Public Disclosure Authorized

Public Disclosure Authorized

185L

tee

/'qL

World Bank Discussion Papers

Military Expenditure

Public Disclosure Authorized

Public Disclosure Authorized

and Economic
DLevelopment
A Symposium on Research Issues
edited by
Geoffrey Lamb
with Valeriana Kallab

Recent

World

Bank

Discussion

Papers

No. 125

HousingReformin SocialistEconomies.Bertrand Renaud

No. 126

AgriculturalTechnology
in Sub-SaharanAfrica: A Workshopon ResearchIssues.Suzanne Gnaegy andJock R
Anderson, editors

No. 127

UsingIndigenousKnowledgein Agricultural
Development.D. Michael Warren

No. 128

Researchon Irrigaticnand DrainageTechnologies:FifteenYearsof WorldBank Experience.Raed Safadiand

Herve Plusquellec

No. 129

Rent Controlin DevelopingCountries.Stephen Malpezzi and Gwendolyn Ball

No. 130

Patternsof DirectForeignInvestmentin China. Zafar Shah Khan

No. 131

A New Viewof EconomicGrowth:FourLectures.Maurice FG. Scott

No. 132

AdjustingEducational
Policies:
Conserving
Resources
WhileRaisingSchoolQuality.Bruce Fullerand AkliluHabte, editors

No. 133

Letting GirlsLearn:PromisingApproachesin Primaryand SecondaryEducation.Barbara Herz, K. Subbarao,

Masooma Habib, and Laura Raney

No. 134

ForestEconomicsand PolicyAnalysis:An Overview.William F. Hyde and David H. Newman, with a contribution

by Roger A. Sedjo

No. 135

A StrategyforFishe;ries
Development.Eduardo Loayza,in collaboration with Lucian M. Sprague

No. 136

StrengtheningPublicServiceAccountability:
A ConceptualFramework.Samuel Paul

No. 137

DeferredCostRecoveryforHigherEducation:StudentLoan Programsin DevelopingCountries.Douglas Albrecht

and Adrian Zidemian

No. 138

Coal Pricingin China:Issuesand ReformStrategy.Yves Albouy

No. 139

PortfolioPe!formance
of SelectedSocialSecurityInstitutesin Latin America.Carmelo Mesa-Lago

No. 140

SocialSecurityand Prospectsfor
Equity in Latin America.Carmelo Mesa-Lago

No. 141

China'sForeignTradeand ComparativeAdvantage:Prospects,
Problems,and PolicyImplications.AlexanderJ. Yeats

No. 142

Restructuring
SocialistIndustry:Poland'sExperiencein 1990. Homi J. Kharas

No. 143

China:IndustrialPoliciesfor
an Economyin Transition.Inderjit Singh

No. 144

ReformingPrices:The Experienceof China, Hungary,and Poland.Anand Rajaram

No. 145

DevelopingMongolia. Shahid Yusuf and Shahidjaved Burki

No. 146

Sino-Japanese
EconomicRelationships:Trade,DirectInvestment,and FutureStrategy.Shuichi Ono

No. 147

The EffectsofEconomicPolicieson AfricanAgriculture:FromPastHarm to FutureHope.William K. Jaeger

No. 148

The SectoralFoundationsof China'sDevelopment.Shahidjaved Burki and Shahid Yusuf, editors

No. 149

The ConsultingProJ;sssion
in DevelopingCountries:A StrategyforDevelopment.Syed S. Kirmani
and Warren C. Baum

No. 150

SuccessfulRural FinanceInstitutions.Jacob Yaron

No. 151

TransportDevelopmentin SouthernChina. Clell G. Harral, editor, and Peter Cook and Edward Holland,
principal contributors

No. 152

The UrbanEnvironmentand PopulationRelocation.Michael M. Cernea

No. 153

FundingMechanismsforHigherEducation:FinancingforStability,Efficiency,and Responsiveness.
Douglas Albrecht
and Adrian Ziderman
(Continued on the inside back cover.)

185

World Baik Discussion Papers

Military Expenditure
and Economic
Development
A Symposium on Research Issues
edited by
Geoffrey Lamb
with ValerianaKallab

The World Bank

Washington, D.C.

Copyright C) 1992
The Intemational Bank for Reconstruction
and Development/THE WORLDBANK
1818 H Street, N.W.
Washington, D.C. 20433, U.S.A.
All rights reserved
Manufactured in the United States of America
First printing December 1992
DiscussionPapers present resultsof country analysisor research that is circulated to encourage discussion
and comment within the development community. To present these resultswith the least possible delay, the
typescript of this paper has not been prepared in accordance with the procedures appropriate to fornal
printed texts, and the World Bank accepts no responsibilityfor errors.
The findings, interpretations, and conclusionsexpressed in this paper are entirely those of the author(s) and
should not be attributed in any manner to the World Bank, to its affiliatedorganizations, or to members of
its Board of Executive Directors or the countries they represent. The World Bank does not guarantee the
accuracy of the data induded in this publication and accepts no responsibility whatsoever for any
consequence of their use. Any maps that accompany the text have been prepared solelyfor the convenience
of readers; the designationsand presentation of material in them do not imply the expressionof any opinion
whatsoever on the part of the World Bank, its affiliates,or its Board or member countries conceming the
legal statusof any country, territory, city, or area or of the authorities thereof or concerning the delimitation
of its boundaries or its national affiliation.
The material in this publication is copyrighted. Requests for permission to reproduce portions of it should
be sent to the Office of the Publisher at the addressshown in the copyright notice above. The World Bank
encourages dissemination of its work and will normally give permission promptly and, when the
reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom
use is granteddtrough the Copyright ClearanceCenter, 27 CongressStreet, Salem,Massachusetts01970, U.S.A.
The complete backlist of publications from the World Bank is shown in the annual Index of Publications,
which contains an alphabetical title list (with full ordering information) and indexes of subjects, authors, and
countries and regions. The latest edition is availablefree of charge from the Distribution Unit, Office of the
Publisher, Department F, The World Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A., or
from Publications, The World Bank, 66, avenue d'[6na, 75116 Paris, France.
ISSN: 0259-210X
Geoffrey Lamb is chief of the Public Sector Management Unit in the Wodd Bank's Europe and Central
Asia and Middle East and North Africa Regions Technical Department. ValerianaKallab was an editorial
consultant for this volume.
Library of Congress Cataloging-in-Publication

Data

Military expenditure and economic development: a symposium on

research issues / edited by Geoffrey Lamb with Valeriana Kallab.
p.
cm. - (World Bank discussion papers ; 185)
Includes bibliographical references.
ISBN 0-8213-2289-3
1. Armed Forces-Appropriations and expenditures. 2. National
security-Economic aspects.
I. Lamb, Geoffrey, 1944II. Kallab, Valeriana. III. Series.
UA17.M55
1992
355.6'22-dc2O
92-34610
CIP

Foreword

The WorldBank's businessis development. Traditionally, therefore, it has not

concerneditself with matters relating to nationalsecurity and the military. In
recent years, however, two trends have combinedto raisereneweddebatewithin
Bank member countries, and within the Bank itself, about levels of military
expenditure(and in some countriesalso military industries)and their impacton
economicdevelopment.
First, global politicalchangesand the relaxationof Cold War tensionshave
held out a prospect-still all too elusive, in the light of post-ColdWar nationalist
and ethnic tensions-of decreased need for military outlays, especiallyin areas
where regional conflictalso had strong Cold War dimensions.
Second, both public and private resources for economicdevelopmentwill
continue to be tightly constrained in most developing countries, and in the
countries in transition from commandeconomies,throughoutthis decade. For
governments and for the Bank, this prompts a hard look at the adequacy of
public money committed to development, and in many countries is already
producingan increasedreadinessto search for economiesin non-developmental
expenditure-includingthe military.
The Bank's leadership has underlined the importance of working
collaborativelywith member governments, and with the IMF, to explore the
fiscal and developmentalissues involved. Where our borrowersare embarking
on reductions in military expenditureor production and seek our involvement,
the Bank also standsready to assist-for example,with programs to facilitatethe
integrationof former militarypersonnel and facilities into the private economy
and, on a market-driven basis, the phasing-downor conversion of military
industries.
Effective action in the sensitive area between national developmentand
national security demands careful appreciationof the complexities. Military
expendituredata are almost always incompleteor misleading. The relationship
between military expenditureand economicgrowth is neither uniform across
countries nor monotonic within them. Converting military industries poses
Hi.

difficult issuesof efficiencyand industrialpolicy, as well as social and political

problems. And, of course, the internal,regional and internationalfactorsdriving
military expenditureare extremelydiverse.
These complexitiespromptedthe Bank to commissionthe survey papers in
this volume, as a contributionto its own thinkingand to the wider debate. Given
the nature of the Bank's interest, the papers focus specificallyon the economic
aspects of military expenditurein developingcountries. Initial versions were
discussedat an internationalsymposiumorganizedby the editor of this volume,
Geoffrey Lamb, in Washington in December 1990. They have since been
extensivelyrevised, in part to take some accountof the rapidity of international
developmentsin 1991and 1992. This activitywas supportedby a researchgrant
from the Bank's ResearchAdministrator.

HarinderKohli
Director
Europe and C0entral
Asia, Middle East and North Africa RegionsTechnicalDeparttment

iv

Acknowledgements

The Editor wishes to acknowledgefinancial support from the World Bank's research
budget, which underwrote the commissioningof these papers and the international
symposiumat which initial drafts were discussed. He is grateful to the discussantswho
gave many valuable comments on the draft papers, and most particularly to the
contributorsthemselves:
NicoleBall, DavidsonSommersFellow at the OverseasDevelopmentCouncil;
Michael Brzoska, lecturer in International Relations in the Political Science

Departmentof the University of Hamburg and Research Fellow on Wars,

Armaments,and Development;
Saadet Deger, Senior Researcher at the Stockholm International Peace
ResearchInstituteand leader of the SIPRIWorldMilitaryExpenditureProject;
Somnath Sen, Senior Lecturer at the University of Birmingham, United
Kingdom, and Senior Researcher at the Stockholm International Peace
ResearchInstitute; and
Robert L. West, until his death in 1991Professor of InternationalEconomic
Relations at the Fletcher School of Law and Diplomacyat Tufts University.
All of those involvedin this volume would like to dedicate it to the memory of
Robert West, colleagueand friend, 1925-1991.

GeoffreyLamb

Contents

Chapter 1. MILITARYEXPENDITUREDATA FOR DEVELOPINGCOUNTRIES:

METHODS AND MEASUREMlENT
by Somnath Sen
1

Introduction.................
Definitionsand Sources.
Issues
OpportunityCosts.14
ConcludingSuggestions.16

2
9

Chapter 2. PATTERNSAND TRENDSIN THE MILITARYEXPENDITURES

OF DEVELOPINGCOUNTRIES
by Robert L. West
Itroduction .19
The Pattern of Military Expendituresin 1980.21
Trends in MilitaryExpenditures, 1970-85.24
StatisticalAnnex.28

Chapter3. MILITARY EXPENDITUREAND ECONOMICDEVELOPMENT:

ISSUES AND DEBATES
by Saadet Deger
Introduction............................................
The Trade-OffBetweenDefense and EconomicGrowth:
The TheoreticalFramework.37
The EconometricEvidence.42
EconomicDevelopmentand Military Expenditure.48
ConcludingRemarks.50

vii

35

Chapter4. ADJUSTDN4G
TO REDUCTIONS IN MELITARYEXPENDITURE
AND DEFENSE PROCUREMENT
by Nicole Ball
Introduction........................................
The Need for Adjustment ...................................
Industrial Production . ...................................
Manpower ...................................
...
VariableEffects ......................................
Two Paths: Conversionand Diversification........
................
Conversion ...................................
...
Diversification........................................
Conversionversus Diversification............................
Transferabilityof SecurityForce Personnel .......
.................
Transferabilityof Technical Skills .........
...................
Limited Spin-off.......................................
Factors Conditioningthe Successof AdjustmentPrograms .....
..........
The Economy .............
...........................
Dependenceon the DefenseSector .........
..................
Defense Specificityof Product or Technology ....................
Nature of the Military Market ...........
...................
Marketabilityof AlternativeProducts ........
..................
Flexibilityof Companyand Manpower........
.................
Role of Governments. ...................................
Conclusion.............................................
Appendix: Recent Adjustmentsin China's Defense Sector .....
.........

53
55
55
56
58
58
59
59
60
60
60
62
63
63
63
64
65
66
66
67
67
70

Chapter5. MILITARYTRADE, AID, AND DEVELOPING-COUNTRY

DEBT
by Michael Brzoska
Introduction........................................
The Data Problem ........................................
Foreign 1'rade Data .....................................
FinancialData ........................................
Other Sources ........................................
InternationalTrade in Arms: An Overview ........................
Trend ...................................
Phasesand Structure ...................................
Comparisonsof Current Magnitude .........
..................
The ChangingFate of Military Aid ..........
...................
Types of Aid.
.........................................
U.S. MilitaryAid ......................................
Soviet Military Aid . ....................................
Other Military Aid . ....................................
Summary.
...........................................
Financingof Arms Imports and Indebtedness .......................
Debt to the U.S. GovernmentDue to Arms Imports .....
...........
EstimatingDebt to the U.S.S.R. Due to Arms Imports .....
..........
Other Suppliers' Direct Military Credit ........................
Estimatesof Total Direct Military Credit .......................
viii

79
80
80
81
81
82
82
82
87
88
88
89
91
93
93
93
95
95
97
97

Opportunity-CostEstimatesof the Military Debt .......

Correlates of the Military-RelatedDebt ........................
The Future of Arms Transfers .............................

.............

98
102
104

Chapter 6. DETERMINANTS OF MILITARYEXPENDITURE

IN DEVELOPING COUNTRIES:
REVIEW OF ACADEMIC RESEARCH
by Robert L. West
113

Introduction............................................
...................
Problemsof Data ..........
Reliabilityof ExpenditureData ...........................
Testing Methods ..........
...................
Problems of EconometricTechnique.........................
AppropriateModelsfor ExplainingPublicExpenditures ....

........

The Determinantsof Military Expenditurein DevelopingCountries ....

GeostrategicConsiderationsand the SecurityEnvironment ....
Arms Race Models ..................................
Threat Perceptionand Inter-StateConflict .....................
The Politics of the BudgetaryDecision-MakingProcess ....
Evidenceof BudgetaryTrade-offs .........................
ExternalPatrons: The Influenceof Donors and Suppliers ....
The Influenceof Financialand EconomicConstraints................
Multi-VariableExplanationsof Military Expenditure ................

.....
........
...........
.........

DemonstratingDifferencesAmong Sub-Groupsof DevelopingCountries ....

DifferencesAttributableto DomesticDefense Industries ....
DifferencesAttributableto FinancialConstraints...................

114
115
116
116
118

..........

119
119
120
122
123
126
128
131
134
.. 137
138
139

Appendix. BACKGROUNDNOTE ON MILITARY EXPENDITURE:

SOURCES AND PRICE CONVERSIONPROCEDURES
147
by Robert L. West ......................................

Li:

MILITARY EXPENDITURE DATA

FOR DEVELOPING COUNTRIES:
METHODS AND MEASUREMENT
Somnath Sen

Introduction
Formal research on the impact of military
expenditure on economic growth and socioeconomicdevelopmenthas been bedeviledby
the problems of data. The quantificationof
the effect of militaryexpenditureon economic
developmentis still a controversialissue, and
without a proper data base, the final
conclusions remain tentative. For various
reasons, including secrecy due to national
security concerns as well more fundamental
statistical problems, transparency in military
expendituredata, particularlyin less developed
countries, is limited. It is difficult to
appreciatewhy expendituredata should be of
major concern in the frameworkof national
security. Rather, details on activeforces and
weapons stocks could have a direct effect on
threat perceptions and trigger so-called arms
races. Data on expenditure is of more
relevanceto domesticconcerns,particularlyin
the evaluationof opportunitycosts.

Note: The author wishes to express special thanks to

Ravi Kanbur, Dan Gillick,and GeoffreyLamb, as well
as to the participantsin the World Bank Symposiumfor
commentsand helpful suggestions.

