bates analytic narratives_intro - cap 5

7/28/2019 Bates Analytic Narratives_Intro - Cap 5

http://slidepdf.com/reader/full/bates-analytic-narrativesintro-cap-5 1/44



ANALYTIC NARATIVS



ANALYTIC NARRTIVES

obert H. Bates

vner Gre

Margaret evi

Jean-aurent osenthal

Bar . Weingast

UV PRINCETON, NEW JERSEY



py © 1998 by Pncen Unvey PePublished by Princeton University Press, 1 William Seet,

Princeton, New Jersey 0850

the United Kngdom: Princeton University Press, Chichester, West Sussex

lRights Reserved

Lby f ne n-n-Pbcn

Analytic narratives I Rober H. Bates ... [et al.].

p. m.

Includes bibliographical references and index.

ISBN 0-691-00128-6 (cloth a paper). ISBN 0-691-00129- (pbk. a

paper)

1 Political stability. 2.Economic history. 3.Inteational relations. .Game

theory. 5.Rational choice theory. I.Bates, Rober H

JC330 .2.A45 1998

320.1'1c21

This book has been composed in Times Roman

Princeton University Press books re printed

on acid-free paper and meet the guidelines

for permanence and durability of the Committeeon Production Guidelines for Book Longevity

of the Council on Library Resources

http/ /pup.prince ton.edu

Printed in the United States of America

10 9 8 7 6 5 3 2

10 9 8 7 6 5 4 3 2 1 (Pbk.)

� if1�i

I

0C

3JO.2

-A�)

1.92

4li LJt

7 Il

98-962

To Dougass North__MENTOR AND FREND



Contents__

Akgm

Introduction

One

Self-Enforcing Political Systems and Economic Growth:

Late Medieval Genoa

G

Two

The Political Economy of Absolutism Reconsidered

Ja-La Ra

Three

Conscription: The Price of Citizenship

Maa L

Four

Political Stability and Civil Wa: Institutions, Commitment, and

ix

3

23

64

109

American Democracy 148

Ba R. Wa

Five

The Inteational Coee Organization: An Inteational Institution 194

RbH. Ba

Conclusion 231

App 239

x 243



Acknowledgments__

WITH nancial support from the National Science Foundation (grant SES-9022192), the Center for Avanced Study in the Behavioral Sciences sponsorethis project, enabling us to work together for the acaemic year 1993-94. n themongs, each of us labore on his or her own project; i the afteoons, wemet, seeking to ientify an elaborate upon the intersection of three els: economics, economic history, an political science. Our iscussions increasingly centere on political economy, institutions, an the uses of game theory. hisbook is the result.

Discussions constitutes too neutral a wor. Explorations, ebates, arguments: these hit closer to the mark. need, by the en of the year, we haachieved no consensus. We herefore continue to meet and to "iscuss for twomore years, ismembering and reassembling rafts of the manuscript. DavidFeatherman employe the Presiential Fun f the Social Science ResearchCouncil to defer a portion of the costs of thes meetings. We also receivesupport from our epartments at Stanfor an arvar Universities an theUniversities of Washington an Califoa an from the arvar nstitute for nteational Development.

Once we sense convergence, we sought to raw others into our ebates. A panel at the1996annual meetings of the American Political Science Association provie one forum. A workshop sponsore by the Weatherhea Center for nteational Aairs at arvar University provie another. n our letter of invitation to this workshop, we stressed that "goo friens can be good critics.Our very goo friens inclue John Alrich, Charles Cameron, Gary Cox,Jorge Donguez, James Fearon, Je Frieen, Peter all, ra Katznelson,Gry King, Egar Kiser, Davi Laitin, Mark Lichbach, an Lustick, Lisa Martin, Ronal Rogowski, Fritz Scharpf, Kenneth Shepsle, hea Skocpol,Steven Solnick, and Margaret Somers. Alison Alter, Smita Singh, an Jeremy Weinstein provie valuable research assistance Kenneth Schultz and Richarcker directe us to several relevant literatures David Collier, Mihir Desai,Georey Garrett, Roberta Gatti, an Leonar Wantchekon provie valuablecommentary.

Besies expressing our gratitue to Philip Converse an Robert Scott of the Center for Avanced Study, we also thank the skilled an supportivesta an the other members of the Class of '94. We oer special thaks toJorge Domnguez of the Weatherhea Center for nteational Aairs, arvardUniversity, whose sponsorship of the workshop contribute so much to theevolution of this project. Robert, Margaret, and JeanLaurent wish to thank the



X ACKNOWLEDGMENTS

Weingast and Greif famlies, and especially Esti and Susy, for welcoming us to

Stanford, and Margaret Bates, Bob Kaplan, and Paula Scott for tolerating our

periodic joueys to Palo Alto.

ANALYTIC NARRATIVES



Introduction________________

THIS VOLUME addresses issues of substance and method Focusing on institutions, the chapters explore the sources of political order; the origins of conict,

both inteat ional and domestic; and the interaction between inteational and

domestic political economy They also raise methodological issues, particularlyconceing the use of game theory in the study of political economy n doing

so, they advance a method of analysis that we call "analytic narratives

The issues we address, both substantive and methodological, cut across eldsof scholarship tudents of development join economic historians in seekng the

political foundations of growth (World Bank 1997). nstitions increasingly preoccupy not only sociologists and political scientists but also students of industrial organization (all and Taylor 1996). The use of game theory hasspread from econocs to poltical science and into anthropolog as well Thechapters in ths volume thus engage the conces- of many disciplines and should

therefore command a broad audience.

The case studies in this volume e motivated by a desire to account for particular events and outcomes For those coitted to the close analysis of

particula r events, the chapters seek to demonstra te the benets to be gained fromthe systematic use of theory For those committed to the development and use

of theory, we seek to demonstrate the retus to be reaped from a close dialoguewith case materials The phrase ny nv captures our conviction that

theory linked to data is more powerl than either data or theory alone

When we refer to theory, we refer to rational choice theory and, most oen,to the theory of games Our advocacy could apply to other forms of theory aswell, however A range of models could serv e as the basis of analytic narratives:those derived from the new institutionalism (all and Taylor 1996) or from

analytic Marxism (Przeworski 1985; Roemer 1986), for example Because of our understanding of institutions and the sources of their power over collectivelifeand because four of the ve authors in this volume mshal game theoretic

reasoning!we focus on the strengths a nd liitations of that genre

Subs Thms

Politcal Order

Contemporary events remind us that stability in political communities cannot

be assumed; it must be a chieved The chapters in this volume therefore retuto one of the classic themes in the social sciences: the sources of political order



4 INTRODUCTION

Avner Greif, himself an economist, joins anthropologists, such as Gluckman(65, Bailey (6, and Colson (4, in examining the construction of polities from famly systems. Much as in contemporary Africa, the Caucasus, andSouth Asia, in twelfth-century Genoa, Greif reveals, powerl clans competedfor wealth and sought to use public power to advance their pvate fortunes.Competition among them sparked civil wars. Greif explores the way in which a political institution, the podesta maintained peace between clans. In doing so,

he examines the political foundations for economic growth in Genoa, exploringhow it was able to extend its political inuence and commercial hegemonytoughout the Mediterranean.

Genoa was a maritime city-state; the United States is a continentwidefederation. As Genoa sought to dominate the markets of the Mediterranean, sotoo did the United States seek to incorporate the territories of North America.And just as Genoa succeeded, so too did the United States, resulting in thecreation of the largest common market on earth. Addressing the politics of the

nineteenth-centry United States, Bry Weingast plumbs a theme previouslyexplored most closely by Karl Deutsch 5 and William Riker 64:

the creation of federal systems. t was Rker who stressed the importanceof collective gains, such as the occupation of territory, in motivating theconstruction of federal systems; it was Deutsch who highlighted the role of beliefs, stressing that centers of power can combine only when convinced of

the peaceful intentions of their partners. Weingast uses his study of the UnitedStates to examine the institutional foundations for federalism. Exploring thewell-known balance rule and using a form of strategic reasoning, Weingast demonstrates how the rule facilitated the achievement of political stability andterritorial expansion.1 Just as the podesta kept peace by enhancing the security

of theatened clans in Genoa, the balance rule prevented civil war by protectingthe security of regional blocs.

Bates also addresses the sources of order, this time at the inteational level.Joining regime theorists (Krasner 82 Keohane 84, Bates explores theway in which sovereign states forged a binding agreement at the global level.Focusing on the inteational market for coee, he examines the origins, operations, and impact of the Inteational Coee Organization (ICO), a politicalinstitution that regulated global trade in coee. By goveng the market, Batesargues, the ICO restrained competition among producers, thereby securing for

them collective benets higher coee pices and the redistribution of incomefrom consumers to producers of coee (see also Bates

In exploring the sources of order in clan-based societies, the foundationsof federalism, and the origins of inteational regimes, these chapters addresssubjects examned at length by others. Distinguishing our work is a central

1The rle required that when a nonslave state was admitted to the Union, it was to be balancedby the admission of a slave state.

INTRODUCTION 5

conce with institutionsfor example, the podesta the balance rle, and theICOand our approach to their analysis. We join those, such as Huntington68, who place institutions at the core of the analysis of political order.However, we dier from many in our understanding of them. Institutions,these chapters argue, do not impose constraints; the order they provide emergesendogenously. Institutions rest upon credible promises, f either reward or

punishment. They therefore can, and should, be analyzed as the equilibria of extensive-form gamesa theme to which we will shortly retu.

Inr Rln

The chapters in this volume also address the stdy of inteational relations. particular, they focus on the relationship between the pursuit of inteationalsecurity and the performance of the domestic political economy. 2

The chapter by Bates focuses on inteational politics. Twice, he documents, politcal organizations replaced market forces in the inteational coee market:once during World War and again dung th cold war. On both occasions,the United States, the largest consumer of coeein the world, cooperated with producers to stabilize market prices. During times of global tension, he argues,economc agents were able to provide political services to the guardians of inteational security. When the guardians of inteational security no longer

required the services of the economic agents, t he guardians defected from t heagreement; in so doing, they precipitated the collapse of the political institutionthat undeinned the cartel. By focusing on the relationship between securityinterests and trade alliances, Bates contri butes to a theme explored perhaps most directly by the contemporary "neorealists in the eld of inteational relations(Baldwin 3 Gowa 4). The inteational political system is characterized

by conict, they contend; the pursuit of security shapes the behavior of nations.Large powers, they argue, nd it in their private interests to provide collective

benets, including institutions at the global level. The chapter both supports andcritiques these arguments, and in so doing engages a well-established literature.

Central to Bates's analysis is the impact of inteational events on domestic politics: clely the theat of nazism and communism rendered securityconsiderations more persuasive, and security specialists more powerl, in thedomestic policy process. Levi amplies and extends this theme. Exploring

broad but nuanced comparisons, she investigates the ways in which variousstates at various times have levied manpower for their litary forces. Bystudying the comparative politics of conscription, Levi highlights the domestictransformations precipitated by war.

2Note the important symposium that addresses many of the following issues: Elman and Elman(1997); see especially Levy (1997.



10 INTRODUCTION

behavior of the other players and the expectation that the game will not actually reach these nodes (i.e., that it will remain on the path of play). A subgame perfect equilibrium imposes an additional restriction.It requires tha in everysubgame, including those never actually re ached (i.e., those that lie o the pathof play), the player's actions constitute a best response to the actions taken byothers in that subgame.The notion of subgame-perfect equilibrium thus requiresthat threats or promises be credible.The players remain on the equilibrium path

of play because it would be in the best interess of others to fulll their threatsor promises should they say o it.Sub game-perfect equilibria thus constiutea subset of Nash equilibria; they exclude those strategy combinations that form

Nash equilibria that include threats or promises that re not credible.Credible treats of punishment thus constitute an essential feature of sub

game-perfect equilibria. A player refrains from straying o the pa because

he or she anticipates being made worse o, given the expected response of other players.Fears of the consequences that await those who stray encourageadherence to equilibrium behavior.Credible teats therefore play a signicant role in generating institutionalized pattes of behavior.

The chapters in this volume thus address the sources of political order, war

fare and interaction between the inteational and the domestic in the politicaleco�omy of nations.In doing so, they explore the interplay between politics andeconomics and stress the role of instituions.They consider ndamental themesin mode political economy. What is distinctive is their use of a particular methodology, analytic narrative, to blend rational choice analysis and narrationinto the study of institutions and of their impact upon political and economic

behavior.

Analytic Narratives

We call our approach analytic narrative because it combines analytic tools that are commonly employed in economics and political science with the narrativeform, which is more commonly employed in history.Our a proach i nativ�; 1it pays close attention to stories, accounts, and context.It 1s analytc m that 1t

\extracts explicit and formal lines of reasoning, which facilitate both expositionJand explanation.

The chapters in this volume constitute in-depth investigations of events that transpired in paricular periods and settings.I this sense, the volume contributes to the ideographic tradition in the social sciences (McDonald 996a,b).The chapters trace the behavior of particular actors, clarify sequences, describestrucures, and explore pattes of interaction.Thus they contribute to the "historical tu in the social sciences, in the words of Terrence McDonald, whichis based "rst and foremost on a commitment ...to producing understandingvia richness, texture, and detail (996a:0, quoting from Ortner 996).

INTRODUCTION 11

In addition to probing the narrative record, the chapters seek "parsimony, renement, and (in the sense used by mathematicians) elegancethe veryqualiies that McDonald explicitly counteoises to those charactesic of the "historic tu ( 996a:0, quoting from Ortner 996). Each constructs,employs, or appeals to a foal model.Greif, Rosenthal, and Weingast constrct their models; Bates takes existing models "o the shelf; and Levi mobilizesformal lines of reasoning.

