Escolar Documentos
Profissional Documentos
Cultura Documentos
A Closer Look*
PAUL F. DIEHL
Correlates of War Project, University of Michigan
The relationship between arms races and war is a critical consideration in both peace research and strategic
planning. This study reconsiders the work of Michael Wallace which has postulated that arms races
significantly increase the probability of a serious dispute escalating to war. A critique of Wallace’s coding
procedures and arms race index precedes an attempt to replicate his findings. In the replication, serious
disputes, taken from the Correlates of War Project, among major powers during the years 1816-1970 serve
as the population to be tested. Adjustments in coding and index construction from the Wallace work are
made. It was discovered that only 25% of those disputes preceded by a mutual military buildup escalated
to war, while almost 77% of the wars in this population were preceded by periods lacking armaments
competition. Controls for inter-century differences and unilateral military buildups failed to alter this
apparent lack of a relationship between arms races and dispute escalation. Differences with Wallace’s study
are analyzed and the implications for peace research discussed.
Conventional wisdom has always presupposed tended the pioneering ideas of Richardson
a link between rapid military buildups and (1960), studying the impact of arms races and
war. The old dictum ’if you want peace, military spending decisions on the outbreak
prepare for war’ offers one perspective on of war. The most interesting work in this
the inter-relationship of military spending area has been that of Michael Wallace. He
and the outbreak of conflict. The spiral model used early Correlates of War (COW) Project
is indicative of a more dangerous connection compilations on major power military ex-
between increasing weapons and war. Whether penditures and serious disputes to investigate
the effect is deterrence or provocation, a nation behavior in conflict generated sit-
nation’s decision to significantly increase uations. In a widely quoted article, Wallace
its military capability could be an important (1979) concluded that the presence or absence
factor in the understanding of interstate of an arms race between two rivals correctly
war. predicted war/no war outcomes in over 90%
Despite the central nature of military of the serious disputes studied. Those results
spending in national security decision-making, are summarized in Table I.
empirical researchers have generally ignored A later study by the same author (Wallace
its possible effect on the initiation of war. 1982), using the same data base, served to
This void in the academic literature noted by reinforce this strong association between arms
Singer (1979) in 1969 remains large today. races and war. The general paucity of alter-
Nevertheless, some recent efforts have ex- native investigations makes Wallace’s studies
the most definitive to date.
*
An earlier version of this paper was presented If Wallace’s findings are correct, the im-
at the Annual Meeting of the International plications for policymaking on arms limitation
Studies Association-South, Atlanta, 4-6 November are clear. The START negotiations must
1982.
The author would like to thank Michael proceed with all deliberate speed, lest a clash
Champion, Miroslav Nincic, Peter Wallensteen, between the superpowers should escalate to
J. David Singer, Bradley Martin and Bruce all-out war. However, certain methodological
Russett for their comments and suggestions. problems cast doubt on the validity of Wal-
In addition, gratitude is expressed to Mary
lace’s conclusions. It is the purpose of this
Macknick and Louis Erste for technical assistance
in this project. paper to detail these difficulties and retest
206
Table I
Wallace’s ’arms races and escalation’
occurred.’ As a result, the strength of the dispute. Yet, it is difficult to believe that
arms race-war relationship stems not from an this brief war was anything but a result of
abundance of distinct cases of dispute the hostilities associated with World War II.
escalation, but merely is a function of To suggest that an arms race in the late 1930s
a coding decision. Wallace’s (1980) response exercised any influence on the outbreak of
to this problem was to re-evaluate his results war in this dispute five years later is premature
using only formal alliance patterns to combine without additional research and runs contrary
certain sets of disputes. This only partly to accumulated historical opinion on the
solved the difficulty of numerical inflation subject.22
of disputes and wars. The follow-up analysis Wallace (1979:8) justifies his choice of cases
failed to consider situations which share and the dyadic coding method by stating:
similar characteristics to those involving
formal alliances, but merely lack a signed ’In the case of those wars which involved more
instrument between the parties. than two powers, each dyad is coded separately.
