INTIMIDATION: LINKING NEGOTIATION AND CONFLICT

• DOI: http://doi.org/10.1111/iere.12398
• Published date: 01 November 2019
• Date: 01 November 2019

INTERNATIONAL ECONOMIC REVIEW

Vol. 60, No. 4, November 2019 DOI: 10.1111/iere.12398

INTIMIDATION: LINKING NEGOTIATION AND CONFLICT∗

BYSAMBUDDHA GHOSH,GABRIELE GRATTON,AND CAIXIA SHEN 1

Shanghai University of Finance and Economics, China; UNSW Sydney, Australia; Zhejiang

University of Finance and Economics, China, and SIBA,SUFE, China

A challenger wants a resource initially held by a defender, who can negotiate a settlement by offering to

share the resource. If Challenger rejects, conflict ensues. During conflict, each player could be a tough type for

whom fighting is costless. Therefore, nonconcession intimidates the opponent into conceding. Unlike in models

where negotiations happen in the shadow of exogenously specified conflicts, offers made during negotiations

determine how conflict unfolds if negotiations fail. In turn, how conflict is expected to unfold determines the

players’ negotiating positions. In equilibrium, negotiations always fail with positive probability, even if players

face a high cost of conflict. Allowing multiple offers leads to brinkmanship—the only acceptable offer is the one

made when conflict is imminent. If negotiations fail, conflict is prolonged and not duration dependent.

1. INTRODUCTION

Interstate conflicts begin when negotiations end. But why do negotiations fail to prevent

conflict even when compromise solutions are available, commitment is possible, and conflict

is likely to be long and painful?2In the crisis bargaining literature, two parties negotiate in

the shadow of an exogenously given conflict.3If the parties’ costs of fighting the conflict are

private information, then each party has an incentive to misrepresent its real cost so as to build

a “reputation.” This process leads to prolonged crises and, possibly, to war. What drives the

choices made during negotiations is the type of conflict the parties expect, its expected length,

and the probability of ultimate victory of each party.

However, how conflict unfolds (its length and the probability of ultimate victory of each party)

is also determined by past negotiations. In fact, by rejecting a generous ultimatum offer, one

party may signal that its cost of fighting is low, which, in turn, makes its opponent more cautious

once the conflict begins. In this article, we argue that this possibility of signaling strength by

rejecting a generous offer weakens the effectiveness of negotiations. In particular, it induces the

parties to deliberately make meager offers that are rejected with strictly positive probability. For

example, the Rambouillet Agreement offered by NATO to Yugoslavia before the onset of the

∗Manuscript received February 2017; revised January 2019.

1We are grateful to three anonymous referees and Masaki Aoyagi for numerous detailed suggestions. We also

thank Gaurav Aryal, Laurent Bouton, Alessandra Casella, Micael Castanheira, Jimmy Chan, Pauline Grosjean, Anton

Kolotilin, Hongyi Li, Qingmin Liu, Youming Liu, Massimo Morelli, Santiago Oliveros, Carlos Pimienta, Ronny Razin,

Santiago Sanchez-Pages, Bal´

azs Szentes, John Tang, and Adam Wong. Ghosh is especially grateful to Costas Cavounidis

and Juan Ortner for extended discussions, and to the National Science Foundation of China (project no. 71850410543)

for financial support. We thank seminar participants at the Australian National University, Boston University, Columbia

University, European University Institute, Georgetown University, University of Montreal, the 2014 ASSA Meetings,

the 2014 Econometric Society Australasian Meetings, the 2013 Australasian Economic Theory Workshop, and the

2014 SAET Conference. Barton Lee provided excellent research assistance. Shen thanks China Institute of Regulation

Research, Zhejiang University of Finance and Economics (ZUFE), Hangzhou. Please address correspondence to:

Caixia Shen, The New Type Key Think Tank of Zhejiang Province “China Research Institute of Regulation and Public

Policy,” Zhejiang University of Finance and Economics, Hangzhou, China. E-mail: shencaixia@gmail.com.

2In a seminal paper, Fearon (1995) poses this as a “central puzzle” that rationalist explanations fail to solve (see

also Reiter, 2003). The key word “rationalist” rules out explanations where the parties to the conflict are entirely or

largely irrational.

