« Outliers | Main | The economics of vending machines »
The "afternoon effect" for artworks
One resilient puzzle identified in the literature is the “declining price anomaly.” This effect was identified by Ashenfelter (1989) and is an obvious repudiation of the law of one price. It refers to the observation that as an auction proceeds, the prices of the lots decline, even for identical goods (e.g., wines). Beggs and Graddy (1997) established the existence of the “declining price anomaly” for heterogeneous goods using data for Contemporary and Impressionist art auctions. This has generated great interest and a number of papers now report somewhat conflicting results in this respect, although the majority still seems to find evidence in favor of this anomaly (see Ashenfelter and Graddy 2003, and Ginsburgh and van Ours 2007.) In light of this controversy, it is of interest to investigate whether or not Latin American art auctions are also subject to the declining price anomaly or the so-called “afternoon effect” (“morning after effect” would be a more appropriate name in this context as Latin Art auctions occur in two parts, the first starting late in the day, say 7pm, and the second starting earlier the following day, usually 10am.) In line with previous research (Beggs and Graddy 1997), we find strong evidence that the “declining price anomaly” holds for Latin Art data, even after controlling for auction and artist unobserved characteristics (dummies) and a huge array of paintings characteristics, including reputation and provenance.
Here is the link and yes I do believe this is true. I believe it is mostly neuroeconomics at work, namely that we are more excited by new offerings than by familiar offerings. Similarly, a painting that has been "shopped around" usually goes for a lower price than a comparable picture coming on the market for the first time in many years; admittedly it is hard to segregate out the selection bias here. So if the same Jasper Johns print is being auctioned at 10 a.m. and 1 p.m., some people just don't want to wait with their bids. I wonder also if there is a theorem about how an asymmetric distribution of risk-averse bidders, fearing they might not get the work at all, could generate the same price pattern,
An alternative hypothesis -- likely true in part -- is that even "identical" artworks differ slightly in quality and the auction houses sell the better one first, if only to create a price precedent and excitement effect for the second one later in the day.
Posted by Tyler Cowen on May 20, 2008 at 06:58 AM in The Arts | Permalink
Comments
Similar effects occur in the stock market.
For example, when a high-quality stock has good news overnight, a great many people will "buy at the open" and prices will open higher. Most of the time, the price will drop from the opening price, often substantially, until daytraders and institutional buyers start picking up the somewhat less expensive shares later in the morning, at which point the stock will often rally.
This pattern does not always hold true, but it is true often enough to make money with. . .
Posted by: Matthew at May 20, 2008 6:44:31 AM
I attend vehicle auctions quite a lot and do not observe this. In my experience, the first of the products get low balled, repeated almost identical products will rise in price, flatten and then usually sink towards the end. Depending on the demand for the product of course.
Posted by: Richard at May 20, 2008 7:04:50 AM
Or is it do with notion that was seemed rare in the opening part isn't so rare if the same sort of good is being auctioned later on. Is it like Eric Cartman find out that Craig & friends now also bought ninja weapons. "Aw man, now every douchebag in town has ninja weapons! Lame!"
Posted by: Gil at May 20, 2008 7:07:52 AM
Maybe there is some sort of self-selection effect. More enthusiastic bidders use up their budget faster, so in the later auctions only the more prudent bidders compete.
Posted by: Stefano at May 20, 2008 7:18:22 AM
For duplicate items, couldn't this just be the winner's curse in action? The first person bought it for more than anyone else at the auction thought it was worth; therefore if a second copy comes up, as long as the first person isn't buying it again, the second winner is very likely to pay less.
For instance, on EBay, two of us bid a rare but obscure record up into the $90 range in one auction. The next time a copy came up, it didn't go for nearly as much, because there was only one of us left who wanted it that badly. Further copies will probably go for even less.
(BTW, my uncle claims prices in general have gone down long term on EBay... same effect?)
Posted by: Sol at May 20, 2008 7:36:03 AM
What am I missing? The bidder willing to pay the most gets the first lot. Once that bidder is out of the picture prices drop. That makes sense to me.
The law of one price doesn't hold in very thin markets, does it?
Posted by: Bernard Yomtov at May 20, 2008 9:03:53 AM
Firstly, I haven't had time to read the study, but I think Sol hits the nail on the head - are we assuming that the same high-value bidders are hanging around for another go, or do they just go home with their purchase. If an auction is doing its job, the high-value bidders get the first lots, and prices decline.
