It has become clear in recent years that history did not end with the Cold War. Instead of “the universalization of Western liberal democracy as the final outcome of human government”,¹ democracy is in crisis and the third wave of autocratization is here.² According to Freedom House, the majority of the worlds’ population live under “partly free” or “not free” conditions meaning that living in fully free democracies is the privilege of a minority.³ Moreover, the number of liberal democracies is in decline.⁴ Freedom House reports that democracy gained ground from the 1980s until roughly 2007. Between 2007 and 2019 however, the share of free and partly free countries declined and that of not free countries increased.⁵ V-Dem Institute’s annual democracy report has similar findings. The decline in the number of liberal democracies is significantly more dramatic when the population size is acknowledged, as autocratization processes have recently taken place in large countries such as India and the United States.⁶

Using V-Dem’s dataset, Anna Lührmann and Steffan Lindberg write that the third wave of autocratization happens mostly in democracies at the same time “democracy is close to its all-time high”. For Lührmann and Lindberg, the trend is consequently “less dramatic than some claim”.⁷ Stating that autocratization happens mostly in democracies is problematic when the dataset used does not properly operationalize developments in already authoritarian countries. Studying democratic transitions, breakdowns or decline in the quality of governance in democracies fails to provide us with the big picture, when variance in the quality of governance in authoritarian regimes is systematically left out or analysed based on concepts and measures developed for understanding politics in democracies. Autocratization processes in authoritarian countries should be conceptualized based on what actually happens in these regimes, meaning for example that developments on the elite level need to be accounted for. There is important new research analysing power concentration in authoritarian regimes, highlighting the need to better integrate internal dynamics of authoritarian countries into studies looking at trends in democracies, which are usually the ones making general claims on global trends. Barbara Geddes, Joseph Wright and Erica Frantz show, that as part of deepening autocratization, personalism in authoritarian regimes has increased significantly since the end of the Cold War.⁸ These trends should not be understated, as higher power concentration is associated with both increased war-proneness and domestic repression.⁹ In short, deepening autocratization in already authoritarian countries is an actual phenomenon affecting hundreds of millions of people worldwide directly, and the rest of the world indirectly. It should be included in studies on autocratization and quality of governance.

This article promotes an approach that enables concept definition and subsequent operationalization on a level applicable to cross-national analysis, while acknowledging the features of both poles of pared concepts such as democracy and authoritarianism. In the following sections, I argue first that the definition of autocratization should include deepening autocratization in already authoritarian countries, rather than focusing solely on regime change. In defining deepening autocratization, my contribution concentrates on the elite level of authoritarian rule. I build on authoritarian power concentration measures developed by Geddes et al. and broaden it outside the military-security sphere by including measures of administrative centralization and state control over economic assets. Exploring links between economic structures and regime type together with the repressive mechanisms has great potential for revealing paths to authoritarian resilience.

Second, the fact that deepening autocratization has been left out of most existing conceptualizations and international datasets highlights the problems of conceptualization and causality in social sciences discussed in this journal. Authoritarian regimes are too often defined using definition by negation as the analytical strategy, leading to omission of many important variables when definitions are taken to the operational level. Moreover, ignoring causal elements in concept construction tends to lead to problems at the application stage.¹⁰ “Authoritarianism” and “democracy” are multidimensional concepts, which are seldom useful when analysed as single variables rather than on the level of their dimensions.¹¹ When looking at how deepening autocratization is linked with authoritarian resilience, empirical evidence shows that there is no “one model fits all” explanation for authoritarian resilience as, according to new research by Geddes et al., this depends on authoritarian regime type.¹² In other words, causal consequences of shock factors may depend on regime types making the overall category of “authoritarian regime” poorly applicable to empirical analysis of authoritarian resilience.

