COMP323 – Introduction to Computational Game Theory

Auctions

Paul G. Spirakis

Department of Computer Science

University of Liverpool

Paul G. Spirakis (U. Liverpool) Auctions 1 / 90

Outline

1 Introduction

2 Second-price sealed-bid auctions

3 First-price sealed-bid auctions

4 Variants

5 Auctions with imperfect information

Paul G. Spirakis (U. Liverpool) Auctions 2 / 90

 Introduction

1 Introduction
What is an auction?
Types of auctions

2 Second-price sealed-bid auctions

3 First-price sealed-bid auctions

4 Variants

5 Auctions with imperfect information

Paul G. Spirakis (U. Liverpool) Auctions 3 / 90

 Introduction What is an auction?

Auctions

In an auction, a good is sold to the party who submits the highest bid.
It is a kind of economic activity used to allocate significant resources,
such as
works of art;
radio spectrum used for wireless communication;
Treasury bills and timber and oil leases.
Auctions take many forms:
bids may be called out sequentially or submitted in sealed envelopes;
the price paid may be the highest bid or some other price;
if more than one unit of good is being sold, bids may be taken on all
units simultaneously or the units may be sold sequentially.
A game-theoretic analysis of auctions helps us to understand the
consequences of various auction designs:
e.g., it suggests the design likely to be the most effective on allocating
resources, and
the design likely to raise the most revenue.
Paul G. Spirakis (U. Liverpool) Auctions 4 / 90
 Introduction Types of auctions

Types of auctions

We will consider the case of a seller auctioning one item to a set of

buyers.
The underlying assumption we make when modeling auctions is that
each bidder has an intrinsic value for the item being auctioned: she is
willing to purchase the item for a price up to this value, but not for
any higher price. We will also refer to this intrinsic value as the
bidder’s true value or simply valuation for the item.
There are four main types of auctions when a single item is being sold
(and many variants of these types).
1 Ascending-bid auctions, also called English auctions;
2 Descending-bid auctions, also called Dutch auctions;
3 First-price sealed-bid auctions;
4 Second-price sealed-bid auctions auctions, also called Vickrey auctions.

Paul G. Spirakis (U. Liverpool) Auctions 5 / 90

 Introduction Types of auctions

Ascending-bid or English auctions

Ascending-bid or English auctions are carried out interactively in real

time.
Bidders are present either physically or electronically.
The seller gradually raises the price.
Bidders drop out until finally only one bidder remains, and that bidder
wins the object at this final price.
Oral auctions in which bidders shout out prices, or submit them
electronically, are forms of ascending-bid auctions.

Paul G. Spirakis (U. Liverpool) Auctions 6 / 90

 Introduction Types of auctions

Descending-bid or Dutch auctions

Descending-bid auctions are also carried out interactively in real time.

The seller gradually lowers the price from some high initial value until
the first moment when some bidder accepts and pays the current
price.
These auctions are called Dutch auctions because flowers have long
been sold in the Netherlands using this procedure.

Paul G. Spirakis (U. Liverpool) Auctions 7 / 90

 Introduction Types of auctions

First-price sealed-bid auctions

In a first-price sealed-bid auction, bidders submit simultaneous sealed

bids to the seller.
The terminology comes from the original format for such auctions, in
which bids were written down and provided in sealed envelopes to the
seller, who would then open them all together.
The highest bidder wins the object and pays the value of her bid.

Paul G. Spirakis (U. Liverpool) Auctions 8 / 90

 Introduction Types of auctions

Second-price sealed-bid or Vickrey auctions

In a second-price sealed-bid auction, bidders submit simultaneous

sealed bids to the sellers.
The highest bidder wins the object and pays the value of the
second-highest bid.
These auctions are called Vickrey auctions in honor of William
Vickrey, who wrote the first game-theoretic analysis of auctions
(including the second-price auction). Vickrey won the Nobel
Memorial Prize in Economics in 1996 for this body of work.

Paul G. Spirakis (U. Liverpool) Auctions 9 / 90

 Introduction Types of auctions

Comparing auction formats

A purely superficial comparison of the first-price and second-price

sealed-bid auctions might suggest that the seller would get more
money for the item if she ran a first-price auction: after all, she’ll get
paid the highest bid rather than the second-highest bid.
It may seem strange that in a second-price auction, the seller is
intentionally undercharging the bidders.
But such reasoning ignores one of the main messages from our study
of game theory: that when you make up rules to govern people’s
behavior, you have to assume that they will adapt their behavior in
light of the rules.
Here, the point is that bidders in a first-price auction will tend to bid
lower than they do in a second-price auction, and in fact this lowering
of bids will tend to offset what would otherwise look like a difference
in the size of the winning bid.
Paul G. Spirakis (U. Liverpool) Auctions 10 / 90
 Second-price sealed-bid auctions

1 Introduction

2 Second-price sealed-bid auctions

Formulation as a game
Analysis of equilibria

3 First-price sealed-bid auctions

4 Variants

5 Auctions with imperfect information

Paul G. Spirakis (U. Liverpool) Auctions 11 / 90

 Second-price sealed-bid auctions

Introduction

The sealed-bid second-price auction is particularly interesting, and

there are a number of examples of it in widespread use:
the auction form used on eBay is essentially a second-price auction;
the pricing mechanism that search engines use to sell keyword-based
advertising is a generalization of the second-price auction.
One of the most important results in auction theory is that with
independent, private values, bidding your true value is a dominant
strategy in a second price sealed-bid auction: the best choice of bid is
exactly what the object is worth to you.

Paul G. Spirakis (U. Liverpool) Auctions 12 / 90

 Second-price sealed-bid auctions

Motivation

Every person is certain for her valuation of the object before the
bidding begins.
Therefore we can assume that each person decides, before bidding
begins, the most she is willing to bid (her maximal bid).
When the players carry out their plans, the winner is the person
whose maximal bid is highest.
How much does she need to bid? To win, she needs to bid slightly
more than the second highest maximal bid.
If the bidding increment is small, we can take the price the winner
pays to be equal to the second highest maximal bid.

Paul G. Spirakis (U. Liverpool) Auctions 13 / 90

 Second-price sealed-bid auctions Formulation as a game

Formulation as a strategic game

We can model such a second-price sealed-bid auction as a strategic game

in which
each player chooses an amount of money, interpreted as the maximal
amount she is willing to bid, and
the player who chooses the highest amount obtains the object and
pays a price equal to the second highest amount.
To define the second-price sealed-bid auction precisely, denote
vi the value player i attaches to the object;
if i obtains the object at the price p her payoff is vi − p.

