Opinion

Felix Salmon

Why you can’t always trust auction results

By Felix Salmon
November 21, 2011
art

Back in May, Sarah Thornton started worrying about the system of third-party guarantees and irrevocable bids at high-profile auction houses. I struggled to see what the fuss was about:

What’s Thornton’s beef with private sales, especially when the final price is public? A $50 million Warhol is a $50 million Warhol, whether the sale takes place in public, in private, or somewhere in between — and whether the sale is deliberate and orchestrated or whether it’s chaotic and unpredictable.

Now, however, Thornton has answered my question, in a new post. It turns out, according to her, that a $50 million Warhol — or, more to the point, a $43,202,500 Lichtenstein — might not be quite what it seems after all.

It is strongly believed, for example, that Guy Bennett, an art advisor acting on behalf of the Qataris, negotiated third-party guarantees on the top lots at both Christie’s and Sotheby’s recent contemporary auctions in New York. During the prestigious evening sale at Christie’s on November 8th, Mr Bennett was seen to make the winning bid of $38.5m for Roy Lichtenstein’s 1961 painting, “I can see the whole room!… and There’s Nobody in it!”. Christie’s normal buyer’s premium (or commission) on this would bring the final price up to $43.2m, which was the price reported by Christie’s. However, high-powered guarantors often negotiate a 50-50 split with the auction house of as much as 30% of the overage (the amount generated above the guaranteed price) and an additional 50% of the buyer’s premium. The market believes the Lichtenstein was guaranteed at $35m. If Mr Bennett, who bought the picture, had negotiated such a deal, the real price he paid would have been $40.3m.

This is more complicated than it needs to be, and of course it’s all based on what “the market believes”, which might not be fully accurate. But in principle, there’s definitely the possibility that something fishy might be going on whenever a guarantor buys a painting at auction for more than its low estimate.

There’s a lot of opacity when it comes to fine-art auctions, but the one thing which is always fully transparent is the final sale price. Sellers can negotiate deals when it comes to the commission they pay the auction house; buyers can’t. The buyer’s premium is set, and is public, and can always be used to find out the total amount of money which was spent by the buyer on the work of art in question. That’s why art-price databases use auction results: they know exactly how much was spent, even if they don’t know who the buyer was.

But now, with the system of third-party guarantees, we no longer know for sure that the reported price was in fact the price actually paid by the buyer.

The reason is this: in return for providing an irrevocable bid for a certain artwork (normally, but not always, at the low estimate), the auction house agrees to split some of the upside with the guarantor. Take that Lichtenstein. Its low estimate was $35 million; let’s say that someone called Guy Guarantor provides an irrevocable bid at that price. In return, he gets a deal from Christie’s. If the painting sells for more than $35 million, he’ll get 15% of the increase in the hammer price over and above $35 million, and he’ll also get 50% of the total buyer’s premium.

As it happened, the Lichtenstein was hammered down for $38.5 million — that’s $3.5 million more than the low estimate at which Guy Guarantor was providing the irrevocable bid. The total reported price for the painting was $43,202,500, including a buyer’s premium of $4,702,500. So Christie’s takes 15% of $3.5 million, or $525,000, and adds it to 50% of $4,702,500, which is $2,351,250. The total — $2,876,250 — is the amount that’s owed to Guy Guarantor under the terms of their deal.

That’s all fine, as far as I’m concerned. The buyer pays a total of $43,202,500, and that money gets divvied up between various parties. Christie’s gets some, Guy Guarantor gets some, and the seller, of course, gets most of it. As long as we know the amount the buyer is willing to spend on the painting, the auction market remains as transparent as it’s ever been.

But what happens if Guy Guarantor is the buyer? According to Thornton, the agreement is still in effect — which means that as he’s handing over a check for $43,202,500 with one hand, he’s simultaneously receiving a check for $2,876,250 with the other. The total amount the buyer pays, in this instance, is not $43,202,500, but rather $40,326,250.

That makes a difference. For one thing, if the work actually sold for $40,326,250 rather than $43,202,500, it would no longer be an auction record for the artist: “Ohhh… Alright…”, a 1964 canvas, sold in the same sale last year for $42,642,500. Is “I can see the whole room” the most expensive Lichtenstein ever sold at auction? According to Christie’s it is. But given Thornton’s reporting, there has to be some doubt about this.