3Notable examples include Esteban and S´

akovics (2008), Fearon (1994), ¨

Ozyurt (2014), Powell (1996, 2004), and

Sechser (2010).

1589

C

(2019) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

1590 GHOSH,GRATTON,AND SHEN

Kosovo War was described as “a provocation, an excuse to start bombing”4that “deliberately

set the bar higher than the Serbs could accept.”5

To understand this two-way feedback relation between negotiation and conflict, we develop

a model where negotiation and conflict are interlinked. In our model, Challenger (she) and

Defender (he) want a resource that yields flow utility. Defender, who initially holds the resource,

can try to negotiate peace by offering a share to Challenger. If all offers are rejected, conflict

begins. In each period of conflict, Challenger chooses whether to attack. If she attacks, Defender

chooses whether to concede the resource to Challenger. If Defender does not concede, conflict

moves to the following period. Attacks are normally costly for both Challenger and Defender,

but with an arbitrarily small probability either party is tough—does not experience the cost

of attacks.

We fully characterize the set of equilibria of this model when the interval between periods

is sufficiently small. As Challenger privately knows whether or not she is tough, her attacks

intimidate Defender into thinking that attacks will continue unless he concedes. Similarly, as

Defender privately knows whether or not he is tough, his nonconcession intimidates Challenger

into thinking that further costly attacks will be useless. This logic of intimidation draws on the

literature on reputation building in bargaining and wars of attrition (Chatterjee and Samuelson,

1988; Ponsat´

ı and S´

akovics, 1995; Abreu and Gul, 2000). The small uncertainty about the parties’

payoffs is magnified by equilibrium play into a significant force that protracts conflict.

The same logic of intimidation also explains why the negotiation cannot succeed with cer-

tainty: Even if Defender can make offers to Challenger before conflict begins, equilibrium offers

are rejected with strictly positive probability by normal (i.e., not tough) Challenger. One might

imagine that the negotiation fails because Defender is afraid to reveal whether he is tough or

normal. We show that this intuition is incomplete by focusing on the case of Defender being

uninformed, in the sense that he does not know his cost of pursuing conflict when he makes

an offer, but will only discover it if and when Challenger first attacks. Even in this case where

offers do not reveal whether Defender is tough or normal, the opportunity to make an offer is

a double-edged sword for Defender. On the one hand, higher offers are better for Defender

because they have a higher probability of being accepted. Indeed, if beliefs were held fixed

both before and after the negotiation, Defender could completely avoid conflict with normal

Challenger by offering slightly more than her expected value of entering conflict. On the other

hand, a generous offer that has a high probability of being accepted increases Challenger’s ex-

pected payoff from conflict because it increases Defender’s belief that Challenger is tough if the

negotiation fails. Therefore, instead of deterring conflict, more generous offers may encourage

Challenger to seek conflict. We show that, in equilibrium, Defender always makes an offer that

is both accepted and rejected with strictly positive probability by normal Challenger. Therefore,

conflict begins with positive probability even if Challenger is normal.

This detrimental effect of generous offers is particularly evident when Defender has multiple

chances to make offers. Indeed, suppose that before the last round, Defender makes an offer

that normal Challenger accepts with positive probability. Then, Challenger strictly prefers to

reject it, as this signals she is tough, thus intimidating Defender into making an even more

generous offer in the last round. In other words, offers that could be successful (but were not)

make subsequent offers more costly. We show that in equilibrium, all offers, except the one

made in the last round, have vanishingly small probabilities of acceptance. Therefore, long

negotiations resolve in brinkmanship: The parties make no progress toward a peaceful solution

up until the last opportunity to negotiate.

This brinkmanship result provides an explanation for negotiation in the shadow of conflict not

making any progress until the very last chance. A possible example is the negotiations leading

up to the Treaty of Porthsmouth, which subsequently gained President Theodore Roosevelt the

Nobel Peace Prize, in the Russo-Japanese War (1904–1905). Negotiations for the treaty were

4Henry Kissinger, Daily Telegraph,June 28, 1999.

5George Kenney, “Rolling Thunder: The Rerun,” The Nation, May 27, 1999.