24. Mechanism Design
Institutions are designed to alter human behavior. To remain effective over
time, institutions need to adapt to changes in the environment or the society
the institution is meant to regulate.
—Jenna Bednar
In this chapter, we show how to use models to design political and
economic institutions. An institution consists of a means through which
people communicate information as well as a procedure for making
decisions, reallocating resources, or producing outputs based on the
information revealed. In markets, people and firms communicate through
prices to execute trades and make production decisions. In hierarchies,
people communicate through written language to organize work plans. And
in democracies, people communicate preferences through votes. Voting
rules then decide policies. Well-designed institutions induce
communications and actions that produce desirable outcomes. Ineffective
institutions do not.
In this chapter, we present a framework for modeling institutions known
as mechanism design. This framework highlights four aspects of real
institutions: information, what the participants know and should be revealed
to them; incentives, the benefits and costs of taking particular actions;
aggregation, how the individual actions translate into collective outcomes;
and computational costs, the cognitive demand placed on participants.
The origins of mechanism design lie in the analysis of general questions
about the allocation of goods, and in particular whether market mechanisms
or central planning best allocates goods. Early models posited behavioral
rules such as price-taking in a market or voting truthfully. The modeler then

worked through the implications of those behaviors, for example, how they
aggregated. That approach was abandoned in favor of one that assumed
optimizing behavior, making the constructions amenable to game theoretic
reasoning. Mechanism designers then solve for Nash equilibria and
compare institutions based on rational behavior.
The framework has proved useful. It can be used to find flaws in
existing rules and procedures, to explain why institutions succeed or fail,
and to predict outcomes. It has also been used to design a variety of
institutions, including the spectrum auctions described in Chapter 2, as well
as many online markets, governmental voting systems, and even the
procedures that allocate space for projects on space shuttle voyages.1
Our treatment consists of six parts. We first describe the mechanism
design framework using the Mount-Reiter diagram. In the second part, we
study the problem of three people choosing between two alternatives. In the
third part, we analyze three auction mechanisms and find that all yield
identical results. In the fourth part, we show that this was not a coincidence
and describe a foundational result, the revenue equivalence theorem, which
shows that any auction mechanism that satisfies certain assumptions
produces the same outcome. In the fifth part, we compare a majority rule
voting mechanism with a pivot mechanism as ways to decide on whether to
undertake a public project. We conclude by broadening our discussion of
mechanisms along the lines introduced in our criticism of Nash equilibria.

The Mount-Reiter Diagram
A mechanism consists of six parts: an environment (the relevant features of
the world), a set of outcomes, a set of actions (called the message space), a
behavioral rule that people follow to produce actions, an outcome function
that maps the actions into outcomes, and a social choice correspondence
that maps the environment into a set of hoped-for outcomes. The social
choice correspondence commonly consists of either the outcome that
maximizes the sum of the participants’ utilities or of the set of Pareto
efficient allocations. An outcome is Pareto efficient if and only if no other
outcome exists that everyone prefers. Pareto efficiency is a low bar.

Pareto Efficiency
Within a set of outcomes, an outcome is Pareto dominated if there
exists an alternative that everyone prefers. All other outcomes are
Pareto efficient.2

The Mount-Reiter diagram captures these essential parts of a mechanism
graphically (figure 24.1). The diagram juxtaposes what we desire and what
exists. Across the top, the social choice correspondence describes the
outcomes that we normatively desire. Along the bottom, we have the
mangle of reality. People apply their behavioral rules to send messages or
take actions. An outcome function maps those actions into outcomes.
Ideally, the lower, more complicated path on the bottom produces the same
outcome as the top path, that is, the desired outcome.

Figure 24.1: The Mount-Reiter Diagram

Not all mechanisms succeed. For example, if the environment consisted
of people with preferences for a public good, the social choice
correspondence maps their preferences to the optimal level of that good.
However, as we saw in Chapter 23, a voluntary contribution mechanism, in
which people pay for as much of the public good as they desire, results in
each person providing units of the public good rather than the optimal N
units. When the outcome produced by the mechanism does not align with
our objective, we say the mechanism fails to implement the social choice
correspondence.
The list of properties that we would like a mechanism to satisfy varies

