Anomalous authoritarianism: what, which, why?

Abstract

Research on autocracies has gained new momentum in the last decade. One element of this research is the observation that some autocracies are characterised by structural conditions that are normally conducive for democracy. These ‘anomalous autocracies’ have high levels of socioeconomic development and democratic neighbour countries. The study of these cases might expose factors that are decisive for autocratic stability and studying them might give us a better understanding of barriers towards democratisation. This paper contributes to the growing literature on autocracies by mapping anomalous autocracies during the third wave of democratisation, thereby paving the way for systematic case selection in future studies. A large-N analysis of 159 cases (1975–2008) identifies Belarus, Chile, China, Cuba, Morocco, North Korea, Peru, Singapore, Swaziland, Togo and Zimbabwe. In a second step, the paper lays out a theoretical framework that centres on actors and institutions. Rulers must establish elite–elite and elite–mass interaction, and this papers argues that they can do so through quasi-compliance of elites and the masses based on traditional institutions woven into a dominant party. The paper uses the framework to tentatively examine the resilience of authoritarian rule in Swaziland and Morocco, two most-different anomalous cases. In both cases, an elaborate traditional institution has co-opted government, business and rural elites and coordinated interaction within elite circles and with the masses, in turn enabling the remarkable regime resilience.

Keywords:

• deviant
• mapping
• nested analysis
• authoritarianism
• state capacity
• informal institutions

Acknowledgements

The author would like to thank participants at the ECPR Joint Session’s workshop on ‘International Dimensions of Authoritarian Rule’, 2013, the Comparative Politics section at the University of Southern Denmark, the editors and two anonymous reviewers for comments on previous versions of the manuscript. The usual disclaimer applies.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. I use the updated data developed by Cheibub et al. (2010).

2. For Iceland I use Norway and Australia for New Zealand.

3. If changes occur over time, I use Ross (2008) data on oil and gas rents per capita (it denotes the value of a country’s oil and gas production, in constant 2000 US dollars, divided by its mid-year population) to identify exactly when a country became dependent on large-scale revenue from resources. Regarding the particular threshold, I follow Ross (2008, pp. 10–13) dichotomous measure of oil income, which identifies countries with more than US$100 per capita of oil income. On this basis, three additional countries have been coded as dependent: Sudan from 2005, Yemen from 1989 and Equatorial Guinea from 1995.

4. Here, a crisp distinction is preferable to a continuous one since the available data are generally not reliable enough to make fine-grained distinctions and as there is much ‘noise’ from year-by-year fluctuations in the world market prices on hydrocarbon.

5. The scores for Sao Tome and Principe, (North and South) Yemen and Czechoslovakia have been calculated based on information from the Joshua Project (www.joshuaproject.net), the Statesman’s Yearbook (http://www.statesmansyearbook.com/) and the 1985-figures made available by Philip Roeder (http://weber.ucsd.edu/~proeder/data.htm), respectively.

6. Following Bernhard, Reenock, and Nordstrom (2004) and Teorell (2010), the settler colonies are excluded from this set of cases (United States, Canada, New Zealand and Australia).

7. I use 1993 as the starting point after the break-up of the Soviet Union to safeguard that explanatory variables are measured after this critical juncture.

8. Polity, in particular, only includes cases with a population of 500,000 or more. The following cases have been excluded: Antigua and Barbuda, Bahamas, Barbados, Belize, Brunei, Cape Verde, Dominica, Grenada, Kiribati, Maldives, Malta, Marshall Islands, Micronesia, Palau, Samoa, Sao Tome and Principe, Serbia and Montenegro, Seychelles, St Kitts and Nevis, St Lucia, St Vincent and the Grenadines, Suriname, Timor-Leste, Tonga and Vanuatu.

9. I run and report robustness tests of four-year periods subsequently, in which it is possible to include the Muslim majority dummy in the analysis based on the DD measure. The dummy has the expected direction (positive) and is significant on a 0.1 level from 1979 onwards.

10. Generally, goodness-of-fit measures are not as strong as those developed for OLS regression. However, I report the ‘Nagelkerke pseudo R²’, which come close to the standard OLS measure (see http://www.ats.ucla.edu/stat/mult_pkg/faq/general/psuedo_rsquareds.htm).

11. The studentised residuals are preferable, but this is not compatible with logistic regression.

12. The threshold for deviant cases is chosen pragmatically, as none of the reviewed large-N analyses are helpful here, nor are Lieberman (2005) or Gerring (2007). Stepan and Skach’s (1993, p. 7) explicit study of democratic underachievers sets the threshold at 1, which is also adopted in this analysis.