I’ve been reading through R.J. Rummel’s blog which, as one of his regular commentators puts it, is kind of a “public seminar” on the virtues of democracy in terms of inter-state and intra-state violence. Rummel’s blog is, indeed, not just a really excellent academic advocacy website but a nice “public seminar.”

Rummel was one of the first contemporary advocates of the Democratic Peace. The basic idea is that if two states are democracies then they will not got to war with one another. He also developed the idea of “democide” and is a tireless advocate of the argument that democracies do not commit mass violence against their own citizens.

Both of these positions are central to the Bush administration’s strategy of aggressive democratic enlargement. Although one could argue – as the Bush administration and past administrations have – that spreading democracy is an intrinsic good, the geostrategic rationale leans heavily on the putatively pacifying role of democratization. As far as I can tell, this is Oxblog’s reason for existence, and Brad DeLong basically embraced it in our recent back and forth. I used to be a firm believer in the Democratic Peace, but I’ve become far more skeptical over the last few years.

There are a number of variants of the Democratic Peace hypothesis. For example:

1) The Democratic Peace is absolute: democracies are not expected to ever go to war with one another.
2) The relationship is probabilistic: democracies are less likely to go to war with one another than (a) democracies and non-democracies or (b) non-democracies and non-democracies.
3) The relationship is linear: the more democratic country A and country B are, the less the chance that they will resolve disputes militarily.

There are also different arguments about what accounts for the Democratic Peace. For example:

1) Interstate democratic norms. Democracies do not believe they ought to fight with other democracies, which they view as inherently legitimate. Democracies, in consequence, tend to form “security communities” with one another, in which war is no longer an acceptable solution to interstate disputes.
2) Domestic democratic norms. Democracies are committed to non-violent dispute resolution in their internal affairs; they are governed by the rule of law. The project these norms outwards: when they encounter other democracies they believe that they will also resolve disputes peacefully. Other democracies see them the same way, hence they will not go to war with one another.
3) Credible Commitments. Democracies have, all in all, an easier time committing to uphold negotiated settlements – for example, they have to go through a lot of political “work” to pass and discard treaty obligations, while authoritarian leaders can simply rip up agreements – and hence they will, when dealing with one another, be able to credibly commit to some mutually beneficial alternative to warfare.
4) Median voters. In authoritarian states, the leadership doesn’t suffer the costs of warfare. The same can be said in democratic states, but in democratic regimes the leadership has to answer to the people who do: the voters. Moreover, the average voter is likely to be prudent about decisions to go to war; elites will have to “sell” their war to the average voter. Thus, when two democracies have a conflict with one another there are strong domestic restraints against resorting to military force that will work to produce a nonviolent resolution.
5) Free trade. Democracies trade more with one another. This trade leads to interdependence and therefore increases the “costs” of conflict (lost trade, lost investments, etc.).

The nagging problem I can’t get over when I consider the Democratic Peace literature comes from the ways its advocates have dealt with anomalous cases. In general, they’ve developed more and more “refined” definitions of democracy that exclude cases that might look like wars between two democracies (e.g., Britain and France against Germany in World War I). For example, a common requirement in the literature is that >=50% of the population ought to be eligible to vote. This excludes the War of 1812. Two or more consecutive “free and fair” elections is also another common standard, as is a threshold of democratic control over foreign-policy making. The latter, by the way, helps to deal with Wilhelmite Germany.

The problem is that not all of the mechanisms discussed above have any logical relationship to the standards laid out in the statistical coding; in many respects, they look like arbitrary ways of preserving the finding. Moreover, despite the increasingly refined statistical tests of the Democratic Peace, I have trouble getting over a couple of significant issues. For one, the vast majority of democracies with contiguous borders or other potential sources of war came into being after 1945. A great many of these existed within the US security umbrella. The current period, marked by unipolarity, has seen almost no interstate conflicts between any regimes, regardless of whether they are democratic or non-democratic. To put it differently, the number of democracies with opportunities, means, and motives to go to war have been extremely small until the last sixty years. After that time, they have been clustered in ways that explain their lack of conflict independent of their regime type. For another, many of the cases of “near inter-democracy warfare” don’t seem to deescalate for reasons related to proposed mechanisms for the democratic peace. It is not even at all clear that these mechanisms reflect domestic and foreign-policy processes in actual democratic states. This suggests, yet again, that the Democratic Peace may be spurious. There also remains the problem of US decisions to subvert democratic regimes. Although covert, these hardly accord with many of the Democratic Peace’s putative mechanisms.

In the end, I’m still on the fence. Any thoughts?

Filed as: democratic peace