William Riker, who pioneered the application of rational choice to the stdy

of politics, tued to the theory of games in order to move, he stated, from "wisdom to "political science ( 962:viii), by which he meant a body of relatedgeneralizations "deduced rom one set of axioms, which ... [oer] a coherent theoretical model (962:3).Riker saw in rational choice theory the promise of a universal approach to the social sciences, capable of yielding general laws of

political behavior.Critics of rational choice likewise point to the abstract andlogical character of game theory, but do so in order to condemn it.The failures of

rational choice theory, Green and Shapiro declare, are "rooed in the aspiration fof rational cho ice theorists to come up with universal th eories of politic s. The result, they argue, is a preoccupation with theor development, accompanied by a striing "paucity of empical applications (994:x).Research "becomestheory driven rather than problem driven; its purpose is "to save or vindicate

some variant of rational choice theory, rather than to account for ... political phenomena (994:6).Clely the chapters in this book are problem driven, not theory driven; they

are motivated by a desire to account for particular events or outcomes. Theyare devoted t o the exploratio n of cases, not to th e elaboration of theory. Inthese ways, they counter the charges raised by Green and Shapiro.By the sametoken neither do the chapters confo to Riker's vision of the role of theory.Although informed by deductive reasoning, the chapters themselves seek nouniversal laws of human behavior.

ere possible, we make use of formal arguments.I pricular, we analyzegames, since we nd them usel in order to create and evaluate explanationsof paricular outcomes. We identi agents; some are individuals, b ut others rare collective act ors, such as elites, nations, electorates, or legis latures. By

reading documents, l aboring trough archives, interviewing, and surveyingthe secondary literature, we seek to understand the actors' preferences, heir perceptions, their evaluation of alteatives, the information they possess, theexpectations they form, the strategies they adopt, and the consraints that liit their actions. We then seek to p iece together the story that accounts for theoutcome of interest: the breakdown of order, he maintenance of peace, thedecision to ght or collude.We thus do not provide explanations by subsuming

Indeed, as with those of McDonald we consider Rer's claims to be based on an "over-

condent and naive vision of the sciences (McDonald 1996a:5).



12 INTRODUCTION

cases under covering laws," in the sense of Hempel (1965). Rather we seek Jto account for outcomes by identifying and exploring the mechanisms thatgenerate them. We seek to cut deeply into the specics of a time and place, andto locate and trace the processes that generate the outcome of interest.

The chapters thus build narratives. But the narratives are analyc narratives.By modeling the processes that produced the outcomes, we seek to capturethe essence of stories. Should we possess a valid representation of the story,then the equilibum of the model should imply the outcome we describeandseek to explain. Our use of rational choice and game theory transforms the

lnarratives into analytic narratives. Our approach therefore occupies a complex J

middle ground between ideographic and nomothetic reasoning.12 /

At least since the writings of Thucydides (1951), the narrative form has been the donant form for explaining human behavior. Recogntion of itslitations, however, led to the adoption of other modes of explanation. Nar ratives, it was recognized, embody explanations. But they often mobilize themythology and hagiography of their times, mixing literary tropes, notions of morality, and causal reasoning in eorts both to justify and to explain (Dray1964; Stone 1979; Danto 1985; White 1987; Fry et al. 1989; Goldthorpe 1991Lustick 1996). Social scientists therefore found it dicult to extract defensible propositions from these complex mixtures.I the face of such diculties, some,

like Robert Fogel, urged the abandonment of narrative accounts of particularevents and the exploration, through statistics, of regular and systematicallygenerated events: births, deaths, and market transactions, for example (Fogeland Elton 1983).

As have others, however, we seek to retu to the rich, qualitative, anddescriptive materials that naatives oer.13 And, as have others, we seek anexplicit and logically rigorous account of the events we describe. Modezation theorists (Lipset 1963), Marxists (Lefebre 1924), world system theorists(Wallerstein 1974), and others (such as the Whig historians, e.g., Trevelyan1938) have composed compelling accounts of qualitative materials. In two

respects, however, we dier in our approach. F irst of all, others sometimes oer metanarratives that document the as-

cendance of political ideas or the emergence of political forces. We pu rsue a

more modest objective. We seek to loca te a nd explore p articulr mechanisms jthat shape the inte rplay between strategic actors and that the reby gene rate )

12As is recognized by Tetlock and Belkin :13See the symposia in Soc Scc Ho (Fall Socooc Mho

Rch 0 (May and Jo of Mhmc Socooy - Relevantconibutions include Aminzade and Somers Others re engaged in projects resem bling ours, for exampl e Abbott Abell and Kiser see so Goldthorpe Also german e are the comme nts of Donald M cClosky 0 For a far less successl eort, seeFerguson

INTRODUCTION 13

Coutcomes.14 Second, most of these literatures are structral: they focus on the

1origins and impact of alignments, cleavages, strctures, and institutions. Ourapproach, by contrast, focuses on choices and decions. It is thus more microthan mac ro in orienta tion. By delti "-p�cf� �echanisms a nd focusingon the determinants and impact of choices, our work diers from that of our predecessors.

However, it is important to clafy and emphasize that we view our approachas a complement to, rather than as a substitute for, structural and macro-levelanalyses. As seen, for example, in the work of Bates, Levi, and Rosenthal, we tooexplore the impact of large-scale political forces: of bipolarity and communsmin the twentieth century, of democratization in the nineteenth, or of absolutismin the Early Mode period. But we focus on the mechansms that translatesuch macrohistorical forces into specic political outcomes. By isolating andunpacking such mechanisms, analytic narratives thus contribute to structuralaccounts. They also draw upon them. The models that we mobilize presumethat some variables are constrained, whereas the values of others can be chosen;they presume that alteatives have been dened, as well as the linkages between actions and outcomes. Analytic narratives and macrostrctural analysestherefore stand as complementary approaches to xplanation.

Building Narratives

When students le research methods, they typically start by leing princi ples o f case selection.15 Like them, we resist this point of departure. As havemany of our stdents, we too have been impelled to do social science by ourfascination with particular cases: World War I, the French Revolution, or theU.S. Civil Wr. In eect, our cases selected us, rather than the other way around.

So compelling do we nd our cases that we wish to ierse ourselves inthem. But we also want to constrct logically persuasive and empirically validaccounts that explain how and why events occurred. In seeking to build suchaccounts, we follow Alexander George in tracing the historical processes thatcharacterized the unfolding of the events of conce (George 1979; George andBennett 1998; see also Lijphrt 1971; Eckstein1975). We identify the actors, the ]decision points they faced, the choices they made, the paths taken and shunned,

14We are of course aware that we may unwittingly be participating in a metanrrative, be itof the extension of rationality or the scientic method. Through vigilance and self-awreness, we

have sought to defend against that possibiity15Collier and Mhoney and King et al dene the state of the rts See the

symposium on this work in Amc Poc Scc Rw (June : -0Przeworski and Teune 0 endures as a major contribution to this literature which of courseextends back a century or more



14 INTRODUCTION

and the manner in which the choices generated events and outcomes. In thewords of Richard Fenno (1990), we "soak and poke. In the phrasing of CliordGeertz (1973), we seek "thick description. In these ways, we respond to thecall of Lawrence Stone (1979) to retu to the narrative tradition or to the urgingof Terrence McDonald (1996b) to take the "historic tu in the s ocial sciences.

But we do not stop there. In an eo to move from apprehension toexplanation, we move from "thick accounts to "thin forms of reasoning.We seek to hghlight and focus upon the logic of the processes that generate the phenomena we study. doing so, we use rational choice theory. We nd game

theoretic models particularly usel ways of exploring the validity of naativeaccounts. The theory is actor-centric; in extensive form, games explicitly take {sequence into account and highlight its signicance for outcomes. They capturethe inuence of history, the impoance of uncertain, and the capaci of peopleto manipulate and strategize, as well as the liits of their ability to do so. Thestructure of extensive-form games thus readily accommodates the narrativeform. To distill the essential properies of a naative, we seek to construct thegame that provides the lik between the proinent featres of the narrative andits outcome.

A narrative possesses a background or setting, a beginning, a sequencei of scenes, and an ending. The construction of an analytic narrative involvesmasteng the elements of the drama. Narratives fascinate, however, in part

because, like dramas, they can be elusive; many possible explanations can exist,and many possible interpretations. How then are we to evaluate our accounts?

Evaluating a Narrative

Once made explicit and recast as a model, the interpretation of the narrative anbe subject to skeptical appraisal. Rendering explanations explicit enables us to

put them at risk. Do their assumptions correspond to what is known? Do their conclusions follow from their preses? Do their implications nd support inthe data? How well do they stand up by compason with other explanations?

And how general is the explanation: does it apply to other, analogous cases? Bycompelling explanations to respond to such challenges, we gain the oppotunityto evaluate them.

To illustrate from the chapters in this volume

1. Do h ssmpos h fcs, s hy kow? Motivated by thisquestion, scholars ight challenge Bates's assessment of the views of the Department of State; Levi's claim about the importance of faess; Rosenthal's portrayal of the preferences of the French elite; or Weingast's account of the

16Milton Friedman (1953) has challenged this criterion. We disagree with his position.

INTRODUCTION 5

beliefs of Southe politicians in the 1850s. Should Bates, Levi, Rosenthal,and Weingast successlly rebut such challenges, ten ther explanations gaingreater credibility.

2. Do cocsos foow fom pmss? Addressing this question involvesapprasing the logic of the model. This is more easily done hen the model isformalized. Note, for example, that the relationship between inteational threat and political cooperation can be more readily checked in Greif's exploration of the politics of Genoa clans than in Bates's exanation of the ICO or in Levi's

analysis of conscription. One need only check Greif's math! Because neither Bates nor Levi provides a mathematical model, their logic is less r eadily subject to scutiny.

s important to point out that the issue is not ndamentally one of mathematics, however; it is one of correct reasoning. The explanations providedby the chapters in this volume employ the reasoning of rational choice theoy. Toaccount for outcomes, game theorists seek to explore whether the choices willoccur in equilibrium. Anticipating the behavior of others, would a rational actor behave in the way that generated that outcome? Would his or her expectationsof the behavior of others be conrmed by their atual choices? Do the choices

represent best responses, given the alteatives and information available at thetime of decision? To satis, the logic must yield implications that are formally

true but also consistent with the reasoning that underlies the theory. It is this reasoning that transfos narratives into analytic narratives.

It is important to sess that such explanations cannot be based upon theoryalone. Theory is not adequate in itself. Repeated games, for example, can yielda multiplicity of equilibria. To explain why one outcome occurred rather thananother, the theorist must ground his or her explanaton in empirical materials,seeking the forces that shaped the selection of a particular equilibrium. Theexistence of multiple equilibria thus compels the analyst to combine theory andempirics. Deductive theory thus becomes an engine of empirical discovery.

Explicit attention to theory strengthens explanation. An explicit appeal tothe narrative is required for the completion of that explanation. In analyticnarratives, the empirical content of the narrative is as important as the logicalstrcture of the model.

3. Do s mpcos d coo h d? In addressing thisquestion, the contributors again came to recognize the role of theoy as a toolfor empirical discovery. For in the process of constructing explanations, theywere orced to recognize the existence and importance of factors that hithertohad escaped their attention. search of explanations consistent with ctel

17The famed folk theorem states that for a suitable rate of time preference, any outcome can besecured as a subgame-perfect equilibrium in an innitely repeated game or in a nitely repeatedgame of uncertain duration (Fudenberg and Maskin 1986)



16 INTRODUCTION

theory, Bates, for example, was compelled to recognize the signicance of the long-term contracts between producers and roasting s. To determine

hy nonelite landholders rst preferred and then witheld support for a policyof purchased exemptions, Levi discovered the existence of complex forms of insurance against losses arising om military service in rural France and the practice of substitution in the United States. And in search of an explanat ion of the tiing of Genoa's political reform and economic expansion, Greif gainedan appreciation of the signicance of te threat posed by the emperors of teHoly Roman Empire. The search for teoreticall consistent explanations tusanimated the process of empirical discovery.

Note that in responding to this third callenge, the authors again blur the conventional distinction between deduction and induction. They also depart from conventional notions of ypothesis testing. Te autors deve implicationsfrom theory; but when the case materials do not conrm their expectations, theydo not respond by rejecting their models. Rather they respond by reformulatingthem and by altering the way in wich tey tink about te problem. Thus Batesmoves from one form of cartel model (that of te Chicago school) to another (that of Oliver Williamson), and Greif incorporates into his model the impact of issing, foreign polic variables. As recognized by Kuhn (1962) and Lakatos(1970), the dominant response to disconrmation is thus reformulation, not

falsication.In responding to challenge (2)Do conclusions follow from premises?e stressed that the theory places constraints upon the narrative; the account isconstrained by the logic of the theory. I responding to challenge (3)Do eimplications of the theory nd conrmation in the data?we nd the teory being shaped by the case materials. On the one and, the response highlightsonce again the intelay between deduction and empiricism characteristic of analytic narratives (George and Bennett 1998). On the oter, it underscoresthe contrast between analytic narratives and interetation. Like all narratives,analytic narratives ae grounded on empirical detail; like all narratives, hey provide interpretations of the data. But being based on rigorous deductive reasoning as well as close attention to empirical detail, analytic narratives are

tightly constrained. They are more vulnerable than other forms of interpretation.They are disciplined by both logic and the empirical record.Looing back, we can see that the construction of analytic narratives is

an iterative process, resembling George's metod of process acing (George1979; George and Bennett 1998) Like George, we seek to convert descripive

histocal" accounts into analytic ones" that are couced in theoretically relevant" language (George and Bennett 1998: 14) Like George, we move backand forth between interpretation and case materials, modifying te explanationin light of the data, which itself is viewed in new ways, given our evolvingunderstanding. What dierentiates our approach from the method of processtracing is a greater emphasis on theory.

INTRODUCTION 17

Initially the theory is formed from the data; it is selected because it appearsto oer a good t. Rendered explicit, the theory ten becomes vulnerable; it can

be subject both to logical appraisal and to empirical testing. Its logic, moreover, renders it a source of new insights, leading to the gathering of new data and

placing the theory at rther risk. The richer the teoy, te greater the number of testable implications, and tus the greater te risk of beig found wanting.

