Foundations of Dominant-Strategy Mechanisms

Author(s): Kim-Sau Chung and J. C. Ely

Source: The Review of Economic Studies , Apr., 2007, Vol. 74, No. 2 (Apr., 2007), pp.
447-476
Published by: Oxford University Press

Stable URL: https://www.jstor.org/stable/4626147

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to
The Review of Economic Studies

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
Review of Economic Studies (2007) 74, 447-476 0034-6527/07/00160447$02.00
@ 2007 The Review of Economic Studies Limited

Foundations of Dominant-Strategy
Mechanisms
KIM-SAU CHUNG and J. C. ELY
Northwestern University

First version received March 2004; final version accepted August 2006 (Eds.)

Robert Wilson criticizes applied game theory's reliance on common-knowledge assumptions. In

reaction to Wilson's critique, the recent literature of mechanism design has adopted the goal of finding
detail-free mechanisms in order to eliminate this reliance. In practice this has meant restricting attention
to simple mechanisms such as dominant-strategy mechanisms. However, there has been little theoretical
foundation for this approach. In particular it is not clear the search for an optimal mechanism that does not
rely on common-knowledge assumption would lead to simpler mechanisms rather than more complicated
ones. This paper tries to fill the void. In the context of an expected revenue maximizing auctioneer, we
investigate some foundations for using simple, dominant-strategy auctions.

1. INTRODUCTION

In the recent literature of mechanism design, there is a research agenda, which is

by the so-called Wilson Doctrine. Roughly speaking, the Wilson Doctrine refers to
articulated in Wilson (1987) that a good theory of mechanism design should not rely
on assumptions of common knowledge:

Game theory has a great advantage in explicitly analyzing the consequences of tradi
rules that presumably are really common knowledge; it is deficient to the extent it assum
other features to be common knowledge, such as one agent's probability assessment abo
another's preferences or information. [... ] I foresee the progress of game theory as
pending on successive reduction in the base of common knowledge required to cond
useful analyses of practical problems. Only by repeated weakening of common knowled
assumptions will the theory approximate reality.

Although there is no clear prescription from Wilson (1987) on how exactly to

dependence on common-knowledge assumptions, the methodology on which the lit
converged is to impose strong solution concepts, which minimize the impact of any s
tion. To understand the intuitive logic behind this methodology, one can consider th
optimal auction design with possibly correlated valuations. The traditional approach
two steps. In step one, the model is closed by adopting the assumption that the dist
valuations is common knowledge. Step two then solves the model by searching for t
Bayesian incentive compatible (BIC) selling mechanism. Step one inadvertently imp
assumptions on bidders' beliefs, which step two then takes literally. The results are
and/or undesirable features of the optimal mechanism. To avoid these perverse resu
giving up step one, one can try to modify step two. In particular, one can replace Bay
tive compatibility with stronger solution concepts that are insensitive to different assu
bidders' beliefs.

This is the approach taken by, for example, Dasgupta and Maskin (2000) and Perry and Reny
(2002) who study the design of efficient auctions in interdependent-value settings. To ensure
that the auction form does not rely on fine details of the bidders' information, they insist that
447

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
448 REVIEW OF ECONOMIC STUDIES

their designs are ex post incentive compatible. Simila

auctions in private-value settings, he also insists that hi
compatible. Both ex post incentive compatibility and d
are stronger solution concepts than Bayesian incentive
This methodology appears, at first glance, misdirec
pendence on simplifying common-knowledge assumpt
relaxing those assumptions directly, stronger solution
der to minimize their effect. In this paper, we shall tr
approach. We focus on private-value auctions and ask
who is unwilling to cling to strong common-knowled
for restricting attention to dominant-strategy mechanism
The usual argument for imposing stronger solution
nisms will then be detail free: the rules would not hav
environment in which it is employed. Indeed, detail fr
Wilson Doctrine. However, a priori, it is not apparent
look as simple as the mechanisms prescribed in the abo
lished intuition in mechanism design suggests that deta
very complicated indeed.
To see why, recall that a mechanism designer can,
that she does not know, and she should do so if the an
consider an auction and an auctioneer who assumes th
about their valuations for the object up for sale. Then
the optimal mechanism depend on the value of p; that
dependence, the auctioneer must construct a more gen
bidders to announce their prior and adjusts outcomes an
The mechanism that results will be free of any detai
that is, their beliefs over their valuations. But this mech
that these beliefs are common knowledge. If the auctio
detail dependence, she should ask more and more ques
to its extreme, a truly detail-free mechanism would b
asking agents to report everything, that is, their whole
Of course, the results that can be obtained from a
the constraint of incentive compatibility. In particular, t
their priors, the mechanism must provide them with a
not be a problem if the auctioneer assumes that this
pose a penalty on the agents if their announcements d
not stake incentive compatibility on such assumption
for truthfully revealing these first-order beliefs will d
order beliefs, a detail-free mechanism is implementabl
announce truthfully his complete belief hierarchy.
It is well known that dominant-strategy mechanism
compatibility; that is, they are detail-free.1 However, d
just one special class of detail-free mechanisms and the
cation (in terms of optimality) for the leap from detai
mechanisms in particular.

1. See Bergemann and Morris (2005) for a modern treatmen

Gerard-Varet (1979).

? 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 449

In this paper, we shall provide a rationale for using dominant-strategy mechanisms,

confronts these problems. Our theory is based on the following often-repeated informa
vation.2 Let v denote the distribution of the bidders' valuations. An auctioneer may h
fidence in her estimate of v, perhaps based on data from similar auctions in the past.
does not have reliable information about the bidders' beliefs (including their beliefs ab
another's valuations, their beliefs about these beliefs, etc.), as these are arguably never ob
She can choose any detail-free mechanism, including those that allow her to ask the bidd
thing about their beliefs that might be relevant. In general, such a mechanism may perfo
under some specific common-knowledge assumptions, but may perform badly if those a
tions turn out to be false. On the other hand, a dominant-strategy mechanism secures
expected revenue, independent of any assumption about the bidders' beliefs. If the aucti
not sufficiently confident in any such assumption, she may optimally choose a dominant-
mechanism.

We call this story the maxmin foundation of dominant-strategy mechanisms, because the
auctioneer chooses among mechanisms according to their worst-case performance. Formally,
the theorem we are seeking is illustrated in Figure 1. In Figure 1, we (heuristically) plot the
performance of arbitrary detail-free mechanisms against different assumptions about bidders'
beliefs. The graph of any dominant-strategy mechanism-and, in particular, the graph of the
best one among all dominant-strategy mechanisms-will be a horizontal line. To establish the
maxmin foundation, we would need to show that the graph of any (potentially very complicated)
mechanism must dip below the graph of the best dominant-strategy mechanism at some point.
Although we believe that Figure 1 captures the intuition of many advocates of dominant-
strategy mechanisms, it turns out to be very difficult to prove in general. With no restriction on
the environment, the set of all detail-free mechanisms is quite rich, and it would be contrary to the
spirit of our investigation to impose exogenous restrictions on the complexity of the mechanism.
Instead, in this paper, we introduce a sufficient condition on the distribution of bidders'
valuations (recall that the auctioneer has confidence in the distribution of bidders' valuations
although not in the distribution of bidders' beliefs). The condition generalizes to the case of an
arbitrary (possibly correlated) v what Myerson (1981) calls "the regular case" in his classical
paper on optimal auctions with independent valuations. It is a familiar condition in mechanism
design and comfortably assumed in many applications.
In fact, under our condition, we are able to prove a stronger result (see Figure 2): there
will be a particular distribution of bidders' beliefs, at which point the graph of every (potentially
very complicated) detail-free mechanism must dip below the graph of the best dominant-strategy
mechanism. We say that this distribution rationalizes dominant-strategy incentive compatibility.
The rationalizing distribution has a simple form. It can be described by a finite type space
in which each possible bidder's valuation is represented by a single type. Moreover, when the
distribution of bidders' valuations, v, converges to a product measure, the bidders' beliefs ap-
proach those that would obtain if v were the common prior. This ties our theorem nicely to the
classical result that there exists a dominant-strategy mechanism that is optimal among Bayesian
mechanisms when valuations are independently distributed.
Clearly Figure 2 implies the maxmin foundation we seek. In addition, Figure 2 is significant
in its own right. To expand on this, let us think about the auctioneer in a different, perhaps more
standard, context.
Imagine the auctioneer as a Bayesian decision-maker. When she needs to choose a mech-
anism, she forms a subjective belief about bidders' beliefs and compares different mechanisms
by calculating the expected performance with respect to that subjective belief. When we observe

2. See, for example, Segal (2003, section VI), who conjectures a result similar to ours.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 450 REVIEW OF ECONOMIC STUDIES

a (potentially very complicated) mechanis

another (potentially very complicated) mech

Cx a the best dominant-strategy mechanism

different assumptions about (distributions of) bidders' beliefs

FIGURE 1

The graph of any mechanism dips below the graph of the best dominant-strategy mechanism at some point

a (potentially very complicated) mechanism

another (potentially very complicated) mechanism

the best dominant-strategy mechanism

an assumption about (or distribution of)

bidders' beliefs at which every other
mechanism performs worse than the
best dominant-strategy mechanism

FIGURE 2

There is a particular point at which the graph of every mechanism dips below the graph of the best
dominant-strategy mechanism

that this auctioneer chooses a dominant-strategy mechanism, we can ask whether such a choice is
consistent with Bayesian rationality, that is, whether such a choice is optimal with respect to some
subjective beliefs. If so, we say that there is a Bayesian foundation for dominant-strategy mech-
anisms. Figure 2 says that, in the regular case, dominant-strategy mechanisms have a Bayesian
foundation.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 451

Note that the existence of a rationalizing belief (Figure 2) is a stronger requ

the maxmin foundation (Figure 1). We mentioned previously that we do not know
maxmin foundation is valid in general (beyond the regular case). However, we do s
ple below that beyond the regular case a Bayesian foundation need not exist. As a
about the rationality of using dominant-strategy mechanisms, we view this as parti
for some distributions of valuations, no Bayesian expected revenue-maximizing au
optimally employ a dominant-strategy mechanism, regardless of her beliefs.
Finally, we relate our results to the widely adopted assumption that all agents
consistent with a common prior. A tempting conjecture is that such consistency i
condition for a rationalizing belief. After all, it is well known that when agents h
tent beliefs there exists a bet, which has positive expected value for all. It would
auctioneer could improve upon any dominant-strategy mechanism by building into
This intuition is incorrect, and the reason uncovers a key insight that is behind
Any bet between the auctioneer and the bidders must be incentive compatible. Eve
are inconsistent, the incentive-compatibility constraint can prevent the auctioneer
from the inconsistency. Indeed, the rationalizing belief we identify in Figure 2 is
consistent with a common prior. The belief is constructed so that incentive compat
any side-bets unprofitable. Because this is a key step in our analysis, following th
our result in Section 3 we present a worked example, which illustrates it.
Nevertheless, the common prior assumption (CPA) is widely adopted in appli
and so it may be of interest to know whether dominant-strategy mechanisms can
by a belief consistent with the CPA. Surprisingly, we show by example in Section 5
not be possible even in the regular case. We do, however, present a positive result t
the valuation distribution is close to being independent. In this case there is a CPA
auctioneer against which dominant strategies are approximately optimal: the assoc
loss relative to the optimal mechanism is vanishingly small.
The rest of the paper is organized as follows. The remainder of this introduct
some important related literature. Section 2 presents the model and formalizes the
main result will be presented and proved in Section 3. Our first example also ap
section. Section 4 interprets our result in terms of a Bayesian foundation and the
example to show that a Bayesian foundation for dominant-strategy mechanisms ne
general. In Section 5, we present our results related to the CPA, and Section 6 then
paper with an observation about the English auction.