.-.-

Perceptionsaboutthe economicburden of
militarizationare based on statistics;hence it
is particularly importantthat these be of the
right order of magnitude and show
approximatelycorrect trends. To take one
example: according to the Stockholm
InternationalPeaceResearchInstitute(SIPRI),
military expenditurein the Third World rose
rapidly, almost without interruption, in the
1970sand in the first half of the 1980s;it has
been declining in the Third World since the
mid-1980s. As a proportionof world military
expenditure,the share of the Third World rose
from 6 per cent in the late 1960s to 18 per
cent in the mid-1980s,after which it fell to 15
per cent in 1989. Currently, over 4 per cent
of aggregate Third World gross domestic
product (GDP) is spent on defense. This
economic burden is higher than in most
industrialcountries. Though defensespending
has fallen, except perhaps in Latin America,
this declinehas not been a result of any arms
control agreementsor any seriousresolutionof
disputes. Rather, it has been a product of
economicproblems. Hence the potential for
expansion is ever present when economic
growtheventuallypicks up (see Deger and Sen
~~~~~~~~~~1990).
Such conclusions can only have
*r

credibility if the data base is structured on

hard information.

MilitaryExpenditureand EcanomicDevelopment

It is widely believed that military

expenditure(hereafter called MILEX) data is
of low quality and based on relatively 'soft'
information. Public discussionis negligible-often nonexistent. There is no clear-cut
distinctionbetweenwhat is the jurisdictionof
military expenditure and what lies in the
domain of internal security spending. In
countries where the influenceof the military
establishment is pervasive, often for good
reason, the distinctionbetween military and
civilian budgetary allocations is even less
distinct. Scheetz(1990) puts it scathingly:
The public and their electedrepresentatives
must someday arrive at a position where
defense questions (especially budgetary
aspects)are a matter of public debate, not
for back-room secret discussions. After
all, countries' potential enemies generally
are quite awareof each other's arms stocks
and operational capabilities. The silence
on budgetaryquestionsis defendedby the
various armed forces as a question of
secrecy. Usually it is simplyto keep their
citizensin the dark.
In evaluating the quality of defense
spending data in less developed countries,
three issues should be kept in mind. First,
questionsregardingreliabilityhave been raised
about all categories of social and economic
data in these countries. In particular,
disaggregated data--say, for components of
GDP such as agricultural output--are often
rough estimates. In highly inflationary
economies, price deflators are usually
approximate. This is partly due to the
structural problemsof underdevelopment:the
collectionand disseminationof informationis
often related to economic development.
Financial informationabout military activity
may be more sensitive but is not necessarily
fundamentallydifferent frromother types of
information. The difference may be one of
degree rather than of kind. Second, for some
countries, particularly in Latin America,

budgetary details are available, provided

analysts can dig them out with care. Some
countriesdo not have a general accountunder
the Ministry of Defense, while others
appropriate funds in other ministries for
purposes that clearlyhave a military function
(such as spendingon the paramilitary,or even
arms imports). Third, secrecy about items of
militaryspendingis not necessarilyconfinedto
developingeconomies. An exampleis given
by Neild (1981) about nuclear weapons
purchase and installation in the United
Kingdom: "Recently, expenditure of 1,100
million on Chevaline, the multiple warhead
developedby Britain for her Polaris missiles,
was concealed,making it the largest matter in
recent years not to have gained mention in the
annualdefenseestimates." However,the level
of independentscrutiny is much lower in less
developedcountries, making it more difficult
to identify expost such missing items.
The purpose of this paper is to evaluate
the problems with military expenditure data
and the methodology employed by the
collecting institutions (for useful background
information, see Appendix, pp. 147-51). A
primary objective is to identify gaps and to
suggestresearch areas where positive changes
can be made. The paper first describes the
alternativedefinitionsthat have been employed
and the various internationalsources for the
data. It next deals with four major issuesthat
need to be analyzedif a proper evaluationand
use is to be made of the information. The
paper then analyzes the use of data in the
evaluationof the opportunitycosts of military
spending, and it concludes with suggestions
for further research.

Defriitionsand Sources
The main purpose of military expenditure
data, whatever the source, is to provide an
easily identifiablemeasure, over time, of the
scale of resources absorbed by the military in
variouscountries. This objectiveof producing

Military ExpenditureDatafor DevelopingCountries: Methodsand Measurement

an input measure is not to be confused with

that relating to the output of the defense
sector, such as military strength or national
security. Expendituredata is only indirectly
and informallyrelated to military capability,
although it can be utilized to judge the
perceptions of governments towards defense
and military strength. This makes it more
amenable to economic analysis--although
formidable problems remain regarding a
proper measureof opportunitycosts.
Two basic definitions exist for MILEX.
The IMF definition, which is the most
explicit, describesdefenseas covering:
all expenditure, whether by defense or
other departments,for the maintenanceof
military forces, includingthe purchase of
militarysupplies and equipment(including
the stockpilingof finisheditems but not the
industrial raw materials required for their
production), military construction,
recruiting, training, equipping, moving,
feeding, clothing and housingmembersof
the armed forces, and providing
remuneration, medical care, and other
services for them. Also included are
capital expenditures for the provision of
quartersto families of military personnel,
outlays on military schools, and research
and development serving clearly and
foremostthe purposesof defense.Military
forces also include paramilitary
organizations such as gendarmerie,
constabulary, security forces, border and
customs guards, and others trained,
equippedand availablefor use as military
personnel. Also falling under this category
are expenditures for purposes of
strengtheningthe public services to meet
wartimeemergencies,training civil defense
personnel, and acquiring materials and
equipment for these purposes. Included
also are expendituresfor foreign military
aid and contributions to international
military organizationsand alliances. This

Sen

category excludes expenditures for nonmilitary purposes, though incurred by a

ministry or department of defense, and
any paymentsor services providedto war
veterans and retired military personnel.
The NATO definitionis similar--thoughmore
vague:
the following items are included: all
current and capital expenditure on the
armed forces, in the running of defense
departments and other government
agencies engaged in defense projects as
well as space projects; the cost of
paramilitary forces and police when
judged to be trained and equipped for
military operations; military R&D, tests
and evaluation costs; and costs of
retirement pensions of service personnel,
includingpensions of civilian employees.
Military aid is includedin the expenditure
of the donor countries. Excluded are
items on civil defense, interest on war
debts and veterans' payments.
There are two basic differences--one a
matter of interpretation,the other substantive.
The status of foreign military aid received is
rather unclear in the IMF definition, and it
may or may not be included in country
submissionsto the Fund. More important,the
IMF definition, based on national account
practices, excludes military pensions, since
this is a transfer payment. Yet when one
calculatesthe burden of defensespending, the
weight of military pensions is borne by the
civilian economy. The same problem arises
regarding some war-related expendituresfor
veterans' payments and interest on war debt.
Both definitions exclude these and have the
merit of consistencyin terms of current costs
in an aggregativeframework. However, in a
two-sector model of an economy, which
characterizes the trade-offs between the
militaryand civiliansectors, such expenditures
by the military do have real costs. This is

MilitaryExpenditureand EconomicDevelopment

particularly true for the developing world,

where over thirty major conflicts raged in
1989.
Increasingly, as discussedlater, there are
demands for more disaggregated data to
conduct proper economic analysis. The
UnitedNations (1986)gives the most detailed
disaggregatedcategorizationin terms of three
major groups: operating costs, procurement
and construction, and research and
development. Member countriesare asked to
fill out details of financial expenditurein each
category and to submit ithese to the U.N.
Reduction of Military Budget Programme.
The U.N. groupingsare as follows:
A. Operating Costs
(1) Personnel: a) conscripts, b) other
military, c) civilian;
(2)

Operations

and

Maintenance:

a) current-usematerial, b) maintenanceand
repairs, c) purchased services, d) rent,
e) others.
B. Procurement and Construction
(1) Procurement: a) aircraft and engines,
b) missiles, including conventional
warheads,c) nuclearwarheadsand bombs,
d) ships and boats, e) armored vehicles,
f) artillery, g) other ordnance and ground
force arms, h) ammunition,i) electronics
and communications, j) non-armored
vehicles, k) others;

Operationsand Maintenance (whichincludes

civilian pay); Weapons Procurement;
Construction; Research, Development,
Testing, and Evaluation (RDT&E). The
first two roughly measure operating
expenditure, while the remainder measure
'investment' in forces. The disaggregation
employedby NATO in revealingfinancialdata
corresponds to expenditures on personnel,
equipment,infrastructure,and other operating
expenditure (which includes R&D not provided separately).
The U.N. definition of disaggregationis
preferable,since it clearlydistinguishesamong
the three functional categories that have
specific opportunity costs: manpower and
operationalitems, investmentin weapons and
current assets, and investmentfor the future.
Discussionson resource re-allocationshould
be based on such divisions, even though

financial resources are largely fungible. The

fundamental distinction that must be
maintained in data reporting is that between
'operating' and 'investment' (or capital)costs.
The former will includeall personnel(military
and civilian),pensions, as well as operations
and maintainence expenditures. The latter
includes procurement and additional defense
industrialexpenses(suchas subsidies),as well
as military research expenditure. When
discussingeconomiceffects, the outcome may
be different dependingon which category of
MILEX-operating or investment--oneuses.

(2) Construction: a) airbases, b) missile

sites, c) naval bases, dl) electronics etc.,

e) personnel, f) medical, g) training,
h) warehouses and depots, i) command,
administration,j) fortifications,k) shelters,
1)lands, m) others.
C. Research and Development
(1) Basic and applied;
(2) Development, testing, and evaluation.

An alternative, following the U.S.

grouping, is to have: Military Personnel;

Having an appropriate definition clearly

does not assure strict adherence to it. Most
internationalsources report official data with
modest adjustments.
Ultimately, the
responsibilityto reveal carefullytheir defense
expenditureand its constituentparts rests with
govermments. There are many reasons why
governmentsmay not be honest in revealing
their data. Defense adds to security, protects

the economy against enemy actions, and in

some cases enhancesstability,.which may be
required for development. These may add up

MilitaryExpenditureDatafor DevelopingCountries: Methodsand Measurement

to societal welfare. On the other hand, the

withdrawalof resourcesfrom more productive
spending reduces growth and social welfare.
Depending on the nature of society's social
welfare function, as reflected in government
preferenceand in terms of public choice,there
may be a tendencyto overstateor under-report
military expenditure. The process is affected
by the position of the military in society, the
signals sent out to adversaries and allies, as
well as technical abilityto monitorall avenues
of expenditure. The contamination of
economic data by the political framework
within which military decisions are taken
creates major problems for evaluating and
filtering the informationavailable.
Six methods are allegedly used to
misrevealdata: doublebookkeeping,excessive
aggregation, extra-budgetary finance,
unidentifieduse of foreign exchange,military
aid, and defense-relateddebt burden (see Ball
1984; Deger 1986). Double bookkeeping
inivolvesthe use of two (or multiple) sets of
accounts,so that publishedbudgetarystatistics
are somewhat different from those used
initernallyby civil servants and politicians.
Excessive aggregation(for example, a single
line entry) impliesthat independentscrutinyis
impossible. Defense-relatedspendingin other
ministries (connected with social security,
housing, education, health, developmental
expenditures)is then also difficult to extract.
Extra-budgetary finance (including offshore
accounts) is particularly relevant when the
military is heavily involved in the civilian
economy.
Unidentified use of foreign
exchange can occur when a part of
governmentexportearningsis not enteredinto
officialaccountingsystemsand is then used to
import weapons. Military aid may not be
reported fully, or it may be obfuscatedwith
economic aid. Interest paymentson militaryrelated debt are invariably difficult to trace
from the recipient country's budget.
Hlowever, it should be noted that donor
countries that provide military assistance

u m -----------.. ............--------------1.......
----

Sen

usually keep better records, and that in

principleit is possible to gather these types of
information. The recent example of Egypt's
repudiation of its military debt owed to the
United States after the Gulf crisis (Sen 1991)
demonstratesthat the values of aid and debt
are known-at least for some countries.
Unless detailed records are studied, it is
not possible to judge the extent of
misrevelation of the data. For most datacollecting organizations, such information
(i.e., the accounts of all the relevant
ministries)simplyis not available. The World
Bank and the IMF have better access,but the
extent of their coverage is not known to
outsiders. A classic examplewas the World
Bank study of Argentina(World Bank 1985),
which showed that during the 1970s actual
militaryspendingwas about50 per cent higher
than that revealed as defense-related
expenditure. The Report claimedthat:
On a number of accounts, central
governmentdefense expendituresthrough
the end of 1982, as shown in official
statistics, may have been understated.
Reportedly,most military constructionand
purchases of equipment were classified
under economicdevelopment,rather than
defense.
Similarly, some personnel
related military costs, such as Treasury
transfers to the military retirement fund,
the construction of military housing,
health care for the military and
dependents, etc., were counted as social,
rather than defense, expenditures.
Finally, some defense expenditures
financed out of extra-budgetarymilitary
resources (special funds, foreign loans,
etc.) may not have been recorded in the
budget (World Bank Country Study:
Argentina, 1985, p.46).
The three international institutions that
regularly publish time series data on military
expenditurein developing countries are: the

MilitaryExpenditureand EconomicDevelopment

U.S. Arms Control and DisarmamentAgency

(ACDA), which publishes World Military
Expenditure and Arms Transfers; the
InternationalMonetary Fund (IMF); and the
Stockholm International Peace Research
Institute (SIPRI), which publishes its data in
SIPRI Yearbook. World Armaments and
Disarmament(see also Deger and Sen 1991).
The ACDA's MILEX data, for "noncommunist countries . . . are generally the
expenditures of the ministry of defense."
Some effort is made to exclude the costs of
internalsecurity when these are known. The
primary sources are those of the files of U.S.
government departments, particularly the
Agencyfor InternationalDevelopment,as well
as the IMF. Details of U.S. governmentreported raw data are not available. SIPRI
data is unique in the sense that it uses open
sources alone, and this lends it a rather strong
credibility. The primary sources are national
budgets (for "large" countries and major
powers) and IMF data (for "small" countries
or those not considered of strategic
significance). Extensive manipulations are
made on the primary data, particularly to
include estimates of military pensions,
spending on paramilitary forces, and the
appropriate allocation of foreign aid. A
specific, thorny question concerns the
appropriatepaymentsmechanismfor imported
arms: cash, aid, barter, compensatorytrade
agreements, and debt-creation. Ad hoc
methodsare used to sort out this problem, but
it remains unavoidable. The nature of IMF
data in the GovernmentW
Finance Statistics
Yearbook is well known and depends on
governmentsubmissions. The only important
point to note is that the Fund provides an
element of disaggregation. However, its
capitalcosts probablydo not include weapons
purchase, since the revealed data does not
resemble anything similar to independent
informationfor many countries. This category
therefore possibly contains items like
construction.

With a few notable exceptions, economic

data are mostly taken from the IMF and the
World Bank. Information on non-member
countriesis collectedfrom nationalsourcesby
SIPRI. In the cases of oil exporters in the
Gulf, ACDA uses its own method of
estimatingGNP. SIPRI providesthree annual
series: MILEX in current prices; MILEX in
constantprice and U.S. dollars for a specific
year (occasionallyupdated); and MILEX as a
share of GDP. ACDA provides a numberof
series: MILEXin constantdollars; MILEXin
current dollars (both series change in every
annual issue due to changes in the base year
measurement);and MILEX as shares of GNP
and central government expenditures. It is
difficult to do long-term time series analysis
with ACDAinformation,but the various series
can be chainedwith some effort.
Finally come the questions of how much
distortion there exists (see Blackaby and
Ohlson1982; Brzoska 1981; Goertz and Diehl
1986) and whether the various sources are
roughly comparable. Amonganalystswho do
formal models with the data, there is a
pessimistic view and an optimistic view.
Scheetz (1990) compares his own budgetary
search data with international sources and
claims: "If one comparesthe Chilean MILEX
figures . .. with the internationaldata sources
(U.S. Arms Control and Disarmament
Agency, International Monetary Fund,
International Institute for Strategic Studies,
and the Stockholm International Peace
Research Institute) the differences are
alarming." On the other hand, Deger (1986)
states that: "the trends observed for defense
spending-whether for major individual
countries (in the developing world), or for a
cross-sectionof countriesat various periods of
time or in different geographicalregions--are
broadlyconsistentamongdifferent institutions
entrusted with preparing the data." By
looking at the series appropriately
transformed, casual empiricismsuggests that
there are some differencesbetween SIPRI and

Military Expenditure Data for Developing Countries: Methods and Measurement

Sen

ZIV.

ACDA data. However, regressing one series

on another,for a numberof countries,reveals
thaltthe trends do tend to be similar, although
individual country variations could be
significant. A generalconclusionis that crosssectional studies, or those using pooled time
series/cross-section data, can get relatively
robust conclusions. For countrystudies, more
care is needed, unless one is certain of the
reliability of the source, and, at the least,
extensive sensitivity tests are required by
comparing the econometric results from
different data sources.

industrial investment goods, whose

productivity is particularly high in less
developed countries. Overall, they divert
scarce foreignexchangeawayfrom competing
needs.

A brief discussion on arms trade data

follows(for a detailedevaluation,see Brzoska
and Ohlson 1987). SIPRI arms imports data
are trend indicator values, which reflect the
trends in major weaponstransfersto the Third
World. These values are constructed from
quantitativeinformationon weaponstransfered
betweencountries, multipliedby a fixed price
vector (dollar values in 1985 prices); they do
not represent financial transactionsor monies
actually paid. The trade is conducted in
various ways--sales,barters, gifts, creationof
debt, and compensatory trade agreements
(CTAs) such as offsets. The data is not
designedto show the economicburden or the
opportunity costs of such transfers. Its main
puipose is to identifyaggregatetrends in terms
of an indexthat is readily understood.

3. Debts are incurred that may require

servicing. It is commonlybelievedthat such
debts, incurredon an officialbasis for political
reasons, are not paid back. However, until
they are canceled or repudiated, interest
payments must continue--and this has an
economic cost like any other form of debt
servicing.

However,rising (or falling)trends in Third

World arms importsdo have indirecteconomic
implications, and these can be unraveled by
using supplementary economic information
(Sen 1991). There are a number of ways in
which the data on trends can be interpretedto
show economiccosts to the recipientsin less
developedcountries.
1. When arms are actuallypaid for, even
at concessional prices, the importers incur
costs that have a negative effect on their
abilityto buy civilianproducts. Arms imports
often compete with those of machinery and

2. Arms imports exacerbate balance-ofpayments problems for less developed

countries. At a time of dwindling reserves
and harsh stabilizationmeasuresto correct for
payments imbalances, these macroeconomic
costs (in addition to the opportunity costs of
trade diversion)could be particularlyhigh.

4. Barter trade does not involve direct

costs.
However, there are indirect or
opportunity costs if the bartered commodity
couldhave been sold in the open international
market. Where there are domesticshortages,
barter exports (for arms) exacerbate such
shortagesand create additionaldifficulties.
5. Even when the arms are given free,
there are additional costs of operations and
support that inevitablyhave to be paid by the
home government. Econometric analysis
shows that military expenditureis positively
and significantlyrelated to arms imports for
some of the major arms importers. The
economic burden of arms-import-induced
defensespendingneeds to be accountedfor.
6. Arms importsoftenlead to the decision
of setting up domestic industries to produce
armaments.
This version of importsubstituting industrialization (ISI) is often
claimed to have large 'spin-offs' for the
civilian sectors. Formal evidence, based on

MilitaryExpenditureand EconomicDevelopment

country studies for some major arms

producers, does not show any positive effects.
Rather, there may be 'spin-ins' where the
civiliansectors are drained of engineeringand
scientificpersonnelas well as R&D capacity.
Absorptive capacity constraints in less
developedcountries can be tightenedthrough
arms industrialization.
SIPRI maintains transactions information
on five categoriesof weapons: aircraft,armor
and artillery, guidance and radar systems,
missiles,and warships. The registersand data
do not include small arms, artilleryunder 100
mm. calibre, ammunition, support services,
components or component technology.
Inclusionof an item in the data base depends
on whetherit is for militaryapplicationalone.
Thus, unarmored military lorries are not
included,althougharmoredpersonnelcarriers
are included.
SIPRI data on the value of arms transfers
tends to underestimate the value of Third
World arms imports. This is becauseit takes
into account the transfers of major
conventionalweapons (and licenses sold for
production) but not that of small arms. In
addition, it concentratesmainly on armaments
purchased by governments and generally
leaves out sub-national actors in the trade.
Since the data base relies on open sources
alone, black-market transactions cannot be
recorded.
ACDA records arms trade for all
categories of weapons. Its reliability cannot
be judged, however, since neither the sources
nor the price system is revealed. In general,
economicanalysis with arms transfer data is
difficultfor reasons pointedout above. There
is, however, a special problem with ACDA
data: Soviet and East Europeanarms exports
are probably valued at U.S. prices and
thereforeadd a fuirtherdistortionto the data of
importingcountriesthat rely on such imports.
If, as widely believed,Sovietarms exportsare

subsidized, then the economic effects are

rather different from those suggested by
ACDA data.
SIPRI has recently begun publicationof a
list of the 100 largest arms-producing
companies in the OECD in order to get a
better idea of militaryproduction. For each
company, informationis given on country of
origin, arms produced, total and weapons
sales, profit, and total employment.
Unfortunately,this data is not comparablewith
nationalarms trade data, so that statistics on
company sales (or production) cannot be
integrated with net exports or national
procurement expenditures. There is also no
comparable data on military industrial
employment, since total employment from
companybalance sheets is not disaggregated
between military and civilian uses. Finally,
this data base is not relevant for socialist
countries. All aggregate estimatesof Soviet
and Chinese arms industries are either
distorted (due to their price distortions) or
relatively rough.
No clear answers are available to basic
questionssuch as the total employmentof the
world arms industry, the net output of the
world defense industry, the capital stock
employed in the weapons industry in
aggregate, or the rate of return on capital.
Substantialfurther research is required about
arms production; this area lags far behindthe
data available for aggregate military
expenditure and the arms trade.
The
quantitative economic characteristics of
weapons industries are a really large
knowledgevoid. Oddly enough, the technological and engineering characteristics are
extremely well known. Once again, the
excuse that national security reasons prohibit
the gathering of information cannot be
substantiated. Much is known about the
physical and technical capability of the
industry's output. Very little is known about

MilitatyExpenditure
Datafor DevelopingCountries:MethodsandMeasurement

the financial and economic costs of the

indLustry's
input.
Issues
Given our basicunderstandingof the problems
relatedto definingMILEX, there remainthree
substantiveissues regarding the data base for
defense spending. The first relates to the
choice of a suitable deflator to derive a
voliume index of the defense effort or cost
(Huisken 1973). The second relates to
comparisonsover time and across countries.
The third relates to the valuationof resources
associatedwith the defensesector.
SIPRIdata emphasizesopportunitycosts of
increased MILEX for society and therefore
uses the consumer price index (CPI) as a
deiaator. ACDAutilizesthe GNP deflatorand
thus implicitly values MILEX in terms of
other elements of national output. If an
alternative assumptionregarding opportunity
costs were made--say,resource transfer from
defense to investment--then an investment
price deflator couldbe used.
In terms of internal consistency,it may be
preferable to construct an explicit military
price deflatorin order to show the real change
in terms of the product mix of the defense
sector itself. There are few examplesof such
price deflators in practice, partly due to
theoretical problems but also because of
secrecy. The theoreticalanalysisis similarto
those obtainedwhen any governmentservices
need to be priced, with a few specific'military
twists.'
There are two alternatives:to use an output
measure or to use an input measure. As
regards the output measure, there is the usual
problem of aggregationcommon to all such
indices. More important, there is little
agreement as to what military expenditure
produces, i.e., how to define the 'output' of
the defensesector. What exactlydoes defense

Sen

produce--security?offensivepower?-and how
can these ever be measured? It has even been
suggestedthat the explosivepower of weapons
couldbe used as a measure, but this gets into
problemsof comparingsmallnuclear weapons
with large conventional forces. Economic
analysisis better used elsewhere!
The input measure closely follows the
economics of the 'ideal price index'
corresponding to the 'ideal cost of living
index' (Samuelsonand Swami 1974). If X;
inputs (i = 1,- n) produce securityS and the
relation is, for example,
a,
a
S = XI . . .Xn
then the price indexP is given by
a,
p = P1

. . .Pn

where the Pi are the prices of the inputs and

the a, are expenditureshares.
Even though log linearityin the S-indexis
not essential (a CES function would give
similar price transforms),the basic idea is that
some implicit security function (preferably
tractable) is required to get a price index as
the aggregation of the individual prices of
inputs that contributeto security. The use of
such 'dual' properties can be used to produce
explicit military price deflators. Despite a
host of difficulties in practice, in principle it
may be worthwhileto considerthe construction of a military price deflator, at least for the
componentsof the arms internationallytraded.
Then imports and possibly domesticproduction costedat 'border prices' will be adequately
deflated to get constant price series for
weaponsprocurement.
As for the comparability of defense
spending data across most countries in the
sample, both SIPRI and ACDA use official

Military Expenditure and Economic Development

exchange rates to transform local currency

values into dollars. Both ACDA and SIPRI
utilize purchasing-powerparities (PPPs) for
the Europeansocialist countries only; for less
developed countries, ACDA uses PPPs for
China alone, whereas SIPRI uses exchange
rates for China as well. As is well known,
exchange rates do not provide adequate
comparability,particularly if they are over- or
under-valued. Massivedevaluationsin the late
1980s--forexample,in Africancountries-have
made their 1988 dollar values (in the latest
SIPRI tables) look smaller relative to the
world total, as compared to previouslyused
exchange rates. On ithe other hand, overvaluationof the peso in the early 1980smade
the domesticcurrencyvalue of Argentinaquite
low becausemuch of defensespendingwas for
arms imports.
Clearlythe optimummethodis to construct
PPPs explicitly for thD defense sector (see
Cars and Fontanel1987; Heston 1990). There
are major difficulties in such an excercise.
Most importantamongthese are the following:
(i) the problems inherent in securing MILEX
data at a sufficiently disaggregated level;
(ii) the difficulties of comparing defense
products bought, produced, and used in one
country with those in other countries, due to
the degree of product differentiation;(iii) the
absenceof well-definedprices,particularlythe
lack of a 'market' where such prices can be
established. We deal with each in turn.
The constructionof adequatePPPs requires
a substantial amountof price and expenditure
data at a disaggregatedlevel. It is useful, for
example, to have information on operating
costs (broken down into personnel, military,
and civilian; and operationsand maintenance)
and capital costs (procurement and
construction)for forces, industry, and R&D;
this is a bare minimum. In each sub-category
it is also necessaryto have highly disaggregate
data in as much detail as possible. One of the

10

%''...

..

principalpurposes of PPP is to have as wide

a representative sample of products for the
expenditurecategoryas can be obtained,with
due considerationbeinggiven to commonand
comparablecharacteristics;the method clearly
is information intensive. Given the general
secrecy with which militarydata is shrouded,
and the genuine difficulties of obtaining
informationeven when available somewhere,
the data set required for efficientcomputation
is not easy to get.
Continuous technological progress,
obsolescence,trade in second-handweapons,
differences in specification due to specific
country needs, small production runs for
specialized weapons, variations in industrial
structuresfor arms producers, and absorbtive
capacity for arms importers all combine to
producewide ranges of productdifferentiation
in terms of innate characteristics and uses.
Comparabilityof like with like is exceedingly
difficult.
Finally come the well knownproblemsof
pricing military equipment and the
fundamentaldifficultyof not having a proper
'market,' as well as other difficulties, such as
those of industrial organization and the
formation of natural monopolies. But even
with cost-pluspricing, it is not easy to impute
and allocate costs for specific equipmentand
functions. The difficultiesare exacerbatedby
hidden subsidies, foreign aid, compensatory
trade agreements, illegal imports and
smuggling against embargoes, purchase of
weapon systems without adequate
considerationof operations and support costs,
and many other such problems.
In the absenceof explicitmilitary PPPs, a
'short-cut' method would be to convert
MILEX by government expenditure PPPs
instead-and then compare. Alternatively, the
GDP parity can be used. Table 1 gives data
showing the rather startling differences

it

Sen

MillitaryExpenditureDatafor DevelopingCountries: Methods and Measurement

_

~~

~.

.****~*

Table 1. Military Expenditure,1975, Using AlternativeConversion Rates

(millionsof dollars)
PPPr for
Government
Expenditure

Exchange
Rate

PPP for
GDP

1,308

2,062

3,099

130

373

473

India

2,843

9,198

16,430

Iran

7,283

12,401

12,854

Jamaica

19

24

24

Malaysia

159

835

871

M[exico

581

981

1,068

Pakistan

726

2,268

4,319

Philippines

487

1,226

2,477

Sri Lanka

18

66

148

Thailand

409

1,097

5,832

Zambia

90

135

190

Country
Brazil
Colombia

'Purchasing-powerparity.
Source: Author's calculations.

between the figures. The second and third

columns, in Kravis-based 'international'
dollars (see Summers and Heston 1991,
pp. 327-68), usually provide much higher
values for MILEX. It is instructiveto note
that the proportion of aggregateMILEX for
these twelve countries as a share of U.S.
MILEX varies enormously,which couldcolor

our perceptionsof how muchthe Third World

is actually spending. Using governmentexpenditure PPPs for conversion, aggregate
MILEX for only these twelve countriesin the
Third World comes out to an incredible52.5
per cent of the MILEX of the United States-the highestmilitary spender in the world.

11

Militayy Expenditure and Economic Development

For market economies, the valuation of

resources generally is not a major problem.
Centrallyplannedeconomieslike China's have
maintaineddifferentialprices for civilian and
military goods--in the process managing to
keep MILEX artificiallylow. But this is not
the general practice, with one type of
exception. In countries where conscripts
rather than voluntary or professionalsoldiers
are used, militaryexpendituremay be low, yet
its economicimpactmay be high. If there is
substantial unemployment, or even underemployment,then clearlythe opportunitycosts
are negligible. If, on the other hand, the
militaryemploysskilled labor, then how these
resources are valued makes a difference.
Conscriptionmakes it possible to understate
MILEX, eventhough the precise effectsdiffer
from countryto country.
It is also importantto calculatethe valueof
militarystocks or assets-both as a consistency
check with flow data on procurement
expenditureand to evaluatethe resource cost
of weaponspurchases. Though the practical
problemsare formidable,the theoreticalissues
should be clarified. Hilderbrandt (1990) has
suggesteda measureof military capitalwhich,
in principle, summarizes the net benefits (in
money terms) obtaineclfrom military assets
over the remainderof the servicelives of such
assets and equipment. The assets are
depreciated according to age. The index of
military capital is also an aggregatemonetary
measure.
Equipment is valued at its
procurement cost at the time of purchase.
Therefore the implicit assumptionis that the
procurement cost represents the lifetime
benefit of the particular equipment.After the
initial entry, later valuationis done according
to the specificdepreciationmethodused. The
levels and changes in the value of military
capital stock, as defined, would be an useful
consistency check on the flow of military
expenditures--specifically on that for
procurementand operations and maintenance
(O&M). Gross inconsistencieswould be

12

revealed quickly. For example, if military

expenditurewere declining over a period of
time when military capital value was rising,
then the postulatedtrend in defense spending
wouldbe highly suspect.
In the absence of such sophisticated
measures for less developed countries, a
simplermethodwouldbe to apply consistency
checks on defense expenditureby comparing
its trend with that of arms imports does. In a
sense, the SIPRI measure of arms imports
does indeedgive a valuationof the changesin
stocksof arms (due to importedweaponsalone
as wellas domesticproductionby license) and
therefore has similaritieswith the (increments
to) military capital stocks discussed earlier.
Further, data on defense spending and arms
trade are derived in conceptually different
ways, and there is no risk of spurious
correlation. Thus the trends in all revealed
military expendituredata series should in the
first instancebe comparedto arms importsfor
the relevant country, and any fundamental
divergenceshouldbe checkedfor consistency.
For most countries, arms imports are
essentiallysimilar to procurementcosts. In a
like fashion, the trends for military spending
should also be compared with the level of
armed forces and some index of personnel
costs. The latter could be approximatedby
multiplying the number of forces by the
industrialwage rate (for a volunteerarmy), or,
alternatively,by some monetary measure of
the cost of living (for a conscript force).
Strong divergence between indices of
aggregatemilitaryexpenditure,arms imports,
and personnelexpenditurewould be a sign of
problematicdata.
Though the specific characteristics of
MILEX data have been mentionedin general,
it may be useful to bring them together. In
particular, it is necessary to emphasize the
differences with other categories of
government expenditure, especially with the
expenditure data on social services such as

AMGilitary
ExpenditureData for DevelopingCountries: Methodsand Measurement

health and education, since comparisonsare

often made to measure resource costs of the
military sector. I mentionthe most important
characteristics that distinguish the military
sector in terms of data evaluation.
First, the differencebetweenconscriptand
volunteer (enlisted)armed forces needs to be
accounted for. The former usually imply
lower personnel costs, although the
opportunitycosts for the civilianeconomyare
similar to the latter case. Inter-country
comparisonsof aggregate military spending,
the defense burden, and disaggregatedcosts
for manpowerneed to be done carefully. This
differentialin humanresourcecosts is peculiar
to military expenditure.
Second, there is a serious need to include
pension payments within total military
spending. There are two reasons for this.
Military pensions tend to be higher than
civilian pensions in less developedcountries.
In addition,various attractiveearly retirement
schemeswith full pension entitlementsfor the
armed forces imply an inflatedbudgetfor this
category of expenditure. More important,
when the opportunity costs of defense are
being calculatedin a two-sectoreconomy,with
the military and civilian sectors being
compared, pension benefitsadd to the cost of
maintainingthe defense establishment. From
a strictly national accounting perspective,
transfer payments should not be included in
governmentconsumption,but such a logic is
inapplicable in a disaggregative model that
tends to measurethe costs of defense.
Third, military procurement is often
import-intensive--muchmore so than other
forms of government socio-economic
spending. Dependingon the way such imports
are financed--whether through foreign
exchange, military aid and assistance,or debt
creation--theiropportunity-costmeasure will
vary. For example, the military-relateddebt
servicing burden is rarely accounted for

Sen

explicitlyin debt statistics;yet this categoryof

repaymentsis importantin measuringdefense
costs.
Fourth, military procurement is the sole
prerogative of the government or the state.
When weaponsare produceddomestically,this
monopsonistic character of the weapons
market creates the possibility of distortions,
which are reflected in the data. For example,
the weaponsmay be priced artificiallylow and
the procurement budget thereby arbitrarily
reduced. Yet defense industries may be
subsidizedfrom other parts of the state budget
(say, industrial ministries), so that the true
costs of production are not reflected in
procurementcosts. At the sametime, if there
are monopoliesin defenseproduction,then the
incentivesfor cost efficiencyare also lost.
Fifth, militaryhardwaretends to be R&Dintensive. However, the costs of such
research, particularly in the basic sciences,
may be borne by civilian institutions, with
defense-related spending thus maintained at
lower levels. The presenceof such sunk costs
implies that unit costs tend to be high if
production runs are not maintained at high
rates. Yet studies of new weaponsproduction
in developing countries show that such
'excessive' costs do not seem to appear in
budgets.
Sixth, technologicalobsolescenceis rapid
in the modern military sector. The existence
of arms races also means that there is a need
to maintain technical superiority with one's
adversary. For the military, 'invention
becomesthe motherof necessity.' This factor
is conspicuous by its absence in social
spending on health or education in most
developingcountries. The presence of such a
factor would imply that procurement costs
wouldbe higher than 'normal,' since write-off
costs need to be included. Yet this rarely
happens. Again, it is highly probable that
industrialsubsidiesfrom non-militarybudgets

13

MilitaryExpenditureand EconomicDevelopment

take care of the potential losses that defense

industries would suffer if proper accounting
practiceswere maintainecl

With respect to comparing the military

burden across countries, it is often believed
that this measure avoidsproblemsof currency
conversion discussed earlier. For instance,
Ball (1984) claims that "For the other
measure, economicburden, measurementsthat
rely on military expenditure in national
currenciesare clearlypreferred to those in US

Even though this method produces

'inflated' values of the military burden in less
developedcountries,and is potentiallya more
realistic measure of opportunity costs, it
suffers from a number of specific
disadvantagesfor the purposes of analyzing
military expenditures. Governmentspending
PPPs tend to attach high prices to U.S.
personnel;in LDC militaries,in constrast,this
is potentially a source of financial savings.
Conscripts are paid little, and even
professional armies tend to be compensated
with 'perks' rather than high wages. Armed
forces are labor-intensive, from a military
point of view, and the elasticityof substitution
tends to be low. On the other hand, imported
weapons equipment could be expensive,
particularly if not paid from foreign aid.
Finally, use of such a measure for any
category of public-sector expenditure would
produce such high ratios. It could then be

dollars . . . as the conversion problem does

claimed that less developed countries have

not exist." However, if military PPPs did

exist and were utilized in calculating the
burden or share ratios, there clearlywould be
a difference.

higher shares of health, education, or other

categories in aggregate GDP than is usually
supposed. Just as the 'burden' of defensewill
look higher, so will the 'benefits' of health,
etc., look higher than standard measures
suggest.

Opportunity Costs
The opportunitycosts of clefensespendingcan
be measuredby using two ratios. The military
"burden" is defined as the;share of MILEX in
GDP or GNP. The military "share"is defined
as the ratio of defensespendingin total central
governmentexpenditure(CGE).

In the absence of explicit military priceconversion rates, use can be made of the
disaggregatedPPPs for governmentspending,
consumption, and capital formation and
observationof the differences. One method
for calculatingthe military burden, advocated
by West (1987), is to transform defense
spending to internationaldollars by using the
government expenditureP]PP. This value is
then divided by GDP, suitably changed
accordingto its own conversionrate, to arrive
at the military burden. As the basic data in
Table 1 shows, the dollar value of MILEX
will be higher when using government
spending PPP compared to that using GDP
parities; hence the burden could be
substantially higher than the usual data
reported by SIPRI or ACDA.

14

An alternative method of measuring the

burden would be to convert weapons imports
into dollars by using U.S. military prices
directly (or some index of international
prices). Added to this would be personnel
costs, which can be transformed by utilizing
the PPPs for consumption. This value of
military expenditurein 'international' dollars
could then be divided by GDP, suitably
transformed, to measure the burden. In a
similar vein, dividing by central government
expenditure (CGE) changed by government
spending PPPs would indicate the value of
militaryshares. It would be interestingto see
how the various measuresof military burden
would differ.

Military ExpenditureDatafor DevelopingCountries: Methodsand Measurement

Recent attemptsto constructmilitaryPPPs

for a broad set of countries (Heston 1990)
show that although the level of developingcountry defense spending in international
dollars is higher relative to exchange-rate
conversion,the differentialis far less than that
for government expenditurePPPs in general.
Tlhus the ratio of military to total centralgovernment expenditure actually falls when
rnilitary and CGE PPPs are used to do the
respective conversions. Since arms imports
play a major role in procurementspending, it
is not surprisingthat internationaldifferentials
are relatively low in the military sector as
compared to, for example, health or education, whichutilize more domesticresources.
When disaggregateddata is available,even
in a broad sense, the distinction between
operating costs (personnel, pensions, and
O&M) and investment costs (weapons
procurement, construction, military R&D) is
important. The differential impact on the
civilian economy of these two types of
resourceuse in the defensesectoris important.
'Calculationsof the military burden utilizing
iLhis distinction are important for further
research.
Table 2 provides information on the
calculation of the military burden using a
numberof options. The first columngives the
norimalfigure, where local-currencyMILEX
is divided by GDP. The military burden in
the second column, Option 1, is calculatedin
the following way. Actual operating costs
incurred are changedinto internationaldollars
by using the consumption PPP, while
budgetary investmentcosts are converted by
utilizingthe capitalformationPPP. The totals
are then added to get aggregate military
spending in international dollars. GDP is
convertedby using its own PPP. Dividingthe
two, we get military burden for Option 1 in
Table 2.

Sen

The third columnprovides informationon

Option 2, which considers a hypothetical
transfer of resources. Suppose that reduced
militaryexpenditurereleasesresourcesthat are
transfered to the civilian economy and
transformedinto consumptionand investment
in the sameproportionas in GDP. Aggregate
military spending is therefore broken up into
two parts, dependingupon the proportion of
consumption and investment in national
output. The separateparts are then converted
into internationaldollars using the respective
PPPs. The total is divided by GDP (suitably
adjusted),and the ratio is the militaryburden.
Comparison of the columns reveals an
interestingfeature. There seems to be little
difference among the three measures for the
countries studied here. It is difficult to say
whetherthis is a specificfeature dependenton
the particularsample chosen, or whether it is
a general characteristic. However, such
calculationsseem to be importantin deriving
the 'true' burden, and rendering military
expenditure more comparable to other
categoriesof government expenditure(health
or education,for example).
If there are indeed differentialimpacts of
military spending on the economy, then the
usuallyasked 'how much?' questionwill have
to be supplementedby a new one: 'in which
way'? Alternativecalculationsof the military
burden, using disaggregated data, and
comparisons with other categories of social
expenditures, would then be useful.
Specifically, operating costs within military
expendituredo sometimeshave an economic
function: human capital formation or
employment. Investment costs are more
inherently wastefuland therefore need to be
reduced first. An indirect effect, of course,
will be that political tensions will be reduced,
since weapons acquisition is the most
destabilizingfactor in inter-state tension. A
virtuous circle may then be set up that stops
an arms race.

15

MilitaryExpenditureand EconomicDevelopment

Table 2. Military Burden, 1975, Under Alternative Hypothesized Options

(percentages)

Country

Burden
(Nonnal)

Burden
(Option1)

Burden
(Option2)

Brazil

1.02

0.98

0.96

Colombia

0.97

0.97

0.97

India

3.21

3.18

3.05

Iran

13.78

11.40

13.60

Jamaica

0.67

0.72

0.65

Malaysia

4.56

4.53

4.48

Mexico

0.66

0.62

0.65

Pakistan

6.49

5.89

6.03

Philippines

3.09

3.20

2.85

Sri Lanka

0.72

0.71

0.69

Thailand

2.79

2.91

2.68

Zambia

3.66

3.34

3.19

Source: Author's estimates from basic data in Bail (1984).

Concluding Suggestions
Future research on data must emphasizemore
transparency, greater detail, deeper study of
budgetary information where available, and
the need for disaggregation,however limited.
In terms of transparency, increasing
democratization should help to speed the
process. Providing more detail could be
technically difficult for governments; more
help can be given them through specialized
agencies and statistical services. A more

16

intensivestudy of budgets is necessaryin the

field of defenseby multilateral agencies such
as the United Nations, the IMF, and the
World Bank. All of the aspects highlighted
for future emphasisare essentiallymatters for
political economy-since, by its very nature,
purely economic analysis of military
expenditureissueswill not be possible.
Disaggregationof data is of more concern
to economists,since the opportunitycosts are
best studied in this context. An uniform

MilitaryExpenditureDatafor DevelopingCountries: Methodsand Measurement

definitionis required, particularlyfor weapon

procurement. It is a potential candidatefor
both 'consumption' and 'investment' in
governmentexpendituredefinitions. In terms
oifthe former category,once militaryhardware
is purchased, it is potentiallydestructible,and
therefore has a depreciation rate of 100 per
cent. However, if one is concerned about
alternative uses-and this is the rationale of
opportunity costs-military weapon purchase
often directly competes with government
irnvestment. Thus there is a strong case for
consideringit as part of military 'investment'

Sen

and seeing whether resource re-allocationis

possible. The case where arms are bought
from domestic industries or paid for by
foreign exchange and debt is even more
importantin terms of investmentcosts.
The use and misuse of data for defense
economics is an important issue. The
information should be used with care and
caution. But with the help of imaginativedata
mining and the study of individual country
cases, a considerable amount of formal
analysiscan be done.

REFERENCES

Cited Studies
Ball, N. 1984. ThirdWorld SecurityExpenditure:A
Statistical Compendiwn (Stockholm: National
DefenseResearchInstitute).
Becker, A. S. 1977. MilitaryExpenditureLimitations
for Arms Control: Problems and Prospects
(Cambridge,Mass.: BaDinger).
Blackaby, F., and T. Ohlson 1982. 'Military
Expenditureand Arms Trade: Problemsof Data,"
Bulletin of Peace Proposals, 13: 4, 291-308.
Brzoska, M. 1981. "The Reporting of Military
Expenditure," Journal of Peace Research, 18: 3,
261-77.
to
Brzoska, M., and T. Ohlson 1987. Arms Transfers
the ThirdWorld(Oxford:SIPRI,OxfordUniversity
Press).
Cars, H., and J. Fontanel 1987. "MilitaryExpenditure
Comparisons,"in Schmidt, C., and F. Blackaby,
eds., Peace, Defense and Economic Analysis
(London:Macmillan).
Deger, S. 1986. MilitaryExpenditurein ThirdWorld
Countries: The Economic Effects (London:
Routledge& KeganPaul).
Deger, S., and S. Sen 1990. MilitaryExpenditure:The
PoliticalEconomyof InternaionatSecurity(Oxford:
Oxford UniversityPress).

Deger,S., and S. Sen 1991. "Tablesof World Military

Expenditure"and "Sourcesand Methods,"in SIPRI
Yearbook 1991, World Armaments and
Disarmament(Oxford:Oxford UniversityPress).
Goertz,G., and P. F. Diehl 1986. "MeasuringMilitary
Allocations: A Comparison of Different
Approaches,"Journal of Confict Resolution,30: 3,
553-81.
Heston, A. 1990. "Real World Military Expenditures:
134 Countries, 1980," Universityof Pennsylvania.
Mimeographed.
Hilderbrandt, G. C. 1990. "Services and Wealth
Measuresof MilitaryCapital,"DefenseEconomics,
1: 2, 159-76.
Huisken,R. 1973. "The Meaningand Measurementof
Military Expenditure," SIPRI Research Report
(Stockholm:SIPRI).
Neild,R. 1981. How to Make Up YourMind About the
Bomb(London: Andre Deutsch).
Scheetz, T. 1990. 'Military Expenditurein Argentina,
Chile and Peru," in Hartley, K., and T. Sandier,
eds., The Economics of Defense Spending: An
InternationalSurvey (London:Routledge).
Sen, S. 1991. "Debt and InternationalSecurity," in
SIPRI Yearbook 1991, World Armaments and
Disarmament(Oxford:Oxford UniversityPress).

17

Military Expenditure and Economic Development

SIPRI Yearbook,World Armaments and Disarmament

(Oxford: SIPRI, Oxford UniversityPress), various
years.
Summers,R., and A. Heston. 1991. "The Penn World
Table (Mark 5): An ExpandedAct of International
Comparisons,1950-1988,"lhe QuartertyJownal of
Economics(May 1991),pp. 327-68.
United Nations1986. "Reductionof MilitaryBudgets,"
DocumentA/40/421(New York: United Nations).

18

U.S. Arms Control and Disarnament Agency (U.S.

ACDA),various years. WorldMiliary Expenditure
and Arms Transfers (Washington, D.C.: U.S.
ACDA).
West, R. 1987. "ImprovedMeasures of the Defense
Burdenin DevelopingCountries,"in Deger, S., and
R. West, eds., Defense Securilyand Development
(London:Frances Pinter Publishers).
World Bank 1985. World Bank Country Study:
ArgentinaEconomicMemorandum.

PATTERNS AND TRENDS IN THE

MILITARY EXPENDITURES OF
DEVELOPING COUNTRIES
Robert L. West

Introduction
This paper examines the patterns of
international distribution and the trends in
military expendituresof groups of countries,
comparing expenditure estimates derived by
the use of convertorsbased on exchangerates
and purchasing-powerparities. Althoughit is
widelyacknowledgedthat the use of exchangerate convertors introduces distortions in the
cross-countryand inter-temporalcomparisons
of real expenditures, the magnitudes of the
deviations are not generally known. The
primary data sourcesfor militaryexpenditure,
and the convertorsthey employ, are discussed
in the Appendixto this paper (see pp. 147-51).
This analysis will focus on: (a) whether
the distortionsare systematicand significantin
magnitude, and (b) the extent to which the
internationalpattern of distributionand trends
in military expenditures described by
exchange-rate-converted
data correspondto, or
deviate from, the patterns and trends revealed
by estimates based on purchasing-power
parities.
For the computation of estimates that
appear in this paper for all countries and all
years, the underlying data on military

expenditures,centralgovernmentexpenditures,
and GNP in local currency, at current prices,
are the same as those employedby ACDA in
computingthe estimatesthat appear in World
Military Expenditures and Arms Transfers
1988.1
ACDA's statistical tables report data for
144 countries.2
Seventeen of these are
centrally planned economies, for which the
methods of data collection and estimation
differ significantlyfrom the market-oriented
countries; for the purposes of this paper, we
omit these 17 nations. Eight other, small,
market-orientedeconomiesare omittedbecause
purchasing-power parities and price levels
have not been computedby the International
ComparisonProject.3
The coverageof the estimatespresentedin
this paper includes 119 (of ACDA's 127)
market-oriented countries. Twenty-six are
classified as "industrialized market-oriented
economies," including all of the nations of
Westernand SouthernEurope (exceptTurkey)
and North America, plus Japan, Australia,
New Zealand, South Africa, and Israel. The
other 93 countries are classified as
"developingmarket-orientedeconomies"; the
geographical regions into which these
countries are further divided are similar to

19

MilitaryExpenditureand EconomicDevelopment

Table 1. Military Expenditures, Central Government Expenditures,

and Gross National Product in Base Year 1980

LowIncome

LowerMiddleIncome

UpperM;ddleIncome

HighIncome
Oil

Countries
(34)

Countries
(36)

Countries
(19)

Exporters
(4)

Exchange-Rate Convertors
10
61
307

21
132
531

MILEXlCGE
CGE/GNP
MILEX/GNP

16.7
19.8
3.3

16.0
24.9
4.0

I
I
I

All
Industrial
Developing Market
Economies

(93)

Economies
(26)

(billionsdoUlars
at 1980currentprces):
MILEX
CGE
GNP

35
262
978

20
77
184
(percentages)
13.2
25.6
26.8
41.7
3.6
10.7

f
I

I
I

I
I

86
532
2,001

291
2,309
7,865

16.1
26.6
4.3

12.6
29.4
3.7

104
1,073
3,094

267
1,871
7,199

9.7
34.7
3.4

14.3
26.0
3.7

170
1,073
3,094

245
1,871
7,199

15.9
34.7
5.5

.13.1
26.0
3.4

Purchasing-PowerParity Convertors
(billions1980internationaldollars;
MILEX convertedat militaryparity):
MILEX
CGE
GNP

19
300
710

32
306
879

MILEX/CGE
CGE/GNP
MILEX/GNP

6.3
42.3
2.6

10.5
34.8
3.6

35
400
1,340

19
66
166
(percentages)
8.8
27.9
29.8
40.1
2.6
11.2

J
j

Purchasing-PowerParity Convertors
(billions1980internationaldoUars;
MILEX convertedat govenmentparly):
MILEX
CGE
GNP

51
300
710

49
306
879

54
400
1,340

17
66
166

(prcentages)
MILEX/CGE
CGE/GNP
MILEX/GNP

20

16.8
42.3
7.1

15.9
34.8
5.6

13.5
29.8
4.0

25.4
40.1
10.2

I
I

Panernsand Trendsin theMilitaryExpenditures

of DevelopingCountries
those defined by ACDA (with Egypt assigned
to the Middle East and Mexico to Latin
America), except that the ACDA "Oceana"
group is included in Asia. The per capita
income classification is that of the World
Bank's World Development Indicators for
1988.4

The Pattern of Military

Expenditures

in 1980

T'able 1 presents estimates of total military

exKpenditures(MILEX), central government
expenditures (CGE), and GNP for groups of
countries--classifiedby levels of per capita
income--inthe base year 1980. It shows the
ratios that are widelyused as indicatorsof the
relative level of effort by governments to
provide for national security: MILEX to
CGE, or the "defense share" of central
government expenditures; CGE to GNP, a
measureof the "size of the public sector"; and
MILEX to GNP, the "militaryburden.
The upper panel of Table 1 contains the
estimates derived by use of exchange-rate
convertorsfor the base year 1980. For each
country, the expendituresand GNP estimates
in local currencyat 1980current prices (from
the ACDA data set) have been converted at
the 1980annual average market/par exchange
rate. The values are in ACDA constant 1980
dollars.
T'he middle panel of Table 1 containsthe
estimatesderived by use of purchasing-power
parity (PPP) convertors. The estimates of
MILEX, CGE, and GNP in local currency at
1980 current prices (from the ACDA data set)
have been convertedby use of their respective
parities. The values are in PPP 1980
international dollars.
The method of
conversion of military expenditures is
consistentwith the concept of measuring the
value of total input resourcesdevoted by each
nation to its military sector.

West

The bottom panel of Table 1 contains

estimates derived in the same way as the
middle panel, except that the military
expenditureshave been converted by use of
the government expendituresparities. This
methodof conversionof military expenditures
is consistentwith the conceptof measuringthe
opportunitycost of each country's budgetary
allocationfor nationalsecurity.
Exchange rates do not adequatelyreflect
the differences among countries in price
levels, and conversion by use of exchange
rates does not take into account the withincountry differences in the structure of prices
for various sectors and purposes, such as
military or central governent expenditures
and GNP. The extent of these differences in
price levelsand in the structuresof prices--for
groups of the 119 countries represented in
Table 1--is illustratedin Table 2.
'able 2 shows (for MILEX, CGE, and
T
GNP) the averages of the price-level indexes
for the member countries of each income
group in 1980. (1he price-levelindexis equal
to the ratio of the purchasing-powerparity to
the exchange rate for the given year and
expenditure category.) As may readily be
seen, there is a systematicdifference in the
price levels; the price-level indexes vary
directly with the per capita income levels of
the countries. But the elasticityof the pricelevel/incomerelationship is much higher for
total central governmentexpendituresthan for
either military expenditures or GNP. The
structure of within-countryprices differs for
countriesin the various income-levelgroups,
and the variation in structure is directly
associatedwith the per capitaincomelevelsof
the countries.
Anotherindicatorof the differencesamong
groupsof countriesin price levels and parities
is the "exchange-rate deviation index," the
ratio of real MILEX, CGE, or GNP,

21

MilitaryErpenditureand EconomicDevelopment

Table 2. 1980 Price-Level Indexes

(U.S. = 100; averages for countries within the group)

Military
Expenditures

Central
Government
Expenditures

Gross
National
Product

34 Low-IncomeCountries
36 Lower-MiddleIncome Countries
19 Upper-MiddleIncome Countries
4 High-IncomeOil Exporters
26 Industrial Market Economies

62.2
75.4
101.7
113.6
115.0

34.2
42.2
62.1
116.3
107.7

60.7
67.2
81.0
111.3
113.7

United States

100.0

100.0

100.0

convertedby use of purchasing-powerparities,

to exchange-rate-converted
values.
Table 3 shows the averages of the
exchange-rate deviation indexes for the
countriesin each of the income-levelgroupsin
1980. As the country groups are arranged in
order of per capita income levels, ascending
from top to bottom of the table, the systematic

relationshipbetweenthe PPP estimatesand the

exchange-rate-derivedfigures can be readily
seen in the table. For MILEX, CGE, and
GNP, the deviation index falls as per capita
real income rises.
The slope of the
relationship is much steeper for central
governmentexpendituresthan for GNP. The
reasons for the systematic variation in the
deviations associated with levels of national

Table 3. 1980 Exchange-RateDeviationIndexes

(U.S. = 11D0;
averages for countries within the group)

22

Military
Expenditures

Central
Government
Expenditures

Gross
National
Product

34 Low-IncomeCountries
36 Lower-MiddleIncome Countries
19 Upper-MiddleIncomeCountries
4 High-IncomeOil Exporters
26 Industrial MarketEconomies

169.4
150.5
115.0
95.5
101.3

345.1
256.0
177.9
90.1
108.6

180.8
159.8
130.6
90.4
93.3

United States

100.0

100.0

100.0

Patternsand Trends in theMilitary Expendituresof DevelopingCountries

West

T'able 4. 1980 Real Military Expenditures

(U.S. = 100; total expenditures for countries within the group)
MilitaryExpenditures
Convertedby:
Exchange
Rates

Military
Parities

Govermnent
Parities

34 Low-IncomeCountries
36 Lower-MiddleIncome Countries
19 Upper-MiddleIncome Countries
4 High-IncomeOil Exporters
25 IndustrialMarket Economies

6.9
14.6
24.3
13.9
102.1

14.1
23.7
26.0
14.1
98.1

41.7
40.1
44.1
13.9
100.3

United States

100.0

100.0

100.0

per capita income has been explored by the

InternationalComparisonProject analystsand
azttributedto the relatively low prices of nontradable goods and services relative to
tradablesin low per capitaincomecountries-a
phenomenon that can be explained by the
'''productivity differential" model. Nontradables, notably including sevices, weigh
much more heavily in government
expendituresthan in GNP as a whole, and this
accountsfor the more pronouncedvariationof
5
C(GEdeviationswith income-levels.
An indicationof the effectof the distortion
introducedby exchange-rateconvertorsin the
cross-country comparison of military
expendituresis providedin Table 4. The table
shows the 1980 magnitudeof total military
expendituresby the developingcountriesthat
aLremembers of the various income-level
groups, and by the 25 industrial market
economies other than the United States-relative to military expendituresby the United
States in that year. We can compare the
estimatesof the real quantitiesderived from
use of military parities and government
parities with the values in ACDA constant
1980 dollars computed by use of exchange

rates. The use of exchange rates very

substantiallyunderestimatesthe real quantities
of inputs devotedto nationalsecurity by lowincome and lower-middle-incomecountries
relative to industrialized countries (compare
the indexes of quantitiesin the exchange-rate
and military-parities columns). The use of
exchange-rate convertors results in a much
greater degree of underestimation of the
opportunitycost of militaryexpendituresby all
low- and middle-incomedevelopingcountries
relative to the military expenditures by
industrializedcountries.
The estimatespresented in Table 1, using
purchasing-powerparities as convertors and
measuring the opportunity cost of military
expenditures(bottom of Table 1), show that
the military burdens of the developing
countries (MILEX/GNP) in 1980 were
inversely related to their per capita income
levels-except for the small group of highincome, oil-exportingcountries. In 1980, the
military burdens for all groups of developing
countrieswere significantlyhigherthan for the
industrial market economies. This rank
ordering of military burdens does not obtain
for the measurementof real inputs devotedto

23

MilitaryExpenditureand EconomicDevelopment

national security (estimated by the use of

militaryparities to convert MILEX), and it is
disguised by the use of exchange-rate
convertors.

expenditures grew rapidly, but the defense

share of government spending declined
sharply.
The military burden declined
throughoutthe decade.

Trends in Military Expenditures,

In contrast, the developing countries,

while also experiencinga high rate of growth
of GNP and of central government
expenditures, had a high growth rate of
military expenditures. There was a resultant
rise in the militaryburden over the decade.

1970-85
Annex Tables A-1 to A-3 (pp. 28-30) contain
estimatesof MILEX, CGE, and GNP and the
ratios widely used as indicators of effort by
governmentsto provide for national security
for 119 countriesin the years 1970, 1980,and
1985. The estimateswere computedby use of
exchange-rateconvertorsand are expressedin
ACDA constant1980dollars. Growthrates of
each of the aggregates and ratios are shown
for the periods 1970-80and 1980-85.
Annex Tables A-4 to A-6 (pp. 31-33)
contain estimates for the same aggregates,
ratios, and growth rates, but the estimates
were computed by use of purchasing-power
parities. Corresponding to the "opportunity
cost" concept of military expenditure, the
government-sectorparities were used for the
conversionof military spending.
The exchange-rateconverteddata in Tables
A-1 to A-3 illustratesthe evidencesupporting
the conventionalwisdom regarding trends in
military expendituresin the 1970sand in the
first half of the 1980s. Comparisonwith the
estimatesbased on conversionusing parities,
shown in Tables A-4 to A-6, reveals some
additional trends and changes that have
occurred in the military expenditures of
developing countriesbut that are less clearly
indicated, or disguised, by the use of
exchange-rateconvertors.
Comparison of industrialized and
developing countries. In the decade of 197080, the industrialized market economies
experiencedgrowth of GNP in excessof 3 per
cent annually but a very low growth rate of
military expenditures. Central government

24

From 1980 to 1985, the GNP and

government expenditure growth rates of
industrializedcountriesdeclinedfrom the high
levels of the 1970s, but the growth rate of
military expenditures rose sharply. This
resultedin an increasein the militaryburden.
Among developing countries, in the first
half of the 1980s, there was a very sharp
decline in the growth rates of all aggregates.
The military burden continued to risealthoughat a rate below that of the 1970s.
Although evidencefor all of these broad
trends-for both the industrialized and
developing groups of countries-can be
identifiedin the estimatesbased on conversion
by exchange rates and by parities, the use of
exchangerates does not show that the growth
rates of MILEX, CGE, and GNP were all
higher in developing countries than in the
industrializedcountries in the period 1980-85
as well as in the preceding decade. As one
consequenceof the persisting growth of the
real military burden in the large group of
developingcountries, the level of the burden
in developingcountrieshas been above that of
the industrialized countries since the early
1970s. The gap betweenthe two continuedto
grow throughout the period under review,
even in the era of rising military expenditures
by industrializedcountriesduring the 1980s.
Figures 1 and 2 illustratethe trends in the
military burdens of the two groups of

Patterns and Trends in the MilitaryExpendituresof DevelopingCountries

West

Figure 1. Military Burden in ACDA 1980 Constant Dollars

5.5

Deveoped

Developing

4.5

0
-0

4
0

1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1960 1981 1982 1983 19S4 1985

Figuire 2. Military Burden in PPP 1980 Constant Dollars

Deloped

Developing

5.5

z
~4.5
0

4
3.5

2.5

1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1960 1981 1982 1983 1984 I9W

25

MilitaryExpenditureand EconomicDevelopment

countries over the 1970-85 period. The

graphs also show that failure to adjust the
expendituresdata for changes in price levels
and structures-as by the use of exchange-rate
convertors--results in disguising the
relationshipbetweenthe real militaryburdens
of the two groups of countries. For the group
of developing countries, the exchange-rate
converted estimates in Figure 1 show much
more erratic year-to-yearvariation, and they
disguise the persistent growth of the real
military burden in the first half of the 1980s.
Trends in the military expenditures of
developing regions. In the decade 1970-80,
there was relatively rapid growth in both
expenditures and GNP in all regions of the
developing world. Generally, the growth of
the governmentsectorwas greater than that of
military expenditures, and both grew more
rapidly than GNP. Only in countries of the
Latin American region was the growth rate of
GNP higher than that of military spending.
The military burden grew in all regions other
than Latin America.
These relationshipscan be inferred from
both the exchange-rate-convertedand paritiesconverted estimates. However, the real
growth rates of the size of the government
sector and the military burden in the Middle
East and in Africa during the 1970sare much
higher in the estimates convertedby use of
parities than in the exchange-rate-converted
data.
In the 1980-85period, there was a sharp
fall in growth rates in both the Middle East
and Africa, and a more modestdeclinein Asia
and Latin America. The result was a reduced
growth rate of the military burden in the
Middle East and Asia and an absolutedecline
in the militaryburden for the Africangroup of
countries, but a rise in the growth rate of the
militaryburden in Latin America.

26

Again, these trends can be observed in

both exchange-rate- and parities-converted
data.
What the exchange-rate-converted
estimatesdo not reveal is that, throughoutthe
1980s, real militaryburdens (as shownby the
estimates converted by use of parities) for
countries in the regions of the Middle East,
Africa, and Asia were all higher than the
military burden of the group of industrialized
countries. Only in Latin Americawas the real
militaryburden in the 1980slower than in the
group of industrializedcountries; but the real
militaryburden in Latin Americawas growing
much more rapidly than in industrialized
countries during the first half of the 1980s,
and this, too, was disguisedby the distortion
in the estimatesconvertedby use of exchange
rates.
Trends in the military expenditures of
d&fferentincome groups of countries. In the
decade 1970-80,growth rates were generally
higher in the middle-incomecountriesthan in
the low-incomecountries. Duringthis period,
generally, the higher the per capita income,
the higher the growth rates of the size of the
government sector and the military burden.
The evidenceof these relativegrowthrates can
be found in both exchange-rate-convertedand
parities-convertedestimates.
In 1980-85,the relation between growth
rates and per capita income levels reversed.
In general,the growth rates were higher in the
low-incomecountries; the rates of growth in
the size of the government sector and in the
military burden were higher in low-income
than in middle-incomecountries. This pattern
can be observed in both exchange-rate-and
parities-converteddata.
Using exchange-rate-converteddata, the
size of the governmentsector (as shown by the
CGE/GNPratio) appearsto be about the same
for groups of countries at different levels of
per capita income. But this is not the case
when differences in the within-country

Paaternsand Trends in the MilitaryExpendituresof DevelopingCountries

s,tructureof prices are taken into account, and

the systematic distortion attributable to
exchange-rateconversionis eliminated. In the
estimatesbased on conversionusing parities,
the real governmentsector size is muchlarger
in countriesat lower incomelevels.
Measured in real opportunity-costterms,
the militaryburden (see Annex Table A-6)

West

was inversely related to per capita income

levelsthroughoutthe periodreviewed, and the
burden rose for all groups of developing
countries from 1970 to 1985. Moreover,
during the first half of the 1980s,the military
burden grew more rapidly in low-incomethan
in middle-incomedevelopingcountries. These
relationshipsand trends cannotbe seen clearly
in the exchange-rate-convertedestimates.

27

Military Expenditure and Economic Development

Statistical Annex

Table A-1
Real Military Expenditures, Central Goverment

Expenditures,

and Gross National Product, 1970, 1980, and 1985

(billions of ACDA constant 1980 dollars; ratios and growth rates in percentages)

1970

1980

1985

Growth Rates
1970-80
1980-85

(percentages)
0.9
4.7
6.0
3.4
3.2
2.2

Industrial MarketEconomies(26)
MILEX
CGE
GNP

266
1,285
5,760

MILEX/CGE
CGEtGNP
MILEX/GNP

20.7
22.3
4.6

291
366
2,309
2,731
7,865
8,757
(percentages)
12.6
13.4
29.4
31.2
3.7
4.2

1
I

86
532
2,001
(percentages)
16.1
26.6
4.3

97
607
2,111

I
I

44
183
425
(percentages)
24.2

49
171
378

f
I

-4.9
2.8
-2.2

1.2
1.2
2.5

10.1
10.7
6.4

2.4
2.6
1.1

-0.5
4.0
3.5

-0.2
1.6
1.4

116.8
18.1
14.2

2.0
-1.4
-2.4

-1.1

3.4

3.4
2.3

1.0
4.4

DevelopingMarket Economies(93)
MILEX
CGE
GNP
MILEX/CGE
CGE/GNP
MILEX/GNP

33
192
1,072
17.0
17.9
3.0

I
15.9
28.8
4.6

1
I

MiddleEastern Countries(13)
MILEX
CGE
GNP

28

9
35
113

MILEX/CGE

26.9

CGE/GNP
MILEX/GNP

30.8
8.3

43.0
10.4

I
I

28.5

45.2
13.0

Paterns and Trends in the MilitaryExpendituresof DevelopingCountries

West

Table A-2
Real Military Expenditures, Central Govermnent Expenditures,
and Gross National Product, 1970, 1980, and 1985
(billionsof ACDAconstant 1980dollars; ratios and growthrates in percentages)

I
1970

1980

1985

5
33
199

10
70
288

8
79
302

AfricanCountries(41)
MILEX
CGE
GNP

I
(percntages)

MILEX/CGE
CGE/GNP
MILEX/GNP

I
I
I
I

Growth Rates
1970-80
1980-85
(percentages)
6.4
7.7
3.7

-2.6
2.5
1.0

-1.2
3.8
2.6

-5.0
1.5
-3.5

6.0
6.4
5.1

15.4
16.7
2.6

13.6
24.3
3.3

10.6
26.2
2.8

10
48
294

20
106
503

26
144
646

1
1

6.6
8.3
5.5

Asian Countries (15)

MILEX
CGE
GNP

(percentages)
MILEX/CGE
CGE/GNP
MILEX/GNP

21.8
16.3
3.5

18.6
21.0
3.9

18.3
22.3
4.1

I
1
I

-1.6
2.6
1.0

-0.4
1.2
0.8

8
76
466

12
174
785

13
213
784

I
I

4.7
8.6
5.4

1.5
4.1
0.0

6.2
27.1
1.7

I
I

-3.6
3.1
-0.6

-2.5
4.1
1.5

Latin American Countries (24)

MILEX
CGE
GNP

(percentages)
MILEX/CGE
CGE/GNP
MILEX/GNP

10.1
16.4
1.7

7.0
22.2
1.6

I
1

29

MilitaryExpenditureand EconomicDevelopment

Table A-3
Real Military Expenditures, Central Government Expenditures,
and Gross National Product, 1970, 1980, and 1985
(billions of ACDA constant 1980 dollars; ratios and growth rates in percentages)

I
1970

1980

1985

Growth Rates
1970-80
1980-85

Low-Income Countries (34)

l
MILEX
CGE
GNP

7
39
229

MILEX/CGE
CGE/GNP
MILEX/GNP

17.3
16.8
2.9

10
61
307
(percentages)
16.7
19.8
3.3

13
83
374
15.8
22.2

3.5

Lower-Middle-Income Countries (36)

MILEX
CGE

GNP
MILEX/CGE
CGE/GNP
MILEX/GNP
Upper-Middle-Income

11
54

320
20.1
16.9
3.4

1
1

6.4
4.0

-0.4
1.6
1.3

-1.1
2.3
1.2

6.8
9.4

2.5
2.0

5.2

1.9

21
132

24
146

531

585

(percentages)
16.0
24.9
4.0

16.4
24.9
4.1

-2.3
4.0
1.6

0.5
0.0
0.6

38
296
1,034

I
I
I

9.7
10.9
6.8

2.0
2.4
1.1

13.0
28.6
3.7

-1.1
3.9
2.7

-0.4
1.3
0.8

Countries (19)

MILEX
CGE
GNP

14
93
510

MILEX/CGE
CGE/GNP
MILEX/GNP

14.8
18.4
2.7

35
262
978

(percentages)

30

f
I

(percentages)
5.2

4.3
4.7
3.0

13.2
26.8
3.6

Patternsand Trends in the Military Expendituresof DevelopingCountries

West

Table A-4
Real Military Expenditures, Central Government Expenditures,
and Gross National Product, 1970, 1980, and 1985
(billionsof PPP 1980internationaldollars; ratios and growthrates in percentages)
I
1970

234
1,093
5,061

245
1,871
7,199

297
2,103
8,101

I
1
I

0.4
3.6

3.9
2.4
2.4

13.1
26.0
3.4

14.1
26.0
3.7

-4.8
1.9
-3.0

1.5
-0.0
1.5

170
1,073
3,094

207
1,286
3,518

I
1

9.5
10.8
5.6

4.0
3.7
2.6

-1.2
5.0
3.7

0.3
1.1
1.4

15.2
17.4
7.1

3.4
-0.7
2.5

I
I

-1.9
9.6
7.5

4.2
-3.1
0.9

I
(percentages)

MILEX/CGE
CGE/GNP
MILEX/GNP

21.4
21.6
4.6

GrowthRates

1985

Industrial Market Economies (26)

MILEX
CGE
GNP

1980

1970-80

1980-85

(percentages)
5.5

Developing Market Economies (93)

MILEX
CGE
GNP

69
383
1,797

(percentages)
MILEX/CGE
CGE/GNP
MILEX/GNP

17.9
21.3
3.8

15.9
34.7
5.5

16.1
36.6
5.9

16
56
270

66
281
538

78
271
608

I
j

Middle Eastern Countries (13)

MILEX
CGE
GNP

(percentages)
MILEX/CGE
CGE/GNP
MILEX/GNP

28.4
20.8
5.9

23.4
52.1
12.2

28.7
44.5
12.8

31

Military Expenditure and Economic Development

Table A-5
Real Military Expenditures, Central Government Expenditures,
and Gross National Product, 1970, 1980, and 1985
(billions of PPP 1980 international dollars; ratios and growth rates in percentages)

I
1970

1980

1985

6
43
202

16
108
293

16
135
318

African Countries (41)

MILEX
CGE
GNP

I
I
I

Growth Rates
1970-80
1980-85
(percentages)
0.5

9.4
9.5
3.8

4.6
1.6

-0.2
5.5
5.3

-3.9
2.9
-1.2

7.4
9.3
5.5

5.6
6.7
5.5
-1.0
1.1
0.0

5.5

2.8
2.7
-0.3

-4.2
3.4
-0.9

0.2
3.0
3.1

(percentages)
MILEX/CGE
CGE/GNP
MILEX/GNP

14.8
21.5
3.2

14.6
36.7
5.4

11.9
42.3

j
|

5.1

94
576
1,467

Asian Countries (15)

MILEX
CGE
GNP

35
171
657

MILEX/CGE
CGE/GNP
MILEX/GNP

20.5
26.0
5.3

72
417
1,121
(percentages)
17.2
37.2
6.4

16.4
39.2
6.4

-1.7
3.6
1.8

20
305
1,124

4.5
9.1

6.5
27.2
1.8

Latin American Countries (24)

MILEX
CGE
GNP

11
112
668

17
267
1,141

(percentages)
MILEX/CGE
CGE/GNP
MILEX/GNP

32

9.9
16.8
1.7

6.4
23.4
1.5

Patterns and Trends in the MilitaryExpendituresof DevelopingCountries

West

Table A-6
Real Military Expenditures, Central Government Expenditures,
and Gross National Product, 1970, 1980, and 1985
(billionsof PPP 1980internationaldollars; ratios and growth rates in percentages)

I
1970

1980

MILEX
CGE
GNP

27
143
501

51
300
710

MILEX/CGE
CGE/GNP
MILEX/GNP

18.7
28.4
5.3

1985

Low-IncomeCountries(34)

I
73
456
940

(percentages)
16.8
42.3
7.1

15.9
48.5
7.7

Growth Rates
1980-85

I 1970-80
I

1
j

I
I
I

(percentages)
6.6
7.7
3.5

7.5
8.7
5.8

-1.0
4.0
3.0

-1.1
2.8
1.6

9.6
11.5
5.9

2.6
2.2
2.2
0.4
0.1
0.4

Lower-Middle-Income Countries (36)

MILEX
CGE
GNP
MILEX/CGE
CGE/GNP
MILEX/GNP

20
104
496
18.8
20.9
3.9

49
306
879
(percentages)
15.9
34.8
5.6

56
342
979

16.2
35.0
5.7

-1.7
5.3
3.5

63
426
1,482

I
I
I

9.9
11.9
6.1

3.1
1.3
2.0

1
1

-1.8
5.5
3.6

1.8
-0.7
1.1

1
I

Upper-Middle-Income Countries (19)

MILEX
CGE
GNP

21
130
743

MILEX/CGE
CGE/GNP
MILEX/GNP

16.2
17.5
2.8

54
400
1,340

(percentages)
13.5
29.8
4.0

14.8
28.8
4.3

33

Military Expenditure and Economic Development

REFERENCES
Notes
l The "ACDAconstantdoUar"estimatesthat appear in
TableI of WMEAT1988are in pricesof the base year
1987. For the purposes of this paper, the localcurrency, current-pricedata set has been reconverted
by the ACDA method for the base year 1980. The
same set of GNP deflators (convertedto base year
1980) and exchange rates used by ACDA for its
WMEAT 1988 computationswas used to derive the
estimates that appear in thlis paper. For aU the
countries includedin this paper, ACDA uses central
government expenditure anid GNP (current and
constantprice) estimates from the World Bank data
bank. See the "StatisticalNotes," WMEAT 1988,
pp. 129-37.
2

Countrycoverageis describedin ibid., p. 129.

4 See World Bank, World DevelopmentReport 1988

(New York: OxfordUniversityPress), p. 212 ff. The

assignmentof developingcountriesis to the categories
of low, lower-middle,upper-middle,and high (oilexporting) per capita income, as in the World
DevelopmentIndicators section of the Report. But
note that, for the purposes of this paper, we classify
7 countries as "industrial market economies" in
additionto the 19 in the World Bank category. These
seven countries are Israel, South Africa, Greece,
Iceland, Luxembourg,Malta and Portugal; they are
excluded from the developing-countryincome and
regionalgroupingsin this paper.
5 For the definitionsof the indexes reviewedhere and
discussion of the systematic relationships, see the
United Nations International Comparison Project
Phase III Report:Irving B. Kravis, Alan Heston,and
RobertSummers,op.cit.,pp. 9-23 and Chapter8. An

The Penn World Table IV data set consists of 130

countries, of which ten (including Afghanistan) are
centraly planned nations omitted in this paper. Hong
Kong, included in the PWT4 data set, is also omitted