Finally, taking into account the multicausal paths to end results in social sciences would greatly improve the research agendas on authoritarian and democratic regimes. There are many ways to reach political equilibrium in authoritarian regimes, and the process is never ready. Somewhat paradoxically, in real-world circumstances, regime stability is about movement. It is precisely the adjustments dictatorships make all the time that keep these regimes in power. To paraphrase Daron Acemoglu and James Robinson, authoritarian governance takes place in a narrow corridor. Staying in that narrow corridor requires a right balance of control: There has to be enough control to avoid revolutions and coup attempts, but not too much control, which might prevent economic activities. Balance maintenance entails a red queen effect, meaning pressure to adapt faster to changing conditions.¹³

Defining autocratization

As the first step in defining autocratization, it is useful to review the conceptual relationship between democracy and authoritarianism to explain why the existing definitions of autocratization are insufficient or misleading.

Problems with operationalization and datasets

Research analysing regime types has moved on from assuming democratic teleology, but earlier concentration on democracies has path-dependent consequences for theories, concepts and their operationalization. Particularly worrying is the practice of defining key concepts used in empirical research on the universal level using definition by negation as the analytical strategy. Here “authoritarian rule” really suffers from an unequal position with regard to “democracy”. Many comparative projects rely on large datasets covering both democracies and non-democracies. Such datasets use variables featuring aspects of democracy but lack measurements for developments present only in authoritarian countries.

V-Dem Institute’s Liberal Democracy Index takes liberal democracy as the starting point, as its name suggests. China is the largest authoritarian country, accounting for almost 20% of the global population. As Figure 1 illustrates, this index depicts some small-scale variance in China’s case, although most of the important changes in an authoritarian regime cannot be adequately illustrated with such a scale. The other countries shown in Figure 1 are there to contextualize China’s variance.

Dynamic dictators: improving the research agenda on autocratization and authoritarian resilience

All authors

Elina Sinkkonen

https://doi.org/10.1080/13510347.2021.1903881

Published online:

06 April 2021

Figure 1. Quality of democracy in example countries (Source: Coppedge et al., “V-Dem Dataset v11”).

Display full size

Figure 1. Quality of democracy in example countries (Source: Coppedge et al., “V-Dem Dataset v11”).

Furthermore, when using datasets, the variables of which have been built on characteristics of liberal democracies, one should be careful about making conclusive judgements on authoritarian regimes. Some datasets may be able to account quite well for fluctuations of social control in authoritarian countries, for example freedom of media and the internet. However, power concentration within the elite is left out as it is not part of the definition of democracy. Furthermore, when autocratization in authoritarian countries is only partly operationalized in cross-national datasets, it can be problematic to apply them to assessing global trends.

Authoritarian resilience, multidimensionality and multicausality

After establishing deepening autocratization as a phenomenon, it is possible to analyse how it is related to authoritarian resilience. Goertz argues that the underlying structure of concepts has causal consequences, and one cannot “neatly separate the ontology of a concept from the role it plays in causal theories and explanations”. As an example, he uses the basic hypothesis that democracies do not fight wars, which involves attributes of democracy such as the possibility to change leaders in elections.³⁵ Goertz’s ontological and realist approach to concepts differs from Sartori’s famous semantic and definitional approach, which does not emphasize the concept-causality connection.³⁶ Sartori also favoured categorical over continuous variables, and while this issue continues to cause controversy, many comparativists have come to the conclusion that it makes sense to treat certain concepts such as democracy or authoritarianism as continuous rather than dichotomous.³⁷

New research results on the relationship between deepening autocratization and authoritarian resilience uncover causal links between the two phenomena. Moreover, they emphasize the need to acknowledge multidimensionality of concepts such as democracy and authoritarianism, and multicausality. The failure to acknowledge multidimensionality can be very costly, as Erik Fritzsche and A. Vogler showcase with the contradictory findings of modernization theory. They argue that “multidimensional concepts of ‘democracy’ should not be implemented as single dependent variables”. Despite decades of research, modernization theory has not managed to produce a clear picture of potential relations between economic performance and democracy, because most studies have treated democracy as a single variable.³⁸ Similarly to democracy and its causes, authoritarian resilience is a complex phenomenon which does not surrender to models based on one main explanatory factor. The nature of scientific inquiry should shift from looking at independent variables and their average effects on the final outcome, into studying the combination of interrelated variables.

