Youth bulge and civil war: Why a country’s share of young adults explains only non-ethnic wars

Abstract

Scholars agree that young men carry out most acts of political violence. Still, there is no consensus on the link between relatively large youth cohorts and the onset of violent, armed intra-state conflicts. In this paper, we examine the effect of youth bulge, a measure of the relative abundance of youth in a country, on the onset of two different types of civil wars—ethnic and non-ethnic wars. Building on and extending three datasets used by other scholars, we theoretically argue and empirically substantiate that, as a result of the negative effects of youth bulge on the economic conditions of the youth cohorts in the country, youth bulge affects the onset of non-ethnic wars, but not the onset of ethnic wars. Possible implications and directions for further research are then suggested.

Keywords

Civil war ethnic conflicts non-ethnic conflicts youth bulge

Young people have fueled discontent and revolutions everywhere (e.g. Moller, 1968). In one of the latest cycles of worldwide discontent beginning in 2011, many people, mostly young, took to the streets to protest ruling regimes, growing economic gaps and general political and social grievances and concerns. Some observers went so far as to argue that the waves of often violent protests in several North African and Middle Eastern countries, referred to collectively as the “Arab Spring”, are, inter alia, the result of the relatively large numbers of young adults in these countries (Hvistendahl, 2011). Indeed, the average share of young adults (ages 15–29) in North African and Middle Eastern countries in 2010 was 42%, higher than 35%, the world’s average that year (World Population Prospects, 2010).¹

Previous studies have already claimed that young men perpetrate most acts of political violence (e.g. Mesquida and Weiner, 1999). Some have argued that the relative abundance of youth, a concept frequently referred to as “youth bulge”, underlies many social upheavals and revolutions (e.g. Moller, 1968), and it has even been suggested that the “youthfulness” in the Muslim world is one of the key drivers of a potential “clash of civilizations” (Huntington, 1996: 117–119). A rise in the proportional size of a cohort of young people tends to strain state and society, with particular impact on the educational system and the labor market, consequentially increasing youth unemployment and lowering the group’s real and expected income in comparison to other cohorts (Bloom et al., 1987; Easterlin, 1987; Korenman and Neumark, 1997; Macunovich, 2000). In short, “other things constant, the economic and social fortunes of a cohort … tend to vary inversely with its relative size” (Easterlin, 1987: 1).

Moreover, studies have shown that youth bulge may contribute to social upheaval through its effects on “civic knowledge” and social cohesion. Countries with large cohorts of young people are characterized by lower levels of “civic knowledge” among their young cohorts (Hart et al., 2004, 2005). Such a lower level of knowledge increases these youth cohorts’ susceptibility to the influences of simple ideologies and charismatic leadership (Hart et al., 2005: 662), making political mobilization easier. Others argue that youth bulge harms social cohesion and facilitates social unrest, as the presence of relatively large numbers of unemployed, less “controllable” young men raises societal “volatility” (Boyden, 2007: 265; Moller, 1968: 254–258).

It seems, therefore, reasonable to expect that dire economic conditions and the low opportunity costs associated with joining a rebellion (Collier and Hoeffler, 2004: 569; Elbadawi and Sambanis, 2002: 309) would significantly ease the recruitment into rebel organizations of alienated young people attracted to simple ideologies “promising perfection in a hurry” (Moller, 1968: 241). This would be more so in countries struggling, unsuccessfully, to meet the demands for education, employment and status of a growing cohort of young adults seeking to secure resources that would enable them to transition to adulthood, that is, marry and establish their own households (Mesquida and Wiener, 1999: 181–183; Urdal, 2004: 2–3).

The youth bulge phenomenon and its potential consequences are attracting the growing attention of public figures, decision-makers (Hvistendahl, 2011) and scholars (Staveteig, 2005). The link between youth bulge and armed intrastate conflicts, however, has not yet been firmly substantiated. In fact, empirical evaluations of the effects of youth bulge on conflict have often showed contradicting results. One study, for instance, finds that youth bulge increases the risk of low-intensity conflicts but not high-intensity ones (Urdal, 2006). Similarly, others demonstrate that youth bulge does not affect the likelihood of civil wars (e.g. Fearon and Laitin, 2003a), while others argue that youth bulge increases the intensity of a conflict once it erupts (Mesquida and Wiener, 1999).

Empirical inconsistencies, we argue, do not necessarily attest to weakness of the theoretical contention; rather they could be the result of a common practice of studying violent conflicts in an overly inclusive way—by collapsing all civil wars into one group instead of theorizing about and studying factors that are more likely to affect one type of war but not another (for an exception, see Sambanis, 2001). Diverging from this common practice, we reason why youth bulge should increase the likelihood of non-ethnic wars but should have no effect on the onset of ethnic wars.

In the next section we develop the theoretical justification for our hypotheses. The second section describes methods, including data and measures. This is followed by the presentation of the main results and then discussion. In the concluding section, we address possible implications of the findings and directions for further research.

Theory and hypotheses

Does youth bulge increase the likelihood of armed conflicts and civil wars? We have already alluded to the fact that studies addressing this puzzle have not reached a firm conclusion (cf. Hegre et al., 2013: 255). Several studies demonstrated that youth bulge increases the likelihood of low-intensity internal armed conflicts (Cincotta et al., 2003: 44–49; Staveteig, 2005; Urdal, 2004, 2006), that is, conflicts with at least 25 battle-related deaths per year (Gleditsch et al., 2002). Others have found that youth bulge does not increase the risk of civil wars (Collier and Hoeffler, 2004; Collier et al., 2009; Fearon, 2011; Fearon and Laitin, 2003a; Urdal, 2006), civil wars being defined as high-intensity internal conflicts that caused at least 1000 battle-related deaths (Sambanis, 2004a). Such results seem to be at odds with evidence suggesting that youth bulge is associated not only with some forms of conflict onset but also with its intensity (Mesquida and Wiener, 1999; cf. Hegre et al., 2013: 260).

These conflicting results could be due to the operationalization of the dependent variable: the aggregation of intense armed conflicts of various types into one civil war category. Indeed, there may be some good theoretical reasons to lump ethnic and non-ethnic wars together, specifically in studies seeking to tap economic motivation common to all intrastate conflicts (e.g. Collier and Hoeffler, 2004; Fearon and Laitin, 2003a).