Paul G. Spirakis (U. Liverpool) Auctions 14 / 90

 Second-price sealed-bid auctions Formulation as a game

Formulation as a strategic game

The players’ valuations of the object are assumed to be all different
and all positive.
The set of players is denoted N = {1, 2, . . . , n} so that
v1 > v2 > · · · > vn > 0 .
Each player i submits a (sealed) bid bi .
If player i’s bid is higher than every other bid, she obtains the object
at a price equal to the second-highest bid, say bj , and receives payoff
vi − bj .
If some other bid is higher than player i’s bid, player i does not obtain
the object and receives the payoff of zero.
If player i is in a tie for the highest bid, her payoff depends on the
way ties are broken: a simple assumption is that the winner is the
bidder of the smallest number (i.e., highest valuation) among those
submitting the highest bid.
Paul G. Spirakis (U. Liverpool) Auctions 15 / 90
 Second-price sealed-bid auctions Formulation as a game

Formulation as a strategic game

In summary, a second-price sealed-bid auction is the following strategic

game:
Players: The set N = {1, 2, . . . , n} of the n bidders.
Actions: The set of actions of each player is the set of possible bids
(nonnegative numbers).
Payoffs: Denote by bi the bid of player i and by b the highest
submitted by a player other i.
If (a) bi > b or (b) bi = b and the number of every
other player who bids b is great than i, then player’s i
payoff is vi − b.
Otherwise, player i’s payoff is 0.

Paul G. Spirakis (U. Liverpool) Auctions 16 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Nash equilibria of second-price sealed-bid auction

The game has many Nash equilibria. One equilibrium is

(b1 , . . . , bn ) = (v1 , . . . , vn ) ,

i.e., each player’s bid is equal to her valuation of the object:

Since v1 > · · · > vn , the outcome is that player 1 obtains the object
at the price b2 , her payoff is v1 − b2 = v1 − v2 > 0 and every other
player’s payoff is 0.
This profile is a Nash equilibrium because:
If player 1 changes her bid to some b10 ≥ b2 , then the outcome does
not change. If she changes her bid to some b10 < b2 then she loses and
receives the payoff of zero.
If some other player i lowers her bid or raises it so some price at most
equal to b1 , then she remains a loser. If she raises her bid to some
bi0 > b1 then she wins the object but receives the payoff of
vi − b1 = vi − v1 < 0.
Paul G. Spirakis (U. Liverpool) Auctions 17 / 90
 Second-price sealed-bid auctions Analysis of equilibria

Nash equilibria of second-price sealed-bid auction

Another equilibrium is

(b1 , . . . , bn ) = (v1 , 0, . . . , 0) .

In this profile, player 1 gets the object and her payoff is zero.
The profile is an equilibrium because:
If player 1 changes her bid, then the outcome remains the same.
If any other player i raises her bid to bi0 , then either the outcome
remains the same (if bi0 ≤ v1 ) or causes player i to obtain the object at
a price that exceeds her valuation (if bi0 > v1 ).

Paul G. Spirakis (U. Liverpool) Auctions 18 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Nash equilibria of second-price sealed-bid auction

In both equilibria we just described, player 1 obtains the object. But there
are also equilibria in which player 1 does not obtain the object, e.g.,

(b1 , . . . , bn ) = (v2 , v1 , . . . , 0) ,

where player 2 obtains the object at the price v2 and every player (including
player 2) receives the payoff of zero. This profile is an equilibrium because
If player 1 raises her bid to v1 or more, she wins the object but her
payoff remains 0, because she pays v1 (the bid of player 2). Any other
change in her bid has no effect on the outcome.
If player 2 changes her bid to some other price greater than v2 , the
outcome does not change. If she changes her bid to v2 or less she
loses and her payoff remains 0.
If any other player raises her bid to at most v1 , the outcome does not
change. If she raises her bid above v1 , then she wins, but in paying
price v1 (bid by player 2) she obtains negative payoff.
Paul G. Spirakis (U. Liverpool) Auctions 19 / 90
 Second-price sealed-bid auctions Analysis of equilibria

Analysis of equilibria
We saw that

(v1 , v2 , . . . , vn ) , (v1 , 0, . . . , 0) , (v2 , v1 , . . . , vn )

are equilibria of the second-price sealed-bid auction.

Player 2’s bid in the last equilibrium exceeds her valuation
(b2 = v1 > v2 ).
If player 1 were to increase her bid to any value less than v1 , player’s
2 payoff would be negative (she would obtain the object at a price
greater than her valuation).
This property does not affect the fact that the profile is a Nash
equilibrium.
But the property does suggest that this equilibrium is less plausible as
the outcome of the auction than the first equilibrium, in which every
player bids her valuation.
Paul G. Spirakis (U. Liverpool) Auctions 20 / 90
 Second-price sealed-bid auctions Analysis of equilibria

Truth-telling is a dominant strategy

The weakness of the last equilibrium is reflected in the fact that player 2’s
bid v1 is weakly dominated by the bid v2 .
Theorem
In a second-price sealed-bid auction, a player’s bid equal to her valuation
weakly dominates all her other bids.

In other words, truth-telling is a weakly dominant strategy.

That is, for any bid bi 6= vi , player i’s bid vi is at least as good as bi ,
no matter what the other players bid, and is better than bi for some
actions of the other players.
A player who bids less than her valuation stands not to win in some
cases in which she could profit by winning (when the highest of the
bids is between her bid and her valuation).

Paul G. Spirakis (U. Liverpool) Auctions 21 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Truth-telling is a dominant strategy

Proof

Theorem
In a second-price sealed-bid auction, a player’s bid equal to her valuation
weakly dominates all her other bids.

Proof.
We need to show that if player i bids bi = vi , then no deviation from
this bid would improve her payoff, regardless of what strategy
everyone else is using.
There are two cases to consider: deviations in which i raises her bid,
and deviations in which i lowers her bid.

Paul G. Spirakis (U. Liverpool) Auctions 22 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Truth-telling is a dominant strategy

Proof

Theorem
In a second-price sealed-bid auction, a player’s bid equal to her valuation
weakly dominates all her other bids.

Proof (continued).
The key point in both cases is that the value of i’s bid only affects
whether i wins or loses, but never affects how much i pays in the
event that she wins (the amount paid is determined entirely by the
other bids, and in particular by the largest among the other bids).
Since all other bids remain the same when i changes her bid, a change
to i’s bid only affects her payoff if it changes her win/loss outcome.

Paul G. Spirakis (U. Liverpool) Auctions 23 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Truth-telling is a dominant strategy

Proof

Theorem
In a second-price sealed-bid auction, a player’s bid equal to her valuation
weakly dominates all her other bids.

Proof (continued). First, suppose that instead of bidding vi , player i

chooses a bid bi0 > vi .
This only affects player i’s payoff if i would lose with bid vi but would
win with bid bi0 .
In order for this to happen, the highest other bid bj must be between
bi and bi0 .
In this case, the payoff to i from deviating would be at most
vi − bj ≤ 0, and so this deviation to bid bi0 does not improve i’s payoff.

