from 8.7% to 6.3%, and the authors concluded that a broad portfolio of paintings is not attractive, though targeting specific styles or top-selling artists may add value. The luxury assets hub covers the wider set of passion assets.

Two long-run studies show gains that survived inflation. Dimson and Spaenjers found in 2011 that British stamp catalogue prices from 1900 to 2008 returned 7.0% a year in nominal terms, or 2.9% in real terms, ahead of bonds and behind equities. The authors’ estimates of after-cost returns for individual investors suggest stamps may rival equities on realised performance. Dimson, Rousseau and Spaenjers estimated in 2015 a real return of 4.1% on top-ranked Bordeaux wine over 1900 to 2012, net of storage costs, which exceeds bonds, art and stamps.

At auction, a painting bought at a $100,000 hammer price costs the buyer $128,000 on Sotheby’s published rates, and a seller at the same house keeps $90,000 of a $100,000 hammer price after a 10% commission. Resold at the price it was bought at, the lot returns 70.3% of what the buyer paid. That is Unmatched Capital’s calculation from the two published schedules.

The answer for a given buyer depends on the price tier, the length of the hold, the seller’s cost schedule and whether the item is what its paperwork says it is. Renneboog and Spaenjers concluded in a 2009 Tilburg working paper that buyers of art should expect non-financial benefits rather than high financial returns. This page is general information, not advice, and the net return calculator lets a reader test a specific purchase.

What have collectibles returned in studies and indices?

The best-measured long-run returns on collectibles are between 2.9% and 5.5% a year after inflation, in studies of stamps, art, wine and coins, and the latest index data for 2025 is mixed.

Category	What was measured	Period	Return reported	What the figure leaves out
Paintings (Korteweg, Kräussl and Verwijmeren)	Index from 32,928 paintings that sold repeatedly	1960 to 2013	8.7% a year; 6.3% after selection correction	The abstract reports no cost deduction
Paintings, prints and works on paper (Renneboog and Spaenjers)	Hedonic index from more than 1.1 million auction sales (working paper)	1951 to 2007	4.03% a year, real US dollars	The authors say “substantial transaction costs” reduce the modest returns
British stamps (Dimson and Spaenjers)	Catalogue prices	1900 to 2008	7.0% a year nominal; 2.9% real	Catalogue prices, not prices achieved in sales
Top-ranked Bordeaux wine (Dimson, Rousseau and Spaenjers)	Arithmetic repeat-sales regression	1900 to 2012	4.1% a year, real, net of storage costs	Wines outside the top-ranked estates
US coins (Obaid, Pukthuanthong and Maslar, as summarised by CXO Advisory)	2,063 coins in a trade price guide	1967 to 2015	9.7% a year nominal; 5.5% real; standard deviation 40.8%	Returns are gross; the average bid-ask spread is 8%
Sports and Pokémon cards (Hilbert, Wooster senior thesis)	Portfolios built from PriceCharting.com sales	January 2021 to December 2023	Sports cards minus 0.39% a year; Pokémon cards minus 4.72%	An undergraduate study over three years
Fine wine (Liv-ex Fine Wine 100, via Knight Frank)	Mid-price of 100 wines scoring 95 points or more	2025	Minus 2.5%; almost 25% below the 2022 peak	The bid-offer gap, because the index uses the mid-point
Watches (WatchCharts, via Knight Frank)	Overall Market Index; Rolex and Patek Philippe indices	Year to early 2026, reported April 2026	Overall plus 5.1%; Rolex plus 4.6%; Patek Philippe plus 12.1%	The article does not say whether dealer margins are included
Handbags (Knight Frank)	Hermès Birkin and Kelly bags	2025	Minus 0.2%	The article does not say how it is built
Classic cars (Hagerty UK)	Price Guide values for more than 3000 models	Hagerty’s December 2025 review of the year (index growth measured against December 2024; no period stated for the 80%)	Only the Hot Hatch and RADwood indices grew meaningfully; nearly 80% of values fell or stood still	Values are estimates set by a team from insured values and sales results, not transaction prices

The broadest recent measure is Knight Frank’s Luxury Investment Index, which closed 2025 down 0.4%. The March 2025 Knight Frank Wealth Report said the values of a basket of 10 leading assets fell by an average of 3.3% in 2024. Over ten years the index has risen 38.6%, or 3.32% a year on Unmatched Capital’s arithmetic (1.386 to the power of one tenth); Knight Frank does not say whether costs are included. It covers ten categories, Elite Traveler reported: handbags, jewellery, coins, watches, cars, coloured diamonds, furniture, whisky, wine and art.

What do collectibles indices leave out?

Collectibles indices leave out the costs of trading, the items that never trade, and in several cases the trades themselves, because many are built from price guides, catalogues or bid and offer quotes. The paintings study shows the first problem most clearly. Its authors describe an asymmetric V-shaped relation between the chance that a painting is sold again and its return, and removing the resulting selection bias cut average annual index returns from 8.7% to 6.3%. A repeat-sales index only sees paintings that came back to market.

