KELLOGG SCHOOL OF MANAGEMENT

Strategic Decision-Making

DECS-452 Professor Bob Weber

Week #6

1. Please carry out the Riverside / DEC negotiation exercise. Submit your results using the form on
the class webpage no later than 11 PM Thursday (day section) / 11 PM Wednesday (evening
section).

2. Read the Managing Negotiations handout (Chapters 6 and 8) and the attached material.

3. (Carryover from Week 4: This time for sure!) Write up your strategy for problem (3) in the
“Common Knowledge” notes. On Tuesday, May 10 (day section) / Thursday, May 12 (evening
section), I will collect the strategies, shuffle them, and then hand them out again, so be sure that
yours is written up clearly enough for a classmate to be able to carry it out.

4. If you’d like to try a few informal exercises:

(a) Analyze the “dissolving a partnership” example (in the attached section of the “Negotiation and
Arbitration” readings) using the approach we took to the first-week bidding experiment

(b) Show that, for either of the “mechanisms” illustrated by the two figures (in the attached section
of the “Negotiation and Arbitration” readings), the best you can do is to truthfully report your own
valuation of the item up for trade.

5. On Friday, May 13 (day section) / Thursday, May 12 (evening section), I’ll distribute the final
project materials. You’ll have three weeks to work on them.
 The “Salty Dog”: Distributive Bargaining

Buyer and Seller Types

$1000 weak strong $4000

(Seller) [-------------|--------------|---------------]
[---------------|--------------|-------------] (Buyer)
$3000 strong weak $6000

seller
Pairings $1,900 $2,900
weak strong
$4,100 strong 3,7,11,… 1,5,9,…
buyer
$5,100 weak 4,8,12,… 2,6,10,…
 The Salty Dog: Summary and Notes

Summary of Final Agreements, 1987-2016:

$1900 $2900
$4100 $3,330 $3,502 strong buyer
$5100 $3,468 $3,723 weak buyer
weak seller strong seller

Likelihood of Agreement:

$1900 $2900
$4100 98.1% 97.6% strong buyer
$5100 98.1% 97.2% weak buyer
weak seller strong seller

In the role-play, each side is given no guarantee that mutual gain is available, although in fact such
gain is available to all pairs. On each side, a party can be in either a strong position (a relatively
attractive alternative to a negotiated agreement is available), or a weak position (the alternative is
relatively unattractive); hence there are actually four different types of pairings: strong-strong (the
narrowest range of mutually-advantageous agreements), strong-weak, weak-strong, and weak-weak
(the widest range). The strong and weak positions are set so that every party knows (privately) that
a mutually-advantageous agreement exists, but doesn't know whether the other party also knows
this.

Note:

(1) Just about every pair reaches agreement.

(2) The actual agreements vary by pair, but, on average, strong parties do better (i.e., reach
agreements closer to their ends of the distributive axis) than weak parties against either
strong or weak partners.

On the surface, (1) and (2) seem reasonable. But (1) and (2) together are inconsistent with
effective negotiating behavior! If every pair “should” reach agreement, weak parties should act
as if they are in strong positions. They will still reach agreements, and the agreements will be
more favorable to them: They should do as well as the strong parties. Alternatively, if strong
parties “should” do better, it must be because weak parties can't afford to mimic their behavior:
That “strong” behavior must involve taking risks that weak parties can't justify, in which case,
strong parties negotiating with other strong parties must sometimes fail to reach agreement.
Strategic Decision-Making (DECS-452 Course Outline)

The Salty Dog case: Distributive bargaining

1. Opening offers

a) If you move first

1) Too aggressive: Antagonistic
2) Too generous: Costly - signals either weakness or softness
3) Just right (as tough as possible without being ridiculous): Fine - gives
mixed signal, either strong or tough, without revealing information; forces
response. Can anchor further negotiations in desirable region, and
increase opposing concern about reaching any agreement.
4) Fuzzy: Difficult to manage, but can provide a way to force opponent to
make first “real” offer
5) Moral - If you have good information about your competitor's best
alternative, start with a tough offer. If you know more about him than he
thinks you do, start below his best alternative and let him drag you up.
b) If he moves first
1) With luck, he'll be too generous at the start
2) Perhaps he'll betray a useful piece of information
3) If he comes in tough, resist anchoring: Respond toughly.
4) Moral - If you are not well informed, try to avoid the first move. If you are
well-informed, but think he may be soft and feel you can undo his
anchoring, make him move first.

