Waves of autocratization and democratization: a rejoinder

ABSTRACT

This is a response to Luca Tomini's “Don't think of a wave!” and Svend-Erik Skaaning's “Waves of autocratization and democratization”. With this rejoinder we make three arguments: First, the question how global waves shape autocratization processes and regime transformations is now more urgent than ever. Since 1994, civil liberties and political rights of one third of the global population have been substantially and increasingly reduced due to autocratization. Second, waves of any concept can only be studied meaningfully if the underlying concepts as well as the waves are clearly defined. We argue that the conceptualization of episodes of regime transformations (ERT) in the ERT dataset provides exactly such a clear state-of-the-art empirical mapping of processes of democratization and autocratization at the national level. In addition, we highlight how our conceptualization of waves builds upon Huntington, 1993. It even improves upon it by allowing for overlap and modelling it closer to its real world counterpart. Third, we view the present debate as a prime example of fruitful scholastic argument. Ultimately, this is what will generate a better understanding of global trends of democratization and autocratization.

KEYWORDS:

• waves
• autocratization
• episodes
• regime transformation
• conceptualization
• measurement

Why waves matter

Since 1992, an unprecedented 36 democratic regimes have broken down and autocratization processes have been set in motion in more countries than ever before.¹ Naturally, such developments call for closer inspections and the effort in this issue “Don’t think of a wave!” (DTW) by Luca Tomini is highly praiseworthy.² There was a similar piece published recently in Democratization by Skaaning, 2020 titled “Waves of autocratization and democratization” (WAD) also engaging with our work in a similarly constructive manner.³ We think these constitute important additions to the broader discussion of how to define and measure waves of autocratization. Yet, we arrive at different conclusions. Further analysing and understanding these differences is what will ultimately move the field forward towards an even better erudition of both democratization and autocratization.

In this response we make three arguments. First, we agree that global processes of autocratization deserve more attention and the wave conceptualization laid out in Lührmann and Lindberg⁴ and further refined in the Episodes of Regime Transformation (ERT) Dataset⁵ actually implements precisely the bottom-up conceptualization approach called for by DTW⁶ and additionally provides state-of-the-art measures on processes of regime transformation. Second, the Skaaning versus Lührmann/Lindberg debate is a prime example of fruitful scholastic argument that advances in science are based on, rather than an “inconclusive debate”.⁷ Third, the concept of a “wave” has a precise definition from the natural world of waves in water, radiology and so on, which we argue brings merits to political science as well if understood properly.

But before we go into those arguments, allow us to clarify two things. First, our endeavour to measure waves of autocratization has always been about institutional configurations at the national level. DTW rightly argues⁸ that drivers at the subnational level likely bear importance for how regimes evolve.⁹ Yet, this does not harm the value of national level and cross-national analysis. It is simply a different approach. We view such different approaches as supplementary rather than substitutes, which DTW seem to do when suggesting that scholars abandon the study of waves for other approaches.¹⁰ Second, since the publication of the “Third Wave” article¹¹ the research team at the V-Dem Institute has amplified and extended the work on episodes. We have now developed a comprehensive approach and dataset on “Episodes of Regime Transformation”.¹² It captures episodes of democratization and autocratization along with detailed information on their evolution and eventual outcome. Maerz¹³ introduces the dataset and early studies already clearly demonstrated its usefulness in assessing how democracies prevail in light of autocratization,¹⁴ how episodes of liberalization unfold in autocracies that may or may not lead to democratic transitions¹⁵ or how such an episode approach can provide better estimates of the economic growth effects of democratization.¹⁶ Thus taking the episode approach introduced in the “Third Wave” article¹⁷ further, we sought generalized answers to two questions that we use in the following sections to organize our response to DTW, while also answering some of the critique levelled in WAD.¹⁸

What is autocratization?

