A second-bid auction is one where all bidders submit their maximum willingness to pay, and then the bidder willing to pay the most pays what the second-highest bidder was willing to pay. An English auction is where bidders submit bids which they will have to pay, with the idea that once the second-highest bidder will stop raising the bid once they pass their threshold.
There’s a lot of theoretical work showing that second-bid auctions are all-around more efficient. English auctions can encourage the highest bidder to overbid, and the winner’s curse refers to the phenomenon that the winner of an auction is generally the person who overestimated its value by the most. Second bid auctions mitigate that by making them pay only the second highest estimate.
If JohnWittle is the only bidder in the auction, then in a second-bid auction he would receive gwern’s soul for free, but because this is an English auction, he has to pay his full bid, and so loses out for dramatically overestimating its market value- like gwern planned all along!
There’s a lot of theoretical work showing that second-bid auctions are all-around more efficient.
I’m don’t specialize in auctions, but this sounds wrong. A second-price auction and an English auction are strategically equivalent in most formal models. Nearly all auctions yield identical revenue and allocations when bidders are risk-neutral expected utility maximizers with independent values. Experimentally, the second-price auction tends to generate more revenue than an English auction, at least in the case of private values.
With common or correlated values (where the winner’s curse shows up), I’d think sealed bid auctions would lead to more winner overbidding than English or Dutch auctions. In these cases though, you really don’t have to worry about efficiency since everyone values the item equally.
I’m don’t specialize in auctions, but this sounds wrong. A second-price auction and an English auction are strategically equivalent in most formal models.
I should have been clearer by ‘all-around’; I meant that the incentives are lined up correctly, the costs are lower (every person only needs to submit one bid, and does not need to expend any effort monitoring the auction), gets exact results without requiring massive numbers of bids, and more information is conveyed by the end of the auction.
If JohnWittle is the only bidder in the auction, then in a second-bid auction he would receive gwern’s soul for free,
Well, yes, technically that’s true… but what prevents/discourages gwern (or his accomplice) from submitting an $N-1 bid (where N is the current sole bid amount)?
Typically, second-bid auctions are sealed, and all opened at once at the end of the auction, so it won’t be known that JohnWittle has bid, or how much he has bid, until the auction is over.
Not the same thing, surely? Submitting an N-1 bid causes the top bidder to pay effectively their bid… in effect turning a second-bid auction into an English auction as defined above. Setting a reserve price sets a floor that has no relationship to the top bidder’s bid.
But sure, the fact that you didn’t set a reserve price also suggests that you wouldn’t take advantage of this loophole in your counterfactual second-bid auction.
This makes this exchange all the more amusing.
I’m obviously missing something, but tally ho, I’ll find out eventually!
A second-bid auction is one where all bidders submit their maximum willingness to pay, and then the bidder willing to pay the most pays what the second-highest bidder was willing to pay. An English auction is where bidders submit bids which they will have to pay, with the idea that once the second-highest bidder will stop raising the bid once they pass their threshold.
There’s a lot of theoretical work showing that second-bid auctions are all-around more efficient. English auctions can encourage the highest bidder to overbid, and the winner’s curse refers to the phenomenon that the winner of an auction is generally the person who overestimated its value by the most. Second bid auctions mitigate that by making them pay only the second highest estimate.
If JohnWittle is the only bidder in the auction, then in a second-bid auction he would receive gwern’s soul for free, but because this is an English auction, he has to pay his full bid, and so loses out for dramatically overestimating its market value- like gwern planned all along!
I’m don’t specialize in auctions, but this sounds wrong. A second-price auction and an English auction are strategically equivalent in most formal models. Nearly all auctions yield identical revenue and allocations when bidders are risk-neutral expected utility maximizers with independent values. Experimentally, the second-price auction tends to generate more revenue than an English auction, at least in the case of private values.
With common or correlated values (where the winner’s curse shows up), I’d think sealed bid auctions would lead to more winner overbidding than English or Dutch auctions. In these cases though, you really don’t have to worry about efficiency since everyone values the item equally.
I should have been clearer by ‘all-around’; I meant that the incentives are lined up correctly, the costs are lower (every person only needs to submit one bid, and does not need to expend any effort monitoring the auction), gets exact results without requiring massive numbers of bids, and more information is conveyed by the end of the auction.
Well, yes, technically that’s true… but what prevents/discourages gwern (or his accomplice) from submitting an $N-1 bid (where N is the current sole bid amount)?
Typically, second-bid auctions are sealed, and all opened at once at the end of the auction, so it won’t be known that JohnWittle has bid, or how much he has bid, until the auction is over.
Ah. (nods) That makes sense.
If I were going to do that, I would simply have set a reserve price.
Not the same thing, surely? Submitting an N-1 bid causes the top bidder to pay effectively their bid… in effect turning a second-bid auction into an English auction as defined above. Setting a reserve price sets a floor that has no relationship to the top bidder’s bid.
But sure, the fact that you didn’t set a reserve price also suggests that you wouldn’t take advantage of this loophole in your counterfactual second-bid auction.