Thursday, November 3, 2011

Why Wouldn't Autocracies Want to Win Wars?

Earlier this week, I asked why democracies would try harder to win their wars.  Of course, harder is a relative term.  BdM2S2 (or BdM and Smith, depending on the citation) argue both that democracies have an incentive to do their best to win and that autocracies have very little such incentive.  I focused primarily in that post on reasons why I'm skeptical of the former claim.  Now I'm going to tell you why we should doubt the latter part of BdM2S2's story.

My argument is twofold.  First, one of the more celebrated results of selectorate theory itself casts doubt on the notion that autocrats can only win wars by taking money away from core supporters, in contrast to the argument put forth by BdM and Smith in The Dictators Handbook when discussing the Six Day War (see previous post).  Second, we should only expect autocrats to find themselves fighting a war when they genuinely care about the issue in dispute.

First, one of prominent results of selectorate theory is that leaders of small coalition systems behave like kleptocrats (see chapter 4 of The Logic of Political Survival).  As a result, such leaders (the ones we typically have in mind when we refer to autocrats) tend to amass rather significant personal fortunes.  For example, it appears that Qaddafi stashed away $200 billion.  Other members of what BdM and co-authors call the Haul of Fame include Indonesia's Suharto, Zaire's Mobutu, the Philippines' Marcos, and Sudan's al-Bashir.  In contrast, democratic leaders are forced to spend just about all their available resources on satisfying their supporters.  It is precisely the inability to amass a rainy-day fund that might help be drawn upon to dole out extraordinary compensation to key supporters during hard times that explains why democrats tend to lose power following economic crises and natural disasters, while dictators manage to cling to power.

So why doesn't the rainy day fund come into play during war?  Why are we supposed to believe that Egypt lost the 1967 Six Day War because trying hard to win would have required Nasser to deprive his key supporters of the kickbacks to which they'd grown accustomed?  Are we to believe Nasser couldnt' have simply drawn upon his own personal wealth to fund the war effort, without requiring any real sacrifice from his supporters?

At this point, you are no doubt thinking, yes, fil, but you forget that we assume that all any leader cares about is remaining in power.  It might be true that Nasser could have tried hard, if he wanted to, but unless he had to try hard in order to remain in office, there's no reason he would.

To which I would respond with two points.

First, selectorate theory tells us nothing about what leaders like to spend their money on.  It is no more unreasonable to think they might not want to influence policy outcomes purely because it brings them personal satisfaction.  The world does not lack for wealthy billionaires who spend their own money trying to shape political outcomes (e.g., George Soros, the Koch brothers), and I suspect that most of us who (sadly) are not billionaires would consider spending at least some of our vast fortunes on changing a few things if we were.

Put differently, it is no more appropriate to say that Nasser would or would not want to use his money to see Israel defeated than it is to say that Nasser would or would not want to use his money to purchase a mansion.

The authors are quite clear, and appropriately so, in stressing that they do not assume that the reason leaders wish to hold power is purely for its own sake, or purely to use the levers of power to enhance their ability to buy mansions and attract wives.  Rather, they assume that all leaders must first and foremost concern themselves with remaining in power because even those who do genuinely hold policy preferences must be in power in order to control policy.  As the phrase "good policy is bad politics" makes clear, their story for why we observe failed policies in autocracies is not that autocrats don't care, or are bad people, but that they couldn't make things better even if they wanted to.

Of course, just because the core assumptions of the theory do not tell us what leaders will want to spend their money on, we might nonetheless ourselves impose the auxiliary assumption that very few dictators will in fact care much about the outcomes of international crises.  Sure, they are permitted to do so under the core assumptions of selectorate theory.  But that doesn't meant that they do.

Fair enough.  And I won't say it surprises me that few dictators draw upon their personal fortunes to build hospitals and schools, though they certainly could.

But consider this: if autocrats don't care about the issues in dispute in the wars they fight, why the heck are they fighting them in the first place?