DTW calls upon us to  … straighten the definition of autocratization up and critiques the “Third Wave” article’s negative definition of the phenomenon – autocratization as a decline of core institutional requirements for electoral democracy.¹⁹ We did discuss different options in the “Third Wave” article, including what Tomini and Wagemann suggest²⁰ and made an explicit choice. We concur that the negative definition has three clear advantages in an analysis of global regime transformations: First, it follows the widely shared intuition where a deterioration of democratic qualities means becoming more autocratic. Second, the approach allows us to conceptualize autocratization and democratization on the same spectrum leading to the third and most important advantage: While the “Third Wave” article focused on autocratization, it is now possible to jointly analyse global movements in both directions along this spectrum, which is only possible when definitions of autocratization and democratization are mutually exclusive, as already discussed in the “Third Wave” piece. The V-Dem dataset is explicitly constructed to cover varieties of democracy.²¹ Once a dataset on a varieties of autocratic institutions that is comparable in temporal, spatial scope, and attribute-scope becomes available, it will most certainly be a fruitful endeavour to further analyse autocratization episodes using it. For the time being, rather than abandoning the idea altogether, we concur that the episode approach allows for original analyses that contribute to improving what we know about current trends in the world.

DTW also disapproves of the use of Electoral Democracy Index (EDI) as a basis for calculating episodes, rather than Liberal Democracy Index.²² That is certainly a reasonable point to make. We take a position in line with much of the literature that electoral democracy forms the core element of a definition of democracy.²³ To us, it therefore provides an adequate starting point for the analysis of episodes of regime transformations. Naturally, episodes of transformations based on the Liberal Democracy Index would also highly interesting and relevant. For this reason, the ERT package available on GitHub allows for flexible adjustment of all thresholds (for example for episode start and end) as well as for the inclusion of different democracy indices as the basis for episodes.²⁴

DTW also suggests

defining autocratization only those processes of change that imply a transition from one regime to another, referring to a robust literature that proposes a widely accepted four-fold distinction between political regimes (…) based on the three dimensions affected by autocratization: executive limitation, public contestation, and political participation.²⁵

We find this old debate between crisp and gradual approaches to how democratization and now autocratization should be understood, somewhat beside the point. Both have their uses for different research questions. Using crisp regime transitions is one relevant way to study autocratization. Meanwhile, a country can undergo significant changes for the worse without breaking the threshold from one regime to another, and this can also be important to study. The episodes approach undergirding the “Third Wave” article allows for both, simultaneously. The now fully developed ERT identifies both cases in which autocratization processes were set in motion but did not culminate in democratic breakdown, and processes that lead to a regime transition. The unique, unified ERT approach is essential as it allows for empirical comparisons with adequate empirical referents. For example, in Boese et al. (2021), we put forward a two stage conceptualization of democratic resilience based on the episode framework.²⁶ In the first stage democracies can be onset resilient by avoiding autocratization altogether (e.g. contemporary Switzerland). Once autocratization has set in, democracies can be breakdown resilient in the second stage by avoiding democratic breakdown. To analyse the correlates (and eventually also drivers) of resilience, this selection into autocratization has to be properly accounted for in any statistical model. No democracy can break down without previously having selected into an autocratization period. Considering only cases that culminated in democratic breakdown implies turning a blind eye to reality in which the end result of democratic erosion is far from predetermined. In fact, our episode approach does not require a transition from one regime to another (as DTW does).²⁷ A period of deteriorating democratic norms within a liberal democracy, for example, can be captured as an episode of democratic regression that starts and ends while the country is a liberal democracy In light of the “new face” of autocratization – gradual erosion rather than abrupt changes²⁸ – including countries where the autocratization process was halted before democratic breakdown is indispensable.

To identify instances of autocratization a central choice is the selection of a threshold of (annual) change to qualify. In WAD, Skaaning passes judgment on the bar as too high in the “Third Wave” article and suggests  … to consider all negative movements on the autocracy-democracy spectrum as instances of autocratization,²⁹ while Tomini in DTW seems to assume that we did what is suggested in “Waves of Autocratization and Democratization”:

The continuous perspective leads, by definition, to reduce the problem of “false negatives”: potentially no case of autocratization will be excluded from the analysis. […] the analysis will be full of cases of alleged autocratization when, in reality, they are not.³⁰

