`We put a wrap on the Spring slate of CC Virtual Workshops discussing an interesting paper by Adam Scharpf that begins with offers the following figure as a puzzle: why would democracies and autocracies violate rights at equal levels when engaged in counter-insurgency when the former violate rights at substantially lower levels when not engaged in COIN operations?
Jessica Di Salvatore, Shanna Kirschner, Jonathan Powell, Jacob Shapiro and Juan Tellez served as discussants of the paper, “Regimes, Organizational Rivalry, and Repression in Counterinsurgency Wars,” and we had a fun, stimulating discussion. You can watch the video here.
Scharpf argues that slack is greater within coercive forces in democracies compared to autocracies, which permits inter-service rivalry to generate human rights violations in the COIN campaigns of democratic countries to rise to the level of those committed by autocratic regimes. That is, Scharpf argues that interservice rivalry produces a focus on short-term measures of “success” by commanders, which we can summarize as a “body count” premium. In autocratic countries leaders need not restrain commanders from a body count approach, beyond limits placed by international scrutiny. As such, violations will be substantial. But the need to “coup-proof” leads autocratic leaders to use coercion as a means to sustain leadership (e.g., bloody purges), and this limit on agency slack among coercive agents in autocracies prevents them from violating rights wantonly (due to international pressure). For democracies, however, both domestic and international pressure incentivize executives to limit rights violations during COIN conflicts. However, they are unable to “discipline” coercive agents like autocrats can, thus where inter-service rivalry exists, the can get a little out of hand.
Put another way, Scharpf expects the median level of rights violations to be equal between autocratic and democratic countries fighting insurgencies, but the former to have lower variance than the latter. The following figure shows that this is the case.
As always, there are measurement issues. Scharpf needs to interact a measure of democracy with a measure of inter-service rivalry. To do this, he creates a dichotomous indicator of rivalry from Pilster & Böhmelt‘s effective number of services measure for the former and selects the Polity dataset for the latter. Both indicators received attention within the CCVW group. Scharpf recognizes that the Pilster & Böhmelt data is a proxy indicator, and the limits that creates for the internal validity of his inferences. Suggestions for alternatives included construction of either a fractionalization or polarization index using the Pilster & Böhmelt data or using a mass killing indicator such as UCDP’s one-sided violence or the PITF Atrocities data.
Other issues concerned the use of democracy itself. Earlier work by Davenport, Armstrong and Moore as well as Conrad and Moore revealed similar dynamics to the Scharpf piece but they paid a bit more attention to the measure of democracy employed. For example, Vreeland has definitively shown that using Polity without extracting the influence of civil conflict on the measure is highly problematic. Additionally, the VDEM project identifies 7 different types of democracies and 300 measures. Gone are the days of just using Polity without confronting this variation. There was a concern that democracies were less likely to see insurgencies and that this selection bias would need to be addressed.
Concerning COIN there was also some discussion about how this was differentiated from/overlapping with state repression. Do not both involve police raids, curfews, interrogation, torture and mass killing? If so, then how could one separate discussion of COIN from repression? Related as civil war involves repressive action/COIN, does focusing the analysis on repression during civil wars muddy the waters even further?
Shapiro explained that he was not persuaded about the internal validity of the inference, and encouraged Scharpf to invest more effort thinking carefully about the joint parameter space of the independent variables, and how he could use that information to identify some specific cases to study qualitatively to really nail down the agency slack—rivalry process generating a focus on body counts. The advantage of this approach is it severs the project’s dependence upon the Pilster & Böhmlet measure as a proxy for a concept it doesn’t measure well, and further permits direct (qualitative) measurement of the other moving parts ignored in the regression equation.
The downside to Shapiro’s recommendation is that such a study would be vulnerable to the retort: “sure, but how do we know those aren’t outlier cases?” Shapiro would explain that is why selecting the cases based on relevant values from the joint parameter space is so important. What we enjoyed about this exchange is the improvement in the quality of this sort of discussion relative to what we would have expected when we entered the discipline back in the 1990s. To his credit, Scharpf wants both external and internal validity. His current paper seeks to score high on the former, but scores low on the latter. The Shapiro proposal would score high on the latter and low on the former. Whether he pursues a multiple paper route or a book project, Scharpf (and the rest of us!) will do well to pursue more than one design to test the implications of his theory, and a single paper does not have the space to do so effectively.
A number of other interesting issues we do not touch on here were raised, and we encourage you to watch the video.
In wrapping up the Spring 2016 CCVW we wish to thank everyone who participated. We are delighted to see so much interesting work being done, and to be able to play a role in both community building and strengthening individual projects. Please keep an eye out for our Fall 2016 Call for Papers in June. We look forward to another great slate of workshops and continuing to learn a great deal about conflict and peace, have some fun and (of course) shake things up a bit.
@engagedscholar & @WilHMoo