On a tangential note, is it in auctioneers' interest to put big, high-publicity (like last week's Freud and Bacon) items up near the start of an auction or the end? Beginning, you might get some decent consolation bidding on later items; End, one might find low-balling to start, but plenty of spare budget to push up the price of the main lot?
Anyway, both issues are linked to whether or not people are planning on bidding for multiple items - how heterogeneous are their preferences, and how substitutable are works of art? Being just a teacher, I don't have the finances to persue any active research! How strong a Veblen effect is there in high art - does it matter which piece of art you have, as long as you have one?
Posted by: Andrew at May 20, 2008 9:18:07 AM
If the auctioneer deliberately constructed the auction to produce non-equilibriating results, then we have to ask why. The auctioneer would seek to construct efficient markets to maximize transaction*value.
He knows how to do this by offering the average item to the average buyers at average intervals so everyone can set their average expectations, then he alternatively produces the odder items as time progresses to fill in the full market function.
Why doesn't the auctioneer do this? Probably because the auction is driven by one or two major sellers wanting to move specific high valued items. In other words, the sellers are essentially a semi-monopoly causing distortion.
Posted by: Matt at May 20, 2008 10:12:37 AM
I wanted to post the same thing as Sol, but he did it better and earlier.
Another effect could relate to multiple item buyers. They come in expecting to spend $X and for the first item, they have all of that $X to spend, even if they know they don't want to spend it all on the first item. As they have less money to spend, what is left becomes more dear.
Posted by: anomdebus at May 20, 2008 12:25:55 PM
I did a similar study but with quarter horse auctions in Oregon. My results showed a similar “declining price anomaly” but it was less of an "afternoon affect" and more of a "mid-day effect." The prices started off low in the early morning, went up throughout the middle of the day, and then back down into the evening. I contributed it simply to the number of bidders in the stands, thus the heightened energy to bidding wars. It could be that horse culture is different than other kinds of auctions (people who come to auctions tend to already have too many horses so they are less likely to make it to the early hours, as they are still at home feeding the others ;o) ). Regardless, the time of day effect is a strange anomaly in auction theory.
Posted by: Hannah at May 20, 2008 1:03:32 PM
I saw the same result as Richard at auctions I have attended (low prices first and then higher). I attribute it to the desire to get a "bargain" and then fear of getting nothing at all. For example at a telescope/camera auction I went to, several identical tripods were auctioned. These retail in the 100+ dollar range. People apparently hoped to buy them at about 20 and so passed when the first ones went for 30. Late in the day the price climbed to 45 and then 55.
Posted by: RobbL at May 20, 2008 1:26:22 PM
Sol and Bernard Yomtov read my thoughts.
Posted by: Dave at May 20, 2008 1:38:03 PM
The declining price anomaly also seems to hold for rotisserie baseball auctions, even though everyone in the auction knows what the player universe is. There could be a winner's curse in play, or the declining number of owners who still need a third baseman.
This might actually become a great source of data now that more roto auctions take place on the Internet (the fantasyauctineer web site for example). Could be a good study source if someone assembled enough sample size.
Posted by: Matt Bruce at May 20, 2008 1:52:36 PM
Isn't this an example of the "law of diminishing returns"? I realize that's an ecomonic term, but in Hollywood, this law suggests that if you have something bad happen to a character in a film, the next bad thing will generate less emotional response. Each time something bad happens, the audience's reaction is less. Same thing with auction prices. Similar to what Tyler mentioned, they lack the newness pizazz of earlier offerings.
Alternately, when an audience in a nightclub is offered a few acts that they get to judge, the later acts are judged better than the earlier acts. The conventional wisdom among those who frequent (or used to DJ for) these events is that people are less critical at the end of the evening, than at the beginning. Could be the beer goggles effect too I suppose. :)
Posted by: Al at May 20, 2008 4:32:53 PM
how does this tie in to the anchoring effect??
Posted by: andres at May 20, 2008 7:31:03 PM
Care to extrapolate this to the auctions for interactive ads?
Posted by: michael webster at May 20, 2008 8:20:38 PM
Purely anecdotal, but I've seen the opposite happen often on ebay. If there are 2-3 of the same item for sale - say a microphone - closing in time in close proximity I've found it's best to go for the buy on the first one. I think alot of folks low ball the first item because they have others to fall back on, but then bid higher as supply dries up.
Time does have an impact and cost here though. Having to wait for future listings, or coming back for auctions a week later quickly becomes less appealing, especially for those of us who bid by sniping in the last seconds.
Posted by: Shane M at May 20, 2008 10:09:19 PM