The “Third Wave” sought a balance between the two that made both theoretical and empirical sense, and the refined ERT inclusion criteria (episode start, annual and cumulative change and episode termination) filter out countries in which small annual changes along the Electoral Democracy Index occurred but no broader process of autocratization has set in. Thereby, we avoid conceptual stretching and instead focus on actual periods of autocratization. Note, that our argument is not solely about measurement uncertainty (if it were, we most certainly could proceed as suggested by Skaaning and include the uncertainty estimates).³¹ We agree that even small changes can be real. However, not every small change is indicative of a broader trend of autocratization or democratization. Institutional configurations such as the ones measured with the Electoral Democracy Index are fluid and evolve ever so slightly during most years. The average of the absolute values of annual change for the Electoral Democracy Index is Δpolyarchy = 0.0145.³² Our criterion for a possible autocratization period to start is Δpolyarchy_start ≤ −0.01.³³ Our episode start criterion is hence of a slightly smaller magnitude than the average absolute change in annual Electoral Democracy and we thus give even small changes “the chance” to be indicative of an autocratization episode. Yet, a simple annual change does not provide enough information to judge whether a country is undergoing an episode of autocratization. For this reason, the variety of other criteria denoted previously (and explained in detail in Episodes of Regime Transformation Dataset (v2.2) Codebook³⁴) need to be fulfilled in addition to the start criterion to constitute an episode of autocratization. That way, we ensure a maximum amount of concept-measurement validity in our coding of episodes of regime transformations.³⁵

What is a wave?

We are a little surprised that the concept of a “wave” is questioned in DTW and WAD.³⁶ After all, it has a precise, universal meaning that comes from the natural world of waves in light, water, radio frequencies, gravitation, and so on.³⁷ In the “Third Wave” article and the work following from it now, it is a metaphor chosen purposefully since the empirics behave in wave – like fashions when it comes to democratization and autocratization. But to suggest that a “wave” is something which it is not as both Tomini and Skaaning do in DTW and in WAD, is arguably not very helpful.³⁸ For example, waves can naturally overlap as one resides at the same time as another builds up. Here we have a different perspective than Huntington, who placed his waves of democratization and reverse waves in different time periods.³⁹ Waves are not mutually exclusive in that one does not have to end before another one gets underway.

The recurring building up of, cresting, and residing of countries undergoing democratization and autocratization as shown in Figure 1 clearly fits such a definition. The dark, solid line in Figure 1 displays the number of countries that experience autocratization as defined by the Episodes of Regime Transformation (ERT) Dataset⁴⁰ – it is almost identical to the original depiction in the “Third Wave” article. Three major waves of both democratization and autocratization are clearly manifest. Even if one were to deny that these are interconnected processes amounting to waves, still the unprecedented amount of countries (and democracies – depicted by the dark dashed line) under autocratization in recent years gives cause for concern. One aim of our investigation of waves of regime transformation is to uncover the regularities and patterns behind this proliferation of both democratization and autocratization processes across space and time.⁴¹

Figure 1. Number of countries experiencing autocratization and democratization since 1900.

When does a wave start?

Irrespective of whether it is an electrical wave, a physical wave, a function like sine or cosine, or other waves – the concept of a wave is inherently part of a cyclical pattern. As such, start or end points of one individual wave as part of the pattern are artificial (and different from when the whole pattern starts and ends). Where does a wave in the ocean start? At the bottom of the sea or once it rises to the water’s surface? Either of these definitions of a start point are valid – yet, what is key is that both agree on the general existence of waves.⁴² The definition of waves put forward by Skaaning, 2020 is at odds with what waves are and how to measure them based on real phenomena in the natural world. A common wave – measure across oceanology, physics, and radiology originates at the lowest, crests at the highest and eventually to ends in/returns to the lowest points of the curve again. Alongside that another wave can be building up. Given the characteristic periodic structure of waves, other choices of start and end points are equally valid. For example, in the context of mathematical functions like sine or cosine waves are often modelled to begin and end with an inflection point. We draw on that intuition in the “Third Wave” article and “define as an autocratization wave the time period during which the number of countries undergoing democratization declines while at the same time autocratization affects more and more countries”.⁴³ To be precise, “We identify an autocratization wave starting when the number of democratization episodes begins to decrease at the same time as autocratization episodes increase for two years in a row. It ends when autocratization episodes decline in number and democratization episodes increase over the next four years”.⁴⁴ We think that is a more accurate foundation than Huntington’s operationalization of a wave of democratization as “a group of transitions from nondemocratic to democratic regimes that occur within a specified period of time and that significantly outnumber transitions in the opposite direction during that period of time”.⁴⁵ Thus, with the ERT episode approach to measuring regime transformations⁴⁶ that builds directly on the “Third Wave” article, we seek high concept-measurement validity beyond using simple changes on a democracy index as WAD does but at the same time capture the wave-like empirical patterns contrary to what DTW seeks to argue. The operationalization of waves in ERT is conceptually focused to derive a measure of waves of regime transformation that follows from science. Returning to the ocean metaphor: Our approach results in waves based on autocratization episodes that can be traced back to actual periods of substantive regime transformations in their respective countries. Having said that, Huntington’s, WAD’s, DTW’s, or the “Third Wave” article’s approach to measuring waves are all legitimate and can be useful endeavours of course, we are united in the intent of empirically measuring a complex multidimensional social phenomenon and ultimately, to improve our knowledge of what drives or slows down these processes of regime transformations. Yet, we concur that it is not helpful to engage in conceptual stretching when it comes to the nature of “waves”, and not to disregard them when they manifest.

