Preconditions for pacted transitions from authoritarian rule

Abstract

This paper investigates why some attempts at pacted transitions from non-democratic rule fail, while others succeed. It determines the composition of opposition organizations that enable pacting. The paper draws on a data set compiled by the author comparing forty-five attempts at negotiations. The qualitative comparative analysis shows that those negotiations that include the opposition with strong organizational capacity succeed and end up with democratization. This strong organizational power of the opposition can be drawn from trade unions or the Catholic Church participating in negotiations, even if the initial regime is personalistic.

Introduction

Successful pacted transitions – those led by negotiations between the ruling elites and the opposition – were once argued to be the most effective way toward democracy (O’Donnell et al., 1986; Karl, 1990; Huntington, 1993; Linz and Stepan, 1996; Colomer, 2000; Stradiotto and Guo, 2014). However, the concept of pacted transition comes from the early wave of studies of regime change – ‘transitology’ – that shared a considerable amount of optimism. Especially in the late nineties, they assumed that every serious attempt at building democratic institutions is very likely to end up with democratization. Therefore, while successful pacted transitions attracted the attention of scholars, there has been far less analytical attention paid to the numerous cases where attempts at pacts failed, or democratization did not follow. For instance, attempts to negotiate pacts broke down in some cases and finished in violence, while some of the seemingly successful pacted transitions brought about only a turnover of leadership but not democratization.

Why do attempts at pacts fail? Relatedly, what are the preconditions for a successful pacted transition? Whereas there is abundant literature on which types of non-democracies are likely to end up with pacted transition (Linz and Stepan, 1996; Geddes et al., 2014, 2018), little has been written about which types of opposition organizations are more likely to facilitate successful pacting. This paper analyzes the composition and characteristics of the organizations representing the opposition during pacted transitions in order to uncover what determines the success or the failure of negotiations and subsequent democratization.

The main argument of this paper is that a successful pacted transition either occurs as a result of trade union involvement in the negotiations or in a regime that permits some degree of pluralism and has established opposition parties and the independent Catholic Church (mostly a party dictatorship). In other words, only those opposition movements that have high organizational capacity – that is, a capacity for credible commitment and credible threat (which can be achieved in several ways) – can negotiate pacted transitions.

Different organizations facilitate negotiations in their own ways. Trade unions can survive in those authoritarian contexts where opposition parties are already banned. Having an established practice to bargain with the state in the economic realm, they are also good at organizing dialogue on political matters. At the same time, strikes are more dangerous to the regime than protests because of their ability to harm the country’s economy instead of just occupying the streets. The Catholic Church can also survive in almost any political setting and, leveraging its symbolic authority, can play an important mediator role when the political crisis becomes too violent. Finally, established political parties are beneficial to pacted transitions when they come from a relatively pluralist party or military dictatorships. However, even if they are supported neither by the Church nor a trade union, these political parties alone should have enough organizational capacity to persuade the leaving leader that he will not be prosecuted in the future, which is only possible in highly liberalized party autocracies.

Knowing the exact preconditions for pacted transitions helps bridge the gap between the structure-based and agency-based theories of democratization. Pacted transitions are more likely to lead to democratic outcomes than any other mode of regime change (see, among many others, Stradiotto and Guo, 2014; Geddes et al., 2018). At the same time, the preconditions for pacts seem not only potentially demanding but also favorable for democratization, even though no study formulates and tests those preconditions. ¹ If the preconditions for pacts turn out to be demanding, it is arguably not the interactions of actors (i.e., pacts per se) that foster democratization via pacted transitions. The success of pacts might merely be a consequence of the presence of strong opposition organizations. If the preconditions for pacts are not demanding and successful pacts are likely to happen despite the weak organizational capacity of the opposition, then it is the agency that makes pacts so effective.

This research is based on a mid-N QCA analysis of historical attempts to negotiate pacted transitions backed with conventional statistical analysis. In doing so, it draws on a unique database of forty-five successful and failed attempts at pacts, covering all countries in which an attempt at a pacted transition was made since 1974. Cases are selected based on the presence of attempts to negotiate a solution to a political crisis, which generally implies the terms of exit for the incumbent elites.

This paper is organized as follows. The next section presents a brief literature review that shows the debate between structure and agency in the studies of regime change and poses a research question. The theory section discusses the types of organizations of opposition, their characteristics, and the types of the initial non-democratic regime that are expected to enable successful pacted transitions. The cases and research design section describes the data and how it was gathered. The analysis section presents a QCA analysis that tests my hypotheses. Then, I illustrate my argument with a vignette study. The conclusion relates the finding to the existing theories of democratization.