I’ve got a call in to Christie’s about this, and we’ll see what they say, but I’ll only really be happy with one thing: a clear statement that a guarantor — or any other buyer, for that matter — cannot get any kind of rebate when he ends up being the high bidder. In order for published auction prices to mean anything, we need to know that the buyer actually paid the sum reported. Right now, we don’t know that — and in the case of the $43,202,500 Lichtenstein, there’s solid reason to believe that the price as reported by Christies, and embedded in places like the Artnet auction database, could actually be a lie.

(Incidentally, the Lichtenstein is a great poster child not only for auction-house opacity, but also for contemporary-art price appreciation. It was last sold for $2,090,000 in 1988 — and before that was sold for $450 in 1961. That’s a 50-year CAGR of somewhere over 25%.)

Update: Christie’s gets back to me, with this statement. The emphasis is mine:

When a third party agrees to finance all or part of Christie’s interest in a lot, it takes on all or part of the risk of the lot not being sold, and will be remunerated in exchange for accepting this risk. The third party may also bid for the lot. Where it does so, and is the successful bidder, the remuneration may be netted against the final purchase price. If the lot is not sold, the third party may incur a loss.

This seems to me an admission that the recorded final purchase price may indeed not be the actual price of the artwork to the buyer. And that therefore we can not trust that, to take just this example, a new artist record has just been set for Roy Lichtenstein. And, I can’t see how the third party will ever incur a loss, unless they never wanted the artwork in the first place. The worst-case scenario, for the guarantor, is that they end up buying the artwork for the amount they said they were willing to buy it for.

Update 2: TGDC and Auros, in the comments, make the case that Guy Guarantor really did pay the full $43,202,500:

He paid in in the form of the $40.3mm in net cash paid PLUS forgoing the $2.9mm he would have otherwise gotten had he not won. The guarantee is a sunk cost/gain since it happens in advance and is irrevocable. At the margin, he decided he’d rather have the painting than the $43mm in cash he’d have if he didn’t make that final bid. That is the market price.

If Guy had decided not to make a further bid above his original irrevocable $35M, he would’ve gone home with a paycheck. The service he provided, for that paycheck, has been renedered, whether he bids or not. The paycheck has to be considered separately from the cost of the art.

I think that TGDC is wrong that Guy Guarantor would have had $43 million in cash if he hadn’t made that final bid. In fact, to make that bid he needed only $40.3 million in cash. But the case here is definitely colorable. It just doesn’t change the fact that the buyer didn’t actually pay the amount he’s reported to have paid.

Update 3: I feel like I’m disappearing down a rabbit hole here, but I’m going to make one more stab at the question of Guy Guarantor’s cash situation. We can basically bring it down to two choices: either he buys the painting, or he doesn’t. (We’ll assume it’s hammered down for $38.5 million either way.) If he buys the painting, he’s down $40.3m in cash, and up one painting. If he doesn’t buy the painting, he’s up $2.9m in cash. So it’s reasonable to say that, in cash terms, the difference between buying and not buying is the difference between -$40.3m and +$2.9m, which is the headline $43.2m figure.

So here’s the question. When you make a purchase, is the amount you spend on that purchase the amount you spend on that purchase? Or does it make more sense to think of it as the difference between the amount of money you have after the transaction, on the one hand, and the amount of money you would have spent if you hadn’t made the transaction, on the other?

In the case of art, there’s a case to be made for looking at it either way. But by convention, sale prices never include sales tax. If I buy a painting at auction in New York and hang it in my New York apartment, I have to pay sales tax on that. But the price of the painting is not the full amount I pay; it’s just the price of the painting, sans tax. If auction results exclude sales tax for those people who pay it, they should also exclude the opportunity cost of guarantors’ foregone profit, I think.

Comments
8 comments so far | RSS Comments RSS

I think your logic is flawed. Guy Guarantor did pay $43mm. He paid in in the form of the $40.3mm in net cash paid PLUS forgoing the $2.9mm he would have otherwise gotten had he not won. The guarantee is a sunk cost/gain since it happens in advance and is irrevocable. At the margin, he decided he’d rather have the painting than the $43mm in cash he’d have if he didn’t make that final bid. That is the market price.