The trivial answer is that BdM2S2 apparently don't believe in anarchy, at least not in the sense that IR scholars typically use the term.  I say this because their proof of the democratic peace fundamentally relies on the assumption that the interactions preceding war can be characterized as a unilateral choice between accepting some "negotiated" outcome (whose terms are given exogenously and need not be accepted by both players) and war.  So if Leader 1 is democratic and Leader 2 is autocratic and Leader 1 doesn't like the terms of the "negotiated" agreement, then we're going to get a war.

But let's set that aside.

BdM and Smith have also argued that democrats often buy policy concessions from autocrats using foreign aid (see here and here).  Given the inefficiency of war, and given that democrats often find it easy to get what they want from autocrats anyway, must we not assume that when a democratic state goes to war with an autocratic state, it clearly indicates that the autocratic leader did in fact care about the issue?  Why refuse to take bribes only to go to war, then once in war, despite having resources available that could have produced victory, sit back and watch your army get crushed?  Even if we believe that war threatens the ability of autocrats to remain in power less so than it does democrats (and it's not clear that this is the case), that's not quite the same as explaining what could possibly possess an autocrat to fight a war it didn't have any interest in winning when there were alternatives available that would have avoided whatever costs the autocrat will incur, and perhaps even put money in their pockets besides.

If you ask me, it simply will not do to say that the reason why Israel defeated Egypt and Syria in 1967 is that the latter didn't care about the issue and had no reason to try to win.  Nor do I find it persuasive to argue that, as a general rule, autocrats fight wars haphazardly, just for kicks, without really caring about the outcome.  I do think that, all else equal, it would seem like the assumption that autocrats don't care much about the outcomes of international crises is a reasonable one.  But if that really was true, I don't see why they'd ever fight wars.  The very fact that they do tells me that they must, at least in some cases, genuinely care about policy outcomes.


  1. Phil,

    First off, thoroughly enjoy your extended posts like this one. Was going to respond to your previous post on the selectorate theory but this will do just as well.

    You raise a good argument against the assumption that autocrats care less about international crises. I would add that since selectorate theory is premised on the assumption that all leaders, democratic or autocratic, care about staying in power (they just face different constraints), then it must be the case, as you've argued, that both types of leaders will care about the outcomes of international crises and wars. The issue then becomes which type of leader cares more. BdM&S argue that it is the democratic leader who does but their reasoning is incomplete.

    I suppose one could introduce audience costs as a fix for the "try harder" argument, i.e. democratic leaders incur higher audience costs if they issue threats and back down (or if they lose wars) but it is still not clear why that would be the case. [I know I'm using terms loosely here.] Sure, democratic leaders have a larger winning coalition-to-selectorate ratio and thus run a greater risk of losing office but no democratic leader was ever killed after losing a war (I think...). On that note, Jessica Weeks has an enterprising research agenda on authoritarian regimes and wars, with which you are undoubtedly familiar. All this is to say that given the threat of death or other nasty consequences of losing wars for autocrats, and that "life" is probably preferred to "staying in power with Maseratis and Lamborghinis", ought not autocratic leaders "try harder", or at the very least, "as hard as" democratic leaders?

    Now one can come back with the following: democratic leaders also know that support for war decreases as casualties mount. This explains why democracies tend to fight shorter wars (in which they tend to be victorious). There's nothing like soldiers in body bags to obliterate a democratic leader's chance of reelection. Consequently, they will "try harder" to win, _and_ to win quickly. However, would not his autocratic adversary know that? There is anecdotal evidence (for the sake of argument) to suggest that autocrats are more than familiar with this, and some would go as far as to fight a protracted war, sacrificing thousands to wear down the democratic leader's support. Would this not count as "trying hard" on the part of the autocrat?

    The "resolve" argument sums all that up. BdM&S (and a fair number of IR scholars as well) claim that democratic leaders are more resolute than autocratic leaders and therefore, are more likely to win wars. This, however, is an assumption, as you've pointed out, and a key one that deserves to be analyzed more carefully. Not to mention resolve is a tricky thing to formalize and measure empirically.