A related problem is Wallace’s inclusion Thus, for example, World War II is coded as an
initiation of Franco-German and Anglo-German
in his population of serious disputes those hostilities in 1939, an Anglo-Italian and Franco-
cases which were not independent of ongoing Italian outbreak in 1940 and a Russo-German and
wars. In some ways, this explains the fact Japanese-American conflict initiated in 1941. This
was done to avoid the practical and conceptual
that the two World Wars account for over
difficulties of aggregating military capabilities
80% of the explanatory capability in his study of nations entering the conflict at different times.’
(Weede 1980). Serious problems are inherent
207
This explanation is open to criticism as the weakened if the threshold is lowered to 50.00.
disputes involving different actors in different Then, ten additional cases would be contrary
years are coded as separate disputes anyway to the escalation model, while its strength
according to COW criteria. If the disputants would not be enhanced by even a single case.
entered the dispute in the same year (thereby While Wallace’s striking findings have
determining the same data points for military dominated this topic area over the past few
expenditures the ten years prior to that year
-
war are indistinguishable from those of arms induced spending. Dispute spending patterns
increases. To include those cases might yield tend to reflect significant spending increases
a false indication of the real effects military whether war results or not. This is not sur-
buildups have on the initiation of war. prising as nations seek to ensure security in
It is all but impossible to accurately estimate a crisis situation.
war-time military expenditures for a nation Furthermore, the index measures only
whose whole economy is devoted to the war spending trends in the immediate past of the
effort. Accordingly, COW treats expenditures dispute. Military expenditures tend to show
during the two World Wars as missing data. greater variation as one moves farther back
Due to this data limitation and the nature in a time-series. Too often, studying a time
of the arms race indicator, cases independent period of five or ten years will cause an
of ongoing wars from 1915-1920 and 1940- overlap into a war period or time frame in
1947 are necessarily eliminated.4 Idiosyncrasies which other disputes influenced expenditure
in the data set resulted in the elimination of patterns. Focusing on the three years prior
another caste.5 Overall, the analysis here to the dispute allows consideration of behavior
considers 86 separate disputes.66 which is more perceptually important than
A measure of mutual arms buildup must comparable spending decisions ten years
reflect significant military increases for both before. Rapid changes in military expenditures
sides in a dispute and yet be able to detect are warning signals not only for peace
instances where only a unilateral buildup is researchers, but are perhaps one of the
present. In addition, an appropriate threshold indicators used by foreign policy elites to
point must be chosen to differentiate between ascertain a large scale military buildup by
incremental spending patterns and those which an opponent.
are abnormally high. These two considerations In this study, I have chosen to designate
were judged to be lacking in the original any instance of both dispute sides increasing
Wallace article. their military expenditures at a rate of 8%
With this in mind, an index of military or greater for the three years before the
growth for each side in a dispute will be dispute as a ’mutual military buildup’. This
constructed from the newly revised COW threshold was chosen because it most perfectly
file on military expenditures (an earlier captured the dividing line between incremental
version was used by Wallace in his work). increases in military outlays and abnormal
This index is the mean rate of change in spending increases. This coding decision allows
military expenditures (expressed in common the World War I dispute to be classified as
currency and controlled for price fluctuations) a mutual military buildup as is the concerted
for each side in the three years prior to the opinion of most historians. In no case,
initiation of the disputes7 however, are both sides in such a buildup
Only expenditures prior to the initiation of below a 1007o threshold often cited as an
the dispute are analyzed. This precludes indicator of high level military spending
consideration of military spending which increases.88
was reactive to the dispute itself. In this The operational definitions of a ’major
way, the index is able to isolate the effects power’, ’serious dispute’, and ’war’ developed
of an arms buildup from the bias of dispute- by Singer & Small (1972) and used by Wallace
are retained in this study.
Table II
Mutual military buildup and escalation
209
about the impact of mutual arms buildup penditures by one and only one side using
on the outbreak of war. There were five the same index construction as before) affects
other disputes prior to World War II which the probability of a dispute escalating to
were preceded by this type of buildup and war is considered in Table III.
yet did not escalate to war. This suggests There seems to be no basis for concluding
that the cases which support the escalation that a unilateral military buildup prior to a
model might only be the product of a spurious dispute increases the chances of war. A Yule’s
association. Q value of -.35 suggests a possible negative
Overall, it appears that most serious association between unilateral buildups and
disputes do not involve previous dual military war but the Chi-square value indicates that
spending increases and most serious disputes the association is not statistically significant.
do not escalate to war; but there does not
seem to be any connection between these Towards a convergence of findings
facts. The Yule’s Q value is .36 indicating The findings presented here are quite contrary
a much weaker positive relationship than to Wallace (1979). This lends itself to a
Wallace reported. However, the more con- number of possible explanations. Immediately
servative 0 coefficient is only .11 and the the differences in arms indices come to mind.