However, there are also good reasons to think of ethnic and non-ethnic wars as quite different, even distinct, phenomena. In fact, one study of the origins of ethnic vs non-ethnic civil war has shown that economic indicators correlate strongly with the outbreak of non-ethnic civil wars (Sambanis, 2001: 276–280), while ethnic civil wars are mostly driven by political and other collective grievances, and much less by economic factors (Sambanis, 2001: 273–280).²

Ethnic wars, according to some, erupt as a result of fear: when members of an ethnic group fear assimilation or the actual physical annihilation of the group (e.g. Horowitz, 1985: esp. chapter 4; Kaufmann, 1996; Lake and Rothchild, 1998: 7–8). Others assert various other motivations for ethnic conflicts, such as achieving self-determination or regional autonomy, favorably changing group position in government or ending discrimination (Cederman et al., 2013: chapter 4; Wimmer et al., 2009a). Pursuing a similar logic, Cederman and colleagues contend that horizontal economic inequality—that is, group-based inequality—increases the propensity for violence of relatively poor ethnic groups, as these ethnic groups perceive themselves as being disadvantaged compared with other ethnic groups in their countries (Cederman et al., 2011, 2013: chapter 5).

Collectively, these studies argue that mobilization to war is driven mainly by a group’s desire to improve its conditions by struggling to change the current status quo. In such a context, fear or hope might help mobilize group members, old and young, to defend, violently if necessary, the future of the group. An opportunity-driven “economic logic” may not be the key factor driving recruitment into combat, as individuals may be eager to fight to preserve their ethnic identity (Sambanis, 2001: 266–267).³ Ethnic conflicts and wars, in other words, involve violent mobilization along ethnic lines, as “the mobilization base for ethnic war is clear and defined by ethnic identity” (Sambanis, 2001: 266; see also Wimmer et al., 2009a: 326; cf. Fearon and Laitin, 2003a: 79), and such mobilization is mostly intended to improve the conditions of an aggrieved ethnic group.

In contrast, while there is no single accepted theory pertaining to the origins of non-ethnic wars, violent mobilization in these conflicts is more closely affected by issues of country-wide distribution of wealth, lack of economic opportunities and the desires of certain groups, irrespective of the ethnic affiliations of the group members, to gain economic leverage by taking control of the government or taking over the extraction of natural resources in resource-rich territories (Sambanis, 2001; see also Humphreys and Weinstein, 2008). Following this logic, it make sense to expect that economic and social distress would exacerbate tensions and raise discontent among all segments of society, irrespective of grievances of any specific group. This would be the case even more so in countries in which demographically large segments of the populations (e.g. young adults) are likely to be affected by the state of the economy, irrespective of their ethnic affiliation. A youth bulge, in other words, should increase the likelihood only of the type of war mostly influenced by country-wide economic and social conditions, that is, non-ethnic war.

It follows that youth bulge may be associated with low opportunity costs for mobilization and recruitment into war, particularly in countries failing to meet the economic and other demands of the large cohort of young people. As a result, rebels may have an ample reservoir of potential recruitees who will agree to take the risk of fighting in the hopes of gaining access to resources they need to guarantee their futures. Under such circumstances the proclivity to join a rebellion should not be in principle related to or affected by recruitees’ ethnic affiliation or their group’s grievances. A relatively large cohort of poor young men struggling to gain access to society increases, therefore, the risk of the onset of non-ethnic civil wars. Stated formally, we expect that:

H1: The higher the youth bulge in a country, the higher the likelihood of onset of non-ethnic civil war.

Turning to the association between youth bulge and onset of ethnic war, we find several scholars suggesting that, in some countries (e.g. Sri Lanka, Russia and Israel/Palestine), ethnic conflicts have been affected, among other things, by the large cohorts of young people in the group engaged in conflict (see Cincotta et al., 2003: 44–47; Huntington, 1996: 259–261; Staveteig, 2005: 16–17).⁴ Young people in these groups may be susceptible to recruitment into insurgent ethnic groups for the same reasons as discussed above—low levels of “civil knowledge” (Hart et al., 2004, 2005) and low opportunity cost.

There are, however, good reasons to expect that countries with youth bulge may not be more likely to experience ethnic civil wars.⁵ Ethnic insurgencies are typically motivated by cultural, political and economic conditions affecting the entire group. The fact that a country has a large cohort of young people may not by itself increase the motivation of members of any one ethnic group to engage in conflict, as group conditions, that is, the specific reasons that drive support for an insurgency, may not be related to the distribution of ages in society at large. Consider, for instance, an ethnic group with a disproportionately large young cohort. This large share of young people in and of itself may have only a marginal effect on this cohort’s economic situation, which is much more likely to be a result of top-down discrimination and marginalization of the entire ethnic group. In addition, a large share of young people in any one group may have only a small or no effect on the social and economic conditions of the young cohort in the entire country, especially if the group in case is small. Lastly, such poorer economic conditions of youth cohorts are most likely to be orthogonal to specific ethnic grievances, as the former relate to all young adults in the country, cutting across ethnic groups (e.g. Macunovich, 2000). Put differently, group grievances may be the result, for example, of exclusion from political power (e.g. Wimmer et al., 2009a) or structured horizontal inequalities among groups in a country (Cederman et al., 2011), and be only marginally related to the conditions of the youth cohort in the entire country.

True enough, attempts to mount an ethnic insurgency may be facilitated by the presence of a large pool of (young) potential recruitees, but in fact ethnic conflicts have broken out in both poor and rich countries (e.g. Sambanis, 2001), and likewise we expect ethnic conflict to break out in societies regardless of the relative number of unemployed and poor young men. Ethnic conflicts are driven, we suggest, by concerns that are not unique to one age cohort or another; group grievances cut across all age cohorts, irrespective of the breakdown of ages in the specific ethnic group or in the entire country. Therefore, unlike non-ethnic conflicts, where older cohorts might feel that they stand only to lose from young men’s struggle to gain resources, ethnic conflicts are likely to find support more evenly distributed among younger and older cohorts, as members of all generations aim to preserve their ethnic identity and/or improve the group’s social, economic and political status. Accordingly, we expect that ethnic wars will erupt in countries with youth bulge as well as in countries with no youth bulge. Put formally, we hypothesize that:

H2: Country level of youth bulge does not affect the risk of ethnic civil war onset.

