Growth and Defense in Developing Countries

Author(s): Emile Benoit

Reviewed work(s):
Source: Economic Development and Cultural Change, Vol. 26, No. 2 (Jan., 1978), pp. 271-280
Published by: The University of Chicago Press
Stable URL: http://www.jstor.org/stable/1153245 .
Accessed: 30/08/2012 10:38

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to
Economic Development and Cultural Change.

http://www.jstor.org
Growthand Defense in DevelopingCountries*

Emile Benoit
Columbia University

It has usually been supposed by economists that defense expenditures

reduce the resources available for investment and so slow down growth.
The evidence available for developed countries is at least not inconsistent
with that view.l
However, in a large study of less developed countries, an opposite
pattern seemed to appear, and this finding was so unexpected and chal-
lenging that it seemed worth exploring in detail. In the course of that
exploration, a number of interesting insights were achieved as to the nature
of economic development and its relation to defense activity.
The study covered the growth rates, investment rates, foreign aid
receipts, and certain other variables of 44 LDCs between 1950 and 1965.
These materials were analyzed primarily by means of correlation analysis.
The 44-country sample covered about three-quarters of the population,
GNP, and defense expenditures of the developing world, other than main-
land China. The sample analysis was supplemented by case studies of
India, South Korea, Mexico, Israel, the United Arab Republic, and
Argentina. (See table 1.)

The Main Finding

Contrary to my expectations, countries with a heavy defense burden
generally had the most rapid rate of growth, and those with the lowest
defense burdens tended to show the lowest growth rates. (Incidentally, I
mainly used civilian growth-the rate of increase in nondefense output-
as the growth index, in order to avoid counting as growth that which was
merely an increase in defense expenditures.)
The strong positive correlation between high defense burdens and
rapid growth rates was initially discovered in quartile analysis, supported
by Spearman rank order correlation analysis, and confirmed by regression
* This articlediscussesthe policy implicationsof data assembledby the author in
his book entitled Defense and Economic Growth in Developing Countries (Boston:
D. C. Heath & Co., LexingtonBooks, 1973).
1 Ibid., appendix 3G. This appendixcovers 19 developedcountriesfrom 1950 to
1965. It shows that defenseburdens(defenseas a percentageof GNP) are inverselycor-
relatedwith growthrates,but with a strength(-.2557) too low to be viewedas significant
at the .05 level.

271
Economic Developmentand Cultural Change

TABLE 1
INDICATORS OF DEFENSE AND DEVELOPMENT FOR SAMPLE COUNTRIES

AB AG' AI AR2

Argentina................ 2.49 3.37 19.47 .01

Brazil.................. . 2.63 5.16 16.43 .53
Burma................... 6.61 4.93 18.06 1.34
Ceylon................... .94 3.29 13.29 .93
Chile.................... 2.69 3.63 10.80 .83
China (Taiwan)........... 11.42 8.12 17.96 5.23
Colombia................ 1.59 4.50 18.04 .04
Costa Rica................ .32 4.85 18.08 1.96
Dominican Republic....... 4.23 4.28 14.49 2.87
Ecuador.. .............. 2.11 4.81 14.10 .34
El Salvador............... 1.41 4.95 12.58 .71
Ghana................... 1.18 4.23 16.96 4.20
Greece........ .......... 5.21 6.58 20.68 3.27
Guatemala............... .89 4.73 11.01 .89
Honduras................ 1.24 3.75 14.33 .04
India.................... 2.46 3.20 12.32 1.22
Indonesia............... 3.10 3.40 15.94 1.33
Iran..................... 4.00 4.51 15.25 1.07
Iraq...................... 5.93 6.39 16.45 1.28
Israel.................... 6.08 10.47 29.51 9.97
Jordan................... 16.75 10.70 14.97 24.70
Kenya ......... .60 4.93 11.50 2.02
Malaysia ................ .......... 2.68 5.00 14.47 .46
Mexico .................. .75 6.15 17.25 .12
Morocco................. 3.84 2.20 12.88 2.54
Nigeria ................. .56 3.75 11.93 1.86
Pakistan................. 3.26 3.65 13.37 3.45
Peru..................... 2.70 5.25 22.39 .21
Philippines............... 1.59 5.12 11.11 1.34
South Africa.............. 1.23 4.77 20.03 -.17
South Korea............... 5.32 6.23 13.17 7.88
South Vietnam............ 10.20 3.00 10.33 14.34
Spain.................... 3.26 6.20 20.03 .89
Sudan.................. 1.59 4.92 12.38 2.10
Syria.................... 7.04 4.97 16.95 1.85
Tanzania ................ 2.11 2.90 12.20 -5.53
Thailand................. 3.38 5.85 17.37 .91
Tunisia.................. 1.78 5.68 21.10 10.22
Turkey................... 4.38 5.72 13.22 2.15
Uganda................. .87 5.04 9.52 3.59
U.A.R................... 6.90 6.17 16.60 2.97
Venezuela................. 1.88 6.65 23.96 -.35
Yugoslavia............... 8.76 8.68 35.85 .60
Zambia 1.45 8.17 23.66 -1.29
...............
3.62 5.24 17.15 2.61
Mean....................
Median................. 2.66 4.94 16.07 1.30

