American Academy of Political and Social Science

Modeling Distributive and Integrative Negotiations

Author(s): Otomar J. Bartos
Source: The Annals of the American Academy of Political and Social Science, Vol. 542,
Flexibility in International Negotiation and Mediation (Nov., 1995), pp. 48-60
Published by: Sage Publications, Inc. in association with the American Academy of
Political and Social Science
Stable URL: http://www.jstor.org/stable/1048208
Accessed: 30-04-2018 21:14 UTC

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
http://about.jstor.org/terms

Sage Publications, Inc., American Academy of Political and Social Science are
collaborating with JSTOR to digitize, preserve and extend access to The Annals of the
American Academy of Political and Social Science

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
 ANNALS, AAPSS, 542, November 1995

Modeling Distributive and

Integrative Negotiations

By OTOMAR J. BARTOS

ABSTRACT: Two main types of negotiation processes can be distin-

guished, distributive and integrative. While the distributive process
consists primarily of concession making, the integrative process
involves both concession making and a search for mutually profitable
alternatives. Thus the meaning of "flexibility" is not always the same:
in distributive negotiations, it means "readiness to make conces-
sions"; in the integrative negotiations setting, it means in addition
"readiness to engage in the search process." Computer simulations
suggest that the distributive process is faster but that the integrative
process is potentially more productive in that it can increase the
chances of an agreement. In order for integrative negotiation to fulfill
its potential, however, the search engaged in must be joint, and
during the search, the adversaries must interact face to face. Dis-
tributive negotiation can also be made more productive by having the
adversaries interact face to face, under conditions that emphasize
their similarities.

Otomar J. Bartos is a professor of sociology at the University of Colorado in Boulder

As a specialist in theoretical and mathematical sociology, with particular interest in
the theory of negotiation, he has published numerous articles as well as two books, The
Process and Outcbme of Negotiations and Simple Models of Group Behavior. Most
recently, he has been interested in future societies and has been working on his next
book, The Sociology of the Future.

48

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
DISTRIBUTIVE AND INTEGRATIVE NEGOTIATIONS 49

EARLY attempts to formalize

arms race model,3thein which each ne-
process of negotiation were
gotiator's tai-
behavior is governed by the
lored to very simple cases, such as following equation:
that of a buyer and a seller.1 More dD = -k*dO-a*D + g O < k, a < 1 (1)
recently, this image has been ques-
tioned as too simple to apply to cases where D is the current demand of the
such as international negotiations, negotiator, dD is the current change
and a new approach has been advo- in his demands (a negative dD corre-
cated.2 This article attempts to exam- sponding to a concession), and dO is
ine the theoretical differences be- the current change in the opponent's
tween the older distributive and the offers (a positive dO corresponding to
newer integrative images of negotia- opponent's concession). Parameter k
tion and to lay a foundation forcontrols a the tendency to reciprocate;
model that, ultimately, could encom- a, the tendency to make unilateral
pass both approaches. The differ- concessions; and g, the level of feel-
ences-and similarities-between ing. (Somewhat confusingly, when g
the two images become clear
is a if we value, the negotiator
negative
convert them into formal models that feels friendly; when g is positive, he
can be used to run computer simula- feels unfriendly.)
tions. This process elucidates the im- To understand fully the role played
plications of each model for flexibleby these three parameters, note that
bargaining behavior. it follows from equation 1-and from
the assumption that parameters k
DISTRIBUTIVE BARGAINING and a are positive numbers-that
each negotiator will make conces-
sions (that is, that dD will be nega-
The image of distributive bargain-
tive) hag-
ing best fits situations involving if the following conditions occur:
gling over the price of a product such
as a used car. The essential theoreti-
- if the opponent has just made a
concession (if dO is positive).
cal problem here is to determine why Thus the model assumes that
and under what conditions the two
the negotiator reciprocates his
parties will make concessions.
opponent's concessions, and k
may be called his reciprocation
A formal model parameter: the larger k is, the
larger will be the negotiator's
To answer these questions, we can
concession in response to his op-
use a modified version of Richardson's
ponent's concession.
1. Sidney Siegel and Lawrence Fouraker, - if the negotiator's last demand
Bargaining and Group Decision-Making (New (D) is high. Since term a, the
York: McGraw-Hill, 1960).
2. I. William Zartman, ed., The 50% Solu-
unilateral concession-making
tion (New York: Anchor Books, 1976); idem, parameter, is assumed to be
The Negotiation Process (Beverly Hills, CA:
Sage, 1977); R. Fisher and W. Ury, Getting to 3. Lewis F. Richardson, Arms and Insecu-
Yes (New York: Penguin Books, 1977). rity (Pittsburgh, PA: Boxwood Press, 1960).