State of the art

Why are some countries have democracies while others are not? Depending on where the answer for this question is sought, the democratization scholarship remains divided between the structuralist and agency-centric traditions. For the former, political regimes are mostly predetermined by economic development, whereas for the latter, actors’ actions play a central role in the choice over a country’s political regime even when the preconditions are absent. The general structuralist argument is that the modernization process – the spread of urbanization, literacy, mass education, and the emergence of mass media, the change in class structure, and so on – increases the likelihood states will democratize (Lipset, 1959; Przeworski, 1999; Vanhanen, 2003; Inglehart, 2018). According to this logic, poor, undeveloped countries are prone to be autocratic, while their wealthy, developed counterparts are very likely to be ruled in a polyarchic way. In contrast to the modernization approach, the ‘transitologists’ argue that at least part of the ‘regime decision’ also depends on the interactions and choices of actors – for instance, over the mode of regime change.

The actor-centric theories of regime change share one pivotal concept – the pacted transition. As opposed to bottom-up revolutions or transitions imposed by the elites, a pact is considered to be the most plausible (albeit not the only) path to democratization. In different theories, the concept of pacted transition has taken different names: cooperative transition, pact, transplacement, transition by agreement, and the like.

The classic typology is offered by O’Donnell et al. (1986) and specifies four modes of regime change. Transitions are divided using two scales – the driver (regime elites or the opposition) and the degree of cooperation between regime elites and the opposition (cooperation or confrontation). This results in different modes of regime change: ‘(1) reform; (2) revolution; (3) pacted; and (4) imposed’ (Schmitter, 2014, p. 7, citing O’Donnell et al., 1986). For O’Donnell and colleagues, reform occurs when a protest movement demands liberalization, eventually forcing the regime to democratize, while revolutions refer to the physical displacement of elites. Moreover, imposed transitions are initiated and undertaken by the government alone without the participation of the masses or opposition groups. Finally, pacted transitions are driven by negotiations between elites and the opposition.

Societies that undergo pacted transitions from authoritarian rule are known to have excellent chances of democratization. The explanations behind this phenomenon-vary. Early ‘transitologists’ (O’Donnell et al., 1986; Huntington, 1993; Casper and Taylor, 1996) see the reason for the success of such transitions in the path dependency they create: they level the playing field, exclude radicals, and help forestall violence. Contemporary scholars of regime change connect the mode of regime change with the type of authoritarian regime in question, and pacted transitions happen in those autocratic regimes that are anyway likely to be followed by democracy (Linz and Stepan, 1996; Geddes et al., 2018).

As I will show in the text, pacted transitions require both the high level of pluralism of the elites in the old regime and the high organizational capacity of opposition. Moreover, this high organizational capacity of opposition might be beneficial for democratization regardless of the mode of regime change. Thus, what remains uncertain here is the origin of these democratic outcomes – is it agency per se or the high organizational capacity of the opposition that enables these pacted transitions? To approach this question, one needs, first, a more precise understanding and operationalization of the organizational capacity of opposition, and, second, an empirically based knowledge to what extent the opposition should be organized to enable the negotiated transition.

While this bunch of regime change scholarship generally emphasizes the role of agency over structural conditions in transition processes, some authors still stress the presence of preconditions for pacted transitions. These requirements can be divided into two groups – those that are based mostly on agency (i.e., previous events), the ‘liberalization’ argument, and those that present more ‘structural’ arguments about the incumbent regime’s type. The early wave of studies of regime change asserted that the opposition would need some degree of achieved development during the ‘liberalization’ phase before pacting could occur (O’Donnell et al., 1986; Casper and Taylor, 1996; Colomer, 2000). Nevertheless, this requirement never assumed a central place in such theories, being neither operationalized nor tested empirically. More recent scholarship on pacted transitions understands the existence of the opposition as a built-in feature of the regime. In a nutshell, leaders of party-based institutionalized regimes are more likely to negotiate their exit (Linz and Stepan, 1996; Geddes et al., 2018).

Categorization omits a part of reality, which in this case implies that a regime may, in fact, embody different party, personalistic, and even military dictatorship characteristics at the same time. Therefore, the latest studies in the field also unpack the ‘regime type’ category and delve into the characteristics of a regime. For example, in their aforementioned book, Geddes et al. (2018), take into account not the regime type but the degree of regime personalization, to describe a regime’s probability of undergoing a pacted transition, where less personalized regimes are more likely to negotiate a pact. My research takes another step in this direction looking at the other counterpart of the negotiations and unpacking what is understood by ‘opposition’ – in terms of its presence and organizational capacity – rather than relying on the ‘regime’ category.