NOTE.-All columns referto averageof all values for availableyearsbetween1950

and 1965.AB (defenseburden) = defenseexpendituresas % of GDP at currentprices;
AI (investmentrate) = gross capital formationas % of GDP at currentprices; AR2 =
net receipts(after repaymentsand outpayments)of bilateraleconomic aid, expressedas
% of GNP at currentprices; AG' (growth in civilian product)= real growth in GDP
minusrealgrowthin defenseexpenditure(afterdiscountingfor pricechanges),expressed
as cumulativerate of annual growthbetweenfirst and last years of the availableseries.

272
 Emile Benoit

analysis. With r = .55, t = 4.2, the correlation was strong enough (given
the size of the sample) for there to be a 1,000-to-1 chance against it being
accidental.
The question arose as to whether the results might not simply reflect
defects in the data. These were derived from published and unpublished
estimates of the United Nations, the Agency for International Develop-
ment, and the World Bank. There is no question that they were defective
in many instances. However, these random defects could not readily give
rise to the regularities underlying the strong correlations that were dis-
covered. But could there be a systematic bias in the data? For example,
might the countries in the sample with military governments tend to have
heavier defense burdens and also to exaggerate their growth rates? Burma,
Jordan, and South Vietnam were mentioned by a critic as possible exam-
ples. However, when a regression analysis was done omitting these three
countries, it did not very much affect the results. The 41 remainingcountries
still showed a strong positive correlation.
In any case, it seems far from obvious that military governments do
generally spend more on defense than civilian governments. The rapid rise
in defense expenditures in Burma was halted and the defense burden
thereafter reduced when the military regime took over in 1958. A similar
pattern appears to have occurred in Indonesia. The defense burden in
Venezuela seems to have remained fairly stable in periods of military and
civilian government alike. In Latin America generally there was no clear
evidence that military regimes spent more than others on defense. It
seemed that military regimes might even be tougher than others in resisting
military extravagance, since they understand military budgets better, can
distinguish better between real and phony claims, and may be more
confident than civilian politicians about sidetracking or suppressing
military revolts.
Nor was it at all clear that military governments would have more
incentive than others to exaggerate their economic growth. Certainly
those dependent on foreign aid might have some hesitation about exag-
gerating the extent to which they could already get along on their own
resources.
A more troublesome possibility was that the simple correlation found
between growth and defense burdens might be technically "spurious,"
that is, it might be accounted for by the action of other factors influencing
both growth and defense burdens in such a manner as to bring about their
apparent correlation. For example, bilateral economic aid of military
allies might enable some LDCs to expand their defense expenditures
while simultaneously increasing their rate of investment and their growth
rate--but in such a case the association between the rise in defense and
the increase in the growth rate might be dismissed as spurious, since it
would be derived not from any direct interaction of these two variables
but from the action of other variables.