Paul G. Spirakis (U. Liverpool) Auctions 24 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Truth-telling is a dominant strategy

Proof

Theorem
In a second-price sealed-bid auction, a player’s bid equal to her valuation
weakly dominates all her other bids.

Proof (continued). Next, suppose that instead of bidding vi , player i

chooses a bid bi0 < vi .
This only affects player is payoff if i would win with bid vi but would
lose with bid bi0 .
So before deviating, vi was the winning bid, and the second-place bid
bj was between vi and bi0 .
In this case, i’s payoff before deviating was vi − bj ≥ 0, and after
deviating it is 0 (since i loses), so again this deviation does not
improve i’s payoff.

Paul G. Spirakis (U. Liverpool) Auctions 25 / 90
 Second-price sealed-bid auctions Analysis of equilibria

Truth-telling is a dominant strategy

In summary:

A second-price sealed-bid auction has many Nash equilibria, but the

equilibrium
(b1 , . . . , bn ) = (v1 , . . . , vn )
in which each player’s bid is equal to her valuation of the object, is
distinguished by the fact that every player’s action weakly dominates all
her other strategies.

Paul G. Spirakis (U. Liverpool) Auctions 26 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Truth-telling is a dominant strategy

Discussion

The fact that truthfulness is a dominant strategy makes second-price

auctions conceptually very clean.
Truthful bidding is the best thing to do regardless of what the other
bidders are doing.
So in a second-price auction, it makes sense to bid your true value
even if other bidders are overbidding, underbidding, colluding, or
behaving in other unpredictable ways.
In other words, truthful bidding is a good idea even if the competing
bidders in the auction don’t know that they ought to be bidding
truthfully as well.

Paul G. Spirakis (U. Liverpool) Auctions 27 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Second-price sealed-bid auction with two bidders

Let us compute all Nash equilibria of a second-price sealed-bid auction

with two bidders.
If player 2’s bid b2 is less than v1 then any bid of b2 or more is a best
response of player 1 (she wins and pays the price b2 ).
If player 2’s bid is equal to v1 then every bid of player 1 yields her the
payoff zero (either she wins and pays v1 , or she loses), so every bid is
a best response.
If player 2’s bid b2 exceeds v1 then any bid of less than b2 is a best
response of player 1. (If she bids b2 or more she wins, but pays the
price b2 > v1 , and hence obtains a negative payoff.)

Paul G. Spirakis (U. Liverpool) Auctions 28 / 90

 Second-price sealed-bid auctions Analysis of equilibria

Second-price sealed-bid auction with two bidders

In summary, player 1’s best response function is

 {b1 : b1 ≥ b2 } if b2 < v1
B1 (b2 ) = {b1 : b1 ≥ 0} if b2 = v1
{b1 : 0 ≤ b1 < b2 } if b2 > v1


and, by similar arguments, player 2’s best response function is


 {b2 : b2 ≥ b1 } if b1 < v2
B2 (b1 ) = {b2 : b2 ≥ 0} if b1 = v2 .
{b2 : 0 ≤ b2 < b1 } if b1 > v2


Therefore
The set of Nash equilibria is the set of pairs (b1 , b2 ) such that

either b1 ≤ v2 and b2 ≥ v1

or b1 ≥ v2 , b1 ≥ b2 , and b2 ≤ v1 .
Paul G. Spirakis (U. Liverpool) Auctions 29 / 90
 First-price sealed-bid auctions

1 Introduction

2 Second-price sealed-bid auctions

3 First-price sealed-bid auctions

Formulation as a game
Analysis of equilibria

4 Variants

5 Auctions with imperfect information

Paul G. Spirakis (U. Liverpool) Auctions 30 / 90

 First-price sealed-bid auctions

Introduction

A first-price sealed-bid auction differs from a second-price sealed bid

auction only in that the winner pays the price she bids, not the
second highest bid.
This means that the value of your bid not only affects whether you
win but also how much you pay.
As a result, most of the reasoning from the second-price sealed-bid
auction has to be redone, and the conclusions are now different.

Paul G. Spirakis (U. Liverpool) Auctions 31 / 90

 First-price sealed-bid auctions

Motivation

A first-price sealed-bid auction models an auction in which people

submit sealed bids and the highest bid wins.
It is also a model for a dynamic auction in which the auctioneer
begins by announcing the highest price, which she gradually lowers
until someone indicates her willingness to buy the object.
A bid is interpreted as the price at which the bidder will indicate her
willingness to buy the object in the dynamic auction.

Paul G. Spirakis (U. Liverpool) Auctions 32 / 90

 First-price sealed-bid auctions Formulation as a game

Formulation as a strategic game

A first-price sealed-bid auction is the following strategic game:

Players: The set N = {1, 2, . . . , n} of the n bidders.
Actions: The set of actions of each player is the set of possible bids
(nonnegative numbers).
Payoffs: Denote by bi the bid of player i and by b the highest
submitted by a player other i.
If (a) bi > b or (b) bi = b and the number of every
other player who bids b is great than i, then player’s i
payoff is vi − bi .
Otherwise, player i’s payoff is 0.

Paul G. Spirakis (U. Liverpool) Auctions 33 / 90

 First-price sealed-bid auctions Analysis of equilibria

Analysis of equilibria

A first-price sealed-bid auction has many Nash equilibria, but in all

equilibria the winner is the player who values the object most highly
(player 1):
In any action profile (b1 , . . . , bn ) in which some player i 6= 1 wins, we
have bi > b1 .
If bi > v2 , then i’s payoff is negative, so that she can do better by
reducing her bid to 0.
If bi ≤ v2 , then player 1 can increase her payoff from 0 to v1 − bi by
bidding bi , in which case she wins.
Thus no such action profile is a Nash equilibrium.

Paul G. Spirakis (U. Liverpool) Auctions 34 / 90

 First-price sealed-bid auctions Analysis of equilibria

Characterization of equilibria

The following theorem characterizes the set of Nash equilibria of a

first-price sealed-bid auction:
Theorem
An action profile (b1 , . . . , bn ) is a Nash equilibrium of a first-price
sealed-bid auction if and only if
the two highest bids are the same;
one of these bids is submitted by player 1; and
the highest bid is at least v2 and at most v1 .

Paul G. Spirakis (U. Liverpool) Auctions 35 / 90

 First-price sealed-bid auctions Analysis of equilibria

Characterization of equilibria
Proof

Proof (=⇒).
A profile of bids in which the two highest bids are not the same is not
a Nash equilibrium because the player naming the highest bid can
reduce her bid slightly, continue to win, and pay a lower price.
Recall that, in any equilibrium, player 1 wins the object. Thus she
submits one of the highest bids.
If the highest bid is less than v2 , then player 2 can increase her bid to
a value between the highest bid and v2 , win, and obtain a positive
payoff. Thus in an equilibrium the highest bid is at least v2 .
If the highest bid exceeds v1 , player 1’s payoff is negative, and she
can increase this payoff by reducing her bid. Thus in an equilibrium
the highest bid is at most v1 .

