Following up on last week’s treatment of the bargaining approach to war, we continued the discussion this week about the (unfortunately?) time-honored dispute over the link between the distribution of power and the probability of war. I won’t belabor the substance of the discussions too much, but two things stood out to me that I thought worth noting today. [Arm raised over dying horse…]
First, as Powell and Wagner have both noted, the old “debate” between power preponderance and balance-of-power types over which distribution was more war-prone was probably able to survive as long as it did because of incomplete arguments. That is, either claim—why fight when you’re unequal? or why fight when it’s not obvious who’d win—makes a certain amount of sense on its face. The problem, though, was that the assumptions required to get war at either distribution of power weren’t explicit; therefore, the logical chains needed to give us a reason to believe either claim weren’t there. The arguments were incomplete, and therefore lacking when it came time to try to arbitrate between them empirically.
Now, it may seem strange on the surface that I’m saying that incomplete arguments are a bad thing, because I’m usually loudly trumpeting the benefits of incompleteness. There’s a difference, though, between an incomplete theory (all are, else they’d not be theories) and an incomplete argument. Theories leave almost everything out involved in producing a given outcome of interest, and that’s their virtue, but in order to have a reason to believe their implications, the arguments that make up the theory must be complete. Else, we’re left with a debate between equal and unequal shares of power being more or less violent with no way to really distinguish between the two. (Put more strongly, we’re left with something generally unproductive.)
Second, and related, there’s great value to uncovering and then completing incomplete arguments. Arguably, that’s what Powell does when he shows that the effect of the distribution of power is conditional on the distribution of beenfits; when states have roughly what they could get by fighting, peace ensues, and when one feels it could get more, war enters the picture. This means that equal or unequal distributions of power should see war occur, but that there’s no independent effect of power on the likelihood of war. Now, previous conjectures about power and war discussed issues of dissatisfaction, or sometimes assumed it to be permanent, but they never completed the argument by linking the decision to fight a war to (a) the actual stuff states might fight over and (b) the bargains they might try to reach in lieu of war. (If we’re being harsher, these arguments nearly neglected the politics of the situation. As a brief aside, that’s why I find the bargaining metaphor for war appealing, because it’s an explicitly political account of war.)
But by doing so, by completing the arguments, we see that there is a link between power and war; it just happens to be conditional on other aspects to the political situation in which countries find themselves. However, looking for an unconditional link, so Powell’s argument goes, will commit the sin of omitted variable bias. If the underlying distribution of benefits determines the effect of relative power—that is, if it’s a sign-changer when included in our theoretical models—then we can’t justify leaving it out of our empirical models, either. For a pretty compelling example, see this.
And if you ask me, that’s a pretty strong case for both making complete arguments and for getting out there and seeing what arguments need completing. It’s as clear a case as one can make for rigorous theoretical work leading to better empirical work…and that’s what a lot of us are in this for, right?