Methods

Data

This analysis is based on data collected annually across 163 countries from 1950 (or from a country’s first year of independence) to 2005. Only countries with a population of at least 500,000 in the year 2000 were included in the dataset,⁶ and it contains a total of 7182 country-years. The 1950 cut-off point is due to the unreliability of demographic estimates prior to that year (Urdal, 2004: 6).

Measures

Dependent variable

Three dependent variables are used in the analysis: civil war onset, ethnic civil war onset and non-ethnic civil war onset. Many scholars have coded the onset of civil wars, but only a few have explicitly separated between the onset of ethnic and non-ethnic wars. In this paper, we rely on coding used in highly cited studies by Sambanis (2000a, 2004a), Fearon and Laitin (2003a) and Wimmer et al. (2009a).⁷

Table 1 presents the number, type and geographical distribution of conflicts according to each of the three data sources. Clearly, there are some noticeable disagreements in the identification of civil wars across the three sources.⁸

Table 1.

Civil war in different regions in 1950–1999, by war-type and war coding scheme⁹

	Non-ethnic war			Ethnic war
	Sambanis	Wimmer et al.	Fearon and Laitin	Sambanis	Wimmer et al.	Fearon and Laitin
Western countries	0	0	0	1	0	1
Eastern Europe	0	1	1	11	8	8
Latin America	11	6	11	2	2	1
Sub-Saharan Africa	11	8	0	41	17 (19)	34
Asia	11	12	9	23	22 (23)	20
Middle East and North Africa	7	4	7	16	8 (9)	9
Total	40	31	28	94	57 (61)	73

In parentheses are the numbers of civil wars in Wimmer et al.’s dataset in the years 1950–2005.

Sambanis identified the onset of 94 ethnic and 40 non-ethnic wars between 1950 and 1999, and using the same time frame, Fearon and Laitin identified 73 ethnic and 28 non-ethnic wars. Wimmer et al. extended the time frame and identified the onset of 61 ethnic and 31 non-ethnic wars between 1950 and 2005 (31 non-ethnic and 57 ethnic war onsets between 1950 and 1999).¹⁰ For the most part the three datasets agree on whether a conflict should be classified as ethnic or non-ethnic; however, they disagree with regard to the type of 20 civil wars.¹¹ Therefore, we test our hypotheses on all three distinct war-coding systems separately in order to gauge whether results are robust to differences in the coding of the onset of war.

Drawing on these coding schemes, we created three dependent variables of each of the data sources: (a) civil war onset; (b) non-ethnic civil war onset; and (c) ethnic civil war onset. For each of the variables, the value 1 indicates the first year of a conflict resulting in at least 1000 battle-related deaths, and 0 otherwise.

Youth bulge: the literature approaches the measurement of youth bulge in two ways. The first is to measure youth bulge as the ratio of men between the ages of 15–24 or 15–29 to the total population (Collier and Hoeffler, 2004; Fearon and Laitin, 2003a). This approach assumes that the relative abundance of men (rather than women)—men being those who commit the majority of crime and violence (e.g. Wilson and Daly, 1985)—is the factor leading to social unrest and to a higher propensity of onset of civil war. Indeed, some researchers even suggest that “fears about youth bulge are really fears of young men” (USAID, 2010: 2).¹²

A second approach argues that a simple measure based on the share of young men in the entire population does not suffice. Such a measure is blind to the fact that in some instances a large share of men between ages of 15–24 or 15–29 relative to a younger 0–14 cohort may be indicative of a steady decline in fertility rates—a decline leading to a “demographic dividend” (Urdal, 2006: 611). It turns out that a rise in the share of people of working age and a decline in the share of infants and children that need to be taken care of can substantially improve the country’s economy (Ashford, 2007; Cincotta, 2008–2009) and reduce the risk of armed conflicts (Urdal, 2006; cf. Fearon 2011: 15–16). Accordingly, proponents of this approach measure youth bulge as the ratio of 15–24 year olds, men and women, to the adult population (15 years and older) in the country (Urdal, 2006: 615).

In developing the measure for the current study, we take note of Urdal’s latter point as well as the observation that people younger than 15 and older than 30 are less likely to participate as combatants in civil wars compared with people between the ages of 15 and 29 (Mesquida and Wiener, 1996: 247; 1999: 181–183). We also consider the argument that, to the extent that violence does break out, men are the most likely perpetrators (e.g. Wilson and Daly, 1985). Consequently, we measure youth bulge as the ratio of men aged 15–29 to the entire adult (15 years and older) male population in the country. Demographic data was taken from the online database of the World Population Prospects (2010) of the United Nations Population Division. We calculated youth bulge estimates for each country-year between 1950 and 2005.¹³ Youth bulge scores in our dataset ranged from 21.0 to 61.6, with a mean score of 42.2 (standard deviation = 7.9).

Control variables

The essence of the argument we have presented so far is that youth bulge increases the likelihood of non-ethnic civil wars because it affects the opportunity structure, particularly for young adults. Eager to guarantee their economic and social future and lacking real educational and financial prospects, young men may be more susceptible to mobilization and recruitment by rebel groups into combat in order to gain economic or political advantages. The presence of a relatively large young cohort in and of itself should not affect the prospects of recruitment in support of an ethnic cause, which is driven by the quest to ameliorate group conditions in general rather than the structure of economic and other opportunities of any one cohort. Since our main argument is that the presence of youth bulge presents rebels with an opportunity (e.g. Collier and Hoeffler, 2004) to mount an insurgency, in order to test our hypotheses we control for several competing explanations (Ray, 2003) for rebels’ opportunities.

First, we control for the country’s level of economic development, as dire economic conditions lower the opportunity costs associated with combat participation and ease the recruitment of young men to the rebels’ ranks (Collier and Hoeffler, 2004; Collier et al., 2009; Fearon and Laitin, 2003a). Moreover, a poor state is constrained in responding to the demands and needs of its citizens, thus contributing to their potential dissatisfaction and hampering the state’s abilities to handle armed challenges posed by rebel groups (Fearon and Laitin, 2003a). Here we use two measures as proxies of the level of economic development. The first is logarithmic transformation of gross domestic product (GDP) per capita in constant 2000 US$ (e.g. Hegre and Sambanis, 2006),¹⁴ which is the most commonly used proxy for level of development in the civil war literature. The second is infant mortality rate (IMR), defined as the proportion of live-born children who die before the age of 1 year (Urdal, 2006: 616; see also Hegre et al., 2013: 256).¹⁵ This is following studies arguing that IMR captures a broader set of developmental factors than standard measures of income levels, such as GDP per capita (e.g. Hegre et al., 2013: 256).

