The Economics of Civil Wars
Before moving to Washington, Paul began his research collaboration with Anke Hoeffler on the causes and consequences of civil wars. The series of papers that they published, beginning in 1998, marks a turning point in Paul's research, with a much greater emphasis on political economy.
The focus on political economy was the fruit of a long-running conversation with Robert Bates, the Harvard political economist whom Paul first met in the late 1980s, and whose work concentrates on violence and State failure in Africa (see, for example, Bates 2014). Paul and Anke’s work highlights economic insights into civil war combatants’ incentives and constraints, which are used to explain the pattern of intra-State conflict observed since the end of the colonial era. It is part of a trend of increasing engagement between development economists and political scientists.[217] Serendipitously, political scientists such as James Fearon, Stathis Kalyvas and Nicholas Sambanis were beginning to take an interest in correlates of intra-State conflict at the same time as Paul and Anke, and this has led to the creation of a genuinely interdisciplinary field. Just as Paul’s earlier work, which applied the methods of mainstream economics to the problems of resource-poor countries, was part of the re-engagement of development economics with the mainstream, so his later work is part of the re-engagement of economics with other social sciences, with development economics and economic history in the vanguard.The conceptual framework underpinning Paul and Anke’s initial research on civil wars, published in 1998, drew on the work of Herschel Grossman (1995) and Jean-Paul Azam (1995). Potential rebels are rational: everything else being equal, the probability that they will take up arms is increasing in the probability of victory, which is decreasing in the incumbent government’s per capita tax-raising powers.
These powers are increasing in per capita GDP and in the proportion of GDP made up of natural resource revenue. (The production of natural resources is geographically concentrated, so the administrative costs of taxing it are lower.) However, potential tax revenue is also a prize that can incentivise rebels to take more risks, so the effect of per capita GDP and the natural resource share on the incidence of civil war is indeterminate a priori, and the empirical relationships could be non-monotonic. The main innovation in the 1998 paper was the idea that the ability of rebels to coordinate opposition to the government depends on the ethnic composition of the country. With complete ethnic homogeneity, there is no opportunity for the rebels to exploit disadvantaged minority groups. On the other hand, if there are very many ethnicities, co-operation between different rebellious groups is likely to be more difficult to achieve. The relationship between civil war incidence and ethnic fractionalisation can therefore be expected to have an “inverted-U” shape. Empirical analysis of cross-country panel data suggested not only a non-monotonic relationship between economic conditions and civil war incidence, but also the inverted-U relationship with ethnic fractionalisation.The better-known 2004 paper is a refinement of the original empirical analysis. Ethnic fractionalisation, now interpreted as a correlate of the propensity for social grievance, is measured in several different ways. This reinterpretation of the ethnic fractionalisation measure is conceptually significant. In the 1998 paper, all of the theory remained within the bounds of rational choice economics, and all of the correlates of civil war were interpreted in terms of the economic costs and benefits of belligerence. In the 2004 paper, there is an implicit acknowledgement that war is at least partly a function of attitudes, and attitudes are shaped by social and political processes. In addition, the paper allows for alternative sources of grievance, including religious fractionalisation, income inequality and inequality in the distribution of land.
It also includes a more extensive set of correlates of the cost of waging war, some of which (e.g. male education rates, a proxy for the reservation wage of soldiers) suggest scope for policy interventions to reduce the long-run incidence of conflict.Hundreds of papers later, this literature is arriving at more nuanced results. For instance, ethnic fractionalisation appears to be an important predictor for certain types of intra-State conflict but not others (see Hegre and Sambanis 2006), and while religious fractionalisation was not found to be a significant correlate of civil war incidence in early studies, certain types of religious difference do now appear to be associated with violence (see Basedau et al. 2016). Economic theories jostle with political theories in interpretation of the data, and reading lists on the subject are correspondingly eclectic. Moreover, as highlighted in another early paper by Paul, civil war is itself a cause of poverty, and so a vicious circle of conflict and poverty can prevent a country from ever developing: in the jargon of development economics, it is a “low-level development trap” (see Collier 1999).
6