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
 50 THE ANNALS OF THE AMERICAN ACADEMY

positive, it creates
Theamodel
negative
of equation 1 implies that
there are three internal
feedback: if the negotiator's lastcauses of con-
demand D was high, his
cession conces-
making: desire to reciprocate
sion dD will be large;
(largeif demand
k), desire to initiate reciproca-
D was low, his concession
tion (large will
a), and be
friendly feelings
small. Thus the model implies
toward the opponent (negative g).
that unilateral concession mak-
The model also implies that the
ing is most pronounced at the
main consequence of flexibility is an
early stages of the negotiation, increase in the opponent's flexibility:
when demands are high. since the parameter k is assumed to
- if the parameterg is negative. be It positive, the negotiators recipro-
is meaningful (as well as cus- cate each other's concessions. In
tomary) to assume that g refers cases when k is actually negative
to the emotional state of the ne- when the negotiator exploits his op
gotiator, a positive g being a ponent by decreasing his own conc
grievance, a negative g repre- sions in response to concessions b
senting friendliness. Thus term g the opponent-flexibility generat
may be referred to as the grievance- inflexibility. But it is hard to conce
friendliness parameter. that such a cycle could last very lon
In general, flexibility not only pr
Flexibility motes further flexibility, but it a
makes an agreement more likely, b
Speaking generally, a negotiator is cause large concessions bring the n
flexible if he is willing to engage in gotiator to an agreement--if an agr
any behavior that promises to facili- ment is indeed possible7-sooner th
tate agreements. Within the context small concessions.
of distributive bargaining, it seems
natural to define flexibility more INTEGRATIVE NEGOTIATION
narrowly as willingness to make
concessions. In the 1970s, theoretical interest
began to shift toward an integrative
While many external conditions--
such as the negotiator's cultural approach.s While different writer
background4 or his preparations and have somewhat different ideas about
what this approach entails and give
relevant bureaucratic factors5--un- it somewhat different names, there is
doubtedly affect flexibility, I will con-
sider here. only conditions located
ence and are influenced by other variables of
within the negotiation process itself.6
the model, such as variables dD and dO of
4. Daniel Druckman et al., "Cultural Dif-equation 1.
ferences in Bargaining Behavior: India, Argen- 7. In the model of equation 1, an agree-
tina, and the United States," Journal of Con- ment is possible only if there is a proposal that
flict Resolution, 20(3):413-53 (1976). has, for both negotiators, payoff at least as high
5. Daniel Druckman, "The Situational as their bottom line, D*, of equation 3.
Levers of Negotiating Flexibility," Journal of 8. Zartman, 50% Solution; idem, Negotia-
Conflict Resolution, 37(2):236-76 (1993). tion Process; Fisher and Ury, Getting to Yes; D.
6. These are the endogenous variables of G. Pruitt, Negotiation Behavior (New York:
the model, that is, variables that both influ-Academic Press, 1981).

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
 DISTRIBUTIVE AND INTEGRATIVE NEGOTIATIONS 51

TABLE 1
PAYOFFS ASSOCIATED WITH THE FIVE-POINT AGENDA

Payoff to

Proposal China United States

1. Noninterference in internal affairs 12 -9

2. No barriers to free trade -1 4
3. Inspection teams to monitor compliance -2 5
4. Protection of human rights -10 13
5. Destruction of nuclear weapons 20 -17