We also controlled for political regimes that are neither democratic nor autocratic. These regimes have been shown to be more susceptible to conflict (Hegre et al., 2001), as scholars suggest that such regimes indicate “political contestation among competing forces and, in consequence, state incapacity” (Fearon and Laitin, 2003a: 81). Such incapacity, in turn, presents insurgents with an opportunity to mount an insurrection against the state. We coded a country as anocracy, that is, a “semi-democracy” (Hegre et al., 2001), if its year-specific Polity IV score (Marshall et al., 2011) was between +5 and −5 (Hegre et al., 2001: 38; Fearon and Laitin 2003a: 81). Population size may also increase the likelihood of the onset of civil wars, since it increases the number of potential recruitees, thereby providing an opportunity to rebel (e.g. Fearon and Laitin, 2003a: 81). We obtained data on country populations from the World Population Prospects (2010) database, and in the analyses use the logarithmic transformation of the country-year population data.

In addition, we also controlled for exclusion from political power, since the larger the share of disaffected groups and individuals who may be willing to fight in order to gain political power is, the higher is the risk of conflict (Cederman et al., 2010; Wimmer et al., 2009a). Data were taken from Wimmer et al.’s (2009b) dataset. Lastly, although unrelated to competing explanations of rebel opportunities, we used a peace time measure, indicating the time in years that has passed since the onset of the last war, in order to control for temporal dependency in the time-series cross-section analysis (Beck et al., 1998; see also Ray, 2003).

Results

To estimate the effect of youth bulge on war onset, we use a binary logit regression model that has been quite commonly used by students of civil war (e.g. Fearon and Laitin, 2003a; Urdal, 2006). The regression’s coefficients indicate the effect of a one-unit change in the independent variable on the log of the odds that a civil war will erupt relative to the odds that it will not (see also Long, 1997: chapter 3).

Table 2 presents models estimating the likelihood of civil wars of any type (both non-ethnic and ethnic). We estimate each model twice for each coding of the dependent variables (Sambanis’, Fearon and Latin’s and Wimmer et al.’s) with different proxies for the level of development: natural log of GDP per capita and IMR. By and large, results are congruent with previous studies (e.g. Urdal, 2006): youth bulge, for the most part, is not a significant predictor of the likelihood of civil war. In only one of the six models, the youth bulge coefficient reaches statistical significance at the 0.05 level (model 2, p < 0.05), and in another (model 4) it is marginally significant (p = 0.07). Overall, the other control variables are in the expected direction and statistically significant for the most part.

Table 2.

The three war-codings—all wars

	Sambanis		Wimmer et al.		Fearon and Laitin
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Youth bulge	0.028 (0.023)	0.051 (0.021)*	0.004 (0.027)	0.039 (0.022)†	0.016 (0.025)	0.032 (0.022)
GDP per capita (log)	−0.434 (0.125)***		−0.617 (0.157)***		−0.401 (0.139)**
Infant mortality rate		0.004 (0.002)*		0.005 (0.002)*		0.007 (0.002)**
Anocracy	1.065 (0.193)***	1.039 (0.191)***	0.743 (0.241)**	0.710 (0.241)**	1.268 (0.223)***	1.255 (0.220)***
Excluded population	1.106 (0.317)***	1.165 (0.328)***	0.883 (0.379)*	0.992 (0.395)*	0.804 (0.372)*	0.764 (0.392)†
Population size (log)	0.249 (0.062)***	0.311 (0.066)***	0.301 (0.073)***	0.390 (0.073)***	0.292 (0.073)***	0.365 (0.078)***
Peace years	−0.009 (0.009)	−0.012 (0.010)	−0.023 (0.010)*	−0.029 (0.011)**	−0.005 (0.009)	−0.001 (0.011)
Constant	−4.473 (1.826)*	−9.821 (1.224)***	−2.493 (2.431)	−10.217 (1.270)***	−4.966 (2.066)*	−10.215 (1.321)***
Number of war onsets	127	126	88	87	97	96
Number of observations	5765	5762	6678	6683	5815	5812
−2 Log likelihood	1104.22	1102.67	844.26	846.06	899.26	887.64
Model χ²	125.21***	131.08***	115.46***	104.24***	99.33***	111.46***
Pseudo R²	0.0951	0.0906	0.0988	0.0866	0.0884	0.0926

†

p < 0.1; *p < 0.05; **p < 0.01; ***p < 0.001 (two-tailed test). Robust standard errors in parentheses.

Models 7–12 in Table 3 confirm the expectation that youth bulge is associated with higher risk of non-ethnic civil wars (H1). In all the models youth bulge coefficients are in the expected direction, and they are statistically significant at the 0.05 level in five of the six models (youth bulge coefficient in model 9 approaches statistical significance; p = 0.135). These results confirm that, in spite of the variations across the three war-coding systems, youth bulge is significantly associated with higher risk of non-ethnic war onset.

Table 3.

Models of non-ethnic wars

	Sambanis		Wimmer et al.		Fearon and Laitin
	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12
Youth bulge	0.097 (0.037)**	0.104 (0.039)**	0.053 (0.035)	0.081 (0.034)*	0.096 (0.038)*	0.073 (0.036)*
GDP per capita (log)	−0.143 (0.231)		−0.431 (0.261)†		0.174 (0.221)
Infant mortality rate		0.009 (0.004)*		0.012 (0.004)**		0.008 (0.005)
Anocracy	0.913 (0.341)**	0.990 (0.355)**	0.654 (0.389)†	0.710 (0.393)†	0.672 (0.417)	0.735 (0.423)†
Excluded population	−0.027 (0.579)	−0.125 (0.584)	−0.156 (0.589)	−0.245 (0.619)	−0.848 (0.762)	−0.995 (0.751)
Population size (log)	−0.032 (0.122)	0.042 (0.121)	0.031 (0.109)	0.145 (0.103)	0.085 (0.108)	0.134 (0.095)
Peace years	0.010 (0.016)	0.022 (0.020)	−0.010 (0.013)	−0.000 (0.017)	−0.002 (0.016)	0.014 (0.022)
Constant	−8.485 (3.390)*	−11.786 (2.155)***	−4.783 (3.745)	−11.756 (2.149)***	−11.849 (3.744)**	−10.940 (2.059)***
Number of war onsets	35	34	28	27	26	25
Number of observations	5673	5670	6618	6623	5744	5741
−2 log likelihood	402.16	384.96	343.00	326.20	321.12	306.84
Model χ²	22.45***	19.77**	24.13***	20.52**	15.14*	18.20**
Pseudo R²	0.0559	0.740	0.0523	0.0707	0.0344	0.0462

†

p < 0.1; *p < 0.05; **p < 0.01; ***p < 0.001 (two-tailed test). Robust standard errors in parentheses.