in this paper. The 17 centrally planned economies
included in the ACDA tables are the 9 Eastern
European states, including Yugoslavia, plus

explanation of the systematic variations in price-levels,

quantities, and parities, in terms of the productivity
differential model, emphasizes that "services, which
are non-traded goods, are very cheap in low-income
countries. Hence, exchange-rate conversions greatly
understate the true quantities of services in lowincome countries relative to those in high-income
countries. The low prices of services in the price

Afghanistan, China, North Korea, Mongolia,

Cambodia,Laos, Vietnam,and Cuba. The 8 marketorientedeconomies(includedin the ACDAtables) for

structureof low-incomecountriesencouragetheir use

and lead to quantities that are high relative to the
quantities of the relatively more expensive

which PPPs are not available are: Cape Verde,

Equatorial Guinea, Guinea-Bissau, Lebanon, Libya,
Qatar, Sao Tome and Principe, and Yemen (Aden).

commodities."
In the case of commodities, the
understatement of real quantities by exchange-rate
conversion is also substantial and inversely related to

3 See Robert Summers and Alan Heston, op.cit., 1988.

income levels, but much less so than in the case of

services.

34

MILITARY EXPENDITURE AND

ECONOMIC DEVELOPMENT:
ISSUES AND DEBATES
Saadet Deger

Initroduction
Military expenditurein the developingworld
has been declining since the mid-1980safter
fifteen years of uninterrupted rise. This
declineis takingplace from a very high peak,
however, and it is still modest and slow.
Third Worldmilitaryexpenditure(MILEX)as
a share of total world military expenditure
rose from 6 per cent in the late 1960sto 18
per cent in the mid-1980s, after which it fell
to 16 per cent in 1990. (Total world MILEX
in 1990 was $950billion dollars.) Currently,
at a conservativeestimate,over 4 per cent of
the GDP of less developed countries is
allocated to the military sector. In addition,
there are hidden expenditures--indirectcosts,
industrial subsidies, and arms-related debt
creation-that are not reflectedin the data. In
1985, there were more than 18 millionpeople
in uniform in the developing world. Warrelated costs are almost never included in
military statistics. Though defense spending

Note: The author extends special thanksto Shuja Nawaz

and GeoffreyLamb for detaLedcommentson this pae,

as well as to participantsin the World BankSymposium
for their suggestions.

has fallen, this has not been a result of any

arms control agreements or any serious
resolutionof disputes, except perhaps in Latin
America. Rather, the decline has been a
product of economicproblems. Hence, the
potential for the expansion of military
expenditureis ever present.
Developmental failures lead to loss of
economic security, and this is potentially
destabilizing.
The legitimacy of the
government is eroded, and this leads to a
decline in the legitimacy of the state--itself
threateningthe democratizationprocessthat is
evolving in the developingworld. Intra-state
conflicts require increasing expenditure on
internalsecuritypolice, the paramilitary,drug
control, and suppression of revolts. The
effectsspill over to neighboringcountries,and
inter-staterelationsare also affectedadversely.
In spite of the fact that the number of major
conflictshas been decliningsince about 1986,
the decline has been painfully slow, and, as
the Gulf crisis shows, the potential for
escalation remains strong. For the Third
World as a whole, the number of major
conflicts in 1990was 31; this was fewer than
the 35 conflictsrecorded in 1986and 1987 as
well as the 34 recorded for 1988 and the 33

for 1989. Althoughthis is a downwardtrend,

lre
totAltneverthess rereensa
the total neverteless

represents a large

35

MilitaryExpenditureand EconomicDevelopment

number of major conflicts. In the future,

developmental failures, environmental
problems, and food scarcity will mean that
resourceconflictsmay increase. In particular,
a number of unresolved disagreementsover
water resources threaten stability in many
parts of the world. There are implicit
conflicts about the waters of the Nile,
Euphrates/Tigris,Jordan, Indus, Salween/Nu
Jiang, Mekong, Parana, Lauca, and Rio
Grande. These involve about twenty-five
Third World countries. In addition, there are
territorial disputes about sea resources and
attemptsto establishexclusiveeconomiczones
around coastal waters--all of which lead to
naval expansion and back again to high
defenseexpenditure.
In this setting, the possibilities for
'spontaneous arms control' and for the
reduction of MILEX need to be increasedby
putting pressure on aid recipients to resolve
disputes and cut defense spending to release
domesticresources for development. But this
cannot be done unless aid levels themselves
are raised. Foreign economicaid can only be
used as an incentive if it is high enough for
developingcountriesto gain substantiallyfrom
such inflows. Otherwise, perceived security
threats will force military expenditure to
increase when economic and budgetary
constraints are relaxed. If the terms of
receiving foreign aid are made stricter,
through defenseconditionality,then there must
be some compensation through higher aid
levels (see Deger and Sen 1991). There is a
need to analyze the potential for the
disarmamentdividend, wherebythe industrial
countriescould pass on a part of their savings
from defense cuts to aid recipients. The
relaxation of the East-West confrontation
allows the possibility of transfering donor
countries'defenseexpenditureand militaryaid
to economicassistanceprograms.
Third World countries are faced with a
rather tragic contradiction between the

36

'security dilemma' and the 'poverty trap.'

They do have genuine military security
problems: history, colonialism, internal
dissension, regional problems, intervention
(both direct and indirect)by the major military
powers, and the weakness of domestic
democratic institutions are all important.
These profound security concerns lead to
governments giving primacy to defense
spending over most other forms of
expenditure. Some econometricsevidencealso
shows that defense spending tends to be
determined relatively autonomously of the
domestic economy.
Clearly there are
exceptions; for example, the rise in oil
revenue may produce an orgy of defense
spending. But even in this case there is no
close connection between wealth and the
amountthat societyis willingto providefor its
defense. Security considerations-the eternal
vigilancetheme-seem to explainthe evolution
of defensebehavior. Given that causality,it is
formally valid to ask what effects defense
expenditurewill have on economicgrowth and
development.
The problem would be simplifiedif there
were a very clear guns and butter, or tank and
tractor, trade-off. If we could be sure that,
always and everywhere, military spending
reduces economic performance, the only
remaininganalyticaltask wouldbe to quantify
the negativemultipliers. But this is not true.
In addition to heuristic evidenceand country
studies, empirical work now exists showing
that defenseexpenditurealso has some positive
effects on the economy. The task of the
analystbecomesmore complicated,sinceboth
the sign and the amount of the effect is
controversial.
The major purpose of this paper is to see
how defense affects growth and development
and to summarize some of the empirical
evidence that has appeared in recent years.
Almost all of the econometric studies are of
recent vintage. In 1978, Kaldor claimed:

MilitaryExpenditureand EconomicDevelopment: Issues and Debates

"Most of the analyses . . . are necessarily

speculative. Very little empirical work has
been done in the field." That seems to have
changed dramatically over the last decade.
The two major variablesused are: the military
burden, or the ratio of defense spending to
national product (GDP); and the rate of
growtlh of GDP (or growth of GDP per
calpita). In addition, the share of the defense
sector in central government expenditure
(CGE) is used as an explanatoryvariable to
analyzethe impactof defenseon growth.
The next three sectionsbriefly discussthe
backgroundto the defense-growthcontroversy,
providethe theoreticalframework,and survey
the econometrics evidence. The paper then
considers wider issues of development and
offers some brief conclusions.

T]he Trade-Off Between Defense and

Economic Growth:

The TlheoreticalFramework
A large literature now exists on the general
relationshipbetweendefenseand development,
A]nearlier formal survey is found in Deger
(1986); the more recent formal survey is in
6robar
and Porter (1990). Descriptive
surveys are availablein Chan (1985) and Ball
(1988). The empirical analysis concentrates
on the rate of growth, since broader
developmentalfactors are not easily identified
empirically. However, increasingattentionis
also being given to socio-economic
development beyond narrow indicators of
economicgrowth. The basic idea can be put
starkly: Defense spending increases national
security and therefore improves the
environment within which growth and
developmentcan prosper. In addition, there
are spin-offs that promote development,
Taken together, these factors imply that,
potentially, military spending could enhance
economicgrowth. On the other hand, military
expenditure reduces resources available for
inivestmentand possiblyfor consumption. In

Deger

addition, it causes distortionsin the economy

that reduce efficiency. Taken together, these
latter factors imply that military spending
reducesthe rate of growth. The debate arises
out of the evaluation of the net effect across
countries and over time--whether defense
retardsgrowthand developmentoveralland by
how much.
Much of the literature is heuristic, but the
formal literature has also grownfast in recent
years. The impetusto the empiricalwork that
focuses on the effect of military spending on
economicgrowth came from Benoit's seminal
work (1973, 1978). Contraryto expectation,
he found that defensespendinghas a positive
effect on the rate of growth of national
product (income)in less developedcountries.
Benoitused cross-sectiondata and estimateda
single-equationsimple trivariate relation with
military burden and investment affecting
growth rates.
In a sense, most of the econometrics
studies on the subject are a response to the
challengeposed by the Benoit result. Benoit
identifieda number of positive and negative
channelsthrough which defenseis expectedto
affect growth. The negative effects are
standard: a) resourcetransferfrom investment
to defense,and b) the fact that the government
sector (including the military) exhibits no
measured productivity increases--hence its
relative expansionlowers the nationalgrowth
rate. But Benoit also identified a host of
factors that directly or indirectly increase
growth. He claimsthat:
Defenseprograms of most countriesmake
tangible contributions to the civilian
economies by (1) feeding, clothing, and
housing a number of people who would
otherwise have to be fed, housed, and
clothed by the civilian economy--and
sometimes doing so, especially in less
developedcountries, in ways that involve
devely coung
th
alanvothe
sharply raising their nutritional and other

37

Military Expenditureand EconomicDevelopment

consumptionstandards and expectations;

(2) providingeducationand medical care
as well as vocationaland technicaltraining
(e.g., in the operation and repair of cars,
planes and radios; in hygieneand medical
care; in construction methods) that may
have high civilianutility; (3) engagingin a
variety of communicationnetworks, etc.,
that may in part serve civilian uses; and
(4) engaging in scientific and technical
specialties such as hydrographicstudies,
mapping, aerial surveys, dredging,
meteorology, soil conservation, and
forestry projects as well as certain quasicivilian activities such as coast guard,
lighthouse operation, customs work,
border guard, and disaster relief which
would otherwisehave to be performed by
civilian personnel. Military forces also
engage in certain R&D and production
activitieswhich diffuseskillsto the civilian
economy and engage in or finance selfhelp projects producing certain manufactured items for combinedcivilian and
military use which might not be
economicallyproduced solely for civilian
demand (Benoit 1978).
Benoit's econometricsis simplistic,since it
follows from a rather basic relation regarding
the effect of defense spendingon growth. A
more complicatedformulationwould have to
take into account quite a few other channels
and balancethe various influencesto produce
the net effect.
For cross-sectionstudies taking many less
developedcountriestogether, at the theoretical
as well as empirical level, there have been
three basic ways of evaluating,analyzing,and
criticizingthe hypothesisthat defensespending
increasesgrowth rates in these countries:
The first method uses growth and
planning models to discuss the possible
channelsthrough which defensespending
can affect growth. After analytically

38

identifying the relevant channels,

the econometric studies try to
measure whether the net effect is
negativeor positive.
* The second method looks at finer subgroupingsand tries to see whetherspecific
economic and developmental characteristics can be identifiedthat point to one or
another type of growth effect of defense
expenditure.
* The third method looks at structuralist
models of developing countries and
focuses on those elements in the
interconnection--forexample,the share of
agriculture or of investment--that
structuralists consider crucial in the
growth process.
Broader measures of economic
development are also utilized, though less
commonly, to measure the effect of defense
spending. The general growth models
sometimesadd on other expenditureshares as
dependent variables (such as government
health or educationspending);more qualitative
variables(infantmortalityrates, quality of life
indicators) are also sometimes considered.
Models of the second type, which emphasize
specific characteristics, introduce wider,
mostlypolitical, notionsof development(such
as governmenteffectivenessor the role of the
militaryin controllingthe state). Structuralist
models can also incorporate broader
considerationsabout economicdevelopment.
For example, the growth of defense spending
crowds out agricultural investment,which in
turn reduces food output and increases
poverty. Alternatively, rapid increases in
military expenditure, say during a war, can
cause differentialprice effects, which in turn
lead to entitlementlossesand possiblyfamine.
In a series of papers, Deger (1984, 1985,
1986) and Deger and Sen (1983, 1987) have
used analyticalgrowth theory to identify three

MilitaryExpenditureand EconomicDevelopment: Issues and Debates

major channels through which defense can

impact on growth. These can be christened
"allocation,"
and
the "spin-off,"
"mobilization"effects. The first includes the
entire host of non-quantifiablefactors,such as
'modernization'or 'discipline' that militarism
can provide to relatively backward societies.
It also includes economic services that the
militarymay put into the economy.For armsproducing countries, there also can be interindustriallinkages,dual-purposeR&D, as well
as technologicalfeed-backsfrom sophisticated
weapons manufacture. The second channel
concerns the use of resources that could go
into investmentbut instead are transfered to
the military. This does not mean, of course,
ithatin the absenceof military spending,all of
the released resources would automatically
flow into investment. Identifyingthe amount
of crowdingout is importantand can only be
,doneempirically. Finally, additionaldefense
spending can mobilize new resources
(taxation, inflation,foreign aid from military
alliancepartners) or reduce saving (by raising
the consumptionpropensity as other public
goodsdiminish, or by reducingforeignsaving
used up for arms imports). The aggregate
effect, combiningboth negative and positive
elements, can be of either sign; it is certainly
complex.
Overall,therefore,from the points of view
of both theory and empirical analysis, the
military expenditure-growth relationship
should allow for the following aggregate
channels:
(1) A direct effect of defense spending on
growth through various spin-offsthat may
on balance be positive (these are
essentiallythe points discussedby Benoit);
(2) An indirecteffectthroughthe savingsrate,
reflectingthe fact that militaryexpenditure
increases government consumption and
reallocates its savings away from
productive investment;

Deger

(3) Resource mobilization and a possible

decreasein the private sector's propensity
to save may be diminished as household
expenditure goes up to compensatefor
lower state spendingon civilian, publicly
funded goods; and
(4) Open economy considerations claiming
that military imports may crowd out
civilianimports and reduceforeignsavings
enteringthe country; on the other hand, if
military aid is used to reduce domestic
burdens or if donor countries give more
aid to strategic allies, then clearly a
positivenexus is set up.
Models of the second type are concerned
with developmentalspecificity and country
sub-groupings according to economic
characteristics. One analytical issue is
concernedwith the non-linearities(the term is
due to Boulding 1974) in the defense growth
relation. Essentially this implies that the
parameter (showing the effect of military
burden on growth) may be negative for one
range of countries; yet for another range of
countries, it might become positive. It is of
course not easy to identify theoreticallywhich
type of country falls within which category.
A workinghypothesismight be that, for small
values of the defenseburden, there may well
be positive effects that overcome the growth
retardation arising out of lost investment.
However, once a 'threshold' is crossed, the
negative role of resource reallocation and
mobilizationdominates. Thus we could expect
countrieswith relatively low military burdens
to have non-negative parameters while
countries with high burdens will show the
standardnegativeeffectsof military spending.
Another related way of looking at nonlinearities(Deger 1986)is to claimthat high-,
low-, and middle-income countries have
different ways of responding to military
expenditure. It is possible that high- and lowincome countries can derive and absorb the

39

Military Expenditureand EconomicDevelopment

benefits of military expenditure so that the

negative impact is modified. On the other
hand, middle-incomecountries tend to suffer
more than they gain from increased defense
spending. The explanationbehind this nonlinearityis important. All countries,whatever
their per capita income, lose out from
allocation effects, with military spending
reducingthe amountof resourcesavailablefor
productive expenditure. But the military also
has spin-off effects. It is possible that highand low-income economies receive more of
the beneficial effects of defense-relatedspinoff than do the countriesin the middle. Lowincome countries at an early stage of
development may gain from the whole
plethora of concepts most generally termed
'modernization.' If the country is not
specificallyat war, the civic action programs
of the military may help in development.
Even if the countryis at war, the militarymay
contribute to the cohesion of the state; for
example, as pointed out by Kennedy (1974),
Nigeria's civil war may have helped to build
the nation rather than divide it, due to the
exemplarybehavior of the central army after
victory. High income-counMries,
on the other
hand, gain from technological spin-off and
effective demand creation (Greece, South
Korea), particularly if they are starting their
own arms-buildingcapacity. Middle-income
countries are sufficiently advanced in
developmentnot to get much leverage from
modernization, yet their economies are not
advancedenough to benefit from the militaryindustrialcomplex. Thus the allocationeffect
dominatesthe weak spin-off, and increasesin
militaryburden tend to depress growth. This
is clearly an area for further empirical
research.
Another variant of such classification
schemes is to group countries as resourceabundant (rich) and resource-constrained
(poor) and see whether there is a difference
between them (see Frederiksen and Looney
1983, 1986). In particular, countries with

40

high endowmentsof foreign exchange, say

after an oil price rise, might withstand the
negative impact of resource re-allocation
towardsthe military, at the same time gaining
the fruits of spin-off.
The third approach to understandingthe
relationship of military expenditure to the
macroeconomy is through the concepts of
structuralism. This is particularly true of
short-mn economic behavior, on which we
concentrate. Instead of dealing with the
aggregate nature of the economic system-a
common foundation of demand management
and supply considerations of classical/
Keynesian models--structuralist theories
emphasize the dualism, or dichotomy, that
characterizes the less developed countries'
socio-economicstructures. The best economic
way to identify the dualism is through the
Hicksianflex-price,fix-price distinction. The
fix-price sub-system behaves in Keynesian
fashion; there is usually surplus capacity, and
aggregate demand (through inventory
adjustments) determines output. Mark-up
pricing, throughunit costs and the profit rate,
determinesthe price level in this sector. Thus
cost inflation is an important ingredient of
overall inflationarypressures. The flex-price
sector, on the other hand, obeys a Walrasian
adjustmentmechanismwherebyexcessdemand
correctionis soughtthrough price adjustments.
Output supply depends principally on
technology(and fixed factors such as land or
capital), while price inflationis motivatedby
demand.
The traditional structuralist literature
labeled these two dichotomous sectors as
manufacture and agriculture. A wider
classificationcoulddistinguishbetweenurban
or rural sectors or between large-scale and
small-scaleproducers of goods. The crucial
distinction is the ability--in terms of cost,
technology, and profitability--to hold
inventories. Smallpeasantfarmers fall within
the flex-pricesector, since they are forced to

MilitaryExpenditureand EconomicDevelopment: Issues and Debates

sell their marketable surplus as quickly as

possible; (paradoxically, these 'subsistence'
farmers are generallyalso the largest suppliers
of agricultural produce in the market due to
their large numbers, financial vulnerability,
and the necessityto pay usuriousinterestrates
in the unorganizedmoneymarket). The same
may also be true for small-scale industrial
producers. On the other hand, major
industries, and even large landholders, can
afford to wait (at least in the short term) until
demand picks up.
Within this framework,the role of military
expenditurein the economybecomescomplex.
It is not simplya vehicleof aggregatedemand;
nor is it just an agent of crowdingout. If it
creates demand in the fix-price sector (say
through arms production), then employment
and income here will rise. This creates
additional demand for the flex-price sector's
output (food, basic consumption),which in
turn will accelerateinflationthere. The feedback through worker's higher wage demands
(due to increases in the cost of living) also
produces inflationarypressuresin the fix-price
sector. If workersare strong enough (as they
usually are in the modern sectors of less
developedcountries), then there even may be
a rise in real wages in the fix-price sector.
Thusit is possible to have a decline in the
relative price of the flex-price sector. If the
latter is defined(predominantly)as agriculture,
then we witness a decline in the agricultural
terms of trade and a fall in agriculturalsupply.
But a real wage increase may also, along the
factor price frontier, erode the profit rate. A
possible decline in profitability will cause a
fail in the growth of capital formation;hence
the economy suffers in the long run. These
postulated effects can take place under any
increases in exogenousdemand--say,coming
from higher governmentexpenditure. What is
so special about defense is its essentially
'unproductive'nature. The structuraldilemma
arising out of the economic dualism is
exacerbatedby the expansionof the defense

Deger

sector--since its inter-industrial linkages are

low, its impact multipliers weak, and its
technological spin-offs for growth and
investmentsmall.
Many of these foregoing theoretical
constructs do not explicitly tackle a central
issue in the military expenditureprocess: the
predominantlypoliticalnature of this category
of government spending. Such political
contaminationof economicrelations makes it
difficultto analyzethe purely economiccostbenefit aspectsof the problem. The politics of
defense can be explicitly integrated into the
economicsof defense in a number of ways,
not all of which are mutuallyexclusive.
First, there must be an explicitrecognition
that defense spending is done to enhance
security. Its use as a fiscal stabilizer, or to
provide an economicmultiplier effect, is not
very important in less developed countries,
which tendto be generallysupply-constrained.
Thus any estimatedgrowth equationwill have
a parameter showingthe effect of defense on
growth that will be heavily influencedby the
fact that security can influence the
environment within which the economy is
operating. When this parameter is positive,
this does not necessarily mean that military
spending has direct economic benefits:
possibly the impact is indirect--through a
better security climate. Alternatively, nonmilitarymeans, rather than defensecapability
alone, could have been used to enhance
security.
Second, it may be useful in empirical
analysisto consider the causes and effects of
defense spending together, preferably in a
simultaneous equation system. The former
will emphasizethe political aspects,while the
latter will underline the economic costs and
benefits. There is certainly a need to have a
joint estimation model, with economic and
military variables analyzedtogether.

41

MilitaryExpenditureand EconomicDevelopment

Third, just as countries are classified by

economic characteristics (high or low
incomes),they should be classifiedby political
characteristics (high or low government
legitimacy,etc.). In analyzingthe causes of
military expenditure in less developed
countries, Maizels and Nissanke (1987)
distinguish between military influence,
political factors, and economiclinkagesat the
domestic, regional, and internationallevels--a
matrix of nine elememts. Rothstein (1987)
makes a distinction beitweenhigh, medium,
and low legitimacyof governmentson the one
hand, and internal,external,and mixed threats
on the other. Such characterizationscan be
integratedwith purely economicones to get a
finer distinction between countries and to
create more disaggregatedcross-sections.
The Econometric Evidence
We now turn to a select survey of the
empirical literature in the light of the three
approachesdiscussedin the previous section:
growth models, non-linearrelationshipsarising
out of specific characteristics,and structuralist
analysis.
For the first approach,both the theoretical
model and the multiplicity of channels
analyzed imply that a simultaneousequation
model is called for. The problem with
econometric modeling of this complex
relationship is partly data unavailability,but
partly also the fact that for identificationwe
need a large number of exogenousvariables
which are unavailable. For most less
developed countries, outside planning
ministries, and central statistical offices, the
required volume of consistent long-run time
series data simply is not available. Crosssection data is therefore the most preferred
vehicle of analysis. But even here, most of
the post-Benoit studies tend to use singleequation estimates--once again to prove or
disprovethe claimedposilLive
effectof MILEX

42

on growth. This is unfortunate, since it goes

against the theory, calls for an undesired
simplification, leaves out some important
avenuesthrough which the relationworks, and
can also lead to specificationerrors. As we
have shown in our earlierwork (Deger,Deger
and Sen, passim) where the data is available
(on a cross-sectionbasis), one should use a
more complex model and use a systems
method such as 3-stage least squares (3SLS)
for estimation. But unfortunately, serious
Third World data problemspartly explain the
paucity of time-seriesanalysis.
It may be argued that a single equation
with growth rate regressedon the appropriate
military variable (usually the burden), plus
other exogenousvariables, is adequate, since
it providesthe reducedform of the appropriate
structural form given by the simultaneous
model. However, a number of problems
surface in the literature that relies on single
equationestimates. Quite often, variables are
used as independentregressors(such as saving
rate affectinggrowth) that should have been
dependentvariables in a more general model.
In addition, even if the growth equation is
properlyspecified, it shouldbe at least jointly
estimatedwith a military equation to ensure
that the Zellner problem of seemingly
unrelatedregressionsdoes not appear.
Avoidanceof a simultaneitybias through
the use of systems estimation is desirable.
However,the researcher is still faced with the
problem of specification errors inherent in
3SLS methods, whereby the parameter
estimatesof the system can be affected by a
specification problem in any one equation.
This is particularly true for the military
expenditure equation, where the exogenous
variables that explain defense resource
allocation are not easy to specify. In
particular, the contamination of economic
variables with political factors, and the added
difficulty of empirically specifying

MiliitaryExpenditureand EconomicDevelopment: Issues and Debates

satisfactorily such factors as 'threat',

'security', 'stability,' etc., means that
specification errors are likely. Careful
researchers must therefore conduct extensive
sensitivity tests and detailed diagnostics
checks, including 'data-mining,' and report
only robust results.
The so-calledCowlesFoundationapproach
to testing the military expenditure-growth
relationship by using cross-sectiondata and
siimultaneousequation models as given by
economic theory has yielded useful
conclusions. To strengthenthe approach,one
should utilize explicit tests for exogeneityof
variables where the relevantequation is overidentified. I have used the Hausman (1978)
tests in the context of the model presented in
D)eger (1986) and find that the military
expenditurevariable can indeed be treated as
exogenous for the growth equation per se,
even though the 4-equation systems method
treats both as endogenous.
The problem of simultaneity and
exogeneity is a recurring theme in the
literature. Joerding (1986) questions singleequation estimations of the military
expenditure-growthrelationby askingwhether
defense spending is exogenous, statistically,
with respect to growth of GDP. Applying
Granger causality tests, he finds that the
hypothesis of Granger non-causality from
MILEX to growth cannot be rejected by his
sample observations. Alternatively, Granger
non-causality from growth to MILEX is
rejected by the data (see also Chan, Hsiao,
and Keng 1982). However,such 'a-theoretical
macroeconometrics' (see Cooley and LeRoy
1985) cannot analyze the form in which
structural macro models should be estimated.
Rather, this type of causality analysisshould
be constructedas a critiqueof single-equation
modelingand a recognitionof the necessityof
simultaneousand/or dynamicsystems.

Deger

Another factor that needs to be explicitly

accountedfor is the impactof lagged effects.
If military spending does reduce growth and
create distortions, it probably does so over
extended periods of time. Moreover, the
adverse impact of defense on development
becomes stronger with time, since there is an
element of cumulation involved. Thus
empirical models should analyze 'delayed'
impacts. The current belief is that crosssection studies across countries with widely
diverse socio-economic features tend to
estimatelong-termparameters. However, the
need for explicitly lagged models--preferably
with some dynamicstructuresinvolved--could
be more fruitful.
Finally, there is a great need for time
series modelingfor individualcountries. This
would face the problem of exogeneity
variables involved, particularly military
expenditureand growth. It would also allow
explicitmodelingof truly exogenousvariables
(such as conflict), which affect both defense
and development. Such modelsalso allowfor
dynamics and the appropriate testing for the
error terms. Much research remains to be
done in terms of appropriate modeling of
individual less developed countries-specificallythose whose data base is relatively
strong.
A comprehensiveeconometric model is
providedin Deger (1986)to which the reader
should turn for more details. It is a crosssection empirical model with four equations,
each identified with a particular channel
discussed earlier. Using data for fifty
countries and time series averages for 196573, the followingmodel is estimatedby threestage least squares, which emphasizes the
structural simultaneityof all the relationships.
(Here g is the growth rate, s is the savingincome ratio, m is the military burden, and B
the trade balance (as a ratio of GDP) to

43

MilitaryExpenditureand EconomicDevelopment

identify foreign saving; Z; are a set of

exogenous variables chosen through data
specification.)
g = ao + a,s + a2m + a3B + a4Z1

(1)

s = bo + b,m + b2g + b3B + b4 Z2

(2)

B = c0 + cIm + c29 + C3Z3

(3)

m =do + d1 Z4

(4)

The fourth equation for 'm' is a function of

strategic, security, and wealth variables--all
representedby the vector Z4.
Consideringm as an autonomousvariable,
the three interdependent equations can be
solved simultaneouslyto get the impactof the
military burden on growth, the saving ratio,
and the trade balance. In each case, the
negative effects are predominant and
significantlyhigh. For example,it was found
that (dg/dm) = -0.36; thus, if the average
military burden for these countries were
reducedby one percentagepoint, then growth
would be increased by over one-third of a
percentage point--a not insignificantamount.
The results for savings behavior and trade
performance are even sitrongerand show that
high improvements in both spheres are
possible with disarmament. Overall, the
results from this more complicated model
suggest that military spending is strongly
detrimental to growth, investment (saving),
and the trade balance when all direct and
indirect effects are taken together.
The econometric results for the second
group of studies are more mixed. Frederiksen
and Looney (1983) claim that resourceabundant countries tend to have positive
effects of military expenditures on growth,
whereas resource-poor countries tend to have
negative effects. Similar results are obtained
for arms-producingcountries (positiveimpact

44

of defensespending)and countriesthat do not

produce weapons (negative impact). An
alternative test within this class of models-proposed by Deger and Smith (1983), is to
classifycountriesaccordingto high-, low- and
middle-incomeranges and then observe the
effects for each group. The results from the
threefold classificationshow that all groups-high-, low-, and middle-income, tend to
demonstratea negativeimpact.
The best structuralistmodels in the field
emanate from the MassachusettsInstitute of
Technology, and a good representativepaper
is by Faini, Annez, and Taylor (1984). The
authors take a series of variables that reflect
the disaggregatedconcerns of structuralism:
investment, imports, industrial production,
agricultural output, and tax receipts--all as
ratios of GDP. These ratios are regressed,
individually,on the military burden as well as
a number of basic developmentalindicators
(GDP per capita, population, etc.) Nonlinearities are tackled by taking logs and
squares of these economicvariables. Using
pooled time-series/cross-sectiondata for a
large sample, they find that the coefficientfor
the defense burden is generally negative and
statisticallysignificantin the regressionswhere
investment and agriculture (as GDP shares)
are the relevant dependent variables. Since
investmentis a major structuralbottleneckand
agriculturethe main lagging sector (the flexprice part of the economy subject to
stagflationaryfluctuations),the negativeeffect
of defensehere is all the more striking. Their
estimatedequationis of the followingtype: (X
stands for the ratio in GDP of (in turn)
investment, imports, industrial production,
agricultural output, and tax receipts; y is per
capita income; N is population; KI is capital
inflow; and m is military burden):
X = ao + a1 (log y) + a2(log y)2
+ a3 (log N) + a4 (log N)2
+ a5 (KI) + a6m

(5)

MilitaryExpenditureandEconomicDevelopment:IssuesandDebates

More generally,Faini, Annez, and Taylor

construct a theoretical structuralist model
where the growth of nationaloutput is related
to the growth of exports, population,and total
capital (reflectingpossibleshortagesin foreign
exchange, labor, and capital), change in
foreigncapital inflows, the level of GDP per
capita as well as the change in the military
burden. The last variable gives the growth
effect of defense. What is significantis that
this is the only model that uses the change
rather than the level of militaryexpenditureas
a share of GDP. This coefficient,representing
the impact of defense on growth, is
consistently negative for less developed
countries as a whole as well as for regional
sub-groupingsof these countries.
Overall,the results are relativelyclear-cut.
There can be no doubt that for some countries,
at some times, militaryspendingmighthave a
positive impacton the economy'sgrowth rate.
The guns and butter trade-offis not an obvious
one. However, as soon as one considersmore
complexmodels and looks at both direct and
indirecteffects, it becomes clear that military
expenditureis a major economicburden and
liability. The empirical and econometrics
evidence for simple models is rather
ambiguous. But the message from more
complex models seems to be clear: defense
spending tends to reduce growth-and in
significantfashion.
What abouttime-seriesanalysisfor single
countries? Here the familiar data problems
mentionedearlier become critical. There are
very few aggregate studies to match the
complexityand estimationsimultaneityof the
cross-sectionones. Faini, Annez, and Taylor
have done some empirical work on India,
1950-72. In this case study, the empirical
results of equation(5) show interestingeffects.
The authors find evidence that military
spendingseems to have helpedcertainareasof
Indian economicperformance. For example,
the investment/GDPratio seems to rise with

Deger

the militaryburden-almost one to one. In the

same way, industrial production is affected
positivelyby the military burden. However,
there is a sizable, and quantitatively the
largest, negativeimpact on agriculture. Why
investment and industrial production are
positively affected by the military spending
process is not made clear in the paper. It
could be because of arms production, since
from around 1960India emphasizeddomestic
weaponsindustrialization(a militaryversion of
import-substituting industrialization, ISI).
However, a disaggregated microeconomic
analysisfor India (Degerand Sen 1983)on the
industrialeffectsof militaryexpenditureshows
that the inter-industrialspin-offof defenseon
the outputof major industrialsectors has been
negligible. The impactof defensespendingon
the value added in five manufacturingsectors
is examined separately, and only for one-metal products-is there a modest positive
impactof military expenditure.
Another recent study of the military
expenditureprocess in the Indian subcontinent
(Deger and Sen 1990)also tries to estimatethe
growth effects of defense spending in the
context of an arms-race model for India and
Pakistan. The implicit assumption of the
economic model is that military spending
affects growth in two ways. The first is the
direct effect embodying the whole host of
spin-off effects, including that of security on
the economy. The second is indirect; this
works through defense spending, affecting
investment. Investmentin any economyis the
sum of the savingsemanatingfrom the private
sector, the government,and the foreign sector
(negativetrade balance). Insofar as military
spending affects each of these elements, it is
bound to have a major role in the investment
process. In addition, greater security due to
defense often impacts more directly on
investmentthrough increased confidenceand
optimistic expectations. Investment, in turn,
will influencegrowth.

45

Military Expenditure and Economic Development

The postulated empirical model therefore

assumes that growth is a function of
investment and military spending--both as
ratios of GDP. In addition, investmentitself
is affected by the military burden. Three
additionalpoints are utilized in the empirical
model. First, due to life cycle effects as well
as the assumptions of the Metzlerian targetsaving hypothesis, we assume that savings,
and therefore investment,is also a functionof
growth. An accelerator type of investment
function will also give the postulated relation
between investment and growth. Second,
structural rigidity may preclude investment
being productive at the period at which it is
done; the installationand use of capitalin poor
countries takes time, principally due to the
lack of cooperative factors--the so-called
absorptive capacity constraint. Hence the
effect of investmenton growth might come
after a lag. Finally, due to considerable
interest in analyzing inflationary effects on
growth, this term is also included for the
empiricalanalysis(see Sen 1991).
The empirical results for India and
Pakistan, using the two equationsfor growth
and investment, are based on (6) and (7).
(Here g is growth, m is the militaryburden, i
is the investmentshare in (GDP,and P/P is the
rate of inflation).
g = ao + a1m + a2i(-]) + a3 (P/P)

(6)

i = bo + b1 m + b2 g

(7)

Taking both equations together, the impact

multiplier of m on g is givenby a1 ; the longrun, steady-statemultiplier is derived from:
(a, + a2b,) / (1 - a2b2).
The empirical results ,showthat the two
countriesvary quite dramaticallyin terms of
the economiceffectsof the militaryburden on
growth (Table 1). For India, the impact
multiplier from (6) is positive, though not

46

highly significant. The long-run multiplier,

using both (6) and (7), is slightly higher.
Overall, the results imply that defense could
have a positive effect on the aggregate
performanceof the economy. Once again, it
may be thought that the positive impact of
defense arises because of substantial arms
production, which has a strongly positive
impacton domesticindustrializationand R&D
as well as other forms of spin-offs. However,
as mentioned earlier, previous work with
inter-industrialdata (Deger and Sen 1983) at
a disaggregatelevel for India shows that the
claims for spin-off are not supported by the
econometrics. A more plausible explanation
may be the one given earlier: with a relatively
low defenseburden, the adverseeffectsdo not
particularly bite--since the industrial and
investmentbase itself is very large.
For Pakistan, the case is potentially
different; a muchlarger burden superimposed
on a smaller investment base can have
detrimental effects. The actual empirical
results for Pakistan are also more clear-cut.
Defense has a small positive impact on
growth; but the coefficient is statistically
insignificant, so that positive and negative
direct effects are cancelling each other out.
But the indirect effect, through investment
crowdingout, is stronglynegative. The final,
steady-state, impact of defense on growth is
negative. The tank and tractor trade-off for
the Pakistan economy is quite crucial. Even
though the military expenditureprocess is not
subservientto the demands of the economy,
giventhe autonomouscharacterof the military
institutions, the feed-back is important. The
cumulationof growth retardationmay make a
long-run arms race untenable, through
economic constraints, although in the intermediateterm it is a major securityissue in the
region.
In anotherstudy, Scheetz(1990),using his
own data set from the national accounts
directly, estimates the impact of defense on

Deger

MilitaryExpenditureand EconomicDevelopment Issues and Debates

Table 1. Military Burden and the Economy: Empirical Results

India

Pakistan

(1)

(2)

(3)

(4)

Constant

-0.06

0.23

-0.07

0.21

rn

1.84
(1.17)

0.67
(2.35)

0.05
(0.07)

-0.80
(-2.51)

i(-l)

0.35
(1.56)

0.65
(1.64)

P/P

0.35
(4.02)

0.16
(1.67)

-0.05
(-0.42)

-0.001
(-0.014)
0.5157
-0.30
(-0.50)

rho

0.9365

0.5376

0.7087

0.98
(2.50)

0.60
(2.85)

0.40
(5.18)

Impact Multiplier

1.84

0.05

Long-Run Multiplier

2.07

-0.46

Note: The t values are provided in brackets.

Source: Deger and Sen (1990a).

growth for Chile, Argentina, Paraguay, and

Peru. The model used is exactly the same as
in equations (1) to (4) except for a few
changes in exogenous variables. A pooled
time-series/cross-section study is also done for
these countries taken together; it yields strong
evidence that the impact multiplier (the direct
effect of the military burden on growth rates)
and the systems multiplier (direct and indirect
effects combined) are both statistically
significant and negative.

This country study for Latin America

gives mixed results. Evidence that defense
spending has had a major negative impact on
economic growth is strongest for Chile (during
1969-87). There is some evidence that the
military expenditure-growth relationship is
negative in the cases of Argentina and
Paraguay. Only for Peru were the results
indeterminate, with both of the multipliers
statistically insignificant. The equations were
also not well specified for Peru, and

.
47

Military Expenditure and Economic Development

specification and diagnostic checks were

unsatisfactory.
Economic Development
Military Expendituire

and

The focus here, followingthe literature in the

field, has been on econtomicgrowth; this is
primarily because growth is subject to
quantification, and this is the minimum
necessary condition for empirical analysis.
Clearly, the resulting picture is only partialsince economic developmentin the broadest
sense must includemany other attributes. The
objective of developmentin less developed
countries was never meant to be growth per
se; developmentwas expectedto improvethe
physicaland social qualityof life, and growth
was a meanstowards that end.
Alternative paradigms to growth have
aboundedin the area of developmentstudies.
The 'basic needs' approach, for example,
emphasizesthe minimumlevels of well-being
that are required as the sine qua non of
development. The 'physical quality of life'
indicatorsare anotherway of measuringsocioeconomicimprovementsover and above the
standard measure of per capita income. A
broader approach--almostan 'umbrella' to all
of the models-has been proposedby Amartya
Sen. He has argued forcefully for a shift in
focus towards the study of 'entitlements' of
individualsand groups of peoplein developing
countries, as well as the 'capabilities' that
these entitlementscan produce. Sen (1983)
states:
Entitlementrefers to the set of alternative
commodity bundles ithat a person can
commandin a societyusing the totalityof
rights and opportunitythat he or she faces.
On the basis of this entitlement,a person
can acquire some capabilities, i.e., the
ability to do this or that (e.g., be well
nourished), and fail to acquire some other
capabilities. The process of economic

48

developmentcan be seen as a process of

expanding the capabilities of people.
Given the functional relation between
entitlements of persons over goods and
their capabilities, an useful-though
derivative-characterization of economic
developmentis in terms of expansion of
entitlements.
Economic developmenttherefore implies
the acquisitionof entilements (food,medicine,
education, etc.), which give rise to better
capabilities (well nourished, healthy,
educationallyqualified, etc.). The problem
thatarises in empiricalmodels is definingthis
rather amorphousconceptof entitlementmore
precisely. We have looked at two types of
indicators: inequality of income distribution,
which gives a good indication of relative
deprivationand possible loss of entitlements
that can be acquired through the market; and
the quality of life as measured by socioeconomicand physicalwelfare.
Table 2 gives data for ten countries on
these measures in the form of ranks, which
avoidsthe problemof wide disparityand noncomparableunits. The first column gives a
ranking in terms of these countries'
performance in an aggregate socio-economic
index(SEI). The SEI is constructedfrom the
following information: (a) percentage of
school agepopulationin school; (b) percentage
of women among university students;
(c) literacy rate; (d) physicians per head of
population; (e) hospital beds per person;
(f) infant mortalityrate; (g) life expectancyat
birth; (h) per capitacalorie and protein intake;
and (i) percentage of total population with
access to safe drinking water.
The
combinationmeasureSEI therefore represents,
in principle, an aggregationover entitlements
(and capabilities where relevant) to basic
education, literacy, women's educational
rights, health care, longevity, nutrition, and
basic sanitation. The best performer gets a
rank 1, and so forth. The second column of

Military Expenditureand EconomicDevelopment: Issuesand Debates

Deger

Table 2. Rank Order for Aggreate Socio-EconomicIndex (SEI),

Income Distribution, Military Burden,
and Growth of Real Per Capita Income

Countries
Argentina

10

Costa Rica

10

Venezuela

SouthKorea

SriLanka

Malaysia

Mexico

Brazil

10

Philippines

10

Peru

Inoome
Distribution

Military
Burden

Growthof Real
Per Capita
Income

Aggregate
SEI

Source: Sivard (1983); Sen (1981);SIPRI data base; and World Bank (1983).

Table 2 ranks the countries according to

simple measures of income distribution (Sri
lanka is the best and Brazil the worst). The
two together give a general view of economic
development in the broadest possible sense.
Clearly, the lower the rank, the better the
situationin terms of development. The third
column gives ranks for the military burden.
To control for growth performance, which
may favorably affect entitlements, we also
have a fourth columnreportingreal per capita
iincomegrowth. Here lower ranks mean
higher levels of the two variables. All basic
data pertain to some periods in the late 1970s

and early 1980s; the military burden data is

for 1975-80;growth data is for 1970-81. It
should be noted that the positions can change
dramatically if there are major security
shocks, such as for Sri Lanka and Argentina;
but the fundamental rankings for most
countries under normal situations remain
similar.
The relationship between economic
development and the military expenditure
process can be appreciatedby looking at the
ranks. Sri Lanka does very well indeed in
terms of economic development in spite of

49

Militairy Expenditure and Eco'wmic Development

havingrelativelylow grolwthrates. The same

is true for Argentina; it has the lowest growth
rate and the best performance in SEI and
income distributioncombined. Brazil, on the
other hand, fails to utilize its growthpotential
to improve its low status in socio-economic
development. Similar staitementscanbe made
aboutMalaysia.
When it comes to looking at the military
burden, we observe that the top five countries
in the first two categoriestend to have a low
military burden. The only exceptionis South
Korea; a high quality of life and more
equitableincome distributioncoexist with the
secondhighestmilitaryburden. For the other
five countries--Malaysia,,Mexico, Brazil,
Philippines, and Peru--the relation works,
generally, in opposite fashion. Peru, for
example, has the highest military burden and
the worst position in terrns of entitlements.
Mexico is an 'outlier'--with a low military
burden failing to compensatefor entitlement
losses.
The author is currently conductingfurther
econometricanalysis,with cross-sectiondata,
to further the qualitative results presented
above. The basic structure is similar to the
four-equation empirical model discussed
earlier. An additional equation, with a
measureof entitlements(essentiallythe Borda
measure--a joint index oiFSEI and income
distributionby addingranks), is used to get a
system of five equationsvwhichare estimated
jointly.
Preliminary estimates of the
multipliers show that military spending
significantly lowers the entitlement index.
Hence,higherdefenseexpenditure,for a given
GDP, lowers the socio-economicdevelopmentallevels of developingcountries.
The central reason for this negative
relationship is the crucial role that
governmentsin less developed countriesplay
in providing a higher quality of life and
greater entitlements. As Sen (1981) has so

50

strongly emphasized, the role of the state,

'public action,' is probablythe most important
ingredientin entitlementenhancementfor most
developing countries. With the exceptionof
certain countries such as South Korea, the
'trickle down' effects take a long time to
appear. Aggregate growth and the market
mechanismby themselvescannotbreak down,
rapidly, the barriers of the poverty trap and
ensure a higher quality of life and egalitarian
incomes for the deprived. Government
interventionis vital, at least in the short run.
But one of the centraltrade-offsoccurs in the
budgetary process between defense spending
and expenditure for social and economic
services. Governments in less developed
countries, enmeshed in regional security
problems, must give first priority to national
security when the need arises. The essential
crowdingout that occurshere betweendefense
expenditure and other spending on health,
education, welfare, and social security
provisions, is a barrier to development. A
specific case study for child developmentin
Sub-SaharanAfrica shows that low defense
spending was associated with a reduction in
infant mortality rates, while rapid growth of
defenseexpenditurewas linked with a slowing
of the rate of decline of infant mortality (see
Deger and Sen 1990). Incidentally, the
history of the Reagan era shows that the
problem is not unique to less developed
countries. But for poor societies, the choices
are more stark and the trade-offmore tragic.

Concluding Remarks
Military expendituredoes have some positive
effects on economic growth in developing
countries. Aggregate demand expansionmay
produce beneficialmultipliers. Certain types
of spin-off may operate. The 'modernizing'
role of a benevolentmilitary, if it exists, may
help in nation-building. The tractor-tank
trade-offis not obvious; nor is the problem a
vacuous one. However, when all direct and
indirect effects are taken together, defense

MklilitaryExpenditure and Economic Development: Issues and Debates

spending,on average, has a serious impacton

growth and development. The aggregate
effect is undeniably negative, and this
conclusion is strongly supported by more
complexeconometricmodels.
The empirical literature has claimed that
three alternativetypes of effects, from defense
to growth, can be identified: zero, negative,
and positive. In the first case, defenseeither
does not affect growth at all, or the contrary
effects cancel each other out. However, this
also implies that resource savings and
judicious re-allocation away from military
spending will have a positive impact on the
economy in terms of both growth and
development. In the second case, the
justification for resource diversion from the
military is very strong indeed, for structural
features show that defense reduction will
stimulate growth. For the countries in the
third case, the argumentsfor militaryspending
reductionsare diluted, even though it may be
arguedthat alternativepublic expenditureswill
have much greater growth multipliers than
defensespending.
The days of estimating single-equation
military/growthestimationsand debatingabout

Deger

positive and negative parameters are over.

Cross-sectional studies must have more
homogeneous data sets (specific region,
common economic characteristics, similar
security perceptions)and need to be based on
simultaneous-equation systems estimation.
Specific country studies are probably more
important. Wheredata problemsdo not allow
adequate coverage, the possibility of pooled
time-series/cross-sectionestimation of a few
countriesover time should be attempted.
As long as developing nations perceive
genuine securityproblems, they will continue
to spend on the military. What is really
needed is a proper cost-benefitanalysis, so
that the economiccosts are starkly presented
as opposedto security gains. The militarylike Ceasar's wife-is often beyond question.
This is unfortunate, since the economics of
militaryexpenditureis too importantto be left
to the military alone. Arms control and
disarmamentmeasures have a greater chance
of success if the costs are highlightedand put
in perspective. Country and regional analyses
are now essentialto demonstratethe massive
opportunitycosts of militarism. They could
demonstratethat disarmamentis a worthwhile
and a major economicalternative.

REFERENCES
Notes
Ball, N. 1988. Security and Economy in the Third
World (Princeton, N.J.: Princeton University
Press).
Benoit, E. 1973. Defense and Economic Growth in
DevelopingCountries(Lexington,Mass.: Lexington
Books).
Benoit, E. 1978. "Growthand Defensein Developing
Countries," Economic Developmentand Cultural
Change.

Boulding,K.E. 1974. "DefenseSpending:Burden or

Boon," War/PeaceReport.
Chan, L., C. Hsia, and W. K. Keng 1982. 'Defense
Expenditureand Economic Growthin Developing
Countries:A TemporalCross-SectionalAnalysis,"
Applied TuneSeries.
Chan, S. 1985. 'The Impact of DefenseSpendingon
EconomicPerfornance: A Surveyof Evidenceand
Problems," Orbis.

51

Military Expenditure and Economic Development

Cooley, T. F., and S. F. LeRoy 1985. "A-theoretical

Macroeconometrics,"
Journal
of Monetary
Economics.
Deger, S., and R. Smith 1983. "Military Expenditure
and Growth in Less Developed Countries," Journal
of Conflict Resolution.
Deger, S., and S. Sen 1983. "Military Expenditure,
Spin-Off and Economic Development," Journal of
Development Economics.
Deger, S. 1984.
"The Developmental Effects of
Military Expenditure in LDCs," in Conference
Proceedings of the International CoUloquiwn on
Armaments,
Development,
Human
Rights,
Disarmament (Paris: UNESCO).
Deger, S. 1985. "Human Resources, Govemment
Education Expenditure and Military Burden,"
Journal of Developing Areas.
Deger, S. 1986. "Economic Development and Defense
Expenditure," Economic Development and Cultural
Change.
Deger, S. 1986. Military Expenditure and Economic
Development: The Economic Effects (London:
Routledge).
Deger, S., and S. Sen 1987. "Defense, Entitlement and
Development," in Deger, S., and R. West, eds.,
Defense Security and Development (London:
Frances Pinter Publishers).
Deger, S., and S. Sen 1990. "Military Security and the
Economy: Defense Expenditure in India and
Pakistan," in Hardey, K., and T. Sandler, eds., The
Economics of Defense Spending: An International
Survey (London: Croom Hehn).
Deger, S., and S. Sen 1990. Military Expenditure: The
Political Economy of International Security (Oxford:
Oxford University Press).
Deger, S., and S. Sen 1990. Arms and the Child: The
Impact of Military Expenditure on Child
Development in Sub-Saharan Africa, A SIPRI
Research Report prepared for UNICEF (Stockholm:
SIPRI).
Deger, S., and S. Sen 1991. "Militay Expenditure, Aid
and Economic Development.' Paper prepared for
World Bank Annual Conference on Development
Economics.
Deger, S., and R. West, eds., 1987. Defense, Security
and Development
(London:
Frances Pinter
Publishers).

52

Faini, R., P. Annez, and L. Taylor 1984. "Defense

Spending, Economic Structure and Growth:
Evidence Among Countries and Over Time,"
Economic Development and Cultural Change.
Frederiksen, P., and R. Looney 1983.
"Defense
Expenditure and Economic Growth in Developing
Countries," Armed Forces and Society.
Frederiksen, P., and R. Looney 1986.
"Defense
Expenditure, External Public Debt, and Growth in
Developing Countries," Journal ofPeace Research.
Grobar, L. M., and R. C. Porter 1990. "Benoit
Revisited: Defense Spending and Economic Growth
in LDCs," Journal of Conflict Resolution.
Hausman, J. A. 1978.
"Specification Tests in
Econometrics," Econometrica.
<
Joerding, W. 1986. "Economic Growth and Defense
Spending: Granger Causality," Journal of
Development Economics.
Kaldor, M. 1978. "The Military in Third World
Development," World Development.
Maizels, A., and M. K. Nissanke 1986.
"The
Determinants of Military Expenditures in
Developing Countries," World Development.
Rothstein, R. L. 1987, "National Security, Domestic
Resource Constraints and Elite Choices in the Third
World," in Deger, S. and R. West, eds., 1987.
Defense, Security and Development (London:
Frances Pinter Publishers).
Sen, A. 1981. "Public Action and the Quality of Life in
Developing Countries," Oxford Bulletin of
Economics and Statistics.
Sen, A. 1983. "Development: Which Way Now?",
Economic Journal.
Sen, S. 1991, forthcoming. Economic Policy and
Structural Constraints: Inflation in the Third World
(Brighton: Wheatsheaf).
SIPRI, 1988. Stockhobn International Peace Research
Institute Yearbook: World Armaments and
Disarmament (Oxford: Oxford University Press).
Sivard, R. 1983. World Military and Social Expenditure
(Leesburg, Virginia: World Priorities).
World Barnk1983. World Tables 1983 (Baltimore, Md.:
Johns Hopkins Press).

4
ADJUSTING TO REDUCTIONS IN
MILITARY EXPENDITURE AND
DEFENSE PROCUREMENT
Nicole Ball

Irntroduction
Armed forces and military industries face
reductions in military expenditure or arms
procurement for a variety of reasons.
Changes in political conditions at the global,
regional, and nationallevel affect the amount
of expenditureon weapon procurement,both
in general and on specific categories of
weapons. Rapprochementbetween East and
West, the termination of regional conflicts,
and the reassessmentof threats by individual
governments have caused at least some
countriesto start to re-examinethe allocation
of resources to their military sectors.
Purely economic factors also play an
important role. A major impetus behind the

Note The authorwishesto thankMiltonLeitenberg,

EJote:

The authorwishes to thankMiltonLeitenberg,

University of Maryland; Michael Brzoska, University of

Hamburg; Gordon Adams, Defense Budget Project;
Arthur Alexander, Japan Economic Institute; and Judith
R.eppy, Cornell University, for their comments on this
manuscript; Raul de Gouvea Neto, University of New
NMexico,for sharing his preliminary draft on defenseindustryconversion in Brail; and Margarita Studemeister
and Lakshmi Murthy Grama, The National Security
Archive, for technical assistance.

changes in Soviet foreign policy that have

significantlyreduced the likelihood of EastWest conflictderives from the urgent need of
the Soviet government to reduce militaryrelated oudays of all kinds, both at home and
abroad. To the extent that the events set in
motionby the new Sovietpolicy reducemajorpower interventionin the developingcountries
and encourage the peaceful resolution of
conflicts, the perceivedneed on the part of a
large number of developing countries to
maintain military expenditure and arms
procurementat current levels may be eroded.
The ability to finance these outlays may also
be affected if, as currently seems likely, the
major powersbecome less inclinedto provide
financial support to the military sectors of
"friendly" countries (the Gulf War
notwithstanding)and if markets for weapons
produced in the developing world contract.
Some developing countries may identify
pressingdomesticeconomicneeds and seek to
shift resources from the military sector, as
China didouring the mi8ss,
Chma did durmg the 1980s.
Another important economic considera-

tion--at least in Western Europe and the

United States--has been the constant increase

in the sophisticationand hence the cost of

mstsince the end of WorldnWarhecoOver
arms since the end of World War II. Over

53

Military Expenditure and Economic Development

Table 1. Arms Production Capabilities of 54 Developing Countries, early 1980s

Country

Ammunition

Small
Arms

Aircraft

Armored
Vehicles

Mssiles

Ships

X
X
X

Algeria
Argentina
Bangladesh
Bolivia
Brazil
Burma
Cameroon

X
X
X
X
X
X

X
X

Chile

Colombia

Congo
Cuba
Dominin Republic
Ecuador
Egypt
Ethiopia

X
m0
X

X
(X)

0)
X
X

Gabon

54

Ghana
Guatemala
Guinea
Hondurs
HongKong
India
Indonesia
Iran
Iraq
Israel
Ivory Coast
Jordan
Korea, Nozth
Korea, South
Madagascar
Malaysia
Mexico
Morocco
Nepal
Nigeria
Pakistn
Panama
Peru
iipph s
SaudiArabia
Senegal
Singapore
South Africa
Sri LAnka
Sudan
Syria
Taiwan
Thailnd
Triidd & Tobago
Tunisia
UpperVolta
Uruguay
Venezuea
Total:
Source:

X
X

00
(X)

(X)

X
X
X
X
X

X
X
X
X
X

(X)
X
X

X
X
X
X

X
X
(X)

X
X

X
X

X
X
X
X
X
X

X
X
X

X
X

X
X
X

X
X
X

X
X

X
X

X
X

X
X
X
X

X
X

X
X

X
X
X

X
X
X
X
X
X
X

(X)
X
X
x

X)
X
X
42

X
27

Michal BrzoskaandTh1m Ohlson, eds., Ams P

16

11

r&wbon
hthe TIn7dWorddandon:

X
X
33

Taylor & Francis, 1986), pp. 16-17.

Adjustingto Reductionsin MilitaryExpenditureand Defense Procurement

employmenthas been generated per unit of

expenditure. In addition, where procurement
budgets have not risen as rapidly as weapon
costs, progressively fewer of any given
wreaponsystem have been procured. Smaller
producers (that is, all those except the United
States and the Soviet Union) have frequently
found that it is more economicalto purchase
sub-systemsand componentsfrom abroadthan
to produce them domestically. This further
reduces the employment generated by local
production. As costs have risen, the life span
of weapons has been extended through
modificationand repair, which has meant that
entire weapon systems are purchased less
frequently. Increased military expenditure,
often includingrising shares of procurement,
and declining defense-industry employment
have occurred simultaneously. The defense
sectorhas becomeless and less able to support
r-ore than a handful of producers for each
kind of weapon. Although some of the slack
h.as at times been taken up by exports,
considerableresources nonethelesshave been
released from the defense sector of the
industrialized countries over the last 35
years.
This paper examines a series of issues
surrounding the process of adjusting to
(lecreased military expenditure and weapon
lprocurementin the developing countries. It
begins by identifyingthe major producers of
military equipment in the developing world
and summarizing the use of manpower by
developing-countrysecurity forces (military
and paramilitary). It then examines the two
rmainstrategies for adjustingto decreases in
defense procurement--conversion and
diversification--and considers the transiferability to the civilian sector of skills
acquired in the course of military service.
'Thecore of the paper is a reviewof the most
important factors affecting the ease or
difficulty that countries, enterprises, and
manpowerexperiencein adjustingto cutbacks
in military expenditure and defense

Ball

procurement. Finally, the availabilityof data

pertaining to these issues for the developing
countriesis assessed.

The Need for Adjustment

Industrial Production
The StockholmInternationalPeace Research
Institute (SIPRI) has identified fifty-four
developing countries that possessed some
domesticarms productioncapacityin the early
1980s, but over half of these produce only
ammunition, small arms (under license),
and/or simplesurface vessels (Table 1). The
ability to manufacturemajor weapon systems
is highly concentrated among developing
countries. Accordingto SIPRI figures, only
the first nine countries listed in Table 2
possessa broad-basedmilitaryindustry. Two
of these, India and Israel, accountedfor over
half of the value of major weaponsproduced
by the developing world (excluding China)
between 1950 and 1984. Altogether, SIPRI
identified fourteen developing countries
(excludingChina) as the main producers of
major weapon systems in the early 1980s.
Another twelve had a much more limited
capacityto produce major weapons.2
The more complicated the technology
involved, the fewer the number of countries
that have been able to use it, even under
license. SIPRI figures show that by the mid1980s, nine or ten developing countries
produced fighter aircraft, eight produced
helicopters, seven or eight manufactured
missiles, five or six produced tanks, and six
manufacturedmajor fighting ships.3 While
there is little information available on the
ability to produce major components for
weapon systems such as engines and
electronics, it is known that only a few
developingcountriespossess the capabilityto
manufacturethese items. Argentina, India,
Israel, Indonesia, Egypt, South Africa, and

s5

Military Ependiture and Economic Development

Table 2. DevelopingCountryProducersof MajorWeapons, 1950-84

Country

Percentage

CnmulativePercentage

India
Israel
South Africa
Brazil
Taiwan

31
23
9
9
8

31
54
63
72
80

North Korea
Argentina
SouthKorea
Egypt
ASEAN countries

6
5
4
2
2

86
91
95
97
99

Othersb

100

I Indonesia, Malaysia, Thailand, Philippines,

b

Singapore.
Bangladesh, Burma, Chie, Colombia, Dominican Republic,Gabon, Madagascar, Mexico, Pakistn, Peru, Senegal,

Sri Lanka.
Source: Michael Brzoska and Thomas Ohlson, eds., Arms Production in the Third World (London: Taylor &
Francis for SIPRI, 1986), p. 10.

Brazil have limited capacity.4 Chinahas for

many years produced aircraft and vehicle
engines as well as electronic componentsfor
its various weapon systems.
Reverse
engineering of Soviet equipment has been
important in building this capability. More
recently,Westernlicenseshave been procured.
The items designed in China tend to be less
sophisticatedthan their Western counterparts,
and serious questions have arisen about the
5
reliabilityof some of this equipment.
Manpower
Precise figures for defense-industry
employment in developing countries are
difficult to obtain.
Nonetheless, the
proportion of the industrial labor force

56

employed by the defense sector in most of

these countriesis (withthe exceptionof Israel)
extremely small (Table 3). While these
employment figures may be subject to a
relatively wide margin of error, the share of
industrial employment they represent is in
most cases so small that even a 100-per cent
margin of error would still produce quite a
small fraction.6
Similarly, the armed forces of most
developing countries absorb a rather small
proportion of the total labor force. Table 4
shows that in about 20 per cent of the
countriessurveyed, less than 1 per cent of the
males in the economicallymost productiveage
group (here definedas 15-44years old) are in
the security forces (military and paramilitary

Adjusting to Reductions in Military Expenditure and Defense Procurement

Ball

Table 3. Share of Industrial Labor Force Employedin Arms Industry,

early 1980s
Employmentin
Arms Industry as Percentage
of Industrial Employment

Estimated
Direct Employment in
Arms Industry

10.3
2.0
1.7
1.5
1.2

90,000
100,000
11,000
55,000
60,000

Egypt
India
Brazil
Pakistan
Indonesia

1.0
0.5
0.4
0.4
0.2

75,000
280,000
75,000
40,000
26,000

Peru
Thailand
Chile
Malaysia
South Korea

0.2
0.2
0.2
0.2
0.1

5,000
5,000
3,000
3,000
30,000

Philippines

0.1

5,000

Country

Israel
SouthAfrica
Singapore
North Korea
Argentina

The data on which arms industry employmentand total industrialemploymenthave been based are
frequentlyquite soft, and the figurespresentedin this table must be treated as estimates.
Sources: MichaelBrzoska and Thomas Ohlson, eds., Arms Productionin the ThirdWorld (London: Taylor &
Francis for SIPRI, 1986),p. 22; and World Bank, WorldDevelopmentReport, 1985 (New York: Oxford
UniversityPress, 1985), Annex Tables 1 and 2.

Not,e:

L_I

forces). In approximatelyhalf the countries,

between 1-5 per cent of this group is serving
in the security forces. In the remaining
countries in the survey, the security forces
absorb more than 5 per cent of the men in this
age group.

economically active population. Central

America and Sub-Saharan Africa have the
lowestratios, althoughin view of the increase
in conflict in several Central American
countriesduring the 1980s,it is likely that the
security force participation rate increased
during the decade in that region.

The Middle Eastern countries have the

highest ratio of security personnel to

57

Military Expenditure and Economic Development

Table 4. Security Forces as Percentageof Males in 15-44 Years Age Group, 1980
(percentagesand number of countries)
< 1.0%

1.0-5.0% 5.0-10.0% 10.1-15.0%15.1-20.0% >20.0%

Armed Forces
(101 countries)

33

50

14

Total Security
(98 countries)

22

51

18

Sources: For demographicdata: United Nations Departmentof Intemational Economic and Social Affairs,
DemographicIrdicatorsof Countries.:Estinates and Projectionsas Assessed in 1980, STIESA/SER.A/82
(New York: UnitedNations, 1982). For security force data: InternationalInstitutefor StrategicStudies,
The MilitaryBalance1981/82(London:nSS,1982);Annuairede l'Afniqueet du Moyen Orient1980: Les
Arnnes et la Dlfense (Paris: GroupeJeune Afrique, 1980); and Defence & ForeignAffairs Handbook
(London:Copley, 1980).

Variable Effects
While the effects of adjusting to decreased
military procurement would be felt by
relatively few developing countries, and in a
fairly limited numberof industrialsectors and
by a rather small proportionof the labor force
in most of those counl:ries,some workers,
communities, and industrial sectors can be
expectedto experiencedifficultyin makingthe
transition. The problems experienced in
absorbing some proportion of the manpower
currently employed by the security forces
dependson severalfactors,such as the number
released, the speed with which the
demobilizationoccurred, the skills possessed
by the released servicemen,the availabilityof
jobs, and the match between the skills of
releasedpersonneland the availablejobs. The
transferabilityof skills acquired in the course
of military service will be examined in a
subsequent section.
Unemployment and
underemployment
are problems of
considerable magnitude in many developing
countries,and the additionof what may appear

58

to be a rather small share of the labor force

may produce disproportionately larger
problems.
The effect that reduced production of
weaponswould have on industrial sectors and
local economiesdependsin part on the relative
weight of arms production, its concentration,
and the link betweenthe major contractorsand
second-tiersuppliers. The more integratedthe
firms that produce major weaponsare into the
local economy--that is, the larger the
proportion of domestically produced
components and materials--the greater the
secondaryeffects of reducingprocurement.

Two Paths: Conversion and

Diversification
Economies are constantly undergoingchange
as entire industriesrise and decline and new
products and production techniques are
introduced. Many analysts believe that:
"Defenseindustry conversionin the 20th and
21st centuriesdoes not differ in its basic form

Adjustingto Reductionsin Military Expenditureand Defense Procurement

from the other industrial transformations

witnessed in the past."7 The process by
which countries adjust to reduced arms
procurement varies according to economic
system and derives from the way in which
economiesare organizedand managed. It
corresponds to the way in which these
economies undertake other structural
adjustments.
Conversion
Defense-industry conversion has become a
controversialconcept in the OECD countries
because proponents of reduced military
expenditure and lower levels of defense
procurementhave argued, largely for political
reasons, that the defense-industrial sector
requires somewhat different treatment than
other industrial sectors undergoingstructural
change. Defense-industryconversion--under
vhich governments mandate advanced
planiningfor the alternative use of defense
production facilities, approveplans developed
at the local level for their re-use, and provide
financialsupport for the adjustmentprocess-is
seen as a tool to promote arms limitationand
arms control by undercuttingpressure from
both labor and management to maintain
stabilityin weaponsprocurement. Conversion
proponents believe that if defense producers
plan for lower militaryexpenditurein advance
of cutbacks, their support and the support of
the communitiesin which they are locatedfor
continued arms expenditurewill be reduced.
'Theyalso seek to assurethe defenseworkforce
that lower defense expenditureneed not be
equatedwith higherrates of unemploymentfor
8
defense employees.
Proponentsof defense-industryconversion
have suggestedthat centralgovernmentscreate
national commissions to oversee the
conversion process, provide financial and
other forms of assistanceto guide that process,
and mandate the creation of "alternativeuse
committees"in each defenseenterprise. The

Ball

latter would be charged with planning, in

advance, the transition from military to
civilian production--choosingnew products
and overseeingthe retoolingof the factoryand
any necessary retraining of the workforce
(includingmanagement).
Diversification
In the OECD countries, most of those
concernedwith adjustingto lower procurement
budgets envision a smaller role for central
governments--onewhich does not go beyond
the programs and policy tools used by
governments to assist economies in making
other transitions. For them, the responsibility
for diversifyinginto new products and markets
rests largely with individual enterprises.
While it is recognized that diversification
requires advance planning, particularly to
identify marketable products, proponents of
diversification (including the defense
producers themselves) strongly object to a
government role in funding or otherwise
facilitatingthe planning process.
Diversificationcan take several forms. In
its simplest and most commonly practiced
form, it involvesthe acquisitionof a company
or division already established in the civil
sector. The company or division that
producedfor the military sector may be sold,
closed or drastically reduced in size. The
plant and equipment of divisions that have
been closed are typically sold or dismantled.
Employeesmust seek work elsewhereif they
are not transferred to another position within
the company. This option is clearly not one
that is available to very small companies
withoutsufficientfinancialresourcesto enable
them to acquire other firms.
Alternatively,an enterprisemanufacturing
military equipment, sub-systems, or
components can identify new markets for
closelyrelated civilian products (such as civil
aircraft for a military aircraft producer). Or,

59

MilitaryExpenditureand EconomicDevelopment

if these do not appear to exist, new products

can be producedusing as many as possible of
the facilities, processes, equipment,and skills
that were used to manufacturedefensematerial
(such as railroad locomotives for a tank
producer). At the micro-level, there is
essentiallyno difference betweenthis type of
diversification and defense-industry conversion. Where these two strategies differ is
in the degree to which govermnentmandates
advanced planning, approves conversion
strategies, and fundsthe transition.9
Conversion versus Diversification
Defense-industryconversionis not likely to be
adoptedin any countryat the presentmoment.
In the Soviet Union, the one countryin which
conversionhas recently been widelydebated,
the concept has diametrically opposed
meanings for the participants in that debate
and has become part of the battle surrounding
the adoption of market-orientedreforms.10
Irrespectiveof which side prevails, it does not
seem likely that reductions in arms procurement will be addressed in a significantly
differentmanner from other structuralchanges
facing the Soviet economy.
In the OECD countries,diversificationwill
be the strategy of choice. The more mobile
labor and other industrial resources are
(whether by choice or by default) and the
more flexibilityeconomicmanagershave, the
more likely it is that diversificationwill take
the form of acquiring new research or
production facilities, as in the United States.
In many European countries (where labor
tends to be less mobile than in the United
States) and among the larger Japanese
corporations, diversificationat the enterprise
level will be the preferred strategy."
Exactly where on this spectrum individual
developing countries fit depends on their
general approach to structural change in the
economy.

60

Transferability of Security Force

Personnel' 2

In the 1960s,the earliest theoristsof the role

of the armed forces in the development
processpositedthat one of the most important
channels through which the security forces
could contributeto the economicprocess was
humancapitalformation--specifically,
training
of administrative and technical personnel.13
The empirical base upon which these
assertions were founded was modest at best
and in many cases non-existent. Although
more research still needs to be carried out to
determineexactlywhich skills are acquired in
the course of military service and the degree
to whichthese benefitthe civilian economy,it
is now possible to make a preliminary
evaluation of the transferability of technical
skills from the armed forces to civilianlife.
Trnsferbility

of Technical Skills

Most soldiers in the army (traditionallythe

largest service) are trained in artillery,
armored, and infantry activities, all of which
are highly military-specific. Relatively few
receive more specialized technical training.
Most of the very few availabledetailedstudies
of developing-country military manpower
pertain to South Korea. One of these found
that only about 20 per cent of the servicemen
dischargedfrom the armed forces each year at
the beginningof the 1970scouldbe considered
"technicallyskilled manpower"(Table 5).
Some analysts have looked at statistics
such as these and have concluded that the
armed forces have made a significant
contribution to the supply of technical
manpower in countries like South Korea.14
Before reaching any conclusions about the
transferability of skills, however, it is
necessaryto examinein more detailthe nature
of the skillspossessedby the 20-25 per cent of
Korean servicemendischargedeach year who

Ball

Adjusting to Reductions in Military Ecpenditure and Defense Procurement

Table 5. Technically Skilled Manpower Released from South Korean

Armed Forces, by Service, early 1970s
Total

Discharged

Total

Arny

Navy

Air Force

Marines

180,000190,000

35,567

19-20

30,500

551

2,609

1,907

Source: Labor Educationand Research Institute, EconomicDevelopmentand Military TechnicalManpowerof

Korea. A Study of ManpowerDevelopmentin the Militaryin Korea (Seoul: Korea University, 1976),
p. 321.

are classified as "technically skilled

manpower." There is no doubt that some
skills are transferable. Many civil airline
pilots, sailors in the merchantmarine, aircraft
mechanicsand radiomen, and skilled workers
in the electronics, automotive, shipbuilding,
heavy construction, and civil engineering
industriesreceivedtheir initial training in the
armed forces.15 At the same time, much of

the training providedto Korean soldiers is of

no benefitwhatsoeverto the civilianeconomy.
The Korean Army Military Occupational
Areaslisted in Table 6 are subdividedinto 144
Military Occupational Specialties (MOS) of
which 55 (or 40 per cent)have no civil-sector
counterparts, probably including a high
proportion of the jobs in the combat and
electronicsareas.'6