enough agreement about

Shouldits essen-
we build an entirely new
tial features to allow us to model it
model, or should we simply add the
formally. search, process to the bargaining
model? Recognizing that even the
A formal model most cooperative negotiators cannot
totally ignore their own selfish inter-
For reasons that will become clearests, I chose to add the search process
shortly, it is not possible to defineto
a the bargaining model.
model of integrative negotiation byEven so, building a model of inte-
means of a simple equation. How-grative negotiation proved challeng-
ing. It soon became clear that the
ever, it is possible to specify the basic
assumptions of the model, inputimagery of buyer and seller is not
very useful, since these actors almost
them into a computer,, andruns.
examine never need to search for new alterna-
its properties by simulation
Only the most basic assumptions and tives. Ultimately, I chose the image of
features of the model will be given two negotiators dealing with a com-
here.9 plex and somewhat odd agenda. It is
odd because it has two sets of alter-
The image of negotiation as dis-
tributive bargaining came to be seen natives, those that are known to the
as deficient because it ignored an im- negotiators when the session begins,
and those that are unknown to them
portant aspect of the negotiating pro-
cess: the fact that stalemates can be at the beginning but are knowable
because they can be discovered.
broken by making new and imagina-
Table 1 clarifies these distinctions.
tive proposals. In other words, the
As the payoff matrix shown in the
bargaining model did not allow the
table was used in actual experi-
negotiators to search for better alter-
ments,10 it may be helpful to describe
natives than those explicitly before
them. Thus to build a model of inte- briefly what transpired in these ex-
periments.
grative negotiation means, above all,
allowing for search behavior.
9. Amore detailed description of the model 10. For a detailed description of the experi-
is given in Otomar J. Bartos, "A Model of Inte-ments, see Otomar J. Bartos, Simple Models of
grative Negotiation" (Paper, University of Group Behavior (New York: Columbia Univer-
Colorado, 1994). sity Press, 1967).

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
 52 THE ANNALS OF THE AMERICAN ACADEMY

The roles of the United States and individual proposals. In all, the un-
China were played by college stu-known agenda consisted of 26 combi-
dents. They were given confidential nations, or package deals, that could
dossiers that contained a detailed de- be created from the five proposals.
Table 2 shows both the known and
scription of five proposals," their gov-
ernment's position on each proposal, unknown agendas.
and some arguments they could use The formal model constructed to
to defend their position. They wereinclude search behavior made the fol-
also informed about their own (but lowing assumptions:
not the opponent's) payoffs. Once they
became familiar with these docu- 1. Each negotiator starts by endors-
ments, actual negotiation started. that option that has the highest
ing
Each side was allowed to make short payoff for him. (This is the same as-
speeches specifying which proposal sumption as in the bargaining model.)
or combination of proposals it would 2. After having heard his oppo-
be willing to agree to "now." nent's first offer, he defines his new
If an agreement was reached level of aspiration. (This is deter-
within the allotted period of time (onemined by equation 1: what in the
and a half hours), each side receivedbargaining model was the negotia-
tor's new demand now becomes his
cash payment equal to the sum of its
own payoffs for the agreed-upon pro-new level of aspiration.)
posals. For example, if the two sides 3. He then selects and publicly en-
agreed on proposals 1 and 4 (nonin-dorses one of the options that lie
terference in internal affairs and pro-within the eligible set, that is, that
tection of human rights), the studenthave for him a payoff smaller than
representing China received $2.00 his last demand but larger than his
(12 - 10 = 2), and the student repre-current level of aspiration.12 (Within
senting the United States receivedthe eligible set, the probability that
$4.00 (-9 + 13 = 4). If no agreementan option will be chosen is inversely
was reached within the allotted time,related to its complexity,13 the sim-
each subject received no money. 12. For example, suppose that China at
It is not difficult to see that each some point endorsed proposal 5, thus demand-
ing the payoff of 20. Now suppose that, as a
subject was presented with both aresult of a U.S. counteroffer, China's level of
known and an unknown agenda. The aspiration D (computed from equation 1)
known agenda, given to him before dropped to 12. Then the eligible set would
include all proposals and package deals that
negotiations began, consisted of the
have, for China, payoffs between 12 and 20. As
five proposals shown in Table 1, with
Table 2 shows, these are the package deals
associated payoffs. The unknown {2,5), {3,5), {2,3,5), and {1,2,3,4,5).
agenda was discovered by him gradu- 13. It is assumed that the negotiator never
ally, as the negotiation progressedcomputes the payoffs from all package deals.
and as he computed the payoffs Instead,
asso- at any given time, he merely searches
ciated with different combinations ofin the vicinity of the package deal he just
endorsed. He accomplishes this by modifying
11. For example, proposal 1 states that that
nei- package deal, by either omitting from it
proposals with the lowest positive payoff or
ther nation will interfere in the internal affairs
of the other nation. adding proposals with the highest negative