2. The middle game

a) Don't let the desire to reach agreement overshadow your search for a good
agreement, i.e., don't become more risk-averse than you truly are. A useful
approach is to formulate an aspiration level for yourself, and to update this as
information comes forth.
b) Recognize the time already spent in discussion as a sunk cost.
c) Carefully listen to him, and continually update your perception of his likely best
alternative.
d) Watch for impatience (usually a sign of weakness), and don't appear impatient
yourself.
e) Moral – Be aware of your own psychological state, as well as his.

3. The close

a) Lock him into a commitment to agreement.

b) Congratulate him on his toughness, and both of you on your mutual success.
c) If it seems useful, pull out the “quivering quill”.
d) If you were prepared to walk, and he calls your threat, walk!
 His ideas were thought reactionary at the time, but Lemuel Boulware has lived to see many of them accepted.

A look back at “Boulwarism”

A chat with Lemuel R. Boulware
by Peter Brimelow (Forbes - May 29, 1989, pp. 246-248)

As part of his long-ago effort to cope with the power of trade unions, Lemuel R. Boulware in the late 1950s hired a movie star
called Ronald Reagan, who bolstered the viewership of General Electric's TV show. In carrying out his assignment, Reagan
visited GE plants where his political views were deeply influenced by the discovery that “we didn't chain the workers to the
machines.” Ronald Reagan went on to use his convictions and his powers of persuasion to help restore America's faith in the free
enterprise system.

Now 94 and retired for nearly 30 years, Boulware is restricted by recent illness to a wheelchair in his Delray Beach, Fla.
oceanfront home. But this spring he talked at length with Forbes about many things, including the past and the future of labor
relations in the U.S.

In the 1950s “Boulwarism” was a household word. It was coined to describe Lemuel Boulware's seemingly rigid style in opening
wage negotiations as head of employee relations for General Electric. He would not bargain, labor leaders complained. He would
simply make whatever offer he had determined was in “the balanced best interest” of company, work force and consumer — and
thereafter refuse to budge. Boulwarism was widely denounced as arrogant paternalism. But in fact it was rather more subtle. As
Boulware pointed out in his book The Truth About Boulwarism, written after he retired in 1961, it could work only where the
company was offering a pay scale that was acceptable by prevailing standards — otherwise resistance would be too intense.

Because Boulware recognized the need to pay competitively, his wage offer was never simply a stonewall refusal to consider an
increase. And, contrary to legend, he says that he always undertook to negotiate further “on getting any old or new information
proving that change would be in the balanced best interest of all.” He reports that in the 14 years he handled labor negotiations
for General Electric, only one of his initial offers was actually accepted in the end without amendment.

But from the labor union leaders' point of view, Boulwarism had a major disadvantage: It made it difficult for them to go to their
members and claim all the credit for the settlement. They went all out to discredit his ideas and succeeded in making Boulwarism
a dirty word to many people.

Because of this essentially political need on the part of the labor leaders, many negotiations had tended in his time to become
“amateur theatricals,” as Boulware puts it. “For instance, if everything pointed to a five-cent increase being about right, there was
a strong tendency among employers in those days to offer nothing at first. Then, under public strike threat pressure, about half
would be offered. Then, after all the union representatives had been called in from the plants and the resulting vote for a strike
had been well-aired in a receptive press, management would 'capitulate' by upping the offer to the full five cents an hour.”

He recollects, “Time and again I was told in private — and even occasionally before mediators and 20 or 30 others at the
bargaining table — that there was nothing wrong with the offer except it was ours and not theirs, and that they had to justify
themselves with their members by showing they could force something more out of US.”

