Today I taught session two of international security, where, if you saw the syllabus I posted here last week, we focused on bargaining and war. I tried to structure today’s readings and discussion around what I see as the arc of thinking about war as a bargaining process—from a useful analogy to way of identifying concrete causal mechanisms and, eventually, closer to developing solid hypotheses. As with last week, I’m not going to run down everything that happened in class, but I will touch on a couple of notable things that came out of discussion. More, as usual, after the jump.
This week focused on bargaining and war, especially formal answers to the inefficiency puzzle of war: why can’t states simply strike the same agreements they reach at the end of the war before fighting and destroying so much stuff? Or losing so many soldiers? Or citizens? Or wasting all that money? We moved from early treatments of the process to later ones that explore the consequences of certain assumptions and modeling choices, which really let us get into some interesting nuts and bolts about testing these arguments. One of my favorites is this: shifts in power large enough to cause war shouldn’t be observable—and therefore able to be associated with war in observational data—because they should be prevented by war. This means that looking at observed power shifts to see if they cause war is tough, and it helps push us into other helpful directions for testing hypotheses based on changing power and commitment problems.
But I digress. One thing I’m trying to emphasize this semester is how we should evaluate the assumptions in other people’s arguments (and, by extension, our own, when it comes to making our own contributions to explaining international politics). In other words, we see an argument—say, Powell’s inefficiency condition that links anticipated shifts in bargaining power to war—and note that it requires a few assumptions. In this case, apart from the basics of costly war and opposed preferences over something of value, we have war when
- there will be a large, rapid shift in the distribution of power
- states can’t commit not to take advantage of this new bargaining leverage
The best way to answer this question, of course, is to think about what would happen if we relaxed it. What if, for example, war only locked in a share of benefits for a single period before we return to normal play, presumably with the other state still growing stronger? Two periods? Thirty? As you might surmise, I had the class do this very thing, and it turns out that the lock-in assumption isn’t all that pernicious: as long as war can give a player some reprieve (i.e. some number of periods more than just the current one) from making painful concessions to a rising opponent in the future, then there will generally be some shift in power large enough to make war attractive. However, the less effective the lock-in (or, perhaps, the less decisive the war outcome? the more easily renegotiable the issue is?) the higher the hurdle any shift in power will need to jump over in order to make some kind of preventive war attractive enough for the declining side to fight.
So what does the “lock in” assumption do? Mostly, it simplifies the solution and clarifies the mechanism driving why war occurs, and it doesn’t seem to be all that distorting, because the sign of the predicted effect of shifting power doesn’t change. Its magnitude might, but its sign doesn’t; it may become a harder constraint to meet the less effective war will be at forestalling or preventing an opponent from growing stronger, but unless war is perfectly ineffective at that (i.e. there’s no lock-in at all), then the essential dynamic identified by Powell holds true. (In a forthcoming paper that used to have some lock-in dynamics in a variant of the main model, I checked the effect of the lock-in assumption in just this way, and I do think I’ll get around to posting this now-removed part of the paper in this space someday…but not today.)
Why do this exercise? Well, yes, it seems obvious that some lock-in is necessary for preventive war, but that doesn’t mean this exercise in general is useless. In fact, I’d argue that it’s responsible any time we come across an assumption that sets off our “now, wait a second, this might be restrictive” alarm bells. Again, every assumption performs the function of restricting what goes on in reality, but objecting to a substantive assumption like this one just because you’re suspicious of it isn’t enough. In other words, it’s not sufficient to just say “I don’t buy it.” And my grad students had better internalize this. (Yes, I’m talking to you.) If we’re to take this enterprise seriously, it’s on both the reader to think hard about the consequences of changing or altering assumptions (if we don’t know that, then we can’t responsibly evaluate them) and, I’d argue, on the individual proposing the theory to know the role their assumptions play, inside and out, in producing their hypotheses. (To digress just a little, if you have an assumption that doesn’t do any theoretical legwork, why even make it?)
To make this last point a bit more practical, imagine that you’ve got a theory that depends on war locking in a share of the benefits. Next, suppose that someone asks you, in a job talk, “Why did you make the lock-in assumption? It seems restrictive, and I suspect it’s restrictive nature is driving your results rather than the parts of the model you’re more interested in.” More often, this will be presented as a starker objection to the content of an assumption, but it’s probably best to think of this as the translation. I told the class that there are two possible answers here, one good and one bad. One answer is, “Well, this is the convention. Other people have done it, so I did it that way, too.” While this answer is common, it’s not a very good one. The person asking the question might not prod you any further at this point, but it’s likely not because she’s satisfied with the answer.
The second (and far better) option is, “You’re right! It is a restrictive assumption, but I’ve verified that my results are robust to something far less restrictive, and I kept it this way to make presenting the logic easier. If I did relax the assumption, we’d still see war occurring for the same reason. The conditions might be harder to satisfy, but my empirical predictions would remain the same.” That’s the kind of answer you want to be able to give about any part of your theory anytime your’re presenting it. Know your assumptions, know what happens if you change them, and be ready to engage in a real discussion—even if it’s just a thought experiment—about what role they play in your theory. Not only does giving this kind of answer engage the person asking the question in a useful and respectful way, it also demonstrates that you know your theory, its assumptions, and—by dint of that—why you make the hypotheses that you do. The person asking the question will come away impressed with you, hopefully, and perhaps more convinced that you’re on to something in your work. And that’s always and everywhere a good thing.