by context. We describe five here. First, we would like the equilibrium
outcome of the mechanism to agree with our social choice correspondence
(Pareto efficiency). Second, ideally participants would apply dominant
strategies, that is, their best actions would not depend on the actions of
others. If so, we say that the efficient outcome is dominant strategy
implementable. Third, we would not want to have to force people to
participate in the mechanism (voluntary participation). Fourth, if the
mechanism involves a transfer or payment of resources, we do not want to
have to put in additional money or destroy resources (budget balance).
Later in the chapter when we analyze mechanisms for deciding on a public
project, we see that this may be difficult to satisfy. Last, in many cases, we
desire truth-telling. We would like the messages that people send to reveal
their true information or their true type. Game theorists call this incentive
compatibility. In most cases of interest, no mechanism can satisfy all of
these desiderata. Thus, one contribution of mechanism design has been in
demonstrating what is possible and what is not.

Majority Rule and the Kingmaker Mechanism
The first class of environments that we consider consists of people voting
on a joint action or piece of legislation. We consider three people, whom we
call Uma, Vera, and Will, who want to see a movie together and must
decide between an action movie, a drama, and a comedy. The same
environment would apply to three members of the military deciding
whether to attack their opponent, defend their position, or cede the land. In
either interpretation, the environment consists of three people with
preferences defined over three alternatives. We write preferences using
orderings. The ordering action
comedy
drama corresponds to the
action movie being most preferred, followed by the comedy and then the
drama. We assume the following preference orderings:
Uma: action

comedy

drama

Vera: comedy

drama

action

Will: comedy

drama

action

In this example, we take the social choice correspondence to be the set
of Pareto efficient choices. Given the assumed preferences, the comedy and
the action movie are Pareto efficient. The drama is Pareto dominated by the
comedy.
We first evaluate majority rule as a mechanism. In the case of a tie, we
assume the choice is made randomly. If people vote sincerely, the comedy
receives two votes. However, suppose that Vera and Will both believe that
the other two people will be split between the drama and the action movie
and each votes for the drama. Suppose also that voting is sequential. Vera
votes first and selects the drama. Will votes second and does the same.
Uma’s vote no longer matters, but suppose that to avoid conflict, she also
votes for the drama. The three votes constitute a Nash equilibrium. No
person has any incentive to change his or her vote. In this case, majority
rule does not always implement a Pareto efficient outcome.
We next consider the kingmaker mechanism.3 In this mechanism, one

person is randomly selected to be the kingmaker. The kingmaker then
selects a “king,” who determines the group’s choice. If Will is the
kingmaker, he must pick between Uma and Vera. Whomever he chooses
becomes king, and that person then selects the movie.
If the person selected as king acts rationally, she will select her favorite
movie. Therefore, the outcome will be Pareto efficient. For this reason, the
kingmaker mechanism implements Pareto efficient outcomes. The
mechanism has the added advantage that if any two people have the same
favorite movie, the mechanism selects that outcome. To see the logic, once
again, assume that Will is the kingmaker. If Uma and Vera prefer the same
movie, then that movie will be selected regardless of Will’s choice of king.
If, on the other hand, Will and Uma prefer the same movie, then Will
should pick Uma.

Three Auctions
Now that we have a basic understanding of mechanisms, we turn to the
study of auctions. Most of us have some familiarity with auctions owing to
the prevalence of online marketplaces like eBay. Auctions are used in other
settings as well, including government contracts, used car markets, and
most web advertising. We restrict attention here to a single seller and many
bidders. The object could be a house, a car, tickets to a soccer game, or a
piece of art. We also assume that each bidder assigns a unique value to the
object to rule out ties. The Pareto efficient outcomes are those in which the
object goes to the bidder with the highest value. Any other outcome will be
Pareto dominated by that outcome. We now compare three types of
auctions: ascending-bid, first-price, and second-price auctions.