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
 DISTRIBUTIVE AND INTEGRATIVE NEGOTIATIONS 53

plest options being most likely

Using to be to determine th
this equation
found and endorsed.) consecutive levels of aspiration (as
4. The opponent's counteroffer sumption 2), the computer generate
again redefines his level of
a series ofaspira-
demands that, according t
tion, and he makes a new endorse- the program, each negotiator woul
ment in the manner specified in as-make.
sumption 3. This process continues More specifically, the negotiators'
until either an agreement is reached opening demands were for the largest
(that is, the negotiators endorse the possible payoff (assumption 1):
same option) or time runs out.14 China advocated the adoption of pro-
posals 1 and 5 (thus, in effect, de-
manding $32); the United States ad-
Computer simulation
vocated the adoption of proposals 2,
These general assumptions pro- 3, and 4 (thus, implicitly, demanding
vided the basis for a computer model $22). Following this opening de-
that simulated the actual experi- mand, the computer proceeded to
ments. To accomplish this, however, identify second demands. Using
two specific assumptions had to be equation 2, it defined the current eli-
gible set (assumptions 2 and 3) for
added. First, the negotiators were as-
sumed to have the payoffs given in each negotiator and selected a pro-
posal or package deal from each set.
Table 2. Second, the general equation
These became the negotiators' second
1 was replaced by a specific equation,
endorsements, and the payoffs asso-
one with parameters that create a
ciated with them became their sec-
reasonable15 path of demands:
ond demands (assumption 3). The
dD = -.3*dO -.2*D + .5 (2) computer continued in this manner
until either each negotiator had
made
payoff. Moreover, he tends to favor the simplest thirty demands or an agree-
package deals, that is, deals with the fewestment had been reached.16
proposals in them. Formally, this is expressed
Since the computer had to make
by assuming that the probability with which a
its
package deal is endorsed is inversely propor- choices in a probabilistic manner
tional to its complexity, that is, to the (assumption 3), each simulation run
number
was
of proposals included in it. For example, if the different. To obtain a typical se-
eligible set includes package deals {2,5), {3,5),quence
of demands, fifty separate
12,3,5), and 11,2,3,4,5), then these four package
simulations were run and then ag-
deals are assigned probabilities w*(1/2,
1/2,1/3,1/5), where w is a number that causes gregated into an average. That is to
the probabilities to sum to 1. That is, 12,5) and say, the first typical demand was the
{3,5) are each assigned the probability of .33, average of all fifty first demands, the
{2,3,5) the probability of .22, and {1,2,3,4,5) the secondtypical demand the average of
probability of .12. One of these package deals
is then selected in accordance with these
all fifty second demands, and so on.
probabilities.
14. Bartos, "Model of Integrative Negotiation."
16. In the simulations using Table 2, an
15. A path of demands was deemed reason-
agreement
able if it converged to an agreement within a was reached when, in their con-
relatively short period of time, such secutive
as 20 speeches, the adversaries endorsed
rounds of endorsements. the same proposal or package deal.

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
54 THE ANNALS OF THE AMERICAN ACADEMY

TABLE 2

THE COMPLETE AGENDA OF THE GAME OF TABLE 1

Payoff to

Alternative Proposals in the Alternative China United States

The Known Agenda (Proposals)

1 1 12 -9
2 2 -1 4
3 3 -2 5
4 4 -10 13
5 5 20-17
The Unknown
6 {1,2} 11 -5
7 {1,3} 10 -4
8 {1,4} 2 4
9 {1,5} 32 -26
10 {2,3} -3 9
11 {2,4} -11 17
12 {2,5} 19 -13
13 {3,4} -12 18
14 {3,5} 18 -12
15 {4,5} 10 -4
16 {1,2,3} 9 0
17 {1,3,4} 0 9
18 {1,4,5} 22 -13
19 {1,2,4} 1 8
20 {1,2,5} 31 -22
21 {1,3,5} 30 -21
22 {2,3,4} -13 22
23 {2,3,5} 17 -8
24 {2,4,5} 9 0
25 {3,4,5} 8 1
26 {1,2,3,4} -1 13
27 {1,2,4,5} 21 -9
28 {1,3,4,5) 20 -8
29 {2,3,4,5} 7 5
30 {1,2,3,5} 29 -17
31 {1,2,3,4,5} 19 -4