Paul G. Spirakis (U. Liverpool) Auctions 36 / 90

 First-price sealed-bid auctions Analysis of equilibria

Characterization of equilibria
Proof

Proof (⇐=). Any profile (b1 , . . . , bn ) of bids that satisfies the conditions
is a Nash equilibrium because:
If player 1 increases her bid she continues to win, and reduces her
payoff.
If player 1 decreases her bid she loses and obtains the payoff 0, which
is at most her payoff at (b1 , . . . , bn ).
If any other player increases her bid she either does not affect the
outcome, or wins and obtains a negative payoff.
If any other player decreases her bid she does not affect the outcome.


Paul G. Spirakis (U. Liverpool) Auctions 37 / 90

 First-price sealed-bid auctions Analysis of equilibria

Dominated bids in first-price sealed-bid auctions

In any equilibrium in which the winning bid exceeds v2 , at least one

player’s bid exceeds her valuation.
Such a bid seems “risky” because it would yield the bidder a negative
payoff if it were to win.
In the equilibrium there is no risk, because the bid does not win, but
(as in a second-price sealed-bid auction) the fact that the bid has this
property reduces the plausibility of the equilibrium.
This potential “riskiness” to player i of a bid bi > vi is reflected in
the fact that it is weakly dominated by the bid vi .

Paul G. Spirakis (U. Liverpool) Auctions 38 / 90

 First-price sealed-bid auctions Analysis of equilibria

Dominated bids in first-price sealed-bid auctions

In a first-price sealed-bid auction, a player i’s bid bi > vi is weakly

dominated by the bid vi .

Proof.
If the other players’ bids are such that player i loses when she bids bi ,
then the outcome is the same whether she bids bi or vi .
If the other players’ bids are such that player i wins when she bids bi ,
then her payoff is negative when she bids bi and zero when she bids vi
(regardless of whether this bid wins.


Paul G. Spirakis (U. Liverpool) Auctions 39 / 90

 First-price sealed-bid auctions Analysis of equilibria

Dominated bids in first-price sealed-bid auctions

However, in a first-price auction, unlike a second-price auction, a bid

bi < vi of player i is not weakly dominated by the bid vi . In fact,

In a first-price sealed-bid auction, a player i’s bid bi < vi is not weakly

dominated by any bid.

Proof.
A bid bi < vi is not weakly dominated by a bid bi0 < bi because if the
other players’ highest bid is between bi0 and bi , then bi0 loses, whereas
bi wins and yields player i a positive payoff.
A bid bi < vi is not weakly dominated by a bid bi0 > bi because if the
other players’ highest bid is less than bi , then both bi and bi0 win and
bi yields a lower price.


Paul G. Spirakis (U. Liverpool) Auctions 40 / 90

 First-price sealed-bid auctions Analysis of equilibria

Dominated bids in first-price sealed-bid auctions

Further, even though the bid vi weakly dominates higher bids, this bid is
itself weakly dominated, by a lower bid! vi

In a first-price sealed-bid auction, a player i’s bid vi is weakly dominated

by any bid bi < vi .

Proof.
If player i bids vi her payoff is 0 regardless of the other players’ bids.
If player i bids bi < vi her payoff is either 0 (if she loses) or positive
(if she wins).


Paul G. Spirakis (U. Liverpool) Auctions 41 / 90

 First-price sealed-bid auctions Analysis of equilibria

Dominated bids in first-price sealed-bid auctions

In summary,
Theorem
In a first-price sealed-bid auction, a player’s bid of at least her valuation is
weakly dominated, and a bid of less than her valuation is not weakly
dominated.
An implication of this result is that

In every Nash equilibrium of a first-price sealed-bid auction at least one

player’s action is weakly dominated.

Paul G. Spirakis (U. Liverpool) Auctions 42 / 90

 Variants

1 Introduction

2 Second-price sealed-bid auctions

3 First-price sealed-bid auctions

4 Variants
All-pay auctions
Multiunit auctions

5 Auctions with imperfect information

Paul G. Spirakis (U. Liverpool) Auctions 43 / 90

 Variants All-pay auctions

All-pay auctions

Some situations may be modeled as all-pay auctions in which every

bidder, not only the winner, pays.
An example is competition between lobby groups for government
attention: each group spends resources in attempt to win favor; the
one that spends the most is successful.
We will study both first- and second-price versions of an all-pay
auction with two bidders, in which both bidders pay the winning price.

Paul G. Spirakis (U. Liverpool) Auctions 44 / 90

 Variants All-pay auctions

Second-price all-pay auction

Two bidders, both pay the winning price

We start with the second-price version of an all-pay auction with two

bidders, in which both bidders pay the winning price, and we will try to
compute the set of Nash equilibria of the game.
The payoff function of bidder 1 is

−b1 if b1 < b2
u1 (b1 , b2 ) =
v1 − b2 if b1 ≥ b2

and that of bidder 2 is


−b2 if b2 ≤ b1
u2 (b1 , b2 ) =
v2 − b1 if b2 > b1

Paul G. Spirakis (U. Liverpool) Auctions 45 / 90

 Variants All-pay auctions

Second-price all-pay auction

Two bidders, both pay the winning price

A profile (b, b) is not a Nash equilibrium for any value of b because

player 2 can increase her payoff by either increasing her bid slightly or
by reducing it to 0.
A profile (b1 , b2 ) with 0 < b1 < b2 is not a Nash equilibrium, because
player 1 can lower her bid to 0 and receive a payoff of 0 > −b1 .
A profile (b1 , b2 ) with b1 > b2 > 0 is not a Nash equilibrium, because
player 2 can lower her bid to 0 and receive a payoff of 0 > −b2 .
It remains to consider the profiles (0, b2 ) and (b1 , 0) where b1 , b2 > 0.

Paul G. Spirakis (U. Liverpool) Auctions 46 / 90

 Variants All-pay auctions

Second-price all-pay auction

Two bidders, both pay the winning price

A profile (0, b2 ) where b2 > 0 is a Nash equilibrium if and only if b2 ≥ v1 :

(⇒) Assume (0, b2 ) is a Nash equilibrium. Player 1 receives a payoff of 0.
If player 1 raised her bid to b2 she would receive a payoff of v1 − b2 .
If v1 − b2 > 0 then (0, b2 ) could not be an equilibrium. Therefore
b2 ≥ v 1 .
(⇐) Assume b2 ≥ v1 . Then (0, b2 ) is a Nash equilibrium:
If player 1 raises her bid to some b10 so that 0 < b10 < b2 she would
receive a payoff of −b10 < 0. If she raises her bid to some b10 ≥ b2 she
would receive a payoff of v1 − b2 ≤ 0.
If player 2 raises her bid, or lowers it to some b20 > b2 > 0, her payoff
remains the same. If she lowers her bid to 0 she would receive a payoff
of 0, instead of v2 − b1 = v2 ≥ 0.

