Felix Salmon links to an important new study that questions the conventional wisdom about art market returns over the long haul. The problem seems to lie with the survivorship bias of the data set which prompts Salmon to take the issue a fair bit further:

I think this works more generally: if you could somehow include in the survey all the art that anybody ever buys, the aggregate returns would certainly be negative. Most art is bought in the primary market and is never sold, mainly because it never can be sold. To all intents and purposes, it has zero monetary value the minute that it leaves the gallery or artist’s studio.

The greatest investors in art never bought art as an investment. If you take the set of all art collectors, statistically speaking some small proportion of them will see their collections grow substantially in value. Those collectors are deemed in retrospect to have had a “great eye”. But people who buy art as an investment are almost certainly going to be disappointed — and even more disappointed if they get someone else to buy art for them.

Returns on Art (Felix Salmon/Reuters)

Buying Beauty: On Prices and Returns in the Art Market