This was exactly what Boulware wanted to prevent. He explains: “We all had 'quite a shock when, at plant after plant in 1946
[when General Electric was idled by a seven-week strike] union officials had proved they had a 'push-button' control over
employees and could cause them not only to strike but also, in too many cases, to do senseless and frightful damage to the
investment in their places of work.”

Assigned to labor relations after the strike, Boulware concluded that union propaganda had persuaded General Electric's
employees that their welfare did not depend on the economic health of their company but on whatever muscle they could apply to
their employers. Managements that allowed themselves to be cast as villains in any such union—scripted charades would merely
increase the labor leaders' power — particularly if, as with General Electric when Boulware took over, the unions were the only
side that would talk to the press. The end result: major corporations becoming “the slaves of unions” — which is exactly what
happened, says Boulware, in the American automobile industry.

Depriving unions of credit for pay increases that were really the result of market forces was only part of Boulware's
counteroffensive. He also launched a company-wide “corrective education” program. A regular bombardment of short, punchy
commentaries and cartoons on elementary economics (see below) began in employee newspapers, and were even included with
mailings to shareholders and distributed to community leaders. (“The clergymen were the worst,” says Boulware. “They were
always against us.”)

In an uncanny foreshadowing of T-groups and Japanese corporate communalism, General Electric supervisors held discussion
groups for employees in their homes, focusing on a specially prepared economics primer that dealt with such intriguing questions
as “Where does the government get its money?” Significantly, Boulware says, the most interested participants were the wives.

Boulware's broadsides may have been simple to read, but they were in fact subtly sophisticated. For example, he never blamed
unions for inflation — high wages could force companies out of business and kill jobs, but an increase in the overall price level
would result only if the government tried to counter this effect by printing money. Thus wage and price controls were irrelevant
to inflation — a point not understood even by subsequent Republican Presidents.

General Electric had no major strikes during Boulware's tenure. But it has been argued that his policy as executed by his
successors eventually resulted in a protracted strike in 1969, after which it was abandoned. Boulware, a loyal company man, says
only: “They thought the job was done.”

Boulware in 1950 and illustration from G.E. employee publication

More than capital and labor were at the bargaining table.

Much of Boulware's experience is now history. Even heavy industrial unionism is a shadow of its former self, with General
Electric's work force sharply deunionized.

What does he think of the current scene? Boulware likes George Bush but says he's surrounded by “go-along-to-get-along types.”
Boulware still thinks that the answer to the nation's problems is civic virtue, which can be injected only by the efforts of the
minority of enlightened businessmen. Alone in his big house since the recent death of his wife, attended by a servant couple and
encroached upon by luxury condominiums, he still receives a procession of emissaries from struggling free market think tanks
and magazines and dispenses advice and money to them, an investment in his country's future designed to pay off slowly and in
the long term.

Of the future for free enterprise, he says: “You have to believe.” He knows that the victory isn't won, but he has lived long
enough to see labor and capital retreat considerably from their formerly adversarial positions and to have watched as Ronald
Reagan went on to apply many of Boulware's basic ideas to the national scene.
 Fish or Shark?
The Ecology of Strategic Behavior

The difficulty in teaching a course on strategic behavior is in distinguishing between what lessons I would
teach if everyone in the world would follow my teachings religiously, what I could teach if everyone would
listen to me but felt free to choose whether to follow my teachings or not, and what I feel I should teach to a
group of managers who will subsequently be dealing with others who haven't heard me.

The finitely-repeated Prisoners' Dilemma illustrates my quandary. If everyone were required to obey my
teachings, the solution would be simple: I would tell you to always “cooperate” (even in the one-stage
game). The world would be a better place for us all. (The problem, of course, is that I can't force even one
person to follow my advice.) If instead, everyone in the world were to listen to me but each had the freedom
subsequently to decide for himself whether or not to follow my teachings, the only thing I could teach is to
always “defect.” If I taught anything else, and you (an individual) thought that everyone else would follow
my teachings, you would find yourself preferring to disobey me. But in the real world, “Tit-for-Tat” makes
sense, and (subject to minor modifications) is what I believe I should teach. As long as there is some
chance that you are facing someone who will cooperate to the bitter end (or as long as you think that they
think you may be such a person, or ...), it is in your interest to cooperate through the early stages, defecting
only very near the end.