Against the backdrop of this discussion, my research question is the following:

What sets and features of opposition organizations account for the success and failure of attempts at pacted transitions?

Theory: types of opposition organizations and successful pacts

By definition, negotiated transitions involve two counterparts among the political elite – those who rule and those who are in opposition (O’Donnell et al., 1986; Huntington, 1993; Casper and Taylor, 1996; Linz and Stepan, 1996; Przeworski, 1999; Colomer, 2000; Stradiotto and Guo, 2014; Geddes et al., 2014, 2018). Among those who represent the opposition, the presence of certain combinations of political parties, trade unions, and the Catholic Church is expected to be sufficient for the success of negotiations and the subsequent democratization. Among the ruling elites, the leaders of party and military dictatorships are expected to contribute to the success of the negotiations and the chances of subsequent democratization, even though it is not necessary that every pacted transition takes place in these regimes.

Political parties in authoritarian regimes, if they are permitted and can participate in parliament, are known to increase the chances of incumbent elites negotiating exit guarantees and to foster further democratization (Wright and Escribà-Folch, 2012, p. 40). However, since parties are present in virtually every pact, it is impossible to test how their absence would affect the negotiations.

Trade unions frequently enjoy a unique position in the authoritarian settings and have enormous capacity to negotiate with state elites. In many cases, unions are the opposition organizations with the densest organizational networks that also are the most capable of mobilization (Valenzuela, 1989, p. 447). A labor strike is a form of mobilization that is the most lethal to a regime because it can cause direct harm to an economy (ibid.) Also, it is harder to repress a strike than it is to repress a street protest. Trade unions can initiate the protests thus allowing the rest of the opposition to step in (Collier, 1999, p. 165). As well as this, the protests organized by trade unions are less prone to fade sporadically. Moreover, at the stage of negotiations, trade unions can help increase contestation by forcing the incumbent to allow left-wing parties (ibid.) to take part.

They also tend to endure for quite a long time, sometimes for many generations, and thus often have a long legacy of existence even under authoritarianism. In this sense, trade unions have experience in negotiating solutions to enduring conflicts with the state. During the regime’s collapse, this legacy ensures for the political elites that the trade union’s promises are credible. In addition, having existed for decades, trade unions have leaderships that are less personalized and more institutionalized. This depersonalization of the opposition movement is expected to work precisely as the depersonalization of the regime‘s leadership (see below). In addition, they tend to have a very subordinated, hierarchical structure that makes them capable of credible commitment.

Finally, the strength of trade unions is socially rooted. The ‘proletarianization’ class-based argument in the literature suggests that trade unions are especially active during certain stages of economic development when the pro-democratic classes (including labor) become stronger than the anti-democratic classes (Rueschemeyer et al., 1992; Collier and Mahoney, 1997; Collier, 1999, p. 10). Thus, the democratization is especially likely in these societies. To sum up, the participation of a strong trade union might be sufficient for democratization even if the rest of the opposition movement is weak.

At the same time, in many cases under the authoritarian rule, the trade unions can turn into top-down vehicles of state control over the labor ² or be newborn, and thus may be not institutionalized and hierarchical. However, in the cases under study, none of these trade unions participated in negotiations (see below).

Depersonalization and hierarchical organization are even more valid for the Catholic Church. During the transitions of the Third Wave, it was frequently the case that its representatives participated in round tables. I only mention the Catholic Church because in the cases under study, no leaders of other religious organizations participated in pacts – except for three borderline cases from the Islamic world ³ .

Traditionally avoiding association with democratic opposition movements, the Catholic Church’s position shifted during the Second Vatican Council in 1965, and it started to promote democracy in those parts of the world where Catholic congregations lived under authoritarian rule (Mantilla, 2010). Nowadays, the Catholic Church is understood as an important mediator that increases the chances that the negotiated outcomes are, in fact, implemented (Paffenholz et al., 2017). What makes it different from any other religious organization is Catholicism’s centralized and hierarchical character that ensures its credible commitment. Therefore, I expect that whenever the Catholic Church participates, the attempt at negotiations always succeeds.