Paul G. Spirakis (U. Liverpool) Auctions 47 / 90

 Variants All-pay auctions

Second-price all-pay auction

Two bidders, both pay the winning price

Similarly, we can show that a profile (b1 , 0) where b1 > 0 is a Nash

equilibrium if and only if b1 ≥ v2 .
Therefore, to summarize:

The set of all Nash equilibria of a second-price all-pay auction with two
bidders, where both bidders pay the winning price, is the set of all pairs
(0, b2 ) and (b1 , 0) where b1 > 0, b2 > 0, b2 ≥ v1 , and b1 ≥ v2 .

Paul G. Spirakis (U. Liverpool) Auctions 48 / 90

 Variants All-pay auctions

First-price all-pay auction

Two bidders, both pay the winning price

We now consider the first-price version of an all-pay auction with two

bidders, in which both bidders pay the winning price, and we will try to
compute the set of Nash equilibria of the game.
In any Nash equilibrium the two highest bids are equal, otherwise the
player with the higher bid can increase her payoff by reducing her bid
a little (keeping it larger than the other player’s bid).
But no profile of bids in which the two highest bids are equal is a
Nash equilibrium, because the player with the higher index who
submits this bid can increase her payoff by slightly increasing her bid,
so that she wins rather than loses.
Therefore
A first-price all-pay auction with two bidders in which both bidders pay the
winning price has no Nash equilibrium.
Paul G. Spirakis (U. Liverpool) Auctions 49 / 90
 Variants Multiunit auctions

Multiunit auctions

In multiunit auctions, many units of an object are available, and each

bidder may value positively more than one unit.
Each bidder submits a bid for each unit of the good.
An action is a list of bids (bi1 , . . . , bik ), where bi1 is player i’s bid for
the first unit of the good, bi2 is her bid for the second unit, and so on.
The player who submits the highest bid for any given unit obtains
that unit.

Paul G. Spirakis (U. Liverpool) Auctions 50 / 90

 Variants Multiunit auctions

Multiunit auctions
Types of multiunit auctions

We will study three types of multiunit auctions, which differ in the prices
paid by the winners.
Discriminatory auction: The price paid for each unit is the winning bid for
that unit.
Uniform-price auction: The price paid for each unit is the same, equal to
the highest rejected bid among all the bids for all units.
Vickrey auction: A bidder who wins k objects pays the sum of the k
highest rejected bids submitted by the other bidders.

Note that the first type generalizes a first-price auction, whereas the
last two generalize a second-price auction.
We will study these auctions when two units of an object are available.

Paul G. Spirakis (U. Liverpool) Auctions 51 / 90

 Variants Multiunit auctions

Two-unit auctions

Assume that two units of an object are available.

There are n bidders.
Bidder i values the first unit that she obtains at vi and the second
unit at wi , where vi > wi > 0.
Each bidder submits two bids; the two highest bids win.
We will show that, in this setting,

In discriminatory and uniform-price auctions, player i’s action of bidding vi

and wi does not dominate all her other actions, whereas in a Vickrey
auction it does.

Paul G. Spirakis (U. Liverpool) Auctions 52 / 90

 Variants Multiunit auctions

Two-unit auctions
Discriminatory auction

In a discriminatory two-unit auction, player i’s action of bidding vi and wi

does not dominate all her other actions.
Proof. Recall that the price paid for each unit is the winning bid for that
unit.
To show that the action of bidding vi and wi is not dominant for
player i, we need only find actions for the other players and
alternative bids for player i such that player i’s payoff is higher under
the alternative bids than it is under the vi and wi , given the other
players’ actions.
Suppose that each of the other players submits two bids of 0.
Then if player i submits one bid between 0 and vi and one bid
between 0 and wi she still wins two units, and pays less than when
she bids vi and wi .

Paul G. Spirakis (U. Liverpool) Auctions 53 / 90
 Variants Multiunit auctions

Two-unit auctions
Uniform-price auction

In a uniform-price two-unit auction, player i’s action of bidding vi and wi

does not dominate all her other actions.
Proof. Recall that the price paid for each unit is the same, equal to the
highest rejected bid among all the bids for all units.
Suppose that some bidder other than i submits one bid between wi
and vi and one bid of 0, and all the remaining bidders submit two
bids of 0.
Then bidder i wins one unit, and pays the price wi .
If she replaces her bid of wi with a bid between 0 and wi then she
pays a lower price, and hence is better off.


Paul G. Spirakis (U. Liverpool) Auctions 54 / 90

 Variants Multiunit auctions

Two-unit auctions
Vickrey auction

In a Vickrey two-unit auction, player i’s action of bidding vi and wi

dominates all her other actions.
Proof. Recall that a bidder who wins k objects pays the sum of the k
highest rejected bids submitted by the other bidders.
Suppose that player i bids vi and wi . We will consider separately the cases
in which the bids of the players other than i are such that player i wins 0,
1, and 2 units.
Player i wins 0 units:
In this case the second highest of the other players’ bids is at least vi .
If player i changes her bids so that she wins one or more units, for
any unit she wins she pays at least vi .
Thus no change in her bids increases her payoff from its current value
of 0 (and some changes lower her payoff).
Paul G. Spirakis (U. Liverpool) Auctions 55 / 90
 Variants Multiunit auctions

Two-unit auctions
Vickrey auction

In a Vickrey two-unit auction, player i’s action of bidding vi and wi

dominates all her other actions.
Proof (continued).
Player i wins 1 unit:
If player i raises her bid of vi then she still wins one unit and the price
remains the same.
If she lowers this bid then either she still wins and pays the same
price, or she does not win any units.
If she raises her bid of wi then either the outcome does not change,
or she wins a second unit. In the latter case the price she pays is the
previously-winning bid she beat, which is at least wi , so that her
payoff either remains zero or becomes negative.

Paul G. Spirakis (U. Liverpool) Auctions 56 / 90

 Variants Multiunit auctions

Two-unit auctions
Vickrey auction

In a Vickrey two-unit auction, player i’s action of bidding vi and wi

dominates all her other actions.
Proof (continued).
Player i wins 2 units:
Player i’s raising either of her bids has no effect on the outcome.
Her lowering a bid either has no effect on the outcome or leads her to
lose rather than to win, leading her to obtain the payoff of zero.