More picturesquely, think of those who blindly cooperate as “fish,” and those who act totally aggressively
as “sharks.” In an ocean without sharks (i.e., in a naively Utopian society), the fish thrive. In an ocean of
sharks alone, everyone starves (at least, in relative terms). God forbid that my goal in life be to make
everyone a shark! Instead, my goal is to teach you to thrive in an ocean containing both fish and sharks (i.e.,
in the “real ocean”), by being a cross between a spiny fish and a dull-toothed shark. Tit-for-Tat starts off
fishy, but bites back.

I have no qualms concerning the message that came from the bidding experiment. In most economic
situations, people do worry too much about getting something, and not enough about how much they get.
(Similarly, in emotionally-charged situations with more than one possible resolution, people frequently
worry too much about the precise form of what they get — Fisher and Ury, in Getting to Yes, properly
emphasize the importance of focusing on issues, rather than on positions.) Optimizing against your best
model of opposing behavior (i.e., being sharkish) is, in many cases, clearly the best way to deal with general
economic confrontations.

But perhaps there are exceptions. If you are negotiating with someone who is obviously a fish, and indeed,
such a fish that you could swallow him whole, should you do so? Society at times says, “No!” (For
example, a divorce judge is supposed to refuse to ratify a mutually-accepted settlement that he deems clearly
inequitable, i.e., unconscionable. The legal presumption is that such a contract was agreed to under duress.)
The line is certainly drawn somewhere short of “conning” the fish.

In a world of pure fish, disputes would be easy to resolve. Have all parties honestly reveal all of their
concerns, all of their alternative opportunities, all of their preferences over tradeoffs between issues (in
multi-issue disputes), and then determine the agreement which yields the greatest aggregate gain and divide
that gain equally among the parties. (The question of interpersonal utility comparisons — “This means
more to me than it does to you” — need not come up: The Nash bargaining solution, or any of several
alternative schemes, provides a reasonable “split-the-difference” procedure.)
Unfortunately, there are sharks in these here waters (not necessarily taught by me — if they didn't exist,
God would evolve them). And in consequence, I cannot counsel totally fish-like behavior. The Salty Dog
case provides a good example. If as a buyer you truthfully and convincingly reveal your reservation price at
the beginning of the negotiations, a seller with any reservation price below yours faces an overwhelming
temptation to exaggerate his own reservation price and offer to split the difference. For in this case, as in
many in the business and personal worlds, the exaggeration will never be subsequently detected. (The shark
need only wear fish clothing.)

What counsel can I give? If I taught you precisely how to formulate an opening offer on the basis of your
reservation price, but never to walk away when a profitable deal is on the table, others could “invert” my
teachings, accurately infer your reservation price from your opening offer, and squeeze you as before.

Fortunately, there is something that I can teach, if you must, as a buyer, make the first offer. Make an offer
(even an invertible one), and make it absolutely clear that any higher counter-offer has some probability of
being rejected, even if your best alternative is worse than the counter-offer. If your statement (threat?) is
believed, then the best the seller can do is maximize his expected gain; if your stated probability of
accepting an offer very close to your reservation price is quite low, his optimal counter-offer will probably
offer you gains of substance. (If he squeezes too hard, he is very likely to gain nothing.) Of course, the
same teachings apply to the seller, if he must move first.

In the real world, it is not socially graceful to be so blunt. We make a starting offer, and signal our walk-
away likelihood through our reluctance to move, or through the size of our concessions. Being able to read
another's signals accurately, and to correctly give the proper signals ourselves, is a skill which comes only
through experience. (Role-playing exercises provide a riskless way to gain some such experience. I don't
have time in this course to do an adequate number of skill-building role-plays, but then, mine is not the only
course at Kellogg which deals with negotiations.)

The bottom line is that, in order to be at least a spiny fish, you must be able to convince others that you are
willing, at times, to walk away from an agreement that offers you gain. And unless you are an accomplished
hypnotist, the only way to be convincing is to be someone who sometimes walks away.