Table 6. Distribution of Servicemen in South Korean Army,

According to Military Occupational Areas, 1968
Combat

Electronics

Electricity

Repairing

Mechanics

45.2a

0.6

3.7

1.0

3.5

Administration

Public
Health

Surveyand
Drawing

Special

Others

11.2

4.0

0.3

2.5

15.4

Vehides
12.6

Note: It is not known what the category"Others"includes.

a Thisestimatemay be low. A 1972study suggestedthat somewhatover 50 per-centof all men servingin the army
belongto occupationalarea 1.
and MilitaryTechnicalManpowerof
Source: Labor Education and Research Institute, Economic
Development
Korea. A Study of ManpowerDevelopmentin the Militaryin Korea (Seoul: Korea University, 1976),
p. 294.

61

MilitaryExpenditureand EconomicDevelopment

Even where technical training that might

be applicable to the civilian sector has been
received, its actual use is in some doubt.17
Surveys suggest that over 90 per cent of
former South Korean enlisted men were
holdingjobs entirelyunrelated to the military
occupational specialties in which they had
received training in the early 1970s. More
strikingyet was the findingof the 1972survey
that 92.8 per cent of the South Korean
servicemen who had undergone specialty
training at branch schools were employed
outside the technical fields in which they had
been trained.'8
A survey of 11,600 members of the
Korean Army Reserve conducted in March
1975 found that some 26 per cent identifiedat
least some relationshipbetween their current
occupations and their MOS while in the
regular Army. (Of these, only 10 per cent
found a "close" relationship.) Over 20 per
cent of those who said that they were working
in the same fields in which they had been
trained by the Army were employed in
"mechanical operation"---presumablyvehicle
operation. Another 9 per cent were in
"rmanagement" and 3 per cent in
"communications."19
While it wouldbe unwiseto draw hard and
fast conclusionsbased on 20-yearold data for
one country, the South Korean case strongly
suggests that there is far from a close or
automatic relationship between training
received in the military and an individual's
subsequent career. Available information
from other countries such as Iran and China
20
supportsthis assumption.
Pre-releasevocationaltraining programs in
place in developingcountriesalso suggestthat
military training transfers poorly. South
Korea inaugurated the S;aemaul, or "New
Village," program in 1970. Under this
program, Army personnel are given a course
in farning techniquesjust prior to their release

62

from the armed forces, and men who have

served in the Navy learn fishing techniques.
While this training has been described as "a
general educationfor culture rather than for
vocational coordination with civilian
enterprises," a 1975 study found that prerelease training had been useful to some 34
per cent of nearly 10,000 former servicemen
surveyed, either in their current job or in a
previous position. A survey of nearly 2.3
million ex-servicemenconductedat the end of
the 1960sfound that nearly 60 per cent were
engaged in agriculture and just under 5 per
cent held positions as manual laborers.2 '
Limited Spin-off
In general, one must conclude that the armed
forces are not a significantsource of technical
manpowerfor developing economies. Many
of the skills taught in the course of normal
training are military/weapon-specific.In many
developing countries, the armed forces have
accessto relativelyfew sophisticatedweapons
and thus have much less of an opportunityto
receive technical training than in a country
such as SouthKorea. If all of the technically
trained personnelin these countries were able
to use their skills in the civilianeconomyonce
they left the armed forces, they would still
accountfor only an extremelysmallproportion
of the civilian workforce. Even in countries
such as South Korea and China, which have
procured large amounts of sophisticated
militaryequipment,only a small proportionof
the armed forces can be definedas technically
skilled.
Furthermore,even when skills that can be
used in the civilian economy may be learned
in the courseof militarytraining, their transfer
to the civil sector is far from automatic. Time
and money are necessaryto match the skills
acquiredby servicemenwith availablejobs in
particular firms, and further training may be
necessary, as indicatedby a description of a
mid-1970sSouth Korean plan to upgrade pre-

Adjustingto Reductionsin MilitaryExpenditureand Defense Procurement

release vocational training.22

In short,
considerably more resources need to be
applied-both by government and by private
industry-before former servicemencan make
a full contributionto the civil sector.

FactorsConditioningthe Success
of AdjustmentPrograms23
With the possible exception of the Soviet
lJnion, no country currently producing
military equipment would suffer serious
economic disruptionif its military industries
experienced a sharp decline in orders. The
U.S.S.R. must be viewed as an exceptionin
view of the extensivedislocationthat currently
characterizes all sectors of its economy.
'Macroeconomic analyses of the likely
economic effects of reductions in defense
budgets in several Western industrialized
countries demonstrate that reductions
comrpensated
by increasedgovernmentoutlays
in the civil sector, by lower taxes, and/or by
an expansionarymonetarypolicy would have
a minor effect on the economy as a whole.
Other studiessuggestthat even uncompensated
reductionswouldhave relativelylittle effect at
the national level.24 A decline in military
procurement can, however, be expected to
createdislocationswithinparticulargeographic
regions and industrial sectors, for individual
enterprisesand for some membersof the labor
force in those enterprises.

The Economy
It is easier for enterprisesand labor to adjust
to change when the economiesin which they
operate are characterizedby a high level of
economic growth and high rates of
employment. The economic situation in the
Soviet Union, Eastern Europe, and the
developing countriesappears, on the surface,
to provide substantial opportunities for
redirecting resources currently employed in
the defense-industrial sector. There are
seriousshortagesof consumergoods as well as

Ball

basic commoditiessuch as food, housing, and

health services. At the same time, there are
seriouspolitical and economicfactorsthat will
constrain adjustment efforts.25
The
developing countries may, for example,
experience severe demand constraints, both
domestic and foreign. Many of the civilsector products that the military industries in

these countries are already producing are

destined for export.26
In periods when
export markets are contracting, this strategy
will prove less successful. In some cases,
domesticmarkets can take up a portion of the
slack. In others they cannot, either because
they are too smallto supportcertainindustries
(Singapore or Israel) or because demand is
artificiallyconstrained(India).
Dependence on the Defense Sector
Dependence on military production varies
among regions, industrial branches, and
enterprises. As in industrialized countries,
arms productionin Israel, India, Brazil, South
Africa, and Taiwan--whichtogetheraccounted
for 70 per cent of all major weaponsproduced
by the developingcountries(excludingChina)
between 1950 and 198427--is concentratedin
the aircraft, shipbuilding, motor vehicle,
electronics, and small arms and ammunition
industries. Changes in defense policies and
weapon technologies can cause shifting
reliance by the defense establishment on
various industrialsectors.
The relativelypoor integrationbetweenthe
military and civil industrial sectors in many
developing countries and the resultant heavy
reliance on imported components, subsystems, and materials, meansthat the pattern
of defense dependence in the developing
countriesmay be somewhatdifferentfrom that
in the industrializedcountries. Some countries
(notablyBrazil)havetraditionallyrelied rather
heavily on external suppliers. Others (for
example, India) have sought to incorporate

63

MilitaryExpenditureand EconomicDevelopment

domestically designed and

componentsin their weapons.

produced

Within industrial branches, both those

heavily reliant on defense orders and those
minimallyinvolvedwith1the military,defenserelated productiontends to be concentratedin
a relatively few firms.2 8
Although a
relatively few companiesproduce the bulk of
the military equipment manufactured in the
developing countries, enterprise dependence
on defenseorders varies considerably. At the
beginningof the 1980s,approximatelyhalf the
output of India's nine defense public-sector
companies was sold to the civilian market.
Several companies, Bharat Earth Movers,
Garden Reach Shipbuilders& Engineers, and
Mazagon Dock sell well over half of their
output in the civilian nnarket. Some Israeli
firms that were originally set up to serve the
defense market--Israeli Aircraft Industries,
Elron, Israel Electro-Optics, Elscint, and
Elta--haveincreasinglynoved into the civilian
market. In South Africa and Brazil, both of
which have built up much of their defense
production capacity on the basis of existing
civil-sectorfirms, the mnilitary-related
sales of
arms producers have increased over time.
Some developing-countryproducers would,
therefore,have a civil-sectorbase on which to
rely if defense procurementwere to decline.
Others would find adjustment more
difficult.29
Defense Specificity of Product
or Technology
There are those who argue that military and
civilian technologieshave divergedfrom each
other so much since the end of World War II
that only minimal transfer can be anticipated
between the sectors.30
It is clear that
investingin militaryproductionis not the most
efficient way to obtain technologicaladvance
in the civil sector. The questionconfronting
enterprises that are forced to adjust to
reductionsin military orders, however, is not

64

what the optimal method of entering the

civilian market may be, but whether the
productsthey manufacturecan be modifiedfor
the civilian market and, if not, what use can
be made of their technology,processes, plant,
and equipment.
The more specializeda piece of defense
equipmentor the technologyused to produce
it, the more difficult it will be for an
enterprise to move into the civil sector. The
degree of specialization varies among
industrialsectors. The machinery and plant,
in the ammunitionindustry tends to be highly
specific, while that used by the military
electronics,vehicles, and aerospace industries
can often be used to produce for both
markets.31 The plant and machineryused by
Swedish defense producers tend to be less
specializedthan those employedby producers
in other countries, particularly the United
States. To the extent that defensefirms in the
developing countries employ more standard
machinery and less specializedtechnologies,
they will find it easier to transferproductionto
the civil sector (on this point, see also the
Appendixto this chapter, pp. 70-73).
Nonethelessthe fact remainsthat manyof
the branches of civil-sector industry most
likely to benefit from products and processes
developed in the military sector are those
which are the most technologically
sophisticated. One of the complaintsheard
from producers of military equipmentin the
SovietUnion is that they are being askedto go
from producing missiles to producing "pots
and pans." In manydevelopingcountries,the
problemis the same. The technologyneeded
to meet the most urgent requirementsof the
population is less sophisticatedthan that used
to produce military equipment. This is one
reason why many of the civilian products
manufactured by defense companies are
exported. It is therefore extremelyimportant
to reduce the defenseproductionprocess to its
basic elements--pieces of equipment,

Adjustingto Reductionsin Military Expenditureand Defense Procurement

processes, technologies--and identify

alternative uses for them.32 The products
that are ultimatelyproduced may bear no or
only a slight resemblance to the original
military end-item. It should be incumbent
upon developing-countrydefense producers
and governments, having made substantial
investments in military industries,to identify
new uses for as many of the resources
containedin these industriesas possible.
However, because so much military
production in the developingcountriesoccurs
under license, reproduction and adaptationof
weapons technology may not be legally
possible. Furthermore, reproduction and
adaptation imply mastery of the technology,
and it is by no means certain that defense
technologies have been mastered in many
developingcountries.33
Nature of the MilitaryMarket
One of the major constraints facing defense
producers seeking to enter the civil sector is
their lack of experience in producing and
marketing non-military goods.
Defense
contractors are familiar with a well-defined,
generally uncompetitive market--a market
dominatedby one customerhavinga clear idea
of the characteristicsof the desired product
and frequently willing to pay a large portion
of its developmentcosts. Price tends to be
less importantthan in the commercialworld.
Technical capability is frequently the key
element in arms production agreements.
Competition,which increasedduringthe 1980s
in at least some OECDcountries, is generally
not as extensivein the militaryas in the civil
sector.3 4 In the civilian market, there are
many customers and little product guidance.
Firms must use their own resourcesto develop
a product that may not suit the market and
may become a commercialfailure.35
In view of these considerations,it is not at
all surprising that the most successful

Ball

conversionsand diversificationsin the OECD

countries have been in the area of hightechnology products sold to government
agencies and other large entities with
reasonablywell definedrequirements,such as
airlines, postal and other communications
systems, and railways.36 This also helps to
explain why diversification through the
acquisition of new companies has been a
favored method of decreasingthe reliance of
defenseorders in the United States.
To the extent that enterprises producing
militaryequipmentin the developingcountries
share the characteristicsdescribedabove, they
will experiencesimilar difficultyin expanding
into the civilian market. Companies that
beganin the civil sector (such as India's Praga
Tools and South Korea's Hyundai) and
continue to sell to the civil sector, or that
produce weapons using componentsthat can
also be used in producing civilian equipment
(suchas Brazil's Engesaand Embraer),should
have an advantagein finding alternativeuses
for their military divisions should defense
orders decline.3 7
This does not mean,
however, that these divisions will not
experience many of the market problems
discussedelsewherein this paper.
To the extentthat public-sectorenterprises
play an important role in the economies of
these countries, and that companies moving
out of the military sector can sell their new
products to these enterprises and government
agencies, problems associated with moving
into a highly competitivecommercialmarket
will be mitigated. Recent movementtoward a
reduced role for the public sector in the
economies of developing countries could
exacerbatethe conversionproblem, at least in
the short term. Privatizationhas, for example,
been identifiedas one likely constraint on the
abilityof Brazilianpolicy makers responsible
for the defense-industrialsector to influence
overall government policies.38 At the same
time, it will limit the support defense

65

Military Expenditureand EconomicDevelopment

producers can expect from the Brazilian

governrment.
Marketability of Alternative
Products
While it is easier for oampaniesto transfer
productionfrom the militaryto the civil sector
during a period of economic growth, it is
equally important that markets exist for the
items chosen to replace military equipment.
Becausethe defensemarketgenerallydoes not
require firms to develop their commercial
marketing capabilities, one of the major
constraints experiencedby defenseproducers
in moving into the civilian market is their
inability to identify, produce, and market
goods and servicesin a commercialsetting.39
Companiesthat are serious about moving
into the civilian market nnakea special effort
to instill in their designersan understandingof
the different needs of products developedfor
the civilian market. This can best be
accomplished by absorbing former weapon
design personnel slowly into an enterprise
already producing for the commercialsector.
Successful product diversification thus may
require the addition of personnel with
commercial experience from outside the
companyor division.

Flexibility of Company
and Manpower
Even under favorable economic conditions,
individualworkers may experiencedifficulty
in transferring from miilitary to civilian
production. Problems can derive from the
defense dependencyof particularjobs (aeroastroengineers,for example),from differences
between the skills and other characteristics
(such as age and salary expectations) of
defenseworkers and those requiredby the new
jobs in the civil sector, and from worker
immobility(voluntaryor involuntary). These

66

problems are likely to arise in all armsproducingcountries.40

Most clerical and unskilled support staff
and skilled and semi-skilled production
workers experienceno trouble transferringto
41
positions in the civilian economy.
The
availability of retraining programs greatly
facilitatesthe adjustmentprocess. Engineers,
scientists,other technicalpersonnel,managers,
and administrators experience the greatest
difficulties in adjusting to civil-sector
production. Defense firms tend to employ a
higher proportion of these categories of
personnel than manufacturing industry as a
whole, althoughproportionsvary from country
to country. A 1978 U.S. study reported that
the two main reasons why administratorsand
managers experience difficulty in moving
betweensectors is the lack of familiaritywith
civil-sectormarketingpracticesand the higher
42
salariesprevalent in military industries.
For scientistsand technicalpersonnel, the
"main barriers to transfer to commercial
work" include:
(1) The lack of cost-consciousnessamong
defenseengineers;
(2) The view of commercial managers that
defense engineers are not well suited to
commercialwork; and
(3) The belief among engineers that the
defense environment requires more
specialists
and the commercial
43
environmentmore generalists.
Not only do these sorts of attitudesencourage
defense employees to lobby against cuts in
procurement. They can also cause serious
problemswhen companiesattemptto diversify
production.
Technical and managerial staff must
exhibitconsiderableflexibilityand willingness

Adfustingto Reductionsin MilitaryExpenditureand Defense Procurement

to acquire new skills if conversion or

diversification efforts are to succeed.
Managementin particular must be willing to
discover what it does not know about
producing for and selling to the civilian
imarket. And it must be prepared to spend
several years, perhaps as many as ten to
fifteen, in moving into new product areas.44
There is no reason to believethat the staff
and managementof companiesproducingfor
the military market in developing countries
will not face the same conditions. What is
lacking, however, is even the most basic
empirical evidence about the defense-sector
workforce in these countries. Very little is
known about the nature of the workforce at
the corporate level (let alone the enterprise
level). What proportionconsistsof managers,
of technical personnel, of scientists, or of
production workers? How specializedare the
skills they possess? How mobile are they?
Thus, when a company such as Engesa is
forced to cut its workforceby two-thirds, it is
impossibleto anticipatewhat the experienceof
the employeeswho havebeen made redundant
might be.
Role of Governments
The role played by governmentsin assisting
defense contractors to adjust to decreases in
orders for military goods and services varies
among countries and mirrors their role in
facilitating other kinds of structural
adjustments. In the United Statesand Western
Europe, programs are in place that provide
subsidies,tax rebates, or retrainingto faciitate
restructuring and maintain employment;
stimulate certain sectors through public
procurement;providegovernmentassistanceto
displaced workers wishing to start their own
firms; and promote the introductionof new
processes and products, among other
objectives.4 5
Local and regional
governments,which are most likely to feel the
greatest effects of reduced defense orders,

Ball

frequentlyplay a role in channelingeconomic

adjustmentassistanceto areas in need. They
can also providetraining, assist firms moving
into the civilian market, and procure new
products produced by former defense
contractors.46
Central governmentsobviouslymust play
the leading role in the conversion of
government-ownedproductionfacilities. The
U.S. government,for example,supervisedthe
closureof two biological warfare facilities in
the early 1970s, followingPresident Richard
Nixon's decision in November 1969 to cease
producing "lethal biological agents and other
methodsof biologicalwarfare."4 7
The roles played by governments in
developing countries with a domestic armsproduction capacity to help enterprises and
employees adjust to reductions in defense
procurementwill similarlymirror the policies
of these governmentsto cope with economic
change in general. It stands to reason that
governmentsshould seek to preserveas much
as possible of the very costly investment in
plant, equipment, technology, and trained
manpowerthat had been made in the militaryindustrial sector, often with government
resources, and to encouragetheir reutilization.
To the extent that developing-country
governments lack the necessary funding to
support the kinds of adjustment programs
availableat all levels in the OECD countries,
adjustmentwill be more difficultand available
resources will not be fully utilized.

Conclusion
The 1990s promise to be a decade in which
much greater attention will be given to the
economic effects of reductions in military
expenditureand arms procurementthan at any
time sincethe end of World War II. There is
a considerablebody of literaturethat describes
and analyzesprevious adjustmentexperiences
in the defense-industrialsector in the United

67

MilitaryExpenditureand EconornicDevelopment

States and WesternEurope. This is now being

supplementedby a growingnumber of studies
and articles on the currenit situation in the
Soviet Union and Eastern Europe. For the
developing world, much less information is
available. There are few detailedprofiles of
developing-countryarms producers and, with
the exceptionof China, little has been written
about previous experienceswith procurement
cutbacksor project cancellations.
The broad outlines of the situation are
nonetneless discernible. The effects of
reducing the size of military and security
forces dependon the numberdemobilized,the
speed with which the demolbilizationoccurs,
the existinglevel of unemployment,the skills
possessed by the former servicemen,and the
match between those skills and the
requirements of the civilian economy.
Available evidence suggests that a large
proportion of the training received in the
armed forces does not provide servicemen
with skills that are in demand in the civilian
economy. Where usuable skills have been
acquired, their transfer from the military
sectorto the civilian economyfrequentlydoes
not occur automatically.
As in the industrialized countries, the
macroeconomic effects of reductions in
domestic arms production will be limited in
the developing world. Problems can be
expected, however, at the local level. The
way in which countries adjust to reduced
domestic arms procurementdepends in large
part on the way in which they adjust to other
structural changes.4 g In order to devise a
strategy for adjusting to cutbacks in
procurement, it is necessary to have some
basic information about defense-industrial
sectors: the companies belonging to it, the
proportionof corporate turnovergeneratedby
defensesales, the share of the corporatelabor
force producing for the military market, the
nature of the technologyemployed, the skills
possessed by employees, Ihe geographic

68

distributionof defenseindustries, and so on.

Some of this informationis available and a
systematiceffort should be made to collect it.
Once that is done, research can be
undertaken to fill in the gaps in existing
knowledge, many of which will be quite
substantial. Given the sensitive nature of the
industries involved, it may prove to be
difficult--asit has been with defenseindustries
in many OECD countries, not to mention the
U.S.S.R. and Eastern Europe--toobtainhighquality informationon all the desired topics.
It is likely, however, that a sufficient amount
of information can be gathered on major
defensecontractorsin most of the developingcountry arms producers (China, Israel, India,
South Africa, Brazil, Taiwan, Argentina,
South Korea, and Egypt) to begin evaluating
the likely impact of reduced output in the
defense-industrialsector.49
This body of information can then be
applied to the issues raised in the preceding
section of this paper in order to evaluate the
likely implications of reductions in defense
orders for the economy, domestic arms
producers, and the defense-industrial
workforce. To evaluatethe impactof the end
of the Iran-Iraq War on the Brazilian
economy,for example, it will be importantto
know, among other factors, what firms are
affected; their dependence on the defense
sector and their ability to support a transition
to civil-sector production from corporate
resources; their geographic distribution; the
strengthof local economies;the abilityof the
domestic(private and public sector) market to
purchase any new products these firms might
manufacture;and economicadjustmentsupport
available from governments at the national,
state, and local levels.
To evaluatewhat is likely to happento the
more than 2,000 workersreleasedduring 1990
by Engesa, the Brazilian manufacturer of
military vehicles, one needs to know, for

Adjusting to Reductions in Military Expenditure and Defense Procurement

example, their geographic distribution; the

skills theypossess; the need for re-trainingfor
each category of employee; the condition of
the local economy; personal characteristics
(such as age, mobility, and willingness to
accept reduced income) that affect the speed
with which individuals are rehired; and the
availabilityof employeeadjustmentassistance,
such ;as retraining and placement programs,
movingcredits, or small business credits.
In order to evaluate what will happen to
Engesa itself, one needs to know, amongother
things, management's interest in seeking to
develop alternative products, the company's
ability to find new applications for the
technologyused to produce military vehicles
shouldthe "easy" productdiversificationsinto
varioulscivilian vehicles not prove profitable,
the existence of govermnent subsidies or
credits to facilitatethis process, and the ability
of managementto understandthe differences
between the defense market and civilian
markets (such as increased risk, relative
importanceof cost) and to adjustto them.
M[oreresearchalso needsto be doneon the
transferability of the skills acquired in the
course of military training to the civil sector.
As shown in the section on Transferabilityof
SecurityForce Personnel,some interestingand
suggestive work has been done on South
Korea. This needs to be expanded,first of all
by identifyingany other studies that may have
been completedfor other countries. If there is

Ball

rather little available information,a series of

case studiescouldbe undertaken. It wouldbe
advisableto study countriesat different levels
of developmentwith armed forces at different
technologicallevels. For each country, the
first step would be to catalog the skills
acquired by each category of military
manpower. (his paper has examined only
technicalmanpower,but it wouldbe useful to
consider the transfer of managerial and
administrativeskills as well.) The secondstep
would be to consider whether the skills
identifiedhave civil-sectorcounterparts. For
those that do, the degree to which former
servicemen possessing these skills actually
make use of their training once they leave the
armed forces would need to be evaluated. If
it turned out that the experience of South
Korea could be generalized, it would be
helpful to considerhow the transferabilityof
skills acquiredin the course of militaryservice
could be enhanced.
In view of the changed international
situationthat has been evolving since the end
of the 1980s and the amount of resources
devoted to the military sector in the
developingcountries, there is every reason to
build up a body of knowledgethat will enable
governmentsand those advisingthem to make
well-informed, realistic decisions about reallocatingresources from the military to the
civilian economy should the opportunity
a
arise.

69

Military Expenditureand EconomicDevelopment

Appendicx
Recent Adjustments in China's Defense Sector
In the second half of the 1970s, the
conjunctionof several political-strategicand
economictrends causedthe Chineseleadership
to re-evaluate China's defense requirements,
and the decision was taken to transfer
resources from the production of weapons to
civilian-sector industry. This decision was
facilitatedby the fact that much of the existing
plant, machinery, and technology in the
defense industry was unsuited for producing
the modern weaponrywith which the military
was to be equippedin the future. It was also
clear that the defenseindustrywas operatingat

minimal. Only a very small proportion of

Chinesedefense-industrialoutput consistedof
products for the civilian market in the mid1970s (Table A-1). Defense enterprises
therefore had virtually no familiaritywith the
civilian market. What is more, individual
defenseenterprisesand the ministriesto which
they belongedall sought to be self-sufficient.
This was both a legacy of the Soviet defenseindustrialmodel that had been adopted by the
Chineseand an outgrowth of China's attempt
to reduce the scope of the task confrontingthe
central planners in organizing the flow of

well below capacity.50

goods and services between sectors and

Reductions in the

level of procurement would only exacerbate

that problem.

geographicalregions. This autarkic behavior

was intensified during the Cultural
Revolution.52

The process of transferingresources from

the military-industrial sector to civilian
industry proceeded slowly until the Third
PlenarySessionof the TwelfthParty Congress
in October 1984.51 During the 1980s,it was
also decidedto reduce the size of the People's
LiberationArmy (PLA)by nearly 40 per cent.
While enough information has become
available to describe in a general way the
conversion process, there is virtually no
informationavailableabout the experienceof
individualenterprises, the processesby which
new products are chosen, the problems faced
in retoolingproductionlines and retrainingthe
labor force, the reseitlement and reemployment experiences of demobilized
soldiers,and so on. What followsthereforeis
just an outline of the changes that have
occurred in China over the last decade.

Industrial Adjustment
Prior to the reforms inauguratedat the end of
the 1970s, the links between the militaryindustrialsector and the civilianeconomywere

70

One of the major problems confronting

Chinesepolicy makers in the early 1980swas
the unwillingness of defense managers to
believe that there really was a need to begin
producing for the civilian market. As in all
other arms-producingcountries, the level of
procurement in China had expanded and
contracted over the preceding three decades.
It took several years for defense producersto
understandthat procurementlevels would not
rebound this time, even though they received
fairly frequent and explicit messagesfrom the
governmentto this effect.53
Once the messagewas received,however,
defenseproducersfaced serious constraintson
their abilityto adapt, which derived to a large
extent from the preferentialtreatmentthey had
previously been accorded in terms of the
availabilityof financial, human, and material
resources.5 4
The economy-wide reforms
introduced in 1984-which were designed to
increase enterprise profitability, the efficient
use of resources, integrationof the economy,

Adjustingto Reductionsin MilitaryExpenditureand Defense Procurement

Ball

Table A-1
Proportionof China'sDefense-IndustrialOutputDestinedfor the CivilianMarket
(percentages)
Year

Civilian Output

1975
1976
1977
1978
1979

6.9
n.a.
n.a.
10
10

1980
1981
1982
1983
1984

16-18
n.a.
20
20
30

1985
1986
1987
1988

45-50
50-60
n.a.
60

1989

66

n.a. = Data not available

Source: Arthur J. Alexander, "National Experiences in the Field of Conversion: A Comparative Analysis," Paper
prepared for the United Nations Conference on Conversion, "Economic Adjustments in an Era of Arms
Reductions," (Moscow, August 13-17, 1990), p. 17.

and an awarenessof consumerrequirements-greatly facilitated the entry of defense

producers into the civilian market. This
process was further assistedby the significant
c:hangein government policy in 1984, which
enabled technologyto be sold at a profit, and
by the implementation of other incentives
(lesignedto move technologyfrom the military
to the civilian sector.
The creation of corporations from
production enterprises and research institutes
in the military sector was important in
55
integrating production and R&D activities.
It also reducedthe power of the bureaucratsin

the ministries that had previously controlled

these functions,since much of the ministries'
responsibilitiesand authoritywere now vested
in the new entities. These corporationscould
exporttheir goods--bothmilitary and civilian-and retain their hard-currency earnings.56
This both provided the corporations with
much-neededforeignexchangeto enablethem
to upgrade their production technology and
equipment and made foreign investmentin
such entities more attractive. By 1987, over
1,000 new enterpriseshad been createdin the
coastal regions of China and the Special
Economic Zones, using the resources of
military producers that had previously been

71

Military Ependiture and EconomicDevelopment

located deep in the Chinese interior, the socalled "Third Front.""

It has been
suggestedthat a not-inconsiderableproportion
of the funds allocatedto conversionin recent
years has been used to facilitate this
relocation.
The proportionof defense-industryoutput
sold in the civilian market has increased
substantiallyover the last decade. By 1989,
some 16 per cent of all defenseproducershad
converted completely to civilian production.
Another 74 per cent were engaged in both
military and civilianproduction,while only 10
per cent produced wholly for the military
market.
The products sold by these
5
Yet
enterprises are extremely diverse."
problemsremain.
Enterprises--both civilian and militaryhave had to continue to operate in a semiplanned system. While efforts have been
made to encourage the freer movement of
some resources, others, notably labor, remain
relativelyimmobile. This means,amongother
things, that the government will not close
inefficientproducersor withdrawall subsidies
from enterprises, losing some of the more
powerful sanctionsat its disposal to alter the
behavior of defense producers. Enterprises
have also been forced to employ as much of
their existingtechnology,plant, and equipment
as possible when it might have been more
economical to build a new plant or acquire
new technologyor equipment.
Other problems include low capacity
utilization, an outmoded technologicalbase,
failure to employ fully the imported
production machinery available, a relatively
high level of unprofitable production,
unmarketableoutput, high-costproducts, poor
marketing skills, shortage of skilled
manpower, and a tendency to reproduce
autarkic patterns of behavior in the civil
sector.59

72

What is more, the data for defenseindustry conversion in China in the 1980s
must be reconciled with the substantial
increase in Chinese weapon exports that
occurred during the decade. To a large
degree, Chinese defense industry took
advantageof the ready market createdby the
Iran-Iraq war (which ended in August 1988).
If domesticChinesedefenseprocurementonly
declined by 10 per cent between 1978 and
1988,60this surge in the export of weapons
certainly would have utilized the spare
capacitycreated,givingrise to questionsabout
the number of weapon-productionlines that
were available for conversion.
Some Western analysts suspect that very
little actualconversionhas occurred. Rather,
now that export orders have dropped, Chinese
defense producers may be maintaining or
mothballingmilitary production lines, just as
Soviet defense producers have done.
According to Chinese sources, very little
mothballinghas occurred. Defense producers
definitelyhave been urged by the government
to re-employ as many of their resources as
possible. Official pronouncements indicate
that the diversificationof defense enterprises
into new productlines has producedsignificant
savings compared to the cost of building up
producationcapabilityfrom scratch.6 1
The deputy minister in charge of
conversion, Huai Guomo, has said privately
that his staff has carried out an evaluationof
the conversion potential of each kind of
machineryin plants in which it was no longer
neededto meet militarycontractrequirements.
Accordingto Huai, only a small percentage
have been mothballed-presumablythose which
were the most defense-specific.
The
remainderreportedly have been converted to
the productionof civilian goods. Given the
relatively unsophisticated nature of the
machineryemployed by the Chinese defense
industry, it is plausible that a high proportion
is general-purposemachinery. And given the

Adjusitingto Reductionsin MilitairyExpenditureand Defense Procurement

unwillingnessof the Chinese government to

sanctionplant closures,the pressureto convert
is great.
Military Manpower
In the 1980s, the People's Liberation Army
(PLA) reduced its manpower by some 1.75
million. During this period, the PLA released
approximatelyone millionconscriptseach year
following the completion of their period of
military service. Many former servicemen
were clearly ill-equipped for civilian life,
having received little more than infantry
training and political education. As a result,
"few enterprises would accept ex-soldiers
because they didn't have the necessary
62
productionand managementknowledge."
At some point in the early 1980s,perhaps
around 1983, the PLA began to provide
servicemen with marketable skills. In early
1985, some two million PLA soldiers were
reportedly undergoingtraining in agronomy,
aquaculture,catering, carpentry, architectural
design, plumbing,bricklaying,machineryand
electric appliance repair,
economic
management, mechanics, transport, food
processing, and other civilianoccupations. In
an unspecified number of PLA units, six
months of their three-year period of service
was reportedly devoted to "learningscientific
and general knowledge and civilian job
skills."6 3
These demobilizationsoccurred at a time
when unemploymentwas high. Youths who
had been sent to the countryside during the
Cultural Revolution were being allowed to
return to the cities, and the economicreforms
of 1983/1984 were reducing the amount of
underemploymentin production enterprises.
Troops were demobilizedto their hometowns
or the towns in which they had enlisted in an
attemptto spread the adjustmentburden. The
responsibility for finding jobs for these
servicemenhad been transferredfrom the PLA

Ball

to the Ministry of Civil Affairs in 1979.

Local authorities signed contracts with
enterprises, administrativeunits, and other
entitiesthat agreed to absorb a given number
of returning servicemen each year. Other
forms of assistance, such as subsidized
buildingmaterials for houses, have also been
made available. Former soldiers have also
been offered incentives (higher salaries,
housing) to relocate in remote areas and
contributeto economicconstructionthere.64
Western analysts have suggested that a
large portion of the demobilization was
achieved by "transferring various noncombatant units of the PLA to civilian
control,' such as the Railway Corps and the
Capital Construction Corps. Security units
became part of the newly created People's
Armed Police Force. Well over 500,000 men
could have been transferred to the civilian
sector in this way.65 During 1986, some 1.2
million soldiers and 21,000 officers were
reportedly resettled. Of these, 250,000 were
given cadre posts in village, township, and
country administration. Some 260,000 found
employmentin rural factories; 110,000 were
placed as laborers (farm and non-farm).
Such statistics tell us nothing about the
success these former servicemen meet in
carrying out these jobs or about their
productivity. Given the high rates of
unemployment in China during the 1980s,
some of the returning soldiers may have
displacedciviliansseekingemployment,and it
would be interesting to know if former
servicemenwere givenpriority for the avilable
jobs. In view of previous Chinese policy
regarding employmentin remote areas, it is
legitimateto wonderhow much choiceformer
servicemenwho acceptedemploymentin these
areasactuallyhad. These are but some of the
questionsit is necessaryto answer in order to
evaluatethe successof the Chinesein assisting
former servicemento adjustto civilianlife.

73

Military Expenditure and Economic Development

REFERENCES
Notes

Threatens
Recession,"
November 4, 1990, p. Hi.

For example, between 1963 and 1981, direct and

indirect employment in the British defense industry
fell by approximately 30 per cent. In 1950, there
were 21 aircraft producers in the United Kingdom.
By 1980, there were only three. See Nicole Ball,
Converting Military F&rilities: Shared Responsibilities and the Needfor Planning, WEP241/WP.1
(Geneva: International Labour Office, October 1985).
2

These twelve are Barngladesh, Burma, Chile,

Colombia, the Dominican Republic, Gabon,
Madagascar, Mexico, Pakcistan, Peru, Senegal, and
Sri Lanka. See Michael Brzoska and Thomas
Ohlson, eds., Arms Prodrction in the Third World
(London and Philadelphia: Taylor & Francis for
SIPRI, 1986), p. 15 and Appendices 1 and 2.

Ibid.,p. 23.

Ibid., pp. 21, 42-43, 60, 86-87, 116, 134-135, 138139, 168, 176.

Stuart Slade, "Chinese Armoured Vehicles: You Get

What You Pay For," International Defense Review,
January 1990, p. 167.

For example, it has been suggested that, by 1987,

employment in the Brazilian defense industry had
reached 150,000, rather than the 75,000 reported in
Table 3 for the early 198Cs. This still accounts for
less than 1 per cent of the Brazilian industrial labor
force. The 1987 figure is reported in Raul de
Gouvea Neto, How Brazil Competes in the Global
Defense Industry, ACDIS Occasional Paper (UrbanaChampaign, Ill.: Program in Arms Control,
Disarmament, and International Security, University
of Illinois, October 1990), p. 4.
The share of arms-industry employment in total
manufacturing employment, has not been calculated
because of data constraints. These ratios would, of
course, be higher than those presented in Table 3.

7 Arthur J. Alexander, "National Experiences in the

Field of Conversion: A Comparative Analysis,"
Paper prepared for the United Nations Conference on
Conversion, "Economic Adjustments in an Era of
Arms Reductions," Moscow, August 13-17, 1990,
p. 1. In the United States, the proportion of the
labor force employed in the manufacturing sector
dropped from 22.3 per cent in 1980 to 17.6 per cent
in 1990. Steven Pearlstein, "Slump in Services

Washington

Post,

See, for example, In Pursuit of Disarmament:

Conversion from Military to Civil Production in
Sweden. Report by the Special Expert Inga 7horsson.
Volume IB: Swnary, Appraisals, Reconmendations
(Stockholm: Liber, 1984), pp. 50-51. The purpose
of this report, written under the guidance of forner
Swedish Under-Secretary of State for Disarmament
Inga Thorsson, was to counter the arguments set
forward by Swedish defense producers that a
domestic defense-industrial capacity must be
maintained for military and economic reasons.
Conversion advocates include Suzanne Gordon
and Dave McFadden, eds., Economic Conversion:
Revitalizing America's Economy (Carnbridge, Mass.:
Ballinger, 1984); and Robert W. DeGrasse, Military
Expansion, Economic Decline (New York: Council
on Economic Priorities, 1983).
9

See John E. Lynch, ed., Economic Adjustment and

Conversion of Defense Industries (Boulder, Colo.
and London: Westview Press, 1987); President's
Economic Adjustment Committee and Office of
Economic Adjustment, Economk Adjustment!
Conversion, Washington, D.C.: U.S. Department of
Defense, July 1985; and "Testimony of Dr. Gordon
Adams before the Senate Armed Services Committee
Subcommittee on Defense Industry and Technology
on Economic Adjustment to Lower Defense
Spending," Washington, D.C.: Defense Budget
Project, May 4, 1990.
One issue not addressed here due to space
limitations is that of converting military bases.
Considerable information on the U.S. experience is
available. See, for example, Chapter 4 in Lynch,
Economic Adjustment, op. cit., and Office of
Economic Adjustment, Civilian Reuse of Former
Military Bases: Swnmary of Completed Military
Base Economic Adjustment Projects, 1961-1990,
Washington, D.C.: U.S. Department of Defense,
April-June 1990.

10 Thus far, very little production has been removed

from the military sector.
Rather, some
diversification has occurred, and the resources and
production responsibilities of the (civilian) Ministry
of Machine Building for Light and Food Industries
and Home Appliances were transferred to the
Ministry of the Defense Industry in early 1988. See
Judyth L. Twigg, "Balancing the Soviet Military and
Civilian Economies," in Eric Arnett, ed., Science

and International Security: Responding to a

74

74

Adjusting to Reductions in Military Expenditure and Defense Procurement

Ball

representative,whether of the firm as a whole, the

shipbuilding industry or former servicemen in

ChangingWorld(Washington,D.C.: AAASProgamm
on Scienceand InternationalSecurity,1990).

general. See Back, "The Role," op. cit., p. 304.

1 See the discussion in Alexander, "National

Experiences,"op. cit., pp. 1-12.
discussion in this section draws on Chapter 8,
"MilitaryManpowerand HumanCapitalFormation,"
in Nicole Ball, Securityand Economy in the Third
World (Princeton,N.J.: PrincetonUniversityPress,
1988), pp. 295-334.

20 Ann TibbiltsSchulz,BuyingSecurity:Iran Underthe

Monarchy(Boulder, Colo: WestviewPress, 1989),

pp. 13840; and Ball, Securilyand Economy, op.
cit., pp. 315-16, 319-21. The Chinese case is
examinedin the Appendixto this chapter,pp. 70-73.

12! The

See, for example, Lucien W. Pye, "Armies in the

Process of Political Modernization," in John
J. Johnson, ed., The Role of the Military in Underdeveloped Countries (Princeton, N.J.: Princeton
UniversityPress, 1962); and Morris Janowitz, The
Militaryin the Poliical Developmentof New Nations
(Chicagoand London:PhoenixBooks, 1964).

13

14

Changsoo Lee, "Civil-MilitaryRelations and the

Emergence of Civiltary Bureaucrats in Korea,"
p. 91, in ChangsooLee, ed., Modernizationof Korea
and the Impactof the West (Los Angeles: East Asian
Studies Center, University of Southern California,
1981), p. 91.

21

Labor Educationand Research Institute, Economic

Development, op. cit., pp. 322, 329-30.
Interestingly, some 300,000 ex-servicemen were
unemployedin 1967, accountingfor just over half
the 590,000 registeredunemployed.

22

Ibid., pp. 330-31.

23 The discussion in this section draws on Ball,

ConvertingMilitary Facilities, op.cit., and Nicole

Bail, "Disarmamentand Industries," in J. Fontanel
and J.-P. Gilhaudis,eds., Le Desarmementpour le
Developpement:
La Courseaux Annementset le Desarmement(Grenoble: Centre d'Etudes de Ddfenses
et de SecuriteInternationale,1986), pp. 141-58.
24 On the UnitedStates,see David Gold, The Impactof

5 JohnP. Lovell, "The Militaryand Politicsin Postwar

Korea," in Edward ReynoldsWright, ed., Korean

Politics in Transition (Seattle and London:
Universityof WashingtonPress, 1975),p. 190; and
Lee, "Civil-MilitaryRelations,"op. cit.
6Astudy using data from 1973 reportedthat

60 per
cent of the MOS have no civil-sectorcounterparts.
See Jong-chunBack, "The Role of the Republicof
Korean ArmedForces in NationalDevelopment:Past
and Future," The Journal of East Asian Affairs,
Vol. 3, No. 2 (Fall/Winter1983),p. 302.
17

'Labor Education and Research Institute, Economic

Developmentand Military TechnicalManpowerof
Korea: A Study of ManpowerDevelopmentin the
Militaryin Korea (Seoul:Korea University, 1976),
pp. 328-29.

IsIbid., pp. 327-28 and p. 73.

19 Ibid., pp. 368-72. An August1981 surveyof 360 of

the 12,230 men employed as engineers,craftsmen,

and office workersby the HyundaiShip Construction
Companyfound that "more than 39 percent of the
respondentslearnedtheir skillsand knowledgein the
service, which they are now exploiting in civilian
occupations." Given the extremely small sample
size, it is not clear if these findings are

Defense Spending on Investment, Productivityand

Economic Growth (Washington, D.C.: Defense
BudgetProject, February 1990). On Norway,West
Germany, the United Kingdom and Sweden, see
Ball, ConvertingMilitaryFacilities,op. cit.
25 The problems facing the Soviet economy are so

extensive that no attempt will be made here to

summarizethem. Two recent publicationsdetailing
many of these problems are: Edward A. Hewett,
Reforming the Soviet Economy: Equality versus
Efficiency (Washington, D.C.: The Brookings
Institution,1988),and AndersAslund, Gorbachev's
Strugglefor EconomicReform: The Soviet Reform
Process,1985-88 (Ithaca,N.Y.: CornellUniversity
Press, 1989).
An extensiveliterature,in both Soviet and foreign
sources, examinesthe specificproblem of defenseindustry conversion.
Many of the Soviet
contributionshave beenpublishedin Englishin New
runes and in translation through the Foreign
Broadcast Information Service. Some recent
publicationson conversionin the Soviet Union are:
Arthur J. Alexander, The Conversion of Soviet
DefenseIndustry,P-7620(SantaMonica,Calif.: The
RAND Corporation,January 1990);ChristopherM.
Davis, "The High-Priority Military Sector in a
ShortageEconomy,"in HenryS. Rowenand Charles
Wolf, Jr., eds., The ImpoverishedSuperpower(San

75

Military Expenditure and Economic Development

Francisco,CA: ICS Press, 1990);JulianM. Cooper,

The Scales of Outpnt of Civilian Products by
Enterprises of the Soviet Defence Industry, SITS
No. 3 (Birmingham,U.K.: Centre for Russian and
East EuropeanStudies, Universityof Birmingham,
August 1988); Julian Cooper, The Soviet Defence
Industryand Conversion: TheRegionalDinension,
WEP 2-411WP.10(Geneva: International Labour
Office, December 1988); and G.K. Khromov,
Problems of Conversionfrom Military to Civilian
ProdwctionFollowingth? Soviet-lnmtedStatesTreaty
on the Elinination of Intermediate-Rangeand
Shorter-RangeNuclear Force: Examples of the
Votkinsk Plant, WE]P 2-41/WP.16 (Geneva:
IntemationalLabourOffice, March 1990).
26

Someinformationon civil sectorproduction,exports,

and foreign-exchangeearnings for India's defense
public-sectorcompaniescan be found in the annual
publication, Government of India, Ministry of
Defence,Report, New Delhi: various years. For
Israel, see Gerald Steinberg, "Israel," p. 297, in
Nicole Ball and Milton Leitenberg, eds., The
Structureof the Defense Industry: An International
Survey (London: Croom-Helm,1983). On Brazil,
see Peter Lock, "Brazil: Arms for Export," p. 86,
in Arms Production,Brzoskaand Ohlson, ed., op.
cit., and de Gouvea Neto, How Brazil, op. cit.,
p. 18.
Brzoskaand Ohlson,ed., Arms Production,op. cit.,
p. 10. See also Herbert Wulf, "Developing
Countries,"pp. 310-43,in Ball and Leitenberg,eds.,
The Structure,op. cit.

28

On WestemEurope, see, for example,Ian Anthony,

Agnes CouradesAllebeck, and Herbert Wulf, West
EuropeanProduction.StructuralChangesin the New
Political Environment (Stockholm: Stockholm
InternationalPeace ResearchInstitute, 1990).

29

Governmentof India, Ministry of Defence,Report,

1980-81 (Delhi: 1981), pp. 47-48; Herbert Wulf,
"India: The UnfulfilledQuest for Self-Sufficiency,"
pp. 132-33, in Brzoska and Ohison, eds., Arms
Production, op. cit.; ThornasW. Graham, 'India,"
p. 159, in James EverettKatz, ed., Anrs Production
in Developing Countries:An Analysis of Decision
Making (Lexington,Mass.: LexingtonBooks,1984);
Lock, "Brazil," op. cit.; and Raul de GouveaNeto,
"From Tanks to Butter: The Conversion of the
Brazilian Defense Industry," Anderson School of
Management, University of New Mexico,
Albuquerque,November1990, preliminarydraft.

76

27

30 The relationshipbetweenmilitary and civilianR&D

is complex. Basic R&D can have both civilianand

militaryapplications,but this categoryoften does not
receivea large share of military R&D budgets. In
1988, "TechnologyBase' R&D accountedfor only
10 per cent of U.S. militaryR&D. Certain civilian
sectors, such as aerospace and computers, have
benefited from early support from the military as
purchaser, promoter of process engineering, and
developer of applicationstechnology. At the same
time, 'In spite of the reasonablygood effectiveness
of technologyinterdiffusionof DOD civil R&D in
the few industrial sectors in which there are
sufficiently similar product and process
requirements,there is no reason to believethat spinoff from defenceR&D can be as effectivein meeting
civil needs as R&D directed specifically at civil
sector requirements."
Alexander H. Flax,
"Interdiffusionof Militaryand CivilianTechnology
in the United States of America," pp. 117-37, in
Philip Gummett and Judith Reppy, eds., The
Relations betweenDefence and Civil Technologies
(Boston:KluwerAcademicPublishers, 1988). The
citationis from page 135.
31

Bernard Udis, Adjustment of High Technology

Organizationsto Reduced Military Spending: The
WesternEuropeanExperience, Main Report, NSFRA-X-75-016(Washington,D.C.P: NationalScience
Foundation, 1974), pp. 484, 486. On the Soviet
Union, see JulianCooper, "The CivilianProduction
of the Soviet Defence Industry, pp. 31-50, in R.
Amannand J. Cooper,eds., TechnicalProgressand
Soviet Economic Development (Oxford: Basil
Blackwell,1986),pp. 31-50;and SeymourMelman,
Barriers to Conversionfrom Military to Civilian
Industry in Market, Planned and Developing
Economies,prepared for the United Nations Centre
for Disarmament,Ad Hoc Group of Governmental
Experts on the Relationshipbetween Disarmament
and Development,April 1980. Mimeographed.

32

See, for example, President'sEconomicAdjustment

Committee and Office of Economic Adjustment,
EconomicAdjustment,op. cit., p. 48.

33 On the reliance of Egyptianand Argentiniandefense

producers on foreign personnel, see Brzoska and

Ohlson,Arms Production,op. cit., p. 284.
34 Michael Brzoska, "The Federal Republic of

Germany,"p. 117, in Ball and Leitenberg,eds., The

Structure,
op. cit; and personalcommunicationfrom
MichaelBrzoska, November16, 1990.

Adjusting to Reductions in Military Expenditure and Defense Procurement

35 Udis, Adjustment of High Technology, op. cit.,

44 See Philip Gummett and Michael Gibbons,

lp. 326, for example, found that a major theme

amongdefenseproducers "was that most of the civil
sectorsare unableto describetheir research needs."
36

"Redeploymentand Diversification at Harwell,"

Manchester: Department of Liberal Studies in
Science, Universityof Manchester,mimeographed,
June 1977, pp. 7-8; and Udis, Adjustmentof High
Technology,op. cit., p. 416, 500-01. A description
of a ten-yearprogramat the Germanaerospaceproducer Dornier is found in President's Economic
Adjustment Committee and Office of Economic
Adjusttnent,EconomicAdjustment,op. cit., pp. 3334. Similarly,SMA, an Italian radar producer, is
following a nearly fifteen-year diversification
programto cut its dependenceon the defense sector
from nearly 100 per cent in 1980 to 50 per cent in
1994.

For a descriptionof the problemsBritishelectronics

firms identifiedin enteringthe civilian market, see
Sir Ieuan Maddock, Civil Exploitationof Defence
Technology,Report to the ElectronicsECD (London:
National Economic Development Office, 1983),
especiallypp. 4-5.

37 Despite their civilian origins, Engesa and Embraer

have followeddifferentstrategiesconcerningcivilian
applicationsof their militarytechnology. By the late
1980s, a substantialportion of Embraer's turnover
was generatedby sales of civil aircraft. In 1989, for
example, Embraer reportedly exported 187 civil
aircraft (Piper modelsand Brasilias)and 52 military
aircraft. In contrast, Engesa has concentratedvery
heavily on the militarymarket. De Gouvea, "From
Tanks to Butter," op. cit.
38
39

45 Lawrence Franko,EaropeanIndustrialPolicy: Past,

Present,and Fture (Brussels: ConferenceBoard in

Europe, 1980);DouglasYuilland Kevin Allen,eds.,
European Regional Incentives, 1984 (Glasgow:
EuropeanRegionalMonitoringUnit, Centre for the
Study of Public Policy, University of Strathclyde,
1984); and President's Economic Adjustment

Ibid., p. 27.

Committee and the Office of Economic Adjustment,

See, for example,the Swedishstudy Civilproduk4on

i f6rsvarsindustrin:
Betankande
frdn
forsvarsindustrikommittin
[Civilianproductionin the
defense industry: Report from the DefenseIndustry
Committee],Ds I 1982:1 (Stockholm:Ministry of
Industry, 1982), p. 210; Melman, Barriers to
Conversion, op. cit., pp. D-14, D-23; and Udis,
Adjusnent of High Technology,op. cit., pp. 311-63.

40 Research carried out for the InternationalLabour

Office duringthe secondhalf of the 1980, primarily

on the industrializedcountriesof East and West, is
summarizedin Peter I. Richards,'Specific Issuesof
Conversion: Manpower," Paper prepared for the
United Nations Conference on Conversion,
"Economic Adjustments in an Era of Arms
Reductions,"Moscow, August13-17, 1990.
41

Udis, Adjustment of High Technology, op. cit.,

pp. 479-91 passim, 505, 513; and "Military Job
Skills: WillThey Transfer,"SLECPBriefing,No. 8,
St. Louis Economic Conversion Project,
August 1981, p. 5.

42 Ibid.
43 U.S. Arms Control and Disarmament Agency,

Ball

EconomnicAdjustment, op. cit., pp. 67-77 and

AppendicesQ and R.
46

On options available to local and regional

governmentsin the United States, see, for example,
Jean Prial Gordus, Paul Jarley, and Louis A.
Ferman, Plant Closingsand EconomicDislocation
(Kalamazoo, Mich.: W.E. Upjohn Institute for
EmploymentResearch,1981); "Deindustrialization:
Restructuring the Economy" [Special Issue], The
Annal5of the American Academyof Political and
Social Science, No. 475 (September 1984); and
Miles Friedman and Deborah Culbertson, "StateLocalEconomicDevelopmentPrograms,"in Lynch,
ed., EconomicAdjustment, op. cit. pp. 175-90.

47 Milton Leitenberg, "Research and Developmentin

(C)BW," Studies on Military R&D and Weapons

Development(Stockholm: Ministry for Foreign
Affairs, 1984). Mimeographed.
4s The developmentliterature containscase studies of

structuraladjustmentin developingcountries. See,

for example, World Bank, Korea: Managing the
IndxstrialTransiion-VohgneI: SelectedTopics and
Case Studies, Washington, D.C., 1987.

The

Economic Impact of Reductions in Defense Spending

(Washington, D.C.: Economics Division, Bureau of
Economic Affairs, ACDA, 1972), p. 24. See also
Maddock, Civil Exploitation, op. cit., p. 4.

49 Information on North

Korea will be extremely

difficult to obtain.

77

Military Expenditureand EconomicDevelopment

Appendix Notes

56 Alexander, 'National Experiences," op. cit., p. 29.

50 In the early 1980s, labor productivity in Chinese

defense industries was 38 per cent lower than in

civilian industries. Metal-cutting machinery in the
defense sector was half as productive as similar
machinery in the civilian sector. Profits on every
100 yuan of output generated by the defenseindustrial sector were sorne 50 per cent lower than
those found in civilian enterprises.
Richard J.
The
Latham, "People's Republic of China:
Restructuring of Defense-Industrial Policies," in
James Everett Katz, ed., Arts Production in
Developing Countries (Lexington, Mass.: Lexington
Books, 1984), pp. 103-22 .
5' Richard J. Latham, "The People's Republic of China:
Profits, Consumerism, and Arms Sales," in James
Everett Katz, ed., The Implications of Third World
Military Industrialization. Sowing the Serpents'
Teeth (Lexington, Mass.: Lexington Books, 1986);
pp. 187-88; and Arthur J. Alexander, "National
E3xperiences," op. cit., pp. 13-16.
52

C.Z. Lin, Employment Implications of Defence

Cutbacks in China, WEP2-41IWP.14 (Geneva:
IntemnationalLabour Office, November 1989, pp. 1819); and Alexander, "National Experiences," op. cit.,
p. 17.

53

Alexander, "National Experiences," op. cit., p. 20.

Norinco, for example, experienced a 60 per cent
drop in Chinese defense orders in 1986. Bai Si
Yeng, "Understanding the Chinese Defence
Industry," Miiay Technology, March 1987, pp. 3653.

54 Alexander, "National Experiences," op. cit., pp. 2122.

Ibid.,

pp.
19-27, and Latham,
Consumerism," op. cit., pp. 192-94.

58 Jin Zhude and Chai Muliang, "China's Experience:

A Case Study," Paper prepared for the United
Nations Conference on Conversion, "Economic
Adjustments in an Era of Arms Reductions,"
Moscow, August 13-17, 1990, pp. 7-10, and
Alexander, "National Experiences," op. cit., p. 21.
59 Ibid., pp. 26-29. In 1989, some one-third of the
aviation and space enterprises were reportedly
producing unmarketable products. Another third
were unable to produce products that were
competitive on the civilian market.
60

Ibid., p. 13.

61

Jin and Chai, "China's Experience," op. cit., pp. 78.

62

Wang Gangyi, "Science Studies Make Better

Soldiers," China Daily, March 1, 1985. See also
Zheng She, "Two-way Training in China's Army,"
China Reconstructs, November 1983, pp. 28-29, and
"Soldiers Urged to Learn Skills," Beijing Review,
August 20, 1984, p. 9 . Numerous articles appeared
in the English-language China Daily during the
1980s discussing the demobilization of one milion
soldiers and their reintegration into civilian society.

S3

Wu Dongfeng, "PLA Teaches Civilian Job Skills to

Servicemen," Xinhua Domestic Service, February
14, 1985, 8:30 GMT.

64 Lin, Employment Implications, op. cit., pp. 8-14.

6m Ibid., pp. 11-12.

55 The seven major corporations in existence by early
1987 are described in Bai Si Yeng, "Understanding,"
op. cit.

78

'Profits,

MILITARY TRADE, AID, AND

DEVELOPING-COUNTRY DEBT

Michael Brzoska

Introduction
Analysesof military and economicphenomena
in Third Worldi countries have long been
kept rather strictly separate. The military
sector often has been viewed as exogenousto
the economyand economicfactors simply as
constraintson military efforts. This separation
has its value, since both sectors must
ultimatelybe judged on their own merits. It
also has a disadvantage, however, in that
public discussions and political decisions
shouldbe based on all the relevantinformation
about costs and benefits.
Stimulatedby the expansion of military
sectors in the 1970sand the counter-intuitive
findingsof the late Emile Benoit, a substantial
numberof studies on the relationshipbetween
military expenditures and the economies of
developing countries were published in the
1980s.2

Note: The author wouldlike to thankparticipantsinthe

Wortd Bauk Symposiumflietor thalnkparticipantsin the
Marc Vogel skilfiullyassistedin the preparationof debt
data.

Less work has been done to try to

establishthe effectsof componentsof military
efforts, including the importation of
armaments.3
This may seem somewhat
surprising, since much attention is given both
by the public and by decision makers to the
internationaltrade in arms. But the interest
focuses on the political issues connectedwith
the arms trade. With respect to economic
effects,there seemsto be a strongpresumption
that they are overwhelmingly negative
whenever they are noticed--for example,
Benoit (1978) made it clear that he expected
negative effects from capital-intensivearms
imports (although he expected positive
economiceffectsfrom large militaryforces).
One reason why fairly little work has been
done on the economicsof the transfer of arms
is that no data series lend themselvesdirectly
to economic analysis. A main point in this
report is that availablearms transfer statistics
are ill-suitedfor economicimpactanalysis-not
only becauseof their inherentqualityproblems
but also becauseof their neglectof the timing
aspect.4 Arms transfered are often not paid
for or paid for only later. Another reason is
that it is not self-evidenthow the importation
of arms could affect Third World economies.
One importantway in which arms transfers
could exert influence on Third World

79

Military Expenditure and Economic Development

economiesis by way of their financing. Arms

importscompetewith other importsin the use
of foreign exchange. They are also an
obviouspossiblesource of the indebtednessof
Third World countries.
The aim of this paper is a) to describe
some of the difficulties surrounding the
economic analysis of arms transfers and
related phenomena,and b) to proposeand test
some ideasaboutthe economicaspectsof arms
imports. Emphasisis placed on consideration
of the financingof arms by importersbecause,
it is proposed here, this is a crucial matter in
the evaluationof the effectsof arms transfers.
The paper starts with a sectionhighlighting
the problems of arms transfer data and then
provides an overview of magnitudes and
trends in the internationaltrade in weapons.
It then addressesthe issue of aid and discusses
the various ways of financingweaponimports
by Third World states. Special attention is
given to the question of indebtedness in
financing weapon imports.
Available
estimates of the military-related debt are
reviewed and examined. The paper closes
with some thoughts about the future of the
internationalarms trade and its financing.
The Data Problem
Foreign Trade Dalta
Arms transfers, like all transfers of goods,
should be registered in trade statistics. In
internationaltrade classificationschemes,there
are special categoriesfor some weapons,such
as guns, tanks, and warships, while other
relevant goods, for example most types of
aircraft and electronics, are lumped together
with civilian goods.
It is of course
possible--and done in some countries--to
separate civilian and military goods in subcategoriesto such categories.

80

Surveys have shown that the large

majority of countries in the world does not
publish complete and detailed data on arms
imports (or arms exports, for that matter)
(SIPRI 1971; Arbeitsgruppe1980). In some
cases, arms transfers are specifically trade
statistics. (his creates an accountingproblem
for balance-of-paymentsstatistics that will be
mentionedlater). In other cases, transactions
connectedto armamentsare not reportedin the
appropriate categories but in one or a few
figures at the end of the trade statistics. In
many cases, it is not clear from the available
statisticswhetherand where arms transactions
are included. Often there are figures in
special categories(such as SITC class 9) and
in those which lump civilian and military
goods together that may or may not be
correct; in these cases, however, it sometimes
is possible to compare this data with
informationfrom other sources and to come
up with some idea of the extent to which arms
transfers may have been included in trade
statistics.
Reporting of arms exports is probably
more completethan that of arms imports. At
minimum, there are single figures for arms
exports in the trade statisticsof several of the
largest arms exporters-including the United
States, France, Great Britain, and West
Germany. It is also generally assumed that
publishedtrade statistics for the Soviet Union
include arms transfers, though there is no
special category for such transfers.5 These
figures are not of much use for analysis of
their economiceffects in recipient countries,
sincethey are highly aggregated.
In some cases, for example the People's
Republic of China, it can be assumed that
arms exports are excluded from export
statistics, but most arms exports to Third
World countriesprobablyare reportedin trade
statistics. Since the same is not true for
importing countries, the arms trade is one
source for the observed divergenciesbetween

Military Trade, Aid, and Developing-Country Debt

exlport and import figures in bilateral and

global comparisonsof trade data (Yeats 1978).
Financial Data
There is little reason to expect to find data
specifically pertaining to arms imports or
related data in financialdata series, since such
data generallyis not functionallycategorized.
Still, some central banks--for example, those
of Brazil and of Israel--do publish separate
data on military-relatedtransactions. Such
data seems to be based on the evaluationof
standard forms that importers and exporters
have to fill out in many countries in
connectionwith foreignfinancialtransactions.
It is not known to this author whether such
questionnaire requirements in any other
countriessimilarlyreveal arms transfers.
In cases where arms transfers are
deliberatelyexcludedfrom trade statistics,and
thereby from the trade balance, it is necessary
to manipulatethe capitalbalance accordingly.
This can be done passively, by letting the
category of errors and omissionstake care of
the problem,or actively,by excludingrelevant
parts of financialtransactionsfrom the balance
of payments. In practice, this means keeping
a second set of books parallel to the official
data. It is not known to this author how
widespreadsuch practicesare.
A relatedmatter is that of indebtedness.It
is not altogether clear to what extent
indebtedness resulting from military
transactions is included in statistics of the
external debt. A number of creditors seem
expressly to separate military-related debt
from the rest of debt and not to report it to the
relevant authorities. The OECD Secretariat
used to specify in the technical notes to its
annual report on developing-countrydebt that
statisticsdid not cover officialmilitarycredits;
it estimated that, for non-OECD developing
countriesas a whole, officialmilitarydebt was
about 10 per cent of total reported long-term

Brzoska

debt, and well under 10 per cent of total

reported debt service.6 In the mid-1980s,
when more information on military-related
debt became available (e.g., by comparing
data sets and during debt reschedulings),the
OECD dropped this note. Now the OECD
Secretariatsimplystatesthat some information
on military debt is considered to be
insufficiently supported by firm data to be
includedin the statistics(OECD 1990,p. 22),
implying that other military credits are
included.
Other Sources
There is a large difference between what is
recorded in trade and finance statistics on the
one hand and what is known about arms
transfers on the other. Political, military, and
economic interests combine to put a quite
considerableamount of informationinto the
public sphere. One set of sources availableto
the public is based on the confidential data
base on arms transfers of the U.S.
government. Various U.S. government
agenciescontributeto this data base. Parts of
it are regularly reported in an aggregateway
by the U.S. Arms Control and Disarmament
Agency (U.S. ACDA) and the Congressional
Research Service of the U.S. Congress
(Grimmett). There are indications that the
data base is somewhatskewed, sincethe U.S.
government's agencies' interest in findingout
about arms transfers is not equally distributed
over all suppliers. In addition, prices for
suppliers other than the United States are
sometimesdifficultto estimate (see below).
Anotherset of sources originatesfrom the
economicinterestsof the sellers of arms and
the military strategic interests of many
recipients to make public what has been
transfered. Such information is regularly
published in a large number of trade journals
and special sales publicationsas well as in the
internationalpress. Becauseof the diversity
of the sources, it is a considerableeffort to

81

Military Expenditureand EconomicDevelopment

collectall the relevantdata into one database.

This is currently done only by the Stockholm
InternationalPeace ResearchInstitute(SIPRI).
SIPRI data is, however, limited to major
weapons,7 since considerablyless information
is available on small arms, components of
weapons, etc. (while the U.S. government's
figures do attempt to cover all transfers of
militarytechnology,includingcomponentsand
equipmentfor arms productionfacilities). The
SIPRI data is also problematic because there
are large gaps in publicreportingof individual
arms deals. The SovietUnion, for example,
did not make any information public until
recently, which probably resulted in underreporting of transfers firom the U.S.S.R.
There is also a strategic-military interest
among recipientsof weaponsin concealingthe
availabilityof certain weapon systems.
Both U.S. governmentand SIPRI data are
collectedin the form of registersof what types
of weapon systems are transfered from one
country to another. While SIPRI publishes
this kind of detailed information, the U.S.
government does not. The raw data is
prepared for publicationin various formats in
both sources by multiplying quantities by
prices. Quite often, actual prices are not
knownor do not reflect the militaryuse value
of the weapons transferedl. In such cases,
prices have to be estimatedon the basis of the
information available. The most difficult
pricing problems concern Soviet arms
transfers.
Thus, in 1986, the U.S.
government decided to substantiallyincrease
its estimate of Soviet weapons transfers
because it had changed its assumptionsabout
the amount of support material that was
suppliedalongsideweapons;it commentedthat
these revisions, "generally do not affect
underlying estimates of the number, type or
value of major military equipmentdeliveries,
but rather the estimated dollar value of
supporting material deliveries, particularly
those to countries engaged in ongoing
hostilities" (U.S. ACDA 1987, p. 145).

82

Through this change alone, the estimate of

total Soviet exports of arms for the period
1981-85 increased from US $56.8 million to
$86.2 million (in constantprices of 1983), or
almost52 per cent (U.S. ACDA, 1986, 1987).
The gaps in the original data and the
difficultiesin pricingplace severelimits on the
data kept by both the U.S. government and
SIPRI. Consequently, the data from both
sources is better suited for trend analysisthan
for inter-countrycomparisons. It is even less
suitable for comparison with economic data
such as trade and financial data. Until better
data is available--anExpert Group appointed
by the United Nations is currently studying a
proposalto createan arms trade register-these
data series are the only ones available. They
will be used here with due considerationof
their limitations.

InternationalTrade in Arms:
An Overview
Trend
Figure 1 shows that from the early 1960sto
the early 1980s,there was rapid growth in the
transfer of arms to countries in the Third
World. This growth is put in perspectiveby
measuring it against other indicators, for
instance the growth in total imports. Such a
comparison shows that the trend growth of
arms imports by Third World countriesover
thisperiodwas sharplyhigher than for imports
as a whole. In the late 1980s,the decline in
general trade was not reflected in a similar
declinein the import of weapons.
Phases and Structure8
Until the early 1960s, most arms were
transfered within military alliances--from
patronsto clients. Weaponstransfered,mostly
for free, were mainly surplus weapons--many
from World War II. Arms transferswere part
of the construction of military alliances,

Brzoska

Military Trade,Aid, and Developing-CountryDebt

Figure 1. Developing Countries' Imports of Arms and All Goods

(Index, 1985 = 100)
200 -

150
\

100 -

.....

-..

SIPRI major weapons

Imports all goods

..................
.. ......

ACDA arms Imports

50
1950

1955

1960

1965

1970

1975

1980

1985

Sources: SIPRIYearbook,variousyears; U.S. ArmsControl and DisarmamentAgency(ACDA),various years;

InternationalMonetaryFund, InternationalFinancialStatistics,variousissues (for data on all imports).

including the ring of containmentbuilt by the

United States on the southern borders of the
SovietUnion and Soviet effortsto destroy that
ring. Thus the first transfer of Soviet
weaponryoutside its bloc after World War II
was to Egypt in 1956.
The emergence of many new nations in
Africaand Asia in the early 1960s,balance-ofpaymentsproblems in the United States, and
the economic revival of Western Europe
slowly changed this structure. The United
States began to ask its allies to pay for the
arms they received. This further motivated
the larger Western European statesto broaden
their arms production activities. In parallel,
they expanded their efforts to sell weapons.
The Soviet Union tried to woo many nations
into its camp through its own continuedfree
transfer of arms, and in the mid-1960s, it
became the most important supplier of

weaponsto the Third World--forthe first time

surpassingthe United States. The position of
the importersof weaponsgot better and worse
at the same time: with more suppliers from
both East and West, they had more choice,but
at the same time they had to pay either in
moneyor with political allegiance.
During the 1960s,the groundwas laid for
the large expansionof the arms trade during
the 1970s. But it took other factors to enable
the arms trade to increase by a factor of 5
within a single decade. First, there was the
rise in the prices of oil and some other raw
materials,which gave some states in the Third
World unprecedented amounts of financial
resources. Second, with lots of money
available in some Third World states that
couldonly investpart of it, and with uncertain
economic prospects in the industrialized
countries,great amountsof financingon cheap

83

MilitaryExpenditureand EconomicDevelopment

terms--in the form of credits on international

financial markets--becameavailable to other
Third Worldcountriesthat were not originally
favored by increases in raw material prices.
A good part of the debt incurred in the 1970s
was used to buy weapons (see below). Third,
the 1973 October War in the Middle East
provided a strong stimulus for rearmament
throughout that conflict-torn and newly
enriched region. In many Third World
countries, arms transfers were riding the crest
of a procurementcycle. The armed forces of
countries that had received World War II
weaponsin the 1950sand early 1960swere in
dire need of replacements. The economic
crisis that in many industrialized countries
paralleled the oil price rises increased
commercial eagerness to make money from
arms transfers. "Arms for oil" became a
catchphrasenot only in Western Europe but
also in the Soviet Union, which greatly
increased its sales of weapons for hard
currency. In the United States, the situation
was also affected by the end of the Vietnam
War, which left some production capacity
unoccupied. The companieslooked for work
abroad. Finally, both the United States and
the Soviet Union changed their attitudes
towards arms transfers as a policy instrument.
The United States, influencedby the Vietnam
experience, formulated a policy--oftencalled
the Nixon Doctrine-of arms transfers to
friendly regionally dominant states in
preferenceto direct military involvement. The
Soviet Union also raised the importance of
arms transfers as an instrument of its foreign
policy.
In addition to the large volume increase in
the arms trade, importantstructural changesin
this trade became manifest in the 1970s.
First, the Third World segment of the anns
market became much more important as
transfers to the Third World grew rapidly
while trade among industrialized countries
stagnated. Second, used or outdatedweapons
were hardly traded any more. States that

84

could pay for more sophisticated weapons

demanded them--in some cases even
contributingfinanciallyto the developmentof
new weaponry (as when Iran partly financed
the developmentof the F-14 aircraft in the
United States). Third, more and more
customers demanded arms production
technology in addition to weapon systems.
Many of them wanted to build up domestic
arms industrieswith foreign help so that they
could become more independentin the long
run. Fourth, as will be reviewed in more
detail below, financingof the arms trade also
changed.
In the early 1970s, the postwar structure
of the arms market--a seller's market
dominatedby the UnitedStates and the Soviet
Union as suppliers with mainly political aims
in their arms transfer policies--reached its
apex. As the 1970s moved along, some
elements became visible that later came to
dominate the different structure of the arms
market in the 1980s: for example, increased
production in the Third World, increased
competition among producers, and a less
hierarchical structure.
Since the early 1980s,the arms trade has
stagnated. Severalfactors have contributedto
this trend.
First, raw material prices,
including those for oil, fell drastically.
Second, the accumulateddebt becamea heavy
burden in many Third World countries.
Third, government deficits that were allowed
to rise in the 1970s--partly to finance
increasing arms budgets--had to be cut in
order to stabilize economies that had gotten
out of fiscal and monetary control. In
addition, domestic arms production in Third
World countries became a factor to be
reckonedwith.
Effective demand nevertheless remained
high in the 1980s. It was fueled by several
conflicts, for examplethe Iraq-Iran War. In
some countries, for example Saudi Arabia,

Military Trade,Aid, and Developing-CountryDebt

Brzoska

'Figure 2. Exporters' Shares of Major Weapons Exports

to the Third World, 1951-89
(percentages of total)
100%

75%
USA
PR CHINA
UK

50%

FRANCE
OTHERS
USSR

25%

0%
51-4

Sources:

55-9

SIPRI Yearbook,

60-4

70-4
75-9
65-9
Years (1951-89)

80-4

85-9

various years.

buildup of the armedforces had been initiated

in the 1970s and was moving into its
hardware-expensivephases. The UnitedStates
and the Soviet Union, faced with clients who
could not pay for weapons, were again more
willingto increasetheir militaryaid. To some
degree, the trend towards commercialization
was undercut. Moreover, while demand
slowed, supplyincreased. In the 1980s,tough
competitionamongsuppliers marked the arms
trade as new suppliers entered the market in
both Europe (Spain, Greece) and the Third
World.
Some structuralchangeswere also notable.
First, the United Statesand the SovietUnion,
whbichwere not willing to fully enter the race
for orders and to abandon their political

approach to arms transfers, lost market

shares--although the Soviets had largely
regained about 45 per cent of the market by
the late 1980s after a drop over the previous
decade (see Figure 2). Whereas in the late
1970stheir combined share in arms transfers
to the Third World had stood at over 70 per
cent, by the mid-1980sit was less than 60 per
cent. This trend was reinforcedby the desire
of many recipientsto obtain weaponswithout
political strings attached-and therefore
preferably not from the United States or the
SovietUnion. Second,two tiers of the market
for complete weapon systems developed: in
one of these, new weapon systems are bought
by rich Third World states; in the other, poor
states are buyingold or simple weapons,often

85

Expenditureand EconomicDevelopment
Military
W

>iWMNWARMW9Ou

Figure 3. Regional Shares of Major Weapon Imports, 1951-89

(percenlages of total)
00%IE
.*...............
..,'il n{........-I

75%

~~

............ .....

... ....

Latin America