273
Economic Developmentand Cultural Change

Multiple regression analysis, utilizing bilateral foreign aid and the

investment rate as well as the defense burden showed the defense burden
to have been a significant determinant of growth in the 1960-65 period
but not for the longer period of 1950-65. The underlying data for the
1960-65 period were undoubtedly more accurate, as well as more compa-
rable, with data for all countries covering the same number of years. On
the other hand, the period may have been too brief for the determinants
to have exerted their full effect on the growth rates. Moreover, in such a
brief period, exceptional conditions such as climatic variations affecting
the first or last year of the series could more seriously distort the estimated
rate of growth. Thus, on the whole, the results for the full 1950-65 period
would appear to be more revealing. And in that period the residual
positive correlation between the defense burden and the growth rate and
its t-value appeared to be too weak to justify regarding the defense burden
as a significant determinant of the growth rate. The correlation between
them appeared likely therefore to have been spurious-an artifact reflect-
ing the action of the investment and foreign-aid variables.
Further and more intensive analysis, however, cast serious doubt on
this conclusion. As shown in table 2, dealing with the full 1950-65 period,

TABLE 2
CHIEFMULTIPLE
REGRESSION
RESULTS:CONTRIBUTIONS
TOCIVILIANGROWTH

De- Inde- Partial Contri-

Equa- pendent pendent Corre- bution
tion Variable Variables lation to R2 t R2
I........ AG' AI .6101 .2644 4.9302 .5540
AB .5366 .1803 4.0718
Total. . ... ... . .4447 F(2,41) = 25.4670 (.1093)*
II....... AG' AB .3512 .0985 2.4018 .3002
AR2 .1222 .0106 .7884
Total.. ... ... .1091 F(2,41)= 8.7958 (.1911)*
III...... AG' AI .6961 .3866 6.2086 .5886
AR2 .5858 .2149 4.6277
Total.. ... ... .6015 F(2,41)= 29.3277 (-.0129)*
IV ...... AG' AI .6612 .3059 5.5736 .6061
AR2 .3418 .0512 2.3003
AB .2065 .0176 1.3351
Total. . ... .3747 F(3,40)= 20.5190 (.2314)*
.....
NOTE.-In eq. I the defense burdencan predicttwo-thirdsas much of the total
variation of AG' (civilian growth rate) as can the investment rate. In eq. III, on the other
hand, AR2, bilateral foreign-aid receipts, can predict only 56% as much as the invest-
ment rate. In eq. II, AB appears as several times as strong a determinant as AR2. In eq.
IV, AB appears at first as too weak to be significant. However, of the total R2 of .6061,
.2314 cannot be confidently ascribed to any of the three determinants. If only one-sixth
of it could be presumed to emanate from AB, this would make AB strong enough to
become a significantdeterminant.In view of the predictivepower shown by AB in the
other equations, it seems a reasonable presumption that it could account for at least
this large a share of the unallocable predictive power of the three determinants taken
together.
* Amount of R2 not covered by total individual contributions to R2. The fourth
and sixth columns relating to R2 refer to the square of the statistical magnitude R, which
measures the degree of variation in the dependent variable that corresponds to variations
in the independent variable or variables.

274
 Emile Benoit

the defense burden (AB) has three times as strong a correlation with
civilian growth (AG') as does bilateral economic aid (AR2) in equation II.
It also has two-thirds as strong a correlation as does the investment rate
(AI) in equation I. The reduction of the correlation between the defense
burden and civilian growth to insignificance in equation IV might be
ascribed to the double counting of the resources given as economic aid,
which then also increased the investment rate. This hypothesis is strength-
ened by the peculiar fact that in equation III the individual contributions
of investment and foreign aid to R2 total more than their combined
contributions. Moreover, when one adds the defense burden as a third
determinant in equation IV, then the total of individual contributions to
R2 Of the investment and foreign-aid determinants drops from .6 to only
.37, and the pool of collective explanatory power of the three determinants
that cannot confidently be ascribed to any one of these three determinants
in particular rises to the high level of .23. If only one-sixth of this pool
could be assumed to be allocable to the defense burden (which seems
reasonable enough, under the circumstances), then its strength would be
more than sufficient to make it a significant determinant of growth for the
1950-65 period as well.

