Interregional Inequality and Civil Conflict: Are Spatial Disparities a Threat to Stability and Peace?

Abstract

This article examines the link between interregional inequality and civil conflict in a panel of 125 countries over the period 1993–2013. The results show that the level of interregional inequality has a positive and statistically significant effect on the incidence of civil conflict, which implies that countries with higher regional income disparities are more likely to experience internal violence. This result is not driven by a specific group of influential countries and is robust to the inclusion in the analysis of a substantial set of covariates that may affect both interregional inequality and civil conflict. Likewise, the observed link between regional income disparities and internal violence does not depend either on the estimation strategy or the measures used to quantify the degree of interregional inequality and the incidence of civil conflict within the various countries. These results suggest that policies designed to decrease the magnitude of regional income disparities may contribute to reducing the incidence of civil conflict.

Keywords:

• Civil conflict
• interregional inequality

Acknowledgements

The author is grateful to one anonymous referee for her/his useful comments and suggestions to an earlier version of the paper.

Notes

No potential conflict of interest was reported by the author.

1 See Blattman and Miguel (2010) for a review of this literature.

2 The sample considered by Lessmann (2015) is mainly composed by middle- and high-income countries. Nevertheless, this author shows that the observed association between interregional inequality and conflict holds when he uses an alternative cross-section regional data-set for the year 2005, which covers 98 countries.

3 Lessmann and Seidel (2017, 128–131) perform various robustness tests that confirm the reliability of their regional income predictions based on luminosity data. In particular, the results of these authors suggest that their predictions are more appropriate for an analysis of real income differentials within countries than observed income data, which are usually based on nominal values.

4 The definitions of all the control variables and their sources are presented in the Appendix.

5 At this point, one may consider the possibility of including country fixed effects in our baseline specification. However, controlling for country fixed effects is not appropriate in our case, as most of the variation experienced by the key independent variable, the measure of interregional inequality, is cross-sectional rather than over time (Ezcurra and Rodríguez-Pose 2013). In fact, the information provided by an ANOVA model shows that 96% of the variation in the population-weighted coefficient of variation is due to differences across countries. As pointed out by Partridge (2005, 371–372), the employment of country fixed effects in this context may produce inaccurate results. Nevertheless, in all regressions throughout the article, the robust standard errors of the estimated coefficients are clustered at the country level.

6 A higher absolute value for this ratio means that the additional control variables included in the full specification, relative to the restricted ones, are not sufficient to explain away the estimated coefficient on the measure of interregional inequality in column (6) of Table 1, so that this coefficient cannot be entirely attributed to omitted variable bias unless the amount of selection on unobservables is much larger than that on observables.

7 In this regression the time-varying regressors (including the measure of interregional inequality) enter as their respective means over the previous five-year period.

8 Note that $GE(\alpha )$ can be transformed into a subclass of the commonly used Atkinson index with $\epsilon =1-\alpha$ for $0\le \alpha <1$, where $\epsilon$ is the (relative) inequality aversion parameter (Lessmann and Seidel 2017).

Additional information

Funding

This research has benefited from the financial support of the Spanish Ministry of Economy and Competitiveness [Project ECO2015-64330-P].