Previous research has built its understanding of authoritarian resilience on multiple factors, which vary in terms of measurability. Svolik’s analysis concentrates on two dimensions of authoritarian rule, authoritarian control and power sharing. In other words, authoritarian resilience depends on controlling society and maintaining a necessary level of elite coherence, and these end states are reached by using different instruments. Svolik further divides authoritarian control into two more specific tools, co-optation and repression.³⁹ Johannes Gershewski adds to this a third element, authoritarian legitimacy, which he admits being difficult to measure.⁴⁰ Even authoritarian systems try to build regime legitimacy as pure coercion would become too expensive in the long run. For Gershewski, authoritarian legitimacy is not an oxymoron, rather it means that non-democratic regimes seek to have active consent, rule obedience or sometimes mere toleration for their rule within the population.⁴¹ Legitimacy can be operationalized through specific and diffuse support. In short, specific support depends on fulfilling popular demands related to socio-economic development and physical security, whereas diffuse support tends to be linked with ideational factors such as political ideologies that help to build long-term support by increasing in-group coherence. Legitimacy strengthens shock resistance, and all regimes aim at forming a widespread belief among the citizenry that the domestic regime is normatively and structurally superior to its alternatives.⁴² Julia Grauvogel and Christian von Soest have developed a dimensional measurement of legitimation strategy, and their empirical analysis confirms that strong claims to legitimacy increase authoritarian persistence.⁴³

Geddes et al. provide so far the most ambitious explanation of authoritarian resilience, which takes interaction effects seriously.⁴⁴ Factors affecting authoritarian resilience are likely to involve interacting causal variables that are not independent of each other. Understanding authoritarian resilience requires a change from an approach which tries to estimate average causal effects of one or more independent variables on democracy/authoritarianism – an approach criticized above by Fritzsche and Vogler.⁴⁵ Instead, we should embrace equifinality, the idea that specific outcomes or end states can be caused by different contributing factors or by different combinations of factors. This means that research should focus on the combination of causes and their interactions rather than effects of individual causes.⁴⁶ By studying the effects of economic downturns in combination with regime types, Geddes et al. conclude that economic downturns increase the likelihood of collapse in dictatorships that lack extensive party networks. Put the other way around, extensive party networks stabilize authoritarian regimes even when the regime is faced with economic crises.⁴⁷ Furthermore, personalization in party-based regimes decreases regime durability, while it increases regime durability in military-based regimes.⁴⁸ In sum, it is not enough to say that economic crises destabilize dictatorships, as the effect of the crisis depends on the type of regime, the situation of the economy prior to the crisis and the level of personalism in the authoritarian regime in question. All this highlights that it is seldom useful to talk about authoritarianism on the universal level when analysing causality, rather one must step down to the middle level on Sartori’s ladders of abstraction and analyse causal relations separately with different types of authoritarian regimes.

Measuring deepening autocratization

When authoritarian leaders reach their power position, they must take active measures to stay there. Understanding this dynamism will help us reach a more comprehensive account of global trends in governance. Currently, dynamic changes within regimes such as fluctuations in the level of autocratization remain poorly conceptualized and measured.

As described above, authoritarian resilience depends on maintaining a necessary level of elite cohesion and domestic control, which can be reached by using tools of co-optation, repression and legitimacy-building strategies. How authoritarian tools can be used and what their outcome will be depend on the regime type. Party-based regimes tend to be more resilient when faced with different types of crises whereas the effect of power concentration on regime durability depends on regime type. These findings are very useful, but there is still the need for better understanding of variation in key sectors of authoritarian rule and how such variance relates to authoritarian resilience in different contexts.

