I’m late to the Jacqueline Stevens’ op-ed party, and don’t think I have much to add on the main issues – about forecasting, capital ‘S’ Science, etc. – discussed by Henry, Erik, and Andrew (or others linked to here). But Stevens also gave an interpretation of the core argument in Laitin’s and my 2003 APSR paper that, for what it’s worth, I think is a misreading. Since I’ve seen this elsewhere over the years, I thought I’d try to speak to it.
Stevens says that our article “claims that civil wars result from weak states, and are not caused by ethnic grievances.” Further, she views it as obvious that many internal conflicts around the world “are largely about political and economic injustice,” an observation that she correctly observes one often gets from the journalists who cover the conflicts.
In fact, in our article we agree that the participants in many or most ethnic and other civil conflicts can be motivated by intense political and economic grievances. We wrote that “Surely ethnic antagonisms, nationalist sentiments, and grievances often motivate rebels and their supporters” (and, I would add, governments as well).
When one asks about the causes of a violent conflict, one is often asking in the first instance about the motivations of the participants. In this sense, we are happy to agree – and also think it is true – that many civil wars are caused in part by the ethnic antagonisms, nationalist sentiments, and other grievances that can motivate participants. They are caused by these things in the sense that if the participants didn’t have these motivations, the wars wouldn’t have happened.
The claim we were making was not about the motivations of civil war participants, but about what factors distinguish countries that have tended to have civil wars from those that have not. We found that several measures that one might think could tap the level of grievances in a country – such as degree of income inequality, democracy, government observance of civil liberties, ethnic and religious diversity, or particular ethnic configurations such as having a majority ethnic group and a large ethnic minority – were not systematically associated with higher civil war risks. (In some cases, the absence of association depends on comparing among countries at similar levels of economic development.)
This was surprising to us, and, I know from the reactions of journalists and others over the years, to many non-political scientists as well. We proposed a possible explanation that I won’t go through in detail here. In a nutshell, we suggested that many if not all countries unfortunately have some intensely aggrieved people or groups, and that what appeared to us to explain more of the variation in civil war propensities across countries are factors that determine whether they have good opportunities to get an insurgency going (which can be done, in the right circumstances, with quite small numbers).
More on grievances and civil conflict: It should be stressed that levels of grievances, ethnic or otherwise, are not randomly assigned across countries or across time within countries. On the contrary, governments and rulers to a large degree choose the level of grievances by choosing policies, and it is clear that they choose the level in some part with an eye on the likelihood of conflict. So grievances are highly “endogenous,” in the current jargon. It follows that associating civil war risk with measures of grievance across countries doesn’t tell us anything about the causal effect of an exogenous amping up grievances on the risk of civil war (not an intervention any IRB would permit, fortunately). I find it very plausible that there would be a positive average effect.
If grievances are endogenous to policy choices, this could be one reason why it is not easier to find strong correlations between broad measures of grievance and civil war risk. If governments tend to “charge what the market will bear” in terms of policies and grievances, then you may often get higher grievances where the aggrieved groups have little capacity to rebel, and lower grievances where they do. For example, Roma are terribly abused in many countries – governments can get away with it because the Roma have little ability to threaten significant rebellion or disorder. This also implies that levels of grievance and factors affecting opportunity or capability to organize for rebellion are causally linked.
The points in the last two paragraphs are based on discussion in a 2007 APSR paper by Kimuli Kasara, Laitin, and me, which was in part a response to a paper by Cederman and Girardin (not the paper Stevens cites, but it’s related). The upshot is that I think it’s hard to know how to interpret the statistically significant coefficients on the policy-based measures of grievances like those considered by Cederman et al. What I would like to do (and am working on) is to get a better fix on the relative importance of shocks to grievances versus shocks to opportunity as triggers for civil war.
But this is getting far afield from Jacqueline Steven’s comment, which I take to be entirely about the question of rebels’ motivations. No, we don’t doubt that participants are often motivated in some part by a sense of political or economic injustice. What we were questioning was whether variation in broadly held grievances explains much of the variation in civil war onset across countries, which is a different proposition. (To be more specific on the subject of the motivations of individual rebels, we think these can be very complex, though they often include anger at injustice. Stevens would probably agree. And she might be interested in the growing literature on “micro” studies of rebels in civil wars that tries to understand rebel motivations, among other things, by doing surveys of and interviews with combatants.)
One more thing. On her blog, Stevens refers the reader to a paper for her full analysis of the problems with the recent cross-national civil war literature. If I read her right (see also Henry Farrell on this paper), she argues that our 2003 article is “conclusively falsified” by evidence that states with strong militaries do not have less and may even have more civil war: “Earlier publications in this field using the Correlates of War dataset report a positive association between military spending and the likelihood of civil war (Henderson and Singer 2000, 290), but this goes unnoted by Fearon and Laitin (2003).”
Suppose Laitin and I had written a paper arguing that higher quality policing and police forces cause lower crime rates, and Stevens argued in reply that this is conclusively falsified because there is a positive association between size of police force and per capita crime rates across US cities. She would be making a mistake. The problem is that whatever the effect of size of the police force on crime is, the level of crime can positively affect the number of police who are hired. (Endogeneity again.) The same is true with military spending and civil war risk – civil war and even the perception of higher likelihood of civil war in the future can cause governments to raise bigger militaries. Or in other words, Germany and Denmark don’t need to have giant militaries to prevent nascent insurgency, in part because their small police forces are very efficient and because they could easily scale up if they needed to. The relevant “state capability” for our argument is the potential of a state to turn money into effective counterinsurgency (or effective policing), and this is not measured well by total military spending because military spending is partly caused by conflict and conflict risk.