Ascending-Bid Auctions
In an ascending-bid auction, the auctioneer calls out a price. Any bidder
willing to pay that price raises her hand. The auctioneer raises the price
until only one bidder remains. That bidder then pays the price at which the
second-to-last bidder lowered her hand. In an ascending-bid auction, a
rational bidder remains in the auction until the price reaches her value.
Dropping out before the price reaches her value creates the possibility of
not winning the object at a price at a good price. Remaining in the auction
after the price exceeds the bidder’s value means the bidder could win the
object but pay more than her value, resulting in a net loss.
If all of the bidders act rationally, then the bidder with the highest value
wins the object and pays a price equal to the second-highest bidder’s value.
As an example, suppose that there exist three bidders with values $30, $60,
and $80. When the price called out by the auctioneer exceeds $30, the first
bidder exits the auction. When the price gets to $60, the second bidder
exits. Therefore, the third bidder wins the auction and pays $60.4
In a second-price auction, each bidder submits a sealed bid. None of the
other bidders see the amount. The object goes to the bidder who bids the
largest amount. However, the bidder only pays an amount equal to the
second-highest bid. The construction of the second-price auction makes
telling the truth optimal. Imagine a bidder who values an object at $80
deciding how to bid in a second-price auction. We can assume that the other
bidders have already submitted their bids. The bidder must consider three
possible cases: the highest other bid could be less than $80, equal to $80, or
more than $80. In each case, the bidder does best by reporting her true
value for the object.
The logic becomes clearer when we work through an example. We will
assume that the bidder’s value for the object is $80. We consider four cases
for the highest submitted bids of the others: $70 (lower), $80 (equal), $85
(just above), or $90 (higher). Table 24.1 shows payoffs for five bid values
ranging from $65 to $95.

Table 24.1: Net Payoff as a Function of Various Bids Given a Value of 80
As can be seen from the table, bidding 80 always gives at least as high a
payoff as any other bid. Bidding her true value is always a best action (a
dominant strategy). The same logic applies to all bidders, so all should bid
their true values (the mechanism is incentive compatible). It follows that in
a second-price auction, the bidder with the highest value wins the auction,
and the amount paid equals the second-highest bidder’s value.
In a first-price auction, each participant submits a bid, and the highest
bid wins, with the bidder paying an amount equal to that bid. As in a
second-price auction, the bids are submitted simultaneously, so no one
knows the others’ bids. A participant’s optimal bidding strategy in a firstprice auction depends on the participant’s belief about the values (and
therefore the likely bids) of the other bidders. We will assume that bidders
do not know other bidders’ values but that they do have correct beliefs
about the distribution over those values. To be specific, we assume that the
bidders’ values are uniformly distributed between zero and $100 and that all
of the bidders know this distribution. Bidders also know that all of the other
bidders know this information as well.
Using calculus, we can show that if the values of bids are uniformly
distributed and if all bidders bid optimally, then with two bidders each
should bid half her true value, and with N bidders, each bidder should bid
N−1
N of her value. A person in an auction with nineteen other people should
therefore bid 95% of her true value. Given this bidding rule, the bidder with
the highest value always wins the object. We can also show that the amount
she pays equals the expected value of the second-highest bidder. Thus, the
ascending-bid auction also produces an efficient outcome and the price
corresponds to the expected value of the second-highest bidder.5
Prior to writing down the model, many of us would have had the insight
that the more bidders in the auction, the more a person should bid. Without

the math, we would not have known the equilibrium bidding rule. The
model gives us an exact expression for how much a person should bid. The
amount increases in the bidder’s value, which implies that the bidder with
the highest value will win the auction, just as in the other two auction
formats.

Revenue Equivalence Theorem
In each of the three auction formats, the bidder with the highest value wins.
Therefore, all three mechanisms produce an efficient outcome. In addition,
the expected amount paid by the winning bidder equals the value of the
second-highest bidder. In other words, all three auctions produce the same
expected revenue and allocate the object to the same person. That is
remarkable. Even more remarkable, it can be shown that the winner and
expected revenue are the same for any auction in which bidders act
optimally, the highest bid wins the object, and a bidder with a value of zero
receives no payoff. In other words, all auctions that satisfy those two
conditions produce the same expected outcome, a result known as the
revenue equivalence theorem.6

Revenue Equivalence Theorem
Any auction in which the bidders have independent private values
drawn from a known, common distribution produces the same revenue
to the seller and the same expected payoffs to the buyers if each bidder
makes a bid that maximizes her expected payoff, the bidder with the
highest bid always wins the object, and a bidder who has a value of
zero has an expected payoff of zero.

