Sold before the gavel
Estimates anchor you, reserves are theatre, and on the biggest lots someone may have agreed to buy the watch before the doors even open. A beginner's guide to the BS that makes up auction records.
This essay took a bit longer than expected but here we are - I usually pick a title at the end, and this is the best I could come up with. I’ll readily admit that I think it’s a bit sh1t. The essay on the other hand, is not sh1t at all, if I may say so myself. I hope you enjoy it, despite the title 😂
Way back in 2017, even before the auctioneer at Christie’s stepped up to the rostrum, Salvator Mundi by Leonardo da Vinci had basically been sold. A third party had agreed to buy it for around $100m regardless of what happened during the actual auction on the day. Which meant there was no option for the sale not to happen, and the only unknown was by how much the gavel would beat the $100m floor. In hindsight, I guess it sucked for the guarantor because it was beaten by a lot. There was a 19-minute bidding war and then the hammer fell at $400 million ($450.3 million with fees). Right now, it’s still the most expensive artwork ever sold at auction. Regardless, the person who provided the guarantee still managed to collect a fat slice of the upside for his trouble, and he never even had to buy the thing for real; all he had to do was promise to buy it, and then got paid for the promise.
Estimated reading time: ~28 mins
Last time I wrote about the running order in an auction, and we also covered where in a sale-sequence the big watch will be placed, and why the crescendo is never usually reached on the final lot. That section in SDC Weekly was all about choreography: the saleroom is an instrument someone has tuned. This time we will talk about what’s beneath the choreography and get into the money plumbing to explore things like estimates, reserves, guarantees, and irrevocable bids.
The romantic idea of an auction is that it’s the closest thing the world has to honest price discovery. There’s no list price, no haggling in a back room, no manufacturer telling you what something is worth… it’s just a public room, which involves some sort of ‘real-time fight’ between people who want the thing, and a number at the end that is, supposedly ‘by definition’, exactly what the market will bear. So, they say it is the truest price there is, and that it was discovered in the open, in front of everyone.
That’s a lovely story, for sure. And much like most other lovely stories about markets, it’s doing a lot of arm-waving to stop people noticing the ‘engineering’ behind it. In reality, by the time a lot crosses the block, up to three separate mechanisms have already influenced that so-called ‘discovered’ price - one anchors it, one truncates it, and one can pre-arrange the whole thing before the doors open. And before you file all this under some art-world disease that watches are immune to, hold that thought, because in November 2023 all of these mechanisms went off at once in a Geneva watch saleroom - in public - and I even wrote about it at the time. We’ll get there eventually.
First, let’s cover the some of these mechanisms - roughly in order of how much they should bother you.
Estimates are anchors
This is the number you see first - the pre-sale estimate. £80,000–£120,000. Mostly, it appears to be some kind of honest forecast (the house’s best guess at where the watch will end up). Of course, it is not really a forecast, but a tool, and it’s doing at least two jobs that have nothing to do with telling you the truth.
The first job you already know if you read the last essay: anchoring, which is the Kahneman & Tversky bias where any number you’re shown, however arbitrary, will drag your own estimate toward it. Beggs and Graddy found it in salerooms specifically, where a lot’s price gets anchored on the figure attached to it, and bidders don’t adjust too far away from that figure. So the estimate was never a neutral guess. The house has chosen your reference point for you, and your sense of ‘a fair price’ will stay tethered to it whether you like it or not.
The second job is a little counterintuitive; you’d assume the house wants a high estimate - meaning it will aim high, anchor high, then sell high… right? Well, in a one-on-one negotiation, sure, that could be a good idea. But an auction isn’t a negotiation; an auction involves a crowd, and with crowds, this logic works in reverse. There’s a paper by Gillian Ku and colleagues with the self-explanatory title ‘Starting Low but Ending High’; they did a bunch of studies and found that in auctions, lower starting points produce higher final prices. You’ve probably seen this in action on eBay yourself, but I’ll share some reasons why this works.
A low number lowers the barrier to entry, which causes more people to start bidding. Once you’ve bid, you’ve sunk a little time and ego into the thing, so you escalate to defend your position (the same instinct that keeps you waiting in queues that aren’t moving - you may have heard the term sunk-cost fallacy). Also, a busy lot looks valuable - a crowd of bidders is itself a signal, so traffic begets traffic.1 When you put those together the cynical strategy seems easy enough; estimate it low, watch a dozen people pile in who’d never even have started at the legit number, and then let the competitive arousal (from the last essay) do the rest. A lower estimate is bait.
A working paper with the superb title ‘Biased Auctioneers’ concluded that pre-sale estimates don’t necessarily reflect what the house expects the hammer to be; they’re shaped by strategy and by the house’s own behavioural biases. Auction houses, the authors note, are brokers of information, and they get to decide how much of it you see. The long-standing trade suspicion, according to people who’d know, is that estimates are routinely set on the low side - on purpose - just to pack the room.