Flexibility in makes concessions, he induces his op-

integrative negotiations
ponent to make them, too. There is no
In integrative bargaining, flexibil- similar assumption about the search
ity has two different aspects: conces- process: one negotiator may search
sion making (since the new model for new alternatives without induc-
still includes distributive bargaining) ing similar behavior in his opponent.
and searching. Anegotiator is flexible The reason for this difference is that
if he is willing to make concessions the conditions conducive to conces-
and to search for alternatives. sion making are inherently different
The bargaining model of equationfrom the conditions conducive to
1 assumes that when a negotiator searching.

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
DISTRIBUTIVE AND INTEGRATIVE NEGOTIATIONS 55

Both models assume that each ne- ternatives, agreement becomes pos-
gotiator will always do what is ra- sible. Let us, then, examine carefully
tional. But concession making is notwhat some writers have claimed are the
(unconditionally) rational: if the ne-advantages of integrative negotiation.
gotiator cannot trust his opponent to
reciprocate, a concession will seem to
Merits of the integrative approach
be giving up something without gain-
ing anything--and he will refuse to Fisher and Ury claim that integra-
make concessions. To be plausible,tive negotiation, which they term
concession making must be made ra- "principled" negotiation, is superior
tional. The bargaining model accom- to distributive bargaining both cogni-
plishes this by assuming that conces-tively and emotionally: it "permits
you to reach a gradual consensus
sions will be reciprocated, namely, by
more efficiently ... [and it makes]
restricting k in equation 1 to positive
numbers. possible an amicable agreement.""
Matters are quite different when Let
it us evaluate this claim.
comes to searching. Searching for
Efficiency. To determine whether
new alternatives is (unconditionally)
integrative negotiation is more effi-
rational; that is, it is potentially prof-
cient than distributive negotiation, I
itable even if the opponent does not
asked a simple question: is it easier
join in. For example, if the negotiator
to reach an agreement when the ne-
discovers that none of the known op-
gotiators engage in search behavior
tions is by itself acceptable to both
than when they do not? To find an
sides (as is the case in Table 1), then
answer, I examined the problem of
he is motivated to search for new
Table 2 under two conditions. First,
alternatives, such as the package the model was run under the as-
deals listed in Table 2. This is ra-
sumptions that all 31 alternatives
tional, since he gains something were known. This meant that search
without having to give up anything. was not needed. Second, the model
Consequently, there is no reason to
was run under the assumption that
assume that one negotiator's search- the five proposals were known, but
ing encourages the other negotiator's the 26 package deals were unknown.
searching behavior, and no such as- This meant that search was needed.
sumption was made. All the simulation runs-with or
This fact puts the causes and con- without search--ended in the same
sequences of searching in a unique way, in an agreement on the package
light. Its main cause is self-interest; deal 12,3,4,5). That is to say, they all
its main consequence, facilitation of ended when both simulated negotia-
an agreement. But this facilitation tors endorsed this package deal, thus
stems from the structure of the situ-
in effect agreeing to eliminate barri-
ation, not from psychological or social ers to free trade (proposal 2), to im-
factors: if none of the known alterna-
17. Fisher and Ury, Getting to Yes, p. 14.
tives is mutually profitable, no agree-
While this claim is widely accepted as correct,
ment is, by definition, possible; if Fisher and Ury do not supply any empirical
search adds mutually profitable al- evidence to support it.

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
56 THE ANNALS OF THE AMERICAN ACADEMY

