Expectations of Power Predict Civil War...and More

Bell and Wolford (B&W) present an innovative approach to studying the commitment problems that that lead to conflict, arguing that oil discoveries provide scholars with an empirical approximation of how expectations of a power shift affect the likelihood of war. Since expectations are particularly difficult to operationalize, this development is quite exciting. The theory and the empirical insight forge a path for studying implications of the commitment problem more broadly.  


Shifting Resources and Civil War 


The simple and insightful theory models the core ideas of bargaining with a commitment problem (see, e.g., Fearon (2003) and Powell (2004)): a government and a rebel group bargain over a pie of a certain size with a common expectation that the state will have an advantage in the next round. B&W characterize the advantage as newly available resources that shift the distribution of power against the rebels, with the rebels’ probability of winning decreasing from p in the first period to p/(1+wr) in the second.  


The authors innovate beyond the basic bargaining model by assuming that the size of the pie to be divided also increases in the second period. The expanding pie increases the advantage to the rebels of fighting in the first period: not only do they avoid a war they are less likely to win in the future, but they also lock in the division of the larger pie they will receive while they still have the military advantage.  


In equilibrium, the rebel group starts a conflict when the costs of war are sufficiently low and their probability of winning in the first period is sufficiently high. B&W interpret the equilibrium condition that rebels reject all offers when p> as suggesting that the relationship between resource shifts and war will not hold if the state is sufficiently powerful to defeat the rebels without the resource shift. This leads to an interesting conditional prediction: When the state is already resource-rich and likely to put down rebellions, discoveries will not affect the likelihood of war. But when the increase in resources occurs for a poor state, the change affects the likelihood of war.  


Broader Implications for Conflict Studies 


B&W’s theory and their argument for using oil discoveries to represent empirical expectations of power shifts opens doors for studying other political phenomena surrounding conflict. Consider, for instance, government repression.  


War occurs in B&W ‘s model when the probability that the rebels will win the conflict is sufficiently high. This is an exogenously given parameter in the model; substantively, we can interpret p as representing how powerful the rebels are relative to the government. 


Instead of treating p as exogenous, we could think of it as manipulable by one or both actors. If the government expects a group to rebel when the state is sufficiently weak gaining power, it has an incentive to try and prevent the challenge. Repression is one option: tactics of repression seek to undermine challenges to the status quo (see, e.g., Davenport 2007, Ritter 2014, Sullivan 2015, Ritter and Conrad [manuscript]) and are frequently cheaper than outright conflict.  


The availability of repressive tactics introduce (at least) one adaptation to the formal model that I assume for this illustration: repression shifts the probability of winning a conflict in the government’s favor, whether by reducing the group’s resources, undermining its collective action abilities, or creating a context of fear that prevents challenges. I assume that: 

  • the group’s probability of winning the conflict is highest in the absence of repression and before the oil discovery becomes fungible, 
  • its probability of winning the conflict is lower if they are repressed but is higher than when the discovery is available, and,
  • the group’s probability of victory is lowest once the discovery is available.  


Introducing this simplistic adaptation to B&W‘s theory creates a new space for predicted outcomes as a function of the expectation of power shifts. [1] This figure replicates Figure 1. The solid black line depicts the cutpoint on d below which the group will rebel, and the dashed line is the cutpoint on p above which they will rebel. The gray shaded region is the area in which we should expect to see war.

The solid and dashed red lines are the new cutpoints created by the additional option available to the government: to repress and improve the government's odds of victory in the post-discovery, pre-availability of resources period.  


The range in which we see civil war is appreciably reduced once repression becomes an option for the state. Under these conditions, newly discovered resources must represent an even greater shift in power to incentivize rebellion—only a larger shift will lead to war. Furthermore, a set of states that would have been sufficiently weak to invoke rebellion (the shaded area between the two dashed lines) will see oil discoveries lead to repression instead of war Oil discoveries may lead to repression—as oil rents more generally have been found to do (see, e.g., DeMeritt and Young 2013)—but this may also prevent wars that might otherwise occur.


In short, the possibility of empirically analyzing expectations of war allows us to study not only the onset of war due to commitment problems, as Bell and Wolford have, but also other phenomena that may arise in expectation of future power shifts.



[1] In an attached document created in Mathematica, I replicate the model and add the repression adaptation to create the figure shown here. Scott Wolford provided the parameter values to replicate Figure 1, in which delta=0.95, d=0.2, and w=1. 






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