Substantively, the models suggest that an increase of 1% in youth bulge raises the likelihood of a non-ethnic war onset by 8–11%. According to model 11, for example, a youth bulge score of 1 standard deviation below the youth bulge mean (i.e. youth bulge of 34%), while holding all other variables constant at their means, yields a 1.6% probability of non-ethnic war onset within a 10-year span. Meanwhile, a youth bulge score of 1 standard deviation above the mean (i.e. youth bulge of 50%) increases that probability to 7.6%. Similarly for all the other models (except model 9), an increase from 1 standard deviation below the youth bulge mean to 1 standard deviation above the mean renders a country between 3 and 5 times more likely to experience non-ethnic war.

A quick look at the control variables shows that, surprisingly, GDP per capita, the most commonly used indicator of economic development, is not significantly associated with the onset of non-ethnic war in two models (models 7 and 11), and only marginally significant in one (model 9, p = 0.099). On the other hand, IMR, the alternative proxy of economic development, is significantly linked to the outbreak of non-ethnic wars in two of the three models (models 8 and 10) and approaches significance in the third (model 12, p = 0.154).¹⁶ To test whether the poor performance of the GDP per capita coefficient is due to the presence of the youth bulge measure in the model, we re-ran models omitting the youth bulge variable from models 7, 9 and 11, and the GDP per capita coefficient became statistically significant in models 7 and 9; that is, when both variables are entered into the same model, the youth bulge variable cancels out the statistical significance of the GDP per capita variable in two out of the three war-coding schemes (cf. Fearon, 2011: 15–16).¹⁷ In the discussion we address these results in more detail.

The remaining control variables are mostly not associated with onset of non-ethnic war. The coefficients of population size (one of the most robust predictors of civil war; e.g. Hegre and Sambanis, 2006), the share of the excluded population and number of peace years are not of statistical significance in any of the non-ethnic models, while only the dummy variable of anocratic regimes is associated with a higher risk of the onset of non-ethnic war (cf. Sambanis, 2001: 276–278).

Models 13–18 in Table 4 provide evidence that youth bulge does not affect the likelihood of ethnic civil wars. Indeed, the youth bulge coefficient is statistically insignificant in each of the six models.

Table 4.

Models of ethnic wars

	Sambanis		Wimmer et al.		Fearon and Laitin
	Model 13	Model 14	Model 15	Model 16	Model 17	Model 18
Youth bulge	0.000 (0.028)	0.032 (0.024)	−0.024 (0.036)	0.022 (0.027)	−0.018 (0.030)	0.014 (0.027)
GDP per capita (log)	−0.533 (0.138)***		−0.670 (0.190)***		−0.639 (0.159)***
Infant mortality rate		0.003 (0.002)		0.003 (0.003)		0.007 (0.003)**
Anocracy	1.178 (0.230)***	1.110 (0.226)***	0.829 (0.301)**	0.757 (0.298)*	1.560 (0.268)***	1.465 (0.261)***
Excluded population	1.585 (0.382)***	1.698 (0.390)***	1.432 (0.478)**	1.638 (0.486)***	1.327 (0.441)**	1.353 (0.457)**
Population size (log)	0.341 (0.068)***	0.404 (0.075)***	0.411 (0.090)***	0.499 (0.089)***	0.365 (0.089)***	0.452 (0.096)***
Peace years	−0.018 (0.011)†	−0.027 (0.011)*	−0.031 (0.014)*	−0.043 (0.014)**	−0.005 (0.011)	−0.006 (0.012)
Constant	−3.791 (2.128)†	−9.999 (1.415)***	−2.566 (3.125)	−10.507 (1.578)***	−2.996 (2.441)	−10.784 (1.622)***
Number of war onsets	92	92	60	60	71	71
Number of observations	5730	5728	6650	6656	5789	5787
−2 log likelihood	826.64	839.77	596.16	609.90	666.77	675.84
Model χ²	144.72***	134.43***	145.27***	95.62***	116.55***	116.97***
Pseudo R²	0.1232	0.1092	0.1290	0.1090	0.1296	0.1177

†

p < 0.1; *p < 0.05; **p < 0.01; ***p < 0.001 (two-tailed test). Robust standard errors in parentheses.

Moving to the controls, GDP per capita in models 13, 15 and 17 is significant and negative, suggesting that rich countries are less likely to experience ethnic wars. While this finding seems to challenge the argument that an “economic logic” plays only a minor role in explaining ethnic war onset (Sambanis, 2001: 266–267), we argue that our results may in fact be in line with Sambanis’s assertion. First, we note that the association of IMR, one of the two proxies of economic development, appears to be unrelated to ethnic conflict outbreak in two of the three models. Second, while GDP per capita is indeed a strong predictor of ethnic war, other indicators, such as regime type (i.e. anocracy) explain, for the most part, a larger share of the variation in the likelihood of ethnic war onset.¹⁸ In other words, there are other important predictors of ethnic war, perhaps even more influential than the level of economic development in the country.

Overall, results in Tables 3 and 4 attest to the different effects of youth bulge on different types of civil wars. Whereas youth bulge is significantly associated with non-ethnic war onset at the 0.05 level in five models, youth bulge is not associated with ethnic war onset in any model. Moreover, averaging the youth bulge’s coefficients and standard errors of Table 3 results in an average of β = 0.084 and standard error (SE) = 0.037, whereas in Table 4 such averaging results in β = 0.004 and SE = 0.029. This suggests that the youth bulge’s association with onset of non-ethnic wars is exponentially larger than its (almost non-existent) association with onset of ethnic wars.¹⁹

The graphs in Figure 1 illustrate the effect of youth bulge, comparing the estimated probabilities of onset of non-ethnic and ethnic war. Panels are identified by the specific dataset from which the dependent variable was drawn, and are repeated twice, once for each proxy of development level. The X-axis represents the percentage of youth bulge and the Y-axis the probability (in percentage) that civil war will erupt in a given year.²⁰ It is quite clear that, as hypothesized, the effects of youth bulge vary across different war-types, increasing the risk of non-ethnic war but not the risk of ethnic war. The probability of non-ethnic war onset increases as a function of the percentage of young men; the increase is steady at low levels of youth bulge and much steeper as the share of young men rises. The evident effect of youth bulge on non-ethnic wars is even more pronounced in comparison to the relatively flat lines plotting the likelihood of ethnic wars.