~~~~~~~~~Africa

....

East
mFar

50%

South Asia
Middle East
25%-

0%
51-4 55-9 80-4 85-9 70-4 751-9 80t-4 85t-9
Years (1951-89)

Sources: SIPRI Yearbook,variousyears.

from other ThbirdWorld countries or from

private arms dealers.
The businessof updatingweaponshas also
gained in importance. The Chinese arms
industry specializedin offeringits Soviet-type
armored vehicles and aircraft integratedwith
Westemn(British, Israeli, U.S.) electronics.
Third, arms production technology partly
substituted for the trade in completeweapon
Thbird Worldi customers are
systems.
demanding and receiving iworkshares in the
production of weapon systems they purchase
and licenses and componentsfor the weapons
they want to produce. Fourth, more recipients
are demanding large "offsets" in order to
lessen the economicburdensof arms imports;
these often take the forim of civilian or

86

militarygoodsthat have to be importedby the

weapon exporter.9 Fifth, the grey and black
markets became more important-at least
during the Iraq/Iran War. Many of the new
exporters rely on the services of private
dealersto enter the highly competitivemarket.
The same is true for established arms producers who do not want to be caught dealing
with problematic customers like Iran, for
example,and prefer to have intermediaries.
The stagnation in the arms market in the
1980s is largely explained by short-term
economicand long-term procurement factors
on the side of the recipients. It is therefore
possiblethat the volumeof arms transfers will
again increase-an issue that will be discussed
in more detail in the last section of this paper.

Military Trade,Aid, and Developing-CountryDebt

Brzoska

Table 1. Magnitude Comparisons of Third World Arms Inports, 1987

(percentages)
I

ArmsTmnortsin Relationto:
Net

SITC
Interest Disburse- Gross
InterClass Payments
ments
All
7
onExternal of ODA national
ImpgortsInpLort PublicDebt(all sources) Reserves
($ millions)
Low-IncomeEconomies
8,715
Middle-IncomeEconomies 19,995
High-IncomeOil Economies 5,115
All CountriesIncluded
33,825-

(percentages)

I
I
l

10
5
14
6

33
15
40
17

230
46
71

54
149
115

26
16
14
16

Note: Countrygroupingsaccordingto World Bank.

* The differencebetweenthe total for all developingcountriesgiven here and the total of $38,980 milliongiven
in U.S. ACDA (1989),p. 73, is due to the exclusionhere of: data ior China(ACDAarms export estimate:$600
million); data for World Bank groupingsof Non-ReportingMembers-including,among others, Cuba, Angola,
and North Korea (ACDA arms export estimate: $4,480 million);and World Bank Memberswith populations
below 1 million (ACDAarms export estimate:$75 million).
Sources: World Bank, WorldDevelopmentReport 1989; U.S. ACDA, World DevelopmentIndicators(1989),
Table 2.

Comparisons of Current Magnitude

There is some discrepancy-as is to be
expected, given the differences in coverage
mentionedearlier-betweenthe averagefigures
for arms imports by the Third World in the
late 1980sgivenby the U.S. governmentand
by SIPRI. While the U.S. government's
figure is above $35 billion, SIPRI's is above
$20 billion (U.S. ACDA 1990; Grimmett
1990; SIPRI 1990). The figures can be
comparedto other data to allow some grasp of
the magnitudes involved. As shown in
Table 1, there are significantvariationsin the
importanceof arms transfersmeasuredagainst
other indicators. The imports of arms by all
of the developing countries included, as
measured by the U.S. government, are quite
sizable-amountingto about 6 per cent of all

imports, and to some 17 per cent of importsof

machineryand transportequipment(class 7 of
the StandardInternationalTrade Classification,
SITC). Arms imports are almost as high as
interest payments on external public debt.
(Thisratio must be interpretedcarefully,since
arms transfers are financed in various ways-discussedbelow-while interest payments are
flows during a certain year.) On average,
arms imports by all developing countries, as
measured by the U.S. government, are
substantiallylarger in dollar terms than all net
developmentaid provided by all aid givers.
(It should again be noted, however, that arms
transfer data does not imply actual financial
flows during the year given.) Another
possible comparisonis to gross international
reserves, where again the quite substantial

87

MilitaryExpenditureand EconomicDevelopment

magnitude of amns imports, as measured by

the U.S. government, is visible.

phenomenon that has been studied in a

substantialbody of literature.

The Changing Fate of MilitaryAid

What is the actual extent of fungibility?It

can be arguedthat aid given for purposesthat
the receivingstate wouldnot finance out of its
own pocket, for instance, is not subject to this
kind of reasoning. Such aid does not allow
the government to substitute among goods
accordingto its own preferences.

Types of Aid
There are two basic conceptionsof military
aid. One is that any kind of support in the
militaryfield shouldbe calledmilitaryaid. In
this understanding, arms transfered are
military aid regardless of whether they are
paid for or suppliedfor free. This concept is
a political one; it starts from the contention
that all support in the mtilitaryfield is some
kind of "help" or aid. The other conceptionis
closer to the general understanding of
development or economic aid. Its basic
premise is afinancial one: the transfer is aid
only where there is a substantial grant
element. It is this second conceptionthat will
be used here, although it is realized that in
practice it might be difficult to establish the
difference.
Various types of financial support fall
under the category of military aid. They can
be, as generaleconomicaid, dividedin several
ways: by extent of the grant element; whether
they concern transfers of capital goods
(weapons, arms production facilities, etc.),

Martin McGuire (1982, 1987, with some

additionaltables) has built an elaboratemodel
of the Israeli economy in order to study
resource allocation in a case where outside
military threats and the inflows of both
military and economic aid are important.
Using data for 1960-79, he tests his model
using a number of mathematicalformulations
and various estimating procedures. Among
the results are first-round estimates of the
fractions of U.S. loans to Israel used as
fungible resources by the Israelis. For
military aid, this fraction, the fungibility
coefficient, ranges from 4 per cent to 18 per
cent; for "economic"aid, the range is from 90
per cent to 100 per cent. McGuire also
considers secondaryeffects of received U.S.
assistance, including political and economic
decision making in the United States, Israel,
and the Arab states. Considering such

operational

linkages, McGuire estimates that, on average,

goods

(ammunition,

logistic

23 per cent of U.S. military assistance has

"leaked"to civilian purposes and 40 per cent
of economic aid has leaked to "military"
purposes.

Then there is the questionof the fungibility

of aid, which has relevance for the military
efforts of recipient states. Aid given for
civilian use can be a means to increase
domestic expendituresfor military purposes.
Obviously, the reasoning can be reversed:
military aid may in fact allow the increase of
imports of civilian goods. Differences
between intended and actual effects of grants
or subsidies are of course a general

McGuire's results confirm the abstract

argument advancedabove. Closer inspection
of the aid data to Israelindicateswhy there are
differences in the fungibilityof military and
economicaid. Military aid has been mostly
connected with the subsidized delivery of
weapon systems, while economic aid has
mostlybeen budgetarysupport.

88

material, foodstuffs),or services;or whether

they are given in exchange for something
(base rights, alliance contributions,etc.).

Military Trade, Aid, and Developing-CountryDebt

U.S. Military Aid

Much imoredetailed informationis available
on U.S. military aid programs than on those
of other states.10 To some extent, the trends
in U.S. military aid also seem to be
representativeof other exportingstates.
The largest program in the past has been
the one originally instituted in 1949 to help
Western European countries, called the
Military Assistance Programme (MAP).
Under this program, weapons, ammunition,
spare parts, and logisticalsupportare supplied
as grants. The importance of MAP has
decreased since the 1960s, as more and more
states were asked to pay for their arms
imported from the United States. After
a trough in FY 1981, with only
reaLchin,g
$104.4 million, MAP grants have, however,
increased again--tomore than $700 millionin
FY 1988, with Turkey and Central American
states the main recipients(U.S. DoD 1990).
Arnotherrelevant program is the Foreign
Military Sales Programme(FMS),which is no
aid program in the narrow sense discussed
above. But since FY 1985, FMS has become
an instrument of military grant aid to Israel
and Egypt. Within the framework of the
FMS, the U.S. Departnent of Defense
organizesthe transfer of weaponsfor recipient
states and selling companies, including the
financial side of the transfer.
In the past, most of the transferswithinthe
FMS financingprogram were contractedat an
interest rate equal to the cost of creditsto the
U.S. Treasury plus a small administrative
surcharge,usuallywith 20-yearrepaymentand
10-yeargrace periods. For governmentsthat
have to pay higher premiums for their credits
on the open markets, this implies that FMS
financing at the Treasury rate carries a grant
element--albeita small one. In the 1980s,
interest rates below the Treasury rate were
increasinglyauthorizedby the U.S. Congress

Brzoska

and subsidized by the U.S. government.

Theseconcessionalrates have variedfrom case
to case, in recent years averaging about half
the going Treasuryrate (U.S. DoD 1989; for
concrete examples see Deger 1990; Platias
1990).
For the purposes of Figures 4 and 5, the
various military programs are aggregated
according to the type of financing arranged.
MAP, forgiven FMS, and other programs
provided at no cost are added up to the
categoryof "grants"proper; FMS financing is
subtractedfrom FMS and shown as "credit";
while the rest of FMS and commercialarms
sales are depictedas "cash." Both Figure 4
and Figure 5 show the decrease of full grant
aid in the 1970s, a slow though fluctuating
increasein credits, and a recoveryof grants in
the 1980s.
There are other aid programswith military
relevance. U.S. military aid from its
beginning has been closely connected to
civilianaid. The originalrationaleof U.S. aid
was to strengthen military and economic
capabilitiessimultaneously.This rationalewas
reiterated in the 1980s by the bipartisan
Commission on Economic and Security
Assistance(1983). One practicaloutgrowthof
this broad understandingof security was the
Defense Support Program of the 1950s that
later became the Economic Support Fund.
Although administered by the Agency for
International Development, these programs
provide recipientswith moneyto be used for
all kinds of purposes, includingmilitary ones.
Under the Defense Support Programs, and
also the PL 480 Food Aid Program, recipient
governmentssold goods--deliveredto them at
no cost--for local currencies to augment their
government budgets. The ESF programs,
quite sizable for countries like Egypt and
Israel, are mostly balance-of-payments
subsidiesby the U.S. governmentto recipient
governments to provide the time needed to
make long-termpolicy adjustments.

89

Military Expenditure and Economic Development

Figure 4. Different Types of Financing in U.S. Arms Sales to the Third World
(billions current US$)
20

Cash Payments
Figure
Share 5.

of

Tpes

iffernt

Grants

US.

Finncingin

CreditsI
'I -90
65

17

90

18

--

Total

5-

0
106 -1965

1970

1975

1980

1985

Figure S. Shares of Different Types of Financing in U.S. Arms Sales

to the Third World (percentages of total)

1965

1970

1975

Credits
_

1980
1980

Grants
3-

0%

Sources:

90

Cash Payments

1\\\\\\

25%

50%

U.S. DoDSAA Fiscal Year Series (1990).

75%

100%

Military 'rrade,Aid, and Developing-CountryDebt

The money givenby the U.S. government

under the ESF program is usually not tied to
specific projects, so that fungibility is high.
Thieseprograms are not military aid proper,
but they had--and in the case of the ESF
continue to have--large military components.
The U.S. governmentregards them as part of
its,security assistancein the broad sense. The
U.S. government reports ESF funds as
"development aid" to international
organizations such as the OECD, which
accordinglyincludesthem in its presentations
of ODA."1 Main recipientsof ESF funds in
the past were Israel, Egypt, and some Central
Americanstates.
Soviet Military Aid
Little is known about Soviet military aid
programs. They are often assumed to have
been large, though such assessmentsseem to
be built upon the broad concept of military
aid.
The U.S. government, which is the only
source that regularly provides estimates of
Soviet arms sales and deliveries,gives no clue
as to the distributionamongaid and trade. All
sales and deliveries are called military
assistance--a practice that many authors
follow. It is therefore confusing to compare
such data with those on Soviet economic aid--

another frequentpractice.
][nthe past, Soviet authoritiesrarely used
the concept of aid or assistance('pomoshch').
'They for long contended that aid was a
compensation paid by the former colonial
powersfor their earlier exploitation;sincethe
Soviet Union was not a former colonial
power, there was no reason for it to give aid.
The relations of the U.S.S.R. with Third
World countries have been portrayed as
'co-operation'with mutualbenefits. It follows
from this official line of reasoning that
outright grants would be rare and that aid
would take other forms, e.g., price subsidies

Brzoska

or credits with grant elements. It is very

difficult to find much discussion or hard
evidence about these issues. Much more
research effort has been put into a different,
though related question, namely, how
importanthard-currencyearningsfrom Soviet
arms transfers havebeen. The three issuesof
price subsidies, concessional credits, and
hard-currency earnings will be briefly
discussedhere.
Some information has been collected of
what are taken to be official Soviet weapon
prices (Efrat 1983). It seems that Soviet
negotiators presented price lists to their
counterparts in countries such as Egypt and
Syria; but it also seemsthat these governments
received weapons at substantially lower
prices--Egypt, for instance, at about 50 per
cent discount. The prices quotedby Efrat are
substantiallyhigher, by factors of 1 1/2 or
more than those used by the U.S. government
and SIPRI. It is difficultto substantiatewith
the informationobtainedby outsideobservers
whetherany of these prices were reflectiveof
the production costs of Soviet weaponry. In
fact, it now turns out that Soviet
economists--andarms sales negotiators--never
knew what individualweapon systems really
cost to produce.12
Somewhat more is known about credit

terms. The Indian government, which has

bought weapons from a large number of
sources,has preferred Sovietweaponsat least
partially because of their price--said to be
40-50 per cent lower than those for
comparable Western equipment. Duncan
reports that arms were usually delivered on
credit terms that had a grant element. It
seems that while they were often denoted in
hard currency, they could be repaid in goods
rather than in hard currency. Interest rates
were 2-2.5 per cent, with credits repayable
over 15-30 years (Duncan 1990, pp. 79-81).
These credit terms seem to have been fairly
universal,althoughindividualcustomersmight

91

MilitaryExpenditureand EconomicDevelopment

have been told that they were the best offers

ever made by the Soviet Union.13 Denoted
in hard currencies, these credit terms imply a
large grant elementfor most of the time since
the 1950s; if the notationsin hard currencies
were only nominal and the credits could in
fact be paid back without the use of hard
currency, the grant element might have been
even higher, dependingon the circumstances
of such repayments.
In additionto makingcredits concessional,
the SovietUnion has also forgivendebt. One
such case concerned Egypt. Moshe Efrat
(1983, p. 450) reports that according to
various Egyptiansources,all Egyptianmilitary
debt to the SovietUnion incurredbefore 1960
was cancelled in 1965. Of the debt incurred
later--Efrat's estimate is 1,500-1,800million
Egyptian pounds ($4-5 billion) by the
mid-1970s--about25 per centhad been repaid
by Egypt before its break with the Soviet
Union.
The question of Soviet hard-currency
income from weapon exports has attracted
considerable attention. Richard Portes has
calculateddata that gives a figure of $18.7
billion for 1970-81(quotedin Deger 1986,p.
167), correspondingto about 40 per cent of
Sovietarms exports as measuredby the U.S.
government. David Hollowayhas stated that
hard-currencyearningswere about$21 billion
between 1971-80(Holloway1983,p. 125), or
60 per cent of total arms exports as estimated
by the U.S. governmentand publishedby the
Central IntelligenceAgency (quotedhere from
Becker 1986, p. 6). Ericsson and Miller
(1979)of the U.S. Central IntelligenceAgency
have estimatedhard-currencyearningsof more
than $6 billion for 1974-79, or about 43 per
cent of arms exports as estimatedby the U.S.
government. Joan Parpart Zoeter (1982) of
the same organization has estimated
hard-currency income at $25.8 billion for
1973-81,or 75-96 per cent of total estimated
Soviet arms exports in individual years.

92

Roger Kanet (1983)has collected information

givinghard-currencyearningsshares between
50 per cent and 75 per cent from variousU.S.
government sources. Wharton Econometric
Forecasting Associates have estimated hardcurrency sales of $22 billion for 1980-83,or
74 per cent of total Sovietarms exports during
the same period.
All of these estimates are based on a
number of assumptionsabout trade volumes,
weapon prices, and payment methods.
Ericson and Miller extrapolated from a
detailed CIA estimate for hard-currency
earnings for 1977 of 43 per cent in all
CIA-recorded Soviet arms exports. Zoeter
does not give much background on the
estimationmethods that resulted in her very
high shares; her figures are not easily squared
with other informationon Soviet balance-ofpaymentsitemsduring the sameperiod. They
would, together with informationabout other
hard-currency incomes, suggest a surplus in
hard currency earnings--althoughat the same
time the Soviet Union incurred substantial
hard-currency debt.
Allan Smith has
suggested that in fact much of the
hard-currency income estimated by the CIA
was credit to be repaid with hard currency (or
commoditiesthat could be exported for hard
currency). In this interpretation,the CIA's
figures would cover credits denominated in
hard currencies as well as actual cash
payments (Smith 1986, pp. 153-54). Robin
Laird and colleagues of the Wharton
Econometrics Forecasting Associates have
estimated that around 70 per cent of hardcurrency sales were on a credit basis.
What can be made of these patches of
information?The Soviet Union has probably
supplied weapons for free to a number of
countries with which it had special political
relations or which it consideredstrategically
important. On many occasions,weaponswere
suppliedto these and other countrieson credit
but without much hope of payment. The

Military Trade,Aid, and Developing-CountryDebt

SovietUnionprobablyalso has aided recipient

states by supplying weapons at prices below
production costs. Credits have been on
concessional terms. On the other hand,
hard-currency income seems to have been
substantial--atleast since the 1970s, although
it is not clear how much of this was actual
cash income and how much was hard-currency
credit.14

Other Military Aid

While extensive military aid is an obvious
policy instrument for arms suppliers with
strong political motives, where commercial
considerations dominate, suppliers can be
expected to ask for full payment of the arms
supplied. Suppliers other than the United
Statesand the SovietUnionhave alsoprovided
military aid for various reasons-though at
much lower levels. One of these reasons is
that many other suppliers also haveused arms
transfers as political instruments--e.g., the
French government in the cases of some
African states, or China and some other
communist states in the cases of liberation
movements. Anotherreason is that, giventhe
tough competitionamongarms exporterssince
the early 1980s, customershave successfully
negotiatedfor package deals--includinggrant
military aid. Direct French military aid has
been on the scale of about $50 million per
year in the 1980s, with two-thirds of this
being equipment aid and one-third, training
5 (Chipman 1989, pp. 114-67);
assistance"
West Germanequipmentaid was less thanhalf
that level.
Since the oil crisistates
of 1973, so of the
oil-rich states of the Middle East also have
become important donors of military aid.
6 But it seems
Detailed data is not available.n
that Iraq, Egypt, and Afghan mudjahedeen
were the main recipients of such aid in the
1980s. One source puts financialaid by Arab
countriesto Iraq in the 1980son the order of

Brzoska

U.S. $ 50 billion (an amount equal to more

than half of Iraq's arms imports in this
period).'7
Summary
Althoughits exact level is not known, military
aid is quite substantialand probably accounts
for a good part of arms imports. A realistic
estimate could start by taking the available
U.S. data, a 40 per cent grant element in
Soviet arms transfers,'8 another 10 per cent
grant element in the transfers of the other
suppliers, and a 10 per cent deduction for
arms subsidiesby Arab states. This implies
that about one-third to one-half of arms
exportedare in effectdeliveredas grants. The
importanceof anns aid decreasedin the 1970s
and increased again in the 1980s. (All this
leaves aside considerationof the fungibilityof
non-militaryaid.)
Financing of Arms Imports
and Indebtedness
Simply put, there are two ways to finance
arms imports if weapons are not suppliedfor
free or financedby some third countrygiving
military aid: either they are paid for
immediately, in cash or kind, or later, via
credit.
In some of the supplier countries, special
creditsare availablefor the foreignpurchasers
of arms. The UnitedStates, for instance, has
the FMS FinancingProgrammentionedabove.
In other countries, weapon exports can be
financed through government owned or
regulatedagenciesset up to financeexports of
all kinds, including arms. Thus the Federal
Republic of Germany, for example, has the
KreditanstaltfUrWiederaufbau(KfW), which
has been involved in the financing of arms
exports--of Tornado aircraft to Jordan, for
instance.

93

Military Expenditure and Economic Development

Table 2. Military Debt Owed to United States, Compared to Total Debt Owed,
Selected Countries (end 1988)

Israel
Lebanon
Egypt
Greece
Palistan
Turkey
Jordan
El Salvador
Tunisia
Somalia
Oman
Botswana
Zaire
Thailand
Morocco
Sudan
Liberia
Korea, South
Honduras
Kenya
Philippines
Dominican Republic
Ecuador
Niger
Indonesia
Panama
Senegal
Cameroon
Peru
Malaysia
Bolivia
Gabon
Colombia
Jamaica
Haiti
Sri Lanka
Total

FMS Debt

Total Debt

Ratio of
EMS Debt to

(Smilolons)

(SmiXions)

Total Debt

8,945
83
5,981
2,595
1,475
2,936
343
89
306
91
89
11
150
346
334
148
18
393
28
50
225
16
37
5
123
12
9
9
29
30
6
3
13
3
1
2

25,990
497
49,890
23,914
16,692
40,932
5,360
1,755
6,747
2,006
2,850
511
8,571
20,707
20,094
11,516
1,629
40,459
3,322
5,967
30,052
3,843
10,500
1,696
52,668
5,302
3,711
4,039
18,118
22,680
5,631
2,549
17,008
4,537
846
4,732

34
17
12
11
9
7
6
5
5
5
3
2
2
2
2
1
1
1
1
1
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

25,171

1,174,180

Sources: U.S. GAO (1989);World Bank, World Debt Tables, 1989;,OECD, Financingand External Debt of
DevelopingCountries:A Survey, 1990.

94

Military Trade, Aid, and Developing-CountryDebt

Availablecircumstantialevidencesuggests,
however,that commercialbanks have been the
most importantsource of credit financingof
arms imports since the 1970s. Hard data on
'the involvementof commercialbanks is not
available. Commercial banks are generally
unwilling to discuss business details; and in
many cases not even the lending banks were
aware of the final destinationof the money
they transfered to govermnents. Thus
commercialbank credits, often in the form of
credits to governments,point towards another
fungibilityissue (discussedabove).
The followingsection will focus first on
direct military credits and then present an
approach to calculatingthe maximumuse of
credits for militarypurposes.
Debt to the U.S. Government
Due to Arms Imports
The U.S. governmentnot only publishes the
most detailed data on arms transfers and
militaryaid but has also released data on debt
owed due to FMS financingprograms.
Compared with total debt, FMS debt is
low, exceptfor a few countrieswith which the
United States has some type of political or
strategicalliance(Table 2). Some of the FMS
debt to these countrieshas been rescheduled
and even forgiven--e.g., in late 1990, the
Egyptian FMS debt of $7.2 billion
accumulated by that time was substantially
eased in reward of the Egyptian support
against the Iraqi invasion of Kuwait (Fadil
1990).
Estimating Debt to the U.S.S.R.
Due to Arms Imports
There is no comparabledata for the U.S.S.R.,
the largest exporter of arms in past decades.
In fact, Soviet debt data of any kind has only
become available recently (Izvestiya 1990),
and it is not clear how good this data is-for

Brzoska

instance, whether or not military credits are

included.19 Since, according to Western
estimates, arms exports have accounted for
40-50 per cent of all Soviet exports to Third
World countries, creditsto finance importsof
Soviet arms should be a large share of all
credits.
In Table 3, available data on developingcountry debt owed to the Soviet Union, as
reported by the Soviet government, is
comparedwith various trade data series; some
correlationswere calculatedbetweenthesedata
seriesand debt data (Table4). The data in the
second column of Table 3 (and the first of
Table 4) is the SIPRI estimate of Soviet
exports of major weapons for 1951-89. The
third columnof Table 3 is based on the same
data, but here it is assumedthat these transfers
were made on a credit basis with a 10-year
grace period and 20 equal yearly payments
thereafter (SIPRI Weighted). The fourth
column is the U.S. government's estimate of
Soviet arms transfers for 1983-87 (ACDA).
The last columnof Table 3 shows transfers of
all goodsfor the same time periodas givenby
official Soviet trade statistics--althoughthis
data in all probabilitydoes not include arms.
The strongest bivariate association exists
betweenthe export and debt series--withan r'
of .82--indicatingthat most of the debt in the
official statistics is related to civilian trade
agreements. The correlation coefficients
betweenthe arms transfer and the debt series
are also fairly large: .34 for SIPRI
Unweighted,.32 for SIPRIWeighted,and .47
for ACDA.20 Higher regression coefficients
result from a multivariateanalysis, including
both trade and arms transfer series.
Standardizedcoefficientsfor trade are about
twice the size of coefficients for the arms
transfer series. If the official debt data were
complete, these correlations would indicate
that about twice as much of the debt resulted
from civilian as from military trade. Given
the highershare of arms exportsin total trade,
this would imply that the share of arms

95

Fb

pyr-PpPppppoo.opp-oSeopv-p^o-reSJoo-.olo9.@PpPPp,pyww

&l

Xt~~

81t
'X

f1' 1.. .1 !m1

n.