The revenue equivalence theorem implies that an all-pay auction—in
which every bidder, even the losing bidders, pays the amount of her bid—
produces the same outcome as the second-price auction.7 Even a crazy
design such as a third-price auction, where the highest bid wins and pays an
amount equal to the third-highest bid, produces the same winner and same
revenue. The revenue equivalence theorem does not imply that the auction
rules do not matter. In an actual auction, bidders may not use optimal
strategies, or, in a first-price auction, bidders may have different beliefs
about the value distributions of other bidders. If either condition holds—
non-optimizing bidders or diverse beliefs—then revenue could vary across
the types of auctions. Empirical and experimental tests do show some
differences in how auctions perform.
As would be expected from our discussion of when to expect rationality,
the higher the stakes and the more sophisticated the bidders, the more likely
it is that people act rationally. In online auctions for consumer goods, we
might expect some people to follow rules of thumb or suffer from biases
(such as bidding in increments of $10). In a multimillion-dollar oil lease
auction, bidders probably have access to full information and the requisite
skills.
Also, the type of auction could influence the number of bidders. In
timber auctions, first-price auctions attract more small bidders than do
ascending-bid auctions because small bidders have some chance of winning
if the bigger bidders submit low bids. The small bidders have no chance in

an ascending-bid auction, as the bigger firms can see the smaller firms’ bids
and outbid them.8
Auctions also differ in the cognitive demands they place on participants.
In some auctions, optimal behavior is easy to learn. In an ascending-bid
auction, a bidder should remain in the auction until the price reaches her
value. Other bidders not following optimal strategies could cause a bidder
to have a higher or lower expected payoff, but they do not change the
optimal strategy: a bidder should stay in as long as the price is less than her
value. Similarly, in a second-price auction, a bidder should always follow
the same strategy of bidding her true value. However, figuring out that
truthful bidding is optimal requires multiple steps of logic.
Recall that dominant strategies are optimal regardless of the strategies of
others. Both ascending-bid and second-price auctions have dominant
strategies. First-price auctions do not. In a first-price auction, changes in the
bidding strategy of one bidder can change the optimal strategy for another
bidder. If one bidder always bids either zero or 50, then the other bidder
should always bid either 1 or 51. There would be no reason to bid 60 or 70
as the winner would pay more for the object than necessary. Given the
behavior of the other bidder, whenever a bid of 60 wins the auction, so does
a bid of 51.
Even if an auction has a dominant strategy, not all dominant strategies
are equally easy to deduce. In an ascending-bid auction, the strategy of
staying in as long as the price is less than a bidder’s value requires a single
step of reasoning: if the price is less than your value, buy it at this price. In
the second-price auction, a bidder has to think through multiple
contingencies to see that truthful revelation is optimal. Of course, once
someone bids in several second-price auctions, she should learn that the
optimal bid is to tell the truth.
A last feature to consider is whether the auction encourages non-optimal
behavior. In first- and second-price auctions, bidders submit their bids
without knowing others’ bids. In an ascending-bid auction, bidders can see
the price rise and are aware of who remains in the bidding. This could cause
a bidder to attach some value to winning and to overbid. Auctioneers in
charity auctions try to raise bids by emotional appeals, perhaps by showing
a video of children frolicking on new playground equipment that your bids

will support.
The success of strategies depends on the sophistication of the bidders. It
is difficult to imagine bidders in a timber auction being persuaded to bid
more than their forecasted valuations. It is less difficult to imagine a person
at a charity auction overbidding in light of the cause. Whether or not
bidders change their values during the bidding process is a matter of
conjecture. We need only recognize that it could happen. In first- and
second-price auctions, bidders make a single bid, which allows no
opportunity for emotional appeals during the auction.
Finally, in the first-price auction and the ascending-bid auction, the price
equals the highest bid. In the second-price auction, it equals the secondhighest bid. This leaves the appearance that the seller could have received a
higher price and, in part, explains why governments do not use second-price
auctions. Imagine the headline if a government received three bids for oil
rights, one at $6 million, one at $8 million, and one at $12 million:
“Government Gets $12 Million Bid but Sells Land for $8 Million.” Anyone
who knows auction theory would know that had the government run a firstprice auction or an ascending-bid auction, then the top bid would not have
been $12 million. It would have been $8 million.
As has been highlighted throughout the book, formal models reveal
conditions necessary for a result to hold. The revenue equivalence theorem
does not say that all auction mechanisms produce the same outcome. It
states that all auctions with optimizing bidders in which the highest bidder
wins the object and a bidder with zero value has an expected payoff of zero
are equivalent. A seller could raise more money by relaxing one of those
three assumptions. A seller would have difficulty making people act against
their self-interest, and she would also probably not be able to extract money
from someone who does not value the good. This leaves as the only
possibility not selling the object to the highest bidder. One way to do this is
to not sell the good at all. If the seller knows the distribution of values, she
could set a reserve price, a minimum bid. Under some conditions, this can
increase her expected revenue. Suppose that seller is certain that the three
bidders for an object have values of 5, 10, and 60. Using any of the three
auctions above, the winner bids $60 and pays $10. The seller could earn
higher revenue by imposing a reserve price of $60 and running a first-price
auction.