Figure 1.

Marginal effect of youth bulge on the probability of onset of non-ethnic and ethnic war. (a) Sambanis’s war-coding; (b) Wimmer et al.’s war-coding; and (c) Fearon and Laitin’s war-coding.

Robustness tests

In general, the reported results are in line with our research hypotheses: the percentage of young male adults correlates with the outbreak of non-ethnic civil wars and is unrelated to the onset of ethnic civil wars. Results hold when we control for other well-established predictors of civil wars, and when using three considerably different lists of civil wars. We nevertheless conducted a series of robustness tests to investigate whether results hold with changes in model specifications and model estimations.²¹

First, we ran all baseline models with two alternative measures of youth bulge: the ratio of men aged 15–24 to the entire adult male population in the country; and the percentage of young adults (men and women) between the ages of 15 and 24 (Urdal, 2006). Results were practically the same as the results of the youth bulge variable in this paper (see Tables 1–4 in the Online Appendix 2). Second, we added an additional variable to the baseline models—the rate of economic growth—as an additional and competing explanation for economic opportunities (Collier and Hoeffler, 2004; see Tables 5 and 6 in the Online Appendix 2). This addition did not change the substantive effect of youth bulge on the onset of non-ethnic war. We also ran additional models in which we changed the specifications of the baseline models: in several models we used alternative measures of two control variables (logged GDP per capita and anocracy), and in other models we added control variables to the baseline models (e.g. dummy variables for decades). By and large, results in these models were very close to those of the baseline models (see Tables 7–16 in the Online Appendix 2).

In order to gauge the possible distinct effect of youth bulge on armed conflicts in general, and not just civil wars, we used Wimmer et al.’s (2009a) coding of ethnic and non-ethnic armed conflicts, conflicts with at least 25 battle-related deaths per year (Gleditsch et al., 2002). Results are consistent with the expectations. Youth bulge is positively and statistically associated with the onset of non-ethnic armed conflicts, but not with onset of ethnic armed conflicts (Table 17 in the Online Appendix 2).²² All in all, these results provide further support for our hypotheses.

In addition, we repeated the analyses using different statistical estimation techniques. First, we reran all models with robust standard errors clustered by countries instead of heteroskedasticity-robust standard errors, in order to control for autocorrelation (see, e.g. Cederman et al., 2011). Second, owing to the relatively small number of non-ethnic civil wars, which could bias our estimation, we ran the logistic regression using the Rare Events Data estimation method (King and Zeng, 2001a). Overall, results of the alternative estimation techniques were quite similar to those of the baseline (see Tables 18–21 in the Online Appendix 2).

Following Ray’s (2003: 20–27) suggestions, we also examined youth bulge’s effects on civil war across space and over time, that is, between and within countries (see Tables 22 and 23 in the Online Appendix 2). The coefficients of the across-space youth bulge measure are positively associated with onset of non-ethnic war in all six models, as in three of them it is also statistically significant at the 0.1 level. In contrast, the over-time measure is overall not associated with onset of non-ethnic war. This suggests that, as could have been expected, higher levels of youth bulge in comparison to other countries are predictive of onset of non-ethnic war while within-country changes in youth bulge, irrespective of the youth bulge score itself, are not.

Finally, in the Methods section (note 11) we explained that, in 20 civil wars, we found discrepancies between the coding of type of war used by Sambanis, Fearon and Laitin, and Wimmer et al. Additional analyses tested the sensitivity of the results to discrepancies in the coding of the dependent variable. We ran additional models in which we used coding of war onsets that appeared in at least two out of the three datasets (Table 24 in the Online Appendix 2), and models in which we determined, according to a majority among the coding systems, the war type (non-ethnic/ethnic) of the 20 civil wars whose type was controversial (Table 25 in the Online Appendix 2).²³ Consistent with our hypotheses, results were very similar to those of the baseline models.

Discussion

Disproportionately large shares of young adults—youth bulges—are prevalent in developing countries around the world, mainly in Sub-Saharan Africa, but also in the Middle East, North Africa, Latin America and Asia (World Population Prospects, 2010). Youth bulge has been associated with increasing social unrest (Moller, 1968) as well as with terrorism, rioting, violent demonstrations and the outbreak of armed conflicts (Urdal, 2006).

Concerning youth bulge’s effect on civil war, some studies attest to strong effects of youth bulge on the onset of armed intrastate conflicts (e.g. Cincotta et al., 2003; Urdal, 2006), but the theoretical argument has not been, thus far, firmly substantiated (e.g. Collier and Hoeffler, 2004; Fearon and Laitin, 2003a). In this paper, we reason that youth bulge affects only the mobilization patterns leading to a non-ethnic war. Youth bulge pressures state and society by increasing economic, educational and other demands where the availability of such opportunities may be quite scant. The likelihood of non-ethnic wars, which are much more likely than ethnic wars to be driven by country-wide economic and social changes, may therefore be higher in countries with large cohorts of young people. In contrast, youth bulge need not be associated with the ethnic wars. These are driven more directly by group-specific political and cultural aspects (e.g. Sambanis, 2001) and, at times, economic considerations pertaining to the conditions of specific ethnic groups (e.g. Cederman et al., 2011).

Empirical analyses support such reasoning. Results suggest that youth bulge affects only one type of civil war: conflicts that are not motivated by ethnicity. These results are robust to different definitions of youth bulge, to various alternative model specifications, and to different war-coding systems. In addition, our results support the argument that, at the very least, systematic differences between ethnic and non-ethnic civil wars should be investigated more thoroughly (Sambanis, 2001).

The presence of a high proportion of young men in a country, a phenomenon whose economic effects are mostly country-wide, is associated with non-ethnic war onset but not with ethnic war onset, which gives further support to the suggestions that ethnic war erupts more due to ethnic grievances. In non-ethnic wars, rebel groups take advantage of certain political and demographic characteristics that make it easier for rebel groups to mount an insurgency against the state (Collier and Hoefller, 2004: esp. 563–570), regardless of the ethnic composition of the group.