1.1. Related literature

This paper is not the first to offer a foundation for dominant-strategy mechanisms. Bergem
and Morris (2005) offer an alternative foundation for ex post incentive-compatible mechanis
which in private-value settings are equivalent to dominant-strategy mechanisms. The main d
ference between Bergemann and Morris (2005) and the present paper concerns the type of me
anisms being considered. Bergemann and Morris (2005) focus exclusively on mechanisms in
which the outcome can depend only on pay-off-relevant data. These mechanisms are natur
suited to study efficient design. On the other hand, we are interested here in revenue maxim
tion for a seller. The optimal mechanism for such a designer will almost always depend not j
on the valuations, but also on pay-off-irrelevant data, such as beliefs and higher-order belie
This is why the results of Bergemann and Morris (2005) do not apply in our setting.

3. For instance, see the auctions depicted below in Figures A. 1-A.6.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 452 REVIEW OF ECONOMIC STUDIES

Neeman (2003) is similar in spirit in that he performs a

glish auction (which is a dominant-strategy mechanism). H
the English auction to the benchmark of full-surplus extr
is called "effectiveness". He shows that the effectiveness
high, and in fact close to 1 for a wide variety of distribu
full-surplus extraction was used despite the fact that this
for the optimally chosen mechanism,4 mainly because det
environment as general as he considers is a daunting task.
is to show how to derive the optimal auction in the worst-
We are thus able to compare dominant-strategy mechanism
and show that the optimal dominant-strategy auction perf
We discuss another connection with Neeman (2003) in foot
case.

Finally Jehiel, Meyer-ter-Vehn, Moldovanu and Zame (2006) study e

patibility in environments with interdependent valuations. They pres
5.1), which suggests that foundations for ex post incentive compatibilit
valuations may be different from dominant strategies in private-value
ample assumes an ad hoc objective for the seller and also assumes that
the object (something that a revenue maximizing seller will typically
comparison with our analysis is unclear. Indeed, Bikhchandani (2006) sh
no externalities, and the auctioneer can retain the good, the limits imp
compatibility are less severe than Jehiel et al. (2006) find for the gen
of the foundations for ex post incentive compatibility in general is an i
direction for further research.

2. PRELIMINARIES

2.1. Notation

If {Xi }N 1 is a collection of sets, then X denotes the Cartesian product xi Xi, or the set of "pro
files" of elements of {Xi}. We write X-i = x j i Xj. If x E X, then xi refers to the i-th coordinat
and we use x-i to denote the element of X-i obtained by removing xi. Likewise, if { fi }iN is
collection of mappings fi : Xi - Yi, then f-i denotes the "product" mapping f-i : X-i - Y
where f-i(x-i) = (fl(xl),..., fi-(xi-1), fi+l(xi+1),...---, fN(xN)). If Y is a measurable set,
then A Y is the set of all probability measures on Y. If Y is a metric space, then we treat it as
measurable space with its Borel a -algebra.

2.2. The auction environment

A single unit of an indivisible object is up for sale. There is a set N of risk-neutral bidders with
privately known valuations competing for the object. Each bidder has M possible valuations and
for notational simplicity, we suppose that the set Vi of possible valuations is the same for each

bidder i and that Vi = {1, 02 .. M}, whereom m-1 = y for each m and some 7 > 0.5 The
bidders' valuations are distributed according to a given probability distribution v e A V. Note that
we are allowing for correlated values and that the independent private-value model is included as
a special case when v is a product measure. We assume that v has full support.

4. In fact, Neeman (2004) showed this.

5. These notational conventions simplify the statements of results and notation, but are entirely innocuous. As-
sumptions of asymmetry in the bidders' valuation sets, or differing gaps between valuations would not affect any of our
results.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 453

A bidder i with valuation vi receives expected utility pi i - ti if Pi is the prob

which he will be awarded the object and if his expected monetary payment is ti. A
of V is v, and a typical element of Vi is v-i.
We consider distributions satisfying a generalized version of Myerson's (198
condition. Let Fi (vi, v-i)= ii v (i i, v-i) denote the cumulative distribution f
valuation conditional on the opponents having valuation profile v-i. The virtual
bidder i at profile v is
1- Fi ()
v (o)

Definition 1. We say that v is regular if the virtual valuations satisfy the single-crossing
condition: for each v, i E {1,..., N}, and j E {0,..., N}, j 4 i,

yi(D) >j(u) yi(Oi,ui) > yj(Oi,OV-i)

for every vLi > vi, where yo0() - 0 denotes the auctioneer's value for the object.

Our definition extends Myerson's (1981) regularity condition to correlated v, but reduces to
his original condition when v is independent. To see this note that if v is independent, then the
virtual valuation of bidder j depends only on vj. Thus, the single-crossing condition reduces to
the requirement that the virtual valuation of each bidder i is increasing.
The regularity condition is stated directly in terms of virtual valuations. A familiar set of
sufficient conditions on v is given below. First, the monotone hazard rate condition is satisfied

if for each i and vi, the hazard rate, hi (^i I _i) = -(,v is an increasing function of
vi. The valuations are affiliated if for each pair of profiles , o', v (v v o')v(v A o') > v(v)v(v'),
where v V' is the component-wise maximum and v A o' the component-wise minimum of the
two valuation vectors.6
We prove the following in Appendix B (see Proposition 3).

Proposition 1. If v satisfies both the monotone hazard rate condition and affiliation, then
v is regular

2.3. Types

To characterize the (equilibrium) behaviour of the bidders who compete in some given auction
mechanism, it is not enough to specify the bidders' possible valuations or even the probability
distribution from which they are drawn. In addition, we must also specify their beliefs about
the valuations of their opponents (called the first-order beliefs), their beliefs about one another's
first-order beliefs (called the second-order beliefs), etc.
The standard approach to modelling the bidders' information is to use a type space. A type
space, denoted Q = (, f, fi, gi)iN is defined by a measurable space of types Qi for each player,
and a pair of measurable mappings fi : i 4-> Vi, defining the valuations of each type, and gi
-i -+ A* _i, defining each type's belief about the types of the other bidders.

6. Affiliation is a strong form of positive correlation. In the worst-case analysis of Neeman (2003), the distribution
of valuations itself was a free variable. He showed that the worst-case distribution of valuations involves negative corre-
lation. It is thus not surprising that we use a condition such as affiliation. Furthermore, our counterexample in Section 4
also involves negative correlation. While the performance measure used in Neeman (2003) is not the same as ours, the
similarity between this aspect of the two results suggests some deeper connection.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 454 REVIEW OF ECONOMIC STUDIES

A type space encodes in a parsimonious way the beliefs

bidders.7 One simple kind of type space is the naive type
tribution v. In the naive type space, each bidder believes
from the distribution v, and this is common knowledge. In
troduced above, this is modelled as follows. For each vi e V
with the property fi (o'i) = vi . The belief gi (woti ) is defi
probability v(. I vi) over V-i is derived from v, then this i
a belief over the other bidders' types, so that the probabili
for the opponents is given by gi (o"i)[-coi] = v(v-i I v i).
associated with valuation distribution v.

The naive type space is used almost without exception in auction theory and mechanism de-
sign. The cost of this parsimonious model is that it implicitly embeds some strong assumptions
about bidders' beliefs, and these assumptions are not innocuous. For example, if the bidders' val-
uations are independent under v, then in the naive type space, the bidders' beliefs are commonly
known. On the other hand, for a generic v, it is common knowledge that there is a one-to-one
correspondence between valuations and beliefs. Myerson (1981) characterizes the optimal auc-
tion in the independent case and Cremer and McLean (1985) in the other case. Which of these
cases holds makes a big difference for the structure and welfare properties of the optimal auc-
tion. These and similar issues have been raised in Morris (2002), Neeman (2004), Bergemann and
Morris (2005), Dekel, Fudenberg and Morris (2006), Heifetz and Neeman (2006) and Weinstein
and Yildiz (2007). The spirit of the Wilson Doctrine is to avoid making such assumptions.
Instead, as explained in the introduction, the common approach is to maintain the naive
type space, but try to diminish its adverse effect by imposing stronger solution concepts. To pro-
vide foundations for this methodology, we have to return to the fundamentals. Formally, weaker
assumptions about bidders' beliefs are captured by larger type spaces. Indeed, we can remove
these assumptions altogether by allowing for every conceivable hierarchy of higher-order beliefs.
By the results of Mertens and Zamir (1985) (hereafter MZ), there exists a universal type space,

* = (*, f* g )iEN with the property that, for every valuation vi and every infinite hierarchy
of beliefs hi, there is a type of player i, coi, with valuation vi and whose hierarchy is hi. Moreover,
each Q* is a compact topological space.9
Another sense in which Q* is "universal" is the following. Certain simple type spaces are
essentially "subspaces" of Q* as captured by the following proposition, which will be used in the
proof of the main theorem.

Lemma 1. Let Q be a type space in which the mapping oi -4 (fi(wi),hi(woi)) is

one-to-one. Then there exist subsets -i c tQ and bijective mappings mi : i -- 4 i such that
1. fi*(mi (Ci)) = fi (Ci)for all coi E Qi,

7. Consider a type woi E i. Its first-order belief is a probability distribution over the valuation profiles of i's
opponents. We can uncover this probability distribution as follows. The probability type coi assigns to a given valuation
profile _-i is gi (ci)({co-i : f-i (o-i) = v-i }); that is, the probability wi assigns to the set of types of the opponents with
valuations v-i. Next, for any profile p_-i of first-order beliefs for i's opponents, let _-i (P-i) be the set of types with
first-order beliefs (as derived previously) P-i. Then the second-order beliefs of woi assign probability gi (oi)[f il (oVi)n
-i (P-i)] to the profile (vi, P-i) of valuation/first-order belief pairs for the opponents. This procedure can be repeated
to compute all higher-order beliefs of each bidder.
8. This terminology originated in Bergemann and Morris (2005).
9. See Appendix A for the details on the MZ construction and how it is applied to our setting. To be precise, the
universal type space includes all hierarchies that satisfy a natural coherency property. Also, in the MZ universal type
space generated by V, there would exist types who are uncertain about their own valuations. Our private-values model
corresponds to the subspace of the universal type space in which it is common knowledge that each bidder knows his
own value. Heifetz and Neeman (2006, section 2.2) call this the private values universal type space and Bergemann and
Morris (2005) the known own-pay-offs universal type space.