What is a shark? Someone who can convince others that he is prepared to permanently walk away, and
still come back gracefully if they don't budge. A typical tactic is to let an affiliated party “override” your
decision: A lawyer, having walked, can later return saying, “My client wants me to try again.” An arms
negotiator can be “sent back to the table” by the President. A car buyer can be “overruled” by a spouse.
Alternatively, you can come back yourself, saying, “I just realized that there is another way we can structure
the deal that might be acceptable.” (For example: “Oh, the car comes with tires? Then maybe I can raise
my offer a bit.”) Still, Lincoln's caution is of relevance: “You can fool all of the people some of the time,
and some of the people all of the time. But you can't fool all of the people, all of the time.” If you never
truly walk, you're definitely exploitable.

It is overly simplistic to say that concerns about long-term relationships or reputation should restrain a
shark. It's worth noting here that “walking away” can happen in the middle of a continuing relationship:
Rather than settling a disagreement quickly and amicably, a husband and wife might subject each other to a
day or two of sulking silence. Or a union might strike. But the relationship endures. The occasional
failure-to-settle maintains a healthy atmosphere of mutual respect.

Of course, many (nay, most) people indeed never walk away from gainful agreements. And others do
exploit this. And you, in turn, can exploit them. An effective salesman will expect you (a “typical”
purchaser) to walk only if the price on the table is above your reservation price. He'll start well above and
slowly come down, reading you all the way. As soon as he thinks you're no longer willing to walk (i.e., as
soon as he thinks he's hit your reservation price), he'll freeze. Feeling that he's the shark and you're the
fish, he'll come all the way to his reservation price only if he feels that your reservation price is very near
his. So you, a spiny fish, wear your pure-fish clothes. You do your research beforehand, deciding what you
want and what reservation price the salesman is likely to have. You approach him, and indicate a desire to
spend some amount less than this. You let him boost you to more expensive models by showing disinterest
for those listed near your “expressed” budget, but you never admit the ability to pay any more. (He'll
assume you can be dragged up somewhat.) When he finally shows you what you want, the game begins.
He makes some concessions, you come up in a few tiny steps, then freeze just above his reservation price.
If necessary, you begin to walk. Don't worry: If you did your research well, and froze anywhere above his
reservation price, he (not knowing that you're spiny) will mistakenly conclude that he's hit your reservation
price, and will close the deal.

Must you lie about (or at least, “strategically misrepresent”) your position? Not necessarily. But in practice
most people back themselves into situations where they must. The salesman asks, “What is the most you
can afford?” An indignant “If I told you that, you'd drag me up to it!” simply alerts him to your spinyness.
Without prior thought (and practice!) you'll probably cite a falsely-low amount. (He expects this: Most
people will. In that case, are you really lying?) A more principled approach might be to turn the question
around: “Well... How far down can you come?” But the turn-around itself is a bit of a tipoff that you know
what you're doing.

Fish working with other fish build excellent agreements, and share equally in the joint gains. But fish
working with disguised sharks (and most true sharks do wear fish clothing) build pretty decent agreements,
and then the sharks take most of the gains for themselves. If the sea contains both fish and sharks (as, in
fact, it assuredly does), the fish frequently lose out. But sharks against sharks get nothing.

A shark wants it all; a spiny fish recognizes the needs of other parties. Spiny fish build decent agreements
with other spiny fish, and, after dancing the strategic dance, share in the gains. Spiny fish are relatively safe
from sharks (and can sometimes turn the tables).

The biggest problem a spiny fish (i.e., an effective negotiator) faces is in deciding how to deal with pure fish
(i.e., those less-skilled). Does he mimic a pure fish (sacrificing potential gain to protect the other party), or a
shark in pure-fish clothing? I wish I had an easy answer. But it seems clear that, once we abandon the
unrealistic dream of an all-pure-fish ocean, the best we can work for is a sea in which most of the fish have
spines. (Perhaps we can even starve out the sharks.) Wearing our own spines proudly, and nipping at the
pure fish without devouring them, leaves us in excellent shape and encourages them to grow their own
spines. And maybe, just maybe, the right ecological balance can be achieved.
Example 8 (dissolving a partnership). Two individuals jointly own a piece of property. They have
decided to sever their relationship, and for one of the two to buy the land from the other. Each knows how
valuable the land is to him, but is unsure of its worth to the other. They agree that each will write down a
bid; the high bidder will keep the land, and pay the amount of his bid to the other.