------

~----------------------------------------------------------

-,

oo.ooooo._ws_swWoo_wovoO
- - - - - - - - - - - - - - - - - - - - - - - -

9fi*to_

- - - - - ---

_ _ ________________

iJ!

~~~

;0oof

PP

>

..

*.<kU

ooPo

- - - - - - -

Is

- - -

_________________

:E=CEC:=-=,-

Wi

Military Trade,Aid, and Devloping-Counny Debt

Brzoska

Table 4. Correlates of Developing-Country Debt to the Soviet Union (end 1989)

Arms Transfer Data

SovietExports
R2

SlPRIUnweighted

SIPRIWeighted

ACDA

0.30
0.59
0.90

0.30
0.60
0.91

0.32
0.56
0.89

Sources: See Table 3.

transfers financed by means other than

credits--throughgrants or cash payments-was
higher than that share for civilianexports. Of
course, official debt data may not correctly
reflect the military debt--though it seems
unlikely, given the calculated regression
coefficients, that military debt is totally
excludedfrom the official debt data.
Other Suppliers' Direct
Military Credit
What little information we have on other
countriesis circumstantial. It suggestsa trend
rather similar to the U.S. case: credits
assumedan im portant place in financingarms
importsbeginningin the 1970s. In the caseof
WesternEuropean suppliers,however, credits
seem to have come mosdy from commercial
banks and not direcdy from governments.
Estimates of Total Direct
Military Credit
Although not enough data is available to
actuallycalculatethe total direct militarydebt,
there are various ways to estimate its size.
One can simplyassumethat there is not much
difference between the extent to which the
exportsof all goodson the one hand and arms

on the other are financedthroughcredits. The

share of direct military debt in total debt can
then be assumedto be about6 per cent for the
last two decades--thesame as arms imports in
total imports. Consideringthat some countries
with large foreign-exchangeincome, such as
the oil-rich countries of the Middle East, had
over-proportionallylarge arms imports, there
is good reason to put the estimate somewhat
lower (Krause 1985b).
Another estimate of total military credit,
made by Walter Kitchenman,is based on the
extrapolationof the share of credits in U.S.
arms transfers over the period 1961-79. This
percentage is applied to total arms exports
from all suppliers. The result is a share of
approximately 12 per cent of direct military
credit in all credit and about 9 per cent of
directly military-relateddebt service in total
average yearly debt service over the period
studied (Kitchenman, 1983, p. 14). For
another estimate,detailedstatisticson military
aid and military credits were compiledfor the
1970sand early 1980s. The estimateis biased
for several years by rather high assumptions
about U.S. military aid. For years in which
this does not play a role, the estimated share
of military credit in new debt is close to the
figure estimatedby Kitchenman.

97

MilitaryExpenditureand EconomicDevelopment

As mentioned earlier, in the early 1980s,

the OECD used to state in its surveys on
external debt that military credit amounted to
about 10 per cent of total debt and debt
service for military creditsto less than 10 per
cent of total debt service.
Compared to the U.S. government's data
on arms imports, the reported estimates of
direct military debt correspond to 8-16 per
cent of the volume of arms transfers.
Opportunity-Cost Estimates of the
MilitaryDebt
Estimatesof total direct militarycredit are one
answer to the question of how much of the
total debt is debt for the purpose of arms
imports. There is, however, another way to
look at the matter: how muchlower could the
debt have been with lower or no arms
imports?
The starting pointfor estimatingmaximum
military debt is the assumptionof full fungibility of credit. The importantconsideration
then becomes whether a country's debt
increasedthrough its arms imports. The same
kind of reasoningof course can be appliedto
all other imports into a capital-importing
country. Arms imports are here singled out
for analyticalpurposesonly.
The simplestway to obitainan opportunitycost estimate is to compare military
expendituredata with debt data for a particular
year. The Kreditanstalt fur Wiederaufbau
calculateda share of military expendituresin
the gross inflowof externalresourcesfor 1979
for a number of countries. The calculated
shares ranged from 19 per cent for Sierra
Leone to 103 per cent for Pakistan. All Third
World military expenditurewas estimatedto
be on the order of 90 per cent of all net
external financial flows in 1980 (KfW 1983,
pp. 66-67).

98

But it is misleading to use military

expendituredata. Although there is a high
correlationbetweenarms imports and military
expenditures(Deger 1986, p. 145; McKinlay
1989, pp. 52-3), military expenditures are
predominantlyspent domestically. The usual
and better starting point for opportunity-cost
estimates of the military debt is countryspecific data on arms imports.21 Three ways
of financingsuch importsmust be considered:
direct outsidesubsidiesin the form of military
aid, cash payments, and credit payments.
Direct military aid has to be subtracted from
the arms importdata. Arms that are given for
free do not increase credits, capital imports,
and indebtedness. True cash sales in the
opportunity-cost sense discussed above can
only be made to countriesthat have the means
to import arms without increasing their net
debtor position. The rest of arms imports can
then be called credit-financedarms imports.
The calculationof the opportunitycost of
arms importscan be made for individualyears
or for a longer period and using different
financial aggregates. Helmut Maneval and
Pasi Rautsolaused various financialaggregates
for a simple opportunity-costcalculation,and
they used major weaponsimports for 1971-85
(as measuredby SIPRI) to make some simple
opportunity-costestimates for a number of
highlyindebtedcountries(see Table 5). They
assumed that capital imports in all countries
were higherthan arms importsthroughoutthe
period measuredand that there was no military
aid, and they ignored accumulationof debt.
The first assumption(which was not true in
the case of Venezuela, for instance)tends to
overstatethe opportunity-costof arms imports,
as does the second (which, for instance, was
wrong in the case of Peru, which obtained
large amounts of Soviet weapons on
concesssionalterms in the period considered).
The final assumptiontends to understate the
total opportunitycost of debt.

Military Trade, Aid, and Developing-Country Debt

Brzoska

Table S. Maneval and Rautsola's Country Estimates of the Opportunity Cost

of Arms Imports, 1971-85
(percentage shares)
Major WeaponImportsas Percentageof:
Public
External
Debt

Total
Exports

Total
Inports

Current
Account
Deficits

Argentina
Bolivia
IBrazil
Chile
]Ecuador
Colombia
Morocco
Mexico
Nigeria
Peru
Philippines
Venezuela

21
15
7
26
32
21
41
1
25
56
10
26

12
5
3
9
12
4
8
1
2
17
2
4

9
5
4
8
11
5
16
1
2
16
3
3

46
20
5
16
44
15
30
2
30
65
8
-23

Unweighted average

23

22

Sources: Maneval/Rautsola1990; Brzoska/Ohlson1987; IMP 1987; World Bank, World.DevelopmentReport

1987.

The ratios calculated by Maneval and

Rautsola vary greatly among countries, but
their unweighted averages come close to
averages for all Third World arms importers
determined by other authors (Brzoska 1983,
Tullberg 1985). All ratios can be interpreted
in an opportunity-cost way. With all other
things being equal, for instance, without the
observed expenditure on arms imports, the
highly indebted countries' debt could on
average have been 25 per cent lower by 1985.

government. True cash payments are assumed

for those countries which imported weapons
without producing an increase in their net debt
position. In cases where there was some
increase--but one amounting to less than total
arms imports--it was assumed that the
difference between reported arms imports and
net debt increase was cash-financed. The rest
of the arms imports that were not accounted
for by aid or cash sales were assumed to be
credit-financed in an opportunity-cost sense.

Table 6 is calculated using a procedure

developed earlier (Brzoska 1983): estimates of
military aid are subtracted from the total of
arms transfers, as given by the U.S.

The estimated maximum share of credits

associated with arms imports (imports by
countries increasing their indebtedness) grows
in the second half of the 1970s--mostly

99

Military Expenditure and Economic Development

Table 6. Estimates of Military-Related Debt, 1970-87

(millions current dollars)
Estimatesof Paymentsfrom:

1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987

Anrs Transfer
to Third World
Countries

U.S.
Military
Aid

Soviet
Military
Aid

Countries
Not Increasing
Indebtedness

Countries
Increasing
Indebtedness

4,205
4,735
7,430
9,745
8,430
9,190
12,340
15,380
17,545
19,285
39,520
43,654
45,616
45,273
44,422
34,974
32,822
35,954

1,434
1,370
2,440
3,485
1,590
1,580
270
240
215
225
610
610
1,194
1,587
2,013
3,333
3,758
4,031

900
960
2,630
3,180
2,460
2,400
3,200
3,900
4,400
6,200
11,530
10,830
11,183
11,059
10,600
8,659
8,915
11,040

800
240
880
440
1,690
3,250
2,300
2,960
4,050
3,840
7,860
12,200
13,990
13,680
15,850
7,230
8,630
6,680

1,070
2,165
1,480
2,640
2,690
1,960
6,590
8,280
8,880
9,020
19,520
20,194
18,850
20,947
15,979
15,752
11,519
14,203

Sources: Author's calculations(seetext) basedon Brzoska1983(for 1970-79);ACDA,variousyears; Grimmett,

Trendsut Convendonal
Anns Tranfers, 1988 and 1989, WorldBank,WorldDebt Tables,1989, and (for
1980-87),OECD, Ftnancingand Debt of DevelopingConries, 1990.

because the increase in total imports was not

accompaniedby increases in military aid. It
reaches a high level in the early 1980s,and it
decreases later in the 1980s, due both to an
increase in military aid and to smaller
increases in net debt for a number of armsimportingcountries. Nonetheless,the share of
total arms imports is rather high (about31 per
cent for 1985 to 1987, based on U.S. ACDA
data.)

o.

100

We can extend this discussion of debt

relatedto arms transfers, definedas above, to
interest and amortization. Assumptionshave
to be made about appropriateinterest rates and
debt-repaymentschedules. For the sake of the
argument, very low interest rates are chosen:
a fixed 5 per cent and one-half the Libor rate
for one-yearcredits. Averageinterest rates on
military credits are lower than the one forcommercial credits because of U.S.

Military Trade,Aid, and Developing-CountryDebt

Brzoska

Table 7. Estimates of Accumulation of Maximum Debt Due to Arms Imports,

Under Various Assumptions, 1970-87
(biUlionscurrent dollars)
Equal Repayments Over.
A
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
Sum diD

1
3
4
6
8
9
15
21
27
33
48
62
72
83

87
88
84
82

10 wears
B
C

0
0
0
0
0
0
1
1
1
2
2
3
4
4
4
4
4
4

0
0
0
0
1
1
1
2
3
4
4
6
8
10
12
14
15
17

1
3
4
7
9
11
17
24
32
39
56
73
88
103
113
122
126
132

0
0
0
0
0
0
0
1
1
2
3
5
5
4
5
4
3
3

20 years
B
C
0
0
0
0
0
1
1
1
2
2
3
4
4
5
6
6
6
7

0
0
0
0
1
0
0
1
1
2
4
6
6
5
7
6
4
5

100

0
0
0
0
0
1
1
1
1
2
2
3
4
5
6
7
8

1
3
5
7
9
11
17
25
33
41
59
77
93
110
122
133
140
148

J
50

30 years
B
C
D
0
0
0
0
0
1
1
1
2
2
3
4
5
6
6
7
7
7

0
0
0
0
1
0
0
1
2
2
4
6
6
6
7
6
5
6

0
0
0
0
0
0
0
1
1
1
1
2
3
3
4
5
5
6
33

A: Maximumcumulateddebt at year end due to arms transfers after deductionof principal repayment(maximum
new annual debt, taken from Table 6).
B: Interest on A at fixed 5 per cent interest rate.
C: Interet on A at half Libor rate for US S, one year.
D: Principal.
Sources:Author's calculationsbased on Table 6. For Libor: IMP.

concessional FMS financing and the low

iinterest rate for Soviet credits-but the
estimates chosen here are still conservative
lbecauseof the extent of credits at commercial
rates. It is further assumed that the debt is

repaid in equal installments over specified

periods. Table 7 presents the results for
calculationsassumingten, twenty, and thirty
equal annual repayments.

101

MilitaryExpenditure
and EconomicDevelopment

Figure 6. AccumuilatedMilitary-Related Debt,a as Share in New and Total Debt

1979

Share In New Debt

. . lE In Total Debt
Share

<\

1980

1981

Repayment/Arms

1982
1983
1984
1985

1988
1987
0

10

30

20

40

50

e0

70

"Here at 5 per cent fixed interestrate, 20 years repaymentin equal installments.

Sources: See Table 6 and text.

Figure 6 providesa fiutherperspectiveon

the magnitude of estimated maximum debt
connectedwith arms importson the basis of a
selected set of debt terms. On these
assumptions, new military debt ranged from
22 per cent of total new debt in 1980to over
50 per cent in 1984, after which it fell back
sharply to around 10 per cent by the late
1980s. The share of accumulateddirect and
indirect military debt (here a 5 per cent fixed
interest rate and 20-year repayment in equal
installments) slowly increased through the
1980s. The ratio of amortization(interest and
principal repayment) to arms imports-using
the same assumption as to interest and
principal--increasedquite dramaticallyin the
1980s, as did the ratio of repaymentto new
maximummilitarydebt.22

102.
102

i.

Correlates of the
Military-RelatedDebt
Regressionanalysisis yet another approachto
establishinglinks betweenmilitary activityand
Third World countries' indebtedness. The
most prolific author in this domain is Robert
Looney, who tested various possible relationships in a number of papers, using basically
the same cross-nationaldata for seventy-seven
countries based on readily available statistics
for 1981 (withsome exceptions). Looneyuses
absolute dollar values for the dependent
variable (debt) and many of his independent
variables (including, among others, military
expendituresand arms imports). This choice
of variables raises problems of multicollinearity and autocorrelation, which Looney

...

Military Trade,Aid, and Developing-CountryDebt

notes but does not try to correct. In one

paper, Looney (1987) reports results from
multiiple regression analysis with military
expenditures among a number of factors
argued to influence indebtedness--including
also gross domestic product, merchandise
exports, merchandiseimports,current-account
balance, and gross internationalreserves. The
most important independent variable,
explaining almost 70 per cent of variance
alone, is gross domestic product (note the
possibilityof simultaneity,however). Robert
Loorney finds no statistically significant
correlation with public domestic debt for his
77-countrysample.
This lack of result is not surprising.
Some countrieswith large debts had very little
expenditure on the military. Even more
important, a number of countries financed
arms imports without creating debt because
they had large surpluses in their trade
balances. In a mix of both capital-importing
and capital-exportingcountries,we should not
expect to find a correlation between arms
importsand indebtedness.
RobertLooneythen goes on to distinguish
sub-groups of his sample. In one exercise,
countriesare separatedby performinga factor
analysis with 34 variables that are all
somehowconnectedto internationaltrade and
financial transactions. The factor analysis
reduces these to seven major measures of
resource scarcity.
Using these seven
variables, a discriminant analysis puts the
countries into two groups: a "restrained"one
rnade up of countries usually smaller, less
economicallydynamic, and more reliant on
foreign external debt23 ; and an "unrestrained"group--consistingof severalmajor
oil exportersand several of the more dynamic
newly industrializingstates--characterizedas
less reliant on external debt (although the
group includes large borrowers such as
Mexico and Brazil). For the "unrestrained"
group, Robert Looney finds no statistically

Brzoska

significant correlation between military

expenditures and indebtedness. For the
"restrained" group, however, military
expendituresseem to be the most important
factor amongthose included in the regression
analysis. The coefficient is positive, large,
and significant.
Another separation is made between
military producers and non-producers, using
Stephanie Neuman's dichotomic distinction
(Neuman 1984).24 The result here is no
significant correlation for producers; and a
positive and significantcorrelationcoefficient
for non-producers (although t-values in this
regression are only half those reported in the
regression distinguishing the groups of
restrainedand unrestrained countries).
There is large overlapbetweenthe groups
of non-producers and restrained countries
(with several exceptions, such as nonproducers Saudi Arabia, Kuwait, and Syria;
and producers Pakistan, Sri Lanka, and the
25
Dominican Republic).
Robert Looney
concludesthat it is not the question of arms
producers versus non-arms producers that
creates differences;instead, it is the ability to
mobilize diversified sources of foreign
exchangethat determinesboth arms production
and the ability to avoid indebtedness(Looney
1987, p. 26). In another paper, Looney
extendedthis approachto includea test for the
relationship between arms transfers and
indebtedness(Looney1989b). He first repeats
the results obtained in the paper mentioned
earlier, finding a significant positive
correlationbetween military expendituresand
public external debt for the constrained
countries--butnot for the unconstrained,and
not for the total sample. He then proceedsto
present results of multiple regression analysis
with arms imports (ACDA data) in U.S.
dollars as a dependent variable and public
externaldebt in U.S. dollars amonga number
of independentvariables. For the group of
resource-constrainedcountries, he reports a

ZiMMM1A

103

MilitaryExpenditureand EconomicDevelopment

significant, positive coefficient and for the

unconstrainedcountries,a significantnegative
coefficient. For the total sample, the
coefficientis insignificant(p. 228).
Robert Looneyhas also tried to find out
whethertype of regime affectsthe relationship
between military factors and indebtedness
(Looney 1988b, pp. 25-26). Using data
collected by Ruth Sivard for 1982 (Sivard
1983), he separates the countries into two
groups, according to civilian or military
regime. The dependent variable he chooses
for his multipleregression analysisis military
expenditures;public externaldebt is one of his
independentvariables. This variablecarries a
significant,positive coefficientfor the military
regimes group and a significant negative
coefficientfor the civilian regimes. For the
total sample, there is no significant
correlation.
Louis Pilandon (1985) also has tried to
establish correlations between military
variablesand indebtednessby using regression
analysis. His data consists of various
indicators of debt as the dependent variable
and total military expenditures and military
expendituresper head as independentvariables
for thirty-seven Third World countries for
1965-80. He also distinguishestwo country
groups, one called "less industrialized"
(having less than 20 per cent of the
economically active population in industry)
and the other, "intermediate" countries.
Pilandon finds rather small correlations
between his variables. Only one involving
2 6 : the
debt data is statistically significant
relationshipbetween the total military budget
and the bilateral debt for the intermediate
countries (1985, p. 79).
Looney's work has been criticizedon a
number of grounds.27 'rhe most important
problem in the context discussed here is that
the fundamental difference between capitalimporting and capital-exportingcountries is

104

overlooked. For the latter, arms imports

reduce the abilityto import civiliangoods,but
probably will not lead to changes in the debt
position. It is only in capital-importing
countries that we would expect to see effects
of arms imports--and only small ones,
compared with other factors that were
instrumentalin buildingup Third World debt.
Louis Pilandon's lack of a range of significant
results is therefore plausible.

The Future of Arms Transfers

Some of the factors that contributed to the
stagnationof arms imports in the 1980swill
most probably also shape the future of arms
transfers. Some structural influenceson arms
transfers--suchas domestic arms production,
economicdifferentiationamong Third World
countries, and political developments within
Third World countries--might preclude a
continuationin the arms market of the 1990s
of the high levels of the 1980s.
An important additional factor that
influences future arms transfers is the
relationshipbetweenthe United Statesand the
Soviet Union. There are signs that they will
be less willingto provideweaponsfor free, or
on concessionalterms, in the future. A U.S.Soviet agreement on how to handle their
rivalry in the Third World, which mightput a
cap on military aid and thereby lower the
supply of weapons, seems closer now than
ever. On the other hand, many conflicts are
going on in Third World areas, fueling
demand. Procurement cycles are producing
militarypressuresto buy new weaponsystems.
In addition to such political factors, the
economicsituationis important. The costs of
economic restructuring lower the financial
maneuvering space of many governments.
Sizable debt repayments are another heavy
burden. SomnathSen has calculatedcombined
burdens of foreign debt interest and principal
repayments and military expenditures at the

Military Trade,Aid, and Developing-CountryDebt

Brzoska

level of 40-80 per cent of current government

revenues in many countries (Sen 1990,
pp. 2134).
A greater awarenessof the economiccosts
of arms transfers, includingtheir contribution
to, indebtedness, might influence the future
course of arms imports. After all, arms
imports come about after corresponding
decisionshave been made by those in power.
The high debt burden could in this sense be a
"window of opportunity"(Sen 1990, p. 213).
Aid and credit-givingagenciesand institutions
influencethe possibilityof many governments
to buy weapons. Military activityis internally
and externallyfinanced,and Louis Pilandonis
right when he writes: "In summary, foreign
aiidand bilateral credit allow countriesthat get
sujchfundsto have larger militarysectors than
would be possible without them" (1985,
p. 86).

But the relationshipis not a simple, linear

Arms imports seem to exhibit even
more resilience than military expenditures.
Currentlyavailablestatisticsdo not, however,
allow firm statementsabout the relationships
between arms imports and the economic
situation--especially concerning external
financing.
one.28

To be able to make better predictions on

the effects of the inflows of various types of
foreign financing on arms imports, we need
better data--on both arms transfers and the
financial arrangements that make them
possible. Such data could come from a
numberof sources, includinga specialregister
at the United Nations, better trade and
balance-of-paymentsdata, more detaileddebt
data, and government-backedexportinsurance
agencies.

REFERENCES
Notes

3 For exceptions, see Ball 1982, 1988; Deger 1986, pp.

146-8; Portes 1989.

The term "Thir World' is used here interchangeably

with "developing countries."
The preferred
geographical coverage is aU of the states in Latin
Amrenca, Africa, and Asia (except Japan).
exact
geographical coverage reported in tables and figures
follows the sources given.
Whenever U.S.
government data is used, Portugal, Greece, Turkey,
Yugoslavia, Bulgaria, and Romania are also included
as developing countries.
2 For summaries, see, e.g., Nawaz 1983; Chan 1985;
KoU1ner1987; Ball 1988; Harris 1988; Grobar and
Portes 1989; Frederikson and Looney 1989; Fontanel
1990; West 1990.

.B

On lags in the impact of military expenditures, see

Joerding (1985); Babin (1989); and Frederiksen
(1989).
The
5 There is some debate as to the inclusion of Soviet
arms transfers in official statistics (see, e.g., Smith
1986).

See, e.g., OECD 1984, p. 78. The same remark can

be found in later versions of this report.
7 Aircraft, missiles, ships above lOOt, armored
vehicles, electronic equipment such as guidance
equipment, and radar.

105

MilitaryExpenditureand EconomicDevelopment

8 Ibis sectionis an abbreviatedand updatedversionof

Brzoska (1988). For a raview of early post- World
War II conventionalweaponsproliferation,see SIPRI
(1971); for accounts of the current status, Klare
(1987);Catrina(1989);Neuman(1989);and Anthony
(1990).
9 Neuman1985;U.S. lTC 1985;U.S. GAO 1986;and
Griffinand Rouse 1986.
U.S. DoD; U.S. GAO 1982; McKinlayand Muhan
1984, pp. 30-50; U.S. CRS 1985; Grimmett1985a;
Grimmett1985b;Louscherand Salomone1987;Ball
1988, pp. 244-94.
1 For a critique, see Ball 1986, p.251.
12

This lack of oversightbecamea majorproblemwhen

Soviet authorities tried to calculate figures for
domestic military expenditures that included
procurementcosts in 1989(SIPRI1990, pp. 164-66).

13 Amongthe paperscaptured by the U.S. government

after the invasionof Grenada.

14

In later calculationsof the opportunitycost of debt,

I will assumethat, of the deliveriesreported in arms
trade statistics,60 per cent was actuallyor practically
granted for free (includingprice subsidiescompared
with the prices given by SIPRI and the U.S.
government),while 60 per cent had to be paid for in
hard-currency,either directlyor later.

15 French statistics on military assistancealso include

the substantial cost of thie overseas stationing of

Frenchtroops, which is not militaryaid in the sense
used here.

19 The 1991 "Joint Study" uses very similar data,

converted into U.S. dollars. (The total of 28,587
millionrubles comes to U.S. $121,656million. The
one importantdifferenceconcerns Cameroon. While
debt reported by 7zvestiya is below 200 million
rubles, the Joint Study figure is $1,520 million.
IMF, World Bank, OECD, European Bank for
Reconstructionand Development, A Study of the
Soviet Economy,Vol. 1, (Washington,D.C.: IMF,
1991, p. 118).
20 Correlations between the arms transfer series are
fairly high: .46 in the case of SIPRI Unweightedand
ACDA and .62 in the case of ACDA and SIPRI
Weighted. Coefficientsbetween arms transfer and
exportsare small: .11 for SIPRI Unweighted,.9 for
SIPRI Weighted,and .24 for ACDA.
21 Brzoska 1983; Tullberg 1985-slightly different
versions: 1987a, 1987 b; Krause 1985;
Maneval/Rautsola1990.
22 The share of accumulated debt in total debt is
obviouslyvery sensitive to the assumptionsabout
repayment.In an earlier paper, using Libor rates and
10-yearrepayment,I estimatedthis shareto be 22 per
cent already in 1979 (Brzoska1983, p. 274).
23 The distinction between these two groups of
"restrained" and "unrestrained" is a major
contribution by Robert Looney (and his frequent
co-authorPeter Frederiksen)to the general debateon
the economiceffects of the military sector in Third
World countries, which they have used in multiple
regression analysis for a large number of variables
(see, e.g., Looney 1986, 1988, 1989a; Looney and
Frederiksen1986, 1989;West, Forewordin Looney
1988; West 1990).

16 Officialdetailed data on giants from Arab countries

seems to omit military transactions(Arab National

and RegionalDevelopmentInstitutions,1983). In an
OECD study on Arab grants, it is said that some of
the reportedtransfer to frontline statesprobablywas
for military purposes; for about one-third of known
bilateral aid, the OECD could not determine the
geographicaldistribution(OECD 1983, 12).
17 Figure taken from EuropaischeWehrkunde,August

1990, p. 459.
I The Soviet estimate is comparativelyhigh because
ACDA data is used for arms transfers, which
probably overstates Soviet arms prices. For a
different view, see Efrat, 1985.

106

24 Stephanie Neuman's classificationhas as its main

criterion the ability to produce at least one type of

major weapon system (Neuman 1984). This is a
highly questionable criterion. It is not very
demanding to produce a simple naval craft, or to
fabricatea light plane. Her list includesa number of
producers with at maximuma few hundredpersons
working in arms production, like the Dominican
Republic,Mexico,and Colombia-alongsidecountries
with very large arms industries employingtens of
thousands of workers, like India and Egypt.
Neuman's criteriontends to distinguishcountrieswith
some industrial base--and thereby almost
automaticallyamong arms producers, at least for
simple weapons-fromcountrieswithoutany industrial
base. The latter ones are mostly in Africa, while in
Latin Americamost states are countedas producers.

Military Trade, Aid, and Developing-Country Debt

25 In another paper, Looney has used multiple

discriminantanalysis to find what combinationof
factors can best predict Neuman's two groupings.
Several measures of the size of inflow of external
resourcesare, after the size of militaryexpenditures,
the most important factors in such an exercise
(Looney1988c,70-71).Looneyhas some reservation
against the robustnessof this result because of the
many resource rich non-producers in the
non-producergroup (p. 77). It might be, as has been
suggestedby Alexander,Butz, and Mihalka (1981)
and Brzoskaand Ohlson(1985)that eithertoo much
or too little foreign exchange hamper the
development of a domestic arms industry. Where
there is much foreignexchangeavailable,direct arms
imports are a less roundaboutway to get implements
of war. Where too little foreign exchange is
available,the often substantialforeign-exchangecost
of domestic arms productioncannot be met (Terhal
1982;Brzoskaand Ohlson1986). It may be true, as
Robert Looney has stated, that domesticproduction
on average reduces armns imports (one of his
estimatesimplies an average reductionof 8-12 per
cent in arms imports for arms producers, Looney
1989c,23). But there may be a shilt fromthe import
of complete weapon systems to the import of preproducts and production technology not properly
included in the available arms transfer statistics. A
detailed analysisof the Indian case showedthe high
foreign exchange demand of domestic arms
production(Terhal 1982).
26 It may be noted in passing here that the only other

significant one he finds is between military

expenditures and public development aid for less
industrializedcountries,p. 79.
27 Robert West has
some doubts about the
RobertWest has expressed some doubts about the
validity of his discriminant analysis in separating
country groupings(West 1990). Fontanel finds his
econometricssimplistic and without sensitivity to
manytechnicalproblems(Fontanel1990). Maneval
and Rautsola(1990)find his ad hoc use of data series
ingenious and insist on the development of a
theoreticalmodel before testing. he use of absolute
dollar values for independentand dependentvariables
insteadof some ratio has been criticizedby a number
of authors,e.g., Pearson(1989). NicoleBall (1988)
has expressed serious criticismon the use of crossnational studies in a field greatly influenced by
various kind of specifics-especiallywiththe kind of
data that is available. She also questionswhetherthe
causalityissue-an issue that has been systematically
treated by Wayne Joerding (1986; see also
Frederikson (1989, p. 382)-can be ignored. To
these points shouldbe addedthe conclusion(reached

Brzoska

earlier in this paper) that arms importsrecordedin a

specific year do not mean that the corresponding
financialtransactionsoccurred in the same year.
25 To give one (somewhatodd) example:It seemsthat
more fiscal control, at least partially stimulatedby
requests from international financial institutions,
raised the importance of arms transfers as an
instrumentto fill sludgefunds for politicians.While
such funds can be fed from various sources, under
fiscal restraintarms transfersare especiallyattractive
because of the secrecy surrounding them in many
cases. Evidenceof increaseduse of arms transfersto
collectcommissionsby politiciansis sketchy,though
pervasive,and includesNigeria (Jaguar aircraft and
other cases), India (Bofors guns and HDW
submarines),and Greece(F-16 and Mirageaircraft).

Cited Studies
Alexander, Arthur J., William P. Butz, and Michael
Mihalka 1981. "Modeling the Production and
International Trade of Arms: An Economic
Framework for Analyzing Policy Alternatives,"
Rand Note N-1555-FF/RC(Santa Monica: Rand
Corporation).
Anthony, Ian 1990. "The InternationalArms Trade,"
Disarmament,13: 2, 231-56.
Arab National and Regional DevelopmentInstitutions
1982.
"Statement of Financial Operations,"
December 31, 1982 (Kuwait: Arab Fund for
Economicand SocialDevelopment).
Arbeitsgruppe Riistung und Unterentwicklung,
Universitit Hamburg(AGRU)1980. "Data in the
Socio-Military Field," Report prepared for the
United Nations Study on Disarmament and
Development,Hamburg.
Babin, Nehema 1989. "Military Spending, Economic
Growthand the Time Factor," Armed Forces and
Society, 15: 2, 249-62.
Bach, QuintinV. S. 1987. Soviet EconomicAssistance
to the Less Developed Countries (Oxford:
ClarendonPress).
Baek, Kwang-l1,Ronald D. McLaurin, and Chung-in
Moon, eds. 1989. The Dilemma of Third World
Defense Industries. Supplier Controlor Recipient
Autonomy(Boulder,Colo.: WestviewPress).
Ball, Nicole 1983. Third WorldSecurityExpenditure:A
Statistical Compendium (Stockholm: National
DefenceResearchInstitute).
Ball, Nicole 1988. Securityand Economyin the 7hird
World (Princeton:PrincetonUniversityPress).

107

Military Expenditure and Economic Development

Becker, AbrahamS. 1986. The Soviet Unionand the

Third World: The Economic Dimension, Rand
OccasionalPaper, February 1986.
Benoit,Emile 1973. Defenseand EconomicGrowth in
Developing Countries (Lexington, Mass.:
LexingtonBooks).
Benoit, Emile 1978. "Growth and Defense in
Developing Countries," Economic Development
and CulturalChange, 26: 2, 271-80.
Brzoska,Michael1982. "Arms TransferData Source,"
Journal of ConflictResolution,26: 1, 77-108.
Brzoska, Michael 1983. "Research Communication:
The Military-Related
E5ernal Debt of Third World
Countries," Journal of Peace Research, 20: 3,
271-77.
Brzoska, Michael 1986.
Rastungsexportpolitik
(Frankfurt:Haag und Herchen).
Brzoska, Michael 1988. "On ConventionalWeapons
Proliferation,"in EdwardA. Kolodziejand Patrick
M. Morgen, eds., Security and Arms Control
(Westport,Conn.: GreenwoodPress)
Brzoska, Michael, and Thomas Ohlson, eds., 1986.
Arms Production in the Third World (London:
Taylor & Francis).
Brzoska, Michael, and Thomas Ohlson 1987. Arms
Transfers to the Third World 1971-85 (Oxford:
Oxford UniversityPress).
Catrina, Christian 1989.
Arms Transfers and
Dependency(London:Taylor & Francis).
Chan, Steve 1985. "The lmpalctof DefenseSpendingon
EconomicPerfornmance:
A Surveyof Evidenceand
Problems," Orbis, 29: 2, 403-34.
Chipman,John 1989. FrenchPowerin Africa (Oxford:
Basil Blackwell).
Copper,J. F., and D. S. Papp, eds. 1983. Comtmunist
Nations' Military Assihtance (Boulder, Colo.:
WestviewPress).
Deger, Saadet 1985. "SovietArms Sales to Developing
Countries: The Economic Forces," in Robert
Cassen, ed., Soviet Interests in the Third World
(London:Sage Publications).
Deger, Saadet 1986. Militajy Expenditurein Third
World Countries (London: Routledge & Kegan
Paul).
Deger, Saadet 1990. "The EconomicCosts and Benefits
of U.S.-TurkishMilitaryRelations,"in Jane Sharp,
ed., Europe After an American Withdrawal
(Oxford:OxfordUniversityPress), 243-73.
De Masi, Paula, and Henri Lorie 1989. How Resilient
are Military Expenditures?IMF Staff Papers,
36: 1, 130-65.
Duncan, Peter J. S. 1990. The Soviet Unionand India
(London:Routledge).
Efrat, Moshe 1983. "The Economicsof Soviet Arms
Transfers to the Third World-A Case Study:
Egypt," Soviet Studies, 35: 4, 437-56.

108

Ericsson, P. G., and R. S. Miller 1979. "Soviet

Economic Behavior in a Balance-of-Payments
Perspective," in Joint Economic Committee,
Congressof the UnitedStates, The Soviet Economy
in a Time of Change (Washington, D.C.:
GovernmentPrinting Office), 222-34.
Fadil, Magda Abu 1990. "Cairo Cashes In," The
MiddleEast, December1990, 32-33.
Fontanel, Jacques 1990. "The Economic Effects of
Military Expendituresin Third World Countries,"
Journalof Peace Research,27: 4, 461-66.
Frederiksen,Peter C. 1989. "The RelationshipBetween
Defence Spendingand Economic Growth: Some
Evidence for Indonesia, 1964-85," Contemporary
SoutheastAsia, 10: 4, 375-84.
Frey, Bruno 1975. "Weapon Export and Aid to
DevelopingCountries,"Journal of Peace Science,
1: 2, 1-15.
General Agreement on Trade and Tariffs (GATT),
annual. InternationalTrade(Geneva: GATT).
Griffin, John C., and William Rouse 1986.
"Counter-Tradeas a Third World Strategy of
Development," Third World Quarterly, 8: 1,
January, 177-204.
Grimmett, Richard F. 1985a. "An Overview of the
United States Military Assistance Programs,"
CongressionalResearch Service Report 85-91 F
(Washington,D.C.), April 23, 1985.
Grimmett, Richard F. 1985b. "The Role of Security
Assistance in Historical Perspective," in Ernest
Gravesand StevenA. Hildreth,eds., U.S. Security
Assistance(Lexington,Mass: LexingtonBooks).
Grimmett, Richard F. annual. Trendsin Conventional
Arns Transfers to the Third World by Major
Suppliers (Washington, D.C.: Congressional
ResearchService).
Grobar,Lisa M., and Richard C. Porter 1989. "Benoit
Revisited:DefenseSpendingand EconomicGrowth
in LDCs," Journal of Conflict Resolution, 33: 2,
318-45.
Harris, GeoffT. 1988. "EconomicAspectsof Military
Expenditure in Developing Countries: Survey
Article," ContemporarySouth East Asia, 10: 1,
82-102.
Hdbert, Jean-Paul 1988. Les Ventes d'Armes (Paris:
Syros).
Holloway,David 1983. TheSoviet Unionand the Arms
Race (New Haven, Conn.: Yale UniversityPress).
Holtzman, R. 1989. "Politics and Guesswork,"
InternationalSecurity, 14: 2, 101-31.
Hosmer, Stephen T., and Thomas W. Wolfe 1986.
Soviet Policy and Practice Toward Third World
Conflicts(Lexington,Mass.: LexingtonBooks).
Hudson, George E. 1987. "The Economicsof Soviet
Arms Transfers: A Policy Dilemma," in David J.
Louscherand Michael Salomone,eds., Marketing

Military Trade, Aid, and Developing-Country Debt

Security Assistance (Lexington, Mass.: D. C.

Heath).
International Monetary Fund (IMF). International
FinancialStatistics, monthly (Washington,D.C.:
IMF).
Izvestiya1990. "Press Focus: Unique Document-Who
'Owes' Us 85.8 BillionRubles?"March 2, 1990,
p. 3 (in Russian; translated in English in:
FBIS-SOV-90-043,March 5, 1990).
Joerding,Wayne 1986. "EconomicGrowthand Defense
Spending: Granger Causality," Journal of
DevelopmentEconomics,21: 1, 3540.
Kaldor, Mary, and Asbj5rnEide, eds. 1979. The World
MilitaryOrder:The Impactof MilitaryTechnology
on the Third World(London:Macmillan).
Kanet, Roger E. 1983. "Soviet and East European
Arms Transfers to the Third World: Strategic,
Politicaland EconomicFactors," Paper presented
to 1983NATO Colloquiumon ExternalRelations
of CMEACountries, Brussels.
Katz, James, ed. 1986. The Implicationsof ThirdWorld
Military Industrialization: Sowing the Serpent's
Teeth (Lexington,Mass.: LexingtonBooks).
Kitchenman,Walter F. 1983. "ArmsTransfersand the
Indebtednessof Less DevelopedCountries,"Rand
Note N-2020-FF (Santa Monica, Calif.: Rand
Corporation).
Klare, MichaelT. 1987. "The Arms Trade: Changing
Pattems in the 1980s," Third World Quarterly,
9: 4, 1257-81.
K611ner, Lutz 1987. Rdstung und Entwicklung:
politische, wirtschaftliche und finanzielle
Voraussetzungenund Folgen in Entwicklungsklndern (Munich:WeltforumVerlag).
Kreditanstaltfur Wiederaufbau(KfW) 1983. Bericht
dber das Geschaftsjahr1982 (Frankfurt:KfW).
Kramer, Mark N. 1987. "SovietArms Transfersto the
Third World," Problems of Communism,36: 5,
52-68.
IKrause, Joachim 1985a. "Aufzeichnungbetr: Der
Internationale Hamdel mit Waffen und
Riustungsgiitern,"SWP- AZ 2437, (Ebenhausen:
StiftungWissenschaftund Politik).
Krause, Joachim 1985b.
Die sowjetische
Militarhityepolitikgegenaber der Dritten Welt
(Baden-Baden:Nomos).
Laird, Robin, and Wharton EconometricForecasting
AssociatesStaff 1983. Soviet Arms Transfersto
the Non-CommunistThirdWorld in the 1970sand
1980s (WashingtonD.C.: Wharton Econometric
ForecastingAssociates).
Lock, Peter 1974. "BundesdeutscheRuistungsexporte
und Militirhilfe," in StudiengruppeMilitarpolitik,
ed., Antiweissbuch(Reinbek:Rowohlt).

Brzoska

Lock, Peter 1979. New InternationalEconomic Order

and Armament,Vierteljahresberichteder Friedrich
Ebert-Stiftung,p. 77.
Lock, Peter, and Herbert Wulf, 1979. "Consequences
of the Transferof Military-OrientedTechnologyon
the DevelopmentProcess," in Mary Kaldor and
Asbj6rnEide, eds., TheWorldMilitaryOrder:The
hnpact of Military Technologyon the Third World
(London:Macmillan).
Looney, Robert E. 1986. The Political Economy of
LatinAmericanDefenseExpenditures:CaseStudies
of Venezuelaand Argentina (Lexington, Mass.:
LexingtonBooks).
Looney, Robert E. 1987. "Impact of Military
Expenditureson Third World Debt," Canadian
Journal of DevelopmentStudies, 8: 1, 7-26.
Looney, Robert E. 1988a. Third World Military
Expenditures and Anrs Production (London:
Macmillan).
Looney, Robert E. 1988b. "The Political Economyof
Third World Military Expenditures: Impact of
RegimeType on the DefenseAllocationProcess,"
Journal of Politicaland MilitarySociology,16: 1,
21-19.
Looney, Robert E. 1988c. "Economic Environments
Conducive to Indigenous Third World Arms
Production," The Singapore Economic Review,
33: 2, 63-78.
Looney, Robert E. 1989a. "Intemal and External
Factors Affecting Third World Military
Expenditures,"Journal of Peace Research,26: 1,
33-46.
Looney, Robert E. 1989b. "The Influenceof Arms
Imports on Third-World Debt," Journal of
DevelopingAreas, 23: 2, 221-31.
Looney, Robert E. 1989c. "Have Third-WorldArrns
Industries Reduced Arms Imports?," Current
Researchon Peace and wilence,13: 1, 15-26.
Looney, Robert E., and Peter C. Frederikson 1986.
"DefenseExpenditures,Extemal Public Debt and
Growthin DevelopingCountries,"Journalof Peace
Research,23: 4, 329-38.
Looney, Robert E., and Peter C. Frederikson 1989.
"RecentResearchon DefenseSpendingand Growth
and Implicationsfor Third World Disarmament,"
The JerusalemJournal of InternationalRelations,
11: 1, 1-11.
Louscher,DavidI., and MichaelSalomone,eds. 1987.
MarketingSecurityAssistance (Lexington,Mass.:
D. C. Heath).
Luckham, Robin 1985. "Militarizationin Africa," in
SIPRI, WorldArmamentsand Disarmament,SIPRI
Yearbook 1985 (London: Taylor & Francis),
295-328.

109

Military Expenditure and Economic Development

Maizels, Alfred, and Machiko Nissanke 1986. "The

Determinants of Military Expenditures in
DevelopingCountries,"WorldDevelopment,14: 9,
1125-40.
Maneval, Helmut, and Pasi Rautsola 1990.
Rdstungsimporte und die Weltschuldenkrise,
Jahrbtcher fir Nationalokonomieund Statistik,
207: 2, 130-42.
McGuire, Martin C. 1982. "U.S. Assistance, Israeli
Allocationand the Arms Race in the MiddleEast,"
Journal of Conflict Resolution, 26: 2, June,
199-235.
McGuire, Martin C. 1987. "U.S. Assistance, Israeli
Allocationand the Arms Race in the MiddleEast,"
in Christian Schmidt, ed., The Economics of
MilitaryExpenditure(London:Macmillan).
McKinlay, Robert D. 1989.. Third World Military
Expenditure: Determinants and Implications
(London:Pinter Publishers).
McKinlay,Robert D., and A. Mughan 1984. Aid and
Arns to the Third World: An Analysis of the
Distributionand Impact of U.S. OfficialTransfers
(London:Pinter Publishers).
Ministerstvo Vneshnikh EkonomicheskikhSviazanieiy
SSSR, Statisticheskiy
SSSR. VneshnaiaTorgovolaia
Finansy i Statistika).
Sbornik, annual (Moscowv:
Nawaz, Shuja 1983. "Economic Impact of Defense
Expenditures,"Finance 2ndDevelopment,20: 1.
"International
Neuman, Stephanie G. 1984.
Stratificationand Third 'WorldMilitaryIndustry,"
InternationalOrganisation,38: 1.
Neuman,StephanieG. 1985. "Coproduction,Barterand
Countertrade: Offsets int the InternationalArns
Market," Orbis, 29: 1, 183-213.
Neuman, StephanieG. 1989. "The Arms Race: Who's
on Top," Orbis, 33: 4, 509-21.
Organisation for Economic Co-operation and
Development(OECD), Financing and External
Debt of Developing Countries:A Survey, annual
(Paris: OECD).
Organisation for Economic Co-operation and
Development (OECD) 1983. Aid from OPEC
Countries(Paris: OECD).
Organisation for Economic Co-operation and
Development(OECD) 1990. The Export Credit
Financing System in OECD Member Countries,
fourth edition(Paris: OECD).
Pearson, Frederic S. 1989. "The Correlatesof Arms
Importation,"Journal of Peace Research,26: 2,
153-63.
Pilandon, Louis 1985.
"Influence des relations
6onomiqueNord-Sud suir les depenses militaires
des pays en voie de d6veloppement,"
Revue lutudes
Internationales,16: 1, 75-86.
Platias,AthanassiosG. 1990. 'Costs and Benefitsof the
U.S. Military Bases in Gireece,"in Jane Sharp,

110

ed., EuropeAfter an AmericanWithdrawal(Oxford:

OxfordUniversityPress), 218-42.
Schmidt,Christian1984. "D6pensesmilitaires,industrie
d'armement et endettement du Tiers Monde,"
DefenseNationale,40: 12, 75-84.
Sen, Somnath 1990. Debt, Financial Flows and
International Security, SIPRI Yearbook 1990
(Oxford:OxfordUniversityPress).
Shubik, Martin, and Paul Bracken, 1983. "Strategic
Purposeand the IntemationalEconomy,"Orbis, 27:
3, 567-89.
Sivard, Ruth. WorldMilitaryand SocialExpenditures,
annual (Leesburg,Va.: World Priorities).
SIPRI (StockholmInternationalPeace ResearchInstitute)
1981. The Arms Trade With the Third World
(Stockholm:Almquist& Wicksell).
SIPRI. WorldArmamamentsand Disarmrnament:
SIPRI
Yearbook, annual (Oxford: Oxford University
Press).
Smith, Alan 1986. "The SovietEconomyand the Third
World," in Robert Cassen, ed., Soviet Interests in
the Third World(London:Sage Publications).
Terhal, Pieter 1982. "Foreign ExchangeCosts of the
Indian Military 1950-72," Journal of Peace
Research,19: 3, 251-60.
Commissionon Securityand EconomicAssistance1985.
A Report to the Secretary of State (Washington,
D.C.: Departmentof State).
Tullberg-McWilliams,Rita 1985. "Military-Related
Debt in Non-OilDevelopingCountries," in SIPRI,
World Armamaments and Disarmament; SIPRI
Yearbook1985 (Oxford:OxfordUniversityPress),
445-58.
Tullberg-McWilliams,Rita 1987a. "Military-Related
Debt in Non-Oil Developing Countries," in
ChristianSchmidtand Frank Blackaby,eds., Peace,
Defense and Economic Analysis (London:
Macmillan),302-16.
Tullberg-McWilliams,Rita 1987b. "La deuda por
Gastos Militares en los Paises en Dessarollos no
Petroleros1972-1982,"CommercioExterior,37: 3,
196-203.
U.S. Arms Control and DisarmamentAgency, World
MilitaryExpendituresand Arms Transfers,annual
(Washington,D.C.: GovemmentPrintingOffice).
U.S. Congressional Research Service (CRS), U.S.
Military Sales and Assistance Programs: Laws,
Regulations,and Procedures,Report prepared for
the Subcommitteeon Arms Control, International
Security and Science of the U.S. House of
Representatives, Committee on Foreign Affairs
(WashingtonD.C.: Government Printing Office,
July 1985).
U.S. Department of Defense, Defense Security
AssistanceAgency, Comptroller. ForeignMilitary
Sales, Foreign Military Construction Sales and

Military Trade,Aid, and Developing-CountryDebt

Military Assistance Facts, annual. Defense

Security Assistance Agency, Data Management
Division,Washington,D.C.
U.S. Department of Defense, Security Assistance
Agency. CongressionalPresentationfor Security
AssistancePrograms,annual, Washington,D.C.
Ul.S. Department of Defense, Security Assistance
Agency, annual. Fiscal Year Series. Data
ManagementDivision, Washington,D.C.
U.S. General Accounting Office (GAO) 1978.
"Summaryof Effortsto Recover U.S. Government
Costs in ForeignMilitarySales," ID-77-56,GAO,
Washington,D.C..
U.S. General AccountingOffice (GAO) 1982. "U.S.
Security and Military Assistance: Programs and
Related Activities," GAO/ID-82-40, GAO,
Washington,D.C..
IU.S. General AccountingOffice (GAO) 1983. "U.S.
Assistanceto the State of Israel," GAO/ID-83-51,
GAO, Washington,D.C..
IJ.S. GeneralAccountingOffice (GAO)1986. "Military
Export: Analysisof an IntragencyStudyon Trade
Offsets," GAO/NSIAD-86-99BR, GAO,
Washington,D.C..
lJ.S. GeneralAccountingOffice(GAO)1989. "Security
Assistance: Foreign Military Sales Debt

Brzoska

Refinancing," GAO/NSIAD-89-175, GAO,

Washington,D.C..
U.S. InternationalTrade Commission(USITC) 1985.
"Assessment of the Effects of Barter and
Countertrade on U.S. Industries," USITC
Publication1766, Washington,D.C.
West, Robert L. 1990. "Problems of Third World
NationalSecurityExpenditures,"Paper prepared for
presentationat U.S. Instituteof Peace Conference
on ConffictResolutionin the Post-Cold-WarThird
World, Washington,D.C., October3-5, 1990.
WhartonEconometricForecastingAssociates,Centrally
PlannedEconomies,Vol. 1, 1981.
World Bank. WorldDebt Tables, annual (Washington,
D.C.: World Bank).
Wulf 1980. Banken verdienen an der Aufriistung,
epd-Entwicklungspolitik,
18-19.
Yeats, Alexander 1978. "On the Accuracyof Partner
Country Trade Statistics," Oxford Bulletin of
Statistics, 4: 341-61.
Zoeter, Joan Parpart 1982. "U.S.S.R.: Hard Currency
Trade and Payments," in Joint Economic
Commitee, Congress of the United States, Soviet
Economy in the 1980s: Problems and Prospects,
Part 2 (Washington,D.C.: GovernmentPrinting
Office), 479-506.

111

6
DETERMINANTS OF MILITARY EXPENDITURE
IN DEVELOPING COUNTRIES:
REVIEW OF ACADEMIC RESEARCH
Robert L. West

Introduction
This paper is both a review and an assessment
of recent academic literature on the
determinants of military expenditure in
developing countries. The objective of the
literaturereview is: a) to identifythe theories
and hypothesesin recent scholarlyresearch on
the causes of military expenditure,and b) to
summarize the empirical evidence in the
literature on these hypotheses, highlighting
areas of consensus and controversy. More
specifically,the literaturereviewis intendedto
do the following:
* Review, analyze, and synthesize the
research and empirical evidence on the
determinants of military expenditure in
developingcountries;and
* Assess the important distinctions among
country sub-groups with different
conditionsthat are likely to influencethe
causes of military-sectorexpenditures.
Recent scholarly literature on these two
topics is reviewed in parts II and III of this
paper.

The literature reviewed is almost entirely

confinedto items that have been publishedin
the English language during the past two
decades. The purpose is to ensure that the
reader will, to the extent possible, find the
literaturecited to be readily accessible. While
a substantialbody of writing on these topics
has appeared in other languages, for some
decadesEnglishhas quite clearlyprovidedthe
dominantmedium for publicationin this area
of scholarly research.
Distinctive
contributions in other languages promptly
appearin Englishtranslationor interpretation.
Other important criteria of selection are
the emphasisin this paper on studiesthat have
systematically evaluated empirical evidence
and on the work of researchers who have
subjectedexplicitpropositionsor hypothesesto
empirical tests. Statistical analyses of the
causes and consequences of military
expenditurehave grown rapidlyin volume and
sophisticationover the past several decades;
this constitutesa clear and importanttrend in
scholarly research on the relationship of
security and development. Nonetheless, in

113

MilitaryExpenditureand EconomicDevelopment

selecting studies for review, empirical

orientationand formal testingprocedureshave
been given more than proportional
representation.
A survey of the literature shows that most
of the recent statistical work on the
determinantsof security expenditures in the
Third World has been based on analysis of
relationships observed in a selection of
countries. There are several reasons for this.
Researchers have found the cross-country
studiesto be more useful than case studiesand
longitudinalinvestigationsin seekingto verify
patterns applicable to a broad spectrum of
countries. Cross-sectionalresearch has also
been encouragedby the development(over the
past twenty years) of regular data assembly,
editing, and reporting on military-sector
activity in a large number of countries--a
cooperative international enterprise in data
publicationthat is discussedbelow.
This emphasisamong investigatorson the
use of macrostatistical and cross-sectional
research methods has led to the increased
employment of econometric techniques in
studying both the causes and the effects of
military expenditures. Econometricmethods
have become increasingly popular among
investigatorsin this subjectarea because they
permit the adoption of familiar techniquesto
survey simultaneously many countries and
multipleexplanatoryvariables.
For both of the topics reviewed in this
paper, research employingeconometrictesting
modelsand a variety of relatedmacrostatistical
methods is found to be highly influential in
fashioning the body of widely accepted
evidence concerning the relationship of
security and developmentin the Third World.
This demandsattention to the questionof the
reliabilityof statisticalinferencesderived from
econometrictests of cross-countryexperience
in this area. In this review of the literature,
readers are urged to assess with care the

114

reliability of findings based upon

macrostatistical methods of empirical
investigationas these have been used in this
subject area. This concern is a reflectionof a
vigorous and unresolved debate in the
literature with respect to the degree of
confidence with which readers may accept
conclusions about the causes of military
expendituresthat rest upon evidenceobtained
from cross-sectionalstatisticaltests. A broadranging critique of the reliability of the
research methodsthat have been employed in
this literature may be found in Chapters3 and
4 of NicoleBall's recent volume, Securityand
Economy in the Third World (Ball 1988,
pp. 84-157). The parameters of the debate
between the critics and defenders of the
disputedresearch methodswere defined more
than a decade ago in an exchange in Orbis
betweenStephanieNeumanand Dan and Ron
Smith (see Neuman 1978; Smith and Smith
1979).
Questionsaboutthe reliabilityof research
findingsderive from problemsconcerningthe
data employed in the analysis and from
problemsrespectingthe researchmethodology.
Both sets of problems--brieflydescribedin the
remainder of this Introduction--arise in
evaluatingthe findingsreviewed in this paper
relating to the determinants of military
spending(part II) and how these determinants
differ in sub-groupsof countries (part III).

Problemsof Data
The past two decades have seen a very
significantimprovementin the availabilityof
quantitative information about the level of
military-sectoractivity in a large number of
countries. Key indicators of this activity in
virtually all countriesof the world--including
estimates of military expenditures--are
currently assembled, adjusted to improve
internationalcomparability,and reported on a
regular basis by the International Monetary
Fund (IMF), the Stockholm International

Determinantsof MilitaryExpenditurein DevelopingCountries

Peace ResearchInstitute(SIPRI),and the U.S.

Arms Control and Disarmament Agency
(ACDA). (See Appendixto this publication,
Pp. 147-51.1) The ready availabilityof this
information, and growing confidence in its
validity, has spurred the growth of scholarly
attention to analysis of the securitydevelopmentrelationshipin the Third World;
it has contributedimportantlyto the research
emphasison cross-countryanalysisand to the
trend toward adoption of econometric and
other quantitativemethods.
Reliability of Expenditure Data
Accuracy of Observations. The scale and
persistence of the international effort to
assembleinformationon militaryexpenditures
by a large number of countries and to report
the data on a comparable basis across
countriesand over time has been only partially
reassuring to users of the data. Skepticism
with respect to the reliability of the data is
based in large part on the fact that, regardless
of the efforts made by internationalreporting
agencies,the primary sources for information
on defense spending are the national
governments--andgovernmentsare believedto
have a variety of politicaland securitymotives
for under-reporting defense spending and to
employ a number of mechanismsto disguise
the level of their security expenditures.
Investigation of the mechanisms used by
governmentsto obscuretheir securityactivities
and other internalevidenceof inaccuraciesin
reporting has nourished the attitudes of
skepticismamongresearchersin this field but
has gained little headway in estimating the
magnitudeof the reporting errors.2
Compositionof Securiy Expenditures. A
second source of serious concern about the
reliabilityof availableinformationon military
expendituresby many countries is the highly
aggregatednature of the reported data. For
many countries, the international reporting
agencies provide no more than a single figure

West

for defense expenditures in a given year.

More detailedinformationon the composition
of military expendituresis availablefor only a
limited number of countries, and reporting
categories frequently change from year to
year.
This weakness has been the particular
target of the U.N. Department of
Disarmament Affairs, and of a sequence of
Expert Groups which, since 1975, have guided
United Nations efforts to induce member
governments to publish a unified reporting
schedule of security expenditures data in a
disaggregatedform.' Adoption of the unified
security expenditure reporting system by
member governmentshas been very partial,
but there has been a gradual increase in the
reporting of disaggregated data and in the
provision of additional information on the
compositionof military expenditures.4 There
is now a sufficient volume of reporting on
military expendituresin a disaggregatedform
to support experimental efforts to compute
military expenditure price levels and to
estimate the real quantitiesof military inputs
for a substantialcross-sectionof countries, as
describedbelow.
Price Levels and Methods of Conversion.
In the absence of information about the
composition and the price levels of- military
expenditures, reliable methods are not
available to compare year-to-year changes in
real quantities represented by the defense
expenditures of a given country, nor to
convert local currency data into values
expressed in a numeraire or base currency.
Data convertedto a numeraire are needed to
aggregate and compare the various national
expenditures in different countries. This
inabilityto present reliable indicators of real
quantitiescorrespondingto reported military
expenditures that can be compared through
time and across countries may be the most
serious data problem affecting study of the
relationshipsreviewed in this paper. Gradual

115

MilitaryExpenditureand EconomicDevelopment

progress is being made to overcome this

critical problem of data reliability.
The need to introducea reliableconversion
methodology for military expenditures has
long been recognizedby scholarsas well as by
ACDA. A decade ago ACDA includedin its
annual World Military Expenditures and Arms
Transfers report a discussion of the use of
exchangerates and purchasing-powerparities
(PPPs) to make international expenditure
comparisons. It acknowledgedthat PPPs are
recognized as the most reliable means of
convertinglocal currency data into a common
denominator in order to make valid
expenditurecomparisonswithin countriesand
amongcountriesfor either a particularyear or
a series of years. Lacking an adequatedata
base of PPP conversion ratios, ACDA and
other reporting agencies have generallyused
exchangerates, sincethese have beenthe only
conversionrates availablefor most countries.
For many countries,however, exchangerates
do not accurately reflect the relative
purchasingpower of the currency. Moreover,
exchangerates do not readily adjustto varying
inflationrates in differentcountriesbut tend to
move abruptly with currency revaluationsor
devaluations. Exchangerates are particularly