Regarding measures of domestic control, repression is already quite well measured in existing research. Following Christian Davenport, repression can be defined as the “actual or threatened use of physical sanctions against an individual or organization, within the territorial jurisdiction of the state, for the purpose of imposing cost on the target as well as deterring specific activities”.⁴⁹ Repression can be understood as high or low intensity repression, as many datasets distinguish civil liberties and political rights.⁵⁰ For example, the Varieties of Democracy project includes multiple indicators for both civil liberties and political rights in its dataset.⁵¹ To ensure domestic control, the state has to allocate necessary resources to its repressive machinery. In this respect, it may be useful to study budgets of national police, paramilitary forces or similar coercive institutions which the leadership can task to supress opposition. Sheena Greitens argues that authoritarian leaders must simultaneously deal with two internal threats: threats from the population and threats from elites, who can be part of the coercive apparatus itself. While a unitary and inclusive coercive apparatus is effective vis-à-vis mass unrest, it also forms a coup risk for the leadership. Consequently, dictators sometimes opt for exclusive coercive mechanisms such as secret police, which is less effective for responding to threats from the population but less risky for the leadership.⁵²

When it comes to power concentration on the elite level, there is more work to be done in order to integrate authoritarian regimes into research analysing governance trends globally. Fortunately, there is promising new research looking at aspects of governance missing from democracies and big democracy focused datasets. The problem is that so far scholars studying variation of power concentration in authoritarian regimes have created their own datasets on authoritarian conditions, but international datasets including democracies do not communicate enough with this authoritarianism focused branch of research. Svolik argues that contrary to popular beliefs, maintaining elite cohesion in authoritarian countries is much more important for regime survival than societal control.⁵³ While more recent research states that this characteristic may be changing, as popular protests have become more significant a challenge to authoritarian regimes,⁵⁴ elite cohesion remains important. Moreover, there is likely to be variance on the balance between elite versus public management with regard to authoritarian survival, depending on the country context.

Measuring power concentration

To measure power concentration, we should decide which aspects are relevant, measurable and possible to code in a way amenable to comparative research. At least, personalization, administrative centralization and state control over key economic assets fill these criteria. With these three, there are certainly measurement difficulties, but they should not be insurmountable. In fact, there is already a measurement for personalization as Geddes et al. have filled in an important gap in the study of authoritarian regimes by developing one.

In this section, I build on work by Geddes et al. on the military-security sphere and broaden it to economic issues. Economic factors tend to be discussed in the context of how they relate to democratic transitions. There is less systematic analysis on how economic factors help to stabilize authoritarian regimes which lack natural resources.

As was discussed briefly above, modernization theory has studied economic performance and democracies, but due to measurement-level issues, failed to form a clear picture of causal relations.⁵⁵ Dimensions of democracy and their relations to components of economic performance would likely produce more useful findings. Acemoglu and Robinson have been able to find robust results, but they also look at the economy from a democratization frame in linking distribution of wealth and regime type.⁵⁶ Geddes et al. concentrate on authoritarian regimes, but they are interested in how economic shocks affect different types of authoritarian regimes, not how economic factors can help to stabilize non-democratic regimes.⁵⁷ This broader context influenced the proposal to study administrative centralization and state control over economic assets. They can both be studied in democratic and authoritarian regimes helping to form a broad picture. Administrative centralization and state control over economic assets are components of economic performance, helping to avoid the pitfalls of treating multidimensional concepts as single variables.⁵⁸

Figure 2 summarizes measures authoritarian leaders can take with regard to elites and the wider society. It is notable that there can be simultaneously policies increasing and decreasing autocratization, meaning that not all aspects that can vary will necessarily move in the same direction at the same time. On the elite level, authoritarian leadership can increase or decrease power concentration, which can be measured with personalization, administrative centralization and state control over economic assets. On the societal level, authoritarian leadership can increase or decrease repression. Leaders can also increase public support by building legitimacy though measures enhancing in-group coherence and applying welfare policies. As mentioned above, some scholars of authoritarian regimes use co-optation as one tool to maintain authoritarian stability.⁵⁹ In what is discussed below, parts of the often-used co-optation measures are included in the personalization index (security apparatus), administrative centralization and state control over economic assets (capital, labour and land ownership). As the role of the economy in regime maintenance is broader than buying off powerful groups, it is not enough to talk about co-optation.

Dynamic dictators: improving the research agenda on autocratization and authoritarian resilience

All authors

Elina Sinkkonen

https://doi.org/10.1080/13510347.2021.1903881

Published online:

06 April 2021

Figure 2. Dimensions of authoritarian rule and autocratization.