Non-ethnic wars therefore seem to be more likely in times of political and societal turmoil, when rebels aiming to achieve non-ethnic goals identify winning opportunities and have a wider potential of recruits across ethnic groups. Moreover, we take the observation that GDP per capita loses its statistical significant effect once youth bulge is added to the equation (see Table 3) as a tentative indication that the percentage of young men, rather than a country’s relative economic wealth, accounts for the outbreak of non-ethnic wars.

The findings that population size is a strong predictor of ethnic war but not of non-ethnic war might also attest to different causes for ethnic and non-ethnic war, as the mobilization to ethnic war could be facilitated as the population increases, providing every ethnic group with more potential recruits for its ranks. In the case of non-ethnic war, it is not the population size but rather the youthfulness of the population that provides rebel groups with opportunities to rebel.

Anticipated declines in fertility rates and higher life expectancy in many developing countries may lower youth bulge rates (Ashford, 2007). If indeed the relationship between youthful age structures and violent conflicts is as strong as some have argued (e.g. Cincotta et al., 2003; Urdal, 2006), then future demographic trends should result in fewer armed conflicts and civil wars in these countries (see also Hegre et al., 2013: 260). While this may be true for non-ethnic conflicts, we argue that ethnic conflicts may continue to erupt driven, inter alia, by political marginalization and economic inequalities that breed perceptions of relative deprivation (e.g. Cederman et al., 2010, 2011; Wimmer et al., 2009a), regardless of a country’s age structure.

Indeed, long-term demographic trends may lower the risk of war, but in the short run the prediction that youth bulge is associated with the onset of non-ethnic war should convince decision-makers in countries with high rates of youth bulge to attempt to address and resolve societal and economic grievances and promote the inclusion of the younger cohorts into society. Such efforts may be even more important in countries that possess a combination of dire economic conditions, soaring unemployment rates and a high proportion of young adults, as indeed is the case today in many countries in the Middle East and North Africa.

This paper is not without its limitations. First, while we investigated the direct effect of youth bulge on civil war, this effect may be contingent on the effect of other explanatory factors. Urdal (2006), for instance, examined several interaction effects, but without differentiating between ethnic and non-ethnic wars.

In this paper our main motivation was to examine the main effect of a phenomenon that in recent years has attracted the attention of many public figures and scholars, as well as to resolve an apparent contradiction in the literature pertaining to the effect of this phenomenon on civil war onset. Such desires notwithstanding, we acknowledge that future research could help explain the effects of youth bulge when taken in conjunction with other factors in different war-types. Also, in this paper we investigated youth bulge’s effect on different war-types throughout the world, but it could be that its effects are more, or less, salient in certain continents or regions, perhaps owing to idiosyncratic cultural traits. Such possibilities certainly await further research.

Another direction for future research of youth bulge could be the investigation of the social effect of the age structure. Thus far, only two studies have rigorously examined non-economic effects of youth bulge (Hart et al., 2004, 2005), and while several scholars argue that youth bulge could be harming social cohesion and creating more “volatile” societies (Boyden, 2007; Cincotta, 2008–2009; Moller, 1968), there are no studies corroborating this. Such claims no doubt deserve further investigation as part of the academic effort to more fully understand the youth bulge phenomenon.

In conclusion, in this paper we have shown that youth bulge affects the onset of only one type of civil war: the non-ethnic war. In light of the significant role young people have played in recent years in many events around the world (e.g. Hvistendahl, 2011), and in light of the results of this paper, which show the effect of youthful age structure on the outbreak of different types of civil war, more research on youth bulge is definitely needed. A more profound understanding of the youth bulge phenomenon and its consequences might help us understand not only recent events of the Arab Spring, but other political and societal phenomena elsewhere in the world.

Footnotes

Acknowledgements

We would like to thank Jeremy Bajema, Pazit Ben-nun Bloom, Danielle Hanley, Maayan Mor, four anonymous reviewers and the editor of Conflict Management and Peace Science, for their helpful comments and insights.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Ashford

(2007) Africa’s youthful population: Risk or opportunity? Population Reference Bureau. Available at: http://www.prb.org/Publications/PolicyBriefs/AfricasYouthfulPopulation.aspx

Beck

Katz

Tucker

(1998) Taking time seriously: Time-series-cross-section analysis with a binary dependent variable. American Journal of Political Science 42(4): 1260–1288.

Bloom

Freeman

Korenman

(1987) The labour market consequences of generational crowding. European Journal of Population 3: 131–176.

Boyden

(2007) Children, war and world disorder in the 21^st century: A review of the theories and the literature on children’s contributions to armed violence. Conflict, Security and Development 7(2): 255–279.

Cederman

L-E

Wimmer

Min

(2010) Why do ethnic groups rebel? New data and analysis. World Politics 62(1): 87–119.

Cederman

L-E

Wimmer

Gleditsch

(2011) Horizontal inequalities and ethnonationalist civil war: A global comparison. American Political Science Review 105(3): 478–495.

Cederman

L-E

Wimmer

Gleditsch

(2013) Inequality, Grievances, and Civil War. New York: Cambridge University Press.

Cincotta

(2008–2009) Half a chance: Youth bulges and transitions to liberal democracy. Environmental Change and Security Program Report, 10–17, issue 13.

Cincotta

Engelman

Anastasion

(2003) The security demographic—Population and civil conflict after the Cold War. Report for the Population Action International, Washington, DC.

10.

Clogg

Petkova

Haritou

(1995) Statistical methods for comparing regression coefficients between models. American Journal of Sociology 100(5): 1261–1293.

11.

Collier

Hoeffler

(2004) Greed and grievance in civil war. Oxford Economic Papers 56(4): 563–595.

12.

Collier

Hoeffler

Rohner

(2009) Beyond greed and grievance: Feasibility and civil war. Oxford Economic Papers 61(1): 1–27.

13.

Easterlin

(1987) Easterlin hypothesis. In: Eatwell

Millgate

Newman

(eds) The New Palgrave: A Dictionary of Economics, Vol. 2. New York: Stockton.

14.

Elbadawi

Sambanis

(2002) How much war will we see? Explaining the prevalence of civil war. Journal of Conflict Resolution 46(3): 307–334.

15.

Esty

Goldstone

Gurr

Surko

Unger

(1995) State failure task force report. McLean Science Applications International.

16.

