Artnet’s silly indices
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.
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.