Lecture 10 - Advanced Auction Theory Flashcards
Describe “The Winner’s Curse”
- When bidding in an auction where Common Values apply, whoever wins the auction will be whoever has the highest valuation
- The true or market value is likely to be the average of those market valuations.
- The highest valuation is thus likely to be too high.
- The winner of the auction will have over bid, thus becoming cursed
In the Winner’s Curse, how does a bidder view a value of the good?
- Each bidder has an unbiased private estimate (signal si) of the true value v.
- E(V/si) = si
If in bidder believe that errors are equally likely to be positive or negative, what does this say about the distribution of estimates?
They will be symmetric
Where are bidders estimates drawn from?
A probability distribution
Do all distributions look like a symmetrical bell curve?
No, some look like a straight horizontal line
What is on the 2 axes of a probability distribution curve?
Frequency and si
What is true about the error size of the most “optimistic” estimate?
This estimate will have the largest positive error
What are bids a function of/dependent on?
The estimates, i.e. bi= f(si)
Which side of a distribution is the winning bid drawn from? What is relevent about this?
The right hand side, hence is is likely higher than the true value of the good. i.e. b>v
What happens to a distribution if everyone revises their estimates downwards, conditional on them winning?
It shifts to the left
What is the optimal strategy to avoid The Winner’s Curse?
- Revise estimate (si) downwards, conditional on this being the highest.
- Calculate the optimal bid from the revised estimate
What happens to the size of the downward revision when the numbers of bidders increase?
The more bidders there are, the greater the downward revision
When trying to bid optimally, what di we assume about estimates (of si) and their range?
That they lie in the range 0 to 2v and are distributed uniformly
Is a person’s error more likely to be positive or negative?
Equally likely that error is positive or negative
What is the only circumstance that matters?
If si is the highest estimate
What are we assuming about each bidder?
That they are seeking to maximise their expected surplus
What is effectively applied to the private estimate of si when trying to bid optimally?
A discount factor
What happens to the revision when N tends to infinity? What does this show?
the revision tends closer to si/2, where the higher the N, the greater the downward revision
What discount factor is applied to the private estimate si?
si(N+1)/2N
Is the highest bidder of 100 bidders or 10 bidders likely to be further from the true value?
100 bidders
Theoretically does the sale price converge to as the number of bidders increases? Why?
The true value, as more information is acquired and ‘aggregated’, even though any individual bidder has only partial information.
What do the Law of Crowds state?
A large enough crowd is generally more correct in predictions than expert economists.
In research, does the sale price of a good converge to the true value as bidders increase?
Sometimes, but not generally, particularly where information is costly, and thus remains “private”
What does the optimal bid depend on?
The auction format.
What should the optimal bid be in a First Price Sealed bid auction for a common value good?
The optimal bid would be the original estimate si revised downwards twice by
[(N-1)/N] and [(N+1)/2N]