This article summarizes an empirical investigation of the link between poverty and the incidence of civil war. The thesis examined is a simple one: at the individual level, low living standards tend to enhance militaristic nationalism and confidence in the armed forces and therefore, on average, more support for attempting military solutions to social conflicts. At the national level this means that, assuming central governments need some popular support in order to start and sustain military campaigns, poorer countries will tend to attempt a military solution to a wider range of conflicts than would richer nations. Thus, the oft reported correlation between low GDP per-capita and civil war incidence is partly explained through the effects living standards have on the attitudes of the population and the resulting effect on the ability of governments to sustain a military-solution approach to conflicts. This claim involves a certain shift of emphasis. When thinking about the relation between income-distribution and war, one often thinks about the conditions under which the exploited poor rebel against the rich. But civil wars today often appear to be fought over other issues and moreover, it is often the poor who support the government in fighting the insurgents. Thus, while studying the rebels’ motivations and decisions is obviously of major importance, the present paper sets these concerns aside in order to focus on the population from which the government side of the conflict must draw its support.

In recent years, several attempts have been made to empirically identify the sources of civil wars using cross-country data sets, mostly from the second half of the 20th century. The empirical studies typically attempt to estimate a model where the probability of the eruption of civil war in a given country at a given time interval is determined by various aggregate measures at the country level such as GDP, natural resources, ethno-linguistic fractionalization, education levels, civil liberties, democracy and inequality measures. The strongest and most robust finding in this literature seems to be that societies at low levels of economic development have suffered much more from societal warfare than prosperous societies. This seems to suggest that higher national income significantly reduces the risk of civil war. But it is equally well argued that civil wars are detrimental to economic growth: causality probably runs both ways, and it seems hard to separate the effects in a cross-section of countries. Moreover, granted that low levels of economic development are conducive to civil war, it is not quite clear why this should be the case - and cross-country data seem insufficient to answer this question. Collier and Hoeffler (2001), for example, view income per capita as a proxy for the cost of recruiting rebels: low per capita income thus facilitates conflict by making rebellion cheaper. Fearon and Laitin (2003), on the other hand, while finding a similar relationship between GDP per capita and civil wars, claim that low GDP per capita is related to “weak states,” which in turn attract insurgency. But in general - since GDP per capita is correlated with so many social, economic, political and international factors that are not easily controlled for - it is hard to point to any particular mechanism as driving the statistical relationship. In order to do that, the aggregate-level analysis probably needs to be complemented by a disaggregated one. A preliminary attempt is presented here.

Before summarizing the results I should emphasize that they may not apply to the civil wars afflicting Africa, as data on Sub-Sahara Africa is scant and save South Africa, none of the African countries that recently experienced civil war take part in the empirical analysis.

I start with the cross-country patterns. The figure shows the mean national level of confidence in the armed forces plotted against GDP per capita. The level of confidence is taken from the World Values Surveys (Inglehart et al. 2000, henceforth WVS), performed in the early and mid 1990s. Each point represents the estimated mean from one country at one year. The figure suggests that some of the observed correlation between GDP per-capita and civil wars might indeed be due to the former picking up the effect of confidence in the armed forces. This possibility was explored using three different measures of civil wars and civil conflicts, taken from Fearon and Laitin (2003), and Gleditsch, Wallensteen, Eriksson, Sollenberg & Strand, 2002. Confidence in the armed forces was measured by the proportion of the population professing the highest level of confidence in the armed forces, taken from the WVS. The two main results are as follows:
1. The prevalence of confidence in the army is strongly and positively related to the risk of experiencing civil war, even after controlling for a host of other factors that are commonly held to account for civil war risk.1
2. Confidence in the army is found to be partly responsible for the relationship between GDP per-capita and civil war incidence.
As in most cross-country regressions, a causal link between confidence in the army (or other covariates) and civil war risk cannot be established based on these regressions alone. An obvious reason is that in countries experiencing - or even expecting - war, support for the army may tend to rise, which may be reflected in the reported confidence in the army. If this is the case, it would be very hard to separate such effects from the effect these attitudes in turn have on the eruption or perpetuation of the war. But the results do suggest that part of the observed relationship between GDP per capita and civil war risk is due to widespread confidence in the army.