Fearon

(2011) Governance and civil war onset. World Development Report 2011. World Bank.

17.

Fearon

Laitin

(2003a) Ethnicity, insurgency, and civil war. American Political Science Review 97(1): 75–90.

18.

Fearon

Laitin

(2003b) Ethnicity, insurgency, and civil war. Replication data. Available at: http://www.stanford.edu/group/ethnic/publicdata/publicdata.html

19.

Gates

(2002) Recruitment and allegiance: The microfoundations of rebellion. Journal of Conflict Resolution 46: 111–130.

20.

Gleditsch

Wallensteen

Eriksson

Sollenberg

Strand

(2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39: 615–637.

21.

Hart

Atkins

Markey

Youniss

(2004) Youth bulge in communities. Psychological Science 15(9): 591–597.

22.

Hart

Atkins

Youniss

(2005) Knowledge, youth bulges, and rebellion. Psychological Science 16(8): 661–662.

23.

Hegre

Sambanis

(2006) Sensitivity analysis of empirical results on civil war onset. Journal of Conflict Resolution 50(4): 508–535.

24.

Hegre

Ellingsen

Gates

Gleditsch

(2001) Toward a democratic civil peace? Democracy, political change, and civil war, 1816–1992. American Political Science Review 95(1): 33–48.

25.

Hegre

Karlsen

Nygard

Strand

Urdal

(2013) Predicting armed conflicts, 2010–2050. International Studies Quarterly 57: 250–270.

26.

Heston

Summers

Aten

(2006) Penn World Table, version 6.2. Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania.

27.

Horowitz

(1985) Ethnic Groups in Conflict. Berkeley, CA: University of California Press.

28.

Humphreys

Weinstein

(2008) Who fights? The determinants of participation in civil war. American Journal of Political Science 52(2): 436–455.

29.

Huntington

(1996) The Clash of Civilizations and the Remaking of World Order. New York: Simon & Schuster.

30.

Hvistendahl

(2011) Young and restless can be a volatile mix. Science 333: 552–554.

31.

Kaufmann

(1996) Possible and impossible solutions to ethnic civil wars. International Security 20(4): 136–75.

32.

King

Zeng

(2001a) Logistic regression in rare events data. Political Analysis 9(2): 137–163.

33.

King

Zeng

(2001b) Improving forecasts of state failure. World Politics 53(4): 623–658.

34.

Korenman

Neumark

(1997) Cohort crowding and youth labor markets: A cross-national analysis. Working Papers: 6031. Washington, DC: National Bureau of Economic Research.

35.

Lake

Rothchild

(1998) Spreading fear: The genesis of transnational ethnic conflict. In: Lake

Rothchild

(eds) The International Spread of Ethnic Conflict: Fear, Diffusion, and Escalation. Princeton, NJ: Princeton University Press.

36.

Long

(1997) Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks, CA: Sage.

37.

Macunovich

(2000) Relative cohort size: Source of a unifying theory of global fertility transition? Population and Development Review 26(2): 235–261.

38.

Marshall

Jaggers

Gurr

(2011) Polity IV Project. Available at: http://www.systemicpeace.org/inscr/inscr.htm

39.

Mesquida

Wiener

(1996) Human collective aggression: A behavioral ecology perspective. Ethology and Sociobiology 17: 247–262.

40.

Mesquida

Wiener

(1999) Male age composition and severity of conflicts. Politics and the Life Sciences 18(2): 181–189.

41.

Moller

(1968) Youth as a force in the modern world. Comparative Studies in Society and History 10: 237–260.

42.

Ray

(2003) Explaining interstate conflict and war: What should be controlled for? Conflict Management and Peace Science 20(1): 1–31.

43.

Ross

(2006) A closer look at oil, diamonds, and civil war. Annual Review of Political Science 9: 265–300.

44.

Sambanis

(2000a) Partition as a solution to ethnic war: An empirical critique of the theoretical literature. World Politics 52: 437–483.

45.

Sambanis

(2000b) Partition as a solution to ethnic war. Replication data. Available at: http://go.worldbank.org/08ZHJ2TPI0

46.

Sambanis

(2001) Do ethnic and nonethnic civil wars have the same causes? Journal of Conflict Resolution 45(4): 259–282.

47.

Sambanis

(2004a) What is civil war? Conceptual and empirical complexities of an operational definition. Journal of Conflict Resolution 48: 814–858.

48.

Sambanis

(2004b) What is civil war? Replication data. Available at: http://pantheon.yale.edu/~ns237/WhatIsCivilWar.zip

49.

Sambanis

(2004c) Using case studies to expand economic models of civil war. Perspectives on Politics 2(2): 259–279.

50.

Staveteig

(2005) The young and the restless: Population age structure and civil war. In: Population and Conflict: Exploring the Links. Environmental Change and Security Program, issue 11, pp. 12–19.

51.

Toft

(2005) The state of the field: Demography and war. In: Population and Conflict: Exploring the Links. Environmental Change and Security Program, issue 11, pp. 25–28.

52.

United Nations Demographic Yearbook (2013) The United Nations Statistics Division. Available at: http://unstats.un.org/unsd/demographic

53.

United States Census Bureau (2013) International Data Base page. Available at: http://www.census.gov/population/international/

54.

Urdal

(2004) The devil in the demographics: The effect of youth bulges on domestic armed conflict, 1950–2000. The Social Development Department of the World Bank, Working Paper no. 14.

55.

Urdal

(2006) A clash of generations? Youth bulges and political violence. International Studies Quarterly 50: 607–629.

56.

USAID (2010) Youth bulges and conflict. Technical brief. Available at: https://c.ymcdn.com/sites/www.allianceforpeacebuilding.org/resource/collection/9E786610-CF07–4073–9E34-A03818697B2E/USAID_Youth_Bulges___Conflict.pdf

57.

Wilson

Daly

(1985) Competitiveness, risk taking, and violence: The young male syndrome. Ethology and Sociobiology 6: 59–73.

58.

Wimmer

Cederman

L-A

Min

(2009a) Ethnic politics and armed conflict: A configurational analysis of a new global data set. American Sociological Review 74(2): 316–337.

59.

Wimmer

Cederman

L-A

Min

(2009b) Ethnic politics and armed conflict. Replication data. Available at: http://hdl.handle.net/1902.1/13825

60.

World Population Prospects (2010) The United Nations Population Division. Available at: http://esa.un.org/unpd/wpp/