The coin study drew its prices from a trade price guide, and its summariser notes that the reported returns are “gross, not net”. The stamp study used catalogue prices. Hagerty UK’s Price Guide tracks more than 3000 models of classic and collectable car and updates quarterly, and a team sets each price using insured values, public and dealer sales results and other data. That is a considered estimate of value, which differs from what a particular seller received in a particular sale.

Liv-ex prices each wine in the Fine Wine 100 at its Mid Price, the mid-point between the current highest bid and the lowest offer, and only wines with a critic’s score of 95 points or above and a regular market on its platform qualify. A buyer pays the offer and a seller receives the bid, so the index sits between the two prices a trader sees.

The coin study found an average bid-ask spread of 8%, rising from 6.1% at the start of its 1967 to 2015 sample to 10.9% at the end, and CXO Advisory, summarising the study, cautioned that the market may be thin enough for large trades to move prices. The paintings authors describe their subject as an illiquid asset that trades endogenously, and tie that to the sale-probability pattern above.

Knight Frank’s article names ArtTactic, MyArtBroker, WatchCharts, Liv-ex, LUXUS and the Fancy Color Research Foundation among its sources for individual categories. The page does not say how the sources are weighted in the index or whether buyer’s premiums and other costs are included, so a reader cannot treat the 38.6% as a return a buyer received.

What does it cost to buy, hold and sell a collectible?

At a major auction house, buying a collectible costs a buyer’s premium of 22% to 29% on lots up to $8 million, holding it costs insurance and storage, and selling it costs a seller’s commission. Grossman LLP’s May 2026 review of recent changes lists three tiers at each of three houses, all in US dollars on the hammer price:

Auction house	27% to 29% tier	22% tier	15% tier
Sotheby’s (effective February 2026)	28% up to $2 million	$2 million to $8 million	Above $8 million
Christie’s (effective September 2025)	27% up to $1.5 million	$1.5 million to $8 million	Above $8 million
Phillips (effective date not stated)	29% up to $2 million	$2 million to $8 million	Above $8 million

Sotheby’s own fee page states that its standard seller’s commission is 10% of the hammer price, with a minimum of $500 on any lot with a low estimate of $5,000 or less and a 2% success fee on lots that hammer above the high estimate. The page is undated, and rates for other houses and for private sales differ.

Masterworks, which sells fractional art shares and so has a commercial interest, says fine-art insurance commonly runs about 0.1% to 2% of insured value a year, and can price below 0.5% for a high-value, well-documented collection in a low-risk, well-secured home. That is a vendor’s estimate rather than a measured figure.

The worked example below is illustrative and uses only those published schedules. A buyer purchases a painting at a $100,000 hammer price and pays $128,000 under Sotheby’s 28% tier. The buyer insures it at 1% of the price paid, $1,280 at the end of each year, and sells after ten years at Sotheby’s 10% commission. The calculation ignores storage, conservation, sales tax, artist’s resale royalty, the success fee and tax on any gain, all of which lower the net return further.

Hammer price growth, per year	Hammer price at sale	Seller receives after 10% commission	Net return per year after premium, commission and insurance
0%	$100,000	$90,000	Minus 4.6% (loss of $50,800)
4.0%	$148,024	$133,222	Minus 0.6% (loss of $7,578)
6.3%	$184,218	$165,796	Plus 1.7% (gain of $24,996)
8.7%	$230,301	$207,271	Plus 4.1% (gain of $66,471)

The growth rates of 6.3% and 8.7% are the paintings study’s corrected and uncorrected index returns, applied here to one lot’s hammer price as an illustration; they are not a forecast. The hammer price must rise 42.2% to $142,222 before the seller’s proceeds equal the $128,000 paid, and 56.4% to $156,444 once ten years of insurance are added, which is 4.6% a year. Every figure in the table is Unmatched Capital’s calculation from the assumptions above.

Wine and classic cars carry their own versions of these costs. The wine study’s 4.1% is stated net of storage costs, and Hagerty’s quarterly values describe the price of a car, not what it costs to restore, store and insure one.

How big a risk are fakes and weak provenance?

Fakes and weak provenance are a risk that none of the indices on this page prices, and the best-documented case in this research is wine. Rudy Kurniawan counterfeited high-end wines and sold them at auction and in private sales for about a decade, and in August 2014 a US federal judge sentenced him to 10 years in prison, Wine Spectator reported. When FBI agents searched his home in 2012 they found hundreds of bottles, corks, stamps and 18,000 fake wine labels. The court ordered $28.4 million in restitution to seven victims and forfeiture of $20 million in property. He was the first person tried and convicted in a US federal court for wine counterfeiting.

That case bears on returns because a price index built from sales records a sale of a counterfeit bottle as a sale of the genuine one until someone discovers it. That is Unmatched Capital’s reading of how the indices are built, not a finding of any study cited here.

The research for this page found no measured rate of fakes for any collectibles category, and the shares quoted for sports memorabilia and art forgery on other pages could not be traced to a source that could be read, so this page gives none. Without a rate, a buyer has the item’s paperwork to go on. Before paying, a buyer can ask a seller who has owned the item and how each owner can be documented, who has examined it and whether that person is independent of the seller, and whether the auction house or dealer will refund the price if the attribution fails.