plement inspection teams (proposal

creates a more amicable atmosphere?
3), to protect human As rights
we have seen,
(proposal
there is some truth
4), and to destroy all in this claim,weapons
nuclear since search, being a
(proposal 5). As Table 2 shows,
jointly this can be-
profitable enterprise,
means that China always ended
come cooperative andupthus create a
with a payoff of 7 friendly
and the United
atmosphere. But it is possi-
ble 5.
States with a payoff of to engage
This in search either indi-
finding
is theoretically important in
vidually or its Suppose
jointly. own that the
way, both because thisnegotiators have reachedis
agreement an impasse
the Nashis solution ofand the
adjourngame
to reconsider
andtheir op-
because it is the agreement reached
tions. Each negotiator consults with
his("live")
in 68 percent of actual advisers and power brokers and
experi-
creates a new option.
ments.'9 Thus the simulation When this hap-
runs
pens, the new
provide additional support for the option becomes simply
claim that the Nash solution is the another proposal and does little to
most reasonable solution to conflicts promote amicability. On the other
of interest. hand, if the negotiators adjourn, go to
For the purposes of this article, how- dinner together, and engage in brain-
ever, another finding was most impor- storming, then the cooperative effort
tant: the simulation runs with search increases the level of friendliness.
were, on the average, longer than the To see why search creates friend-
runs without search. Searching often ship only if it is joint, let us consider
involves following false leads. Thus wethe theory of friendship suggested by
find what one might have known all Homans.21 Its essence can be repre-
along:20 integrative negotiation is not sented as shown in Figure 1. As is
as cognitively efficient as one might customary, the arrows in the figure
have supposed. At best, a habit of inte-represent causal relationships. For
grative negotiation sensitizes the ne- example, the double arrow between
gotiator, so that he can recognize ear- interaction and similarity may be ex-
lier that searching for new options may pressed as "The more two persons
be necessary. interact, the more similar they be-
come (in their beliefs, values, and
Amicability. But how about the norms), and vice versa."
claim that integrative negotiation The complete theory of friendship
18. John Nash, "Bargaining Problem," is more complex than Figure 1 sug-
Econometrica, 28(1):155-62 (1950). gests. It further assumes that fre-
19. Bartos, Simple Models of Group Behav-
quent interaction can have either
ior, p. 275. The Nash solution was reached in
positive
all simulated but not all actual experiments or negative consequences. If
because in the simulated negotiations the the
two individuals are fairly similar to
start
adversaries behaved in the same fashion, both with, interaction starts a posi-
obeying equation 2, whereas actual negotiators tive cycle: it leads to greater similar-
are bound to be different from each other.
These differences make reaching the Nash so-
lution less likely. 21. George C. Homans, The Human
20. Needless to say, what is obvious after it Group (New York: Harcourt & Brace, 1950);
is explained may be quite obscure before anidem, Elementary Social Behavior (New York:
explanation is found. Harcourt & Brace, 1974).

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
DISTRIBUTIVE AND INTEGRATIVE NEGOTIATIONS 57

FIGURE 1
HOMANS'S FRIENDSHIP PROCESS

Interaction

Similarity Liking

modified.
ity and friendship. If It should remain
they start as is,out
ex-
very dissimilar, cept that parametersinteractio
however, g and k should
starts a negative cycle:
no longer it It
be constant. leads to
should have
a built-in query:
increasing hostility and atdissimilar
time t, are the
ity. Negotiators negotiators who search interacting face
for to face?
new
alternatives share an interest in If they are, unfriendliness (g) is de-
creased
making agreement possible and are by some amount-say, from
in this sense similar. Therefore, .60 un-
to .50--and tendency to reciprocate
less their cultural differences are (k) is increased-say, from .20 to .25.
To see the consequences of this
very great, their interaction is likely
modification, let us consider the ne-
to start the beneficial cycle ofincreas-
ing similarity and friendship. gotiator's bottom line, the lowest de-
mand he will make. Equation 1 sug-
Returning to equation 1, we note
that two parameters are relevantgests
to that the lowest demand, D*,
the process of Figure 1. The first is when dD = 0 (there is no
occurs
change in the negotiator's demands)
the grievance-friendliness parame-
ter, g. As its name suggests, itand
ex-dO = 0 (there is no change in the
opponent's offers). Substituting these
presses how friendly the negotiators
values into equation 1 and solving,
are toward each other: ifg is positive,
we
the negotiator has a grievance against see that the lowest demand, D*, is
his opponent; if it is negative, he feels
D* = gla (3)
friendly toward him. The second pa-
rameter is the reciprocation parame- This equation provides
ter, k: the closer k is to 1, the more
formation about the co
equal will be the concessions ofjoint
the search: since it d
two negotiators. Thus parameter g
level of unfriendliness
the bottom line, D*, l
represents the level of unfriendliness;
coefficient k, the level of similarity.
22. It is perhaps clear from
This suggests how our modelifof parameter a remains cons
integrative negotiation ought to be
creases in size, D* must decr