In te end, we, like George, acheve a match between theory and casematerials (but see below). Unlike George, we acieve that outcome in a waythat renders the process not just a means of tracing but raher a means of testing.Moreover, the nished product does not just report our nal account; rather it shows our work," rendering our process more transparent and our conclusionsmore open to scrutiny.

We stop iterating when we have run out of testable implications.A implication of our method is that, in te last iteration, we are left uncertain; ironically

perhaps, we are more cetain when teories fail than when tey t the casematerials.18 To reduce ther the level of uncetainty, we therefore seek furter

challenges and pose two additional questions.

4 How well does the theo stand up by comparon with other explanations?

Tus far we have discussed our response to failures of prediction. But oware we to respond to models that appear to t? The problem is troubling

because several models can t he data generated by a particular case. ChrlesTaylor (1985) has stessed tat multiple interpretations can exist for a givenobservation, each inteally coherent and each tting the facts as they are own(but see Ferejohn 1991) So too has Kun (1957)

Selecting among theories is easier te more explicit and logically coherent they ae. Explicit theories render anomalies easier to detect. As seen in chapter 5, for example, e contrast between theory and outcome, strkly highlighted because of the explicit nature of the theory, impels Bates to assess a vaetyof alteative theoies. Thus our preference for analtic narratives, which rest upon explicit and logically rigorous models.

As already discussed, however, it is naive to believe that the answer liesin falsication. Even with explicit and logically rigorous accounts, multiple

explanations will persist; as tey ae observationally equivalent, we will not beable to choose among them. One response is that of Greif. He oers a politicaleconomic rather than a culural account of cooperation in Genoa because it can account for periods of cooperation and discord, whch, e notes, a culturalaccount cannot. Greif tus cooses the teory tat is the more powerl.

Levi and Weingast follow an alteative strateg. Rather han arbitratingamong alteatives, they instead subsume them. They endorse their explanations because of heir ability to reconcile alteatives. Levi nds that standard

18 v f G Kg -



18 INTRODUCTION

rationales for the end of "buying outone economic, the other political provide only partial explanation. She therefore oers a synthetic model, in

which the growth of the electorate is combined with variation in the opportunity

costs of militry service to account for changes in the means of military recruit

ment. Weingast's approach enabes him to explain why traditional historians

have pointed to savery as the dominant national issue of mid-nineteenth

century U.S. politics, while the new political histori ans consider the issue to play

but a inor roe. Levi and Weingast thus respond to the existence of multiple

expanations by seeking a more general explanatory framework, capable of

subsuing the competing approaches.

5. How gal i th xplaatio? Do it apply to oth a? This last

question merits extended discussion. It is the stu of which conclusions are

made. We therefore retu to it, following presentation of the chapters.

References

Abbott, Andrew. 1993 "Measure for Measure: Abell's Nrative Methods Joual of Mathematical Sociology 18 (2-3): 203-14

Abell, Peter 1993 "Some Aspects of Narrative Method. Joual of Mathematical

Sociology 18 (2-3): 93-134

Aldrich, John H, and Kenneth A. Shepsle. 1997 "Explaining Institutional Change:

Soaking, Poking, and Modeling in the US. Congress. Paper presented at a conference

i honor of Richard Fenno, University of Rochester

Almond, Gabriel. 1988 "Separate Tables: Schools and Sects in Political Science. S.:Political Science and Politics 21: 828-42

- 1990 A Discipline Divided Newbury Park, Calif: Sage.

Ainzade, Ronald. 1993 Ballots and Barricades. Princeton, N.J.: Princeton Unversity

Press.

Bailey, F G. 1969 Strtagems and Spoils. Oxford: Basil Blackwell.

Baldwin, David, ed. 1993 Neorealism and Neoliberlism: The Contempo Debate.New York: Columbia Unversity Press

Bates, Robert H. 1997 Open-Economy Politics: The Political Economy of the World

Coee Tade.Princeton, N.J: Princeton University Press

Cameron, David R. 1978 "The Expansion of the Public Economy: A Comparative

Analysis American Political Science Review 72: 143-61Collier, David, and James Mahoney 1996 "Insights and Pitfalls: Selection Bias in

Qualitative Research World Politics 94 (2): 6-91Colson, Elizabeth 1974 Trdition and Contct. Chicago: Aldine.

Danto, Arthur. 198 Narratives and Knowledge. New York: Columbia University Press

Deutsch, Karl 197 Political Communi and the North Atlantic Area. Princeton, NJ:Princeton University Press.

Dixit, Avinash, and Barry Nailbu. 1991 Thinking Strategically. New York: Norton

Doyle, Michael W 1983 "Kant, Liberal Legacies, and Foreign Aairs, Parts 1 and 2Philosophy and Public Aairs 12: 20-4, 323-3

INTRODUCTION 19

Dray, William. 1964 "Conicting Interpretations in History: The Case of the English

Civil War. In Heeneuti�s: Questions and Prospects, edited by Gary Shapiro and

Alan Sica. Amherst: University of Massachusetts Press

Eckstein, Harry. 197 "Case Study and Theory in Political Science. In Handbook of Political Science, vol 7 edited by Fred I. Greenstein and Nelson W. Polsby Reading,

Mass: Addison-Wesley.

Elman, Colin, and Miriam Fendius Elman, eds. 1997 "Symposium on Diplomatic

History and Inteational Relations. Inteational Securiy 22 (1): 3-8Evans, Peter, Harold Jacobson, and Robert Putnam, eds 1993 Double Edged Diplo

macy: Inteational Bargaining and Domestic Politics. Berkeley: University of Califoia Press.

Fenno, Richard 1990 Watching Politicians: Essays on Participant Obseration. Berke

ley: Institute of Govemental Studies, University of Califoia

Ferejohn, John. 1991 "Rationality and Interpretation: Parliamentary Elections in Early

Stuart England. In The Economic Approach to Politics, edited by Kristin Monroe.

New York: HarperCollins.

Ferguson, Niall, ed 1996 Vrtual Histo. London: Picador.

Findlay, Ronald. 1991 "The New Political Economy: Its Explanatory Power for LDCs.

In Politics and Policy Making in Developing Countries, edited by G. M. Meier San

Francisco: Inteational Center for Economic Growt.

Fishlow, Albert, Catherine Gwin, Stephan Haggard, Dani Rodrik, and Robert Wade.

1994 Mircle or Design: Lessons from the East Asian Experience.Washington, DC.:

Overseas Development Council.Fogel, Robert, and G. R. Elton. 1983 Which Road to the Past. New Haven, Conn: Yale

University Press.

Friedman, Milton. 193 "The Methodology of Positive Economics. In his Essays onPositive Economics. Chcago: University of Chicago Press.

Fry, Brian, Eugene Golob, and Richard Vann, eds. 1989 Louis 0. Mink: HistoricalUnderstanding. Ithaca, N.Y: Coell University Press.

Fudenberg, Drew, and Eric Maskin 1986 "The Folk Theorem in Repeated Games with

Discounting or with Incomplete Information Econometrica 4: 33-4Gates, Scott, and Brian Humes 1997 Games, Information, and Politics. Ann Arbor:

University of Michigan Press.

Geertz, Cliord. 1973 Interetation of Cultures. New York: Basic Books

George, Alexander L. 1979 "Case Studies and Theory Development: The Method

of Stuctured, Focused Comparisons. In Diplomacy: New Approaches in Histo,Theo and Policy, edited by Paul Gordon Lauren New York: Free Press.

George, Alexander L, and Andrew Bennett 1998 Case Study and Theo Development.Cambridge, Mass: MIT Press.

Gibbons, Robert. 1992 Game Theo for Applied Economists.Princeton, NJ.: Princeton

University Press

Gluckman, Max 196 Politics, Law and Ritual in Tribal Societies. Oxford: Basil

Blackwell.

Goldthope, John H 1991 "The Uses of History i Sociology British Joual of Sociology 42 (2): 220-2

- 1996 "The Integration of Sociological Research and Theory. Paper resentedat the ICS Inteational Conference, University of Groningen, June



22 ODUO

Stinchcombe, rthur. 1963. Constructing Social Theories. New York: Harcourt Brace

and World.

Stone, Lawrence. 1979. "The Revival of Narrative. Past and Present 85: 3-4.

Taylor, Charles. 1985. Philosophy and the Human Sciences. Cambridge: Cambridge

University Press.

Tetlock, Philip, and Aaron Belkin. 1996. "Counterfactual Thought Experiments in World

Politics. Counteactual Thought Experiments in World Politics, edited by Philip

Tetlock and Aaron Belkin. Princeton, N.J.: Princeton University Press.

Thucydides. 1951. Histo of the Peloponnesian Wa translated by John H. Finley, Jr.

New York: Mode Library.Tirole, Jean. 1988. The Theo of Industrial Organization. Cambridge, Mass.: MIT

Press.

Trevelyan, Macaulay George. 1938. The English Revolution. London: Oxford University

Press.

Trimberger, Kay Ellen. 197. "A Theory of Elite Revolutions. Studies in Comparative

Inteational Development 7 191-07.

VanDamme, Eric. 1987. Stabiliy and Peection of Equilibria. Heidelberg: Springer-

Verlag.

Wallerstein, Immanuel. 1974. Capitalist Agriculture and the Origins of the European

World Economy. New York: Academic Press.

- . 1987. The Content and the Fo. Baltimore: Johns Hopkins University Press.

Williamson, Oliver. 1985. The Economic Institutions of Capitalism. New York: Free

Press.

World B 1993. The East Asian Mircle. New York: Oxford University Press.

- . 1997. World Development Report: The State in a Changng World. New York:

Oxford University Press.

One___Self-Enforcing Political Systems and Economic

Growth: Late Medieval Genoa

VN GF

THIS CHAPER analyzes the political institutions of the commune of Genoa

during the late medieval period. By the end of the thirteenth century, Genoastood second, perhaps, only to Venice among the Italian maritime cities in

terms of wealth and power. In seeking a better understanding of the se of Genoa, the chapter addresses several broader themes, most notab ly the processof state formation, the determinants of political and economic performance,and the politics of resource mobilization and o goal attainment by polit

ical communities. Mobilizing theoretical argument and empirical evidence,the chapter seeks to cut more deeply still, to explore the factors that ren

dered Genoa's political institutions self-enforcing. By exposing these factors,the chapter identies the conditions under which political actors follow the

rules that gove the political order, rather than ignoring them or resorting

to other means, such as violence. It detenes as well the conditions under which the rules will change, leading to the riseor collapseof the political system.

A key nding of this chapter is that the degree to which Genoa's political

system was selfenforcing depended on the degree to which its economic systemgenerated appropriate political incentives. the erly period of Genoa's rise,

economc growth would have eroded, rather than enhanced, the economicfoundation for its maintenance. It was only after a long period of leng and

experimentation, and because of particular historical circumstances, that the

Genoese altered their political system and introduced a poltical innovation

I gratefully acknowledge the support of the National Science Foundation (grants 9009598-01

and 9223974) and the Center for Institutional Reform and the Informal Sector at the University of

Mryland a t College Prk. I have beneted om the comments of James Alt, Masa Aoki, Rober t

Bates, Mrgaret Levi, Paul Milgrom, Boaz Moselle, Jean-Laurent Rosenthal, Barry Weingast, and participants in a joint Harvard-MIT economic historyevelopment semin, an all-deprtment seminr at Tel Aviv University, an all-deptment seminar at Northweste University, and the

Social Science History Workshop at Stanford University. Some of the ideas expressed in this paper

were formulated while I was a fellow at the Center for Advanced Study in the Behavioal Sciences,Stanford, Califoa; the Center's kind hospitality greatly facilitated this study. As always, Debbie

Johnston provided eective and indispensable assistance.



Five______ _The Inteational Cofee Organization:An Inteational Institution

ROBERT H. BATES

CURNTLY worth roughly $10 bilon per year, coee expors stand next tothose of oil as the most valuable from the tropics For many developing nations,coee constitutes a major source of foreign exchange (see table 1)

From 1962 to 1989, the Inteational Coee Organization (ICO) regulatedthe world's exports of coee. he organization was formed under the terms of

the Inteational Coee Agreement, and its members accounted, on average,for 90 percent of the world's total consumption he consuming nations that

remained outside the ICO included the socialist countries of Asia, EasteEurope, and the Soviet nion, which lacked the converible currencies so

desired by tropical exporters, and the Middle Easte countries, which possessedsuch currencies but consumed little coee (see table 2)

he ICO set target prices for coee. In the later years of its existence, the

target lay between $1.20 and $140 a pound he agency then set quotas for coee exports so as to force market pces into the trget range When itsindicator of market prices rose above $140 a pound, quotas were relaxed;

when it fell below $120 a pound, they were tightened At times of extremeincreases i prices, such as after major fr osts in Brazil, quotas were abandoned

altogether, until production retued to normal levels and trading resumedwithin the target range.

Quick calculations highlight the domestic signicance of the ICO Consider,for example, ganda, one of the major suppliers of robusta coee: for the coee

The materials this chapter are drawn om Bates (1997). The chapter was written while I wasat the Center for Advanced Study in the Behavioral Sciences, Stanford, Califoa. I acknowledgethe support of Harvard and Stanford Universities as well as the National Science Foundation forits grants to the Center for Advanced Study in the Behavioral Sciences (SES-9022192) and toRobert H. Bates (SES-8821151). I am also gratel for the comments and cticisms receivedom Mihir Desai, Roberta Gatti, Avner Greif, Margaret Levi, Ronald Rogowski, Jean-LaurentRosenthal, Fritz Schaf, and Bay Weingast and the superb research assistance I have receivedon this project om Laura Alvarez, Rosalba Capote, Carlos Coneras, Amy Curry, Gina Dalma,Catherine Elkins, John Hall, Donald Da-Hsiang Lien, Brian Loynd, Andrew Mason, J. MuthengiMusunza, John Nye, Dixie Reeves, Daniel Resepo, and Michael Thompson.