Chi-square value is not significant at any However, Wallace’s cardinal spline function
meaningful level. is heavily weighted toward changes in military
It is possible that this analysis, aggregated expenditures in the three or four years prior
over a two-century period, may hide a relation- to the dispute, much as the index used in
ship that is present in only a portion of this study. In applying the Wallace measure
Table 111
Unilateral buildup and escalation*
*
Table includes only those cases which fail to meet the criteria for a mutual military buildup.
210
to the data set used in this study, similar finds no significant joint arms increases
findings canbe reported where coding rules occurring.
between the studies were not in conflict. Where Differences in military expenditure figures
differences did exist, the variation can be might account for opposite conclusions in a
explained by reference to other factors beside few cases. Wallace replaced some interpolated
index construction. Thus, I conclude that the data points with his own estimates of military
differences in findings are not attributable appropriations. This may explain why dif-
to differences in the military spending indices. ferent results are obtained in the 1866 dispute/
Another hypothesis is that differences war between Germany, Italy and Austria.lo
between the two studies’ data sets led to While revision of research files is to some
divergent findings. Wallace’s list of disputes degree a continuing process, it is presumed
was only in its early stages of completion (pending comparison) that the data used in
when his article first appeared. The population this study are more complete and accurate
of serious disputes used in this replication than previous compilations.
attempt represents a more recent and complete Most of the remaining conflict in the aggre-
version of that list. In comparing the two gate findings of the two studies can be traced
versions (prior to any coding decisons), the to differences in coding procedures. Ten
disparities do not seem to be extraordinary, cases which were not independent of ongoing
at least not to suggest radically different wars, yet exhibited covariation of spending
conclusions. The newer data set includes a increases and escalation, were eliminated in
few more cases of pre-World War II disputes this study. In addition, the non-dyadic coding
which were preceded by dual arms increases method used here resulted in the collapse of
but did not escalate to war. The Korean ten cases, which fit the escalation hypothesis,
War, actually preceded by spending cutbacks, into three integrated disputes. In each case,
is another instance of a dispute not covered the two World Wars account for almost all
by Wallace. Most of the other changes in the instances. In effect, the strength of the
the new file are additions or deletions of arms race-war relationship cited by Wallace
’no buildup-no war’ disputes. As a whole, rests heavily on the two World Wars. The
the empirical validity of the escalation relationship seems absent in any other cir-
hypothesis is weakened when tested with the cumstance and gains statistical significance
updated file, but the changes alone are in- only through an artificial division of an
sufficient to reject Wallace’s conclusions. integrated situation.
Beyond simple changes in the number of
cases, the new file contains some corrections. Conclusions
The Russo-Japanese War of 1904 had its This study retested Wallace’s (1979) findings
dispute beginning in 1903 according to the latest that a mutual military buildup between major
file. This is an update from the Wallace powers increased the probability of a serious
report that the dispute began in 1904. Wallace dispute escalating to war. Using a modified
considers spending increases in this dispute set of assumptions and indicators, it was
through the actual first year of the dispute discovered that only one-fourth of the disputes
(1903). It is not surprising then that he con- preceded by mutual military buildups resulted
cludes that an arms race took place before the in war, while ten of thirteen wars occurred
war. However, according to the corrected in the absence of joint arms increases by the
files, one might infer that spending in 1903 dispute participants. Therefore, it was con-
was reactive to the dispute and that the cluded that mutual military buildups did not
Wallace index would yield a false indication exercise any general impact on the initiation
of prior military competition. Looking only of war under the limited conditions studied.
at military spending patterns prior to the This lack of a relationship between military
beginning of the dispute (pre-1903), this study spending and dispute escalation remained
211