Still, the relationship between confidence in the army and national income levels may well be spurious. That is, it may be the case that in rich countries everyone has relatively low confidence in the army while in poor countries it is the reverse, and that this is due to some other factors affecting both GDP and attitudes. I therefore examine whether the relationship also holds at the individual level, both in rich and in poor countries. Are richer people less likely to have a high level of confidence in their country’s armed forces, regardless of whether they live in a poor or a rich country?

To answer this question I use WVS data from a reasonably diverse sample of 27 countries during the 1990s, all using the same confidence-in-the-army question. To measure living standards in a comparable way, I use income per household member, converted to 1996 dollars using the PWT 6.1 consumption-purchasing-power-parities.2 The proportion of the population with the highest level of confidence in the army ranges in these surveys from below 5% in Austria, Belgium, Latvia and the Netherlands to over 64% in Turkey. Income per household member in 1996 PPP dollars ranges from $300 to $40,000.

The results are rather striking in that in almost all countries where data are available, the estimated effect of income on confidence in the army is negative. There is no clear difference between richer and poorer countries with respect to the marginal effect of income. There appears to be a strong negative relationship between income and confidence in the army in countries as diverse as Austria, Brazil, Canada, Hungary, Portugal and Turkey. Further, the size of the effect appears to be large enough to account for half of the association between national income and confidence in the army illustrated in thefigure. This leaves plenty of room for other, national factors - such as recent conflicts - to simultaneously affect both average income and average support for the army. But the association at the individual level seems to suggest that the cross country association is not all due to such factors.

A concluding remark. Military and political leaders engaged in violent conflict often seem to devote considerable efforts to try to enhance popular confidence in their armed forces. It seems plausible to assume that such efforts are not unreasonable, in the sense that higher confidence in the armed forces can lead to higher popular support for the war being fought (or anticipated). As we have seen, the claim that widespread confidence in the army may facilitate the practice of civil war is consistent with the available data. Yet confidence in the army is not determined just by governmental propaganda. The present article tried to point to the fact that the living standards of the population may also be related to the levels of confidence they have for the army. Could it be then that the extensive popular confidence that the army enjoys in countries like Turkey and India is not just a result of the long conflicts in which they have been involved, but also a factor that prolongs them - and that such confidence is partly due to the low income levels of much of the populations of these countries? I tried to show here that this possibility is not rejected by the available data, but to be able to say more we need to identify the mechanism underlying the association of income and confidence in the army. This is the subject of an ongoing research into the economics of nationalism.

ENDNOTES
1. The controls include measures of GDP per capita, population size, proportion of land mountainous, noncontiguous states, oil exporting, instability, democracy and ethnic fractionalization. The sample of countries with both WVS and civil war data available consists of only 71 to 74 country-years, depending on the specification. The results obtained from this sample appear to be reasonably comparable to the larger-sample results in the literature, as the effects of most variables (before controlling for confidence in the army) are generally similar to those reported in the literature.
2. Since information on the income categories used in the WVS is not available for all surveys and since for a few countries no reliable PPP exchange rates exist, we are left with 31 national surveys (four countries have two surveys at two points in time) and four surveys from Spanish regions.

REFERENCES
Collier, Paul, and Anke Hoeffler, 2001, “Greed and Grievance in Civil War” World Bank. http://econ.worldbank.org/programs/library (November 18, 2002).

Fearon, James D. and Laitin David D., 2003, “Ethnicity, Insurgency, and Civil War”, American Political Science Review Vol. 97, No. 1 February 2003.

Gleditsch, Nils Petter; Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg & Håvard Strand, 2002. “Armed Conflict 1946-2001: A New Dataset”, Journal of Peace Research 39(5): 615-637.

Heston, Alan, Summers, Robert and Aten, Bettina, 2002, Penn World Table Version 6.1, Center for International Comparisons at the University of Pennsylvania (CICUP), October 2002.

Inglehart, Ronald, et al. 2000, WORLD VALUES SURVEYS AND EUROPEAN VALUES SURVEYS, 1981-1984, 1990-1993, and 1995-1997. ICPSR version. Ann Arbor, MI: Institute for Social Research, 2000.

Moses Shayo is a graduate student in economics at Princeton University.

This article is an excerpt from the prize-winning essay in the fifth Isaac Roet Essay Competition. This year’s topic was “The Distribution of Wealth and Income: A Question of War and Peace.” Mr. Shayo received the prize of €5000. All finalists received a free one-year membership in ECAAR.