? 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 455

2. g!(mi(coi))[m-i (o-i) = gi (coi)[co-i]Ifor all woi E Qi and co-i E Q-i,

where mi (coi) = (ml (cl),..., mi-1 (coi-1), mi+1 (w9i+1),..., mN (CON)).

The lemma shows that simple type spaces in which each possible valuation/belief-
pair is held by exactly one type of each player can be embedded in the universal typ
way that preserves all of the relevant structure.
When we start with the universal type space, we remove any implicit assumption
bidders' beliefs. We can now explicitly model any such assumption as a probability di
over the bidders' universal types. Specifically, an assumption for the auctioneer is a
ji over Q*.
In this paper we will mainly deal with two varieties of type spaces, naive type spa
universal type space. Once the information of the bidders has been specified through
of type space, the seller's problem is to design a selling procedure in order to maximi
We turn to this in the next subsection.

2.4. Mechanisms

An auction mechanism consists of a set Mi of messages for each bidder i, an allocation rule
p : M _ [0, 1]N, and a payment function t : M -> RN. Each bidder will select a message from
his set Mi, and based on the resulting profile of messages m, the object is awarded according to
p(m), and payments are exacted according to t (m). Player i receives the object with probability
pi (m) and pays ti (m) to the seller.
We consider environments in which the seller cannot compel the bidders to participate in
the auction, so we require that each Mi includes the non-participation message Oi. Selecting Oi
is equivalent to "opting-out" of the auction, and so we assume that for any profile m in which
mi = Oi, the allocation and payments rules satisfy pi (m) = 0 and ti (m) = 0. A direct-revelation
mechanism for a given type space Q is one in which Mi = Qi U {Oi).
The auction mechanism defines a game form, which together with the type space constitutes
a game of incomplete information. The mechanism design problem is to fix a solution concept
and search for the auction mechanism that delivers the maximum revenue for the seller in some

outcome consistent with the solution concept. The now-widely adopted approach to implement
the Wilson Doctrine and minimize the role of assumptions built into the naive type space is to
adopt a strong solution concept which does not rely on these assumptions. In our private-value
setting the often-used solution concept for this purpose is dominant-strategy equilibrium.
By the revelation principle, an outcome can be implemented in dominant-strategy equilib-
rium if and only if it is dominant-strategy incentive compatible.

Definition 2. A direct-revelation mechanism F for type space Q is dominant-strategy in-

centive compatible (dsIC) if for each bidder i and type profile wo e 2,

and
Pi (0j) fi (oi ) - ti (w) > 0,

Pi (w)fi(w1) - ti (0) pi (6Oi, cO-i) fi (0i) -ti (di,wo-i),

for any alternative type Oi E E 0-i.

Definition 3. A dominant-strategy mechanism is a dsIC direct-revelation mechanism for

the naive type space KY. We denote by (D the class of all dominant-strategy mechanisms.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
456 REVIEW OF ECONOMIC STUDIES

When the type space is the naive type space, we

compatibility constraints for dsIC depend only on valu
is dsIC with respect to OV if and only if it is dsIC wi
f2V'. So we can always discuss whether an auction mech
without referring to any specific naive type space.
To provide a foundation for this indirect approach t
shall compare it to the direct route of completely elim
about beliefs. We maintain the standard solution conc
enlarge the type space all the way to the universal type s
the set of resulting outcomes is equal to those that arise
mechanisms for the universal type space.

Definition 4. A direct-revelation mechanism F for

for each bidder i and type wOi E Ki,

and
S(Pi (c)fo (i) - ti (o))gi (oi)do-i >0,

(Pi(c)ti(c))
i(co) - tii(o))g(i)d
ci Qc-i > , o-i) fi (woi) - ti (, oj-i)) gi (i)do-i
Ji > (p(pi8i
for any alternative type 6i E -Qi.

A mechanism, which does not rely on implicit assumptions about higher-

should be incentive compatible for all belief hierarchies. In other words, it s
ative to the universal type space.

Definition 5. Let 'P be the class of all BIC direct-revelation mechanism fo

type space. We say that such a mechanism is detail free.

2.5. The auctioneer as a maxmin decision-maker

The given valuation distribution, v, represents the auctioneer's estimate of the bidders' valuations
An assumption p about the distribution valuations and beliefs of the bidders is consistent with
this estimate if the induced marginal distribution on V is v. Let M (v) denote the compact subset
of such assumptions. For any mechanism F, the ui-expected revenue of F is defined as R, (F)
fn,* i ti (co) dl (co).
Unlike the standard formulation of the optimal auction design problem, we do not assume
that the auctioneer has confidence in the naive type space as her model of bidders' beliefs. Rathe
the auctioneer considers other assumptions within the set M (v) as possible as well. An auc-
tioneer who chooses an auction that maximizes the worst-case performance solves the maxmin10
problem of
sup inf R,,(F). (1)
FET jEM(v)

10. Another way to think about this formulation of the problem is to view the auctioneer as uncertainty averse. The
beliefs of the bidders are ambiguous to the auctioneer and this ambiguity is modelled by supposing that the auctioneer
holds all possible priors q.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 457

If the auctioneer uses a dominant-strategy mechanism, then her maximum re

be

iD (v) = sup Rv (F),

FeD
where R, (F) = 1 v (v) Ji ti (v) for any dominant-strategy mechanism F e -.
As we show in the following lemma, any dominant-strategy mechanism can be extended
to a revenue equivalent detail-free mechanism. Thus, the optimal dominant-strategy revenue is a
lower bound for the maxmin value in (1).

Lemma 2. The auctioneer can do no worse than the optimal dominant-strategy mecha-
nism; that is,

sup inf R, (F) > HD(V). (2)

FEW P1EM(v)

Proof Let F = (p, t) be any dominant-strategy mechanism. It induces a m

(p', t') for the universal type space as follows. For all wo e -*, set p'(
t'(ow) = t (f* (w)). In other words, F' is defined over the universal type space,
valuations and not on beliefs. Since for each realized valuation profile, F' pro
come as the dominant-strategy mechanism F, it follows immediately that
for any u E=-M (v),

R, (F') = Et'= Eut of* = It(v)P({cw: f*(wo)= v}) = It(v)v(v) =

veV veV

Thus inf,,,M(,) R,

The maxmin found

can do no better t
equality. We will sh

shall prove that, wh

1DD(v) = sup R,~ (F), (3)
FE'I'

which implies

ID (v) = sup R,* (F) > inf sup Ru (F) > sup inf R. (F),
Fj-TEM(v) F=T FET E=AM(v)

so that (2) holds with equality. For this reason, if p* satisfies (3) then we say that
the use of dominant-strategy mechanisms.

3. THE MAIN RESULT

We can now state the main result.

Theorem 1. If v is regular, then the use of dominant-strategy mechanisms has a m

foundation; that is,
sup inf R,(F)= D(v).
FeT PEAM(v)

The proof of Theorem 1 is in the appendix. Here we shall use a simple example to illus-
trate the ideas behind the proof. Consider an auction with two bidders, each with two possible
valuations. Bidders' valuations are correlated according to the distribution v depicted in Figure 3.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 458 REVIEW OF ECONOMIC STUDIES

vl = 4 vl = 9
v2 = 11 3/10 1/10
v2=5 L 3/10 L 3/10

FIGURE 3

The distribution v of bidders' valuations

vl=4 v = 9
v2 = 11 a = 2, tl = 0, t2 = 11 a = 2,
v2=5 a=0, tl =0, t2=0 a=1,t = 9, t2=0
FIGURE 4

The optimal dominant-strategy mechanism F

bl = 2/5 al = 1/4
a2 = 1/4 3/10 1/10
b2 = 2/5 3/10 3/10

FIGURE 5

Deriving the assumption *

The optimal dominant-strategy mechanism is depicted in Figure 4. In Figure 4, "a = i" is

the shorthand for "allocating the object to bidder i" (i.e. pi = 1 and p-i = 0), and "a = 0" means
no sale.

We first verify that v is regular. Note that the virtual valuation of the high-valuation type
is equal to the valuation itself. Thus, the single-crossing condition will be satisfied provided the
high valuation of bidder i exceeds the low valuation of bidder -i, and this is indeed the case in
our example. Hence, according to Theorem 1 there exists an assumption ,u* consistent with the
distribution v such that equation (3) holds. To illustrate the issues that are involved, we construct
one such assumption below, keeping our exposition informal.
It will suffice to consider assumptions that have a simple form. For each valuation of bid-
der i, there will be exactly one type with that valuation in the support. We write ai (bi) for
the first-order belief held by a high-valuation (low-valuation) type of i that the opponent -i
has high valuation. Figure 5 depicts a probability distribution over the four possible profiles of
valuation/first-order belief pairs.
Figure 5 uniquely defines an assumption u * as follows. We first derive the belief hierarchies
from Figure 5 by induction. For example, for a low-valuation type of bidder 1, the second-order
belief assigns probability 2/5 (3/5) to bidder 2 having high (low) valuation and holding first-
order belief a2 = 1/4 (b2 = 2/5), and a high-valuation (low-valuation) type of bidder 2 has a third-
order belief that assigns probability 3/4 (3/5) to bidder 1 having low valuation and having such
a second-order belief, and so on. Thus, we derive a unique belief hierarchy for each valuation.
The assumption pu* is the measure which attaches the probabilities in Figure 5 to the resulting
four valuation/hierarchy profiles. It is obvious that this assumption up* is consistent with the
distribution v.