Conclusion

Does deepening the debate about waves risk “taking the entire discussion down a blind alley”?⁴⁷ Most certainly not. At the most fundamental level, constructive clashes of minds are what scientific advances are predicated on. Disagreement forces us to sharpen our arguments, improve on the data we have, and question our methods in the never-ending striving towards improved knowledge. Second, waves of any concept can only be studied meaningfully if the underlying concepts are well defined. A proper conceptualization of autocratization (such as the one underlying the ERT dataset) therefore is key to the study of waves, alongside an appreciation for the nature of waves as they appear universally in the natural world and captured in the natural sciences.

Third, we view research (figuratively speaking) as a process of going down dark alleys and shedding light on previously unexplored terrain. The question how global waves shape autocratization processes and regime transformations in general is precisely such an understudied topic. Shedding light on it is critical. Since 1994, civil liberties and political rights of one third of the global population have been substantially, and increasingly reduced due to autocratization,⁴⁸ in what actually takes the shape of a wave building up while the previous wave of democratization is subsiding. For these reasons, we believe the study of waves of regime transformations is now more urgent than ever. We encourage others to follow DTW and WAD to engage in further constructive exchanges on waves of regime transformations.⁴⁹

Author contribution statement

Vanessa Boese took a lead on conceptualizing and drafting this response. Anna Lührmann and Staffan Lindberg provided extensive input at all stages and made substantial contributions to revisions of the text. We thank Palina Kolvani for skillful research assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

We recognize support by the Knut och Alice Wallenbergs Stiftelse to Wallenberg Academy Fellow Staffan I. Lindberg [grant number 2018.0144]; by H2020 European Research Council [grant number 724191], PI: Staffan I. Lindberg; as well as by internal grants from the Vice-Chancellor’s office, the Dean of the College of Social Sciences, and the Department of Political Science at University of Gothenburg.

Notes on contributors

Vanessa A. Boese

Vanessa A. Boese is a postdoctoral research fellow at the Varieties of Democracy (V-Dem) Institute at the University of Gothenburg, Sweden. Her research covers how to measure democracy, democratic resilience, regime transformations, and the interaction between trade, development, democracy, and peace. Her articles are featured in journals such as the British Journal of Political Science, Democratization, the Economics of Peace and Security Journal or the International Area Studies Review.

Staffan I. Lindberg

Staffan I. Lindberg is Professor of political science and Director of the university-wide research infrastructure V-Dem Institute at the University of Gothenburg, founding Principal Investigator of Varieties of Democracy (V-Dem), founding Director of the national research infrastructure DEMSCORE, Wallenberg Academy Fellow, author of Democracy and Elections in Africa as well as other books and over 60 articles on issues such as democracy, elections, democratization, autocratization, accountability, clientelism, sequence analysis methods, women's representation, and voting behavior. Lindberg also has extensive experience as consultant on development and democracy, and as advisor to international organizations.