    So, if both types of leaders may be equally resolute, what explains varying war outcomes? Well, there is another critical aspect that affects war outcomes: capabilities. Now this is something more tangible, and there is an abundance of research on this. Perhaps democracies are more likely to prevail in conflicts because (1) they have better military technologies; (2) they are better able to raise the funds for war, either by borrowing (credit) or raising taxes; (3) they tend to have higher-educated populations for research and development, etc.

    Returning to the example of the Six-Day War, I concur with you, Phil, that the reason for Israel's victory is not because her adversaries didn't care about the war outcome. A more plausible (and straightforward) reason may simply be the difference in war capabilities writ large. Now the interesting question is why that would be the case. I think the selectorate theory may be able to provide answers, though not in its current (yes, even the reformulated) form.

  2. Thanks, TT.

    You raise a lot of good points.

    I'd be careful about conflating audience costs with a punishment for losing a war. Costs for backing down have very different effects from costs for losing. If democracies are more likely to be punished for losing wars, we might well expect them to try harder and win more often. But if they are punished for backing down after threatening to use force, that could well make us expect them to lose the wars they fight more often.

    There's some debate about whether support for war decreases as casualties mount. If it was as simple as that, then WWII should have been the least popular war in US history, rather than the most popular one. Gelpi, Feaver and Reifler offer a strong argument that if the public continues to believe victory is likely, they will support very high levels of casualties (e.g., WWII) and when they believe defeat is likely, they will not tolerate even relatively low levels of casualties. There's also evidence that casualties as such are not what matter, so much as opposition rhetoric (which admittedly may be a function, in part, of casualties). See Berinsky's 2007 article or his book on this point.

    I think these quibbles matter a lot because, though it is true that, *on average*, democracies fight shorter (interstate) wars, 5 of the 7 longest wars that occurred since 1945 involved democracies. There's a great deal of variation there that isn't easily accounted for by the conventional wisdom. And if we consider extrastate wars, it becomes really hard to say that democracies only go to war when they expect a quick and easy victory while keeping a straight face.

    You're absolutely right that one of the reasons we should be skeptical of the claim that democracies fear defeat in war more than autocracies is that autocrats tend to have a pretty terrible fate waiting for them after they lose power, whereas former democratic leaders live the good life of speaking tours, writing books, or, if they so desire, quiet retirement. Hein Goemans has a lot of really good work on this, including the piece I linked to above.

    Reiter and Stam do take CINC scores into account, as well as the quality ratio (milex per soldier), so I'm not sure capabilities as such would explain the supposed difference between democracies and autocracies when it comes to differences in war outcomes (a difference I'm not sure is anything but a statistical artifact). Granted, it's hard to measure non-material causes of martial effectiveness (R&S *do* find that democratic armies exhibit greater leadership and initiative, for what it's worth). They also do not account for variation in access to credit. So there could be something here, if we think of capabilities much more broadly than CINC scores. That's a good point.

    In his analysis of war outcomes in Win, Lose or Draw, Stam finds that surprise attacks can make a difference, but it's rare that states achieve them. That is, surprise does not account for a large amount of variation in the war outcomes that we've observed, but where it comes into play, it can have a big impact. The Israelis conducted surprise attacks in 1967, and destroyed the entire Egyptian air force in a matter of hours. I think that's got to be part of the story. Having established air superiority from the very outset would not have been a trivial advantage, I don't think.

    I'm certainly open to the idea that domestic politics plays a role in explaining how states behave in war, and when they go to war. But more and more, I find that simply contrasting democracies with autocracies isn't very useful. There's too much variation within the behavior of each group, and too much of what's been said when trying to establish simple patterns rests upon fairly weak arguments or problematic empirics.