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
 58 THE ANNALS OF THE AMERICAN ACADEMY

making agreement moretegrative

likely.view
Notetends to hide it by em-
that, although joint search increases
phasizing the benefits of creativity
the tendency to reciprocate (k), this
and problem solving but ignoring the
change does not affect fact
D*,these
since k
processes do not alter the
does not appear in equation 3. It could
underlying differences in payoffs.
Moreover, the
be shown, however, that increasing re-reasons for the su-
ciprocation affects the periority
time pathof the of
integrative approach
the demands: low demands are are often misunderstood. Our discus-
reached earlier, thus bringingsion
thesuggests
ne- that the reason for its
gotiators to an agreement earlier.
efficiency is rather unspectacular: it
can deal with situations for which the
IMPLICATIONS bargaining model was not designed.
It can handle (known) agendas con-
Formal models of distributive bar-
sisting of proposals that (1) are dis-
gaining and integrative negotiation crete, (2) have payoffs that are not
allow us to make some observations
related in a simple--that is, linearr-
about the way in which the process of
manner, and (3) cannot lead to an
negotiation and flexibility are under-
agreement. It deals with such agen-
stood by adherents of the two compet-
das by allowing the negotiators to
ing images of negotiation. search for new alternatives.
Finally, while a stalemate often
The process of negotiation makes searching for new alternatives
rational, search by itself does not cre-
Some authors draw sharp distinc-
ate an amicable atmosphere. It will
tions between the two images. They
do so only if it is joint, if it involves
see the distributive view as assuming
face-to-face interaction between simi-
that negotiators are adversaries who
lar adversaries.
reach agreement through a series of
Thus the negotiation process is best
concessions; the integrative view, as
seen as a combination of concession
assuming that negotiators are part-
making and search. When we deal
ners who cooperate in searching for a
with simple cases such as that of a
fair agreement. Moreover, they see
used-car dealer and a potential cus-
the distributive view as seriously flawed
tomer, search may be unnecessary, and
and the integrative view as inher-
the process becomes a special case of
ently better. Our discussion suggests
simple bargaining. When the agenda is
that this conception is oversimplified.
complex and no single proposal is jointly
To begin with, we should recognize
that there is a fundamental similar-
profitable, negotiation has to involve
both concession making and search.
ity between the two views. Both as-
sume, explicitly or implicitly, that ne-
gotiators behave rationally, each
trying to maximize his own payoff. 23. For example, the payoffs of one negotia-
tor P and the payoffs of a second negotiator Q
But while the bargaining view makes
could be related in a linear fashion such as P =
this assumption obvious, by portray-
-Q + 3. In Table 2, the payoffs are related in a
ing the negotiators as adversaries
complex fashion because their relationship
who struggle with each other, the in-
cannot be expressed by a single equation.