Under this assumption pu *, there are at least two potential ways to improve upon the optimal
dominant-strategy auction F in Figure 4. First, according to pu*, conditional on bidder 1 having
low valuation, the conditional probability that bidder 2 has high valuation is 1/2. This is different
from the first-order belief of the low-valuation type of bidder 1, which is bl = 2/5. In other words,
p * is not consistent with a common prior. This suggests the possibility of a mutually acceptable

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 459

bet between the auctioneer and the low value type of bidder I about the realized
2. One possible way to improve upon F is to build this bet into the mechanism.
Second, since the two types of bidder 1 hold different beliefs, another pot
improve upon F is to introduce lotteries in the spirit of the surplus extraction m
Crdmer and McLean (1985). Note that in the dominant-strategy mechanism F,
unsold when both bidders have low valuation resulting in a deadweight-loss of
If the auctioneer were to try to capture some of that surplus by selling the g
dominant-strategy incentive compatibility would require that the high-valuation
1 earn "information rents". On net, the auctioneer finds this unprofitable and t
auctioneer witholds the object when dominant-strategy incentive compatibility is
So the auctioneer may try to improve upon dominant strategies by adding p
depend on the reported valuation of bidder 2. Due to the differences in beliefs of
of bidder 1, such payments can be found that induce self-selection between these
thereby relax the constraint of incentive compatibility.
However, incentive compatibility prevents the auctioneer from profiting from
manoeuvres, as we now show. First, consider a bet between the auctioneer and
the realized valuation of bidder 2. Let x and y be the amount bidder 1 pays the a
event bidder 2 has low and high valuations, respectively. This bet will be accepta
auctioneer and the low-valuation type of bidder 1 only if

(1/2)x + (1/2)y >and0,

(3/5)(-x) + (2/5)(-y) > 0,

with at least one inequality being strict unless x = y = 0. But then the high-valuation type of
bidder 1 would find the bet acceptable as well because

(3/4)(-x) + (1/4)(-y) = (5/2)[(3/5)(-x) + (2/5)(-y)] + (3/2)[(1/2)x + (1/2)y],

which is strictly bigger than the zero rent for the high-valuation type of bidder 1 under F. Thus,
offering the bet to the low type but not the high type would violate (Bayesian) incentive compat-
ibility. And when both types of bidder 1 accept, the bet turns sour for the auctioneer, as

(3/5)x + (2/5)y < 0.

This explains why introducing the first type of modification does not help.
Second, consider introducing a lottery in the style of Cr6mer and McLean (1985) to separate
the high- and low-valuation types of bidder 1. By offering a bet (x, y) depending on the realiza-
tion of bidder 2's type, the seller may be able to relax the downward incentive-compatibility
constraint and sell to the low-valuation type of bidder 1 without leaving extra rent for the high-
valuation type. If such a modification is successful then we must have

(3/5)(4 - x)+ (2/5)(-y) > 0, and

(3/4)(9 - x)+(1/4)(-y) < 0.

The first inequality would be the individual rationality constraint of the low type of bidder 1,
and the second would be the incentive-compatibility constraint of the high type. However, these
together imply that any bet like this cannot be profitable for the auctioneer, as

(1/2)x + (1/2)y = (2/3) [(3/4)(-x) + (1/4)(-y)] - (5/3)[(3/5)(-x) + (2/5)(-y)] < -1.

This explains why introducing the second kind of bet does not help either.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
460 REVIEW OF ECONOMIC STUDIES

More generally, these two types of modifications

there are conceivably a variety of other potential wa
strategy mechanism F. However, in the formal proof
show that in fact there is no modification of F that c

4. BAYESIAN FOUNDATIONS FOR DOMINANT-STRATEGY MECHANISMS

In this section, we shall investigate another possible foundation of dominant-str

nisms, namely, the Bayesian foundation. Imagine the auctioneer as a Bayesian dec
When she needs to choose a mechanism under uncertainty of bidders' beliefs, she f
jective belief in M(v). She evaluates any mechanism according to, instead of i
performance, its average performance with respect to this subjective belief. When
observers observe that this auctioneer chooses a particular mechanism, say a domin
mechanism, we can ask whether such a choice is consistent with Bayesian rationa
whether such a choice is optimal with respect to some subjective belief. If the an
then we say that such a choice is rationalizable. Given the predominant role of Bay
nality in the literature of mechanism design, it seems even more natural to pursue
foundation.

To investigate the possibility of the Bayesian foundation, we only need minim

our setting. We now interpret a probability measure over the universal type space a
by the auctioneer over the belief hierarchies held by the bidders. Such a belief is co
the given distribution of valuations v if it belongs to M(v). A Bayesian foundation f

strategies then exists if there is a belief u ** e M.(v) such that

IID(v) = sup R,*(F);

FET

that is, (3) holds.

It follows from the proof of Theorem 1 that there exists a Bayesian foundation for th
dominant-strategy mechanisms when the distribution of valuations is regular. However
show by example below that beyond the regular case, a Bayesian foundation need not e
a negative result about the rationality of using dominant-strategy mechanism, we view
particularly strong: for some distributions of valuations, no Bayesian-rational auctione
optimally employ a dominant-strategy mechanism, regardless of her subjective belief a
ders' beliefs.
In this example, there are two bidders and each has two possible valuations. The dis
of valuations v is represented in Figure 6.11
The optimal dominant-strategy mechanism is depicted in Figure 7, where we fol
convention from the previous example and use "a = i" as the shorthand for "allocating t
to bidder i".
It is helpful to pay attention to a few noteworthy aspects of this environment and th
dominant-strategy mechanism. Notice that the valuation of bidder 1 is always higher
of bidder 2. Nevertheless, the auctioneer chooses to sell to bidder 2 when bidder 1 h
valuation. This is optimal because conditional on bidder 2 having low valuation, the pro
that bidder 1 has high valuation is greater than 1/2. This means that it is optimal to ex

11. The distribution v in this example does not have full support. This simplifies the exposition of the ex
the conclusion would be the same if the event l01 = 10, 02 = 41 had positive (but small) probability.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 461

v1 = 5 v1 = 10
v2 =4 1/6 0
v2 =2 1/3 1 1/2
FIGURE 6

The distribution v

v = 5 v1 = 10
V2 = 4 a = 2, t1 = 0, t2 = 2 = 1, tl = 10, t2 =
v2 = 2 a = 2, tl = 0, t2 = 2 a = 1, tl = 10, t2

FIGURE 7

The optimal dominant-strategy mechanism F

low-valuation type of bidder 1 to relax the incentive constraint and sell to the high-valuation type
at his reservation price. Given this, the auctioneer may as well sell to bidder 2 when bidder 1
has a low valuation. If monotonicity were not a constraint, the auctioneer would choose to sell to
bidder 1 when bidder 2 had high valuation. Thus, the monotonicity constraint binds here, and in
order to satisfy it, the object is sold to bidder 2 in this case.
The following proposition says that, when bidders' valuations are distributed as in Figure 6,
the dominant-strategy mechanism in Figure 7 can never be optimal regardless of the auctioneer's
belief. It should be obvious from the proof of the proposition that this example is robust.

Proposition 2. For the distribution v depicted in Figure 6, the maximum revenue achiev-
able by any mechanism is uniformly bounded away from the maximum revenue achievable by
dominant-strategy mechanisms regardless of the auctioneer's subjective belief; that is,

inf sup R, (F')> VD(v).

CEM4(v) r'e'

The proof of Proposition 2 is in the appendix. Here we give a verbal sketch of the argu-
ment. There are a few different ways the auctioneer could conceivably improve on the dominant-
strategy mechanism and for any belief of the auctioneer at least one of them will indeed improve.
The outcome of the mechanism could be made to depend on the first-order belief of bidder
2. In particular, the mechanism could ask bidder 2 to report his belief in the probability that I has
a low valuation. Suppose 2 were to report that his own value is low and that 1 is quite likely also
to have a low valuation. In an incentive-compatible mechanism, 2's report can be assumed to be
truthful. But this only means that 2 truthfully believes that 1 is likely to have a low valuation.
What matters is the inference made by the auctioneer about l's valuation conditional on learning
that this is 2's belief. There are two possibilities depending on the auctioneer's belief.
The auctioneer may disagree with bidder 2. But if this is the case, then a mechanism, which
involves a bet between the auctioneer and bidder 2 about 1's valuation, would improve the seller's
revenue. Alternatively, the auctioneer may agree with bidder 2. In that case, conditional on learn-
ing that both bidders have a low valuation, the object should be sold to bidder 1 (who is willing
to pay more) contrary to the outcome of the dominant-strategy mechanism.
Therefore, only if the seller believes that a low-valuation bidder 2 would never believe that
bidder 1 has a low valuation could it be optimal to use a mechanism which, like the optimal
dominant-strategy mechanism does not depend on the beliefs of bidder 2. By a symmetric argu-
ment, only if the seller believes that a high-valuation bidder 2 would never believe that bidder 1
has a high valuation could a dominant-strategy mechanism be optimal.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
462 REVIEW OF ECONOMIC STUDIES

But this means that the auctioneer must believe that th

must have a strong difference in beliefs. In such a situatio
mechanism by including Cramer-McLean separating bets t

5. REMARKS ON THE CPA

The validity, in the regular case, of the maxmin and Bayesian foundations for the use of dom
strategy mechanisms was shown by construction of a particular assumption about bidder
liefs. It is noteworthy that the assumption constructed in the proof of Theorem 1 is incon
with the widely adopted CPA.
Loosely speaking, the CPA says that there is a common probability measure (the com
prior) from which each bidder derives his belief by computing the conditional probabil
opponents' types conditional on his own "signal" or "information". In our current setting,
relate any assumption i to the CPA as follows. For any subset A e = , we shall write i (A
a short hand for jp (A x Q*i). In other words, we abuse notation and use the same notatio
probability measure as well as its marginal distributions.

Definition 6. We say that an assumption u is a CPA-assumption if for any measurab

subsets A c E27 and B C ii,*

I g((oi)(B) pl(doi) = i(A x B).

It is apparent that the particular assumption p/* we used in the proof of Theorem 1
a CPA-assumption. Is it possible to rationalize dominant-strategy mechanisms using only
assumptions? We investigate this possibility in the present section.
The following notation will be convenient. Let Q be a type space which can be embedd
Lemma 1 by some mapping m in K2*, and let p be any common prior over Q. The correspo
prior over Q* is defined by p o m-1. We denote it by m(p). If 0 is the naive type space
abuse notation and use v to denote also the common prior over Qv, and use m (v) to deno
corresponding distribution over W*. If p takes the form of m (p), where p is the common pr
some type space Q embeddable in the universal type space Q*, then p will be a CPA-assum
in the sense of Definition 6.

In Appendix C, we use an example to demonstrate that, when v is far from being a indepen-
dent distribution over V, the answer is negative. To gain some intuition, let's recall the example
in Section 3. The second step of that example argues that it would be too costly for the auction-
eer to introduce a lottery in the style of Cremer and McLean (1985), even though such a lottery
can help separate the two types of bidder 2. As shown by CM, had there been a common prior,
introducing such a lottery would not have been costly at all, and hence our second step would not
have gone through.
In this section, we shall present some positive results for the case when v is close to an
independent distribution. We begin by noting that when v is a product measure (i.e. the play-
ers' valuations are drawn independently), then the regular case reduces to the familiar monotone
hazard rate condition. In this case, we can consider the naive type space Qv with the prior v,
and it has been shown that the optimal BIC mechanism can be implemented in dominant strate-
gies. When we embed Qv in the universal type space Q*, the image of v (i.e. m(v)) will be
a CPA-assumption that rationalizes the use of dominant-strategy mechanisms. We record this
observation as a lemma for ease of reference.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 463

Lemma 3. Let v be regular and independent, and let m(v) be the corresponding dis
bution over the image of Q' in the universal type space. Then m(v) is a CPA-assumptio
suPF~T Rm(v)(F) = rD(v).