INI CFF GNIZIN 195

Ll

Percent of Total Exports Composed of Coee Exports i

Selected Counies

Year Brazil Colombia El Salvador Guatemala Kenya Cte d'Ivoire Uganda

0 637 778 88 66 74 00 288

8 776 88 64 4 287

2 738 803 87 76 24 626 28

3 707 826 8 68 23 2 344

4 607 837 876 708 22 64 32 3 83 86 70 3 2 478

6 6 76 77 73 4 637 37

7 608 824 80 722 348 7 44

8 3 84 72 726 33 6 36

ources: Inteational Monetary Fund (1980, 1990, 1992, 1993); Food and Agriculture Organization of the United Nations (198891).

year 1981/82, the ICO granted ganda a quota of roughly 28 million bags(Inteational Coee Organization 1982:136, table VI-2) Had members of

the agency permitted ganda a 5 percent higher quota, the country could haveeed an additional $20 million per year in export markets. en more bags of

coee, shipped out by a gandan peasant, would have generated an additioalincome of over $200, an increase equivalent to the average annual per capta income in ganda at that time

his chapter presents a narrative of the birth, behavior, and collapse of the

ICO then analyzes this narrative, employing a variety of analytic approachesOf particular interest is that several of these approaches fail Either they f �il

to t imporant features of the empirical setting or, to succeed, they requreconditions that are logically inconsistent with their premises From observingthe failure of various analytic approaches, however, we le we are driven

to a new appreciation of the signicance of feares of the case materia �sand to revisions of our approach. he interplay between data and analyss(George 1979) promotes a renement and deepening of our uderstanding of

the political economy of the coee market and of the origins and signicanceof its institutions.

IThroughout, I employ the standardized measure adopted by the ICO, according to which one bag contains 60 kilograms of green coee.

. . . . Based on an average price for robusta of $1.06 a pound. See the pce data conted m Lcht(1993:G.l5).

.Even assuming that taxes, bribes, and ansport consumed over 80 percent of the border pce

costs. A bag of coee weighs 60 kilograms, or 132 pounds. With a price at the border of $1.00 per pound, ten bags would be worth about $1,320. The extent of goveent taxation is documentedSears et al. (1979).



'0

-

0 V

-

-

t \

-

\ VV- --

\ \\- --

'"5·c0e

N V

N oV

V:

OV

\V

-\V

THE INTERNATONAL COFFEE ORGANIZATON 197

The Narrative: Part I

The origins of the ICO lie in the emergence of Brazil as a donant producer and its subsequent eorts to inuence prices in inteational markets. In thenineteenth century, Brazil became the primary producer of coee. When leaf

rust spread to the plantations of the Dutch East Indies, coees from that region became scarce and their prices rose. As t he supply of Brazilian coeesincreased, their prices declined (Delm Netto 1959:155). As the price dier

ential between coees from the two origins widened, consumers increasingly switched to Brazil's cheaper coees. By the mid-1800s, 50 percent of the coee purchased in world markets came from Brazil, a gure that rose to 70 percent by the early 1900s (gure 5.1). With the collapse of exports from the east, Brazil

became a giant in the world coee market. It became the dominant producer of

the world's coee (see Delm Netto 1959; Holloway 1975; Bates 1997).Brazil was quick to seize the opportunity oered by its market power.

Led by political leaders from So Paulo, the largest of its coee-producingstates, Brazil in 1906 launched its famous valorization scheme, purchasing andstockpiling nearly eight million bags of coee (Delm Netto 1959; Holloway

1975). Brazil's eorts to "stabilize the pce of coee soon attracted newentrants into the market, and Colombia soon became its major competitor. As

shown in gure 5.2, Colombia increased its exports from fewer than treehundred thousand bags in the early 1890s to over tee million bags in the earl y 1930s. In its eorts to exploit its power in the inteational coee market, then,Brazil provoked entry.

In the mid-1920s, Brazil once again sought to determine the inteational price of coee. Announcing a policy of "permanent defense, the govement built warehouses, controlled the movement of coee by rail and by sea, created a nancial system to enable the purchase of the crop, and constructed a regulatory bureaucracy tha superintended the stockpiling and movement of coee, retingit from the exteal mrket and thereby driving up its price in global markets.However, Brazil fered the entry of additional producers into the market. It therefore sponsored a series of inteational conferences at which it sought to

secure agreements to lmit the production of coee for export. In these meetings,its diplomats repeatedly threatened economic sanctions should its rivals failto restrict production and exports (Rairez campo and Perez Gomez 1986).When other nations failed to endorse a formal marketing agreement, Brazil acted

on these threats: increasing monthly shipments by 50 percent, on November 8,1937, Brazil dumped its stocks of coee (Taunay 1943:60).

As a result of Brazil's actions and the loss of the European market in thelate 1930s, Colombia and other Latin American producers joined with Brazil tolimit exports and thereby stabilize the inteational price of coee. The end of World War II led to a sharp upward shift in t he demand for coee. The resultant postwar rise in coee prices was brought to a climax by the Brazilian frost of



9

I 80> 7Q

6

' Q

eQ

a 4

3( C 1 " r ( C 1

0 " r ( 1 1 1 r r r

FIGUR 5.1. Brazil's share of world coee exports by year 1853-1917 Sourc: Beyer (1947, appendix table V). Reproduced from Bates(1997



I<.

l111

.

I11

.

l1.

6..

lL.

IO.

lL.

IO.

l1.

Io,..

oc e <oc

S6eq 6-9 J S!W

C

s0

"("

�:

�

Nc

0"§C :(C

�;r�s0("'0

B.

8!C

�(<

.:0c

TH TTL C ZT 203

1978; Kreps and Wilson 1982 Analysis of this game provides insight into thediculties faced by the dominant producers in preventing entry.

The chain store paradox refers to competition between an established rm, a chain store, and a potential entrant. I n order to retain its monopolistic status, thechain store seeks to deter entry; it therefore threatens to ht any seekingto enter the market. Fighting is costly. The new arrival, or entrant, can choosewhether to enter the market or stay out. The incumbent can choose whether to incur the costs of ghting the new entrant, thereby retaining its monopolistic

status, or let it enter and then share the mrket. The new rm moves rst.In gure 54 the new entrant is designated and the monopolist I (for incumbent). The new rm can enter (E or refrain om doing so ("E, where"signies negation); the incumbent can then choose to retaliate R or not (RThe payos to the entrant are listed rst and those to the incumbent second.Should the incumbent deter entry, it then reaps monopoly prots x Should it fail to do so, the enteng rm receives positive prots 1 > y > ) Fightingincts costs of 1 on both rms.

The lessons of the chain store paradox come in two stages: those generated bythe game when it is played a nite number of tim es and those generated when it is played an innite number of times (or when it is ternated probabilistically).

Played once, the game yields a clear outcome: despite the incumbent's threats

to retaliate, it will fail to deter entry. The monopolist will gain, should the newrm refrain from entry; but once the new rm has chosen to enter, the monopolist does better by acquiescing, for ghting is costly. The monopolist's threats aretherefore not credible.

This result lays the foundation for the paradox. Intuitively, when the gameis repeated, it would seem plausible that the incumbent would ght. A chainstore, for example, might be expected to pay the costs of punishment in onemarket so that it could render its threats credible in others. But analysis of thegame shows that such intuition is violated. In the last market, knowing that the incumbent cannot prot from ghting, the new rm enters. So too in the penultimate market . Thro ugh backward induction, the pro cess then unravels,such that in the rst market the chain store chooses to share the mrket rather

y O

NO�O�

y O

y - -

FGU 54 A g h h gm bbrviation w ; I,um; E ; E ; R l; "R



204 CHAPTER FIVE

than to contest eny. Knowing at no future periods of monopoly prots await it in other markets, the incumbent will not incur the costs of ghting in the rst.The dominant rm-the chain store-is therefore, pradoxically, powerless.Even in repeated play, its threats are not credible.

When the game is innitely repeated (or randomly terminated), however,then the analysis suggests that for a suciently high discount factor, thedominant producers will ght to deter eny. The costs of ghting will yield a stream of benets over a suciently long period that, if not heavily discounted,

will more than compensate for the shortterm losses. In the face of an innite(or uncertain) time horizon, the monopolist might therefore be willing to paythe cost of ghting (Fudenberg and Maskin 1986). The dominant producer will

treat the costs as an investment in creating a valuable reputation. nowing that it will ght, potential competitors refrain from entry; they are deterred from

competing in the market.In applying this analysis to the coee market, we can note that Brazil,

Colombia, and other producers were locked in an innitely lived game, or one that would be terminated only though a random act of fortune-such as

by the diseases that ha d earlie r befallen the Ea st Indies. And yet, although t he

incumbent, Brazil, sought to deter entry in the 1930s, it refrained from doing soin the 1950s. The logic of the model thus fails to nd conrmation in the data.

However, its failure proves fritl; it motivates us to ret once again to thenrrative and to a reappraisal of the case mateals.One possible lesson is that goveents possess interests that dier from

those of nations. The producing countries may be innitely lived, but their goveents re not. Goveents may therefore not be willing to incur shortterm costs to achieve gains that will accrue over time. The more politically

insecure the goveent, the more heavily it will discount the ture. Such reasoning highights the signicance of domestic politics to the inteational

behavior of nations. It highlights in particular the importance of the politicalinsecurity of the goveent of Brazil and especially of its pressing needfor foreign exchange (Bates 1997). But this attempt to render the analytics

consistent with e narrative rns afoul of an obvious diculty: the fact that

the Brazilian goveents of the 1930s were surely as insecure as those of the 1950s, but nonetheless contested entry.

We are thus driven to take yet another look at the data. both the 1930s and

the 1950s, Brazil constituted the dominant producer in the inteational coeemarket. In the 1930s, when Brazil faced competition from Colombia, it faced

prospective, but not present, threats from Africa as well. Aer World War ,

however, Brazil actually faced competition from that quarter during the war,

African producers had expanded their plantations and captured a large portionof the European market. They had already entered the coee market.

Retung to the logic of the model, we rapidly gain an appreciation of the political signicance of these economic facts. Coee production involveshigh xed costs. Prior to the late 1960s, coee trees required ve to seven

THE INTERNATIONAL COFFEE ORGANIZATION 205

years o reach ll production, and variable inputs, such as pesticides and labor,constituted less than 30 percent of the total costs of production. After producersincurred such high xed costs, they would naurally be reluctant to close down production in response to a fall in price; in this case, they would be reluctant touproo their plantations, and instead would simply reduce eir inputs of labor.The implication for the model is that when a region enters production, the

game becomes a oneperiod game; for any reasonable rate of discount, the new player must be viewed as innite ly lived. The chai n store model demonstrat esthat with nite repetition (such as one period of play), entry cannot be deterred

by costly threats; in such setings, threats lack credibility. Changes that took place during the 1940s thus rendered the market of the 1950s the equivalent of

a singleshot game.In studying the origins of the ICO, we have taken recourse to one theory of

imperfect competition a theory of entry deterrence. From it, we have gained a deeper appreciation of the political signicance of the economics of the coee

industry and, in particulr, of the way in which the technology of productioninuenced the sategic behavior of the producing nations. We have also gainedinsight into why dominant producers failed in their attempts to cartelize themarket and why they therefore sought an alteative way of restricting market

competition-one based not on entry deterrence, but rather on thirdpartyenforcement.

Return to the Narrative

When Brazil and Colombia realized that they were unable to prevent free riding

on the part of small producers, they tued to the dominant consumer, the UnitedStates, to police the agreement by imposing restrictions on coee imports. In sodoing they were aware that econoic arguments would not work. The UnitedStates was a consumer, not a producer, of coee. Brazil and Colombia therefore

based thei r appeal for U.S . coopera tion on the tea t of communism. January 1959, Castro entered Havana; by the middle of 1960, he had made

clear his opposition to the United States and his commitment to a MarxistLeninist philosophy. the presidential elections of 1960, John Kennedy andRichard Nixon competed to see who could pose as the greater crusader against

comunism and, in particular, "Castroism, its local variant. The goveents

of Brazil and Colombia were able to lik the defense of coee prices to thedefense of hemispheric security. Indeed the United States' statement of support for the coee agreement was incorporated into the proclamation of the Alliance

for Progress, the Kennedy Administration's pncipal response to the communist threat in Latin America.5

5T bs ms f s s cd Fs (1972). S ls Wg (1970)d Ls d Os (1970).



2 APTR V

For the United States to become a member of the ICO, the executive branch's support for an inteational coee agreement was a necessary, but not a sucient, condition, however. Membership required passage of a treaty, andthus action by the Senate. It also required the passage of enabling legislation bythe House of Representatives, empowering U.S. customs authorities to monitor and police shipments by private rms (gure 55 Although the executive

branch and the Senate may have been willing to trade o economic costs for political gains, the House was much less willing to do so. Indeed, because of

its conces with ination and the growth of govement regulation of privatemarkets, the House rst delayed and then defeated the legislation necessary for U.S. participation in the ICO. Only with the convenng of a new Congress didit pass the legislation, after having delayed U.S. entry into the ICO for over two years. Even then, repeated assurances from major corporations that the

regulation of the market would not ha their interests were necessary beforethe House would act.

The Department of State sought to regulate inteational trade in coee soas to raise its pce and thereby inject much-needed foreign exchange into theeconomes of Latin America. However, it lacked the expetise about commercial

practices that would enable it to regulate trade eectively, and so it tued tothe large coee-roasting rms for information. These rmsGeneral Foods

Procter and Gamble, and otherspurchase "green coee from the coee�

producing nations, which they then roast, grind, and sell at retail in consumer markets. The executive branch also needed to secure the support of Congress,and the rms' cooperation proved ctical for this puose as well.