Anna Lührmann

Anna Lührmann is Assistant Professor in the Department of Political Science at the University of Gothenburg and Senior Research Fellow at the Varieties of Democracy Institute (V-Dem). Her research on autocratization, elections and democracy aid has been published among others in American Political Science Review, Electoral Studies, International Political Science Review, and the Journal of Democracy.

Notes

1 Boese et al., “How Democracies Prevail.”

2 Tomini, “Don’t Think of a Wave!”

3 Skaaning, “Waves of Autocratization and Democratization.”

4 Lührmann and Lindberg, “A Third Wave of Autocratization.”

5 Edgell et al., “Episodes of Regime Transformation Dataset.”

6 Tomini, “Don’t Think of a Wave!”

7 Ibid., 2.

8 Ibid., 6.

9 Subnational processes still influence national level processes over time. Our aim is to study those patterns across time.

10 Tomini, “Don’t Think of a Wave!”

11 Lührmann and Lindberg, “A Third Wave of Autocratization.”

12 Edgell et al., “Episodes of Regime Transformation Dataset.”

13 Maerz et al., “A Framework for Understanding.”

14 Boese et al., “How Democracies Prevail.”

15 Wilson et al., “Successful and Failed episodes.”

16 Boese and Eberhardt, “Democracy Doesn’t Always Happen Overnight.”

17 Tomini, “Don’t Think of a Wave!” 3–4; Lührmann and Lindberg, “A Third Wave of Autocratization.”

18 Tomini, “Don’t Think of a Wave!”; Skaaning, “Waves of Autocratization and Democratization.”

19 Lührmann and Lindberg, “A Third Wave of Autocratization.”

20 See Lührmann and Lindberg, “A Third Wave of Autocratization,” endnote 59.

21 Coppedge et al., “V-Dem Country-Year Dataset.”

22 Tomini, “Don’t Think of a Wave!” 3.

23 Teorell et al., “Measuring Polyarchy.”

24 Maerz et al., “ERT – an R package.”

25 Tomini, “Don’t Think of a Wave!” 4.

26 Boese et al., “How Democracies Prevail.”

27 Tomini, “Don’t Think of a Wave!” 4.

28 Lührmann and Lindberg, “A Third Wave of Autocratization.”

29 Skaaning, “Waves of Autocratization and Democratization,” 3.

30 Tomini, “Don’t Think of a Wave!” 4.

31 Skaaning, “Waves of Autocratization and Democratization,” 3.

32 This average displays the magnitude of change. It takes the absolute value of all changes and thus considers negative and positive changes jointly.

33 Edgell et al., Episodes of Regime Transformation Dataset (v2.2) Codebook, 20.

34 Ibid.

35 For a comprehensive overview and systematic evaluation of a range of different approaches, see Pelke and Croissant, “Conceptualizing And Measuring Autocratization Episodes.”

36 Tomini, “Don’t Think of a Wave!”; Skaaning, “Waves of Autocratization and Democratization.”

37 The Encyclopedia Britannica defines a wave as the “propagation of disturbances from place to place in a regular and organized way”. See https://www.britannica.com/science/wave-physics.

38 Tomini, “Don’t Think of a Wave!”; Skaaning, “Waves of Autocratization and Democratization.”

39 Huntington, The Third Wave.

40 Edgell et al., Episodes of Regime Transformation Dataset.

41 Spill-over effects to neighbouring countries in democratization processes are already well documented (see e.g. Brinks and Coppedge, “Diffusion is No Illusion”; Gleditsch and Ward, “Diffusion and the International Context”). Yet, we need to know more about both how and why they spread out across time and space. Most importantly, we need to analyse how autocratization and democratization join together as highly multidimensional processes of regime transformations.

42 This is in stark contrast to the end of history line of thought/Hegelian dialectic. These two should not, as DTW does, be confused.

43 Lührmann and Lindberg, “A Third Wave of Autocratization,” 1102.

44 Ibid., 1111 (footnote 77).

45 Huntington “The Third Wave,” 15.

46 Edgell et al., Episodes of Regime Transformation Dataset.

47 Tomini, “Don’t Think of a Wave!” 1.

48 Maerz et al., “State of the World 2019.”

49 Tomini, “Don’t Think of a Wave!”; Skaaning, “Waves of Autocratization and Democratization.”