When v is close to an independent distribution, but not independent itself, Lemma

dramatically. Indeed, the Crdmer and McLean (1985) mechanism extracts all buyers' su
while the optimal dsIC mechanism must yield some information rent to high-value buyer
m(v) itself cannot be used as a rationalizing CPA-assumption. However, we show that wh
v is regular and close to an independent distribution, there exists a CPA-assumption u
"almost" rationalizes the use of dominant-strategy mechanisms. Precisely, the optimal do
strategy mechanism achieves nearly the same revenue as the optimal detail-free mechanis
assumption u.
To state the general result, we introduce some necessary notation. We say that val
distribution v is e-close to V3 if there exists some i such that v = (1 - e)i3 + e? .

Theorem 2. For any regular and independent Vi and 6 > 0, there exists e > 0 such t
v is e-close to V^, then there exists a CPA-assumption p E (v) such that

HD(v) > sup R (F) - .

FET

Proof We begin by constructing a type space with a common prior. For each bidder i, and
valuation vi there are two types cJ'i and ci', with fi (cV'i) = fi (io"i) = Di. There is a common
prior p over the set Q of type profiles defined as follows:

(1 - e) (lv ,.., o) if t)i -= 6Oi' for each i,

p(o)=- e(vo,..., on) if =)i -= ot)i for each i,
O otherwise.

That is, with probability 1 - e, the valuation

and with the remaining probability from
closed subspaces corresponding to the valu
and f2. We define the belief mappings gi
can be embedded by some mapping m in
i with the same valuation has a distinct h
corresponding distribution over the image
tion and Definition 6 that u is a CPA-ass
M = m(Vi) and i = m(i). Note that I =
Let F be any mechanism in T. We have

R, (F) = (1 - e)R (F) +eR

Hence,

F=T
sup f-cT
Rp (F) < (1 ('cT
- e) sup R1 (F) + e sup R (F)

= ((11 r)D61ueVi3)?+ sup RE i (S),

where the equality follows from Lemma 3.

12. Consider the types belonging to C2. For these types it is common knowledge that values are drawn from i.
A type in Q can have such a hierarchy if and only if the valuation is indeed drawn from i. Thus, the remaining types
(i.e. types in 0) have different hierarchies.

? 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 464 REVIEW OF ECONOMIC STUDIES

From the maximum theorem, for any K > 0 we can c

VD( 0) - K. Thus,

D (v) _> sup R,(F) - e[sup Rp () - IHD()] -K.

Because ji is a CPA-assumption, we have the standard accounting identity: total surplus equals
buyers' expected utility plus R1 (F). Because F e 'P satisfies individual rationality, buyers' sur-
plus is non-negative, so Rf (F) is bounded by the total surplus, which is itself uniformly bounded
by vM = max V < oc. We thus have

vID(i) > sup R (F) - evM - ,

which yields the statement of the theorem when we take x = 6/2 and e < 5/(2vM). II

Theorem 2 shows that when the auctioneer believes that valuations are distributed nearly
independently, then the use of dominant-strategy mechanisms has an approximate maxmin foun-
dation even if we limit ourselves to CPA-assumptions. In this case, any slight loss in revenue
might be compensated for by the other virtues of dominant-strategy mechanisms, for example,
simplicity and transparency of equilibrium play for the bidders.

6. CONCLUSION

We have identified a sufficient condition, a direct generalization of the regular case i

(1981), under which dominant-strategy mechanisms can be rationalized as optimal me
either by appeal to maxmin or Bayesian optimality criteria. Let us conclude by pointi
additional implication of this result. Suppose that in addition to the regularity assum
distribution of valuations v is symmetric, a natural assumption for a seller who does
the identities or characteristics of the bidders. Then the English auction with a suitab
reserve price is an optimal dominant-strategy auction.13 We have thus shown that in s
regular environments, the widespread use of the English auction as a selling mechanis
justified as an optimal response to uncertainty about the bidders' beliefs.

APPENDIX A. UNIVERSAL TYPE SPACE

In this appendix, we review the MZ construction of the universal type space and show how to apply it in o
In general, the set of possible first-order beliefs for bidder i is

1A Vi,

and the set of all possible k-th-order beliefs is

7k -A(Vxi X 7TkTl).

An infinite hierarchy of beliefs for bidder i is a sequence hi = (h , h2,...) satisfying hk E ik

13. Lopomo (2000) proved that the English auction is the optimal dominant-strategy mechanism in (almost) this
setting.
14. It has been recently discovered that the MZ belief hierarchies are limited in a certain sense: some rationalizable
and equilibrium behaviours can arise when modelled using a small type space but cannot be captured in the universal
type space. See, for example, Ely and Pqski (2006) and Dekel, Fudenberg and Morris (2007). To capture all possible
assumptions that are relevant for Bayesian equilibrium behaviour, the universal type space would have to be enlarged.
This issue can arise only in models with a common-value element. They would not alter any of the results in our private-
value setting.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 465

The projections k : Tk _ k-1, defined inductively by q2(h2)(v-i) = h({v_i } x T7i), and for each measurable
subset {v_i } x BC V-i x Tk-2

(hk)({v-i } x B) = hk({-_i } x ($T1)-I(B)),

demonstrate that each k-th-order belief for bidder i implicitly defines beliefs at lower orders as well. A hierarchy is
said to be coherent if these implicitly defined beliefs are consistent with those explicitly defined at lower orders; that is,
Ok(hk) = hk
Recall (see footnote 7) that for any type wi in any type space, it is possible to identify the hierarchy of beliefs of ci .
Let hi (oi) represent this hierarchy.

Lemma 4. There exists a type space K2* such that for each player i, each value vi e Vi and each coherent infinite

hierarchy
compact of beliefs
topological space. hi over V_i, there is a type 0wi E *. such that fi (coi) = vi and hi (oi) = hi. Moreover, each 2*. is a

This lemma is a straightforward application of the results in MZ which we briefly sketch. We take the space of basic
uncertainty (what MZ call the parameter space) to be V. The main theorem in MZ (theorem 2.9) shows the existence of
the "universal belief-space" y generated by V. Because all possible beliefs are included in y, it allows for the possibility
that player i is not certain of which element of vi has been realized. Thus y is too large for our purposes. Instead, MZ's
remark 2.17 derives a compact belief-subspace C in which it is common knowledge that each player i knows his own
value.

Formally, C is a compact space such that there exist for each i, spaces15 Qi where Qi c A(V_i x Q-i) such that

C Vx R1 X ... X 2N

where - denotes homeomorphism.16 Each Qi is a set of possible beliefs for i about the valuations and beliefs of the
others. The hierarchies derived from C will have the property that it is common knowledge that each player knows his
own valuation. Moreover, the MZ construction of C (see remark 2.17 and property 6 in MZ) ensures that every profile of
hierarchies of beliefs satisfying this restriction is represented in C.
To obtain our set-up, we take K2* = Vi x xi, and let fi* : Q* -* Vi andnd gi* :Q" - i be the projection mappings.
Because Vi is finite and Qi is compact, we have that Q* is compact. Our universal type space (corresponding to the

"private values"
Also, Lemma 1 isuniversal
an immediatetype spaceofof
consequence Heifetz
property andMZNeeman
5 from (2006))
when we note is non-redundancy
that the then 2* = (Qcondi-
, fi*, g7)icN"
tion of MZ is satisfied if the mapping wci - (fi (w9i), hi (oti)) is one-to-one.

APPENDIX B

Proof of Theorem I
In this section, we shall first review the properties of the optimal dominant-strategy mechanism design
will be used in the proof of Theorem 1. We use a version of a standard argument to show that the do
incentive-compatibility constraints can be replaced by a monotonicity constraint on the allocation rul
that regularity implies that the monotonicity constraint is not binding in the optimal dominant-strategy m
problem. This sets the stage for the proof of Theorem 1. The latter proceeds by constructing an a
which the optimal BIC mechanism design problem reduces to the same objective function but without t
constraint. It follows that the optimal values in the two problems are the same.
We can formulate the optimal dominant-strategy mechanism design problem as follows:

max 1 v () Zti () (B.1)

p(.),t(.)
uiV i=1

subject to: V i = 1 ..., N, V m, I = 1 ..., M, V

Pi (om,o-i)um --ti (om

15. MZ call these type spaces. Our use of that terminology is th
in mechanism design.
16. The product structure of C is not explicitly noted in MZ, b
and property 6 in MZ.

? 2007 The Review of Economi

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 466 REVIEW OF ECONOMIC STUDIES

Pi (vom,-i)om --ti (om,-i) > Pi(ol,o-i

We omit the proof of the following standard lemma which establi
by a single monotonicity constraint on the allocation rule.

Lemma 5. Say that an allocation rule p is dslC if there exists a tra

(p, t) satisfies the constraints in (4). A necessary and sufficient condi

condition: Vi = 1 ..... N,
Pi (m, Vi)> pi (om-1, -i), V m = 2,.
It follows from standard arguments that in an optimal dominant-

(DICm-_m-1) are binding and (given that p is monotonic) all other c

equalities, we see that when the other bidders report valuation profile

Ui (vol, oi) =0

for type v 1 and

m-1

Uix(m,'m-i) -= Pi(m-l'?-i)(vm - m-1)+Ui(m-1, -i) -=-

mr'=1y Pi(m"-i)

for type om, m > 1. By definition, the total transfer received by the auctioneer is the total surplus generated by any sale
of the object less the rent received by the bidders. Thus, an equivalent formulation of the problem is to choose a dsIC (i.e.
monotonic) allocation rule to maximize the expected value of this difference.

NM m-1
max I m'=1
i=l m-l v_V-i 3 I v(om,
subject to(Mi), i = 1 ..., N

A typical approach to solving a problem

and check whether the solution satisfies th
guaranteed to be the case under the regular

Proposition 3.