U.S. participation thus rested upon a domestic political coalition that included the executive branchthe White House and the Department of Stateand Congress. The large rms played a major role in constructing that coalition.The consensus of all who have researched the history of the ICO is that, hadthe rms opposed the agreement in ther testimony before Congress, the treatywould not have passed and e ICO would never have been created (Krasner 1971, 1973a,b; Bruchey 1987).

Anyzn Nrrv: u II

The Chicago School

Having employed one modelthe chain store model of entry deterrencetoexane the orgins of the ICO, we now tu to another: the model of third party enforcement oered by the so-called "Cicago school of regulation

(Stigler 1971; Peltzman 1976). This approach stresses that cartels are inherentlyunstable; producers face grave diculties in restrcting entry and in restrainingcompetition. In response, the argument holds, producers shft their eorts

T TRATOAL O ORAZATO 27

E Cg

N + 0

+ +

FIGU 55 d d f p f x p Ud Sb: , pp; , d bj

from the marketplace to the political arena. By marshaling publc-interest argumentssuch as the need to enhance quality or to protect public health private rms secure goveent regulation of an industry. By thus bending the regulatory process to their own ends, they employ public power to secure privateobjectives: the cartelization of the market and the redistrbution of income fromconsumers to themselves.

The Realst School

The ar gument of the Chcago school seeks t o p r ovide a neoclassical model

of ca r tel for mation. But when a p plied t o the case mate r als, the a r gument

confr ont s a majo r anomaly: the United States is a consume r , not a p roduce r, of

co ee.6 Being neoclassical, the model requi res rat ionality in individual choice.

A majo r reason fo r ap plying models t o naatives is to secu r e logically consist ent

ex planat ions fo r events o r out comes. The Unit ed Stat es' expendit ur e of cost ly

e or t to est a blish an ag reement t hat raised prces to consume rs violates the

premises of the model.

As was t he case in ou r discussion of the chain sto r e paradox, ths f ailu r e of

analysis is nonet heless inf o rmative it casts light upon impo rt ant p ro pert ies of

the data. In this instance, the failure highlghts the signicance of the comunist

t r eat t o U.S. secu rit y. Just as the pu blic's conce wit h ai r safety today ena bles

p roduce r s t o secur e the regulat ion of com pet ition among ai rlines, so, by analogy,

e Un ted Stat es' conce with the communist t h reat enabled the p roduce rs of

co ee to secure the regulation of com petit ion in the co ee ma rket. Adding

" political p ref e rences t o the Unit ed States' ob ject ive f unction t hus rest or es

logical consist ency; it highlghts political benets t hat com pensate for the

econoc costs r esulting fr om the f o rmation of the car tel The neoclassical

model thus joins the set of ex planat ions o e red by the "r ealist school of

int eational relat ions, which tr eat s st ates as rational acto r s and argues that

large states, o r so-called hegemons, willingly bea r t he p rivate cost s of pr oviding

the inf rastructu r e that pr ovides collect ive goods (e.g., Kindle be r ge r 1973;

Waltz 1979).

6With the exception of the reativey smlamount of coee production in Hawaii.



208 CHAPTER FIVE

The facts of the naative again intrude, however, and mark as prematuresuch eorts at closure. The explanation is saved by noting the magntude andintensity of threats to security. But the narrative discloses that even at the height of the cold war-even when the introduction of Soviet missiles into Cuba

underscored the agility of U.S. security and the dangers of communismthe Congress of the United States delayed and defeated the admnistration's

eorts to implement the ICO. There is thus an awkward lack of t. Giventhe supposed magnitude of the stimulus, the response was uncertain and long

delayed; a major reason for the delay was that multiple domestic actors were responsible for foreign policy. Our attempt to presere the n eoclassical vision

of policymaking by identifying political benets thus fails to t the data.

Turning to Domesc Poltics

The performance of the Chicago model of cartels thus inspires us to searchfor a more decentralized model of politics. We shift from a unitary actor to a

perspective based on collective choice. Phrased another way, we shift from a focus on inteational relations between presumably unitary states to a focus

on the domestic politics of inteational policymaking.The United States entry into the ICO required subscribing to the terms of

a treaty between states. Enabling legislation was required to implement the

treaty, and participation in the ICO therefore required the concuence of bothhouses of Congress. To model the behavior of the United States, we can adopt a spatial framework: one that adheres to the prese of rationality in choice but

that analyzes how collective polcy emerges from the strategic intelay amongdecentralized interests.

Spatial models are neoclassical; they locate the preferences of actors as pointsin a policy space, which represent ideal or bliss points, i.e., points of maximum

utility. Polcy outcomes can also be represented as points in that space, and their

distance from an ideal point can be interpreted as a measure of utility losses,the extent of the loss being proportional to the distance. As secuty could be

purchased by paying higher coee prices, the policy space can be cast i n on e

dimension: a greater interest in security implies a greater willingness to pay for coee imports.

In gure 5.6, I construct a spatial portrait of the interactions among theexecutive branch (E), the Senate (S), and the House of Representatives (H). Thestatus quo (SQ) denotes the level of security generated by prevailing policies.

Inteational security constitutes the underlying dimension; more specically,

the dimension depicts the willingness to commit resources to counter the threat posed by c ommunism. Panel A represents the distribution of preferences in the

early 1950s. The location of the ideal points of the House, Senate, and executive


branch suggests a low conce with security, the executive branch being themost conceed. The point labeled "ICO signies the level of resources that would be channeled into the ght against communism by entering into the ICO.

Panel A suggests hat the executive branch would be most inclined to join theICO. Its ideal point lies mdway between the status quo and the ICO policy outcome. The ideal points of the House and Senate, however, lie closer to the

status quo than to the ICO, suggesting that, given their low assessment of thesignicance of the counist threat, they would not be willing to shift to thenew policy_?

Panel B suggests the change brought on by the rise of Caso. The executive

branch now shifts to supporting the ICO, as does the Senate. Some membersof the House likewise reappraise the signcance of the communist threat;

some remain unwilling to ask American consumers to nance the war against communism by paying higher consumer prices. But the median of the Housedoes shift to the right; its preferences track those of the large roasters, who have

now come out in favor of the inteational agreement. As the two houses of Congress as well as the executive branch now prefer the outcome under the

ICO to that under the status quo-i.e., as their ideal points lie closer to the point

marked ICO than to the point marked SQ-the necessary conditions for theadoption of the policy now prevail and the collective choice now shifts in favor of the agreement.

seeking an explanation for the formation of the ICO, we have thus movedfrom he study of entry deterrence (or its failure) to the study of cartel behavior.Although it provided a useful starting point, the Chicago model of regulation encountered a major anomaly, one that challenged its neoclassical foundations. To

t the naative, while also preserving logical consistency, we recast the model,

7rt is impotant to justi the specication of pefeences. One way is by appealing to casemateials. As descibed by Kasne (191), Buchey (198), and Bates (199), the passage of thereo was managed in the Senate by the Foeign Aais Cottee, whose membes wee deeplyconceed with inteaonal elations. Pio to addessing the question of the reo, the committeehad investigated and epoted upon the oigins of political instability in Bazil, and in doing so it hademphasized the destabilizing pact of low commodity pices and shotages of foeign exchange

the House of Repesentatives, the Ways and Means Comittee inoduced the legislation. Littleconceed with foeign aais the cottee focused instead on domestic taxation.Anohe way of justifying pefeences is by appealing t eason. A senato is elected fo a six

yea tem, a congessman fo two. One-thid of the Senate is up fo election evey two yeas, but allmembes of the House stand fo eelection on that schedule Econoic shocks ae theefoe moelkely o have an impact upon the elections of a majoity of the membes of the House than uponthose of a majoity of the membes of the Senate. The House is thus moe likely to pay attention toeconoc issues Moeove, it is easie to launch a pesidential campaign om the Senate than omthe House. Motivated by ambition, senatos ight be moe willing than membes of the Houseto endose costly pogams that yield enhanced national secuity. It is notable, in that egad, that

sevea of the senatos on the Foeign Aas Comittee at the time of its endosement of therCOfo example Senatos James W. Fulbight and Hubet H Humpheywee maneuveing fo places on the national ticket of the Democatic Paty



213



212

TABLE5.5

CHPR FIV

Members of the oreign airs ommittee of the ational oee ssociation,960-6

Year Six largest roasters Six largest green coee rms Small rms

960 0 96 0 496 96 4

Source: Krasner (1971:255).

Aais Comttee to monito and advise the goveent as it pepaed tointevene in coee mkets. The comttee intensied the association's eotsto educate the bueaucacy about the economic and coecial ealities of thesemkets. It communicated by lette,8 by dispatching delegations to the Depatment of State,9 and by hosting Depatment of State epesentatives at businessconfeences.1 At the depatment's invitation, the Foeign Aais Committee of the NCA paticipated in the Coee Study Goup, a coittee oganized by theDepatment of State to develop the egulations that constituted the famewokfo the Inteational Coee Ageement.11As shown in table 5.5, the lage oasting ms held half o moe of the positions on the Foeign Aais Cottee.

The industy was active not only in the United States but also in Latin Ameca. Even a casual eading of the commecial pess and the minutes of oganizations within the Latin Amecan coee industy yields an appeciationof the pominence of its executives. The rival in a poducing county of, say,Geoge Robbins, head of the coee division of Geneal Foods, amounted to astate visit, with fomal eceptions, ceemonial dinnes, and extensive coveageby the national pess on the political, economic, and society pages. And Robbinsdid visit, egulaly touing the geat coee centes: Rio, Santos, So Paulo,Manizales, and Bogot.1

One of the most signicant of his visits took place in Bogot in July 63.The teaty had been atied by the U.S. Senate in Decembe 62. Geneal

8See, for example, the correspondence between Edward A bo, president of the NCA, and thesecretry of state, contained in le 21.2333/1-1653, Unted States Archives.

See the discussions in Fisher (1972) and Bruchey (197:13.0The NCA holds an annual conference in Boca Raton, Florda, to which it invites goveent

ocials from the United States and abroad.I!Information on the organzation of the Coee Study Group is contained in le 392333/2-

1955, United States Archves.So constant was the presence of Robbins that I failed to record each notice of his arrival. For

a Brazilian example, see the coverage given hs 194 visit in Revista do Comrcio de Caf de Rio

de Janeiro, February and March 194. For examples from Colombia, see Atas No. 3 and 42-44,October 3-December 1, 1952, in Comit Nacional (1952). See as well records of the visit of hssuccessor, Paul Keating, to Manizales in Federacin Nacional de Cafeteros (19).

H INRNION COFF ORGNIZION 213

Foods had paticipated in the tade advisoy committee that had helped o d t the accod in Washington and in the delegation that helped to negotate ts atication by membe nations. Its executives also seed on the delegation epesenting the United States at meetings of the ICO. In the me

_etings in Boot

in July 63, Robbins, fomey a membe of the U.S. delegatn, enteed todiect commecial negotiations with the managing diecto of Colombia's coeeagency. And the executive comttee of that agency convened a seies of secet sessions to discuss his poposal.1

the ely 60s, Genal Foods was puchasing ,00,000 bags of Colom-

bian coee annually, o nealy onequate of Colombia's total poduction andnealy twothids of its expots to the United States. It now oeed

to inceae

its annual puchases by 500,000 bags, with a coitment to contue at thsinceased level fo thee yeas. In exchange, Geneal Foods wanted a eductionof $0.02 a pound in the pice chaged fo the additional bags puchased, ove and above its usual discount of 0 pecent. It also oeed to launch a new bandof coee, to be called Yuban, that would featue "pue Colombian coee.

The leades of Colombia's coee industy teated Geneal Foods' oe withgeat caution. They had long sought to build consume� loyalty t � olombiancoee in the United States and had even paid a leadg advetsg m to pomote it; Geneal Foods' oe to featue a "pue Colmian label woul,they felt, inspie emulation by othe oastes, and so gVe a boost to the

mketing campaign. Howeve, they also saw disadvantages to the poposal.One was the dange of becoming even moe dependent on Geneal Foods;anothe was the isk of being seen by othe poduces as cutting the pice of coee, just as an accod to stabilize pices was being put in place. In the en�, theexecuive committee decided to go ahead with the deal. They did so, I believe,in ode to stengthen, not weaken, the inteational ageement.

.Shotly afte the initiation of postwa eots to build an accod, the doant

poduces had signed bulk contacts with the lage coee oasting ms.

Colombia appes to have signed its st such contact with Gene o?ds 5;Bazil followed suit in May 60.14 The evidence suggests that la contactswee signed with othe majo oastes: Nest, Rots, and Pocte and Gamble.The contacts committed the oastes to the puchase of xed amounts of coee pe quate50,000 bags, in the case of Geneal Foods in the ealy 60sinexchange fo discounts on the pice of coee. Unde the tems of each c�ntact,the discount was applied to the costs of tue puchases. Because the dscount was not paid immediately, but athe ebated in subsequent peiods, the maket fo coee was tansfomed fom a spot maket. Unde the tems of the bukcontacts, the poduces and oastes enteed into long-tem elationships.

This account is drawn om Federacin Nacional de Cafeteros, Comt Nacional (1963)Acta No. 4 de la sesin del da 2 de enero de 196 and Acta No. 1 de la sesin del da

12 de mayo de 1965 Comit Nacional (196, 1965).



214 CHAPTER FIVE

Of course, nancial costs and benets constitute a signicant component of the exchange between the dominant producers and the large roasting s.The $0.02 per pound price reduction in the purchase of Colombian coeeenabled General Foods to pay nearly $2 million less for its 1,500,000 bagsof coee and Colombia to generate a 17 percent increase in its sales in theNorh American market. But Colombia got even more out of the bargain. Aspart of its relationship with General Foods, it also secured political services.