Mechanisms for Deciding on a Public Project
We next compare two mechanisms for deciding whether to build a public
project such as a school, a new highway, or a sports arena. We assume each
person has an individual value from the project and that the project has a
collective cost.

A Public Project Decision Problem
Let (V1, V2,···, VN) denote the monetary values that N people attach to
a public project with cost C. The project should be undertaken if and
only if

C < V1 + V2 + ··· + VN

We first consider the majority-vote equal sharing mechanism. In this
mechanism, individuals vote whether to undertake the project. If a majority
vote yes, the cost of the project is divided equally among the population.

Majority-Vote Equal Sharing
Individuals vote for or against undertaking the project. If a majority
vote for the project, the project is undertaken and each pays a cost .
As the following example shows, this mechanism can violate
efficiency and voluntary participation.

We know from the spatial voting model that whether or not the project is
undertaken depends on the preferences of the median voter. In this case,
that will be the person with the median value for the public project. By
construction, the mechanism satisfies the budget balance condition and
incentive compatibility. However, the mechanism need not satisfy either
efficiency or voluntary participation as can be seen in an example. Suppose
that there exist three people with values $0, $120, and $150 for a public
project that costs $300. The efficient outcome is that the project should not
be undertaken because the total cost of $300 exceeds the sum of the
individual values. However, given that costs will be split equally, each
person votes on whether to undertake the project at a cost of $100 each. It
follows that two of the three individuals will vote for the project, and it will
be undertaken, the inefficient outcome. Moreover, the individual with value
$0 receives a payoff of -$100, so the example also violates voluntary
participation.
In our second mechanism, the pivot mechanism, each individual submits
a valuation for the project. If the sum of the valuations exceeds the cost of
the project, the project is undertaken. Otherwise it is not. The amount that
the individual will be taxed equals the cost of the project minus the sum of
all of the other individuals’ valuations. If the valuations of the other
individuals exceed the cost of the project, the individual pays nothing.

The Pivot Mechanism
Individual i submits a valuation for a project of cost C. If the sum of
the individual valuations exceeds the cost, then the project is
undertaken.

Individual i pays no tax if
and
otherwise. The mechanism is incentive compatible (
),
efficient, and individually rational. It also implements the efficient
outcome in dominant strategies. As the following example shows, this
mechanism can violate budget balance:
Example: (V1, V2, V3) = (60, 120, 150) and C = 300.
The project should be undertaken given that 300 < 60 + 120 + 150.
Individual 1 pays taxes of 30, the cost minus the sum of the other
valuations (300 − 270); individual 2 pays taxes of 90; and individual 3
pays taxes of 120. The total taxes generated the sum of 240, less than
the cost of the project.

This mechanism satisfies incentive compatibility by a logic similar to
that of a second-price auction. Suppose that the project has a cost of $300
and that an individual values the project at $80. There are three cases to
consider. If the other valuations sum to less than $220, the individual has no
incentive to submit a value more than $80, as he would have to pay that
amount. If, at the other extreme, the sum of the others’ valuations exceeds
$300, then he pays nothing, and he could submit any valuation. If, though,
the sum of the valuations of the others lies between $220 and $300, then if
the individual submits a valuation of $80, he will pay $300 minus that sum
and the project will be undertaken (the efficient outcome). He would not
want to submit a valuation of, say, $70 because the sum of the other values
could be $225 and his low valuation would prevent the project from being