1. If v is regular then any solution to th

2. If v satisfies both the monotone hazard

Proof First, consider the relaxed problem

and notice that the derivative of the max

v-i)/v(v). It will be optimal at valuation p
negative virtual valuation, with the object
more bidders tie for the greatest non-negati
For part 1, suppose that the virtual valu
that solves the unconstrained maximiz

i (v)
of > maxj#i
bidder i to i3i > yj
vi. (v). Fixsingle-crossing
By the v such thatcondition,
pi (v) >yi0,(o)(so that yj(v)
> maxj1i yi(v)and_ hence
maxjpi(o)
yj (v)) and shows
= 1. This consider
that an increase in the valuation
(Mi) is satisfied.
For part 2, suppose that both affiliation and the monotone hazard rate condition are satisfied and let v be a valuation

profile at which yi (v) > yj (v). Consider an increase in the valuation of bidder i to 0i > vi. Write 3- = (3^i, v-i). It is well
known that affiliation implies that this "increases" the conditional distribution of other bidders' valuations in the sense of
V(VU6_j) v(O)v U)
the monotone likelihood ratio ordering. That is, for any pair of valuations V' > vj, v( ) (
The new virtual valuation for any bidder k is()
The new virtual valuation for any bidder k is

1 - Fk (A)
Yk(vi, o-i) = kk - Y v(i)

17. For related derivations, see Lopomo (2000) and Segal (2003).

? 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 467

By the monotone hazard rate condition yi () > yi (v). By affiliation, for each bidder j : i,

I -Fj () (() o, -j)

= II)
1)V >0 (

- v( j)
). >0j

vFj(v)
v (V)

and this implies yj (V3) < yj(v). And for the seller (j = 0), the latter inequality holds by definition.
Combining these results we have yi (03) > yj (3). Since j was arbitrary, this proves that the single-crossing condition
holds. I|

We are now in a position to prove Theorem 1. The structure of the proof is as follows. We begin by supposing that
v is regular and satisfies an additional condition, called non-singularity. We show that the maxmin foundation exists for
dominant-strategy mechanisms in this case. Next we show that we can find a sequence of such distributions to approach
any v satisfying the hypotheses of the theorem. We then apply a limiting argument to show that the maxmin foundation
for dominant-strategy mechanisms exists in this case as well.

Given
associated v, writedistribution
de-cumulative vm for function.
the marginal
Let am = probability of valuation
v (v m) be the conditional vi = over
distribution vm,theand writeofGi
valuations (m)= _,m/=v for the
bidders j : i conditional on bidder i having valuation vm. Say that v is non-singular if the collection of vectors {aim m 1
is linearly independent.
Say that a type space is simple if for each player i and valuation oi there is a unique type for i with valuation vi;
that is, the mapping f is one-to-one.18 By Lemma 1, a simple type space can be embedded via a mapping m into the
universal type space. Say that an assumption p is simple if it concentrates on the image in Q* of a simple type space.
In this case, for any mechanism (p, t) defined over Q* we can consider the reduced mechanism (p, t) defined over V,
where

Pi(0) = pi(m(f-l(0))), ti (0) = ti (m(f-l())).

A further notational simplification will be convenient. Letm and denote respectively the vectors (Pii (v1, ))-i V-i
and (ti(v ,))-ieV_-i
Suppose in RMN-
v is non-singular and regular. We begin by constructing a simple type space which will then be embedded
in the universal type space using Lemma 1. Let the set of types for player i be equal to the set of possible valuations; that

is,bidder
of Qi = Vi. We the
i about take fi to
types of be
the the identity,
other bidders. and for notational ease we will write =- gi (um) for the beliefs of type vm
These beliefs are defined as follows:

vi Vm T =Gi(m) m'= .
Thus, conditional on having valuation om, bidder i's belief over opponents' valuations is the average of the auctioneer's

beliefs conditional on i having valuation at least om. 19 Note that the collection {rIzm' m=1 is linearly independent by the
non-singularity of v. The following equivalent recursive definition of rnt is useful:

IM =M
1 1

1IM
Gi (m)
(vmau+Gi(m
Finally, because it is simple, the type space 0 = (=i, fi, gi)iAN can be embedded in the universal type space

by Lemma 1. Let 9i C K* be the image in the universal type space, and write aoi= -mi(v'7). Type wm has valua-
tion fi* (Wcm) = vm and belief g* (w7m) = r (up to the relabelling). We can now define the assumption p * by setting
S*(m(f-1 (v))) = v(U). Clearly ~* ~ MA4(v). We will show thatp * is a rationalizing assumption.

18. The naive type space Qv is one example of simple type spaces.
19. Thus, each bidder type has beliefs which are a distortion of those that would be derived from v, except for the
highest valuation type, where there is "no distortion at the top".

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 468 REVIEW OF ECONOMIC STUDIES

Under the simple assumption u *, the optimal BIC auction desig

max v v(o)ti (v) (B.7)

p(), t() i= iveV

subject to: Vi= 1,..., N, Vm= 1...,

7m i
(PIm mP,
m-ti ) > 0, (IRm)
I

ZI .Pi UmI - I P,
m
We have used the inner product notation such as i
the IR and IC constraints for all types outside of the
Say that an allocation rule p is BIC if there exists a
the constraints in (B.7). Because the beliefs of those
every allocation rule is BIC. Indeed, by exploiting th
rationality constraints can be satisfied by building into
intended type and arbitrarily large negative expected va
and McLean (1985), and we shall omit the details.
While the above argument shows that any allocatio
rule, we can further sharpen the conclusion and argu
without cost to the auctioneer. To begin with, each "u
N-1

Indeed, because bidder i's beliefs are linearly independent, there exists a lottery 2 e RM such that rm ? 2 = 0 for all
m > 1 and rm . 1 < 0 for all m < 1. Since by (B.6) a !is a linear combination of Tr and rT+1, we also have a. 1 = 0.
By adding (some sufficiently large scale of) 2 to tf, each (ICmT>') for m < 1 can be relaxed. No other constraints are
affected and the resulting change in the auctioneer's revenue is U- - 2 = 0.
We next show that for any auction mechanism (p, t) that satisfies the remaining constraints, there exists an auction

mechanism (p', t') which satisfies the constraints (IRm), for m = 1,..., M, and (ICm~nm-1), for m = 2,..., M, with
equality, and achieves at least as high an p*-expected revenue as (p, t) does.
To prove this, fix any auction mechanism (p, t) that satisfies the remaining constraints. Suppose (ICmm-1 holds

with strict inequality. Let r denote the matrix whose M rows are the vectors {rm M =1 and let (r-m m-1) be the
matrix obtained by replacing the m-th row of r with the vector um-1 Note that the matrix (r -m m-l) hasrank M.
We can thus solve the following equation for 2:

(r-m m-1)- xm

where xm denotes the m-th elementary basis vector in RM. Note that because rm-1 -2= 0 < m-1 2, and because

rm-l is a convex combination of m-1 and rm according to (B.6), we have m -A < 0.

We will add the vector eX to t~m-1 for some scalar e > 0. Because ri 2 -= 0 for m' 4 m, no constraints for types
i are affected. As for type wm, the constraint (IRm) is unaffected. The only incentive constraint of type om that is
affected is (IC m-mm-1), and this constraint was slack by assumption. Let Sm > 0 be the slack in (ICm-m-1), and
choose e = -SSm/(rm -2) > 0. Then, with the resulting transfer rule, (ICin- ) holds with equality. Finally, because
em-1 M 1 A> 0, the auctioneer profits from this modification.
We next show that each (IRm ) can be treated as an equality without loss of generality. Define Sm7 = m (tIm m -
t,) > 0 to be the slack in (IRmn). Construct a lottery 2 that satisfies

m . S, m=l,..., M.
i , ...

By the full-rank
will be affected,
profits from this

20. More precise

upper bound for

@ 2007 The Rev

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 469

M M-1

m (a m - ) = (Gi
m=l m=l

= Gi(1)ri1 .iA

= Gi(1)Sl
>0.

The proof for the non-singular case is now concluded as follows. Based on the preceding arguments
the modified programme in which the constraints (I R'1) and (IC7m m-1) are satisfied with equality. W
constraints to substitute out for the transfers in the objective function and reduce the problem to an u
timization with the only choice variable being the allocation rule (recall that any allocation rule is BIC).
objective function will be identical to the objective function (B. 1) for the dsIC problem. Thus the only dif
the two problems is the absence of any monotonicity constraint in the BIC case. It then follows that (i) th
lem and hence the original problem (B.7) will have a solution, and (ii) this solution will be the same as t
the optimal dominant-strategy mechanism design problem by Part 1 of Proposition 3. In particular, eq
and p * rationalizes the use of dominant-strategy mechanisms.
We rewrite the objective function in (B.7) as below, and impose the constraints as equalities:

N M

max vI m I mu - (B.8)
(.),t(.) 1 =lm=1

subject to Vi=l ....N, Vm=l ....M,

S nm -t) = 0, IRT)

rim (nm tm)= rm Pm-1

By definition, aM = M, so (-hm) becomes aM .-M = M a. pM N

m m = I [Gi(m)rI - Gi(m +1)rn=1 ]iim

vmGi

= I Gi(m)vmrm - Gi (m +l)rm+1 -m 3 m+l))m+1 ' m+1.

= [Gi (m)vmzm -mGi(m +1)vm+lrm+l'.3].

In the first line we used the recursive definition in (B.6), in the second line we used (I
-m+ 1
third line we used (IRm+ ).
Substituting the constraints into the objective function, it becomes

N M-1

i=1 m=Pi
v i~l
MvM i"M=l
- p[ V m Gi (m) rim -m -- m+1Gi (m + l)rz+1./ -m
N M-1

V= M vMuM.
i=l m=l + [vrm ((vmur +Gi(m+1)rr+1). i3m vm+l Gi(m + l)rizm+l. i]
N M M

i=1=m=l mrVM
m=2 m" Pt -

@ 2007 The Review

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 470 REVIEW OF ECONOMIC STUDIES

Applying the definition of rm, the objective function becomes

N M M=2m=m
vmVm Pi
i=1 m=1 m=2 m'=m I "m IPi
/M1 -Mm . -m- 1
N M M m

i=
=
m=l
mv1m
m=2m'=2
. p-_

NM m

? M/_2

NM m-1
i=i v (vm,- i)e
i=1m=- ve V-i m1=1

This is identical to the objective function in (B

that equation (3) is satisfied for any regular, non
Now consider an arbitrary regular v, not nece
that each vn is non-singular. Moreover, for vn cl
will be preserved, and hence vn will all be regul
space n exactly as in the first half of the proof
in. Passing to a subsequence if necessary, take r
=i = Vi, fi is the identity, and gi (vm) = rm. By
type space. Let 'On and Q be the corresponding
constructed as in the first part of the proof. In
Let F = (p, t) be any mechanism in '. As befo
respectively, where each w is an element of 0, t
satisfy the constraints in (B.7). From the vectors
the constraints in the Vn-version of (B.7), as foll
We will say that F(n) satisfies BIC(n) if the fol

r (n) -(pim - tirm (n)) > 0, (IRm (n))

rm (n)? (pinm -t(n)) (n) n)? (Pim -

For each i, m, and n, let

Sm (n) = max{0, rm (n) - (in - p' om.v)}

be the amount by which the (IRm (n)) constraint is violated by the transfers t- for type wom (n). Because (5-, t) satisfies
the constraints in (B.7), and Trin(n) -> rm, we have S" (n) -- 0 for each i and m.
Let tn (n) be the sum of the vector tF and the constant vector -Sm (n). Next, for each i, m, 1, and n, let

Lm-1 (n) = max{0, Tr (n) - (povm - m," (n)) - im (n). (pmom - (n))}

be the amount by which (ICT-l1 (n)) is violated by the transfers ii (n). Note that Lm'm (n) = 0. Again, because (/-, t)
satisfies the constraints in (B.7), and because i(n) -? t, we have L'-~l(n) -> 0 for each i, m, and 1. For each n, we

construct 2! (n) to solve the system

r'n(n).1 i(n)= Lm--->l(n), V i,m,1.