Paul G. Spirakis (U. Liverpool) Auctions 57 / 90

 Variants Multiunit auctions

An application of multiunit auctions

Internet pricing

A proposal to deal with congestion on electronic message pathways is that

each message should include a field stating an amount of money the
sender is willing to pay for the message to be sent.
Suppose that:
During some time interval, each of n people wants to send one
message.
The capacity of the pathway is k < n messages.
The k messages whose bids are highest are the ones sent.
Each of the persons sending these messages pays a price equal to the
(k + 1)st highest bid.
Note that the auction differs from the two-unit auctions we studied before
because each person submits only one bid.

Paul G. Spirakis (U. Liverpool) Auctions 58 / 90

 Variants Multiunit auctions

An application of multiunit auctions

Internet pricing

However, the situation may be modeled as a multiunit auction in which

k units are available;
each player attaches a positive value to only one unit and submits a
bid for only one unit;
the k highest bids win; and
each winner pays the (k + 1)st highest bid.
By a variant of the argument for a second-price auction, in which highest
of the other players’ bids is replaced by highest rejected bid, we can show
that
A player’s action of bidding her value weakly dominates all her other
actions.

Paul G. Spirakis (U. Liverpool) Auctions 59 / 90

 Auctions with imperfect information Models of imperfect information

1 Introduction

2 Second-price sealed-bid auctions

3 First-price sealed-bid auctions

4 Variants

5 Auctions with imperfect information

Models of imperfect information
First-price auction with two bidders
Extensions

Paul G. Spirakis (U. Liverpool) Auctions 60 / 90

 Auctions with imperfect information Models of imperfect information

Imperfect information

Auctions with imperfect information:

So far, we have assumed every bidder knows every other bidder’s
valuation of the object for sale.
We will analyze auctions in which bidders are not perfectly informed
about each others’ valuations.
General setting:
We assume a single object is for sale, and each bidder independently
receives some information (a signal) about the value of the object to
her.
If each bidder’s signal is simply her valuation of the object, we say that
the bidders’ valuations are private.
If each bidder’s valuation depends on other bidders’ signal as well as
her own, we say that the valuations are common.

Paul G. Spirakis (U. Liverpool) Auctions 61 / 90

 Auctions with imperfect information Models of imperfect information

Imperfect information

Examples:
The assumption of private values are appropriate for a work of art
whose beauty than resale value interests the buyers. Each bidder
knows her valuation of the object, but not that of any other bidder,
and the other bidders’ valuations have no bearing on her valuation.
The assumption of common values is appropriate for an oil tract
containing unknown reserves on which each bidder has conducted a
test. Each bidder i’s test result gives her some information about the
size of the reserves, and hence her valuation, but the other bidders’
test results, if known to bidder i, would typically improve this
information.

Paul G. Spirakis (U. Liverpool) Auctions 62 / 90

 Auctions with imperfect information First-price auction with two bidders

First-price auction with private values

We want to capture a setting in which bidders know how many

competitors they have, and they have partial information about their
competitors’ values for the item. However, they do not know their
competitors’ values exactly.
We will focus on auctions with private values in which:
bids for a single object are submitted simultaneously (sealed-bid);
the participant who submits the highest bid obtains the object; and
the winner pays the price she bid (first-price auctions).

Paul G. Spirakis (U. Liverpool) Auctions 63 / 90

 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

The model

In the simple case, suppose that:

There are two bidders.
Each bidder has a private value that is independently and uniformly
distributed between 0 and 1.
Note that the fact that the 0 and 1 are the lowest and highest possible
values is not crucial; by shifting and re-scaling these quantities, we
could equally well consider values that are uniformly distributed
between any other pair of endpoints.
This information is common knowledge among the two bidders.

Paul G. Spirakis (U. Liverpool) Auctions 64 / 90

 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

Strategies

A strategy for a bidder is a function s(v ) = b that maps her true value v
to a non-negative bid b. We will make the following simple assumptions
about the strategies the bidders are using:
1 s(·) is a strictly increasing, differentiable function; so in particular, if

two bidders have different values, then they will produce different bids.
2 s(v ) ≤ v for all v : bidders can shade their bids down, but they will

never bid above their true values. Notice that since bids are always
non-negative, this also means that s(0) = 0.

Paul G. Spirakis (U. Liverpool) Auctions 65 / 90

 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

Strategies

A strategy for a bidder is a function s(v ) = b that maps her true value v
to a non-negative bid b. We will make the following simple assumptions
about the strategies the bidders are using:
1 s(·) is a strictly increasing, differentiable function; so in particular, if

two bidders have different values, then they will produce different bids.
2 s(v ) ≤ v for all v : bidders can shade their bids down, but they will

never bid above their true values. Notice that since bids are always
non-negative, this also means that s(0) = 0.
These two assumptions permit a wide range of strategies:
The strategy of bidding your true value is represented by s(v ) = v .
The strategy of shading your bid downward to by a factor of c < 1
times your true value is represented by s(v ) = c · v .
More complex strategies such as s(v ) = v 2 are also allowed (although
we will see that in first-price auctions they are not optimal).
Paul G. Spirakis (U. Liverpool) Auctions 65 / 90
 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

Strategies

The two assumptions help us narrow the search for equilibrium strategies:
The assumption of strictly increasing strategies restricts the scope of
possible equilibrium strategies, but it makes the analysis easier while
still allowing us to study the important issues.
Since the two bidders are identical in all ways except the actual value
they draw from the distribution, we will narrow the search for
equilibria in one further way: we will consider the case in which the
two bidders follow the same strategy s(·).

Paul G. Spirakis (U. Liverpool) Auctions 66 / 90

 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

Payoffs

The assumption of strictly increasing strategies says that the bidder

with the higher value will also produce the higher bid.
If bidder i has a value of vi , the probability that this is higher than
the value of i’s competitor in the interval [0, 1] is exactly vi .
Therefore, i will win the auction with probability vi .
If i does win, i receives a payoff of vi − s(vi ).
Putting all this together, we see that i’s expected payoff is

u(vi ) = vi (vi − s(vi )) .

Paul G. Spirakis (U. Liverpool) Auctions 67 / 90

 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

Equilibrium strategies

What does it mean for s(·) to be a (symmetric) equilibrium strategy?

s(·) is an equilibrium strategy if, for each bidder i, there is no incentive for
i to deviate from strategy s(·) if i’s competitor is also using strategy s(·).

It is not immediately clear how to analyze deviations to an arbitrary

strategy.
Fortunately, there is an elegant device that lets us reason about
deviations as follows: rather then actually switching to a different
strategy, bidder i can implement her deviation by keeping the strategy
s() but supplying a different “true value” to it.

