it determines the capacity to mobilize and possibly arm people. We will see that this notion
of “wealth as capacity” — as opposed to the idea of “wealth as loot” — generates a more nu-
anced and qualified relationship between inequality and conflict. Specifically, as a given group
becomes wealthier (and depending on the way this increased wealth is distributed within the
group) it might more easily be tempted to expend part of the additional resources into fighting
the opponent party. While this outcome is independent of whether the group in question is eco-
nomically weaker or stronger to start with, its effect on the opponent depends crucially on this
feature. Specifically, we show in Proposition 2 that if the weaker group expends more resources,
so will the stronger group, leading to an escalation of conflict. The opposite is true if the initial
impetus comes from the stronger group: conflictual activity is deterred for the opponent. It is
possible, then, that the closing of the wealth gap between two groups — rather than its widening
— might ignite conflict instead.
This latter point seems compatible with the abundant evidence — see Melson and Wolpe
(1970), Olzak and Nagel (1986) and Tellis, Szayna and Winnefeld (1998) — that economic mod-
ernization fuels rather than moderates ethnic conflict. The process of modernization might gen-
erate resources to fundamentalist segments (or cynical opportunists) which would then be chan-
nelled into financing (a thus far latent) conflict.4
Indeed, under the approach we take, a balanced growth of wealth within a group is less potent
in generating conflict than a wealth increase that accrues to the e´ lite in that group (see especially
Propositions 3 and 5). Such disequalizing wealth changes, we argue, are more likely to trigger
greater activism for the group in question. [As we’ve already said, whether this leads to es-
calation or deterrence depends on whether this group was economically weaker or stronger to
start with.] Our approach is therefore in line with Bates (1999), who emphasizes within-group
inequality as a potential source of increased conflict against the opposing ethnic group. The
emergence of an economic and cultural e´ lite appears as a critical factor in substantial escalations
in many ethnic conflicts in Africa. They provide the leadership and the means that facilitate the
escalation. Thus, accordingly with Bates’ arguments conflict will be higher the more uneven is
the distribution of the benefits of modernization within each rival group. This view is also shared
by Horowitz (1997, pp 439 and 457) who stresses the fact that the e´lites hold the same or even
more hostile views. Having more resources does not necessarily make people more moderate.
Disequalizing wealth changes are more dangerous precisely because they put resources in
the hands of potential contributors, while at the same time they do not increase the cost of mobiliz-
ing activists. By increasing the opportunity cost of conflict all round, balanced wealth changes
dampen this phenomenon. Thus our findings are perfectly consistent with the findings of Col-
lier and Hoeffler (2002), Fearon and Laitin (2003), Miguel, Satyanath and Sergenti (2004) and
others that overall income increases are negatively related to conflict.
It may be worth relating these observations to the nature of the link between polarization and
conflict. Esteban and Ray (1994) axiomatize a measure of polarization based on the intuition that
“polarized societies” have clusters of individuals that draw their strength from a sense of within-
group identification, but also display substantial antagonism across such groupings. Within this
framework a heightening of within-group “similarity” boosts the sense of group identification
and hence aggregate polarization, assuming, of course, that there is more than one group.
This intuition appears to run against the results we obtain in this paper, as within-group
income heterogeneity has the effect of increasing the severity of conflict. Let us discuss this ap-
parent contradiction in some detail. One of the novelties in the present paper is the modeling
4See Bourguignon (1998) for a careful quantitative analysis of the realtionship between growth, inequality, and
conflict.