Measure for Measure: Concept Operationalization and the Trade Interdependence-Conflict Debate
| DOI | 10.1177/00223433030405004 |
| Date | 01 September 2003 |
| Author | Quan Li,Erik Gartzke |
| Published date | 01 September 2003 |
| Subject Matter | Journal Article |
553
Introduction
Recent dyad-level quantitative studies of the
consequences of trade for international
conflict behavior report what appear to be
incompatible results. Barbieri (1995, 1996,
1998b) finds that bilateral trade increases the
probability of militarized interstate disputes
(MIDs). Other research makes the opposite
assertion, offering evidence that bilateral
trade reduces the likelihood of MIDs in econ-
omically liberal dyads (Bennett & Stam,
2000; Gartzke, Li & Boehmer, 2001; Oneal
et al., 1996; Oneal & Ray, 1997; Oneal &
Russett, 1997, 1999a,b; Russett, Oneal &
Davis, 1998). The source of contradictory
findings is a subject of considerable debate,
and has been identified variously as dispari-
ties in data collection, econometrics, model
specification, control variables, and the
choice of temporal and spatial domain
(Barbieri & Schneider, 1999). We argue that
the disparity in findings can be at least partly
explained by features inherent to the variable
constructions used by the competing
approaches. How researchers construct
measures of dyadic interdependence can help
to determine what they find in empirical
© 2003 Journal of Peace Research,
vol. 40, no. 5, 2003, pp. 553–571
Sage Publications (London, Thousand Oaks, CA
and New Delhi) www.sagepublications.com
[0022-3433(200309)40:5; 553–571; 035589]
Measure for Measure: Concept Operationalization
and the Trade Interdependence–Conflict Debate*
ERIK GARTZKE
Department of Political Science, Columbia University
QUAN LI
Department of Political Science, The Pennsylvania State University
While most quantitative studies find a negative relationship between economic interdependence and
interstate disputes, research by Barbieri finds that interdependence precipitates conflict. Participants in
the debate suggest several causes, but we show that alternative variable constructions are sufficient to
account for the discrepant findings. A simple formal equivalence unites respective operationalizations
of dyadic interdependence used by Oneal & Russett (trade dependence, tradeij/GDPi) and Barbieri (trade
share, tradei/tradei) with the consensus construction of monadic trade openness (tradei/GDPi). We also
show that Barbieri’s trade share is negatively correlated with openness.Arguments in the article are
verified through large-sample quantitative regression analyses of the two competing dyadic variable
constructions and trade openness on MID onset. The results of these dyadic regression analyses show
that trade share increases the probability of MID onset, trade dependence decreases the probability of
MID onset and, correspondingly, that trade openness is negatively correlated with MID onset.
*Equal authorship implied. We thank Han Dorussen,
John Oneal, Bruce Russett, and several anonymous referees
for comments and suggestions on earlier versions of this
article. We also thank John Oneal and Katherine Barbieri
for help with data. Alex Braithwaite provided valuable
research assistance. The authors can be contacted by e-mail
at: gartzke@columbia.edu and quanli@psu.edu. Data
employed in this article can be obtained at http://www.
prio.no/jpr/datasets.asp. Statistical analysis is conducted
using Stata7.
04 JPR 40-5 Gartzke (JB/D) 23/7/03 8:37 am Page 553
analyses of the relationship between trade
and conflict.
Interdependence really is complex.
Economic linkages involve interactions at
the systemic, dyadic, and national levels that
are by themselves intricate. The political per-
mutations of economic linkages further
amplify complexity (Mansfield & Pollins,
2003). How economics matters to political
processes is ultimately an empirical question,
but it is a question with significant analyti-
cal prerequisites.1Since researchers cannot
actually see the value politicians place on the
economic relationships that span borders, we
must guess at the impact these relationships
have on political decisionmaking.2Dyadic
analysis of trade–conflict relationships
requires an explicit interpretation of the
nature of this political impact. One approach
to the question is to ask whether a given
linkage is valuable, relative to other trade
relationships (Barbieri). Another approach is
to assess whether a given trade relationship
matters, relative to a state’s overall economic
performance (Oneal & Russett). These two
operationalizations of trade interdependence
tap into different dimensions of the under-
lying multidimensional concept. Yet, if each
measure relates to the same fundamental
process, then the measures must also be
related to each other. The relationship
between measures of dyadic interdependence
is an underexplored and potentially import-
ant element in how conflict interacts with
economics.
We examine the relationship between the
two major competing dyadic measures of
interdependence in this article. After review-
ing the two competing measures offered by
Barbieri and Oneal & Russett, we show how
each is compatible with the other and con-
stitutive of the same core concept. We also
show that the effect of one measure (trade
share) is inversely related to the consensus
measure of openness (monadic trade inter-
dependence), capturing the disconnected-
ness from the world economy or dependency
aspects of trading relations. The debate in
the literature about the consequences of
economic interdependence may be due in
part to differences in variable construction.
Explicating issues in the debate, and evalu-
ating competing claims, depends on under-
standing the underlying identity uniting
formulations of dyadic trade interdepen-
dence. We then employ our conceptual
reasoning to compare empirical results based
on these measures. Statistical results substan-
tiate our theoretical claims. Finally, we offer
a summary and some conclusions from our
results.3
Debating Multiple Measures of One
Concept
Several indicators have appeared in the litera-
ture to measure trade interdependence and
its consequences for dyadic conflict
(Polachek, 1980; Gasiorowski, 1986). Two
of the most discrepant ongoing research
programs, in terms of their findings about
the effect of trade interdependence on the
same dependent conflict variable (militarized
interstate disputes, MIDs), are based on the
different variable constructions offered by
Barbieri (1995, 1996, 1998b) and Oneal &
Russett (cf. 1997, 1999a,b; there are other
disparities between these two research
programs including different data). Table I
illustrates the steps for constructing each
dyadic measure. Barbieri’s composite variable
uses the proportion of bilateral trade to each
journal of PEACE RESEARCH volume 40 / number 5 / september 2003
554
1Mansfield & Pollins (2003) and McMillan (1997) offer
reviews of the literature.
2One approach to better approximating the value of trade
linkages is to examine price elasticities. See Crescenzi
(2003), Polachek (1997), Polachek & McDonald (1992),
and Polachek, Robst & Chang (1999).
3Oneal (2003) also explores aspects of the mathematical
identity among the three measures, such as the inverse
relationship between trade share and openness. The two
studies developed independently.
04 JPR 40-5 Gartzke (JB/D) 23/7/03 8:37 am Page 554
To continue reading
Request your trial