Paul G. Spirakis (U. Liverpool) Auctions 68 / 90

 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

The revelation principle

Here is how this works.

If i’s competitor is also using strategy s(·), then i should never
announce a bid above s(1), since i can win with bid s(1) and get a
higher payoff with bid s(1) than with any bid b > s(1).
So in any possible deviation by i, the bid she will actually report will
lie between s(0) = 0 and s(1).
Therefore, for the purposes of the auction, she can simulate her
deviation to an alternate strategy by first pretending that her true
value is vi0 rather than vi , and then applying the existing function s(·)
to vi0 instead of vi .
This is a special case of a much broader idea known as the Revelation
Principle; for our purposes, we can think of it as saying that deviations
in the bidding strategy function can instead be viewed as deviations in
the “true value” that bidder i supplies to her current strategy s(·).
Paul G. Spirakis (U. Liverpool) Auctions 69 / 90
 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

Equilibrium

We can now write the condition that i does not want to deviate from
strategy s(·) as follows:

vi (vi − s(vi )) ≥ v (vi − s(v )) ∀v ∈ [0, 1] .

In order for s(·) to satisfy the above inequality, it must have the
property that for any true value vi , the expected payoff function
u(v ) = v (vi − s(v )) is maximized by setting v = vi .
The first derivative of u(·) is u 0 (v ) = vi − s(v ) − vs 0 (v ).
Therefore, vi should satisfy u 0 (vi ) = 0 or equivalently

s(vi )
s 0 (vi ) = 1 − .
vi

This differential equation is solved by the function s(v ) = v /2.

Paul G. Spirakis (U. Liverpool) Auctions 70 / 90
 Auctions with imperfect information First-price auction with two bidders

First-price auction with two bidders

Equilibrium

Thus, if two bidders know they are competing against each other, and
know that each has a private value drawn uniformly at random from
the interval [0, 1], then it is an equilibrium for each to shade their bid
down by a factor of 2. Bidding half your true value is optimal
behavior if the other bidder is doing this as well.
Unlike the case of the second-price auction with complete
information, we have not identified a dominant strategy, only an
equilibrium. In solving for a bidder’s optimal strategy we used each
bidder’s expectation about her competitor’s bidding strategy. In an
equilibrium, these expectations are correct. But if other bidders for
some reason use non-equilibrium strategies, then any bidder should
optimally respond and potentially also play some other bidding
strategy.

Paul G. Spirakis (U. Liverpool) Auctions 71 / 90

 Auctions with imperfect information Extensions

First-price auction with many bidders

Payoffs

Now suppose that there are n bidders, where n ≥ 2.

Each bidder i draws her true value vi independently and uniformly at
random from the interval [0, 1].
The assumptions on the strategies still imply that the bidder with the
highest true value will produce the highest bid and hence win the
auction.
For a given bidder i with true value vi , what is the probability that
her bid is the highest?
This requires each other bidder to have a value below vi ; since the
values are chosen independently, this event has a probability of vin−1 .
Therefore, bidder i’s expected payoff is

u(vi ) = vin−1 (vi − s(vi )) .

Paul G. Spirakis (U. Liverpool) Auctions 72 / 90

 Auctions with imperfect information Extensions

First-price auction with many bidders

Equilibrium strategies

The condition for s(·) to be an equilibrium strategy remains the same

as it was in the case of two bidders.
Using the Revelation Principle, we view a deviation from the bidding
strategy as supplying a “fake” value v to the function s(·).
Given this, we require that the true value vi produces an expected
payoff at least as high as the payoff from any deviation:

vin−1 (vi − s(vi )) ≥ v n−1 (vi − s(v )) ∀v ∈ [0, 1] .

Paul G. Spirakis (U. Liverpool) Auctions 73 / 90

 Auctions with imperfect information Extensions

First-price auction with many bidders

Equilibrium strategies

We can derive the form of the bidding function s(·) using the
differential-equation approach that worked for two bidders.
The expected payoff function u(v ) = v n−1 (vi − s(v )) must be
maximized by setting v = vi .
Setting the derivative u 0 (vi ) = 0 we get
(n − 1)v n2 vi − (n − 1)v n−2 s(vi ) − vin−1 s 0 (vi ) = 0
or equivalently
 
s(vi )
s 0 (vi ) = (n − 1) 1 − ∀vi ∈ [0, 1] .
vi
This differential equation is solved by the function
 
n−1
s(v ) = v .
n
Paul G. Spirakis (U. Liverpool) Auctions 74 / 90
 Auctions with imperfect information Extensions

First-price auction with many bidders

Equilibrium strategies

So if each bidder shades her bid down by a factor of (n − 1)/n, then

this is optimal behavior given what everyone else is doing.
Notice that when n = 2 this is our two-bidder strategy.
The form of this strategy highlights an important principle in
first-price auctions: as the number of bidders increases, you generally
have to bid more “aggressively”, shading your bid down less, in order
to win.
For the simple case of values drawn independently from the uniform
distribution, our analysis here quantifies exactly how this increased
aggressiveness should depend on the number of bidders n.

Paul G. Spirakis (U. Liverpool) Auctions 75 / 90

 Auctions with imperfect information Extensions

General distributions

In addition to considering larger numbers of bidders, we can also relax

the assumption that bidders’ values are drawn from the uniform
distribution on an interval.
Suppose that each bidder has her value drawn from a probability
distribution over the non-negative real numbers.
We can represent the probability distribution by its cumulative
distribution function F (·): for any x, the value F (x) is the probability
that a number drawn from the distribution is at most x.
We will assume that F is a differentiable function.

Paul G. Spirakis (U. Liverpool) Auctions 76 / 90

 Auctions with imperfect information Extensions

General distributions
Payoffs and equilibrium condition

Most of the earlier analysis continues to hold at a general level:

The probability that a bidder i with true value vi wins the auction is
the probability that no other bidder has a larger value, so it is equal
to F (vi )n−1 .
Therefore, the expected payoff to vi is

F (vi )n−1 (vi − s(vi )) .

Then, the requirement that bidder i does not want to deviate from
this strategy becomes

F (vi )n−1 (vi − s(vi )) ≥ F (v )n−1 (vi − s(v )) ∀v ∈ [0, 1] .

