Course: POLSCI 353-01 Globalization of Democracy, Duke University
Instructor: Herbert Kitschelt
Teaching Assistant: Peng Peng
“Dictatorship” or “authoritarian regimes” is a residual category for political regimes (Svolik 2012: 20), meaning anything that cannot satisfy the definitions of democracy goes into this category as a dictatorship. Due to this nature, authoritarian regimes differ greatly from one another, with various shapes and sizes, and even demonstrates greater internal diversity than their distinction from democracies (Frantz 2018: 64). For example, there are apparently huge differences among the authoritarian regimes of China, Saudi Arabia, and Singapore, but all of them are categorized as “dictatorships” without sufficient subdivision from each other. It is more than necessary and definitely worthy to draw conceptual distinctions among this residual category of authoritarian regimes.
The Most Convincing Conceptual Strategy
Out of many typologies of authoritarian regimes, I find the dimensional subdivision introduced by Milan W. Svolik most convincing. Rather than creating a vague democracy-dictatorship gradient, he precisely separates the authoritarian regimes from unambiguous democracies based on a simple dichotomy, following Przeworski et al. (2000), Boix (2003), and Cheibub et al. (2010) (Svolik 2012: 22). He then constructs four dimensions at the country level to assess the authoritarian regimes – military involvement in politics, restrictions on political parties, legislative election, and executive elections (Svolik 2012: 32). Then he sets comprehensive scales on each dimension to divide all regimes into different types or levels of restriction or involvement, for instance dictatorships are categorized into “corporate”, “personal”, “indirect”, and “civilian” on the scale of military involvement in politics (Svolik 2012: 34).
One deterministic advantage of this typology is that it leaves no “blank space” for exceptions: all subcategories are collectively exhaustive (Svolik 2012: 38). Since the concept of “dictatorship” or “authoritarian regimes” is already residual – containing the exceptions or “others” – in nature, we cannot allow any cases being left out as an anomaly that does not fit in any subtypes when we conduct a subcategorization. Being criticized in Svolik’s article, the threefold division advanced by Barbara Geddes is a great example of non-exhaustive typology (Svolik 2012: 29). Geddes divides authoritarian regimes into three subcategories: military, personalist, and single-party (Geddes 1999: 123-125). However, it is difficult to fit certain authoritarian regimes into any of these three categories, which are coincidentally missing from Geddes’s original dataset (Svolik 2012: 31). The example used by Svolik to disapprove Geddes’s typology are absolutist monarchies like Saudi Arabia, which fail to satisfy any of these categories (Svolik 2012: 31). Similarly, Frantz (2018) completely excludes monarchies from its typology of authoritarian regimes, claiming that the number is too small and it’s “difficult to make meaningful inferences” (Frantz 2018: 72).
Another profound leverage of Svolik (2012) over the alternatives is the clear-cut boundaries of individual subcategories: it’s mutually exclusive (Svolik 2012: 38). As an example, Svolik constructs three subtypes on the dimension of “restriction on political parties”: parties banned, single part, and multiple parties (Svolik 2012: 32). Apparently, the boundaries between each other are extremely clear, as zero, one, and multiple, that anyone can easily fit any regime into its corresponding category without any potential controversies; and intuitively we can tell that there couldn’t be any overlap among the subcategories. On the contrary, we would have a much harder time to label certain authoritarian regimes like China, Syria, and Thailand when using Geddes (1999) or Frantz (2018), thus generating detrimental overlaps where the threefold typology becomes invalid as it fails to distinguish one from another within the overlapping area.
A third reason to favor Svolik’s typology is its objectivity and quantifiability. Along all of its four dimensions, political regimes are categorized based on simple facts and numerical values, such as number of political parties and type (direct/indirect) of military involvement. Svolik’s categorical typology does not require us to put values and weights on individual indicators, which is the case for most continuous typologies such as the Freedom House and Polity IV scores. These continuous typologies that assess “how authoritarian” each regime is suffers from certain degrees of bias due to the subjectivity of judgement. Although this typology uses a nominal measure (Clark et al. 2017: 166), its coding can be highly quantifiable. Intuitively, the subcategories on the “restriction of political parties” dimension can be readily quantified as , , and (Svolik 2012: 32). Using a combination of specific point values and open intervals, the process of categorization shall be considered highly precise and simple. This also creates potentiality for the use of machine learning and artificial intelligence in categorizing or assessing political regimes.
Why is it Worth?
A conceptual distinction within the group of authoritarian regimes is fundamentally beneficial to the studies of regime behavioral patterns. It is generally easy to track and analyze the regime behavioral pattern in one specific country or historical case and predict its possible future outcomes, but it is much harder to generalize these patterns and outcomes to a broader crowd without a precise and effective typology (Frantz 2018: 67). If there is no internal distinction or division inside the authoritarian family, it is nearly impossible to generalize effective trajectories of behaviors and outcomes applicable to all regimes but only apply some obvious, universal traits to make inferences, most of which tend to be strongly biased especially towards “anomaly cases” in the authoritarian family. Under a defined but flawed typology, however, generalized trajectories could be applicable to the majority of cases within a subtype, but the exceptions and those lie on overlapping areas still suffer from biases and imprecision. Only a clear-defined, mutually exclusive, and collectively exhaustive typology can make the largest number of regime behavioral patterns and outcome predictions generalizable to the largest crowd possible.
On the practical level, a subcategorization of authoritarian regimes would provide much more insights for policy suggestion and planning. Taking the U.S. foreign policy on “advancing freedom and democracy” as an example, we may easily notice that the U.S. government does not only ignore the diversity of democracy regime types but also employ uniform policy frameworks to all authoritarian regimes, disregarding the enormous internal distinction among different regimes. The U.S. State Department’s biannually Advancing Freedom and Democracy Reports list out 11 topics that the U.S. government devises its foreign policy to support democratization elsewhere (Bureau of Democracy, Human Rights, and Labor 2018), but it fails to detect the variability of authoritarian regimes, thus the diversity of other countries’ needs for support, and to devise country-specific policy frameworks for supporting democracies. If a precise and effective typology of authoritarian regimes can be broadly accepted by not only the academia but also policy institutions such as government, international organizations, and civil societies, global and local supports and policy suggestions for democratizations could become less biased and more down-to-earth.
Conclusion
Drawing conceptual distinction among authoritarian regimes is not only favorable but necessary in consideration of the precision of regime behavioral studies and the effectiveness of policy suggestions and actions. Svolik’s typology of authoritarian regimes is theoretically sound and practically viable for us to gain deeper understanding of authoritarian regimes and to take actions for changes.
This paper has references from 6 external sources. Please see the original document to view the full list.