We can now define t(n) by setting
ti (n) = tm - Sm (n) + A (n).

By construction, t(n) satisfies BIC(n), and together with p satisfies the constraints in the Vn-version of (B.7).
Because each vn is regular and non-singular, and each p* is an assumption that rationalizes the use of dominant-strategy
mechanisms, the first part of the proof implies that

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 471

Y)n()t-i(n)(v)< HD(vn)
for each n.

Because the constraint set in the optimal dominant-strategy mechanism design problem (B.5) is co
mum theorem implies

n-D(vn) --- ID(v).

Hence,

flD(v)- R , (F) = lim [I-D(vn) - R*,(F)]

n--- 00 lu

lim
i=1
[Evn, i (n) - Evi (n)]
N

nlim [Evn (Ti (n) - i) + Evn ti - Eti]

i=1

= lim[Evn (i (n) - ti)]

i=-1

N M

=Znli-
i=1 m=1oI m (n) - (Sm (n) - ]m (n))
= 0.

The last equality follows because S' (n) --> 0 for every m and rim (n) 1 (n) - 0 for each m and I implies by (B.6) that
a"' (n) - 2m (n) - 0 for each m.
This establishes that * rationalizes the use of dominant-strategy mechanisms for the distribution v and thus con-
cludes the proof.

APPENDIX C. AN EXAMPLE FOR SECTION 5

In Section 5, we claim that there exists a distribution v that satisfies the regularity condition, and such tha
no CPA-assumption u under which equation (3) holds. We shall provide an example of such a distribution here
Consider the same example as in Section 3, where there are two bidders, and each bidder has two possible v
The distribution of valuations is as depicted in Figure 3, and the corresponding optimal dominant-strategy mec
as depicted in Figure 4.
Suppose there exists a CPA-assumption p E MA(v) for which equation (3) holds. We shall prove that there
detail-free mechanism that generates higher su-expected revenue than F does. This would contradict the suppos
equation (3) holds.
It suffices to work only with bidder 2's first-order beliefs in order to complete this proof. So, following the co
in Section 3, we shall continue to use a (b) to denote the first-order belief of a high-valuation (low-valuation)
bidder 2 that bidder I has high valuation. Let b = sup{x e [0, 1]: u (b < x) = 0}.
First, observe that b > 4/9. Suppose, on the contrary, b < 4/9. Then pick any number z between b and
consider the mechanism F (z) as depicted in Figure A. 1.
It is obvious that F(z) is BIC for the universal type space. The only difference between F(z) and F is in the
null) event of b < z, in which case F(z) generates p-expected revenue of 4, whereas F only generates u-expecte
of 9i(vol = 91b < z) < 9z < 9(4/9) = 4, where the first inequality comes from the fact that u is a CPA-assumpti
this would have contradicted the supposition that equation (3) holds, we must have b > 4/9.
Then, consider the mechanism F" as depicted in Figure A.2.
To see that F" is BIC for the universal type space, it suffices to observe that, for low-valuation types of
with b > 4/9, truth-telling gives them a non-negative rent of (5 - 11)(1 - b) + (15/2)b > (-6)(5/9) + (15/2
Since b < 4/9 is a p-null event, F" generates p-expected revenue of 9(4/10) + 11(6/10) - (15/2)(4/10)
whereas F only generates p-expected revenue of 9(3/10) + 11(4/10) = 71/10. This proves that equation (3
hold, a contradiction.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 472 REVIEW OF ECONOMIC STUDIES

vl = 4 vl = 9
a [0, 1] a = 2,tl = 0, t2 = 11 a = 2, t = 0, t2
b > z a= 0,t = O, t2 = 0 a = 1,tl = 9, t2
b < z a = 1,t = 4, t2 = a = 1, t = 4, t2 = 0

FIGURE A.1 I

The mechanism F(z)

v -=4 V1 = 9
a E [0, 1] a = 2, tl = 0, t2 = 11 a = 1, t =
b > 4/9 a = 2, tl = O, t2 = 11 a = 1,tl =
b < 4/9 a = 0, tl = 0, t2 = 0 a = 0, tl =
FIGURE A.2

The mechanism F"

APPENDIX D

Proof of Proposition 2
We shall first prove a weaker version of Proposition 2.

Proposition 4. For the distribution v depicted in Figure 6, the optimal dominant-strategy mechanism F d
in Figure 7 cannot be rationalized by any element in M (v); that is, Vp E M(v),

sup Rp (F') > D(v).

FT'E'

Proposition 2 further strengthens Proposition 4 by asserting that supF,-eW R1, (F') is uniformly bounded away from
VD(v) for all p e -M(v). This second result will be proved after we have proved Proposition 4.
We prove Proposition 4 by way of contradiction. Fix any element p in M (v) that rationalizes the optimal dominant-
strategy mechanism F, we shall prove that there exists a mechanism in 'P that generates higher p-expected revenue than
F does. This would contradict the assumption that p rationalizes F and completes the proof.
The proof proceeds by a sequence of lemmas. In each we derive conditions that must be satisfied by p. Finally we
show that no p can satisfy them all.
For the purpose of this proof, it suffices to work only with bidder 2's first-order beliefs in order to arrive at a
contradiction. So we shall maintain the notational convention used in the example of Section 3 and summarize bidder
2's belief by his first-order belief that bidder 1 has high valuation. The belief of a type with high (resp. low) valuation
is denoted a (b). Now because there may be many types in the support of pu * with the same valuation, we need some
notation to refer to different sets of types. For any (measurable) subset A c [0, 1], we shall use "a e A" to denote the
event that 2 has high valuation and believes with some probability in A that 1 has a high valuation. Likewise "b e B" is
the event that 2 has low valuation and believes with some probability in B that 1 has high valuation.21
The first lemma says that, conditional on any p-non-null subset of low-valuation types of bidder 2, the p-conditional
probability that bidder 1 has high valuation cannot be too low, otherwise the auctioneer can improve upon F by selling to
some low-valuation types of bidder 1.22

Lemma 6. For any x E (0, 1] such that p(b = x) =0, if p(b < x) > O, then p (v = 10b < x) > 3/8.

Proof Suppose there exists x e (0, 1] such that p(b < x) = p/(b < x) > 0, and yet p(vl = 101b < x) < 3/8.
Consider the mechanism F (x) as depicted in Figure A.3.
To see that F(x) is BIC for the universal type space, note that (i) truth-telling continues to be a dominant strategy
of bidder 1, (ii) low-valuation types of bidder 2 always have zero rent regardless of what they announce, and (iii)
high-valuation types of bidder 2 would not announce the (newly added) message "b < x" as that gives them zero rent.

21. Formally, for any type cw2 of bidder 2, if f2 (02) = 4 (i.e. if v2 = 4), a denotes g (2)[(fl*)-1 (10)] and a A
denotes the event {cow : f (2) = 4, g* (2)[(fl*)-1(10)] A}.
22. In Lemma 6 (and similarly in Lemmas 7-9), the seemingly redundant requirement of p (b = x) = 0 is a null-
boundary property used only in the proof of Proposition 2.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 473

vI = 5 v1 = 10
a E [0, 1] a = 2, ti = 0, t2 = 2 a = 1, tl = 10, t2 = 0
b > x a = 2, t1 = O, t2 = 2 a = 1, t1 = 10, t2 = 0
b< x a = 1, tl = 5, t2 = 0 a = 1, tl = 5, t2 = 0

FIGURE A.3

The mechanism F(x)

v1=5 vI = 10
aE [0, 1] a = 2, ti = 0, t2 = 2
b > x a = 2, tl , t2 = 2 a = 1, tl = 10, t 2= 0
b < x a = 0, tl = 0, t2 = -2 a = 1, t1 = 10, t2 = 2(1 - x)/x
FIGURE A.4

The mechanism F'(x)

The only difference between F(x) and F is in the (/u-non-null) event of b < x, in which case F(x) generates p
expected revenue of 5/z(vl = 5ib < x) + 5/u(vl = 101b < x) = 5, whereas F only generates p-expected revenue
2p (vl = 5ib < x) + 10/(vl = 10b < x) < 2(5/8) + 10(3/8) = 5, contradicting the assumption that / rationalizes

The second lemma says that for any low-valuation type of bidder 2 that is possible under /*, the first-order belie
b also cannot be too low, otherwise his belief would be too different from the auctioneer's belief, so much so that th
auctioneer can improve upon F by betting against him.

Lemma 7. uz(b < 3/13) = 0.

Proof Suppose not. Then pick x < 3/13 such that p (b < x) > 0 and u (b = x) = 0,23 and consider the mechanis
F'(x) as depicted in Figure A.4.
To see that F'(x) is BIC for the universal type space, note that (i) truth-telling continues to be a dominant strategy
bidder 1, (ii) low-valuation types of bidder 2 would have strict incentive to announce the (newly added) message "b <
if and only if the resulting rent of 2(1 - b) - [2(1 - x)/x]b = 2(1 - b/x) is positive, or equivalently if and only if b < x
and (iii) high-valuation types of bidder 2 would not announce the (newly added) message "b < x" as that gives th
rent of 2(1 - a) - [2(1 - x)/x]a = 2(1 - a/x), which is lower than the rent of 2(1 - a) if they tell the truth.

The only difference between F'(x) and F is in the (/z-non-null) event of b < x, in which case F'(x) collects fr
bidder 2 an /p-expected amount of

(-2)/u(v o = 51b < x)+ [2(1-x)/x]z(vo1 = 101b < x)

> (-2)(5/8)+[2(1-x)/x](3/8)

= 3/(4x) - 2

> [3/4(3/13)]-2

= 5/4

(where the first inequality follows from Lemma 6), whereas F only collects from bidders 2 an jp-expected amount of
2/ (vl = 51b < x) < 2(5/8) = 5/4, contradicting the assumption that p rationalizes F. II

The third lemma says that the first-order belief a of high-valuation types of bidder 2 cannot be too low. Otherwise
beliefs held by high- and low-valuation types of bidder 2 would be too different, and this would enable the auctioneer to
improve upon F by introducing Cr6mer and McLean (1985) bets to separate these types and relax incentive-compatibility
constraints.

