Artnet’s silly indices

By Felix Salmon
May 24, 2012
" data-share-img="" data-share="twitter,facebook,linkedin,reddit,google" data-share-count="true">

A couple of weeks ago, Artnet officially launched Artnet Indices — what it calls “the world’s first comprehensive set of art indices“. According to the press release:

It is now possible to measure price performance and other important market metrics for individual artists and artworks with the same rigorous standards used in financial indices.

Artnet’s Thomas Galbraith is quoted in the release as saying that “the artnet Indices provide quantitative market reports on the performance of artists like Andy Warhol or Damien Hirst, just as you might track a Fortune 500 company”.

I had a long lunch with Galbraith on the day that the indices were launched, and I’ve been going back and forth with him since then, trying to get a feel for how they really work. And as you might imagine, I have quite a few problems with these things.

To put this in perspective, here’s the chart that Artnet loves to send out to reporters, featuring its first index, the C50 index of contemporary art.

artnet C50 versus S&P500.jpg

The message of this chart is very clear. Contemporary art is an asset class, it’s a strongly performing asset class, and if you go back to 1988, it has significantly outperformed the S&P 500. If you start them both at 100 in 1988, for instance, then by 2009 the S&P would only have reached 354, while the C50 would have reached 578 — even after a big plunge from almost 1,000 in 2008.

In fact, however, an investment in the S&P 500 would have done much better than that: it would be 638 in 2009, thanks to the fact that stocks (unlike art) pay dividends. If you chart the C50 against the S&P 500 with dividends reinvested, the outperformance shrinks markedly:


What’s more, this chart takes the C50 at face value, as a vaguely investable index — when it simply isn’t. Here, for instance, are the top 15 artists in the C50 right now: there are lot of names there (Zao Wou-Ki, Zeng Fanzhi, Chu Teh-Chun, Zhang Xiaogang, Wang Yidong) who simply weren’t investable in 1988, and certainly weren’t in the C50 index back then.

I can’t show you a chart of how the 50 artists in the C50 index would have fared if you just bought those 50 artists and held them, because Artnet’s tools won’t let me combine more than 10 artists in one list. But here’s the next best thing: the middle 10 artists from the C50 list in 1988, charted, again, against the S&P 500. These are pretty big-name artists: Alexander Calder, Jim Dine, Helen Frankenthaler, Franz Kline, Robert Motherwell, Louise Nevelson, Kenneth Noland, Theodoros Stamos, Cy Twombly, and Richard Lindner. If contemporary art in general has done well, you’d expect these names to have done well. And, they have! But they haven’t outperformed the S&P 500.


Now the components of the S&P change over time, too — but the changing components have much less effect on the S&P’s performance than they do on the C50′s. And in fact, if you just buy and hold all the components of the S&P 500, you’re likely to outperform the index as a whole. Hot stocks enter indices, and undervalued ones drop out: I don’t have a chart here for the performance of the 500 components of the S&P 500 in 1988, but it would probably do better, not worse, than the index.

Not that that matters: the S&P 500 is investable. You can buy index funds or ETFs which very closely track the performance of the index, with stocks going in and out: they’ll sell the stocks which drop out, and buy the ones which come in. Since September 1989, there have been a total of 587 additions to the S&P 500: that’s about 25 per year, or 5% of the total.

By contrast, since 1988, there have been 111 additions to the C50: that’s about 5 per year, or 10% of the total. Which means that the C50 churns twice as fast as the S&P 500. And in the S&P 500, that churn can be positive: it can happen when when one constituent gets acquired. By contrast, churn in the C50 only occurs when one artist drops out and is replaced by another.

The result is massive survivorship bias. To demonstrate just how massive the bias is, here are the middle 10 artists of the C50 in 1988, charted against the middle 10 artists of the C50 in 2012: Alexander Calder, Damien Hirst, Roy Lichtenstein, Joan Mitchell, Pierre Soulages, Wang Guangyi, Christopher Wool, Rudolf Stingel, Liu Xiaodong, and Liu Wei. You can see that the current members of the index, had you bought them back in 1988, would have performed spectacularly well. The performance of the C50, then, is largely a function of the fact that hot artists keep on getting added — after they’ve become hot. It’s a classic case of investing with hindsight: if you only bought things which performed extremely well, then you would have made lots of money. Well, thanks for that.


The difference here — the 1988 artists end up at 477 in 2012, while the 2012 artists end up at 2,183 — makes a mockery of the idea that contemporary art is some kind of homogenous and investable asset class, or that someone who simply bought contemporary art in 1988 would have seen their assets perform in line with the C50 index.

What’s more, you’re actually seeing treble survivorship bias here. Artnet’s art indices are created by combining its individual-artist indices, and those individual-artist indices have their own survivorship bias built in. That’s because they break down an artist’s work into groups of “Comparable Sets”, and then combine the Comparable Sets in a price-weighted manner to get the artist index. As a result, if Gerhard Richter abstracts, say, suddenly go on a tear, then those abstracts will start making up an ever-greater part of the overall Gerhard Richter index. Both on an artist level and on the index level, whatever does well becomes highly weighted, and things which don’t do well essentially get ignored. (For instance, you can’t even draw up a chart on Artnet of the bottom 10 artists of 1988, because for some of them, Artnet hasn’t even bothered to put together an index yet.)