Assume that each is equally likely to value the land at any level between $0 and $1200, and that both know
this. Then the unique Bayesian equilibrium point of the bidding game is for each to bid one-third of his own
valuation. If, for example, one of them values the land at $300 and believes the other to be following the
indicated equilibrium strategy, then by bidding $100 he has an expected payoff of 1/4 ⋅ $200 + 3/4 ⋅ $250;
he expects to win with probability 1/4, and when he loses, he expects the other's (winning) bid to be
between $100 and $400. This private strategy is optimal for him, given his belief about the other's behavior.
(Given his belief that his partner will bid a third of the partner's valuation, his own expected payoff, when
his valuation is v and he bids b, is

(3b/1200) ⋅ (v-b) + (1 - 3b/1200) ⋅ (b+400)/2.

In general, this is maximized by taking b = v/3 .)

Observe that this bidding arrangement always yields a Pareto-efficient result, i.e., the individual who values
the land more highly always ends up in possession of it. Hence, the appropriate choice of a dispute-
resolution procedure can, at times, circumvent inefficiencies of the type which arise in the lemon problem.

Note also that an intervenor could suggest the use of this procedure, if the parties found themselves unable
to work out an agreement on their own.

4.3 The revelation principle

There are, of course, numerous other procedures that an intervenor could suggest in order to resolve the
dispute in Example 8. Let us consider the (seemingly appalling) general question of what outcomes can
result, at equilibrium, from any procedure which might be used to resolve a given dispute.

A simple, yet conceptually deep, type of analysis has become standard. Consider any equilibrium pair of
strategies in a particular game. Each party's strategy can be viewed as a book, with each chapter detailing
the private strategy of one of that party's types. Given the two actual types, a pairing of the private strategies
in the two appropriate chapters will lead to an outcome of the game.

Next, step back from this setting, and picture the two parties in separate rooms, each instructing an agent on
how to act on his behalf. Each agent holds in hand the strategy book of his side; all he must be told is which
chapter to use. From this new perspective, the two parties can be thought of as playing an
“agent-instruction” game, in which the strategy books are prespecified and each must merely tell his agent
his type (or, equivalently, point to a chapter in his strategy book). An equilibrium point in this new
“chapter-selection” game is for each to tell the truth to his agent. Otherwise, the original strategies could
not have been in equilibrium in the original game.

Consequently, anything which can be accomplished at equilibrium through the use of any particular
dispute-resolution procedure, can also be accomplished through the use of some other procedure in which
the only actions available to the parties are to state their (respective) types, and in which it is in equilibrium
for each to truthfully reveal his type.

- 12 -
Example 9. Two parties must share the cost of a public works project (e.g., the planting of a tree on the
boundary line separating their homes); if they cannot agree, the project will not be carried out. The project
will cost $100. Both parties are risk-neutral, and it is known by both that Party A will derive $90 in benefit
from the project. Party B knows the benefit he will receive, but all that is known to A is that there is a
50% chance it is worth $90, and a 50% chance it is worth only $30, to B. What possible agreements could
they reach?

The revelation principle tells us that any agreement which could be arrived at through any negotiation
procedure will be an agreement which could also be achieved in a formally-structured game in which each
simply names his type, and each has no incentive to lie. (Since A's type is known to both, only B will
actually have a move in this game.)

An outcome of the revelation game will be, most generally, a probability that the project will be carried out,
and a sharing of the $100 cost if it is indeed carried out. Since a different outcome might result from each
of the two type-declarations B might make, the full spectrum of possible agreements can be characterized
by four parameters: pH and pL, the probabilities of project commencement given that B announces his type
to be “high” ($90) or “low” ($30), and eH and eL, the payments to be made by B given his announcement
and that the project is carried out.