23. It is always possible to pick such an x, as any distribution over [0, 1] can have at most countably many mass
points.

? 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 474 REVIEW OF ECONOMIC STUDIES

vl =5 vl = 10
a < y a = 1, tl = 5, t2 = -2x(1 -
a y a = 2, t = 0, t2= 2 a = 1, t1 = 10, t2 = 0
b < x a = 1,tl = 5, t2 = -2x(1 - y)/(x - y) a = i, tl = 5, t2
b > x a= 2, tl = 0, t2= 2 a = 1,tl = 10, t2 = 0
FIGURE A.5

The mechanism F(x, y)

v1 = 5 vl = 10
a 2 1/12 a=- 2, tl = 0, t2 = 123/61 a = 2, tl
a < 1/12 a = 2, t = 0, t2 = 2 a =, i t =
b [0, 1] a = 2, tl = O, t2 = 2 a = , t = 10, t2 = 0

FIGURE A.6

The mechanism F'

Lemma8. u(a < 1/11) =0.

Proof If not then let y < 1/11 such that /p(a = y) = 0 and tu(a < y) > 0. Notice that y < 1/11 implies y <
3y/(2y + 1) < 3/13, and hence we can also choose x between 3y/(2y + 1) and 3/13 such that p(b = x) = 0. Consider
the mechanism F (x, y) as depicted in Figure A.5.
To see that F (x, y) is BIC for the universal type space, note that (i) truth-telling continues to be a dominant strategy
of bidder 1, (ii) low-valuation types of bidder 2 would have strict incentive to announce the (newly added) message "b <
x" if and only if the resulting rent of [2x(1 - y)/(x - y)](1 -b) - [2(1 - x)(1 - y)/(x - y)]b = 2(1 - y)(x -b)/(x - y)
is positive, or equivalently if and only if b < x, and (iii) high-valuation types of bidder 2 would have strict incentive to
announce the (newly added) message "a < y" if and only if the resulting rent of [2x(1 - y)/(x - y)](l - a) - [2(1 - x)
(1 - y)/(x - y)]a = 2(1 - y)(x - a)/(x - y) is strictly higher than the truth-telling rent of 2(1 - a), or equivalently if
and only if a < y.

Since the event of b < x is a/ -null event by Lemma 7, the only real difference between F (x, y) and F is in the
(pa-non-null) event of a < y, in which case F(x, y) generates p-expected revenue of

5 - 2x(1 - y)/(x - y)

= 5 - 2(x - y + y)(l - y)/(x - y)

= 5 - 2(1 - y) - 2y(l - y)/(x - y)

> 5 - 2(1 - y) - 2y(l - y)(2y + 1)/[3y - y(2y + 1)]

= 5 - 2(1 - y) - 2y(l - y)(2y + 1)/[2y(l - y)]

= 2,

whereas F only generates pt-expected revenue of 2, contradicting the assumption that p rationalizes F. II

Finally, the fourth lemma says that the first-order belief a of high-valuation types of bidder 2 cannot be too high.
Otherwise the beliefs of such types would be too different from the auctioneer's subjective belief, and this would enable
the auctioneer to profit by offering an incentive-compatible and individually rational bet. Obviously Lemmas 8 and 9
deliver the contradiction and thus prove Proposition 4.

Lemma 9. /p(a < 1/11) > 0.

Proof. Suppose y (a < 1/11) = 0. Consider the mechanism F' as depicted in Figure A.6.
To see that F' is BIC for the universal type space, note that (i) truth-telling continues to be a dominant strategy of
bidder 1, (ii) low-valuation types of bidder 2 would not announce the (newly added) message "a > 1/12" as that gives
them strictly negative rent regardless of what bidder 1 announces, and (iii) high-valuation types of bidder 2 would have
weak incentive to announce the (newly added) message "a > 1/12" if and only if the resulting rent of (4 - 123/61)(1 -
a) + (4 - 233/61)a is weakly higher than their original rent of 2(1 - a), or equivalently if and only if a > 1/12.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
 CHUNG & ELY FOUNDATIONS OF DOMINANT-STRATEGY MECHANISMS 475

Since the event a < 1 / 12 < 1/ 11 is a pu -null event by assumption, the only real difference between F
(p-non-null) event of a > 1/12, in which case T' generates p-expected revenue of 123/61 > 2, whereas
p-expected revenue of 2. This proves that p does not rationalize F. 1I

This completes the proof of Proposition 4. We now prove the remaining part of Proposition 2 thro

Lemma 10. Suppose K is a compact topological space and that ~7 is a family of real-valued
such that, for each x e K, there is some fx e F which is continuous at x and satisfies fx (x)
infxeK supf,~ f (z) > 0.

Proof For each x e K, there exists an open neighbourhood Ux such that, for each y E Ux, w
fx (x)/2. The collection {Ux : x e K} forms an open covering of the compact space K, and hence th
sub-covering. Let {Ux, ..., Uxn } be a finite sub-covering and let e = min{fxl (xl), ... fxk (Xn)} >
we have x E Uxl for some 1 = ,..., n so that supf3 f(x) fx (x) > fxl (xt)/2 > E/2 > 0. II

Lemma 11. Suppose 1, - - -... On are disjoint open subsets of K* such that p(U Co) = 1, an
bounded real function that is constant on each 01. Then the mapping

p -> J,t p'(dt)

is continuous at the point p.

Proof Fix any e > 0. Let t > 0 be an upper bound for Itl. The function pI' - p'(Oi) is lower semi-continuou
(see Aliprantis and Border, 1999), hence we can set

tn2

and find a neighbourhood U of p such that, for all p' e U, p'(Ol) > pd(01) - for 1 = 1 ..., n. Since p (UO1) = 1, it
follows that p'(O1) < p (01) + (n - 1)6 and p '(K* \ UOl) < p (Q* \ UO) + n6 = n6.
We can write
n

tdp' =p'/=1
(01)JQ
t (01) +*\(w) dp',
*\U01
so that

Cii n~(i)/~'uO)lnr i'd~

1=1 1=1

n n

/=1 1=1

S [p(O) --]t(01) -ndT < tp'(dw) < [p (01)+ (n - 1)6]t(01)+n5t

= - t (01)

-? -2ndT <
Jo * fJl*

This proves that fI.t tp'(dw) - fj. tp (dw) < max f2nOd, n26i C = c.

Proof of Proposition 2. Notice that, for each of the mechanisms used in the proof of Proposition 4, the total
transfer (tl + t2)(t) satisfies the conditions of Lemma 11. For example, consider the mechanism F(x) in Lemma 6. For
any (0l, 02), the set of universal type profiles in which the valuation pair is (vl, 02) is open in the product topology
with p-null boundary. Moreover, since p (b = x) = 0, the event b < x is also open in the product topology with p-null
boundary. Therefore, we can take 1, ..., 06 to be the interiors of the sets represented by the cells of the table in Figure
A.3. These open sets are disjoint, have p-null boundaries, and have total p-measure equal to 1 as required.

@ 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms
476 REVIEW OF ECONOMIC STUDIES

Thus, for any p E M(v), there exists a mechanism F(p) e T su

S R1P ,,(FT()) - liD(v) is continuous at the point p' = p. We c

S = {R(.)(F) -rD(v):FE-TJ. II
Acknowledgements. Thanks to Larry Epstein, Stephen Morris
also thank the Editor and two anonymous referees for helpful suggest
Science Foundation under grant #SES 99-85462 to author Jeffrey C

REFERENCES

ALIPRANTIS, C. D. and BORDER, K. C. (1999) Infinite Dimensional Analysis: A Hitchhiker's Gui

Springer-Verlag).
BERGEMANN, D. and MORRIS, S. (2005), "Robust Mechanism Design", Econometrica, 73, 1521-153
BIKHCHANDANI, S. (2006), "Ex Post Implementation in Environments with Private Goods", Theoreti
11 (3), 369-393.
CREMER, J. and MCLEAN, R. P. (1985), "Optimal Selling Strategies Under Uncertainty for a Di
Monopolist when Demands are Interdependent", Econometrica, 53, 345-362.
DASGUPTA, P. and MASKIN, E. (2000), "Efficient Auctions", Quarterly Journal of Economics, 115 (2
D'ASPREMONT, C. and GERARD-VARET, L.-A. (1979), "Incentives and Incomplete Information", Jou
Economics, 11, 25-45.
DEKEL, E., FUDENBERG, D. and MORRIS, S. (2006), "Topologies on Types", Theoretical Economics, 1
DEKEL, E., FUDENBERG, D. and MORRIS, S. (2007), "Interim Correlated Rationalizability", Theoretic
(forthcoming).
ELY, J. C. and PJSKI, M. (2006), "Hierarchies of Belief and Interim Rationalizability", Theoretical Ec
19-65.
HEIFETZ, A. and NEEMAN, Z. (2006), "On the Generic (Im)possibility of Full-Surplus Extraction in Mechanism
Design", Econometrica, 74, 213-233.
JEHIEL, P., MEYER-TER-VEHN, M., MOLDOVANU, B. and ZAME, W. R. (2006), "The Limits of Ex-Post
Implementation", Econometrica, 74 (3), 585-610.
LOPOMO, G. (2000), "Optimality and Robustness of the English Auction", Games and Economic Behavior, 36,
219-240.
MERTENS, J. F and ZAMIR, S. (1985), "Formulation of Bayesian Analysis for Games with Incomplete Information",
International Journal of Game Theory, 14, 1-29.
MORRIS, S. (2002), "Typical Types" (Mimeo, Princeton University, http://www.econ.yale.edu/sm326/typical.pdf).
MYERSON, R. B. (1981), "Optimal Auction Design", Mathematics of Operations Research, 6, 58-73.
NEEMAN, Z. (2003), "The Effectiveness of English Auctions", Games and Economic Behavior, 43, 214-238.
NEEMAN, Z. (2004), "The Relevance of Private Information in Mechanism Design", Journal of Economic Theory, 117,
55-77.
PERRY, M. and RENY, P. J. (2002), "An Efficient Auction", Econometrica, 70 (3), 1199-1212.
SEGAL, I. (2003), "Optimal Pricing Mechanisms with Unknown Demand", American Economic Review, 93 (3
509-529.
WEINSTEIN, J. and YILDIZ, M. (2007), "Finite-Order Implications of Any Equilibrium", Econometrica (forthcom
WILSON, R. (1987), "Game-Theoretic Approaches to Trading Processes", in T. Bewley (ed.) Advances in Econom
Theory: Fifth World Congress, Ch. 2 (Cambridge: Cambridge University Press) 33-77.

? 2007 The Review of Economic Studies Limited

This content downloaded from

150.84.238.59 on Sat, 25 May 2024 04:36:07 +00:00
All use subject to https://about.jstor.org/terms