Antiquities at auction (1)

The Art Newspaper is reporting that for the auction houses Sotheby’s and Christie’s ‘Higher-volume, lower-priced business in the middle market could be the saleroom mantra for 2016’ as profits from the high-end market are squeezed. It reminds us that the auction houses are active commercial agents. They are not in business simply to facilitate transactions between buyers and sellers – they are in business to make a profit. With that rather obvious fact in mind, it is illuminating to look back over Sotheby’s New York sales records for non-Islamic antiquities from 1985 to 2013. The following series of graphs shows quite clearly how Sotheby’s has played the market.

Auctions 1

This first graph (above) shows how starting in 2001 Sotheby’s began offering fewer lots for sale annually, until about 2009, when the figure averaged out at just under 200 per year. Before then, in the 1980s and 1990s, the number of lots offered for sale annually had fluctuated in the region of 600 to 800. (It is interesting that the number of lots offered annually did not increase appreciably after the cessation of Sotheby’s London antiquities sales in 1997). At the same time, catalogues began including more information about provenance. By 2008, some information was included for nearly every lot offered for auction. That is not to say that the information provided was always useful for tracing back the ownership history and thus legitimacy of the lot offered for sale. Quite often, the ‘provenance’ might consist only of a name or a date. These minimal provenances might be interpreted as an honest attempt by Sotheby’s to meet customer concerns about provenance, or looked upon more cynically as an example of creative compliance, creating the appearance of meeting customer concerns while in reality carrying on business as usual. Either way, they do offer some reassurance to customers mildly worried about provenance, and maybe that is the point. But while such limited information might be of little use for investigating the legitimacy of a lot, it should not be dismissed out of hand. A false provenance is a fraudulent provenance, and fraud is a criminal offence. So any increase in provenance information does increase risk for an auction house, and is a step in the right direction towards a transparent, legitimate market. The nature of this provenance information will be explored further in a future post.

Auctions 2

The second graph shows that as the number of lots offered annually has declined, the mean price per lot sold has increased in real terms (all monetary values are standardized to 2005). Perhaps Sotheby’s has been been selling a smaller number of better quality objects. That would make sense. It might also mean though that the objects themselves were increasing in value through time as the market was increasing in value. That would also make sense. These two alternatives will be investigated further in a future post. (The 2007 price ‘spike’ is due to the sale on 7 June of a Late Hellenistic/Early Roman Imperial bronze statue of Artemis and the Stag for $28,600,000 (lot 41) and on 5 December of the Elamite ‘Guennol Lioness’ for $57,161,000. The 2010 price ‘spike’ is due to the sale on 7 December of a Roman Imperial bust of Antinous for $23,826,500).

Auctions 3

The third graph is particularly interesting. It examines the relationship between pre-sale price estimates provided by Sotheby’s in its catalogues and actual achieved prices. If the estimates were unbiased, it would be expected that achieved prices would distribute normally around estimate prices. This does seem to have been the case until about 2001, when the number of lots offered annually begins to decline. After 2002, a progressively higher number of lots achieve prices higher than the estimate. It is hard to believe that this is happening by chance, or that Sotheby’s specialists are consistently and accidentally undervaluing material. It seems more likely that estimates are being intentionally kept low to draw in potential buyers. ‘Come-hither’ estimates are an old-established auction practice to drum up custom. Once the potential buyers are bidding of course, they pay little attention to the estimate. But why would Sotheby’s want to attract more buyers?

Auctions 4

Traditionally, auction houses made their money from charging a seller’s commission, effectively a service charge levied on a seller. In 1975, in addition to the seller’s commission, both Sotheby’s and Christie’s began charging a buyer’s premium – a service charge levied on a buyer. From 1975 to 1992 the buyer’s premium remained steady at 10%. From 1993, however, Sotheby’s (and Christie’s) began to increase it, and to charge proportionately more for lower-priced lots. By 2013, Sotheby’s was charging 25% on prices up to $50,000, 20% on prices between $50,000 and $1 million, and 12% on prices in excess of $1 million. The fourth graph shows the effect of the increasing buyer’s premium from 1985 to 2013 on two notional lots priced respectively at $10,000 and $500,000. The auction houses have been forced to increase the buyer’s premium because in a competitive marketplace they are vulnerable to potential consignors ‘shopping around’. With auction houses keen to secure business, consignors can negotiate a deal to reduce or even dispense with seller’s commission. Auction house costs then have to be recovered from the buyers, who are less willing or able to negotiate.

Auctions 5

Graph number five shows the increasing revenue to Sotheby’s derived from the buyer’s premium. Notice the linear trend line increasing from about $350,000 in 1985 to about $4 million in 2013 (standardized to 2005 values).

Auctions 6

Finally, the sixth graph shows that the increasing value through time of lots being sold is real, and not just a function of the increasing buyer’s premium. (All sales data analyzed are from auction house results sheets, which incorporate the buyer’s premium into published prices).

This series of graphs strongly suggests that the changing configuration of auction sales through time is an outcome of a deliberate commercial strategy on Sotheby’s part to increase profitability by (1) reducing the number of lots offered, thereby decreasing associated handling costs; (2) drawing in more buyers with better provenanced lots and come-hither estimates, probably to increase prices by more competitive bidding; and (3) charging buyers progressively higher premiums.

This pattern is indicative of Sotheby’s sales strategy more generally, which since 2002 has been to focus more on the high end of the market [1]. Between 2002 and 2007, across the company, it halved the number of auction transactions and shed staff. Over that period, in a broader reflection of the antiquities sales data discussed here, the mean price per lot sold for all categories of object increased from $35,000 to $50,000. If the Art Newspaper is to be believed, however, we might be about to see this trend put into reverse.

Scholars (such as myself) who would like to use time series auction data as a proxy measure of illicit trade or regulatory impact would do well to take full account of the fact that auction houses are active commercial agents and it might be their commercial agency that is primarily structuring the data.


  1. Thompson, Don, 2008. The $12 million Stuffed Shark. New York: Palgrave, at 100.