Effects of Growth on Defense

Thus there seemed a good probability that the correlation between defense
burdens and growth rates in LDCs was not only positive but strongly
positive enough to make one a significant determinant of the other and to
reveal the existence of direct interaction between them. A question arose,
however, about the direction of this interaction. Might not the correlation
be explained by the influence of growth rates on defense expenditures
rather than vice versa? Countries with rapid growth might feel better able
to indulge themselves in the luxury of elaborate defense programs-just as
rich families are usually more adequately insured than poor ones. More-
over, rapidly rising national incomes might generate an even more rapidly
rising level of tax revenues of which the powerful defense lobby might be
expected to secure a proportional (or rising) share.
These plausible hypotheses did not appear to be supported by the
evidence. No significant correlation was found between income per capita
and defense burdens. Nor were tax revenues, total government expendi-
tures, or the ratio of defense to total government expenditures closely
linked to the rate of economic growth. And in multiple regression analysis,
economic growth did not emerge as a significant determinant of the defense
burden.
What did appear to be the main determinant of the size of the defense
burden was the expectation of political and military leaders of the need for
forces to deter, to threaten, or to engage in combat. Basically the defense
burden was high in areas where combat had occurred or threatened to
occur or which were on boundaries between rival power blocs. Every

275
Economic Developmentand Cultural Change

country in the top quartile of defense burdens had fought a war during
this period or had been threatened with one. But Latin America, where the
experience and expectation of war was low, had a median defense burden
only one-third as high as those in the Middle East, where the experience
and expectations of war are high and where a statistically demonstrable
arms race was found to have occurred-with a correlation over time
between changes in real defense expenditure of the main rivals running as
high as .9.
Thus growth appeared to exert only a weak influence on defense
burdens. Money is not spent on defense just because it happens to be
available: there are always competing needs, and the military, like other
claimants, must justify its claims. Countries which had wars or defense
crises during the period covered which were succeeded by peace or periods
of greater security (like China [Taiwan], Greece, Indonesia, South Korea,
and Yugoslavia) achieved marked declines from their peak defense burdens.
To be sure, money cannot be spent on defense unless it is available, and in
this sense the rise in GDP sets an upper limit on how much of a defense
burden a nation can afford. But the point is that this is an extremely
elastic limit: the defense burdens actually carried in the sample of countries
range from one-third of 1% to nearly 17%. What countries can afford is
determined more by the level of income than by their rate of growth, and
nations may in a crisis spend more than they can really afford, just as in
quiet and secure times they may be glad to spend less than they can afford,
if it seems safe to do so, so as to have more for other pressing needs.

Benefits of Defense for Growth

Thus the direct interaction between growth and defense burdens seems to
run primarily from defense burdens to growth rather than vice versa. It
seems clear that in the sample countries higher defense burdens stimulate
growth, at least to the extent of fully offsetting any adverse growth effects
that defense expenditures may have had. In understanding how this might
be possible, it is important to remember that in LDCs only a small part of
any income not spent on defense is put into highly productive investments.
Most goes into consumption and much of the rest into social investment
such as housing, which may contribute more to consumer satisfaction than
to increasing future production, or into "productive investments," which,
however, are so badly conceived or managed that they operate at un-
economically high costs and contribute less to real growth than they may
appear to do. And on the other side, defense programs, though not
intended to contribute to the civilian economy, may in fact do so in
indirect ways that have not been hitherto adequately recognized. The
optimum civilian programs for growth would no doubt make a much
greater contribution. But one must compare the defense programs with
their objectively probable substitutes, not with their optimum substitutes,
if one seeks to estimate realisticallytheir opportunity costs and net benefits.