undertaken. Had he submitted a valuation of $80, it would have been
undertaken at a cost to him of only $75.
Given that the pivot mechanism satisfies incentive compatibility, it
follows that it also satisfies efficiency. The project is undertaken only if the
sum of the valuations exceeds the cost. Note that because reporting one’s
true value is a dominant strategy, the efficient outcome is also dominantstrategy implementable. Also, because each individual pays at most her
value for the project, the mechanism satisfies voluntary participation.
However, as shown in the box below, the mechanism need not produce a
balanced budget. In fact, only in rare cases will it do so.
For the problem of deciding on a public project, no mechanism will
satisfy every criterion we might desire. The fact that we can use models to
prove that can save us a lot of time trying to do the impossible. Just as
engineers do not waste time trying to build perpetual motion machines,
mechanism designers do not seek incentive-compatible, individually
rational, efficient, budget-balanced mechanisms for the public project
problem. No such mechanism exists.
The pivot mechanism is about as good as we have, but it fails to satisfy
budget balance. That problem cannot be fixed by raising the amount of
taxes that people pay for the project, as that would make the mechanism no
longer either incentive compatible or individually rational. Individuals
would have an incentive to lie, and some might be asked to contribute more
than their value for the project. One possible workaround is to raise taxes in
some other way and to have a pool of money available for projects. That
would itself create incentive issues, though not as directly. A better solution
is to have some other source of money. For example, a university that both
has a large central endowment and consists of colleges that have separate
endowments could use this mechanism to decide whether to construct a new
student union. Each college dean would have an incentive to truthfully
reveal her value for the union and the university chancellor could make up
for any shortfall. A business composed of subunits that have budget
authority could do the same. A project to switch to a cloud-based system
could be decided upon with the pivot mechanism, and any shortcoming
could be covered by upper management.

Summary
The mechanism design framework enables us to compare mechanisms
across a variety of criteria. Does a mechanism produce efficient outcomes?
Do people tell the truth? Would people voluntarily participate? Does the
mechanism produce a surplus or loss? Using the mechanism design
framework, we can also derive what is possible. It may not be possible to
satisfy all of the desired criteria within the same mechanism. In these cases,
modelers become engineers. We use models to try to construct workable
solutions.
As technology changes, so too can our mechanisms. Take, for example,
the auctions used by internet search sites such as Google. Originally,
Google charged a fixed price per thousand clicks. That mechanism was not
optimal given changes in information technology that allowed Google to
run millions of auctions simultaneously. By using auctions, Google
increased its revenue, and it allocated ad space more efficiently. Google
now uses a generalized second-price auction. Each bidder submits a perclick bid to advertise for a keyword—say mesothelioma, a cancer caused by
exposure to asbestos. The highest bidder receives the first ad slot, the
second-highest bidder receives the second slot, and the third-highest bidder
receives the third slot. The prices they pay are determined as in the secondprice auction.
Suppose that the top four bids are $10, $7, $6, and $3 per click. The
third-highest bidder will pay a price equal to the bid of the fourth-highest
bidder, $3. The second-highest bidder will pay a price equal to the bid of
the third-highest bidder, $6. And the highest bidder will pay $7.9 After
learning the valuations of advertisers, Google could have set a reserve price
and raised even more money. But that outcome would not necessarily hold
if the bidders knew this was Google’s plan. A bidder who thinks he is likely
to be the high bidder would not want Google to know his valuation. Placing
reserve prices would also harm Google’s reputation. A reserve price would
be seen as non-cooperative behavior because Google cannot claim to have a
reserve value for spots on a webpage. The top advertising slot on a keyword
search has little value to Google unless it is sold. That is not true for

someone selling a collection of vintage albums or a used car. Those items
have values to their sellers, so a reserve price is justified. However, Google
values its reputation, and setting a reserve price to extract more revenue
might anger advertisers.
To summarize, mechanism design models can aid in the design of and
choice among institutions. With them, we can deduce what is possible and
not possible to implement. It may not be possible to construct a mechanism
that produces efficient outcomes, induces people to tell the truth, and results
in a balanced budget. If so, we should not waste time and effort trying to
design the impossible. Better that we devote our energies to thinking
through how best to trade off efficiency, truthful revelation, and a balanced
budget
We can also use mechanism design to explore bigger questions, such as
when we should use a market, when we should vote, when we should rely
on a hierarchical mechanism, and when we should turn to a voluntary
collective to allocate a resource or decide on an action.10 Each of these four
institutions—markets, democracies, hierarchies, and collectives—functions
well in some settings and not so well in others. For example, we would not
want to vote on what goods people buy, nor would we want to use a market
to decide on our political leaders.
Within society writ large as well as within an organization, we see each
of these institutional forms. A university confronts a market for professors,
relies on a democracy to hire faculty, assigns course assignments through a
hierarchy, and develops strategic plans using collectives. Nonprofits, forprofits, and government agencies are also a mashup of these various
institutional forms. Using the tools of mechanism design, we can formally
compare how each of these institutions functions, and in doing so, make
better assignments of institutions to tasks.