Display full size

Figure 2. Dimensions of authoritarian rule and autocratization.

It is out of the scope of this article to finalize a coding scheme or to test these ideas with a dataset. My purpose here is to evoke discussion on factors that should be counted as aspects of power concentration and how to best measure them. The three factors listed below arise from existing research on comparative authoritarianism and from characteristics of authoritarianism in China, the world’s largest authoritarian country. In other words, the list is not intended to be final.

Personalization

A dictator wants the minimum sufficient support to survive, but no more, as he has to compensate support either by sharing power or giving out resources.⁶⁰ Thus, he has incentives to concentrate power in his own hands as much as possible. While personalization can sometimes refer to the building of a personality cult of some sort, here personalization signifies measurable features of power concentration. Geddes et al.’s personalization index includes “dictator’s personal control of the security apparatus, creation of loyalist paramilitary forces, dictator’s control of the composition of the party executive committee, the party executive committee behaving as a rubber stamp, dictator’s personal control of appointments, dictator’s creation of a new party to support the regime, dictator’s control of military promotions, and dictator’s purges of officers”.⁶¹ In their research, Geddes et al. calculate a yearly personalism score for each regime making it possible to show variation during a single dictator’s tenure. Before this, “personalization has never been so carefully and consistently measured across regimes and across time”.⁶²

Research on authoritarian regimes has also looked at the level of personalization and its relation to regime endurance, level of domestic repression and foreign policy. Geddes et al. report that dictators who score high on the personalism scale during their first three years in office are much more likely to stay in power than dictators with low early personalism scores.⁶³ According to Bruce Bueno de Mesquita et al., intensity of oppression is linked with the type of the winning coalition: the bigger the difference between leaders’ payoffs and the payoffs of the winning coalition, the more widespread and intense oppression measures within the elite tend to be.⁶⁴ Frantz et al. find that personalization of power in dictatorships increases observed repression. They also point out that at the end of the Cold War, 23% of all authoritarian countries were ruled by personalist leaders, but by 2020 the share of personalist regimes had increased to 40%.⁶⁵ Furthermore, Jessica Weeks has shown that dictators with higher levels of power concentration tend to start more wars than more constrained authoritarian rulers.⁶⁶

It is unfortunate that Geddes et al.’s data coverage ends in 2010, as in many countries authoritarian tendencies have intensified since then. In China, Xi Jinping’s predecessor Hu Jintao experimented cautiously with a more consultative approach by promoting intra-party democracy and expanding the number of official actors influencing leadership decisions on foreign policy, which as a policy area has been severely restricted.⁶⁷ Xi started by creating new political bodies – such as the National Security Commission (NSC) in January 2014 – to supervise both foreign and domestic security issues. Xi’s aim in establishing the NSC was to concentrate power in his own hands and to improve policy co-ordination.⁶⁸ According to Xinhua, 13 March 2018, China rearranged its ministries and diminished the total number of ministerial-level bodies by eight and vice-ministerial-level bodies by seven. As the most telling example of increased personalism, the National People’s Congress abolished term limits for the president in March 2018, which indicates that Xi might be planning to stay in power after 2023.⁶⁹

Administrative centralization

While personalism is certainly a key aspect of power concentration in authoritarian regimes, it is not the only one. There is also variation regarding power sharing between regional administration and the central government. Steve Hess argues that “modern authoritarian cases might be disaggregated into centralized and decentralized types”.⁷⁰ Hess defines decentralization as the transfer of state authority and resources from national governments to the subnational level.⁷¹ For a numeric measurement for decentralization, it is possible to use the International Monetary Fund’s cross-national fiscal decentralization database, from which one can track the percentage of revenues controlled by subnational governments.⁷²