What is the best collectible to invest in?

No collectible comes out as the best in the data, because the studies cover different periods and none deducts the same costs, but the evidence does show what separated items that held value from the rest. Ranking the categories in the table above would compare a 4.1% real return on wine over 1900 to 2012 with a 5.1% rise in a watch index over one year.

The studies point to quality within a category. The paintings authors found that a broad portfolio was unattractive while specific styles or top-selling artists might add value. Renneboog and Spaenjers reported a positive masterpiece effect, meaning that high-quality works performed better. Dimson, Rousseau and Spaenjers found that young wines from strong vintages earn the best returns while still maturing, and that past maturity, wines from famous châteaux deliver growing non-financial benefits to their owners.

Recent car and watch data show wide gaps inside a category. In 2025 Hagerty UK found only two of its indices growing meaningfully, both tracking cars from the 1980s and later, with nearly 80% of values down or static, and wrote that the very best cars still seemed to be in demand at almost every price point. The top end did not escape: Jaguar fell 21.4% over the 12 months to December 2025, largely because of a £3.6 million fall in the mean value of the XKSS. WatchCharts’ indices put Patek Philippe up 12.1% over the year to early 2026 against 4.6% for Rolex, Knight Frank reported.

The cost hurdle falls as the price rises, and this is the publication’s analysis rather than a finding of any study cited here. Sotheby’s premium drops from 28% on lots up to $2 million to 15% above $8 million, as Grossman LLP lists it, so a buyer of an expensive lot needs a smaller price rise to break even than the 42.2% a $100,000 lot needs. The studies’ evidence on high-quality works and top-selling artists concerns price growth measured before costs, so a lower hurdle and better growth can stack for a buyer at the top of the market. The rule that follows is to work out the hurdle at the tier of the lot on offer, because 42.2% belongs only to the 28% tier.

The strongest counterargument is concentration. An expensive item is a single-asset bet, and a buyer who can pay $8 million for one lot carries the risk of that one lot, including a fake or a disputed attribution. The paintings authors themselves say only that targeting top-selling artists “may” add value.

Cost structure differs between categories as well. The coin study found an average bid-ask spread of 8%, while the painting above carried a 28% buyer’s premium and needed a 42.2% rise to recover the premium and the commission. The studies separate items by quality within a category, and Unmatched Capital’s arithmetic adds the cost hurdle, but neither gives a basis for naming one category the best.

How are collectibles taxed?

In the United States, net capital gains from selling collectibles such as coins or art are taxed at a maximum 28% rate, according to the IRS (page reviewed 24 September 2026). The page counts a gain as long-term when the asset was held for more than one year, and says individuals with significant investment income may also owe the net investment income tax. It does not define “collectible”; the IRS’s instructions for Schedule D (2025) say collectibles include works of art, rugs, antiques, metals, gems, stamps, coins, alcoholic beverages and certain other tangible property. The same instructions define a collectibles gain as a long-term gain from selling a collectible, so the 28% maximum applies to items held for more than one year.

In the United Kingdom, a person may have to pay Capital Gains Tax on a possession sold for £6,000 or more, and must work out the gain to find out whether tax is due, according to GOV.UK. The page says a car is not taxed unless used for business and that items with a limited lifespan, such as clocks, are exempt unless used for business. Sets of items such as matching vases are listed among possessions that may be taxable, and a share of a jointly owned item is exempt on the first £6,000. The page has no date.

Tax rates, allowances and the treatment of dealing as a trade change, and this page does not give rates for the United Kingdom. A buyer with a large purchase should take advice from an adviser who knows the item and the jurisdiction.

What this means before you buy a collectible

A buyer who intends to treat a collectible as an investment can check four numbers before paying, and each one is available from the seller or from a published schedule. The first is the all-in price, which on Sotheby’s rates as Grossman LLP reported them is the hammer price plus a premium of 28% on lots up to $2 million (effective February 2026). The second is the exit cost, which at the same house is a 10% seller’s commission, so the lot must rise 42.2% before the seller’s proceeds equal the price paid.

The third is the carrying cost. Masterworks, a vendor that sells art shares, puts insurance at about 0.1% to 2% of value a year, and a broker’s quote for the actual item replaces that range. The fourth is provenance: independent documents that trace the item’s owners and examiners, and a written refund term if its attribution is overturned.

The first three, together with an assumed rate of price growth and a holding period, give a net return and the yearly price rise needed to break even. The net return calculator takes them and returns both. It charges insurance on the item’s value at the start of each year, whereas the worked example above charges it on the price paid, so the two will not match to the decimal. The fourth input decides whether the calculation applies at all.

A buyer who is collecting for enjoyment faces a different calculation. The research cited here measures only money, and the 2009 working paper on art concluded that buyers should expect non-financial benefits. The cost to set against them is the worked example’s $50,800 loss on a lot whose hammer price never moves.