276
 Emile Benoit

Defense programs of most countries make 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 LDCs, in ways
that involve sharply raising their nutritional and other consumption
standards and expectations; (2) providing education and medical care as
well as vocational and technical training (e.g., in the operation and
repair of cars, planes, and radios; in hygiene and medical care; in con-
struction methods) that may have high civilian utility; (3) engaging in a
variety of public works-roads, dams, river improvements, airports,
communication networks, etc.-that may in part serve civilian uses; and
(4) engaging in scientific and technical specialties such as hydrographic
studies, mapping, aerial surveys, dredging, meteorology, soil conservation,
and forestry projects as well as certain quasi-civilian activities such as
coast guard, lighthouse operation, customs work, border guard, and
disaster relief which would otherwise have to be performed by civilian
personnel. Military forces also engage in certain R & D and production
activities which diffuse skills to the civilian economy and engage in or
finance self-help projects producing certain manufactured items for com-
bined civilian and military use (e.g., batteries and tires) which might not be
economically produced solely for civilian demand.
Unfortunately, most of the evidence for the growth-stimulativebenefits
of defense programs is imprecise, anecdotal, and difficult to evaluate.
Defense budgets are generally not broken down into elements that would
be most helpful in measuring such effects. For the most part, the secondary
economic effects of defense programs are unreported. Even "civic action"
programs in which defense forces and equipment are used in such a way
as to make a direct contribution to the civilian economy have been so far
only sporadically and impressionistically reported in the open literature.
But the size of regular military forces as a percentage of population does
turn out to be strongly correlated with economic growth.
The militaryestablishment, which in developed countries often appears
as rather archaic and tradition bound, can in the LDC context have quite
a different significance, as an important force for modernization. If it is to
be at all effective, it has to be largely utilitarian and efficiency oriented in
its approach. It also inculcates in those it trains a great many modern
attitudes and aptitudes (e.g., using and conserving machinery; following
and transmitting precise instructions; living "by the clock"; earning,
spending, and saving money; understanding and working with bureau-
cracy). Even more fundamental is its revolutionary effect in destroying
unquestioned acceptance of local custom and tradition, in sometimes
substituting a national for a local, ethnic, tribal, or caste consciousness
and including modern ideas and interests.
Also of considerable possible importance is the contribution made by
the military to "nation building" by inspiring pride and itself symbolizing

277
Economic Developmentand Cultural Change

the nation, by including a significant portion of the male population in an

organized and disciplined confraternitydedicated to national preservation.
Military training may be peculiarlyeffectivein promoting certain moderniz-
ing attitudes and ways of life: in part because it has available the instru-
ments of compulsion; in part because the value which justifies its activity-
national security-exercises a strong influence on most individual con-
sciences and appears to justify imposing difficult and often painful adjust-
ments on the individual. This process is facilitated by the fact that the
military conscript is isolated from his family and community and living in
an artificialcommunity where he is under continuous government control.
A second possible link between defense and growth was through
inflation. I had originally supposed that, even if heavy defense burdens
caused inflation, this could only pull resources away from civilian uses,
not increase their total. I observed, however, that in India the mainland
Chinese attack of 1962-63 resulted in rapid increases in defense spending,
a liberalization of monetary fiscal policies, a considerable rise in prices,
and a substantial speedup in the rate of real increase in civilian goods and
services-until the very bad harvests of 1965 and 1966.
It appeared that somewhat similar forces might have been at work in
a good number of other countries. It was found that with four hyperinfla-
tion countries excluded (those which averaged a more than 15% per an-
num increase in prices) there was a significant positive correlation between
the rate of inflation and the rate of real growth and also between defense
burden and the rate of inflation. It seemed likely that the need to finance
heavy defense burdens had induced certain countries to relax their strict
monetary fiscal policies, leading to more inflation than would otherwise
have occurred-but this inflation, unless extreme, had succeeded in pulling
into economic use unused or underutilized resources which contributed to
real growth.
There are also certain conceivable psychological links between defense
and growth. Military tension has sometimes led nations to cooperate
more effectively and work extra hard. The most obvious example in this
sample is Israel. Fourteen countries that engaged in wars or were in a
peculiarly exposed position had a growth rate one-quarter higher than the
sample as a whole. Creating a common enemy is an ancient technique of
governing, and a willing acceptance of a national leadership may be an
important factor in promoting development.
On the other hand, some countries lacking such military stress (e.g.,
Mexico and Zambia) have had outstanding civilian growth rates, whereas
others which have had serious military stresses or confrontations (e.g.,
India, Indonesia, and South Vietnam) have among the lowest civilian
growth rates in the sample. Thus the true significance of military stress as
such for economic growth remains as yet uncertain.
Another somewhat similar type of hypothesis is that a high level of
general competence and modernization in LDCs, associated presumably
278
 Emile Benoit