Finally, it’s no coincidence that Artnet’s first public index is its contemporary art index — the one part of the art world which has been on fire of late. It’s the third level of survivorship bias: if and when Artnet starts publishing its Old Masters index, say, you can be sure the numbers won’t look nearly as impressive.

But even within the contemporary art world, I would be shocked if one collector in a hundred actually saw the kind of returns that Artnet is implying are typical. The thing about the S&P 500 is that it’s meant to be reflective of the market as a whole: while some stocks will do better and other stocks will do worse, broadly speaking stocks perform pretty much in line with the S&P 500. And that’s simply not true of the C50. The overwhelming majority of contemporary art does not perform nearly as well as the C50. Even if you confine yourself to works bought at auction, if you hypothetically bought every work of contemporary art that was sold at auction in 1988, you wouldn’t come close to matching the performance of the C50 since that date.

In other words, stock indices like the S&P 500 are useful precisely because they act as a benchmark: something an investor can reasonably hope to achieve. No sensible contemporary-art collector, by contrast, could ever reasonably hope to see their collection appreciate in value in line with the C50.

The real point here is that contemporary art is always full of here-today-gone-tomorrow art stars, who create art which goes from being white-hot to being pretty much unsellable. In 1988, for instance, the C50 included where-are-they-now names like Theodoro Stamos, Pierre Alechinsky, James Havard, Jean Fautrier, and even Saul Steinberg, the New Yorker illustrator, who appeared just above Robert Rauschenberg on the list. Last year, the most expensive Steinberg sold at auction reached just $28,750.

And that was a much more staid time, when very few really contemporary artists ever appeared at auction. (There was no Basquiat on the list, for instance; no Schnabel, no Fischl.) Today, the list is not only very China-dominated, but also includes names like Rudolf Stingel, Christopher Wool, Mark Tansey, and Glenn Brown — true heirs to the kind of hype that surrounded the likes of Schnabel in the 80s. You can buy their art at auction, if you really want. But you’d have to be insane if you really thought you were making an investment.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

It seems some key elements are missing. I definitely agree with the (obvious) point on survivorship bias, as we all know, it infects all equity indices as well, to different degrees. An important remark is that from what I understand, artnet produces the C50 from auction data, so that it is impossible that an artist not already traded publicly in 1988 could influence the index. This shrinks the survivorship bias relative to the primary market substantially. Second, I think, like for all indices, the survivorship bias is an issue depending on the use of the index. It sure is an issue for economists who want to compare things that people cannot really invest in, maybe less for art dealers and other professionals of the art market who roll their stock of artworks continuously and are permanently exposed to constituents of the C50. Isn’t it natural for dealers to check past market behavior of what they currently own? Finally, thanks to the survivorship bias, the C50 can be considered a best case scenario for contemporary art, and just that information is already very precious for people daring to invest in contemporary art.

Posted by FabianB | Report as abusive

When it comes to art, the best advice has always been buy what you love. If it becomes worthless, you can still love it. If it skyrockets, you might love it more.

But to take something that is so subjective and personal as tastes or preferences and sink into it like a commodity is just asking to be beaten. To be broad, it’s like having a child in hopes he or she will be a huge sports star and bring money back home to papa, and that’s the ONLY reason why you had the kid.

That’s the extreme example of course, but to commoditize a drawing or painting like this is to reduce it to the material sitting on top of the material, encapsulated in another material for presentation, and has no relationship to intent anymore.

Posted by DwDunphy | Report as abusive

“It’s a classic case of investing with hindsight: if you only bought things which performed extremely well, then you would have made lots of money. Well, thanks for that.”

You have index that’s different? That tells the future? Pls advise, woild like to become rich.

“Silly,” indeed.

Posted by Quasimodo3000 | Report as abusive

Felix Solomon has detailed several key flaws with Artnet’s Indices, including the lack of comparables since S&P dividend reinvestment not factored.

Another key issue is that when it comes to buying art, averages are irrelevant. You need to see Art to buy Art, and buying average Art is never a wise idea. The new Artnet Indice appears to be an unnecessary bell & whistle.

Posted by JKLFA | Report as abusive

I gotta say, Felix, I don’t really see what your problem is. I mean, you make several cranky comments about how the Artnet indices are figured, and then what? You got nothing, really. The fact is that lots of people buy art, lots of people think of art as a kind of asset, and lots of art has been steadily increasing in price. Is it so nutty to try to quantify this kind of thing? Not only is it not “silly,” it’s human nature.

Posted by Anonymous | Report as abusive

Over the past twenty years, the group of top performing artists in the Contemporary art Market has changed and any index that aims to accurately track a market must adapt to reflect the shifting composition of that market. This is common practice, and evidenced by the S&P index delisting hundreds of stocks over the same time period covered by our Contemporary Index. Market indicating indices are macro level views, and we urge our customers to consider the more nuanced artist level indices. Indeed, should a collector or investor wish to view only a group of artist that were present at a particular point in time (for example, Felix’s consideration of how the 1988 C50 artists performed), artnet allows for the creation of unique custom indices. artnet’s new product allows users to create reports for a single artist or a group of artists, an extremely useful tool for collectors who want to monitor the performance of their art assets. Ultimately, the reports are very much in line with artnet’s core business philosophy of bringing much needed transparency to the art market. Something we don’t find silly at all.

Posted by aAnalytics | Report as abusive