“Oil Discoveries, Shifting Power, and Civil Conflict:” A formal and quantitative replication

Oil is associated with all kinds of bad things: Civil war (Ross 2004, 2006), Human rights violations (DeMeritt and Young 2013), and poor economic development (Karl 1997).  These studies all examine how oil wealth covaries with these outcomes, but not how the discovery of new reserves alters the processes that produce violence or poor economies. A new paper in International Studies Quarterly by Curtis Bell and Scott Wolford, “Oil Discoveries, Shifting Power, and Civil Conflict” does just this. Bell and Wolford offer a novel theoretical claim to explain how the discovery of petroleum reserves has nonlinear effects on civil conflict.  They show that oil discoveries in poor states influence the balance of power and make regime commitments to peace less credible given a prospective shift in state capabilities. The paper also uses a formal model to generate these claims, and then a statistical model to test them.

This symposium will examine the paper from two perspectives. First, Emily Ritter from the University of California, Merced replicates the formal results from the paper.  She clearly explains how Bell and Wolford expand the basic bargaining model of civil war made popular by Fearon (2003) by showing how new resources influence future bargaining. [1] She extends their claims and suggests power for the rebels could be made endogenous to the model leading to claims beyond simple conflict onset.

A clear benefit of formal models is the transparency of assumptions and how they lead to hypotheses. Ritter is able to replicate and extend the results. I am not sure how often this kind of replication occurs, but it is a first for these symposia. Published in the paper or an appendix, proofs of the math are often readily available. But what are the incentives to replicate the model? Clearly the bargaining model as applied to conflict has been replicated and used by a many scholars. The extent to which less famous models have been thoroughly examined, however, is not clear.  In the current discussions around research ethics and transparency, this is a useful exercise.

Beyond the formal replication, we asked Adrian Florea, who will be a Lecturer at the University of Glasgow in the fall, to replicate the quantitative results. Florea was able to replicate the main results from Bell and Wolford, tried a re-coding of some of the conflict onset cased, and using some different specifications, the results were fairly consistent. To probe and extend, Florea examines a new dependent variable—civil war continuation. Strikingly, the results again are similar to the civil war onset model of Bell and Wolford. This suggests an application of the theory different domains, which Florea encourages Bell and Wolford to pursue.

In the final installment of the symposium, Bell and Wolford respond. They gladly engage in a “modeling dialogue” both statistical as well as formal. They laud Ritter’s extension of the model for explaining potentially conflicting evidence on repression that suggests a clean empirical test. They clarify a point about the mechanism in response to Florea. Finally, they ask questions about the implications of Florea’s replication and suggest the need for future extensions of the theoretical and empirical model.

Since Collier and Hoeffler and Fearon and Laitin’s seminal pieces on civil war onset in the early 2000s, many scholars have examined the link between state capacity and the onset and duration of civil war. Bell and Wolford’s piece moves beyond this literature in a useful way by thinking about how future expectations of actor capabilities will influence the onset of conflict. If Florea is right, their theory might extend into duration and other areas as well. While international studies often rewards novel theories like Bell and Wolford’s, the work Ritter and Florea did is just as important. We hope this symposium is part of building this edifice of solid empirical work that can serve as a foundation for future research.


[1] Powell (2004) extended to revolutions, international conflict, and other areas where an inefficiency mechanism may be at work.


Full Text PDF   Find Replication Data
Discuss this Article
There are currently no comments, be the first to post one.
Start the Discussion...
Only registered users may post comments.
ISQ On Twitter