Paul G. Spirakis (U. Liverpool) Auctions 77 / 90

 Auctions with imperfect information Extensions

General distributions
Equilibrium strategies

The equilibrium condition

F (vi )n−1 (vi − s(vi )) ≥ F (v )n−1 (vi − s(v )) ∀v ∈ [0, 1]

can be used to write a differential equation just as before, using the

fact that the function of v should be maximized when v = vi .
The derivative of the cumulative distribution function F (·) is the
probability density function f (·); proceeding by analogy with the
analysis for the uniform distribution, we get the differential equation
 
0 f (vi )vi − f (vi )s(vi )
s (vi ) = (n − 1) .
F (vi )
Finding an explicit solution isn’t possible unless we have an explicit
form for the distribution of values, but it provides a framework for
taking arbitrary distributions and solving for equilibrium strategies.
Paul G. Spirakis (U. Liverpool) Auctions 78 / 90
 Auctions with imperfect information Extensions

Seller revenue

Let us now try to compare the revenue a seller should expect to make in
first-price and second-price auctions.
There are two competing forces at work here:
In a second-price auction, the seller explicitly commits to collecting
less money, since she only charges the second-highest bid.
In a first-price auction, the bidders reduce their bids, which also
reduces what the seller can collect.

Paul G. Spirakis (U. Liverpool) Auctions 79 / 90

 Auctions with imperfect information Extensions

Seller revenue

To understand how these opposing factors trade off against each

other, suppose we have n bidders with values drawn independently
from the uniform distribution on the interval [0, 1].
Since the seller’s revenue will be based on the values of the highest
and second-highest bids, which in turn depend on the highest and
second-highest values, we need to know the expectations of these
quantities.
Computing these expectations is complicated, but the form of the
answer is very simple. The basic statement is:

Suppose n numbers are drawn independently from the uniform distribution

on the interval [0, 1] and then sorted from smallest to largest. The
k
expected value of the number in the kth position on this list is k+1 .

Paul G. Spirakis (U. Liverpool) Auctions 80 / 90

 Auctions with imperfect information Extensions

Seller revenue
Comparison between first- and second-price auctions

If the seller runs a second-price auction, and the bidders follow their
dominant strategies and bid truthfully, the seller’s expected revenue
will be the expectation of the second-highest value.
Since this will be the value in position n − 1 in the sorted order of the
n random values from smallest to largest, the expected value of the
seller’s revenue is
n−1
.
n+1

Paul G. Spirakis (U. Liverpool) Auctions 81 / 90

 Auctions with imperfect information Extensions

Seller revenue
Comparison between first- and second-price auctions

If the seller runs a first-price auction, then in equilibrium we expect

the winning bidder to submit a bid that is (n − 1)/n times her true
value.
Her true value has an expectation of n/(n + 1) (since it is the largest
of n numbers drawn independently from the unit interval), and so the
seller’s expected revenue is
n−1 n n−1
· = .
n n+1 n+1

The two auctions provide exactly the same expected revenue to the seller!

Paul G. Spirakis (U. Liverpool) Auctions 82 / 90

 Auctions with imperfect information Extensions

Revenue equivalence

The fact that the two auctions provide the same expected value to the
seller is a reflection of a much broader and deeper principle known in the
auction literature as revenue equivalence, which, roughly speaking, asserts
that
A seller’s revenue will be the same across a broad class of auctions and
arbitrary independent distributions of bidder values, when bidders follow
equilibrium strategies.

Paul G. Spirakis (U. Liverpool) Auctions 83 / 90

 Auctions with imperfect information Extensions

Reserve prices

So far, we have implicitly assumed that the seller must sell the object. But
how does the seller’s expected revenue change if she has the option of
holding onto the item and choosing not to sell it?
We assume that the seller values the item at u ≥ 0, which is thus the
payoff she gets from keeping the item rather than selling it.
Clearly, if u > 0, then the seller should not use a simple first-price or
second-price auction: in either case, the winning bid might be less
than u, and the seller would not want to sell the object.
Instead, the seller announces a reserve price of r before running the
auction.
The item is sold to the highest bidder if the highest bid is above r ;
otherwise, the item is not sold.

Paul G. Spirakis (U. Liverpool) Auctions 84 / 90

 Auctions with imperfect information Extensions

Reserve prices
Models

Models of auctions with a reserve price:

In a first-price auction with a reserve price, the winning bidder (if
there is one) still pays her bid.
In a second-price auction with a reserve price, the winning bidder (if
there is one) pays the maximum of the second-place bid and the
reserve price r .
As we will see, it is in fact useful for the seller to declare a reserve price
even if her value for the item is u = 0.

Paul G. Spirakis (U. Liverpool) Auctions 85 / 90

 Auctions with imperfect information Extensions

Reserve prices
Second-price auctions

We focus on the the case of a second-price auction with a reserve price.

It is not hard to go back over the argument that truthful bidding is a
dominant strategy in second-price auctions and check that it still
holds in the presence of a reserve price.
Essentially, it is as if the seller were another “simulated” bidder who
always bids r ; and since truthful bidding is optimal regardless of how
other bidders behave, the presence of this additional simulated bidder
has no effect on how any of the real bidders should behave.

Paul G. Spirakis (U. Liverpool) Auctions 86 / 90

 Auctions with imperfect information Extensions

Reserve prices
Second-price auctions

What value should the seller choose for the reserve price?
If the item is worth u to the seller, then clearly she should set r ≥ u.
But in fact the reserve price that maximizes the seller’s expected
revenue is strictly greater than u.
To see why this is true, we will first consider a very simple case:
a second-price auction with a single bidder, whose value is uniformly
distributed on [0, 1], and
a seller whose value for the item is u = 0.

Paul G. Spirakis (U. Liverpool) Auctions 87 / 90

 Auctions with imperfect information Extensions

Reserve prices
Second-price auctions

With only one bidder, the second-price auction with no reserve price
will sell the item to the bidder at a price of 0.
On the other hand, suppose the seller sets a reserve price of r > 0. In
this case:
with probability 1 − r , the bidder’s value is above r , and the object will
be sold to the bidder at a price of r ;
with probability r , the bidder’s value is below r , and so the seller keeps
the item, receiving a payoff of u = 0.
Therefore, the seller’s expected revenue is r (1 − r ), and this is
maximized at r = 1/2.

Paul G. Spirakis (U. Liverpool) Auctions 88 / 90

 Auctions with imperfect information Extensions

Reserve prices
Second-price auctions

If the seller’s value u is greater than zero, then her expected payoff is
r (1 − r ) + ru (since she receives a payoff of u when the item is not
sold), and this is maximized by setting r = (1 + u)/2.
So with a single bidder, the optimal reserve price is halfway between
the value of the object to the seller and the maximum possible bidder
value.
With more intricate analyses, one can similarly determine the optimal
reserve price for a second-price auction with multiple bidders, as well
as for a first-price auction with equilibrium bidding strategies of the
form we derived earlier.

Paul G. Spirakis (U. Liverpool) Auctions 89 / 90

Further reading

Martin J. Osborne: An Introduction to Game Theory. Oxford

University Press, 2004.
David Easley and Jon Kleinberg: Networks, Crowds, and Markets:
Reasoning About a Highly Connected World. Cambridge University
Press, 2010.

Paul G. Spirakis (U. Liverpool) Auctions 90 / 90