In order for truth-telling to be optimal (i.e., a best response to A's null action) for B, these parameters must
satisfy two incentive constraints:

(90 - eH)⋅ pH ≥ (90 - eL)⋅ pL (the $90-type must prefer announcing “H” over “L”)

(30 - eH) ⋅ pH ≤ (30 - eL) ⋅ pL (the $30-type must prefer announcing “L” over “H”)

Furthermore, in order for A, and for both types of B, to be willing to agree to the procedure, it must satisfy
the following participation constraints:

1/2 ⋅ pH ⋅ (eH - 10) + 1/2 ⋅ pL ⋅ (eL - 10) ≥ 0 (for A to participate)

eH ≤ 90 , eL ≤ 30 (for both of B's types to participate)

It follows (algebraically) from all this that we must have pH ≥ pL and eH ≥ eL ; that is, when B reports
himself to be the $90-type (in practice, when he acts as if he is that type), the project is more likely to be
carried out, and he will be charged a larger share of the cost.

There are agreements which will lead to the project always being done, i.e., agreements with pH = pL = 1.
However, such agreements must have eH = eL ≤ 30, and hence A must bear at least 70% of the cost,
independent of B's type. Any alternative agreement which lessens A's burden must have pL < 1, and hence
must require that the project is sometimes not carried out. For example, if gains from the project are to be
split evenly between A and the announced type of B, then we must have eH = 50 and eL = 20; if the project
is to be certainly carried out (pH = 1) when the $90-type is announced, then pL can be at most 4/7.

One interpretation of this result is that efficiency and equity are, at times, at least partially incompatible.
Only the threat by A of not doing the project can “separate” the two types of B, and this threat is only
viable if, when B claims to be his $30-type, A sometimes actually carries it through.

- 13 -
Example 10 (bilateral trade). One classical type of bargaining problem involves a seller and a buyer, each
uncertain of how the other values an object currently held by the seller. Assume that each believes the other
to be equally likely to value the object at any amount between $0 and $300; each, of course, knows his own
valuation. According to the revelation principle, the possible agreements which can result from any choice
of negotiation format can be characterized by a pair of functions p(vS,vB) (the probability that trade takes
place when the seller announces his valuation to be vS and the buyer announces his to be vB) and e(vS,vB)
(the amount to be paid by the buyer when these announcements are made and trade does take place). This
pair of functions must satisfy a continuum of incentive constraints: Each buyer or seller type must prefer
announcing truthfully to making any other announcement. Furthermore, the functions must satisfy a
continuum of participation constraints: Every seller type must expect to be paid at least his valuation when
trade takes place, and every buyer type must expect to pay no more than his valuation.

Consider one particular format for arranging a sale. Each writes down a price. If the seller writes a higher
price than the buyer, no trade occurs; otherwise, the object is sold at the average of the two amounts. It is
simple to show that it is not in equilibrium for both to write truthfully their valuations: If either is truthful,
the other can gain by exaggeration. One natural equilibrium pair of strategies is for the seller to write down
2/3 ⋅ tS + 75 , where tS is his actual valuation, and for the buyer to write down 2/3 ⋅ tB + 25 , where tB is his
valuation. Notice that when the buyer's valuation is only slightly greater than the seller's, trade does not take
place; indeed, if the seller's type is greater than 225, there is never a trade.

Consider an alternative mechanism, wherein each writes down an amount (vS and vB, respectively), and
trade takes place only if vS ≥ vB - 75, at price of (vS+vB)/3 + 50. It can be verified that it is in equilibrium
for both to tell the truth; furthermore, every pair of types faces (at equilibrium) the same outcome here as
they did in the previous mechanism. This latter mechanism is, in fact, the “revelation game” derived from
the former game and equilibrium point using the approach outlined at the beginning of this section. (It is
known that the mechanism given here maximizes the traders' ex ante (before they learn their types) expected
joint gains from trade; if one were organizing a market within which such traders would be forced to deal,
this mechanism would be the natural choice.)
300

vS + vB
+ 50
3

vB
no trade

75

0
0 225 300
vS

- 14 -
4.4 Incentive-efficiency

It would seem foolish for an arbitrator to suggest a mechanism for dispute resolution which left every type
no better off, and some types worse off, than some other mechanism would have. In other words, the
suggested mechanism should be incentive-efficient, i.e., efficient subject to the incentive constraints. A
generalization of Nash's solution to the complete-information bargaining problem should therefore select a
particular incentive-efficient mechanism.