General Foods and the other large roasters were not merely purchasersand processors of coee. They were also lobbyists and members of national

delegations. They provided the executive branch with the information necessaryto aintain and to regulate the coee trade and testied before Congress onbehalf of U.S. intervention. From Colombia's point of view, the discount in theprice of coee to General Foods bought compensating advantages: the help of a large roaster in securing the Unted States' enforcement of the InteationalCoee Agreement, thereby checking opportunistic behavior by the competitivefnge-and raising the average pce of coee.

In tu General Foods and the other large roasters who had signed bulkcontracts received major economic advantages. The large roasters operated inthe national market. Taking into account their costs for advertising, distribution,and promotion, the purchase of raw materials constituted a comparatively smallportion of their expenses. However, for the small, regional roasters with whom

they competed-Chock Full O'Nuts in Baltimore and Atlanta or BreakfastCheer in Pittsburgh-the price of raw materials constituted a relatively larger percentage of total costs (Hilke and elson 1989; Sutton 1991). By strcturingthe regulation of the market so as to increase the price of raw materials, and bysecuring rebates from the dominant producers of those raw materials, the larger roasters were able to increase the costs of raw mateals to their competitors,thereby achieving a cost advantage. 15

Analyzing the Narrative: Cut I

In describing the behavior of the major coee rms, this narrative has portrayed

behavior analyzed by Oliver Williamson and others a form of market rivalryin which one rm seeks to maximize its prots by raising the costs incuedby its rivals.6 Within the United States, the large roasters backed the causeof the agreement. Insofr as the U.S. govement enforced it, it raised thecosts of raw materials to all coee rms; however, insofar as the large roastersreceived discounts from the producers, the increase in costs aected only their competitors. They thus secured a cost advantage.

15The situation most clearly resembles the Penington case (Williamson 968). The impact of the induswide shift in the costs of production on competitors in the industry is magnied in thisinstance by the rebates (Salop and Scheman 983)

1See Williamson (968) and Salop and Scheman (983)


A major question then arises: why would such an agreement be stable? Assuggested in the logic highlhted by the entrydeterrence game, the answer must e in the credibility of the theats that support it. This reasoning focusesattention on the relationship between those who demanded the agreement (thelarge coee producers) and those who organzed its supply (the coee roastingrms). Examining the outcome of their behavior in a collusiv equilibrium, wend that when a large roaster, like General Foods, helped the U.S. goveentto police illicit shipments of coee from the competitive fringe-say, CentralAmerica-it drove up the price of those coees. The dierential between theprice of Colombian and other coees then declined. Given the dierentialsin quality between Colombian and other mild coees, buyers switched toColombian coee. As more buyers tued to Colombian coee, Colombia'sdependence on any one given buyer of its coee declined. The strategies pursuedby the parties in the collusive equilbrium thereby lowered the costs to producersof implementing teats to the roasting rms. The reduced costs to Colombia of canceling its contract with any given rm ren dered Colombia's threats-but thistime, against rms-more credible. The collusive agreement could thereforebecome selfenforcing.

In an interview (October 27, 1982) with the head of the coee division of Procter and Gamble, the second largest roaster of coee in the United States,I gained insight into the extent and nature of the fears that deterred movementaway from the collusive equilibrium. Procter and Gamble cultivates a reputationfor a strong commitment to "allAmerican values: patiotism, capitalism, andcompetitive markets. Yet when I questioned the head of its coee divisionabout the agreement he indicated that "the free market would be O.K. "Wemay in fact testify in Congress against [the agreement], he added. "How aboutBrazilan reprisals? I asked. "Would they be likely to punish you by cancelingyour contracts? here was a long pause before the executive replied, "Don'teven breathe that possibility to anyone else. I will have to explain to our Chief Executive that the Brazilians may force us on board. I would be less than honestif I didn' t say this to him. The Brazilians price coee so attractively to us-wego with that contract and buy big and use it. We buy all we can get. But thenthey can put the screws on us.

The chief executive ocer of Procter and Gamble preached the vtues of the free market; the head of its coee division had to deal with the realities of the coee market. I take the latter's comments as conrmation of hs belefsas to what would happen were his company to stray from the path o f collusionand violate the terms of its relationsp with the organizers of the accord.

seeking an analytic framework for the narrative oered by the ICO, Ihave reached for models of imperfect competition. Most fail, but each doesso in a interesting and iformative manner. As with the chain store paradoxand the Chicago model of regulation, their failures highlight important featuresof the data-features whose signicance has thus f been insuciently appreciated. Their failures also inspire important reformulations of the models,



216 CHATER FIVE

either to improve their goodness of t or to restore consistency in their logic.Ecoomic models of collusion become models of political economy, or, aswith the realist school, models of inteational relations give way to models of domestic politics.

In the end, we arrive at a complex vision, albeit still one of a cartel. Formallythe ICO constituted an agreement among sovereign states. In reality it was a

political and economic alliance engineered at the subnational level. The ICO represented a coalition among bureaucrats, politicians, and rms that used the

power of states to restructure markets. The result was enhanced political securityfor public ocials and the enhanced protability of rms, which secured wealthin retu for political service.

The ICO as an Institution

Thus fr we have expored the creation and maintenance of the IC0.17 We nowtu to its behavior

Efectveness

It is dicult to measure dectly the impact of the ICO. Being unable togenerate reliable estimates of the impact of the ICO, I have therefore settledfor a less direct assessment of its impact. For the ICO to work, it needed tolll two necessary conditions: it had to restrict arbitrage between the memberand nonmember markets and competition among the producers of coee. Theevidence suggests that it succeeded at both.

RESTRCTG AGE

Figure 57 presents evidence conceing the eectiveness of the ICO. Thehorizontal axis displays years, including the dates during which quotas were in

eect: August 1965 to October 1972 and October 1980 to August 1985. Thevertical axis records the ratio between coee prices in the markets of member

17The operations of the ICO create three measurement problems. The imposition of quotaslimits supplies when prices are declining, thus censoring the data and biasing estimates of thedeterminants of market prices; mrket forces determine prices when they lie above the target rangeand quantity restrictions when they enter that range, thus imposing a change in regime; and pricescan induce a change in quotas, just as quotas can induce a change in prices, thus creating a problemof endogeneity The attempts of other scholars to assess the pact of the ICO have failed to addressthese issues. I therefore could not use their estimates. My own attempts to consuct a properlyspecied model of the regulated coee mrket foundered on the absence of data on key variables,especially in monthly time sees.

\vv·

v<II

l[.

�

p

I

1�

I-

•

vt'

�ocoooooooo1<I"MN.omc��..

JU:d

LB

8

98

PBS

£8

ZB

08

L

BL

LL ��

L :09L

PL

£L

OL

8

L

9

19



18 CHPR FIV

and nonmember nations, expressed as a percentage.18 When the ratio exceeds100 percent, then prices in member markets exceed those in the markets of nomembers; arbitrage has been prevented and barriers to trade eectively putin place. The evidence in gure 5.7 suggests that when the ICO imposed quotasit was able to enforce them, yielding a higher price in the coee markets of member nations.

Although visually satisfying, were these eects statistically signcant? Toanswer this question, I calculated the means and variances of coee prices inmember and nonmember markets during periods in which the ICO did and did

not impose quota restrictions.19 I constrcted a vaable, D, which measures thedierence in the mean value of prces in the member and nonmember marketsin each period. D can be transformed into a test statistic that possesses a t

distribution. The null hypothesis is that the mean of D is 0, or that there is nostatistically signcant dierence between the mean of the prices in the twomarkets. We can reject the null hypothesis if the test statistic falls into a critical range-a range within which its value would be highl y unlikely (one chancein one hundred) to fall were it determned by chance. In this instance, I chosea critical range appropriate to a onetailed test I wished to reject the null whenthe value of the test statistic was highly and signicantly positive, i.e., when themean price of coee in the member market exceeded that in the nonmember market by a level signicantly greater than would be likely by chance. Given

the number of obserations available, the critical region in which one can rejectthe null hypothesis begins at 233 The results of the test, presented in table 5.6

enable us to reject the hypothesis of no dierence in the mean price of coee inthe two markets at the .level of condence. The imposition of quotas appearsto generate a signicantly higher price in member markets.20

RIVALRY

Thus far I have focused on the impact of the ICO on average prices. At leastas important is its impact on relative prces, for competition between coee

8The measure of coee "prices is actually the unit value of exports, as recorded in Inteational

Coee Organization, W.P-Agreement No / Rev 2.9With the help of Ms (now Dr.) Dixie Reeves, my research assistant For a development of

the method followed, see Larsen and Marx ( )20As seen in gure 57 and table 56 when quotas were not in place, pces in member mkets

fell below those in the mrkets of other counies At such times, exporters sought hard currencyengs and competed for sales in Europe and the United States

The greater tendency for producers to sell lowquality coees in the nonmember marketconstitutes a plausible alteative explanation of the price dierentials. Research by economists atthe Department of State concluded, however, that this behavior could not account for the magnitudeof the dierential (interview, Washington, DC, June ) And a reanalysis of the data underlyingtable 56 suggests that the dierential was as great for each qualty of coee, i.e., as great for

robusta as for lds

H INRNION COFF ORGNIZION 19

TABLE

Testing for the Impact of the Coee Quota

Number of observations Mean forD -Statistic

Periods in which quotas were in place

August to October

October 0 to August

Periods in which quotas were not in place

November to September

September to August

Source: Monthly price data from the les of the ICO

00

40

40

34

4

-

-0

-4

-3

producers takes the form of competitive prce cutting by producng ntions.The market appears to recognize four qualities of coee. Colomban lds, awashed arabica, are positioned at the top of the market.21 Then come the other milds, the washed arabicas produced in Central America. Brazil's unwashedarabicas occupy the third rng in the quality rankngs.22 Africa's robustas slotin at the bottom of the market.23 Despite the quality dierences, consumersare willing to substitute among these vrieties, given sucient incentive. Acompetitive cut in the prce of other mld coees would therefore pr ?d

.uce a

lowering of the price of Colombian coee, for consumer woud be.ll toswitch from Colombia's highquality coee, given a sucent pce derental.A second test of the eectiveness of the ICO, then, is its impact on relative

prces. Figure 5.8 illustrates the patte of relativ prce for a peod durngwhich quotas were in eect. By and large, price derentals remaed stale.The patte contrasts with that exhbited in gure 5.9 whch

. portrays relate

prices in a perod durng which quotas were suspended followg a catastrop�cfrost in Brazil. Given Brazil's size in the market, the frost produced a massve

rise in prices (compare the prices on the vertical axes of gures 5.8 nd59). It also produced a massive distortion in relative prices. After recoveng,Brazil sought to reposition its coee at a price level intermediate betwenthat of the other milds and that of robusta coees. But, as can be seen

gure 5.9 Afrcan producers resisted, cutting their prices to keep Brail f �omclosing the gap. So did the producers of other milds, who lowered ther pcesbelow those charged by Brazil. This competitive response posed a challengeto Colombia as well, for the growing price dierentials between other mildsand Colombian coee encouraged consumers to abandon the higherquality

21 The vast bulk of which are produced in Colombi a, of course But Kenya and the Kilimanjro region of Tanzania also produce coees classied i n this group.

22Ethiopia also produces unwashed rabicas; its crop is insignicant by compason w1th that

of Brazil.23Indonesia is also a major producer of robusta coees.



0

.cQ

:

4400

4200

4000

300

3600

3400 +

3200

-� �- �

3000+-Sep Nov Dec Jan Feb Mar Apr May Jun Jul Aug

1-Combian -Olds Brazii�:usta

FIGU 5.8. C ps, 1967-68 Source: Mnly pi data fm th ls f th C Rpdd fm Bats (1997).



222 CHTER FIVE

Colombian product. Colombia therefore joined in the price cutting. In the face of the kind of competitive behavior portrayed in g ure 5.9, the dominant producerssought to reimpose the quantity constraints that would generate the evenlyspaced price dierentials exhibited in gure 5.8. They succeeded in doing so.

The ICO thus appears to have worked. It was apparently able to preventarbitrage across markets and competition between producers of dierent kindsof coee. It appears to have been able to regulate behavior in the coee marketin ways that enabled it to achieve higher average prices. Indicative of this power is the behavior of coee prices following the breakup of the ICO. As will berecounted in the narrative that follows, in June and July 1989 the ICO collapsedAs seen in gure 5.10, when the ICO broke up, coee prices plummeted. Thespot price for Colombian milds, for example, fell from $1.80 a pound in Juneto less than $1.00 by July-and stayed below $1.00 a pound for several yearsthereafter.

Operaon

In addition to having been able to regulate the coee market eectively, theICO appears to have been rule goveed. The bestdocumented example of the impact of the rules comes from the negotiations that led to the assignment

of quotas in 1982.4 Quotas had been suspended when prices rose above thetarget range because of shortages resulting from the Brazilian frost of 1975.As prices retued to the level at which the ICO would seek to defend them,Brazil proposed a new set of quotas. Under this proposal, Brazil would secureover 33 percent of the market, a share held in the 1960s but long since lost, particularly since the frost of 1975. The primary victim of Brazil's proposalwould have been Colombia, which would have been constrained to less thana 19 percent share of the market, after having captured over 22 percent of itfollowing the frost. Colombia therefore set out to block Brazil's proposal bycrafting one of its own.

Under the rules of the ICO, quotas had to be voted upon, and Article 13of the Inteational Coee Agreement goveed the allocation of votes. Under

Article 13, the proportion of total exports for the period 197677 to 1979/80accounted for by a given producer, i (call them Pli, where 1 stands for the time

period above, i for the producer, and P for export performance) determined the proportion of the votes controlled by that producer (call t hat proportion Wi).This rule can be represented as

(1)

24This section draws heavily upon Bates and Lien (1985) and Lien and Bates (1987). The articlescontain the data underlying this analysis. For eatments of closely related issues, see Fielding andLiebeck (1975) and Hosli (1993)

0

.'gE o

8t

+ +

0<

C

:c: EN (

0 0 0 0 0

N 0 <

p t ·s·n

066 JBf

t

066 q:

C(1

066 er s":(

696 oa C

89

696 N 0u�

(-

696 POC

(

696 d -s

696 5v �":(

C·c

696 1r t

696 r !c

696 A Bf

�

696 Jd\ 6u�

(

-696 Jf

696 q .

