Getting the opposition together: protest coordination in authoritarian regimes

ABSTRACT

It is widely recognized that unified oppositions present a bigger threat to dictators than divided oppositions. In this paper, we use micro-level data on opposition protests in Putin-era Russia to examine the factors that facilitate co-operation among different opposition forces. In particular, we focus on what leads so-called systemic opposition parties – those who have been granted some institutional accommodation by the regime – to join forces with non-systemic opposition forces. We propose a novel permutation-based method for analyzing protest coordination using event count data and find that coordination is most likely on issues of fundamental importance to the systemic opposition’s base supporters. We also find that state co-optation reduces the extent of coordination. These findings illustrate the politically precarious position of “loyal” oppositions under autocracy; they must simultaneously show fealty to the state and maintain some measure of credibility as an opposition party that cares about its supporters’ demands.

KEYWORDS:

• Protest
• authoritarianism
• Russian politics
• regional politics
• opposition

Supplementary Material

Supplemental data for this article can be accessed here.

Notes

1. As such, the IKD does not report on events organized solely by the KPRF. Events that are co-organized by KPRF and non-system groups do appear in the data.

2. Semenov (2017) worries more that the Lankina data overstate the proportion of political events than that the IKD understates them.

3. In the models presented in the main text, lagged unemployment is operationalized by the one-year lagged unemployment rate averaged over the time in the region-convocation, appropriately weighted for the proportions of different years within the convocation. In the supporting materials (online; Section IV), we present several other measures of unemployment and show that our findings are robust to alternative operationalizations.

4. This measure is from the Glasnost Defense Foundation’s three-point scale of press freedom in the Russian regions: http://www.gdf.ru/map/.

5. In Section V of the online appendix, we also examine the effects of a history of protest on the likelihood of coordination. While a number of studies suggest we should see more coordination in places with a history of more protest (Diani, Lindsay, and Purdue 2010; Obach 2010; Roth 2010) there is little statistically reliable evidence in favor of this explanation in the Russian case.

6. For much of our data we are missing information on the specific location of the protest within the city. However, the vast majority of protests in regional cities occur on the main square or street. The exception is Moscow and St. Petersburg, and our results are robust to excluding these regions from the analysis.

7. We assessed the models using various exposure rates (e.g. minimum protest days for the KPRF or IKD, number of protest days for the IKD, number of days in the convocation). Results across these different models were quite variable. Having no statistical grounds on which to choose one exposure as more important than the other, it is important to try a more principled approach.

8. As with all of the figures, we focus on civil rights and material demands here, but present the relevant information on all of the demands in the online appendix (Sections I and II).

9. We normalize by the number of KPRF protests because we assume the marginal distribution of protest for both groups is fixed and the maximum level of KPRF coordination is the total number of events in which it participated. This produces a measure of what proportion of KPRF events were coordinated with the IKD.

10. As the supporting material shows, extra-random coordination is also observed on political and historical issues (9 of 167 legislative convocations).

11. For the controls that vary within convocation, we simply take the within-convocation average. This creates a between-convocation-region design.

12. Here, unemployment is operationalized as the average one-year lagged unemployment. In the supporting material, we operationalize unemployment in several other ways. Our substantive findings are sufficiently similar that the choice of operationalization is clearly unimportant.

13. The results are not interestingly different if we treat the permutation mechanism more systematically. Here, coordination is a random variable and we know its distribution $p\left(y_i\right)$. We could treat our modeling exercise as a Bayesian one where we are modeling $p\left(\beta |y,X\right)=\underset{y}{\int }p\left(\beta |y,x\right)p\left(y\right)d\left(y\right)$, with flat priors over the support for model parameters, integrating over the uncertainty in $y$, via the Monte Carlo method. The posterior mean coefficients for KPRF leadership (Bayesian one-sided p-values in parentheses) are – All events: −0.053 (0.024); Material demands: −0.055 (0.043); Civil Rights demands: −0.0003 (0.0.50).

Additional information

Funding

This article was prepared within the framework of the hse University Basic Research Program and funded by the Russian Academic Excellence Project “5-100.”