Authoritarian regimes are generally reluctant to decentralize. However, there are a few notable exceptions such as China and Kazakhstan.⁷³ In addition to economic benefits, the logic behind decentralization is often that by granting local officials greater authority within their jurisdiction, the centre avoids blame for local authorities’ official misdeeds and their use of repression. Relative to their more centralized counterparts, decentralized regimes tend to exhibit higher levels of durability in the face of popular challenges.⁷⁴ It is part of the Communist Party of China’s (CCP) deliberate strategy to blame local governments for general problems and enjoy support at the top. According to public opinion surveys, support and trust in the leadership is indeed polarized: the Chinese tend to support the central government more than local leadership.⁷⁵

China has decentralized a great deal in the past decades, as – thanks to fiscal decentralization – subnational expenditures jumped from 45% in 1981 to 85% in 2013. This is exceptional among non-democracies where government expenditures at the subnational level tend to be below 18%.⁷⁶ Yet, centralization and decentralization go in tides. Decentralization has had clear advantages as it has contributed to economic performance since the early 1980s.⁷⁷ During Xi’s regime from 2012 onwards, China has recentralized its administration and the anti-corruption campaign has constrained local-level initiatives.⁷⁸ The centralization tendency has also been strengthened by a provincial debt crisis, to which central authorities have responded by initiating reforms which considerably limit financial resources available for local governments.⁷⁹

State control over economic assets

In many authoritarian regimes it is not enough for the leaders to control the security apparatus, which is included in Geddes et al.’s measurement of personalization. Often economic elites need to be controlled as well. Authoritarian leaderships have to balance between supporting economic growth in their country and avoiding opposition forces, which can form around powerful economic actors. In China this has been done by reserving key sectors for state-owned companies and trying out new policies gradually and first in special economic zones. Through its nomenklatura system the CCP can control all key nominations, including company CEOs and university rectors, making sure that only loyal people can get top positions.⁸⁰ There is a revolving door mechanism for cadres, circulating them between Party positions and state-owned business positions, both in order to enhance their experience and to restrain their personal power by making sure that nobody can stay too long in the same entity.⁸¹ Furthermore, the party is integrated in business life as 53% of private and 91% of state-owned enterprises have CCP organizations.⁸²

In China’s case, it would not be impossible to collect data on CCP involvement in economic structures, as there is data on state-owned enterprises, their share of GDP and so on. However, making such measures comparable with those of other authoritarian countries would be more difficult. For example, Vladimir Putin’s Russia relies on companies owned by oligarchs, who remain politically loyal to Putin. While the companies are not state-owned, they can be described as being under state control. Without Chinese-style institutionalized control mechanisms for economic actors, managing the economic elite can sometimes lead to drastic measures. For example, some argue that Putin’s decision to annex Crimea was partly a coup d’état against domestic elites, because after the event powerful Russian elites holding mostly liquid assets lost their wealth and only those with significant fixed assets were able to maintain their economic positions.⁸³

Furthermore, power concentration in the economic realm takes novel forms thanks to digitalization. In China, the information technology-driven social credit system (SCS) is probably the most significant new addition to the authoritarian toolkit in ensuring party control. By collecting data from different sources, the social credit system can monitor, asses and change the behaviour of both citizens and companies.⁸⁴ At the moment, there are multiple co-existing SCSs for different purposes at different administrative levels rather than one coherent system. Individuals and companies have different systems. In addition, some systems concentrate on creating infrastructure for economic and financial activities whereas others are linked more with social governance. The People’s Bank of China is in charge of creating a financial social credit system whereas Beihang Credit is the only commercial company dealing with commercial credit rating services. In the social governance realm, there are both national and municipal blacklist/redlist systems for both individuals and companies.⁸⁵ The corporate social credit system is more developed than the system designed for individuals, although the sanctioning mechanism remains somewhat fragmented when it comes to co-operation between local and central-level actors. Similar to individuals, companies are rated, and blacklisted entities can be found on the credit system’s webpages. Companies that get poor ratings can get penalties and restrictions to market access. The system was supposed to be ready in 2020, but as there are problems of data-sharing between different parts of the system and sanctioning mechanisms remain underdeveloped, the deadline of 2020 was not met.⁸⁶

Concluding remarks: red queens rule everywhere

This article has argued that rather than only comparing authoritarian regimes with democracies, politics of authoritarian rule should be taken as the starting point of analysis on authoritarian resilience to better understand the building blocks of non-democratic regime survival. This means that we should include measurements for characteristics that are specific to authoritarian regimes in our models on autocratization. These include dimensions of power concentration such as personalization, administrative centralization and state control over economic assets. Understanding deepening autocratization would also help us to form a complete understanding of global trends in governance.