Posted by TGDC | Report as abusive
 

“However, high-powered guarantors often negotiate a 50-50 split with the auction house of as much as 30% of the overage (the amount generated above the guaranteed price).” I’m a little confused here… isn’t 15% of the overage the same thing as “a 50-50 split of 30% of the overage”?

Posted by AdamJ23 | Report as abusive
 

I think I agree with TGDC, above. If Guy had decided not to make a further bid above his original irrevocable $35M, he would’ve gone home with a paycheck. The service he provided, for that paycheck, has been renedered, whether he bids or not. The paycheck has to be considered separately from the cost of the art.

If I work for an auction house full time for a salary of $100k, and I decided to let my employer auction off my deceased grandmother’s estate, and that costs $10k, would you then say that “really” I am making $90k? No; these are two completely separate, unrelated transactions. (Certainly the IRS will be upset if I claim to be making that much less in taxable wages!)

Posted by Auros | Report as abusive
 

AdamJ23: The guarantor promises that if nobody bids over $35M, he will pay that much and take the art. In exchange for providing a guarantee that the art will not sell at an embarrassingly low price, he gets two things: First, if the art does sell for more than $35M, he gets 15% * (Actual Price – Guarantee Price). Second, the auction house is charging the buyer a 30% fee on the ENTIRE final bid, and guarantors who work with the auctioneer frequently can negotiate to take half of that as well.

So the guarantor’s total take is:

( 15% * (Actual – Guarantee) ) + 30%/2 * Actual
= 30% * Actual – 15% Guarantee

Which basically means that in exchange for avoiding an embarassing “short sale”, the auction house is cutting their own take from 30% of the actual sale price, to 15% of the guaranteed price.

Posted by Auros | Report as abusive
 

I am actually wondering if the auction house can take more than one guarantor. Competition always makes things more complicated.

Posted by history_student | Report as abusive
 

“I think that TGDC is wrong that Guy Guarantor would have had $43 million in cash if he hadn’t made that final bid. In fact, to make that bid he needed only $40.3 million in cash.”

Um, Felix? Math check here?

$40.3 cash previously + $2.9 payment for guarantee = $43.2 cash if he forgoes bidding.

What I don’t understand, though, is why anybody would care. The fact that somebody is willing to pay $43M for a painting doesn’t make the painting worth $43M (as anybody who bought Las Vegas real estate in 2005-2007 can attest). The painting is either worth:

(1) The value it brings by its existence and possession, independent of what it has sold for in the past. (Hard to imagine that being worth the cost of providing a high school education for 1,000 students, but we know the 1% have skewed values.)

(2) The value you could sell it for tomorrow. Which depends on having ANOTHER buyer willing to pay that price.

Posted by TFF | Report as abusive
 

I think you’re conflating two concepts, hence the confusion. One is the price of the painting, and the other is what the buyer paid for it.

It helps to think of the guarantee as a derivative contract (yes, everything really is a derivative). The guarantor has entered a contract with a payoff determined by the price set in the auction. If the guarantor didn’t participate in the auction, this is just a side bet. If the guarantor actually bids in the auction and wins, the contract was a hedge, so it certainly makes sense to add in this payoff when determining what the buyer actually paid.

In either case, the price that cleared the auction should be reasonably reflective of what the market is willing to pay for the painting.

Posted by niveditas | Report as abusive
 

TFF, the fact that a single buyer was willing to pay price X at a public auction actually gives you a substantial amount of information about what the painting would sell for to another buyer. If no other buyers were interested in the painting at between 35M and 38.1M, it would have sold for the low guarantee of 35M. So we know that there was at least one other buyer willing to pay some amount between those two numbers. If we know more about the bidding process, we could further refine that number. If the bidding is fine grained enough, then using the purchase price as a “value” is a reasonable rough estimate.

Felix: I think others are correct that 43.2 is essentially the real price, since it is the difference in the buyer’s financial assets (minus the painting) between buying and not buying. I have one niggle, though. Since the bidding is not infinitely fine-grained. What matters is what price the painting would have sold for if the gaurantor had not made the final bid. If this is, say 100k less than the final purchase price, then 65% (or whatever the deal is) of that difference should be taken off the price. If the steps were much wider, and this difference was 500k or even 1million, that ends up making a pretty big difference in the actual purchase price.

Posted by michaelsullivan | Report as abusive
 

Post Your Comment

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 http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/
  •