�='

696 er l0 "

);��

224



224 HAPTER FIVE

Article 30 of the agreement goveed the allocation of quotas. Claims for quotas, this article stated, were to be based on average exports over one of twothree-year peods: 1968/69 to 197172 or 197677 to 1979/80. Claims for quotas were transformed into actual quotas only when voted upon, however;quotas were set by a (distributed) two-thirds majoty vote of the member nations. Article 30 can therefore also be represented in the form of a function,wherein the quota assigned to a particular nation, i (qi), is a nction of past export perfomance (Pi where the subscpt 2 refers to the time period notedin this pragraph, rather than that used in the allocation of votes, as discussed

earlier) and of votes (Wi). Thus

(2)

Rule-goveed environments produce opportunities for sophisticated behavior. In the presence of voting rules, actors can behae strategically. In majority rle environments, they can convert endowments of votes into political power by converting coaitions into political majorities. Given a set of rules that denesa winning majority, the Shapley vue of an actor provides a measure of that actor's power: the percentage of all coalitions among members of the groupwhich that actor, given its allocation of votes, can convert into a winningmajority.26 Once again, the reasoning can be summzed:

(3)

where Si stands for the Shapley value of nation i and Wi for its endowment of votes, as noted previously.

The rles that goveed the assignment of the quotas by the ICO thus can

be recast as a system of equations:

Equation

1. w i = f(P l i)2. qi = g(Pi, Wi)3. si = h(Wi)

Justication

Article 13Article 30Behavioral assumptions

In eect, this system constitut�s a model of how quotas (qi) would beassigned under the rules of the ICO, were actors ming sophisticated use of the powers conferred upon them by those rules. An advantage of this formalizationis that it suggests a way of estimating the impact of the rules. We can estimate

25The producer nations could choose either period. Brazil, for example, chose the rst andColombia the second, as the former lost market share following the ost of 197 while the latter gained.

26For a discussion, see Luce and Raia (197). For illuminating critiques and discussions, seeRoth (1988)


the parameters of the model, and thus the impact of the rules of the agreement, by usng a two-stage least squares approach, in which the constant terms, P1

and P are used as instruments to eliminate the corelation between Si and qiindirectly produced through their association with export performance.27

Equation (4) presents the results for the allocation of the quota proposed by Colombia:

qi = -0.2279+0.8281Pi+0.2175Si 6 6

(4)

The gures in parentheses represent standard errors of the estimated pa rameters. The standard errors of the estimated coecient relating Colombia'sShapley value to its quota is one-third the value of the estimated parameter, andit is therefore statistically signicant.

These results can be interpreted as suggesting that Colombia took cognizanceof the rules and their eect on e distribution of power and allocated quota entitlements so as to accommodate the voting power of particular nations.Indeed documentry evidence underscores the hypothesis that the Colombiandelegation counted otes and craed its proposal accordingly.28 The rles thusappear to have shaped the behavior of member nations, and, in particular, theallocation of rights to export in inteational mkets.

By creating the ICO, the dominant producers sought to erect a goveancestrucre that enabled them to reduce the risks of oppornistic behaviorrisksthat undermined their capacity to incur costs with the certainty that they, andothers, would reap the benet of higher coee pices. The results of the estimatesalso provide a test of their ability to do so. The Shapley value provides a measureof the ability of a nation to convert coalitions into winning coalitions, i.e., intoones that command electoral majoities. Under the rles of the agreement,

bigger producers were endowed with greater numbers of votes; intuitively they should therefore have possessed more power. Equation (5) provides the test weseek. As seen in that equation, the relationship between the voting strength of a nation and its Shapley value is indeed nonliner; it is quadratic, such that as thenumber of votes controlled by a nation increases, its Shapley value rises more

than proportionately:

Si = 0.8227Wi+ 0.0152W ?3

(5)

The lesson is clear: by controlling more votes, lger nations gained more"Shapley power, as shown in equation (5) and therefore could extract greater quotas, as shown in equation (4). This estimate of the impact of the rles thus

27Da-Hsiang Donald Lien and I have performed this analysis (Bates and Lien 198)28MemorandumNo. 001, enero de 198, pp. 1-, Convenio Inteational de Caf, Copador

Memoandos 198.



228 CHAPTER FIVE



228 CHAPTER FIVE

shied their policy preferences away from the ICO and toward the free market

policies that had formed the earlier status quo.The collapse of the ICO marks the last stage of this naative. The institution

did not endure. But our understanding of it apears to have survived the test

posed by its deise. Changes in the value of explanator y variables acted

decisively, and in expected ways, upon the politics of the coee market.

Conclusion

As a political institution that regulated inteational markets, the ICO was very

rare indeed; few other bodies have eectively goveed the behavior o f st ates

in the global arena. As with other rare occurrences, the history of the ICO can be described it possessed a birth, a l ife, and a death . In analy zing ths case,

however, I have sought to move beyond mere na ration. I have s ought to e xtrac t

systematic nowledge.In an eor to do so, I have sought to t analytic frameworks to the history

of this institution. ne framework was that of noncooperative game theory

and, in particular, the chain store model of entry deterrence. Like the naative,

the game analyzed actors and choices. Being starker than the naative itself, it

highlighted featres of the nrrative. It shaened our underst anding of the lo gic underlying the events and heightened ou r awareness of what was problemat ic

and therefore in need of explanation. It also rovided an engine of discovery,

directing our attention to iportant features of the data.

The analysis also drove us to recognize that the creation of the ICO was

not the result o f purely econoc behavior, but rath er of political and eco

noic choice aing. eeping with the Chicago theoy of ctel behavior,I sought to see why a third party-the goveent of a consuming nation, the

United Stateswould provide enforcement for a collusive agreement among

producers. Using the tools of spatial analys is, I exlore d the olitics of coalition

formation and isolated the domestic political coaltion that explained why the

goveent of a consuer nation should suport a cartel.

The last part of this chapter emphasized an additional advantage of movingfrom historical accounts to analytic naatives: the transfomation helps to

render such accounts aenable to testing Moels exlcitly searat e exogenous

from endogenous factors and specify the causal lnkages between them. They

constitute a clai that changes in the exogenous variables should, of necessity,

generate changes in those that are endogenous. Models therefore set the stage

for testing. ne way is by estimating relationshs between vaiables. Another is by maing obserations of data generated by events that take lace outside

the original sample. Thus did the death of the ICO rovide an oportunity to

assess ou r account of its creation.


References

Arango, Mariano. 1982. El Cafe n Colombia, 13-13 Bogot: Crlos Valencia.

Bacha, Edmar, and Robert Greenhill. 1992. 15 Yars of Co. Rio de Janeiro:

Marcellno Mtins and E. Johnston.

Bates, Robert H. 1997. Opn Economy Politics: Th Political Economy of th World

Co Trad. Princeton, N.J.: Princeton University Press.

Bates, Robert H., and Da-Hsiang Donald Lien. 1985. "On the Operation of the Intea

tional Coee Agreement. Intational Organization 39: 553-59.Beyer, Robert Carlyle. 1947. "The Colombian Coee Industry: Origins and Major

Trends, 1740-1949." Ph.D. diss., Department of History, University of Minnesota.

Bruchey, Stuart. 1987. Amrican Businss and Forign Policy. New York: Grland.

Comit Nacional. Vrious years. Atas, Acurdos, Rsolucions.

Delm Netto, Antnio. 1959. Problma do Cafe No Brsil. So Paulo: University of

So Paulo.

Federaci6n Naciona] de Cafeteros. 1980. Sminario Sbr Economa Cafta. Man

izales, Colombia: orporaci6n Autonema Universiatria de Manizales.

Federaci6n Nacional de Cafeteros, Comit Nacional. 1963. Atas II (July-December).

Fielding, Geo, and Hans Liebeck. 1975. "Voting Structure and the Square Root Law.

British Joual of olitical Scinc 5: 249-63. Fisher, Bart. 1972. Th Intational Co Agrmnt: A Study in Co Diplomacy.

New York: Praeger.

Food and Agriculture Organization of the United Nations. Vrious years. Tad Yar

book. Rome: Food and Agriculture Organization of the United Nations

Fudenberg, Drew, and Eric Mas in. 1986. "The Folk Theorem in Repeated Games with

Discounting or with Incomplete Information. Economtrica 54 (3): 533-54.Gallop, Frank M., and Mark J. Roberts. 1979. "Firm Interdependence in Oligopolistic

Markets. Joual of Economtrics 10: 313-31.George, Alexander L. 1979. "Case Studies and Theory Development. In Diplomacy:

Nw Approachs in Histo, Tho, and Policy, edited by P. G. L. Lauren. New York:

Free Press.

Harvard Business School. 1987. Unitd Stats National Co Markt. Cambridge,

Mass.: Hrvard University Press.

Hilke, John C., and Philip B. Nelson. 1989. "Strategic Behavior and Attempted Monop

olization: The Coee (General Foods) Case. In Th Antitrust Rvolution, edited by

John C. Kwoka and Laurence J. White. New York: Scott Foresman.

Holloway, Thomas H. 1975. Th Bazilian Co Valorization of 16 Madison: State

Historical Society of Wisconsin for Department of History, University of Wisconsin.

Hosli, Madeline 0. 1993. "Admission of European Free Trade States to the European

Counity: Eects of Voting Power on the European Comunity Council of Min

isters. Intational Organization 47 (4): 6293.Inteational Coee Organization. 1982. Quartrly Statistical Bulltin, July-Sptmbr

1 London: Inteational Coee Organization.

Inteational Monetary Fund. Vrious years. Intational Financial Statistics. Wash

ington, D.C.: Inteational Monetary Fund.

Kindleberger, Chrles. 1973. Th World in Dprssion, 1-13 Berkeley: University

of Califoa Press.

23



23 FIV

Krasner, Stephen D. 1971. "The Politics of Primary Commodities: A Study of Coee

1900-1970. Ph.D. diss., Department of Goveent, Harvard University.

-. 1973a. "Manipulating Inteational Commodity Markets: Brazilian Coee

Policy 1906 t o 1962. Public Policy 21 (4): 493-523.

. 1973b. "BusinessGoveent Relations: The Case of the Inteational Coee

Agreement. Inteational Organization 27 (4): 495-516.

Kreps, David, and Robert Wilson. 1982. "Reputation and Imperfect Information.

Joual of Economic Theo 27: 253-79.

Larsen, Richard J ., and Morris L. Marx. 1990 . Statistics. Englewood Clis, N.J.: Prentice

Hall.Levinson, Jerome, and Juan de Onis. 1970. The Alliance That Lost Its Way. Chicago:

Quadrangle Books

Licht, 0 1993. Inteational Coee Yearbook, 199 Ratzeburg, Germany: 0.

Licht.

Lien, DaHsiang Donald, and Robert H. Bates. 1987. "Political Behavior in the Coee

Agreement. Economic Development and Cultural Change 35 (3): 629-36.

Luce, Duncan, and Howard Raia. 1957. Games and Decisions. New York: John Wiley

and Sons.

Peltzman, Sam. 1976. "Toward a More General Theory of Regulation . Joual of Law

and Economics 19 (2): 211-40.

Ramirez Ocampo, Jorge, and Silervia Perez Gomez. 1986. 8 Aos de Poltica Cafetera

Inteacional y la Participaci6n de Colombia en este Pceso. Bogot: Federacin

Nacional de Cafeteros.

Roth, Alvin E. 1988. The Shapley Value: Essays in Honor of Lloyd S. Shapley. Cam

bridge: Cambridge University Press.

Rowe, J. W. T. 1963. The World' Coee. London: Her Majesty's Stationery Oce.

Salop, Steven C., and David T. Scheman. 1983. "Raising Rivals' Costs. American

Economic Review 72 (2): 267-71.

Sears, Dudley, et al. 1979. The Rehabilitation of the Economy of Uganda, vol. 2. London:

Commonwealth Secretariat.

Selten, Reinhrt. 1978. "The ChainStore Pradox. Theo and Decision 9: 127-59.

Stigler, George. 1971. "The Theory of Economic Regulation. Bell Joual of Economics

and Management Sciences 2 (1): 3-21.

Sutton, John. 1991. Sunk Costs and Market Structure Price Competition, Advertising,

and the Evolution of Competition. Cambridge, Mass.: MIT Press.

Taunay, Afonso de E. 1943. Taunay, Historia do Caf No Bsil, vol. 15. Rio de Janeiro:

Departamento Nacional do Caf.

Wagner, Haison R. 1970. United States Policy toward Latin America. Stanford, Calif.:

Stanford University Press.

Waltz, Kenneth N. 1979. Theo in Inteational Politics. Reading, Mass.: Addison

Wesley.

Williamson, Oliver E. 1968. "Wage Rates as Barriers to Entry: The Penington Case in

Perspective. Quarterly Joual of Economics 83 (1): 85-116.

Conclusion_________________

THE CHAPTERS in ths volume have explored the sources of political order,the origins of conict, and the intelay between inteational and domestic

politcal economy. Throughout their discussion of such themes, they havefocused on the nature and signicance of institutions. Because the chapters

engage such ndamental issues, they are of general signicance. Those seeknginsight into state fomation, the political foundations of development, and the role of instiutions can read them with prot. But although the themes andinsights are general, the generality of the explaa is far less apparent. It isto ths issue that we now tu.