unreliable indicators of purchasing power in
developing countries;their use introducesvery
serious distortionsin the estimationof "real"
quantities for purposes of international
comparisons,aggregationacross countries, or
the study of real changesover time.5
The U.N. Department of Disarmament
Affairs is also a strong advocate for the
constructionof military price indexesand the
use of purchasing-power parities for the
international comparison of military
6
expenditures.
Efforts coordinated by the
United Nationshave produced a critical mass
of disaggregatedmilitary expendituredata, as
noted above, and this has made possible the
computationof an initial sel:of military price
levels and real (PPP) military expenditure

116

estimatesfor a large numberof countries. But

expenditure estimates based on purchasingpower parities have not been widely available
for use in the research reviewedin this paper.
The distortionsintroducedby use of exchangerate converteddata in cross-countrystatistical
analyses seriously compromisethe reliability
of the reported findings.

Testing Methods
In making statistical inferences from
econometrictests, a trade-off exists between
cross-sectionalanalysis, which gives general
results with little specific applicability, and
individualcountrytime-seriesanalysis, which
produces highly specific results but little
insight into broader principles. Multivariate
investigations of the relationship between
security and developmenthave pursued both
avenues, although the cross-sectional design
and use of macrostatisticalmodels has been
predominant.
There are serious questions about the
appropriate use of econometric methods to
evaluate cross-sectionaldata in some lines of
this research, and the reliability of statistical
inferencesbased on these testing methods has
been called into question. We will encounter
these questions in evaluatingthe reliability of
research results in the review of the
determinantsof military expenditure and the
identificationof sub-groupslater in this paper.
The common criticism of the use of
econometric methods in the securitydevelopmentliterature, generallyapplicableto
both topics, is describedbriefly in this section.
Problems of Econometric
Technique
Problems of econometric technique
experienced by security and development
researchers fall into two categories: (1) the
specification of the theoreticalI regression
model, and (2) violations of t1e' Classical

Determinantsof MilitaryExpenditurein DevelopingCountries

Assumptionsof the Ordinary Least Squares

(IDLS)regression analysis.7
Model Specdfjaion. Correct specification
of a structural model involves choosing
independent variables based on theoretical
expectationsthat they are causal determinants
of the dependentvariable. Regressionanalysis
by itself is only a statisticaldevice providing
evidenceof correlation. Attemptsto uncover
relevant variables through step-wise
regressions or sequential search based on
coefficientt-testsmake the results inadmissible
as proving causation. These techniques
involve recreated regression "runs" that
increase the chance of deriving a statistically
spurious result. In order to infer economic
causation,the a priori selectionof independent
variables must be combined with rigorous
hypothesis-testingof the signs of the resulting
coefficients.
Two other consequences of poor
specification seen in the literature include
omissionof relevantvariablesand inclusionof
irrelevantones. The former causesbias in the
estimated coefficients of other independent
variables correlatedwith the omittedvariable.
Thbelatter lowers the significanceof the other
independentvariables.
The choice of functional form for the
regression equation must also be based on
available theory on an a prioi expectation
presented before the actual estimation. Most
econometric work reviewed in this paper is
linear in the variables. Deviations without
explanationmust be scrutinizedcarefully.
OLS Assumptions. Violations of the
Classical Assumptions prevent the OLS
r egression method from producing minimumvariance, linear, unbiased estimates for the
coefficients of the independent variables.
Problems of multicollinearity and heteroskedasticity crop up in individual studies
reviewedbelow, but problemsof simultaneity

West

are far more pervasive and have proved to be

a major barrier to the attainment of
econometric reliability in this field of
application.
Simultaneitybias occurs in specifications
that fail to account for feed-back effects and
dual causalitybetween independentvariables
and the dependentvariable. An independent
variable that is jointly determined with the
dependentvariableis an endogenousvariable.
If the other half of the causal relationship
(i.e., the effect of the dependent variable on
the independentvariable)is excludedfrom the
structural model specified by the researcher,
the coefficientsof all variables in the equation
are subjectto potential bias. A similar result
is obtainedwhen the interactionis includedin
the model, but it is then essentiallyeliminated
by the specification of a reduced form for
actualestimationpurposes.
In order to allow estimation of multiequation simultaneous models, regression
techniques such as Two- and Three-Stage
Least Squares (2SLS and 3SLS) are used.
Unfortunately, for a number of reasons-includingthe continuedbias of the coefficients
for smaller samples--thesetechniques do not
provideresults that are as readilyinterpretable
as OLS.
Two contentious issues involving
simultaneityare raised in the literature:(1) the
existenceof multiple channelsthrough which
military expenditures influence economic
growth rates, and (2) the endogeneity of
militaryexpenditureswith respectto economic
growth. Most researchers have agreed that
defense spendinghas a direct and significant
impacton economicgrowth. Only a few have
undertaken research acknowledging that
militaryexpendituresmay also have an impact
on other macroeconomicvariables, such as
savings, investment, and the balance of
payments--whichin turn have their own direct
relationshipwith economicgrowth.8
WA

117

Military Expenditure and Economic Development

The omission of structural equations

specifying these indirect macroeconomic
impacts of military expenditureresults in a
failure to measure the overall impact of the
military burden on growlh. Simultaneitybias
occurs when single-channel/single-equation
modelsare used, becausegrowthitself is often
jointly determinedwith these macroeconomic
variables.
Imperfect multicolliriearity (collinearity)
occurs when independentvariables, such as
bilateral aid and the military burden, are
highly correlated yet included in the same
regression equation. This results in an
increase in the variance of the coefficients,
lowering their t-statistics and making results
very sensitive to specificationerrors and the
nature of the sample.
Heteroskedasticityoften occurs in crosssectionalstudies such as those reviewedin this
paper. If values diverge widely from one
observation to the next, error term variances
will differ across the sample, and OLS will
overstate t-statistics, leading to errors in
interpretationof the signifricanceof variables.

the existenceof lagged response times in the

relationship between the independent and
dependent variables. Determiningthe nature
of these relationshipsusing cross-countrydata
averages is extremely difficult. Further
research on this topic may have to come from
standardizedtime-seriesanalysisonce general
agreement on a theoretical framework is
reached.
Appropriate Models for Explaining
Public Expenditures
A basic question concerningthe appropriateness of econometric techniques to provide
satisfactoryexplanationsof changes in public
expenditures was first raised in 1961 by
Peacock and Wiseman. Given the variations
in institutional structure and purpose over
time, they posited that econometricmodeling
techniques are inappropriate for the
investigation of the determinants of
government expenditure. The underlying
assumptionof econometricanalysisis that all
other conditioning factors can be held
constant. This assumption is violated if
significant legal, social, and political
discontinuities,called "displacementeffects,"
exist during the sampledtime period (Peacock
and Wiseman 1961).

DeterminingCausaliti. The existenceof

a causal relationship between growth and
military expenditureremains a disputedtopic.
Some analysts (Joerding 1986) have used
methods such as Granger Causality to
demonstrate that military spending is not
exogenous. Othershave estimatedsignificant
relationships for the level of development
expressedby per capitaincomebut have failed
to show the significanceof growth ratesper se
(Deger 1986). Again, if the militaryburden is
truly endogenous with respect to growth, an
equation specifyingthe determinantsof these
expendituresmust be includedin the structural
model, or a simultaneity bias will be
introduced.

One solution to this dilemma is to find

consistentgroupingsof countryor time-series
data in which the variance in institutional
factors is minimized. Statistical techniques
such as factor analysisallow the reduction of
a large number of potentially determining
factorsto a few significantexplanatoryfactors.
Once underlying structural differences have
been removed by successful sub-grouping,
econometricanalysiscan correctlyassumethat
the only factors that are in flux and therefore
influencing the dependent variable are the
chosen independentvariables.

Another obstacle to reliable econometric

estimates related to questions of causality is

Peacock and Wiseman (1979) present this

potentialsolutionto their original problem but

118

Detenninants of MilitaryExpenditurein DevelopingCountries

West

hesitateto endorseit fuly. They note that this

form of analysis only mechanistically

that turns inputs into outputs, or into the

process that determinesthe growth of public

constructs a black box. It does not actually

expenditures.

provide insightinto the socialtransformation

The Determinants of Military Expenditure

in Developing Countries
During the past decade, a large quantity of
literature has been devoted to explaining
differencesin militaryexpendituresamongthe
developing states. Various hypotheses and
findings have been proposed concerning the
determinantsof militaryexpenditurebehavior,
with growing recognition of distinct patterns
of military spendingbehavior with regard to:
replacementof weapon stocks that have been
depleted during armed conflict; acquisition
patterns intrinsic in the replacementof major
wfeaponsystems; and resources applied to the
development, routine maintenance, and
operation of military institutions. Empirical
tests of the various hypothesesconcerningthe
determinants of military spending behavior
have focused on the level of expendituresfor
recurrent operationsand routine maintenance.
This part of the paper will review the
research, key questions, and empirical
evidenceconcerningthe determinantsof military expenditure in Third World countries.
T'he discussionhas been divided accordingto
five different explanatory hypotheses:
(1) geostrategicconsiderations;(2) budgetary
politics; (3) the influence of arms suppliers
and domestic arms production; (4) financial
and economic factors/constraints; and
(5) multi-variable explanations of military
expenditure.

GeostrategicConsiderationsand the
SecurityEnvironment
This literature postulates the existence of
linkagesbetweenand among nationalsecurity
and threat perceptions (both internal and
external),defenseexpenditures,and economic
development.
These relationships are
complex, and the difficultiesof analysis are
furthercomplicatedby the probableabsenceof
unidirectionalcausality, which is to say that
there may be feedback and reverse effects
(Deger and West 1987).
Both internal and external threats (or the
perception of suchthreats) are hypothesizedto
conduce to larger defense expenditures.
Defense spending, in turn, may influence
economic growt
either positively or
negatively.
McKinley contends that a variety of
studiesare limitedby their adherenceto purely
system-level perspectives or simple ideoThe
logical/political/economic interests.
author argues that interstate conflictis . .
product of a comparative calculation of the
rewards and costs accruing to the variety of
different means of promoting or protecting
some interest" (McKinley1989, p. 84). This
leads to the contentionby many authors that,

*M1m
119

Military Expenditureand EconomicDevelopment

in order to effectivelyanalyzethe causes and

effects of military expenditurein developing
states, it is important to specify the many
different influences affecting resource
allocation.
A number of analystshave addressed the
hypothesisthat militaryspendingis determined
by geostrategicconsiderations. Severalof the
key questionsthat have been studied are:
(1) Does the existence of international
conflict/threatperceptions explain levels
of military spending? Do military
expendituresfluctuatebased on different
levels of conflictintensity?
(2) Do Third World states determine their
military expenditurelevels in responseto
their neighbors' military spending?
(3) Is military expenditure a response to
interstate conflict and if so, does the
response itself stimulate further
expenditures?
(4) Do arms race models indicatetrends that
can explain systematic variations in
military expendituresacross Third World
countriesand over time?
Arms Race Models
One empiricallytested approach used in the
literature to explain patterns of inter-country
variationin militaryexpenditureshas been the
"arms race model." In accordancewith these
models, military expendituresare influenced
by political, psychological,and international
parametersthat emphasizeeach government's
perceptionof its adversariesand the behavior
of its allies and neighbors. Although nearly
all of the armed conflictsthat have occurred
since World War II have been fought in the
developing world, much of the arms race
literature has focused principallyon the great
powers. Nonetheless,some researchhas been

120

conducted on the developing states (Holist

1977).
Arms race theory can be divided into two
broad models, namely the arms-using model
and the arms-buildingmodel. The arms-using
model describes how armaments, military
forces, or nationalresources are consumedin
armed conflict. This model addresses the
arms race question indirectly by assessing
what national resources may be required to
assure victory, national survival, stability, or
some other purposerequiring the use of arms.
The dependent variable in this model is the
surviving amount of military forces or
resources available following an armed
conflict. Moll and Luebbert (1980) criticized
this model as being too limited in merely
describing how wars are fought rather than
how they are preparedfor. The arms-building
modeladdressesthe latter issue by focusingon
the resourcesand effortthat a nationutilizesto
developand maintain its defenseforces.
The seminalwork in arms-buildingtheory
is Richardson's (1960) mathematicalmodel.
Richardsonargued that a rate of change in a
nation's armaments could be explained by
three factors: (1) external threats,
(2) economic burden and fatigue, and
(3) grievance. Richardson'smodel is based on
competitivefactors between countries and is
known as an "action-reaction"model.
The arms race literature is no longer
limited to Richardson-type "action-reaction"
models. During the 1970s,the arms-building
models became more sophisticated and
multivariate. Research also came to focus on
the analyses of arms expenditures. During
this period, researchers integrateda range of
independentvariables. Their modelsincluded
resource constraints (among others, Choucri
and North 1975; Hollist 1977; and Ferejohn
1976); budget parameters (Russett 1970;
Lucier 1979);prior militawyexpendituresas an
indicatorof.future expenditures(Lucier 1979;

Determinantsof Military Expenditurein DevelopingCountries

West

Table 1. Arms Race Model? and Military Expenditure

4
Level of Analvsisc IAnalvticalRelatinshio
IS
NS PB I SOP PYM GM RC

Authors

Pub.
Date

Time
Period

Regionb

Ostrom

1978

1955-73

US

Russett

1970

Global

Lucier

1979

1918-38

AS,EUR,NA

Choucri and North

1975

1871-1914

EUR

Wallace and Wilson

1978

1870-1914

EUR

Hollist

1977

c. 1976

AS,ME,EUR,NA

Rattinger

1975

Global

Ferejohn

1976

Global

Taiylor

1979

Global

I
X

I
X

IX
X

X
X

X
X

All modelslisted in this table are 'arms-building"modelsexcept Taylor (1979), who presentsan
"arrns-using"model.
b AS = Asia; EUR = Europe; ME = MiddleEast; NA = North America.
c IS = InternationalSystem level; NS = Nation-Statelevel; PB = Political/Bureaucraticlevel.
d SOP = Changesin budgetaryprocedures; PYM = Previousyear's militaryexpenditure;
GM = Growthin military expenditure;RC = Resourceconstraints.
a

L_

C'houcriand North 1975; Wallaceand Wilson

1978; and Rattinger1975);and GNP (Choucri
and North 1975; Wallace and Wilson 1978;
and Ferejohn 1976). In additionto economic
and fiscal variables, these authors and others
also incorporated political/bureaucraticand
organizational factors on the domestic and
internationallevel into arms-buildingmodels.
Moll and Luebbert (1980) defined four
classificationsused to analyzethe determinants
of an arms race. Three of those classifications
are illustratedin Table 1 under the "Level of
Analysis" heading. The internationalsystem
level (IS) focuses on such issues as the
presence or absenceof arms races, stabilityof
arms races, military aid, and the role of the
superpowers. The nation-state level (NS)
focuses on national attributes and behavior.

National trends in areas such as arms

expenditures,military personnel, and weapon
quantitiesare emphasized. At the politicalbureaucraticlevel (PB), emphasisis placed on
domesticbureaucraticinstitutionsand political
organizationsand their respective influences
on defensepolicy.
Table 1 lists a representative sample of
arms race modelsdevelopedduring the 1970s.
Each study is identified by the system-level
classificationas noted by Moll and Luebbert
(1980)and the analyticalrelationshipsthat are
central to the working of each model. As
Table 1 indicates,the models of the majority
of analysts utilizing the arms-building
perspective were more sophisticated and
multivariatethan were the arms-usingmodels,
as represented by Taylor (1979). For

121

MilitaryExpenditureand EconomicDevelopment

example, Choucri and North (1975) and

Wallace and Wilson (1978) utilized both the
international(IS) and national (NS) levels of
analysiswhile integratingwithin their models
three of the four analyticalrelationshipslisted
in the table. In contrastto the modelsof these
arms-buildinganalysts,Taylor's (1979) armsusing model lacks substantive empirical
support--a common weakness in arms-using
models.
Although the arms-building models are
more sophisticatedthan the arms-usingones,
many analysts have concluded that the arms
race models do not have a high degree of
explanatory power with respect to the
mobilizationof resources over time or across
countries. This may be due to the lack of
statistical categoriesmatchingthe conceptsof
the theories underlyingthe models.
There is an extensive range of literature
investigatingarms race models,of which only
a few exampleshave been discussedhere. A
more comprehensivereview of this literature
can be found in the Moll and Luebbert (1980)
study.

Threat Perception and Inter-State

Conflict
An hypothesis concerning national security
behavior and one that is linked to arms race
behavior is the proposition that inter-state
conflict or the perception of a threat is
associated with variation in military
expenditure levels among Third World
countries. In accordancewith this approach,
the military capabilitiesof neighboringstates
often are seen as both the source of and
responseto a perceivedthreat.
In a recent influential study, McKinley
(1989)conducteda cross-sectionalanalysisfor
a large number of Third Worldcountriesover
the period 1950-82. The author determined
that military expenditurerises in response to

122

an inter-stateconflictand decreasesfollowing
the cessation of the conflict. This would
appear to be the expected finding and to
confirm the common sense interpretationof
behavior in "arms-using" circumstances. It
can be interpreted as showing weaponsinventorymanagement:a surge of replacement
expenditures occurs when the inventory is
drawn down by hostilities. But McKinley's
expendituredata is not disaggregatedto show
categories of expenditure that can be
associated with "arms-building"behavior or
the routine maintenance of military
establishments. It is not clear whether
McKinley's analytical results show any
associationbetween inter-stateconflictand the
military expenditures addressed by armsbuildingmodelsof behavior.
In a similar vein to McKinley, Weede
(1986) argued that international competition
and threats to nationalsecurity lead to higher
militaryparticipationratios and larger military
outlays. Weede supportedhis conclusionwith
cross-nationalregression analysis of data for
31 developingstates during the 1970s. Again,
the lack of disaggregatedmilitary expenditure
data makes it difficult to determine whether
Weede has demonstratedanything more than
the working of inventory replacement and
material replenishmentcycles. Confirmation
of "arms-using"behavior is not trivial, but it
leaves without satisfactory explanation the
long-term trends in "arms-building'
expenditures of the recent past as well as
importantdimensionsof both inter-countryand
inter-temporalvariations in defense spending.
The literature addressing the hypothesis
that military spending is determined by
geostrategic considerations is inconclusive.
Several writers contendthat perceived threats
or other indicatorsof the security environment
can account for variance in Third World
military expenditures, but this has not been
conclusivelydemonstratedempirically. While
the arms race models have become more

Determinantsof MilitaryExpenditurein DevelopingCountries

sophisticatedand multivariate, the empirical

testing of these models shows ambiguous
results.
Many authorshave postulatedthe existence
of a correlation between geostrategicfactors
(domestic or internationalconflict and threat
perceptions) and military spending levels.
However, the literature to date has been
unable to isolate specific expendituresin such
as way as to demonstrate the relationship
between securityconsiderationsor geostrategic
eventsand their associatedsegmentof military
expenditures. Total spending and resource
allocation levels need to be disaggregatedin
order to determine the systematicinfluences
affectingmilitaryexpenditurelevels. It is also
necessaryto distinguishbetweenarms use and
non-use circumstances and how the
disaggregated military expenditures are
influencedunder both conditions.

The Politics of the Budgetary

Decision-Making Process
A number of researchers have sought to
explaindefensespendingbehaviorin terms of
bureaucratic and political processes,
emphasizingorganizationalfactors and chiefly
domestic (economicand political) sources of
influenceon the scale of defenseefforts. This
body of research begins with the proposition
that the military is a criticallyimportantactor
in domesticpolitics in the ThirdWorld. Thus
the perspectives and activities of the military
are seen to be principally domestic, and the
most importantinfluenceson the military are
hypothesizedto be domesticbureaucratic and
political interactions. They are consideredto
be bureaucraticin the sensethat the militaryis
involvedin defendingits institutionalinterests
against those of other, usually governmental,
organizations. They are considered to be
political in the sense that the military is
constantlyengagedin definingits relationships
and power to political institutionsand political
forces in the country. In this research,

West

international factors such as conflict, arms

races, or arms transfers are consideredto the
extent that they appear to influence the
domesticbureaucraticand politicalactivitiesof
the military. Thus, to understandand explain
militaryexpenditures,researchneedsto assess
the domestic political activities, bureaucratic
politics, and institutionaldevelopmentof the
military (Grindle 1987).
Increasingly,researchershave focused on
budget allocationsbecausebudgetary data are
available and quantifiable. There also has
been an assumptionthat the regime in power
has controlover governmentexpenditures,and
that therefore differences in the patterns of
budgetary allocations may be explained by
particular regime characteristics or types.
Thus, much research has focused on the
relationship between types of regimes and
levels of military spending.
A familiar intuitive proposition is that,
when in power, the militarywill allocatemore
to the defense sector--on the assumptionthat
military officers in power will follow their
corporate self-interest. Many analysts have
studiedthe spending behaviorof military and
civilian regimes and have addressed several
key issues, including:
(1) Are there systematic differences in
spending patterns between civilian and
military regimes? Do military regimes
allocate more resources to the military
than their civilian counterparts?
(2) Can regimes be distinguished by the
degree of consensus existing between
civilianelites and the military?
(3) Are there systematic differences in
budgetarytrade-offsthat can be attributed
to military versus civilianregimes?
(4) Does militaryspendingincreasefollowing
a coup d'etat?

123

Military Expenditureand EconomicDevelopment

(5) Are regimes better units of analysisthan

countries for studying expenditure
patterns?
Several studies (as shown in the fourth
column of Table 2) have examined defense
expenditures following the occurrence of a
coup d'etat. These authors have attemptedto
determine the validity of the hypothesis that
military spending is increased following a
coup d'etat. No consensus exists among
authors about post-coup military spending.
Schmitter(1973), for example,concludedthat
military coups are associatedwith changesin
military expenditures,buttthat the directionof
change is indeterminate. This is based on a
study of post-coup defense spending of 19
Latin American countries over the period
1950-70. An emphasison Latin America is
characteristic of the majority of the authors
who analyze the relationshipbetween regime
type and military allocation, as noted in
Table 2. This Latin American emphasisalso
obtains in the studies that investigate the
correlationbetweenmilitaryand socialwelfare
expenditures, listed in Table 3. The
information presented in, Tables 2 and 3 is
largely drawn from Alexanderand ElliotBerg
(1991).
In common with other investigators,Zuk
and Thompsonfound no relationshipbetween
the occurrenceof coups(shownas the military
variablein Table 2) and military expenditures
as a proportion of central government
expendituresas the dependentvariable. These
authors applied a GLS r5gression procedure
(pooling cross-sectional and time-series
observations) to data for 66 developing
countriesover the time period 1967-76. (For
each of the studies listed in Tables 2 and 3,
the size of the country sample is shown in
brackets in the "Region" column, while the
period of observationis shown in the column
"Time Period.")

124

The analyticalresults of this literature are

contradictory concerning the relationship
between regime type (civilian vs. military
regime) and the level of military spending.
These conflicting findings are the result of
differentmethodologies,definitionsof military
regime, and sources of data. An extensive
assessmentof the literature on the relationship
of military expenditure and regime type
appearsin the Berg and Berg study.
As indicated in Table 2, several analysts
have found a positive correlation between
military expendituresand regime type. The
majority, however, have concluded that no
relationship exists between regime type and
patterns of defense spending. In general,
these authors were testing the hypothesisthat
systemicdifferencesexist in spendingpatterns
attributableto regime type.
Two recent studies by Grindle(1986) and
Looney(1988a)take a fundamentallydifferent
look at the relationshipbetween regime type
and defenseexpenditures.
Looney attributes the lack of evidence
concerningthe differencebetweencivilianand
military regimes in past studies to the
problemsof measurementin standard indexes
used to represent the defense burden. His
analysis merges two areas of explanation,
economic (ability to spend) and political
(willingness to spend), and concludes that
military regimes tend to develop the military
to levels not warrantedby size of the national
economy. This is accomplished through
increased foreign borrowing, mobilizationof
foreign exchange earnings, and price
distortions that facilitate increased defense
spending.
Utilizing an aggregate cross-national
research design for 18 Latin American
countries over the period 1967-80, Grindle
found that regimes explain more as a unit of
analysis than countries do in accounting for

Detervinants of MilitaryExpenditurein DevelopingCountries

West

Table 2. Relationship between Regime Type and Defense Expenditure

Authors

Pub.
Date

Time
Period

Nordlinger
Schmitter

1970
1971

Weaver

1973
1973

Region'

Military
Variable"

LDCs (74)
LA (20)
LA (8)
LA (19)
LA (6)
LA (2)
LDCs (32)
LDCs (41)
Brazil

Influence
Influence
Influence
Influence
Presence
Influence
Coups
Presence
Presence

lThompson
Kennedy
Hayes
McKinlay
and Cohan

1973
1974
1975

1957-62
ca. 1960
1950-67
1945-70
1960-70
1961-70
1946-66
1960-70
1950-67

1976

1961-70

Global (101)

Presence

Ames and Goff

1975

1948-68

LA (17)

Influence

Tannahill
Dickson
Hill
Whynes
]Pluta

1976
1977
1979
1979
1979

1948-67
1961-70
1946-65
1959-75
1961-70

LA (10)
LA (10)
LDCs (104)
LA (10)
LA (10)

Presence
Presence
Influence
Coups
Presence

Ravenhill
Zuk and
Thompson

1980

1960-73

AF (33)

Presence

1982

1967-76

LDCs (66)

Grindle

1986

1967-80

LA (18)

Looney

1987

1961-82

Argentina

Dependent
Variable'

Findings:
Relationship

ME/GNP
Positive
ME/CGE+GNP Weak positive
ME/GNP
Positive
ME
None
ME/GDP
Negative
ME/GDP
Positive
Change in ME/CGE
Positive
ME/CGE
None
ME/CGE
None
ME/GNP and
military size
ME
ME/CGE
ME/GNP
ME/CGE
ME
ME/GDP
ME
ME and
military size
ME

Presence
ME
Presence
ME growth rate
Coups
ME/CGE
C-M
ME/CGE
Regime change ME/CGE
Regime tenure
ME/CGE
Regime change ME/CGE

None
None

None
Weak positive
Positive
Positive
Inconclusive
None
Weak positive
None
None
Inconclusive
Positive
Positive
None

LA = Latin America; AF = Africa. Numberof country observationsin parenthesis.

Influence= High military influencein the politicalprocess;Presence = Presence of military in top executive
positions;C-M = Civil-militaryrelations;Regimetenure = Duration of rule by a single regime.
c ME = Militaryexpenditure;CGE = Central governmentexpenditure.
Source: Drawn from Alexanderand ElliotBerg (1991).
m
b

variations in defense spending. Her analysis

incorporates regime change, regime tenure,
and civil-militaryrelations. In her assessment,
the budgetary process is influenced by the
power and bargaining skills of government

competitorswho strive to gain a larger share

of the budget. The strategies and resources
available to the military, which ultimately
determine its ability to extract resources for
defensepurposes, varies due to characteristics

125

Military Expenditureand EconomicDevelopment

of civil-militaryrelations. These relationsare

determined by the degree of military
institutional exclusivenessand the degree of
consensusamongcivil and militaryelites about
the role of the military. Regime change and
civil-military relations, rather than regime
type, are found to be more useful explanatory
factors for variations in defenseexpenditures.
Evidence of Budgetary Trade-offs
A number of authors have examined the
contention that military expendituresinvolve
opportunitycosts, causingreductionsin social
programs to support military outlays. While
several authors contend that a negative
correlation exists betweenmilitary and social
expenditures (see last column of Table 3),
there remains a paucity of empirical evidence
indicatingsystematictrade-offs.
An inspection of the "social welfare
expenditures" column in Table 3 shows that
education and health expenditures are often
used as indicators of a government's
commitmentto socialwelfare. Ames and Goff
(1975) compared education and defense
expenditures in sixteen Latin American
countries for the period 1948-68 and found
that both sectors experienced fluctuationsin
funding based on the level of central
government revenues.
Verner (1983)
concludedthat trade-offsare both country-and
time-specific and cannol: be generalized in
cross-country analysis. Verner (1983) and
Hess and Mullan (1988) found that both
education and military spending have strong
constituenciesthat prevent cuts in either area.
Kennedy (1974) found no significant
evidence of budgetary trade-offs between
military and non-militaryregimes in his study
of spending patterns in forty-one developing
nationsduring the 1960s. His conclusionthat
regime change rather than regime type
precipitates a reallocation of budget shares
supportsGrindle's (1986) findings.

126

Table 3 summarizes the key findings in

selected studies on the correlation between
militaryand social welfare allocations. These
studies were testing the hypothesis that
systematic budgetary trade-offs can be
attributed to regime type. As the findings
listed in Table 3 indicate,the majorityof these
studies were unable to establish a systematic
correlationthrough time or across countries.
Only one of the studies (Harris et al 1988)
found a positive relationship,meaningthat as
military expenditures were raised, education
and health expendituresalso increased. Four
of the studies found a negative relationship,
but only Deger has significant empirical
evidence. Regardingthe negativefindings of
Tannahill,Dickson,and Pluta, Berg and Berg
note that ". . . sincethe methodologyof these
[three] studies is a simple comparison of
means, the results are questionable,and the
issue remainsopen" (Berg and Berg 1991).
Many analystsapparentlybelievedthat, by
focusingon militaryexpenditures,the relative
political clout of the military could be shown
in the budgetaryprocess; they expectedto find
politicalinfluencecorrelatedwith regime type.
The determination of military expenditures
through the political-bureaucraticprocess was
thought to be a particularly clear exampleof
interest-grouppolitics and to demonstratehow
resources are mobilizedby the public sector.
There is insufficientempiricalevidenceto
confirm a systematic effect across countries
attributableto militaryregime influenceon the
process of budgetarydecision makingand the
allocation of resources. Recent research
indicatesthat defenseallocationsare probably
influencedmore by other variablesthan by the
degree of military influence in the
government. It appears that changes in
regimes over time in individual countries
provide more useful insights into the
budgetary decision-makingprocess, including
the characterof systematictrade-offs.

West

Determinantsof MilitaryExpenditurein DevelopingCountries

'rable 3. Correlation of Military and Social Welfare Expenditures

SocialWelfare Findings:Relationship

Pub.

Time

Authors

Date

Period

Regiona

Deger

1986

1967, 1973

LDCs (34)

Agriculture

Negative

'Dabelkoand
McCormick

1977

1950, 1972

LDCs (76)

Education and health

Weak negative

Lyttkens and
Vedovato

1984

1984

Expenditures to MilitaryExpenditures

(no empirical data presented; authors used production

possibility graphs to estimate opportunity costs)

Ames and Goff 1975

1948-68

LA (16)

Education

None

Kennedy

1974

1960-70

LDCs (30)

Education and health

None

Hess and
Mullan

1988

1982-83

LDCs (77)

Education

None

Peroff and
Podolak

1979

1929-74

USA

Education

None

All welfare

None

1950-62 EUR,NA (12)

Pryor
Grindle

1987

1970s-80s

LA (3)

All welfare

Mixed

Harris et al

1988

1967-83
1980

AS (12)
AF,AS (12)

Health
Education and health

Mixed
Positive

Tannahill

1976

1948-67

LA (10)

Social spending
ratio to CGE

Negative

Dickson

1977

1961-70

LA (10)

Education

Negative

Pluta

1979

1961-70

LA (10)

Education and health

Negative

" AF = Africa; AS = Asia; EUR = Europe; LA = Latin America;NA = North America.

Source: Drawn from Alexanderand Elliot Berg (1991).

There is also some evidence that budgetary

trade-offs may have different characteristics
depending on the degree of elite consensus
concerning the role of the military. In some
recent experience in Latin America, what
appears to influence the budgetary process in
ways affecting the pattern of functionalexpenditure trade-offs is not the type of
regime (in any variant of the classical

distinction between military and civilian), but

rather the strength of civil and military elite
agreement on terms of the "contract" defining
the role of the military in society and polity.
As this elite consensus wanes or increases in
strength, the relative success of the military in
mobilizing public resources varies, and the
pattern of expenditure trade-offs changes.

127

Military Expenditure
and EconomicDevelopment

External Patrons: The Influence of

Donors and Suppliers
External or internal threats can propel
developing states into the internationalarms
trade, but unless these countries have the
capabilityto produce major weapon systems
themselves,they willbe forcedto purchasethe
armamentsabroad-and indeed all developing
states to some extent must so do. If the
weapons cannot be obtained through a
concessional grant, the recipient will be
required to purchasethe arms at a cost in use
of credit or foreign-exchangeearnings that
may hinder economic growth. Deger and
West state that: "Many LDCs have adopted
an import-fedgrowth strategy (rather than an
export-led one). This iimpliesthat capital
formationand output increasedependcrucially
on intermediate imported goods such as the
latest vintage machinery. If additionaldefense
spending on foreign arms reduces the
importation of such intermediates, growth
suffers. . . The arms trade is therefore a
cruciallink betweennationalsecurity,external
threat and economicgrowth" (Degerand West
1987, p.11).
The hypothesis that military spending is
influencedby the actions and preferences of
patrons--dominantexternal suppliersof major
weapons systems and donors of military
assistance--hasintroduceda numberof related
questions that have been explored in the
literature. These include the following:
(1) Do arms suppliers exercise a dominating
influenceover the military sector activity
of Third World recipients?
(2) Are countrieswith greater arms imports
also those with higher domestic defense
spending?
(3) Are wealthier and less dependent
developing countries more prone to
develop a domestic arms industry than

128

poorer states?
What is the
relationship between domestic arms
production and the level of military
expenditure?
Some observers-for example Harkavy
(1975), Arlinghaus (1984), and Brzoska
(1987)-suggest that as a result of growing
competitionin the arms trade, the non-armsproducingnations have increased leverage in
bargaining with the suppliers, and their
dependency has declined.
Conversely,
Neuman contends that economic and other
constraintsamong developing states promote
dependency upon the suppliers. Although
Neuman argues that the superpowers have
"substantial" influence on recipients'
procurement activities and expenditures,
exercisedthrough their control of major arms
transfers, she acknowledgesthat "the evidence
is circumstantial' (Neuman 1988,p.55).
Both Neuman(1988) and Brzoska (1987)
arguethat arms transfer activitiessignificantly
influencethe militaryprocurementpracticesof
purchasers and recipients. Brzoska contends
that arms transfer practicesare " . . . largely
explainedby short-term economic conditions
and cyclicalprocurementfactorson the side of
the recipients" (Brzoska 1987, p.176).
Accordingly,as weaponsbecomeoutdatedand
economicconditionsimprove, many developing states can be expected to increase arms
imports. This cyclicalprocurementpattern is
evident in the 'boom' years of the 1970s,
followed by the relative decline in arms
transfers during the 1980s(Brzoska 1987).
The evidenceof empirical studies to date
appears to indicate the existenceof a binary
situation. Externalpatrons can and do have a
dominant influence over military-sector
activity, and military expenditures, in a few
developingcountries. But significantexternal
patron influence on the level of defense
spending cannot be detected in many other
countries. The evidencedoes not support the

Degterminants
of MilitaryExpenditurein DevelopingCountries

West

contentionthat this kind of external influence

can be described as having a systematiceffect
across all developingcountries. In those cases
where determinant external influence exists,
there is evidence that the mechanics of the
recipients' decision-makingprocess may be
altered by the introduction of special
institutional arrangements (such as extraordinary budgets or special accounting
procedures). The purpose of these arrangements is to implementthe patron's influence
over the recipient's budgetary behavior and
change the composition of government
spending. Further study of these cases is
needed to identify reliable indicators of the
degree of influencethat a supplier is able to
exercise over a recipient's planning and
budgetarydecision-makingprocess.

producers based on whether a country was

capableof producinga major weapon system.
Looney concluded that, for non-producers,
arms imports were closely and positively
correlated with overall imports and inversely
related to total military expenditures. In
contrast, arms producers tended to expand
arms imports in association with overall
increases in military expenditures, with
reductions taking place during periods when
total imports increased. Further, unlike nonproducers, these countries increased arms
importswith overall expansionof government
consumption relative to GDP.
Looney
concluded that it appears from this analysis
that arms producers do not have to make
major sacrificesin socioeconomicexpenditures
to achievea desired level of securityinputs.

Brzoska (1987) found a correlation

between national income growth and major
weaponsimports. Countrieswith a per capita
GNP of less than $440 substitute imported
arms for personnel expenditures, while
countries at higher GNP levels produce some
arrnaments domestically to substitute for
imports (Brzoska relied on data supplied by
SIPRI, the IMF, and the World Bank for his
analysis).

Table 4 depicts the relationshipbetween

arms production and military expenditures.
The majority of authors describe arms
production (AP) as a positive determinant of
military expenditures(ME). Generally, these
authors view arms production as a means of
facilitating industrialization (AP > ME).
Hence, arms productionis consideredpart of
a broader developmentprogram and is not
determined simply by the level of military
expenditure.Political, military, and economic
factors interact with the relationshipbetween
arms productionand military expenditure.

Increasingdomestic arms productiondoes

noit necessarily lead to a reduction in arms
imports, as attested by the large domestic
production and importationof arms by India
and Israel. Deger and West (1987) note that
some developingstatesmay link arms imports
and exportsby utilizingcomparativeadvantage
to produce weapons for export, while relying
on the increased foreign-exchangeearnings
generatedby arms sales to import armaments
that (for technical or other reasons) their
domestic industries are poorly equipped to
produce at home.
Looney (1988b) studied 104 developing
countriesover the period 1961-81and divided
them into arms producers and arms non-

In contrastwith this view, several authors

focus on the importanceof a military power
base in facilitatingboth military expenditure
and a growth in arms production(ME> AP).
The prevailingview in this area of study
emphasizesarms production as a product of
the struggle toward self-sufficiencyand as a
determinant of military expenditure.
However, in the view of a majority of
investigators, causation is not unilateral but
entails feed-backs and interaction with other
factors.

129

Military Expenditureand EconomicDevelopment

_XSOMI~

--------

Table 4. Relationship BetweenRegime Arms Production and Military Expenditure

rune

Findings: Determinants
of Arms Productionb

Authors

Pub.
Date

Period

Region' AP>MEb

Ayres

1983

c. 1977

Turkey

Ball

1986

1969-85

LDCs

1983
1989

1970-81
1968-88

Global
LDCs

+
+

Domestic economicfactors
Drive for independence

1986
1987
1986

1950-84
1981-85
c.1983

LDCs
LDCs
LDCs

+
+
+

Evans

1986

1960-85

LDCs

Frank

1980

1973-79

LDCs

Goulet

1976

c. 1975

LDCs

1984
and 1986

1963-80

LDCs

Looney

1989

1973-88

LDCs

Looney
Looney

1988b
1986

1969-81
1979-80

LDCs
LA

+
+

Moodie
Neuman

1979
1984

1965-75
1979-80

LDCs
LDCs

+
+

Peleg
Rosh
Rosh

1980
1988
1990

1950-77
1969-78
c.1982

LDCs
LDCs
LDCs

+
+
+

Ross

1987

1973-83

LDCs

Global, regional relations

Economicsecurity,independe
Threat perceptions, regional
perception
Cyclical relation; ME fuels
market for AP, which fuels
further ME
Crises, oil income facilitates
ME
Desire for development,
independence
Drive for autonomycoupled
with military power bases in
LDCs
Economic factors over threat
factors; decline in AP fuels
decline in ME
ME through AP earnings
Domesticpolitical/bureaucratic
influence
Drive for independence
Economiesof scale over ME
influence
ME influenceweak
Security concerns
Political, economic, military
mobility
Exaggeratedconcem;AP minor

Wulf

1983

c.1980

LDCs

Ball and
Leitenberg
Brzoska
Brzoska and
Ohlson
Clare
Deger

Katz

ME>Apb
+

Cyclical relation, rise in

dependency
Political, economic factors;
threat perception

activity

Economic, political, military

motives; drive for selfsufficiency

LDCs

Less developedcountries; LA = Latin America.

b AP = Arms production; ME = Militaryexpenditure. AP>ME = Arms production results in growth of

military expenditure. ME >AP = Militaryexpenditurefacilitatesa growth in arms production.

130

Determinantsof MilitaryExpenditurein DevelopingCountries

The Influenceof Financialand

EconomicConstraints
Recent research indicates that economic
variables may provide further insight into the
underlying causes of Third World defense
expenditures. Early studies of this subject
focused on the role of economic factors as
determinants of the size and compositionof
government expendituresas a whole. Many
investigators have adopted a version of
Wagner's law, which asserts that the relative
size of the public sector in the national
economyhas an inherent tendency to grow as
per capitaincome increases. One postulateis
that the increasingly costly requirements of
nationalsecurityand defenseare a contributing
ifactorto the growth of the public sector. By
this reasoning, growth in the resources
required for national security is driven by
social conditions (such as increasing
urbanization) associated with progressively
higher levels of economicdevelopment.
Variants of the Wagner's law hypothesis
are frequently found in the current literature.
For example,Deger states: "The three major
determinantsof the defenseburden are . . .:
(a) long-term developmentalfactors such as
per capita income; (b) the total budget
constraint; (c) displacementvariables such as
wars or structural shifts like an oil price rise
for oil-exportingcountries. It must also be
stressed that these three determinantsaffect
military expenditureas a public good which
produces security" (Deger 1980, p.63).
The linkages among economic factors,
regime characteristics,and other influenceson
military expenditures have been explored in
the context of a budgetary decision-making
model by several authors. Some of the key
questions that have been addressed by these
scholars are:
(1) Does the burden (or share) of military
expendituresrise with increasedper capita

West

income, as postulated by Wagner's

law?
(2) Are military expenditures in all Third
World countriesinfluencedin similarways
by economicand politicalfactors,or is the
set of factors influencing military
expenditures different in sub-groups of
developingstates?
(3) Do resource-constrainednations tend to
have a differentrelationshipthan resourceabundant countries between defense
spendingand economicgrowth?
A variety of different economicindicators
have been asserted to influence military
expenditure levels and trends. Several of
these indicators can be interpreted as
corresponding to variables postulated by
Wagner'slaw to be determinantsof the size of
the public sector. Lotz (1970), for example,
conducteda cross-countrymultipleregression
analysisof 37 developingcountries. Utilizing
military expenditureas a percentage of GNP
(the military burden) as the dependent
variable, he found a significant negative
relationshipwith per capita income measured
in U.S. dollars, and a significant positive
relationshipwith the total governmentbudget
as a percentage of GNP (an independent
variable that may be taken to measurerelative
resource constraint). Lotz also found a
significant positive relationship with the
proportionof urban dwellers--anotherpossible
indicatorof the level of development.
Contrary to Lotz's findings, Maizels and
Nissanke (1986) concluded there is no
significant relationship between per capita
income levels and military expenditure/GDP
ratios. They also determinedthat the level of
urbanizationis not significantlyrelated to the
level of military spending. The most
significantvariable was found to be the share
of the central government budget in GDP.

131

Military Expenditureand EconomicDevelopment

The growth of foreign-exchangeavailability

was an importantconstraint.
Looney and Frederiksen (1988) examined
the linkage between the effect and timing of
certain economic variables and the effect of
past budgetson current defensebudgets. The
regressionanalysis,conductedon an individual
case-studybasis for ten relativelyeconomically
homogeneous Latin American countries,
indicatedthe significanceof fiscal variablesin
accounting for fluctuation in military
expenditures. A tested lag effect was also
significant, indicating that changes in
expenditures or revenues affect the military
budget over time. The results also suggested
that the large regional powers might have a
somewhatdifferent set of fiscal linkages than
smaller countries.
Looney (1986) examinedthe influenceof
external debt on military expendituresamong
a group of 61 developingstates. The author
concluded that:
a) per capita military
expenditure tends to increase in association
with increasesin per capita income, a finding
in support of Wagner's law; b) public external
debt is a significant factor in expanding
military expenditures per capita; and
c) regional differencesin military expenditures
are not as pronounced as the differences
between resource-constrainedand resourceunconstrainedstates.
Looney (1988b) examinedthe effect that
indigenousarms-productioncapabilityhas on
determiningthe level of miilitaryexpenditures
for a sample of 104 developingstates during
the period 1961-81. The results showedthat
differences in military expenditures can be
explained by the level of GDP, the currentaccount balance, and external debt--with
distinct differences between arms producers
and non-producers.
Looney (1989b)also found that the level of
military spending in develcping countries is

132

determined in larger part by economic

constraints than by external threat factors.
This is qualified by Looney's earlier
contention (1988b) that arms producers are
less vulnerable to external factors than nonproducers.
Table 5 summarizes the relationship
between economic/fiscal indicators and
military expendituresas found in a number of
recent studies. The testing by these authors
has proceeded on the hypothesis that the
changes in these economicfactors over time
result in a displacement effect on military
spending. As can be seen in Table 5, the
results of these analyses are not entirely
consistent. This is exemplified by the
previously noted studies by Lotz (1970) and
Maizels and Nissanke (1986). Both studies
used military expenditureas a percentage of
GNP as the dependentvariable and per capita
income as one of the independentvariables.
Lotz found a significantnegative relationship
betweenthe two variables, while Maizels and
Nissanke concludedthat the relationship was
not significantlydifferent from zero.
A wide variety of other economic and
political variables have been examinedin the
literature. While a clear consensushas not
emerged and the need for further research is
evident, some observationsabout the effect of
economic factors on military expenditures
appear to be supported by a broad range of
empirical studies. For one, the proposition
that effects on the level of military
expenditures are attributable to many
influences--politicaland economic, domestic
and international--hasbroad support in the
literature.
Few analysts would now dissent from
Looney's conclusion: "Most importantly,the
analysis indicatesthe usefulnessof examining
the defense burden from an economic
perspective. Despite the wide diversity of
political and strategicsituations in the sample

Determinantsof MilitaryExpenditurein DevelopingCountries

West

Table 5. Relationship Between Economic Indicators and Military Expenditure

Pub.

Time

Independent

Authors

Date

Period

Variables

Lotz

1970 1960-65 Per capitaincome

Urban population
CGE' as % of GNP

Maizels and
Nissanke

Dependent Findings:Relations
Variables'

1986 1978-80 Per capita income

Urban population
For. exch.bconstrained
Harris
1986 1960-80 Inflation
Balanceof payments
Governmentrevenue
O'leary and
O'Leary and
Coplin
1975
GDP
Rival budget levels
Rival arms purchases
Looney and
Frederikson 1988 varied
GDP
CGEa
]Looney
1986 1970-82 Per capitaincome
Public debt
For. exch.bconstrained
For. exch.bunconstrained

Among Variables

ME/GNP
ME/GNP
ME/GNP

Negative
Positive
Positive

ME/GDP
ME/GDP
ME
ME
ME
ME

Not significant
Not significant
Negative
Weak negative
Positive
Positive

ME
ME
ME

Not significant
Positive
Positive

ME
Positive
ME
Positive
per capita ME Positive
per capita ME Positive
ME
Negative
ME
Positive

CGE = Centralgovernmentexpenditures.
For. exch. = Foreignexchangeavailability.
c ME = Militaryexpenditures.
b

of developing countries, economic variables

were shown to account for the bulk of
differencesin per capita military expenditures
across countries"(Looney 1986, p.29).
The extent of domestic and external
resource constraints, as well as the country's
domestic arms productioncapability,appearto
alter the structural relations between other
causal factors and military expenditurelevels.
More generally, the developmental
homogeneityof Third World statescannot be
assumed. Cross-sectionalstudies that have

pooled the experience of countries with

extremely diverse economic and political
conditionsand incorporateda variety of time
periods and testing models have had
indifferent success in identifying the
determinants of inter-country variations in
defense spending. Subdividing countries
according to similarities in economic
characteristicsreveals significantdifferencesin
the relative importanceof various influences
on military expenditures. We will return to
the identificationof significantsub-groupings
in part III below.
is._____133
e .

.... .

133

Military Expenditureand Ecojomic Development

Multi-VariableExplanationsof
Military Expendituri,
Efforts to unravel the determinantsof military
spending using econometric analysis have
centeredon attemptsto verify the corollary of
Wagner's Law for defense spending: that
there is a positive relationship between
economic growth and th,e military burden.
Early uses of multivariate regression
techniquesin this vein were framedwithinthe
context of more general studies of the
determinantsof all public expenditures.
Table 6 summarizes the econometric
analysesof five authorswho have exploredthe
hypothesis that a positive relationshipexists
between economic growth and the military
burden. Each econometric model utilizes a
multivariateregressiontechniqueand employs
several explanatory variables from the four
major classificationslisted in the table.
Lotz, for example, used data for 37
developing countries and employed factor
analysisto determinethe relationshipbetween
the compositionof governrnentspending and
different dimensions of economic and social
structure. Lotz's regressionanalysisestimated
significantpartial correlation coefficientsfor
mineral and oil exports andlsize of the public
sector (both indicators of budget/financial
constraints in Table 6), and for per capita
income and urbanization (indicators of
economicdevelopmentlevel), with respect to
the militaryburden. As shown in Table 6, the
variables explained 37 per cent of the intercountry variation in the militaryburden. Per
capita income level showeda highly negative
coefficient. Lotz concluded that modern
military establishments have technical
economies of scale, and that lower-income
countries must often spend more than they
otherwisewould in order to keepup with their
wealthier, expansionistneighbors.

1_3

134

Lotz's effort suffered from a paucityof a

priori theorizing and the limitations of the
variableshe had availablefor use. His failure
to include foreign-exchange/external-resource
constraintsor internal/externalthreat variables
make his specificationsuspect. Lotz also did
not address the question of the possible
simultaneitybetweenmilitary expendituresand
economicgrowth. As a result, Lotz's results
are questionable.
Tait and Heller, in a broad IMF analysis
of government expenditure in 84 developed
and developing countries, regress per capita
income, urban population share, the public
sector size (net of defense), and population
share under fourteen years of age on the
military expenditures/GDPratio. Explaining
only 15 per cent of the inter-countryvariation
(shownby the r2 of Table 6), Tait and Heller's
estimation fails to return a significant
coefficientfor per capitaincome,but it closely
mirrors Lotz's results for the urbanizationand
governmentsize variables. Constructed in a
snimlarfashion and containingalmostthe same
variables, the work of Tait and Heller falls
victim to the same theoretical and
methodologicalflaws encountered in Lotz's
analysis.
In an update of the Tait and Heller work,
Tait and Diamondreturned to essentiallythe
same analysis,using data for the period 197586. The defense expenditure estimation
explained only 8 per cent of the variation.
Tait and Diamond shed little additional light
on the determinantsof military expenditure.
Maizels and Nissanke (1986) introduced
political, military, and external resource
factors into their analysis. The authors also
specify internal and external economic
linkages that influencemilitary expenditures.
Internally, they allow for per capita income
level, GDP growth rate, and public sector
share of GDP; externally they suggest that
growth of foreign-exchangeavailability, the

Deiterminants of Military Expenditure in Developing Countries

West

I'able 6. Determinants of the Military Burden (Military Expenditure/GNP)

(values of regression coefficients)

Lotz
(1970)
Economic Development Level
Per capita income level
Open economy inverse index

Urban populationshare
Urban populationgrowth rate

-0.006**

Tait and
Heller
(1982)

n.s.

Tait and Maizelsand