with more education, industry, and urbanization, leads to high growth

rates (for more or less obvious reasons) and also to higher defense burdens,
because with a higher level of competence there is more ability to concen-
trate energies and build organizations to obtain high-priority objectives-
such as adequate defense.
This is a plausible though imprecise generalization. It does not take
adequate account of differences in objective military risks, which would
affect the extent to which competent performance would take the form of
larger defense programs. It would also lead us to expect that countries
with higher per capita incomes or a higher ratio of industry to GNP
should have generally higher defense burdens-which turns out not to be
the case.

Policy Implications
One broad implication of the present findings is that development programs
may be putting excessive emphasis on maximizing investment of the usual
sort. Greater relative emphasis might be justified on maximizing the
contributions to civilian production and productivity of noninvestment
and nonindustrial segments of the economy such as government services-
and services in general. Instead of seeking to achieve a sufficiently large
inflow of capital inputs to produce mechanically the high level of savings
and investment required for self-propelling growth, emphasis would be
put on bringing about such "modernizing" changes in attitudes and moti-
vation as would lead to a continuing rise in productivity at whatever
level of capital inputs it proved possible to achieve. One such technique
would involve the teaching of secondary or jointly useful skills and atti-
tudes in nonindustrial areas, like the military, which would later be useful
in industry-and then maximizing the flow of people with these skills and
attitudes into industry.
A major policy implication of the study derived from the finding that
the economic effects of a defense program might depend as much on its
composition as its size. Substantial variations existed among the various
programs in the share of the total allocated to manpower training, im-
ported equipment, domestically produced equipment, etc. Clearly pro-
grams that absorb large amounts of foreign exchange for weapons pur-
chases or which utilize large amounts of domestic resources for indigenous
weapons procurement-particularly if the weapons are of advanced and
sophisticated types-absorb financial or physical resources particularly
strategic for development, and thus heavily burden growth. On the other
hand, programs that offer civilian-utilizable training to a large number of
men and release these men early in their working lives back into the civilian
labor force may make a significant contribution to the productivity of the
civilian sector.
For this reason, military-assistance programs that provide military
equipment free or at greatly reduced costs may reduce the adverse growth

279
Economic Developmentand Cultural Change

effects of defense on the recipients by making their own defense programs

less capital intensive and more labor intensive-a type of specialization
conducive to development. While indigenous weapons production may
provide some useful industrial learning experience, the benefits are likely
to be less than from other types of subsidized industrialization, because
the secrecy and political intervention usually involved in weapons develop-
ment and production reduces the diffusion of the industrial know-how
that is acquired and shields the continuation of inefficiency and high costs.
Civic-action programs which deliberately utilize military forces and
equipment for secondary uses on behalf of civilian economic objectives
would also appear to have important potential economic benefits, whatever
their possible political difficulties. As the defense function becomes
increasingly one of deterrence rather than actual fighting, more resources
are stockpiled in the military sector than are currently requiredfor combat.
Their overhead cost may be reduced by finding secondary and temporary
uses for them which serve other objectives-such as development.
Because of the importance of the composition as well as the size of
defense programs, there would appear to be considerable scope for a
closer integration of defense planning with economic planning, which in
most developing countries have so far very little to do with each other.
The introduction of standards of economic efficiency and cost-benefit
analysis into defense planning might render substantial economic benefits,
often with positive improvements in security as well. The introduction of
this type of thinking into defense planning would also improve the climate
for later arms-control agreements among developing countries themselves,
which so far have been conspicuously lacking.

280