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
DISTRIBUTIVE AND INTEGRATIVE NEGOTIATIONS 59

Flexibility if an agreement is in the cards,24 dis-

tributive bargaining tends to lead to
Within the framework of distribu-
an agreement faster than the inte-
tive bargaining, flexibility may be
grative approach. Its disadvantage is
seen as concession making: the more
that if an agreement is not in the
concessions a negotiator makes, the
more flexible he is. The main internal
cards, the approach does nothing to
improve the chances of an agree-
cause of a negotiator's flexibility is
ment. The advantage of the integra-
the opponent's flexibility: the larger
tive approach is that it can improve
the concessions one makes, the larger
the chances of an agreement: if the
the concessions made by the other.
search for new alternatives involves
The main internal consequence is the
cooperative, face-to-face interaction
increased speed with which demands
between the adversaries, their origi-
and offers converge. It should be em-
nal hostility may be replaced by
phasized, however, that this result
friendly feelings-and this makes an
rests entirely on the assumption that
agreement more likely. Its disadvan-
the opponents are cooperative, that is,
tage is that search is time-consuming
that coefficient k is a positive number.
Within the context of integrative
and may not end in the best agree-
ment possible.25
negotiation, flexibility becomes more
complex. A negotiator is seen as flex- Yet there are some practical impli-
cations. One of these hinges on the fact
ible not only when he makes conces-
that an increase in friendliness makes
sions but also when he is willing to
engage in searching for new alterna-an agreement more likely. While the
tives. The main cause of the search distributive approach does not set the
behavior is negotiators' discovery stage for amicability in the same way
that the known agenda does not in- the integrative approach does, a nego-
clude any jointly profitable propos-tiator can augment the distributive ap-
als. Its main consequence is an im- proach so that it, too, generates amica-
provement in the likelihood that bility.
an How can he accomplish this?
agreement will be reached. This is needs to begin by fully appreci-
He
accomplished in two ways: by findingating the lesson taught by Figure 1:
any two persons are likely to become
new, jointly profitable alternatives,
and-if the search is joint-by in- friendly if they interact face to face
creasing the level of friendliness. and if their values and beliefs are
similar. Since distributive bargain-
Practical implications 24. One might say that an agreement is "in
the cards" if there is at least one proposal or
The practicing negotiator might
package deal with payoffs that, for both nego-
find it difficult to derive useful les-
tiators, are at least as high as their minimum
sons from our discussion. For exam- demand, D*, given in equation 3.
ple, if he were to ask which of the two25. Since, at each point in time, each nego-
tiator chooses his endorsements from the eligi-
approaches is better, he would be told
ble set in a probabilistic fashion (assumption
that each has its advantages and dis-
3), there is no guarantee that the best agree-
advantages. The advantage of the ment-for example, the Nash solution-will
distributive approach is its efficiency: ever be mentioned.

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms
 60 THE ANNALS OF THE AMERICAN ACADEMY

ing emphasizes the differences

authoritarianbe- negotiators merely
tween the negotiator and
prolonghis oppo-
negotiations.
nent, he should attemptSecond,
to interact
the models outlined in this
with his adversary outside that
article may set-
help researchers to deter-
ting, and he should choose a setting
mine whether an actual negotiator is
reciprocating.
that emphasizes similarities, such Theas
theoretical handle
for this
dining together. He might evenproblem
ven- is given by the fact
ture into conversations that, whilesubtly
that, all three parameters shown
and gently, make his in ownequation 1 influence concession
values
seem more acceptable to making,
his only one influences recipro-
adver-
saries. For example, if cation: the reciprocation
the U.S. rep- parameter k.
resentative were to tell an anecdote What statistical methods can be used
that shows how obsessed Americans to determine the size of this parame-
ter? What phrases in the transcripts
are with having access to all secrets,
he might make his Chinese opponentof a meeting-such as saying, in ef-
fect,
more willing and able to explain to "I want to reciprocate"-can be
his power brokers why Americansused
in- to flag reciprocations?
sist on monitoring compliance to Third,
an if we succeed in creating
agreement. even better models of negotiation
than those discussed here, we may
start asking questions that otherwise
Future research
would not be asked. If, for example,
The internal dynamics of negotia-it is true that the search process usu-
ally does not occur until the middle
tion suggest several lines of inquiry.
stage, how would the negotiations be
First, we may explore the external
affected if searching were to occur at
conditions that influence the process.
the very beginning? Is it realistic to
Once such conditions are identified,26
assume that unilateral concession
we may inquire which parameters of
making occurs primarily in the first
the model they influence. For exam-
stage and that it occurs as a result of
ple, if we found that coming from an
anticipating stalemates-or is it
authoritarian culture made negotia- more realistic to assume that unilat-
tors less likely to initiate conces-
eral concession making occurs only
sions-that is, that it decreased coef-
when a stalemate actually occurs?
ficient a of equation 1-then, given
Finally, we may inquire into the
equation 2, we might conclude that
role that search behavior plays with
authoritarian negotiators also tend
respect to the durability of agree-
to have a higher bottom line, D*. If,
ments: does the friendliness gener-
on the other hand, authoritarianism
ated between negotiators who engage
made a negotiator less likely to recip-
in joint search spread to their con-
rocate-that is, if it decreased pa- stituents, thus making the resulting
rameter k-we would conclude that
agreements more binding? Inquiries
such as these will bring further in-
sights into the causes
26. Druckman, "Situational Levers of Nego- and conse-
tiating Flexibility." quences of flexibility.

This content downloaded from 45.32.171.230 on Mon, 30 Apr 2018 21:14:52 UTC
All use subject to http://about.jstor.org/terms