Diamond Nissanke Deger
(1990)
(1986)
(1986)

n.s.

n.s.

0.15**
-0.22*

0.048**

0.05*

0.081*
0.020*

0.10**

0.028*
0.33*

Budget/Financial Constraints
Public sector size CGE/GNP
Mineral export share
Growth of foreign exchange
GDP growth rate

0.21**

0.15**

2.79**
n.s.

Political/Military Influences

War dummy
Oil-countrydummy
Regime-type/violencescore
Arms-supplierconcentration

2.43**
0.65**
0.63**

Other Structural Factors

Populationshare under 14 years
Total population
Ratio of FDI to capital stock
Concentrationof FDI investors
R2

11.35**
3.98**

0.16**

0.098**
n.s.
-1.15**
n.s.

0.37
37

0.15
84

0.08
200

0.65
72

0.87
50

Note: Single and doubleasterisks indicate,respectively,significantat 95% and 99% level (one-sided);
n.s. = not significantat 95% level.

ratio of foreign direct investment to total

capitalstock, and the concentrationof foreign
investmentshould all have explanatorypower
with respect to the military burden.
Maizels and Nissanke carried out their
regression analysiswith military expenditure

data averages for 1978-80for 72 developing

countries. Table 6 shows that both per capita
income and the GDP growth rate fail to have
any significant explanatory power in the
equation. (Both variables are listed in the
table as n.s., or not significantstatistically).
Accordingto Maizelsand Nissanke,the most

135

MilitaryExpenditureandEconomicDevelopment

powerful explanatory variable is the public

sector burden. The incorporationof political
and military factors and external economic
influences significantly improves the overall
explanatorypower of their hypotheses.
The negative, significant impact on the
defense burden of the change in foreigners'
share of the total capital stock has been
interpreted by Gyimah-Brempong(1987) to
show that military expenditure is not
exogenous. Since military expenditureboth
impacts on and is affected by the change in
capital stock, or investment, Maizels and
Nissanke's results appear to suffer from
simultaneitybias.
Joerding (1986) provided additional
evidence of the endogeneity of military
expenditures by conducting a Granger
CausalityTest on data for the 1960sand 1970s
from 57 developing countries. A variable is
said to be directly "Granger-caused' by
another variable if the original variable is
better predictedby using laggedvaluesof both
variables than by lagged values of only the
original variable. Joerding concluded that
economicgrowth does Granger-causemilitary
spending. Since Granger non-causalityis a
necessary condition for considering military
expenditures to be exogenous, Joerding
asserted that he had demonstrated the
endogeneity of military expenditures.
However, Joerding found no evidence that
militaryexpendituresGranger-causeeconomic
growth. While this is not sufficientevidence
to demonstrate the exogeneity of economic
growth (in light of contradictory evidence
presented in other studies), it introduces
uncertainty about the reliabilityof Joerding's
test of the exogeneityof militaryexpenditures.
Deger (1986) carried out a series of
regressions,using data averagedover 1965-73
for 50 developing countries, to test the sign
and significance of per capita income as

136

different explanatory variables are added.

Table 6 shows that Deger found per capita
income to be positive and significant,
confirmingher Wagnerianexpectationthat, as
incomes rise, the share in GDP of a public
good such as military expenditures should
increase.
Deger reportedthat experimentswith other
domestic variables, including growth rates,
showedno significantexplanatoryrelationship
with inter-countryvariations in the military
burden. She concluded that Wagner's Law
cannotbe validatedempiricallywith respectto
economic growth, but only with respect to
levels of economicdevelopment.
Deger also found positive coefficientsfor
two other variables explaining variations in
defensespending:the indicatorsof government
sector size and the total population. Public
sector size may be interpretedhere, as in the
other studies, as an inverse indicator of
relative domestic resource constraint. The
population coefficient is borderline
insignificant.Deger does not addressthe issue
that adding the population variable when per
capita income has already been includedmay
be unnecessaryand introducecollinearityinto
the equation. The questionof the simultaneity
of the public sector size and the military
burden is also ignoredby Deger.
Deger included country-specificdummy
variables for oil exportersand war economies
as measuresof structural displacement. Both
turnedout positive and significant. A final
variable measuredthe degreeof integrationof
a nationaleconomyintothe world economyby
takingthe differenceof incomeper capitaboth
at the official exchange rate and at the
purchasing-powerparity rate. Deger argues
that this open economymeasureis just another
index of development,although she does not
explain why she chose to include this measure
rather than other developmentindexes.

Determinantsof MilitaryExpenditurein DevelopingCowntrs

Although Deger's estimation explains a

substantial 87 per cent of the inter-country
variancein militaryburdens, an assessmentof
her work must point out the absence of an
internal threat variable or a measure of
externalresourceavailabilityother than the oil
clummy. If higher incomegrowthrates do not
explain the variations in the military share of
GIDP,what about inflowsof externalcapitalor
loreign-exchangeconstraints, as modeled by
Maizelsand Nissanke?
Considerable differences remain among
Enalystsconcerninginterpretationof the effects
of economicgrowth and per capita incomeon
the military burden.
Conclusive crosssectionalevidenceof economicgrowthimpacts
on the share of military expenditurein GDP

Wat

have not been found.

Nonetheless, a
relationshipbetween the military burden and
the level of development,representedby per
capita incomeor another indicator,does seem
to emerge. In sum, the results neither fully
confirm nor disprove the Wagner's Law
propositions. This should not be unexpected,
as demonstratingthe corollary of Wagner's
Law with respect to economicgrowth is a
more appropriatetask for time-seriesanalysis
than for cross-sectionalanalysis. Thus far, no
empirically substantive evidence has been
found to counter the intuitive notion that, as
the process of economic growth proceeds,
more resources are freed for the purposes of
public expenditures, including national
defense.

Demonstrating DifferencesAmong Sub-groups

of DevelopingCountries
]Despite improvements in the econometric
methods of research concerning the security
and developmentrelationship, there remains
the question of the appropriatenessof using
econometric techniques to analyze the
determinants of public expenditure. This
difficulty, discussed in an individualcountry
time-seriesframework earlier (in the section
on reliabilityof testing methods), applies to
The
cross-sectional analysis as well.
iinstitutional discontinuities over time are
analogous to the diverse functional and
iinstitutionalcharacteristicsof the countriesin
a broad cross-sectionalsample. If the implicit
segregation of developed from developing
countries is legitimate and significant, what
other sub-groups should be segregated?
Looney (1988c) points out the difficulty of
drawing useful generalizationsfrom studies in
which the samples may include countries as
diverse as Chad and Saudi Arabia.

The multivariateresearch reviewed in the

previous section has often used geographic
regions (Latin America, Africa, and Asia) as
a convenientway of choosingsub-samplesfor
analysis. Unfortunately, the political and
economicdiversitypresent across such broad
regions, as well as the small sample size for
some regions, frequentlyresults in coefficient
estimates of little significance or practical
application(e.g., Lim 1983, Deger and Smith
1983, and Faini et al 1984). It should also be
noted that basing sub-groups on continents
provides little opportunityfor transferbetween
sub-groups, as underlying characteristics of
individualcountrieschangeover time.
Other bases for sub-groupsthat have been
suggested include special sub-region (e.g.,
FrancophoneAfrica), political bloc, war or
peace economy,size of military expenditures,
physical quality of life index, or regime type

137

MilitaryExpenditureand EconomicDevelopment

(Brauer 1988,pp. 29-33). Analysisof regime

type differences has received the most
attention, but as pointeclout in part II above,
there is no decisive evidence that military
spending is correlated with the civilian/
military regime distinction.
Two sub-groups--one based on the
existence of domestic defense industries and
the other on the availability of financial
resources-are worth reviewingfor the salient
insights they offer into the relationship
between security and development in
developingcountries. Robert Looney's work
in both areas is explored because of his
innovative use of statistical techniques in
segregatingcountriesas well as the interesting
results he has obtained.

Differences Attribuitable to Domestic

Defense Industries
For his analysis of the domestic defense
industry sub-group,Looney (1986, pp.31-49)
follows Neuman's (1984b) classification of
developing countries on the basis of the
presence of one indigenouslyproduced major
weapon system. As shown in Table 7,
Looney finds that internal economic factors
accountfor military spendingbehavior among
arms producers. The military's budgetary
allocation in non-producing nations, on the
other hand, is importantly influenced by
external factors. Non-producers, Looney
maintains, do not face the same internal
pressures as do arms-producing nations to
continuedefensespendingin times of minimal
external threat. Thus, there is a kind of
artificial demand stimulation or "military
Keynesianism" at work in arms-producing
nations.
Contrary to assertionsby other analystsof
defenseindustriesin the T'hirdWorld, Looney
does not find economicsize, a thresholdlevel
of per capita income, or an industrialbase to
be necessary or sufficient preconditions for

138

domestic arms production. Instead, his

analysis shows that, due to continued
technologicaldependence on the developed
world, the economic environment, in
particularforeign-exchangeavailability,is the
key determinant. Looney differentiates
between Latin American and non-Latin
American arms producers. Latin American
arms productiondevelopedin the early 1960s,
as export growth and external borrowing
created a large import capacity. The
establishmentof indigenousarms industriesin
other developing countries seems to be
independent of trade performance and to
dependprimarilyon current-accountfinancing
through publicly guaranteed loans (Looney
1987; 1988).
Looney's emphasis on a country's
economic environmentas the primary factor
influencing the installation of domestic
defense-productioncapacity arises from his
efforts to predict the classifications of
countries as producers or non-producers,
based on a discriminant analysis using
socioeconomicvariables. Political variables
proved to add little to the differentiation
between producers and non-producers.
Looney's analysis demonstrates that arms
producers have greater access to financial
resources, particularly foreign exchange.
Therefore, they can afford to finance their
domestic defense industries through external
debt accumulation. Unlike their resourceconstrainednon-producingcounterparts, who
lack debt servicecapacity,arms producerscan
maintain arms production without diverting
domestic resources from productive civil
sector investment.
The arms productionbasis for identifying
sub-groups,whileprovidinguseful insights, is
one-dimensional and static. The use of
discriminant analysis does not determine the
groupings but only uncovers the best
predictors of a de facto segregation of
countries. Given Looney's conclusionthat it

Determinantsof Military Expenditurein DevelopingCountries

West

Table 7. Determinantsof Military Expenditureby Sub-Groupsof Countries

Sub-Group

GNP

Debt

Arms Production

Balance of Payments

Mobilization

(trade deficit)

(externalthreat)

Producers

n.s.

n.s.

Non-Producers

n.s.

(reserves)

(governmentdeficit)

n.s.
+

Resources
Unconstrained
Constrained

n.s.
n.s.

Note: n.s. = not significant.

Source: Looney, 1988c.

is unlikelythat additionaldevelopingcountries
will initiate defense industriesbecause of the
existing debt problem, this categorizationof
countries is also unlikely to mirror changing
economic circumstances among developing
countries. A typology based on economic
variables is likely to be a much better
candidatefor effectivelyseparatingdeveloping
countries into fluid groups with different
functionalcharacteristics.
As noted earlier (in the section on the
reliability
of testing methods), a
methodologicallypreferable solution to the
problem of identifyingcountry sub-groups is
the use of techniques(such as factor analysis)
that do not force the researchers'preconceived
notions of country classification onto the
analysis. Instead, the data is allowedto sort
itself out and define the important
characteristics (factors) that distinguish
With the
significant sub-groups.
discontinuities eliminated, econometric
analysiscan be conducted.

Differences Attributable to Financial

Constraints
This section describes Frederiksen and
Looney's (1982; 1983; 1985) and Looney's
(1986, pp. 3-30; 1990) methodology for
grouping countries on the basis of financial
capabilities. It summarizestheir findings on
the differences between the two groups and
concludeswith an appraisalof the contribution
made by this body of work to understanding
and modeling the security and development
relationshipin developingcountries.
Looney and Frederiksen first argued in
1982 that a fiscal crunch in a resourceconstrainednation causesdevelopmentprojects
to be sacrificed in order to maintain military
expenditures. Theypredictedthat, in a similar
situation,the relativelyless constrainednations
would utilizetheir accessto foreign creditsto
maintain development programming and
expand military expenditures. This behavior
was confirmedin their 1982, 1983, and 1985

139

Military Expenditureand EconomicDevelopment

Table 8. Summaryof Mean Values of DiscriminantAnalysisVariables

Variables

Richto Poor
Ratio

Inflow of public loans as % of exports, 1982

External public debt, 1982
Gross internationalreserves, 1982
External public debt as % of GDP, 1982
Averageannualgrowth in imports, 1970-82
Debt service as % of exports, 1982
External public debt as % of GDP, 1970

1.0 to 3.6
4.5 to 1.0
10.5 to 1.0
1.0 to 2.3
8.7 to 1.0
1.0 to 1.2
1.0 to 2.0

Source: Looney (1986),p.11.

studies. Although the military expenditure

impacton growth in the constrainedcountries
was insignificantin two of the three studies,
the unconstrainedcountries were in all three
cases foundto benefit positivelyfrom military
expenditures.
The early studies used cluster analysis
based on variables representing savings,
investment,foreign-exchangeearnings,import
elasticity, and productivity of investmentto
group countries for the 1965-73 period.
Discriminant analysis then tested the
probabilityof havingcorrectly classifiedeach
country. Finally, single equation OLS
regressions produced estimatesof the impact
of military expenditures on growth.
Unfortunately, in their initial estimates, a
complete data set was available for only 9
countries in the resource-richgroup and 24 in
the resource-poorgroup.
Later work by Looney assembleda large
number of economicand financial variables
for over 60 developing countries during the
1970-82period (Looney1986, pp.3-30). This
research demonstrated a greater statistical
sophistication, employing factor analysis to
select the variables for discriminantanalysis.

140

Looney found seven variables contribute

significantlyto dividing his sample into two
groups, based on their total access to foreign
resources. The results of the discriminant
analysisare shown in Table 8. The relatively
constrained ("poor") group included most of
the African and the poorer Latin American
countries. The "rich" group was made up
largely of Middle-Easternand North African
countries as well as the oil-exporting and
newly industrializingcountriesof Asia, Africa,
and Latin America.
Looking back to Table 7, we may note
that two-thirdsof the arms producers are also
in the resource-richcategory. With groupings
based on access to foreign exchange, external
threat factors are found to be not significant.
This leaves internal economicvariables such
as the government deficit and external public
debt as the significantdeterminantsof military
spending.
Military expenditures per capita in the
larger, more debt-free, financially unconstrained nations show a negative correlation
with externalpublic debt and the government
deficit. Looneybelievesthat this demonstrates
their reluctanceto let militaryspendingexceed

Determinantsof MilitaryExpenditurein DevelopingCownries

their means and to jeopardize their overall

creditworthiness. Meanwhile, among the
more constrained countries, there appears to
be an association among expenditures on
defense,foreign-exchangeshortages,and high
levels of externalindebtedness.
Looney's work has consistentlyaddressed
the crucial question of how to let crosssectionaldata identifysub-groupsof countries,
each with a distinctive set of securitydevelopment relationships. The varied
techniques used by Looney and Frederiksen
have served to disaggregate developing
countries into sub-groups over four different
time periods since the 1950s. One important
result of this researchhas been to demonstrate
that differences in techniques and in time
periods used in the analysis reveal shifts in
country group affiliationsover time. Fifteen

West

of the fifty countries in Looney's factor

analysisfor 1970-82switchedgroupingsin his
latest analysis,which used data for the period
1965-87.
This indicates the need for periodic
reapplication of the best available sorting
procedure in order to make available timely
and appropriate sub-groupings for policy
analysis. As the characteristics underlying
country differences in the securitydevelopmentrelationshipapparentlyfluctuate
over time, both time-seriesand cross-sectional
work must be updated. Econometricanalyses
that are both statistically significant and
theoreticallysound,but that rely on experience
from as far back as the 1960s,may have only
limited practicalsignificancefor policy design
in the 1990s.

.:EFERENCES
Notes

Assembly Document A1411622of September 25,

1986.

1 A very useful descriptionof these sourcesand of the

military expenditures data reported by each also

appears in Ball 1988, pp. 84-87and Appendix2.)
2 For a survey of studieson this subject,see Ball 18,

pp. 111-122;see also earlier reportsby Brzoska1981

and Ball 1984.
3 See United Nations, Department of Disarmament
A,ffairs,1983; Ball 1988, pp. 97-111.
4 These trends are evident in a report on compliance

5 See U.S. Anms Control and DisarmamentAgency

1980, pp. 15-17.
6 See United Nations, Department of Disarmament
Affain, 1986.
7 For a conciseovervicw,see Studenrund and Cassidy
1987.
For cxamplesof multiple-channetestingmodels, see
Deger and Smith 1983;Dcger 1986.

with the U.N. reporting system: U.N. Geneal

141

MilitaryExpenditureand EconomicDevelopment

CitedStudies
Ames, Barry, and Ed Goff 1975. "Education and
Defense Expendituresin Latin America: 19481968," in Craig Liske et al., eds., Comparative
Public Policy: Issues, Theories and Methods
(BeverlyHills, Calif.: Sage Publications).
Arlinghaus, Bruce E. 1984. MilitaryDevelopmentin
Africa: The Politicaland EconomicRisks of Arms
Transfers(Boulder,Colo.:WestviewPress),p. 152.
Ayres, Ron 1983. "ArmnsPlroductionas a Forn of
Import-SubstitutingIndustrialization: The Turkish
Case," WorldDevelopment,11,9: 813-23.
Ball, Nicole 1986. "The Growthof ArmsProductionin
the ThirdWorld," NationalForun, 66, Fall: 24-27.
1988. Security and Economy in the Third
World (Princeton, N.J.: Princeton University
Press).
1984. "MeasuringThird World Security
Expenditure:
A Research Note,"
World
Development,12,2: 157-64.
Ball,Nicole, and MiltonLeitenberg1983. The Structure
of the Defense Industry: An InternationalSurvey
(New York: St. Martin's Press).
Berg, Alexander and Elliot 1991. "The Political
Economy of the NMilitary," in George
Psacharopoulos,ed., Essayson Poverty,Equityand
Growth (Oxford: PergamionPress for The World
Bank),pp. 203-39.
Brauer, Jurgen 1988. "Military Expenditures,Arms
Production, and the Economic Performance of
DevelopingNations." Ph.D. dissertation,Department of Economics,Notre Dame University.
Brzoska, Michael 1981. "The Reporting of Military
Expenditures," Journal of Peace Research,
4,9: 61-89.
1989. "The ImpactoF Arms Productionin the
Third World," Armed Forces and Security,
15,4: 507-30.
Brzoska, Michael 1987. "Current Trends in Arms
Transfers," in Saadet Deger and RobertWest, eds.,
Defense,SecurityandDevelopment(New York: St.
Martin's Press), pp. 161-79.
Brzoska, Michael, and Thomas Ohlson, eds. 1986.
Arms Production in the Third World (SIPRI)
(London:Taylor & Francis).
Choucri, N., and R. North 1975. Nations in Conflict:
National Growth and International Violence(San
Francisco, Calif.: Freeman).
Clare, Joseph 1987. "Whitherthe Third World Arms
Producers?"WorldMilitaryExpendituresandArms
Transfers (Washington,D.C.: U.S. Government
Printing Office),p. 23-28.
Dabelko, David, and James M. McCormick 1977.
"Opportunity Costs of Defense: Some Cross-

142

National Evidence," Journal of Peace Research,

14,2: 145-54.
Deger, Saadet 1986. Military Expenditurein Third
World Countries (London: Routledge & Kegan
Paul).
Deger, Saadet, and Ron Smith 1983. "Military
Expenditure and Growth in Less Developed
Countries," Journal of Conflict Resolution,
27,2: 335-53.
Deger, Saadet, and Robert West 1987. "Introduction,"
Defense,Securityand Development(NewYork: St.
Martin's Press), pp. 1-16.
Dickson,ThomasJr. 1977. "An EconomicOutputand
ImpactAnalysis of Civilian and Military Regimes
in Latin South America," Development and
Change, 8: 325-85.
Evans, Carol 1986. "ReappraisingThird World Arms
Production,"Survival, 28,2: 99-118.
Faini, Riccardo, Patricia Annez, and Lance Taylor
1984. "DefenseSpending,EconomicStructure,and
Growth: Evidence Among Countries and Over
Time," Economic Development and Cultural
Change,32,3: 487-98.
Ferejohn, J. 1976. "On the Effect of Aid to Nationsin
Arms Race," in Dina A. Zinnes and John V.
Gillespie, eds., Mathematical Models in
International Relations: A Perspective and a
CriticalAppraisal(NewYork: PraegerPublishers).
Fontanel, Jacques 1987. "A Note on the International
Comparisonof MilitaryExpenditures,"in Christian
Schmidt, ed., The Economics of Military
Expenditures: Military Expenditures, Economic
Growthand Fluctuations(New York: St. Martin's
Press), pp. 29-46. Proceedingsof a conference
held by the InternationalEconomicAssociationin
Paris.
Frank, Andre Gonder1980. "The Arms Economyand
Warfarein the ThirdWorld," Contemporary
Crises,
4,2: 161-93.
Frederiksen, Peter C., and Robert E. Looney 1982.
"Defense Expendituresand Economic Growth in
Developing Countries: Some Further Empirical
Evidence,"Journalof EconomicDevelopment,July:
113-25.
1983. "DefenseExpendituresand Economic
Growth in DevelopingCountries," Armed Forces
and Society, 9,4: 633-45.
1985. "Another Look at the Defense
Spendingand DevelopmentHypothesis," Defense
Analysis, September1985.
Gyimah-Brempong, Kwabena 1987. "Patterns of
MilitaryExpenditurein Africa,"Draft preparedfor
Rand Corporationand the Departmentof Defense.
Harkavy, Robert E. 1975. The Arms Trade and
International Systems (Cambridge, Mass.:
Ballinger).

Determinants of Military Expenditure in Developing Countries

Harris, G. 1986. "The Determinants of Defense

Expendituresin the ASEAN Region," The Journal
of Peace Research,23.
Harris, Geoffrey, Mark Kelly, and Pranowo 1988.
"Trade-offsBetweenDefenseand Education/Health
Expendituresin DevelopingCountries,"Journal of
Peace Research,25,2: 165-77.
Hayes, Margaret D. 1975. "Policy Consequencesof
Military Participationin Politics: An Analysisof
Tradeoffs in Brazilian Federal Expenditures," in
C. Liske, W. Loehr and M. McCamant, eds.,
ComparativePublic Policy: Issues, Theories,and
Methods (BeverlyHills: Sage), pp. 21-52.
Heller, Peter S., and Jack Diamond1990. International
Comparisons of Governmental Expenditures
Revisited: The DevelopingCountries,1975-1986,
IMF OccasionalPaper No. 69 (Washington,D.C.:
InternationalMonetaryFund).
Hess, Peter, and BrendanMullan 1988. "The Military
Burden and Public Education Expenditures in
ContemporaryDevelopingNations," TheJournalof
DevelopingNations, 22,4: 497-514.
Heston, Alan 1990. "RealWorld MilitaryExpenditures:
134 Countries, 1980," Universityof Pennsylvania
CADEDiscussionPaper, pp. 90-94.
Hill, Kim Q. 1979. "MilitaryRole v. MilitaryRule,"
ComparativePolitics, 11,3: 371-78.
Hollist, L. 1977. "Alternative Explanations of
CompetitiveArms Processes: Tests on Four Pairs
of Nations," AmericanJournalof PoliticalScience,
22: 313-40.
Joerding,Wayne 1986. "EconomicGrowthand Defense
Spending:
Granger Causality," Journal of
DevelopmentEconomics,21, April: 35-40.
Katz, James E. 1984. "UnderstandingArms Production
in Developing Countries," Arms Production in
Developing Countries: An Analysis of Decision
Making (Toronto: LexingtonBooks),pp. 3-13.
1986. The Implications of Third World
Military Industrialization: Sowing the Serpent's
Teeth (Toronto:LexingtonBooks).
Kennedy,Gavin 1974. The Militaryin the ThirdWorld
(New York: Charles Scribners& Sons).
Knavis, Irving B., Alan Heston, and Robert Summers
1982. World Product and Income: International
Comparisonsof Real Gross Product (Baltimore,
Md.: The Johns Hopkins University Press).
Prepared by the Statistical Office of the United
Nationsand the World Bank.
Lim, David 1983. "Another Look at Growth and
Defense in Less DevelopedCountries," Economic
Developmentand CulturalChange, 31,2: 377-84.
Looney, Robert E. 1986. T7hePolitical Economy of
Latin AmericanDefenseExpenditures:CaseStudies
of Venezuela and Argentina (Lexington, Mass.:
LexingtonBooks).

West

1987. "Financial Constraints on Potential

Latin American Arms Producers," Current
Researchon Peace and Violence,10,4: 159-68.
1988a.
"Socio-Economic Budgetary
Contrastsin DevelopingCountries: The Effect of
AlternativePolitical Regimes," Journal of Social,
PoliticalandMilitary Studies, 13,2: 195-218.
Looney, Robert 1988b.
Third World Military
Expenditureand Arms Production(New York: St.
Martin's Press).
1988c. "ConventionalWisdomvs. Empirical
Reality: The Case of Third World Defense
Expendituresand Anns Production." Unpublished
paper.
1989a. "MilitaryKeynesianismin the Third
World:
An Assessment of Non-Military
Motivations for Arms Production," Journal of
Politicaland MilitarySociology, 17,1: 43-63.
1989b. "Internal and External Factors in
Effecting Third World Military Expenditures."
Journalof Peace Research,26,1: 33-46.
- 1990. "Third World Defense Expenditures
and EconomicGrowth in the 1980s." Unpublished
paper.
Looney, Robert, and Peter C. Frederiksen 1986.
'Defense Expenditures,External Public Debt and
Growth in Developing Countries." Journal of
PeaceResearch,23,4.
Looney, Robert, and Peter C. Frederiksen 1987.
"Consequencesof Military and Civilian Rule in
Argentina: An Analysis of Central Government
Budgetary Trade-Offs," Comparative Political
Studies, 20, April: 34-46.
Looney, Robert E., and Peter C. Frederiksen 1988.
"Economic Determinants of Latin American
DefenseExpenditures,"Armed Forcesand Society,
14,3.
Lotz, Joergen R. 1970. "Patterns of Government
Spending in Developing Countries," The
ManchesterSchool of Economicand SocialStudies,
38,2: 119-44.
Lucier, Charles 1979. "Changesin the Valuesof Arms
Race Parameters,"Journal of ConflictResolution,
23,1: 17-41.
Lyttkens, Carl H., and Claudio Vedorato 1984.
"OpportunityCosts of Defense: A Commenton
Dabelko and McCormick," Journal of Peace
Research,21,4: 389-94.
Maizels, Alfred, and Machiko Nissanke 1986. "The
Determinants of Military Expenditures in
DevelopingCountries,"WorldDevelopment,14,9:
112540.
McKinlay, Robert D., and A. S. Cohen 1975. "A
ComparativeAnalysisof the Politicaland Economic
Performance of Military and Civilian Regimes,"
ComparativePolitics, 8,1: 1-30.

143

Military Expenditure and Economic Development

McKinlay, Robert D. 1989. Third World Military

Expenditure: Determinants and bmplications,
(England:Pinter).
Moll, Kendall, and Gregory Luebbert 1980. "Arms
Race and MilitaryExpenditureModels,"Journalof
ConfljictResolution, 24,1: 153-85.
Moodie,Michael1979. "DefenseIndustriesin the Third
World: Problemsand Promises,"in S. Neumanand
R. Harkavy, eds., Arms Transfers in the Modern
World(New York: Praeger), pp. 294-312.
Neuman, Stephanie G. 1978, "Security, Military
Expenditures and Socioeconomic Development:
Reflectionson Iran," Orbis, 22,3: 569-94.
1984a. "Third World Arms Productionand
the Global Arms Transfer System," in James E.
Katz, ed., Arms Production in Developing
Countries: An Analysis of Decision Making
(Toronto:LexingtonBooks), pp. 15-37.
1984b. "InternationalStratificationand Third
World Military Industries,"
International
Organization,38,1: 167-97.
-- 1988. "Arms, Aid and the Superpowers,"
ForeignAffairs, 66,5: 1044-66.
Neuman, StephanieG. 1989. "The Arms Race: Who's
on Top?" Orbis, 33,4: 509-21.
Nordlinger, Eric A. 1970. "Soldiers in Mufti: The
Impactof MilitaryRuleupon Economicand Social
Changein Non-WesternStates,"AmericanPolitical
ScienceReview, 64,4: 1131-48.
O'Leary, M., and W. Coplin 1975. Quantitative
Techniques in Foreign Policy Analysis and
Forecasting(New York: Praeger).
Ostrom, Charles W. Jr. 1977. "EvaluatingAlternative
Foreign Policy Decision Making Models: An
Empirical Test Between an Arms Race Model
OrganizationalPoliticsModel," Journal of Conflict
Resolution,21,2: 235-66.
1978. "A Reactive Linkage Model of the
U.S. DefenseExpenditurePolicy-makingProcess,"
AmericanPolitical Science Review, 72,September:
941-57.
Peacock, Alan T., and Jack 'Wiseman 1961. The
Growth of Public Expenditure in the United
Kingdom (Princeton, N.J.: Princeton University
Press).
1979. "Approachesto the Analysis of
GovernmentalExpenditureGrowth,"Publc Finance
Quarterly,7,1: 3-23.
Peleg, llan 1980. "MilitaryProductionin Third World
Countries: A PoliticalStudy," in P. McGowanand
C. Kegley, eds., Threats, 'Weaponsand Foreign
Polity (London: Sage Publications), pp. 209-30.

Peroff, Kathleen, and Margaret Podolak-Warren1979.

"Does Spending on Defense Cut Spending on
Health? A Time Series Analysis of the U.S.

144

Economy 1929-1974,"British Journal of Political

Science, 9,1: 21-39.
Pluta, Joseph 1979. 'The Performance of South
AmericanCivilianand MilitaryGovernmentsfrom
a Socio-EconomicPerspective,"Developmentand
Change, 10: 461-83.
Rattinger,H. 1976. "Econometricsand ArmsRaces: A
Critical Review and Some Extensions,"European
Journalof PoliticalResearch,4: 421-39.
Ravenhill, John 1980.
"Comparing Regime
Performancein Africa: The Limitationsof CrossNational Aggregate Analysis," The Journal of
ModernAfrican Studies, 18,1: 99-126.
Richardson, L. 1960. Arms and Insecurity: A
MathematicalStudy of the Causes and Origins of
War (Chicago:Quadrangle).
Rosh, Robert M. 1988. "Third World Militarization:
Security Webs and the States They Ensnare,"
Journalof ConflictResolution,32,4: 671-98.
1990. "ThirdWorld Arms Productionand the
Evolving Interstate System," Journal of Conflict
Resolution,34,1: 57-73.
Ross, AndrewL. 1987. "Dimensionsof Militarization
in the Third World," Armed Forces and Society,
13,4.
Russett, B. 1972. The Revol of the Masses: Public
Opinion Toward Military Expenditure (Beverly
Hills, Calif.: Sage).
Schmitter, Philippe C. 1973. "Foreign Military
Assistance,NationalMilitarySpendingand Military
Rule in Latin America," in PhiOlipeC. Schmitter,
ed., Military Rule in Latin America: Function,
Consequencesand Perspectives,(London: Sage),
pp. 117-87.
Smith, Dan, and Ron Smith 1979. "Reflectionson
Neuman," Orbis, 23,2: 471-81.
Studenmund,A. H., and Henry J. Cassidy1987. Using
Econometrics:A Practical Guide (Boston: Little,
Brown & Company).
Summers,Robert, and Alan Heston 1988. "A New Set
of InternationalComparisonsof Real Product and
Prices for 130 Countries, 1950-85," Review of
Income and Wealth,34,1: 1-25.
Tait, Alan A., and Peter S. Heller 1982. International
Comparisons of Government Expenditure, IMF
Occasional Paper No. 10 (Washington, D.C.:
InternationalMonetaryFund).
Tannahill, Neal R. 1976. "Military Intervention in
Search of a Dependent Variable," Journal of
Politicaland MilitarySociology,3: 219-28.
Taylor, J. 1979. "Recent Developments in the
Lanchester Theory of COMSAT," in K. Haley, ed.,

OperationalResearch '78 (Amsterdam:Elsevier

North Holland),pp. 773-806.
United Nations Department of Disarmament Affairs
1983. Reductionof MilitaryBudgets: Refinement

Determinants of Military Expenditure in Developing Countries

of International Reporting and Comnparisonof

MilitaryExpenditures(New York: UnitedNations).
Report of the SecretaryGeneral,GeneralAssembly
DocumentAIS-1217.
1986. Reduction of Military Budgets:
Construction of Military Price Indexes and
Purchasing Power Parities for Comparison of
MilitaryExpenditures(New York: UnitedNations).
Report of the SecretaryGeneral,GeneralAssembly
DocumentA/40/421.
United States Arms Control and DisarmamentAgency
1980. World Military Expendituresand Arms
Transfers 1969-78 (Washington, D.C.: Arms
Control and Disarmament Agency). ACDA
Publication108.
Verner, Joel G. 1983. "BudgetaryTrade-Offsbetween
Education and Defense in Latin America: A
Research Note," Journal of Developing Areas,
October 18, 1983: 77-92.
Wallace,Michael,and Judy Wilson1978. "Non-Linear
Arms Race Modes," Journal of Peace Research,
15,2: 175-92.
Weaver, Jerry L. 1973. "Assessing the Impact of
MilitaryRule: AltemativeApproaches,"in Phillipe

West

C. Schmitter,ed., Miltary Rule in Latin America:

Function,Consequencesand Perspectives(London:
Sage), pp. 58-116.
Weede, Erich 1986.
"Rent Seeking, Military
Participation, and Economic Performance in
LDCs," Journalof ConflictResolution,30,2: 291314.
West, Robert L. 1987. "ImprovedMeasures of the
Defense Burden in Developing Countries," in
Saadet Deger and Robert L. West, eds., Defense,
Security and Development (London: Frances
Pinter),pp. 19-48.
Whynes, David K. 1979. The Economics of Third
WorldMilitaryExpenditure,(London:Macmillan),
p. 165.
Wulf, Herbert 1983. "DevelopingCountries," in Ball
and Leitenberg,eds., TheStructureof the Defense
Industry: An InternationalSurvey (NewYork: St.
Martin's Press), pp. 311-43.
Zuk, Gary, and WilliamThompson 1982. "The PostCoup MilitarySpendingQuestion: A PooledCrossSectional Time Analysis," American Political
ScienceReview, 76,1: 60-74.

145

AppendiJx
BACKGROUND NOTE ON MILITARY EXPENDITURE:
SOURCES AND PRICE CONVERSION PROCEDURES

Robert L. West

Virtually all of the cross-country empirical

testing of propositions concerning the
relationshipof securityand developmentin the
developingworld has been carried out in the
last twenty years. Studies of the worldwide
distributionof military expenditureand of the
trends of military spending in different
geographic regions are also predominantlya
product of the past two decades. These lines
of investigationhave been made possible by
the publicationon a regular basis--beginning
about twenty years ago-of data series
reporting on the scale of national security
expendituresby a large numberof countries.
The fiscal reports of the individual
governmentsare the primary source of these
data. But the assembly, editing, and annual
publicationof data series for a large number
of nations is the activityof a few international
reporting agencies, notably the U.S. Arms
Control and DisarmamentAgency (ACDA),
the Stockholm International Peace Research
Institute (SIPRI), and the GovernmentFinance
Statistics Division of the International
MonetaryFund.'
Becausethe data reportedby these agencies
is readily available, and because users are
confidentof the care that has been exercisedin
assembling the information, the estimates

publishedannuallyby these agencies are very

widelyused to describecontemporarypatterns
and trends of military expenditures in
developingcountries.
The three agencies differ somewhat in
country coverage, in methods of estimation
employed, and in definitions of variables.
Their annual publications are, however,
broadlysimilar in the presentationof military
expenditure (MILEX) estimates with very
wide geographiccoverage,reportingdata on a
consistentbasis for a period of a decade.
Each of the three agencies also publishes
estimatesof associatedexpenditureseries for
each country, accommodatingthe familiar
convention of comparing MILEX with total
expendituresby the centralgovernment(CGE)
and with gross nationalproduct (GNP). The
ratiosof MILEXto CGE (the "defenseshare"
of central governmentexpenditures),of CGE
to GNP (a measure of the "size of the public
sector"), and of MILEXto GNP (the "military
burden")-are widelyused as indicatorsof the
level of effort by a governmentto providefor
nationalsecurity.
The expendituresreported by the national
reporting sources are in local currency at
Ratios of expenditures
current prices.

147

MilitaryExpenditureand EconomicDevelopment

expressed in national currencies-such as the

ratio of military expenditures divided by
national income-reflect the prices in each
country during the period in which the
expenditure occurred. For some analytical
purposes, comparisonsat national prices are
appropriate. For comparisonsof expenditures
over time, however, an adjustmentis required
for changes in price levels. For purposesof
making international comparisons, it is
necessaryto convert the spending in national
currencies to a common numeraire or base
currency, such as the U.S. dollar of some
specifiedbase period.
The convertor most commonlyused for
this purpose is the exchangerate. This is the
convertor used by both ACDA and SIPRI.2
The deficiencies of using exchange rates as
convertors are widely acknowledged. When
appliedto broad nationalaccountsaggregates,
they result in statistics that are far from
satisfactory for making inter-country
comparisonsof real incomeor real product,or
for making comparisons between real
quantitiesover time for groupsof countriesor
for one country. The structure of prices for
goods and services differs widely among
countries, and the changesin these structures
that occur over time are quite different across
countries. Exchange rates do not adequately
reflect these internationaldifferencesin price
levels and structures. ACDA (amongothers)
recognizes that the convertorsthat it uses do
not allow for a number of factors: "any
within-countrydifferences between the price
indexes for military or central government
expendituresand for GNP are not taken into
account. . .. A more seriousproblem is that
exchange rates in many cases do not
adequately reflect the relative purchasing
power of currencies.
To correct for these factors, and to put
expenditures in different countries on a
comparable basis, it is necessary to use
convertorsderived from price comparisonsof

148

the componentsof the expenditureaggregates.

For broad national accounts aggregates, such
as governmentsector expendituresand GNP as
a whole, these convertors--termedparities, or
purchasing-powerparities (PPPs)--have been
computedfor a large number of countriesand
for a long span of years. These parities have
been developed by the International
ComparisonProject, a cooperativestatistical
enterprise coordinatedby the United Nations
StatisticalOffice and supportedby the World
Bank.4
Using parities as convertors results in
expenditureestimatesin a numerairecurrency,
showing the real quantity of goods and
services purchased, valued at a single set of
internationalprices, in each country. The
1980 internationaldollar has the equivalent
purchasingpower of one U.S. dollar, in U.S.
GNP terms, in 1980. Expendituresexpressed
in these PPP dollars can be aggregatedfor
groups of countries and compared across
countries and over time. In all these cases,
the aggregatesand comparisonscorrespondto
real quantitiesof goods and services. These
cross-countryand inter-temporalcomparisons
are not subject to the distortions that result
from currency conversionby exchangerates.
Purchasing-power parities that are
appropriatefor convertingcentralgovernment
expenditures and GNP are provided by the
International ComparisonProject and by the
statistical offices that have continued the
developmentalwork of the ICP. But selecting
the parities for converting the military
expendituresof one country into the currency
of anotherrequiresclarificationof the question
being asked. One possible purposeof making
internationalcomparisonsmay be to measure
the total resources devotedby each nation to
its military sector. But a second and quite
different purpose may be to measure the cost
to each nation of providing the military
establishment that the country chooses to
have.5

Appendix: BackgroundNote on MilitaryExpenditure

][n comparing military expenditures as

measures of the real resources devoted to
national security, the focus is on military
spendingas the purchaseof inputsfor national
defense. The component inputs-manpower,
iFuel,weapons,material-mustbe identifiedand
convertorsderived from price comparisonsof
the components. Improvementin the national
reportingof standardized,detailedinformation
on the composition of military expenditure
categories has been a particular goal of the
United Nations Department of Disarmament
Affairs, whose experts have since 1975
pressed for member governments to publish
disaggregatedsecurityexpendituresdata and to
compute the associated military purchasingpower parities.6
Compliancehas been partial and erratic.
But a sufficient body of data has been
assembled to permit the United Nations to
convert the defense expendituresfor a small
number of countries on the basis of basic
parities derived from detailedprice and salary
comparisons in 1980. Building on these
directly derived estimates of military price
levels, on the civilian price comparisonsfor
the 71 benchmark countries that have
participated in the ICP studies, and on
methodsof indirect estimationof price levels
developed by the International Comparison
Project, Alan Heston has computed estimates
for 134 countriesof real militaryexpenditures
in :1980that have been converted by use of
militarypurchasing-powerparities.'
For Heston's computationof militaryprice
levels and parities, many of the basic parities
for the militaryhave been direcdycarried over
from the basic parities for categories of
civilian expendituresof the 1980 ICP study.
For some items, however, there are no direct
carry-overs, as in the case of military
personneland certainprocurementitems. For
these items, methods to infer defenseparities
from civilianparities have been derived from
the pilot military purchasing-powerstudy by

West

the United Nations Expert Group. For the

estimationof military personnelexpenditures,
the largestsingle expenditurecategoryin most
countries, the pricing of personnel, is
complicatedby the requirement to take into
accountthat:somecountriesuse conscriptsand
some do not.
Heston's experimentalset of militaryprice
level estimatesand parities is available only
for the year 1980. For that base year, it
provides convertors that permit us to make
comparisonsacross a wide range of countries
of the quantitiesof real inputs devotedby each
nation to its militarysector, valued at a single
set of international prices for military
expenditures and expressed in 1980
international-dollars.
A distinctly different purpose is to
comparemilitaryexpendituresas a measureof
the cost to each nationof providingits military
establishment. For this purpose, the proper
measureof cost is that of opportunitycost-the
value of alternative goods and services
foregonebecauseof the decisionsmadeby the
fiscal and budgetary authorities to allocate
expenditures for national defense.
In
conformancewith this concept of cost to each
nation, the comparison should be made of
military expenditures computed as the real
value of the alternativebundle of goods and
services sacrificed by the residents of the
country(that is, the real value of the purchases
that are not made) as a result of the
government'sdecisionto spend the amountof
local currency devoted to national security.
We need indexesof price-levelsfor comparing
the real militaryburdens assumedby different
countries, and their trend through time.
We cannot know what bundle of goods
and services would have been purchased in
each country if the nation had not made the
military expendituresactually observed. But
an estimateof the counter-factualcan be based
on reasonable assumptions and on the

149

Military Expenditure and Economic Development

observation that expenditures for national

security are, in all countries, exclusivelyan
activity of the public sector and the outcome
of a process of central government decision
making with respectto the allocationof fiscal
resources. Military expenditures represent
public use of resources; imoneyexpendedfor
publicpurposesis fungible. For the period of
years addressedin this paper, and for the large
number of countries compared, civil
governmental expenditures were expanding
rapidly and the constrainton growth of public
outlays appeared to be the government's
capacityto mobilizeresourcesfor publicuses.
In these circumstances,for the purpose of
making international comparisons of the

opportunitycost of military expenditures,it is

a reasonable assumptionthat the alternatives
foregone as a result of a given country's
military expenditure can be represented by
equivalent additional spending on the actual
mix of goods and services purchased by the
central government in its civil expenditures
during that year.
The price level
correspondingto this expendituremeasure is
the central governmentprice level. The real
value of the opportunity cost of military
expenditures thus can be represented as the
reported military spending in local currency
converted at the purchasing-powerparity for
general government final consumption
expenditurein 1980internationaldollars.

REFERENCES
Notes

consists of two steps. (1) ACDA deflates each

country's data, expressed originally in national
currency and at current prices, by means of the
country's implicit deflator for GNP as a whole (the
GNP price index is used for all expenditurevariablesmilitary spending, total govemment spending, and
GNP); thus nationalcurrency data for all years are
expressedin the prices of the conversionbase year.
(2) These data are then dividedby the annual average
par/market exchange rate for the base year between
the national currency and the U.S. dollar-and thus
converd into constantbase-yeardollars. See ACDA,
WorldMilitaryExpendituresandArms Transfers1989,
pp. 138-39. SIPRI conversionfollowsthe same two
steps but (generally) employs the consumer price
indexas the nationaldeflator.

The data appear in the followingpublications: U.S.

Arms Control and DisamnamentAgency (ACDA),
World Military Expenditures and Arms Transfers
(Washington,D.C.: U.S. GovernmentPrintingOffice,
annual). Stockholm Intemational Peace Research
Institute(SIPRI),WorldArmnanentsand Disarmament,
SIPRI Yearbook, (Oxford and New York: Oxford
University Press, annual). lnternational Monetary
Fund (IMF), GovernmentFinanceStatisticsYearbook
(Washington,D.C.: IMF, annual).
2

150

The IMF reports expenditures only in national

currencies; SIPRI provides data in both national
currenciesand "dollars"; ACDAreportsonly the data
converted into "dollars." The SIRPI and ACDA
methods of conversion are similar, and both use
exchange rates. The ACDA.conversion approach

ACDA, World Military Expenditures and Arns

Tranfers1989, p. 139. The deficienciesof exchange-

Appendix: Background Note on Military Expenditure

rate convertors and the desirability of using

purchasing-powerparities havealso beenaddressedin
the academicliterature. See, for example, Jacques
Fontanel, "A Note on the InternationalComparisonof
Military Expenditures,"and "Note on the paper by
Fonatanelby Frank Blackaby,"in ChristianSchmidt,
ed., The Economicsof Military Expenditures(New
York: St. Martin's Press, 1987) pp. 29-46, and
RobertL. West, "ImprovedMeasuresof the Defence
Burdenin DevelopingCountries,"in SaadetDeger and
Robert L. West, eds., Defence, Security and
Development(London: Frances Pinter Publishers,
1987) pp. 19-48.
The methodsused for the direct estimationof price
level indexedand parities for participatingcountriesin
benchmarkyears are described in the reports of the
United NationsInternationalComparisonProject. The
Phase III report is: Irving B. Kravis, Alan Heston,
and Robert Summers, World Product and Income:
International Comparisons of Real Gross Product
(Baltimore,Md.: Johns HopkinsUniversityPress for
the World Bank, 1982),esp. Chapters 1-3. The basic
methodological approach is to obtain quantity
comparisons by means of price and expenditure
comparisons. Comparisonsare basedon data supplied
by the participating countries on prices and
expenditures for more than 150 detailed final
expenditure categories.
By combining these
expenditurecategories,price level indices and parities
are computedfor expenditureaggregatessuch as the
main components of GNP (consumption, capital
formation,and government)and for GNP as a whole.
The methods of making indirect estimates for

West

nonbenchmarkyears, and for countries that do not

directlysupply price and expenditureobservations,are
described in Robert Summers and Alan Heston, "A
New Set of IntemationalComparisonsof Real Product
and Prices and 130 Countries, 1950-85," Review of
Income and Wealth, Vol. 34, No. 1 (March 1988),
pp. 1-25.
S

See the discussion of these alternative purposes of

convertingmilitary expendituresin the note on the
paper of JacquesFontanelby Frank Blackby,op. cit.,
pp. 44-45.

See United Nations Department of Disarrnament

Affairs,Reductionof MilitaryBudgets: Refinementof
InternationalReportingand Comparisonof Military
Expenditures(New York: UnitedNations, 1983);and
United Nations Departmentof Disarnament Affairs,
Reduction of Military Budgets: Construction of
MilitaryPrice Indexesand PurchasingPowerParities
for Comparisonof MilitaryExpenditures(NewYork:
United Nations, 1986).

Alan Heston, "RealWorld MilitaryExpenditures:134

Countries, 1980"(Universityof Pennsylvania:ACDA
DiscussionPaper, 1990). In addition to presenting
estimates of 1980 real military expenditures and
military price levels for 122 market-oriented
economiesand 12 centrally planned economies,this
paper contains a descriptionof the data sources and
methods employed. Alan Heston is one of the
prirncipaleconomistsassociatedwith the International
Comparison Project Unit at the University of
Pennsylvania.

151

Distributors of World Bank Publications

ARCENTINA
Caeos Hirsch, SRL
Calera Guemes
Florida 165,4th Floor-Ofc. 453/465
1333 Buence Aires
AUSTRALIA, PAPUA NEW CUiNEA,
FIJI, SOLOMON ISLANDS,
VANUATU, AND WESTERN SAMOA
D-. Books & Jourmals
648WhitehorselRoad
Mitcham3132
Victoria
AUSTRIA
Gerold and Co.
Graben31
A-lOll Wien
BANGLADESH
Micro Industries Development
Assistance Society (MiDAS)
House S, Road 16
Dhanmondi R/Area
Dhaka 1209
Brsnch ozfw
156. Nur Alhmed Sarak
Chittagong 4000
76, K.DA Avenue
Kulna 9100
BELGIUM
jean De Lannoy
Av. du Roi 202
1060 BrusseLs
CANADA
Le Diffuseur
CP. 85 15018 rute Ampire
BoucherviUe Quebec
J4B SE6
CHILE
Invertec ICT S.A
Americo Vespucdo Norte 1165
Santiago
CHINA
uhinaFtnancial & Economic
Publishing House
8, Da Fo Si Dong jie
Beijing
COLOMBIA
Infoenlace Ltda.
ApartadoAereo 34270
Bogota D.E.
COTE DYIVOIRE
Centre dEdition et de Diffusion
Africaines (CEDA)
04 BYP.541
Abidjan 04 Plateau
CYPRUS
Cyprus College Bookstore
6. Diogenes Sbt,
Engomi
P.O. Box 2006
Nicosia
DENMARK
SamfundsLtteratur
Rosenoerns Ali 11
DK-1970 FrederikbergC
DOMINICAN REPUBLIC
Editorm Taller, C por A.
Resturacifn
e Is"b la Cat6lica 309
Apartado de Correos 2190 Z-1
Santo Domingo
EGYPT ARABREPUBLICOF
Al Ahram
Al Galaa Street
Cairo
The Middle East Observer
41. Sherif Street
Caio

FINLAND
Akateeminen Khrjakauppa
P.O. Box 128
SF-0 01 HeLsinid 0

KOREA, REPUBUC OF
Pan Korea Book Corporation
P.O. Box 101. Kwangwhamunu
Seoul

FRANCE
World Bank Publications
66, avenue dq6na
75116 Paris

MALAYSIA
University of Malaya Cooperative
Bookshop. Limited
P.O. Box 1127, Jalan Pantai Baru
59700 Kuala Lumpur

GERMANY
UNO-Verlag
Poppeisdorfer AileeSS
D-5300 Bonn I

MEXiCO
IN5CTiEC
Apartado Postal 22-860
14060 Tiaipan, Mexico D.F.

HONG KONG, MACAO

Asia 2000 Ltd.
46-48 Wyndhamn Street
Winning Centre
2nd Floor
Centrai Hong Kong

NETHERLANDS
De Lindeboorn/InOr-Publikaties
P.O. Box 202
7480 AE Haaksbergen

INDIA
Allied Pubiishers Private Ltd.
751 Mount Road
Madras - 600 002
Branch offic.
15 I.N. Heredia Marg
Ballard Estate
Bombay -400 038

NIGERIA
University Press Limited
Three Crowns Building Jercho
Private Mail Bag.5095
Ibadan

13/14 Asaf Ali Road

New Delhi - 110 002

NORWAY
Narvesen Information Center
Book Department
P.O. Box 6125 Etterstad
N-0602 Oslo 6

17 Chittaranjan Avenue
Calcutta - 700 o07
layadeva Hostel Building
5th Main Road Candhinagar
Bangalore - 56 009
3-5-1129 Kachiguda Crot
Hyderabad - 500 027

NEWZ EALAND
EBSCO NZ Ltd.
Private Mail Bag 99914
New Market
Auckland

Road

Prarthana Flats, 2nd Floor

Near Thakore Baug, Navrangpura
Ahmedabad -380009
PatLala House
16-A Ashok Matrg
Lucknow - 226 001
Central Bazaar Road
60 Baja; Nagar
Nagpur 440010 _
INDONESIA
Pt Indira Limited
JI. Sam Ratulangi 37
P.O. Box 181
Jakarta Pusat
ISRAEL
Yozmot Literature Ltd.
P.O. Box 560S55
Tel Aviv 61560
Israel
ITALY
Licosa Commiesionaria Sansoni SPA
Via Duca Di Calabria, 1/1
Casella Postale 552
50125 Firenze
JAPAN
Eastern Book Service
Hongo 3-Chome, Bunkyo-ku 113
Tokyo
KENYA
Africa Book Service (E.A.) Ltd.
Quararn House. Mfangano Street
P.O. Box 45245
!Nairobi

PAKISTAN
Mirza Book Agency
65, Shahrah-eQuaid-e-Azam
P.O. Box No. 729
Lahore 54000
PERU
Editorial Desarrollo SA
Apartado3824
Lima I
PHULiPPINES
International Book Center
Fifth Floor, Filipinas Life Building
Ayala Avenue,
aikati
Metro Manila
POLAND
International PubiLshing Service
Ul. Piekna31/37
00-677 Warzawa
For s57cription orde:
IPS Journals
UlI. Okrezna 3
02-916 Warszawa
PORTUGAL
Livraria Portugal
Rua Do Carmo 70-74
1200 Lisbon
SAUDI ARABIA. QATAR
jarir Book Store
P.O. Box 3196
Riyadh 11471
SINGAPORE TAIWAN,
MYANMAR,BRUNEI
Infomiation PubOcations
Private, Ltd.
Golden Wheel Building
41, Kallang Pudding #04-03
Singapore 1334

SOUTH AFRICA, BOTSWANA

For single tiflw
Oxford University Pees
SouthernAfrica
P.O. Box 1141
Cape Town 8000
For srdecrition orders:
Intemational Subscription Service
P.O. Box 41095
Craighall
Johannesburg 2024
SPAIN
Mundi-Prensa
Castello 37
28001.Madrid

Ubros SA.

LibTeria Intemational AEDC6

Consell de Cent, 391
08009 Barcelona
SRI LANKA AND THE MALOiVES
Lake House Bookshop
P.O. Box 244
100, Sir Chitatmpalam A.
Gardiner Mawatha
Colombo 2
SWEDEN
For sigle ltitis:
Fritres Fackboksforetaget
Regeringagatan 12. Box 16356
5-1i03 27 Sbockholm
Forsubscrtitun ordcrs:
Wennergren-Wihliams AS
Box 30004
S-104 25 Stockholm
SWtTZERLAND
For single fitle:
Librairie Payot
1, rue de Bourg
CH 1002 Lausanne
For subscri7tion orders:
Librairie Payot
Service des Abonnements
Case postale3312
CH 1002 Lausanne
TANZANIA
Oxford University Pres
P.O. Boa 5299
Maktaba Road
Dar es Salaam
THAILAND
Central Department Store
306 Silom Road
Bangkok
TRINIDAD & TOBAGO, ANTIGUA
BARBUDA, BARBADOS,
DOMINICA, GRENADA, GUYANA,
JAMAICA MONTSERRAT, ST.
KITrS & NEVIS, ST. LUCIA,
ST. VINCENT & GRENADINES
Systematics Studies Unit
#9 Watts Street
Curepe
Trinidad, West lndles
TURKEY
Infotel
Narlabahqe Sok. No. 15
Cagaloglu
Istanbul
UNITED KINGDOM
Microinfo Ltd.
P.O. Box 3
Alton. Hampshire GU34 2G
England
VENEZUELA
Libreria del Este
Aptdo. 60337
Caracas 1060-A

R.ecent

World

Bank Discussion

Papers

(continued)

No. 154

Earnings,Occupational
Choice,and Mobilityin SegmentedLaborMarketsof India. Shahidur R. Khandker

No. 155

ManagingExternalDebt in DevelopingCountries:Proceedings
of aJoint Seminar,Jeddah,May 1990. Thomas M.
Klein, editor

No. 156

DevelopingAgricultural
ExtensionforWomenFarmers.Katrine A. Saito and Daphne Spurling

No. 157

Awakeningthe Market: Viet Nam's EconomicTransition.D. M. Leipziger

No. 158

WagePolicyduringthe Transition
to a MarketEconomy:Poland1990-91. FabrizioCoricelliand Ana Revenga, editors

No. 159

InternationalTradeand theEnvironment.Patrick Low, editor

No. 160

International
Migrationand International
Trade.Sharon Stanton Russell and Michael S. Teitelbaum

No. 161

Civil ServiceReformand the WorldBank. Barbara Nunberg andJohn Nellis

No. 162

Rural EnterpriseDevelopmentin China, 1986-90. Anthony J. Ody

No. 163

The BalancebetweenPublicand PrivateSectorActivitiesin the Deliveryof LivestockServices.Dina L. Umali, Gershon

Feder, and Comelis de Haan

No. 165

FisheriesDevelopment,FisheriesManagement,and Externalities.Richard S.Johnston

No. 166

The BuildingBlocksof Participation:

TestingBottom-upPlanning.Michael M. Cemea

No. 167

SeedSystemDevelopment:The AppropriateRolesof the Privateand PublicSectors.StevenJaffee and Jitendra Srivastava

No. I 68

EnvironmentalManagementand UrbanVulnerability.
Alcira Kreimer and Mohan Munasinghe, editors

No. 169

Common Property Resources:A Missing Dimension of Development Strategies. N. S. Jodha

No. 170

A ChineseProvince
asa ReformExperiment:The CaseofHainan.Paul M. Cadario,Kazuko Ogawa, andYin-Kann Wen

No. I71

IssuesforInfrastructure
Managementin the 1990s.Arturo Israel

No. 172 JapaneseNationalRailwaysPrivatizationStudy: The ExperienceofJapanand LessonsforDevelopingCountries.

Koichiro Fukui
No. 173

The LivestockSectorin EasternEurope:Constraintsand Opportunities.Comelis de Haan, Tjaart Schillhom Van

Veen, and Karen Brooks

No. 174

AssessingDevelopmentFinanceInstitutions:A PublicInterestAnalysis.Jacob Yaron

No. 175

ResourceManagementand PastoralInstitutionBuildingin the West AfricanSahel.Nadarajah Shanmugaratnam,

Trond Vedeld, Anne Mossige, and Mette Bovin

No. 176

Research:Theoryand Experience.Dina L. UmaLi

Publicand PrivateSectorRolesin Agricultural

No. 177

The RegulatoryImpedimentsto the PrivateIndustrialSectorDevelopmentin Asia: A ComparativeStudy. Deena

Khatkhate

No. 178

China:ReformingIntergovernmental
FiscalRelations.Ramgopal Agarwala

No. 179

Nippon Telegraphand TelephonePrivatizationStudy: ExperienceofJapanand LessonsforDevelopingCountries.

Yoshiro Takano

No. 180

China's ReformExperienceto Date.Peter Harrold

N,o. 181

CombattingAIDS and Other SexuallyTransmittedDiseasesin Africa:A Reviewof the WorldBank's AgendaforAction,

Jean-Louis Lamborayand A. Edward Elmendorf

No. 182

PrivatizationProblemsat IndustryLevel:RoadHaulagein CentralEurope,Esra Bennathan and Louis S. Thompson

:No. 183

Participatory
Developmentand the WorldBank: PotentialDirectionsforChange,Bhuvan Bhatnagar and Aubrey C.
Williams, editors

No. 184

Agricultural
Researchin SouthernAfrica:A FrameworkforAaion, Andrew SpurLing,Teck Y. Pee, Godwin
Mkamanga, and Christopher Nkwanyana

The World Bank

Headquarters
1818 H Street, N.W.
Washington, D.C. 20433, U.S.A.

European Office
66, avenue d'lena
75116 Paris, France

Telephone: (202) 477-1234

Facsimile:(202) 477-6391
Telex: wus 64145 WORLDBAN5C

Telephone: (1) 40.69.30.00

Facsimile: (1) 40.69.30.66
Telex: 640651

RCA 248423 WORLDBK

Cable Address:

Tokyo Office
Kokusai Building
1-1 Marunouchi 3-chome
Chiyoda-ku, Tokyo 100japan
Telephone: (3) 3214-5001
Facsimile:(3) 3214-3657
Telex: 26838

INTBAFRAD
WASHINGTONDC

ISBN 0-8213-2289-3