While we should think more about how to improve conceptualization and measurement of key concepts in comparative politics, we should simultaneously think about the future. Technological change and the nature of contemporary capitalism add layers of complexity for building a better understanding of governance in different regimes. Digitalization takes various forms in different types of polities, but there are also common features affecting all countries. A framework that neatly divides countries into authoritarian and democratic types and builds its understanding of the world on these typologies is not likely to meet the demands of the era of big data and technological revolution.

The Arab Spring in 2010 raised high hopes for the roles the internet, mobile phones and social media could play in advancing democratization.⁸⁷ Subsequently, Wikileaks’ revelations about the widespread use of bulk surveillance in Western countries in 2013, the election meddling via microtargeting of voters on Facebook during Trump’s 2016 presidential campaign and the Brexit campaign in 2016 have increased awareness of the negative aspects of social media and citizens’ limited ability to protect their privacy. Moreover, China has never allowed free internet on its territory and uses sophisticated internet surveillance methods.⁸⁸ In China’s case, we do not have evidence of microtargeting in attitude formation, but against the background of relatively influential traditional censorship and propaganda there is certainly potential for it. Studies conducted in China have shown that the media environment influences educated respondents’ views even if they recognize that media is biased in its reporting.⁸⁹

The Chinese government has taken social unrest issues seriously and invested a lot in digital forms of control such as CCTV cameras with automated facial recognition programmes. In Xinjiang, monitoring includes obligatory DNA sampling used for ethnic profiling.⁹⁰ Xi’s regime has built capacity to forecast large-scale popular protests and adapted its political indoctrination to the era of big data by using artificial intelligence (AI) in surveillance and censorship.⁹¹ China also builds high-tech co-operation with Russia with an emphasis on big data and AI, indicating that innovations in either country are likely to be adopted in other authoritarian contexts as well.⁹² These developments imply that there are new forms of repression not included in international datasets.

Authoritarian policy priorities may also influence global governance. China advocates cyber sovereignty, meaning that every country should have control over the internet within its physical borders. To advance an approach highlighting the sovereignty principle in internet governance, China has co-operated with Russia and other members of the Shanghai Cooperation Organization (SCO) and drafted an International Code of Conduct for Information Security, which was originally released in 2011 and updated in 2015.⁹³ In sum, while technology itself can be considered neutral, the hopeful visions from 2010 have since turned more sinister for democracies and authoritarian countries alike.

Something that should also be assessed further is interaction between democratic and authoritarian regimes. Increased political polarization and the rise of populism in Western democracies make democracy seem like a less appealing option for people living in authoritarian countries. Labour markets in the advanced economies have gradually polarized since the 1950s, a process which was accelerated in the mid 1980s due to the spread of information and communication technology. In the United States and Europe, middle-wage workers have been losing both in terms of employment and wage growth in comparison to groups of workers with low or high wages.⁹⁴ Sizes of middle classes are intimately linked with democratic consolidation according to prevalent theories, indicating that if economic structures change drastically, previously stable systems can become unstable. Worsening employment prospects have led to increased dissatisfaction in advanced economies which in turn has fuelled populism. Low income, a subjective sense of economic insecurity and being a member of an older age cohort have been found to be significantly associated with populist and authoritarian attitudes in Western democracies.⁹⁵

Political turmoil in the West may increase authoritarian legitimacy – a pair of terms often seen as an oxymoron. If the crisis of democracy continues, it may enhance the public’s belief in authoritarian countries that their regime functions better, is lawful and should be obeyed. In sum, resilience of authoritarian regimes depends not only on domestic politics but also on the international context. In future studies it would be important to integrate the global systemic level, diffusion of power and qualitative changes happening in the authoritarian toolkit due to technological change in order to understand what authoritarianism is and what makes it resilient.