In Example 9, the incentive-efficient agreements all have pH = 1. Furthermore, if eL ≥ 10, then

pL = (90-eH) / (90-eL).

In Example 10, the mechanism presented can be shown to be one of the incentive-efficient mechanisms.
Furthermore, no incentive-efficient mechanism is ex post Pareto-efficient: It is impossible to arrange for
advantageous trades to always take place.

4.5 Equity and durability

Although parties usually enter negotiations with a primary objective of reaching an agreement advantageous
to themselves, much of the ensuing discussion between the parties concerns the “fairness” of different
proposed agreements.

From where do the parties obtain their notions of what is fair? Certainly, there are commonly accepted
principles which are culturally based. “The greatest good for the greatest number,” “From each according to
his ability; to each according to his needs,” and “Whatever can be obtained from the sweat of the brow” are
examples of such principles; clearly, they stand somewhat in contradiction to each other.

Sometimes, precedent plays a role in perceptions of fairness. Labor negotiations typically take the previous
contract as a starting point, and each party will argue that a concession on one issue “should” be matched by
an opposing concession on another. At other times, a neutral third party will be asked to resolve a dispute in
terms of his external view of equity: The parties will submit their dispute to binding arbitration.

The Nash solution in settings of complete information was derived from a list of desired properties, at least
two of which (individual rationality and symmetry) were directly concerned with equity. Furthermore, the
use of threat-making to establish the original conflict outcome carries with it a notion of equity: Those who
will suffer relatively more if agreement is not reached, receive relatively less from the agreement which is
reached. (In Example 4, Alfred receives less than half of the monetary gains available from trade with
Burton.)

Recently, Myerson has proposed a generalization of the Nash solution to bargaining games with incomplete
information. His approach gives explicit regard to the inter-type competition we have previously discussed.

Example 10 (continued). Assume that the mechanism described earlier (each names a price; trade takes
place if the named prices are compatible, at the average of the two named prices; each follows his specified
equilibrium strategy) is proposed to two traders. At the time of the proposal, each trader of course knows
his own type. If the seller's valuation is greater than $225, he will naturally make an objection. For
example, if his valuation is $250, he expects no trade to take place if this mechanism is used. Instead, he
could commit himself to a first-and-final offer of some higher amount, say, $275. Although the buyer may
be antagonized by this action, if his valuation is greater than $275 and he believes the seller's commitment

- 15 -
then there is some chance that he will accept the offer. Similarly, if the buyer's valuation is less than $75, he
will object to the use of this mechanism.

Even when the seller's valuation is less than $225, and he agrees to use the proposed mechanism, his mere
agreement reveals something about him - namely, that his valuation is less than $225. (Otherwise, the buyer
would expect him to object.) With this extra information, the buyer might choose to reject the proposed
mechanism, and put additional pressure on the seller.

Consequently, this mechanism is not “durable,” in the sense that either it will not be accepted by at least one
party, or it will be accepted and the equilibrium will not persist, since the parties' beliefs about each other
will be changed by their acceptances.

In response to this difficulty, Myerson has proposed a “neutral” bargaining solution, which takes into
account the inter-type conflict which could upset a proposed mechanism. For the buyer-seller problem we
are considering, the neutral bargaining solution is summarized by the figure below: Each trader names his
valuation; trade takes place if the named valuations (vS for the seller, and vB for the buyer) lie in one of the
two triangles, at the price indicated within the triangle. It can be shown that this mechanism is both
incentive-compatible (it is optimal for each to tell the truth, given that the other does so) and
incentive-efficient. Notice that trade takes place more often for the extreme types than under the previous
mechanism, and slightly less often for the intermediate types. This must be so, in order for the extreme
types to not wish to “upset” the mechanism.

300

vS
+ 150
2
225

vB
2
vB

no trade

0
0 75 300
vS

- 16 -