  3. I should also add that Slantchev 2004 found that there is in fact no relationship between democracy and war outcomes once we take into account the relationship between democracy and war duration. I'm not sure why that result doesn't get cited more often. It wasn't his headline result, but still.

    I'd also note that Werner 1999 analyzed the actual provisions of war-ending agreements and found no evidence that democracies get better agreements.

  4. Quite right, Phil. I was sloppy with respect to the distinction between costs for backing down and costs for losing. With regards to costs for backing down (that is, audience costs), I believe Weeks' work does show that (some) autocrats can and do face rather strong audience costs from elites, which goes against the argument that domestic audience costs lead democratic leaders to try harder. On costs for losing, I think that autocratic leaders might incur greater costs for reasons you and I (and Goemans) agree. Taken together, it would suggest that (some) autocratic leaders ought to try harder than democratic leaders. If so, the question remains: if it's true that democracies win wars more often, why?

    Good points on the war support-casualty-war duration argument, and thanks for the references. I'm not as well-versed in that area as I'd like. I ought to revisit Slantchev 2004 as well.

    On capabilities: yes, I do think we should consider capabilities more broadly beyond military or technological superiority, especially in cases of protracted wars. My knowledge is less-than-adequate on how long wars typically last but I'd imagine that the longer the war, the more important a role other capabilities ought to play.

    You bring up an interesting point on surprise attacks. Certainly, Israel's effective destruction of the Egyptian air force (while still on the ground, nonetheless!) must partly account for her swift victory. There is, however, as far as I know, a lack of theorizing about the decision process behind conducting preemptive strikes and their effectiveness in determining war outcomes. I imagine, by most accounts, that a preemptive strike is a declaration of war. Would a preemptive strike reflect a bargaining failure then? Could a nation decide to do so while in the bargaining process; that is, it exists as an outside option? On the latter, when would a country decide to do that? Hmm...a possible topic for a paper perhaps.

    Domestic politics definitely plays a role in explaining war initiation, duration and outcomes. Modeling this, however, would involve new challenges. I too agree that a dichotomous distinction between democracy and autocracy is too simple and obscures variation across each group.

    I see the selectorate theory as an attempt to generalize beyond democracy vs. autocracy; unfortunately, BdM & Co. stop short of examining the variation within each group (thereby neglecting a huge literature in comparative politics). As a (wannabe) formal theorist, I do understand why but there is surprising little work that tries to extend their core model.

    That said, there is emerging work that delves further into different types of W/S ratios within autocracies in comparative politics but that is a minor extension that leaves much of the issues in the model unaddressed. For instance, the points that you brought up in your earlier post regarding public goods and the view of war outcomes as a public good. Or the decision to model the selectorate as a non-strategic actor.

  5. Hey, TT.

    This is a good discussion. Thanks for the thoughts!

    I think that's an interesting point that broader conceptions of capabilities would matter more the longer the war lasts. I'm not aware of any work on that, but it would be easy enough to examine. It would be nice to see someone do that.

    On preemptive wars, I think that one way to think of that is through the lens of commitment problems. I discuss them in that context in my War & International Security class. That's also how Frieden, Lake and Schultz discuss them in their introductory text. First strike advantages can also be thought of as a commitment problem. The FLS text discusses that too.

    I agree with you that selectorate theory offers a nice attempt to go past democracy versus autocracy, in that they characterize governments along two institutional dimensions, both of which vary continuously. But in practice they still end up using democracy versus autocracy language a lot.

    What I'd really like to see is more work that focuses on variation within countries even while institutional arrangements are fixed. If we could do a good job of characterizing behavior without getting into the moving parts, that would be great. It would be really useful. And many people think we can. But, personally, I'm still a bit skeptical of that. I do think there is some value in such an approach, to be sure. But I think that, if we're going to talk about domestic politics (and I do think there's a lot we can explain without even doing so), we might be better off focusing on the actual behavior of actors within the state, rather than just focusing on the institutional arrangements. That's a whole lot messier, to be sure. But I think it would be more useful.