Postdicion

Green and Shapiro (1994) criticize the literature on rational choice for failingto be emprically grounded. We hope that these chapters oer a convincing

rebutal. But in eluding this charge we may run afoul of another. Insof as users

of rational choice theory address empirical evidence, Green and Shapiro contend they construct models that simply account for the original evidence. Mereexercises in "curve tting, such eorts do not oer adequate tests of theory.

In the introduction, we provide one rejoinder: that in constructing our theoies, we were often driven to the discovery of new features of our data.

Thus Greif was compelled to move from the domestic to the inteationallevel, Levi found it necessary to consider the distinction between rral andurban populations in their response to ilit ary recruitment, and Weingast had to

address the revision in Southe expectations following the Democratic Party'sloss of the House elections. The chapters themselves provide another rejoinder:in some, the models fail. Bates, for example, was compelled by the evidence

to discard, or recongure, a variety of approaches to the analysis of cartels.Greif began with a model derived from the spatial literature on voting (Hinich

and Munger 1997)-an approach that he was forced to abandon. Levi had to reconsider a model in which canges in conscription followed diectly fromexpansions in surage. In addition, models posit relationshps among variablesthat render them subject to testing. I Rosenthals chapter, for example, theconditions that me it rational to launch wars suggest relationships betweenthe scal power of elites and the res from ghting, thereby providing ameans for evaluating his explanation.

When models highlight features of the data that hitheto have escapedattention; when they can be contradicted by the evidence; and when they



234 CONCLUSION 235



234 CONCLUSION

value of a critical exogenous variable thus produced a change in the equilibrium

outcome An obvious implication is that propositions derived from a model of the coee market prior to the cold war period could not be tested with data drawnfrom that period The institutions that shaped ade strategies altered; there wasa shi to a new equilibum The behavior of producers therefore altered in thetwo peods, and dierent processes generate prices in inteational markets.It is therefore dicult to test explanations employing data from outside theoriginal sample.

Weingast oers an additional example. Political stability in the nineteenthcentury United States, he agues, was conditioned by the credibility of the

balance rule. Once the rule had been broken, incentives changed, and actorstherefore altered their political sategies. As the counterfactual test of hisargument dramatically illustrates, the model that t behavior in the one periodfailed to t it in the other.

When there are multiple equilibria, the logic that drives stategy choices,given one value of an exogenous variable, may not be the same as the logicthat shapes choices, given other values. When the value of an exogenousvariable alters, people may recompute their best strategies and new strategycombinations may come to form equilibria. Conventional comparative statics,which explores relationships between exogenous and endogenous variables,

therefore cannot be applied to generate propositions about variables that areteae out of the original sample. For the same reasons, investigators maynd it difcult to test their explanations by applying the canons of researchdesign to select additional cases.

One response is to concede the obvious: that analytic naatives are refractoryto some of the most commonly employed reseach methods. They thereforeoften yield outcomes that tend to be "one o and that are relatively rareand infrequent. Many crucial historical events exbit such characteristics.Phrased more positively, by capturing the mechanisms that account for suchproperties, analytic naatives deepen our understanding of why such eventsevade understanding by "normal scientic methods.

A second response is to oer an alteative interpretation of our approach.

Rational choice theory (and in particular the theory of games) oers a theoryof structre: it suggests a way in which structues create incentives that shapeindividual choices and thereby collective outcomes. Insofar as a game yieldsmultiple equilibria, it may be dicult to test its explanatory power But the"force of a game may lie in he properties of strcture that it highlights and in thestrategic problems to which it gives rise. And whereas the specic game may not

be porable, these other attributes may oer opporunities for comparative analysis. They may yield explanations that can be tested in many dierent settings.

Examples of common strategic problems would include the following:Collective action problems Students of class conict (Elster 1985;

Przeworksi 1985), ethnicity (Laitin 1992, 1995; Hardin 1995), and social

CONCLUSION 235

movements (Chong 1991; Kuran 1995; Golden 1996) focus upon such problems, as do students of sectoral politics (Frieden 1991) and conict betweentown and county (Bates 1981; Varshney 1992).

Principal agent problems Resechers into the politics of cental planning (Koai 1995), public sector enterprises (Waterbury 1993), and relations between the legislature and bureaucracy (Ramseyer an Rosenbluth 1993;

Cowhey and McCubbins 1995) concentrate upon the political problems resulting from agency, as do researchers into the inteal politics of politicalparties (Michels 1962; Ramseyer an Rosenbluth 1993; Cowhey and McCub

bins 1995).Problems of credible. commiment A series of recent studies focuses on

the ability or inability of nation-states to bind themselves to inteationalagreements (Eichengreen 1991; Simons 1994; Bates 1997). Researchers havealso focused on problems of credibility in the making of domestic economicpolicy (North and Weingast 1994).

Two-level games Tsebelis's (1990) study of domestic politics in Europeand Putnam's (1988) reserch into interstate policymaing stress the strategicdilemmas arising from the necessity of actors to compete in more than onepolitical arena

Cerain stuctural features of strategic settings also comonly occur, andgame theoretic analyses suggest their signicance for collective outcomes.

These too provide opportunities to focus on diverse cases from a commonvantage point. Such key features would include the following

The power o f the pivot The pivotal decisionmaker is that individual or groupwhose choice decides the outcome3 In his chapter, Bates utilized a commonlyemployed measue of the capacity of individuals to pivot by rendering coalitionswinning, given the rles of the game (the Shapley value; see Shubik 1985);

he employed this measure to analyze the allocation of export rights amongmembers of the ICO Greif's discussion suggested that by creating the podest

the Genoese created an agent that possessed the incentive to side with whichever paty was attacked by another. Because his suppor decides which side winsa conict, he podest is pivotal. Levi's discussion suggested how electoralchanges in the neteenth century changed the identity of the pivotal voter (e.g.,from the elite to the rural properied and urban bourgeoisie), in tu altering thepolicy outcome. Pivots also played a role in Weingast's account of the demiseof the Democratic Party's hegemonic position in antebellum America.

Gatekeeping authori Gatekeeping authority is the power to initiate or to block consideration of proposals within a particular jurisdiction (Shepsleand Weingast 1981; Denzau and Mackay 1983). This authority is typicallyapporioned by institutions and helps to determine the way in which they shape

n a voting contest over a one-dimensional issue space, the pivot is simply the median voter.

n more complex issue spaces however a single median may not exist

236 CONCLUSION CONCLUSION 237



236 CONCLUSION

political outcomes. Gatekeeping authoity conveys the powe to pevent analteation of the status quo. Thus Bates showed how the ules of the ICOenabled the majo poduces to limit pedation by the small, by endowing thefome with sucient votes to block poposals to alte quotas in the ICO.Weingast demonstated how the powe of the Senate to pevent issues fom

being delibeated at the national level helped to keep slavey fom becominga dening issue in the second-paty system. In Schattsneide's pasing, the

powe of the gatekeepe thus helps to account fo the institutionalization of

bias in pol itical systems ( Schattsneide 1960) .

Veto points. Many political systems endow seveal actos with the ight toveto a policy change in a paticula aea (Immegut 1992a,b; Tsebelis 1995).Weingast's chapte exploes the political signicance of such veto points. The

balance ule ta nsfomed the vetoes endowed by the Constitution to the Houseand the Senate into a dual set of vetoes fo both Noth and South. The dual veto

placed self-enfocing limits on the national govement, helping to maintain

the constitutional system of states' ights and fedealismand also helping to peseve slavey. Rosenthal's distinction between paliamenta y and absolutist

egimes hinges on the dieential ability of elites to veto the initiation of wa in the two political systems, wheeas Levi's conce with democatization

emphasizes the changing political locus of veto powe.

Conclusion

The contibutos to this volume have sought to constct analytic naatives andso extact deepe insights fom case mateials. so doing, we have gained a

heightened awaeness of the intimate liks between theoetical easoning andempiical eseach. At least in the wold of stategic behavio, we have leed,

theoy with naative is stonge than theoy alone. And naative with theoyis moe powel than naative alone. Analytic nratives oe a method fo

moving fom the context-ich wold of events and cases to explanations thatae logically igoous, illuminating and insightl, and, if handled with ce,

subject to empiical testing.

References

Bates, Robert H. 1997. Open Economy Politics Princeton, N.J.: Princeton University

Press.

-. 1981. Markets and S tates in Tropical AicaBerkeley: University of Califoia

Press.

Chong, Dennis. 1991. Collective Action and the Civil Rights Movement Chicago:University of Chicago Press.

CONCLUSION 237

Cow hey, Peter F, and Mathew D. McCubbins, eds. 1995. Structure and Policy in Japan

and the United States Cambridge: Cambridge University Press.

Denzau, rthur, and Robert Mackay. 1983. "Gate Keeping and Monopoly Poer of

Comittees. American Joual of Political Science 27 720-62.

Eichengreen, Barry. 1991. Golden Fetters New York: Oxford Universiy Press.

Elster, Jon. 1985. Making Sense of Marx Cambridge: Cambridge University Press.

Frieden, Jeery A. 1991. Det, Development and DemocrcyPrinceton, N. J.: Princeton

University Press.

Golden, Miriam. 1996. Heroic Defeats New York: Cambridge University Press.

Green, Donald P., and Ian Shapro. 1994. Pathologies of Rational Choice Theo NewHaven, Conn.: Yale University Press.

Hardin, Russell. 1995. One for All Princeton, N.J.: Princeton University Press.

Henderson, James, and Richard Quandt. 1971. Microeconomic Theo A Mathematical

Approach New York: McGraw-Hill.

Hnich, Melvin J., and Michael C. Munger. 1997. Analytic Politics Cambridge: Cam

bridge University Press.

· Imergut, Ellen. 1992ealth Politics Cambridge: Cambridge University Press

. 1992b. "The Rules of the Game: The Logic of Health Policy-Making in

France, Switzerland and Sweden. In Structuring Politics Historical Institutionalism

in Compative Analysis, edited by Sven Steinmo, Kathleen Thelen, and Frank

Longstreth. Cambridge: Cambridge Unversity Press.

Koai, Jnos. 1995. The Economics of Shortage Amsterdan: Norh-Holland Press.

Kuran, Timur. 1995. Private Truths, Pulic Lies Cambridge, Mass.: Harvard University Press.

Laont, Jean-Jacques, and Jean Tirole. 1987. "Comparative Statics of the Optimal

Dynaic Incentive Contract. European Economic Review 31 901-26.

Laitin David. 1992. Language Repertoires and State Construction in Africa New York:

Cambridge University Press.

. 1995. "National Revivals and Violence. Archives Europenes de Sociologie

36: 3-43.

Michels, Rober. 1962. Political Parties New York: Collier.

North Douglass, and Barry Weingast. 1994. "Constitutions and Commitment. In

Moneta and Fiscal Policy, vol. 1 Crediiliy, edited by Torsten Persson and Guido

Tabellini. Cambridge, Mass. MIT Press.

Przeworksi, Adam. 1985. Capitalism and Social Democracy Cambridge: Cambridge

University Press.Putnam, Robert. 1988. "Diplomacy and Domestic Politics: The Logic of Two-Level

Games. Inteational Organization 42 427-60.

Ramseyer, Mark J., and Frances Rosenbluth. 1993. Japan' Political Marketplace

Cambrdge, Mass.: Harvard Unversity Press.

Schattsneider, E. E. 1960. The SemiSovereign People New York: Holt, Rinehart and

Winston.

Shepsle, Kenneth A., and Barry R. Weingast. 1981. "The Institutional Fondations of

Comittee Power. American Political Science Review 81 85-104.

Shubik, Marin. 1985. Game Theo and the Social Sciences Cambridge, Mass.: MIT

Press.

238 CONCLUSION



CONCLUSION

Simons, Beth. 1994. W ho Ad justs. Pr inceton, N.J.: P rinceton Unver sity P r ess.Sutton

John. 1991. Sunk C ost s and M ark et Str uct ur e Camb r idge, Mass. MIT P ress.

Tsebes, Geor ge. 1990. N ested Games. Be rkeley Unve rsity of Calif o ia Pr ess. -.

-. 1995. "Decision Makng in Political Systems V eto Playe rs in Pr esidential-sm, Par liamenta rianism, Multicame ralism, and Multipa rtyism. British J oual i Pol itical Science 25: 289- 325.

0

Va rshney, Ashutosh. 1992. Be yond U rban Bias. London Cass.W aterbur y, John. 1993. E xposed to Innumer ble Delusions. Camb rdge: Camb r d e

Univer sity P r ess.g

Appendu__

THE PUROSE of this appendix is to provide an example of a full-inforationgame in extended form and to inoduce the notions of (subgame-perfect)equilbrium, the equilibrium path, and credible treat. The stylized exampleis designed to meet the iediate needs of the reader. For ther informatonabout game theory, the reader is advised to consult more extensive treatments,such as Morrow ( 1994.

Consider gure A., which displays the peacetime army game. The homenation (player H decides whether to have a large or a small army (Lor S, andan opponent (player 0) decides between maintaining peace and intiating war ( orW).

ayos are determined as follows. For the home country:

a army costs

b Life without being conquered is worth

c. Defendig i war costs , if there exists a large army.

d. no large army exists, then is overrun if 0 nitiates war, and receives the

payoO.

For the opponent

a veunning in war is worth

b Iitiating war but failig to overrun costs

c. The payo is otherwise.

w

pL

H

s

FIGUR Al The peacetime army game.

PoH

1 -2

15

1

15



Robert H. Bates is the Eaton Professor of the Science of Goveent and Fac-

ulty Fellow of the Institute of Inteational Development at Harvard University.

Avner Greif is Associate Professor of Econoics at Stanford University

Margaret Levi is Professor of Political Science and the Hay Bridges Chair

in Labor Studies at the University of Washington, Seattle

Jean-Laurent Rosenthal is Professor of Economics at the University of Cali-

foa, Los Angeles

Barry R Weingast is Senior Fellow at the Hoover Institution and Ward C

Krebs Family Professor of Political Science at Stanford University

bates analytic narratives_intro - cap 5

Documents