Reserve is a floor, and some of the climb to it is theatre
Hidden behind the published lot estimates is a number you’re often not allowed to see, called the reserve. This is the minimum the consignor (seller) will accept; below this, the watch just doesn’t sell. By near-universal convention the reserve is set at or below the low estimate - never above it - which is the one reliable and useful fact in this whole system, so file it away. The published estimate range and the secret reserve are joined at the hip, and the bottom of the range is roughly where the floor is.2
Auction theory is a bit counterintuitive regarding the reserve; one foundational result is Roger Myerson’s ‘Optimal Auction Design’ - the paper that, together with Riley and Samuelson’s the same year, basically founded ‘modern auction theory’, and eventually helped Myerson to a Nobel prize in 2007.3 Their main point was that the revenue-maximising reserve is higher than the seller’s own valuation, and - this is the strange part - it’s independent of how many bidders show up. So the theory says you should refuse to sell below a price that’s more than you’d privately accept, and hold that line whether two people or 200 turn up. You’d think setting a floor above your own walk-away number is stupid because you’re choosing, sometimes, to keep the thing rather than sell it - but it works, because the credible threat of not selling makes the people who do want it bid harder. A reserve is the seller’s option to walk away, and the option has value because it’s real.
Real salerooms then deviate from the theory when they keep the reserve secret (which the simplest theoretical models don’t endorse) and there’s some evidence that this secrecy actually costs them. The classic test sold Pokémon cards on eBay - Katkar and Lucking-Reiley compared public minimum bids against equivalent secret reserves and found the secret reserve tended to deter bidders and depress outcomes, because people don’t want to waste effort bidding ‘into the dark’ against an unknown floor. Auction houses keep reserves secret anyway, but this is more about consignor relations and flexibility than revenue maximisation.
Then there are also games which turn the reserve into pure theatre - I went through this back in the day. When the live bidding hasn’t yet reached the secret reserve, the auctioneer can keep bidding anyway, against the room, by taking bids that don’t come from anyone in particular. This is called ‘chandelier bidding’ (and also called ‘bidding off the wall’), where the auctioneer announces successive phantom advances, conjured out of thin air (or ‘off the chandelier’), to build momentum from the opening up toward the reserve. Apparently it’s legal, and it’s disclosed in the conditions of sale. This means that in the early life of a lot, some of the ‘bidding’ you’re watching is a performance - the auctioneer summons invisible competitors to carry the price to the floor below which the watch was never going to sell. Only above the reserve are you guaranteed to be watching a real fight. Below it, you may be watching a man bid against a light fixture. Anyway, if a lot fails to clear the reserve it’s ‘bought in’, and as I covered last time, a bought-in watch often gets burned - everyone saw that nobody wanted it, and that stigma tends to follow the watch to its next sale.
One of the comments under the last essay asked whether the rules on all this phantom bidding differ by venue - the commenter’s understanding was that an American auctioneer may bid against you up to the reserve, while in Hong Kong the house can run you all the way to your maximum. I went and checked, and the printed rules are more boring than that. Christie’s Hong Kong conditions use the same formula as everywhere else: the auctioneer may bid on behalf of the seller ‘up to but not including’ the reserve, will not identify those bids as the seller’s, and may not bid the seller’s interest at or above the reserve. No house’s published terms, anywhere, claim the right to bid you toward your own ceiling. That would be shill bidding proper, and nobody’s lawyers will write that down.
What really differs is the referee. In America there’s a statute behind the contract - UCC §2-328: if the seller’s bidding wasn’t disclosed, the buyer can void the sale, or take the lot at the last honest bid before the seller joined in. New York City used to go further and require a sign at the saleroom entrance - twelve inches by eighteen, letters an inch high - spelling out that the auctioneer could bid for the seller; the city binned those rules in 2022 to lighten the load on the houses. Hong Kong, on the other hand, never had an auctioneer statute to bin. There, the conditions of sale pretty much are the law, with general contract and fraud rules somewhere behind them. So it may seem like the same promise on paper, but you’d probably face very different consequences for breaking it.
But back to the original comment; that commenter’s instinct actually leads us to the absentee bid. If you leave a written bid you will have handed the house your maximum, in advance, and in writing. The terms promise to execute your bid as cheaply as the other bids and the reserve will allow on the day. But the problem is, the chandelier is permitted to climb to the reserve against you, so when it’s a lot with no legit competition, your written bid can still be walked up to the full reserve. You would have ‘won’, against a light fixture, at the top of the house’s floor. And besides, the reserve isn’t some law of nature; that number is what the house agreed with the seller, and they can revisit it before the sale - meanwhile your maximum is locked in the bidding book. Clearly, nobody has to break a single rule for that structure to favour the house. An absentee bid is a letter that names your price, and rather comically, it’s addressed to the people who set the floor.
—
So to recap… the estimate will anchor you, the reserve will truncate the bottom of the price ladder, your absentee bid will inform the machine of your ceiling in advance… and we haven’t even reached the trick that pre-sold the Leonardo yet.