Speaking of the ruling elites, the literature argues that the leaders of party and military regimes are more incentivized and likely to negotiate their exit (Geddes et al., 2018). First, the military or authoritarian legacy parties normally remain after the authoritarian collapse to protect corporate interests. Thus, the party members or military officers are shielded from a potential prosecution by their legacy party. The same is not the case for a personalist dictator. Moreover, for the same reason, party and military cadres are likely to preserve their jobs after the regime collapses. Second, bureaucratized regimes normally allow for more pluralism both within the elites and within the opposition. Thus, each of the three opposition actors will be more developed in negotiations under party and military regimes which under certain circumstances can play a crucial role. Therefore, deriving from a party or military regime is an INUS (an insufficient, but necessary part of an unnecessary but sufficient) condition for the success of negotiations.

Theory: features of the opposition and successful pacts

None of the indicators of organizational capacity offered by the literature on parties, social movements, and trade unions applies directly to this research because of the inherent diversity of non-democratic regimes. First, the literature on political parties normally operationalizes their capacity with votes earned during elections (Tavits, 2008; Hale, 2015). However, this assumes some degree of free and fair elections in the incumbent regime, which is obviously not the case in dictatorships. More than that, the extent to which some elements of elections are present in these regimes varies dramatically. Therefore, it is impossible to compare them with each other. The literature on social movements measures the organizational capacity through the ability to mobilize people and resources, with a trade-off between the two (Della Porta and Diani, 2011). Although the number of people on the streets can be measured, it is unclear in many cases whether they were mobilized by organizations or came on their own. For instance, in both Egypt (in 2011–13) and Ukraine (2013–14), most of the people were brought onto the streets by friends and relatives (Onuch, 2014; Warkotsch, 2014, p. 176). Moreover, there exists no data on the quantity of resources – such as money, offices, or employees – that would cover the whole sample of countries. Finally, the number of formal members does not reflect party strength because not all parties have mass membership.

Unlike with parties, the strength of trade unions can be measured via membership. Although the established trade unions also have the same problem with losing popularity and ability to mobilize when they become too institutionalized, as social movements do, each of the trade unions that participated in negotiations organized major strikes before the regime breakdown and therefore was not co-opted into the regime.

Social movements theories suggest that the protest movements might not have any organizational form when they emerge, but after the initial wave fades, with time, they either disappear or turn into organizations (Della Porta and Diani, 2011). Those organizations might have less support, but they normally have resources – cadres, offices, and money – instead. Therefore, this research considers parties’ organizational existence, which is approximated by the age of the oldest party that participated in the negotiations.

Since the existing theories mostly explain both occurrences of pacted transition and subsequent democratization with the incumbent regime type, it is also included in the analysis. This condition, first, controls for ruling elites will to negotiate because Geddes et al. (2018) expect ruling elites to be willing to negotiate in party regimes rather than in personalistic ones. Second, this will show whether moving from a party regime overlaps with high opposition organizational capacity.

To sum up, the opposition movement’s solid organizational capacity can be reached when it includes established parties participating in the pact, trade unions, and the Catholic Church. In line with the literature, party and military regimes are also expected to be important for successful negotiations. Therefore, in the next sections, I will test the following hypotheses.

Hypothesis 1: the presence of strong organizations is sufficient for the success of negotiations and subsequent democratization. Three types of organizations contribute in this sense: (a) Trade unions; (b) The Catholic Church; and (c) Established opposition parties.

Hypothesis 2: the type of the initial authoritarian regime is an INUS condition for one pathway to the successful transition.

Cases and research design

Outcome variables and unit of analysis

The unit of analysis is an attempt at pacted transition – namely, an episode where, amid a major political crisis, elites-in-power explicitly initiate negotiations with opposition forces, and at least one round of negotiations occurs. The universe of cases of interest includes all the attempts at pacted transitions since 1974 (Huntington, 1993). The outcome is the result of an attempt at pacted transition – whether it is a successful pact with subsequent democratization, a pact without democratization, or a failed attempt at negotiations. My understanding of a pacted transition, adopted from O’Donnell and colleagues, is a transition from non-democratic rule that follows: (1) negotiated agreements among; (2) elites-in-power and elites-in-opposition that are; (3) rule-setting and based on; and (4) mutual ‘vital interest’ guarantees. ⁴ If it leads to democratization, it is coded as a successful pacted transition. If it does not lead to democratization (i.e., the rule-setting component is weak or absent), the case is coded as a pact without democratization. If the negotiations collapse because either the incumbent or the opposition does not hold to its promises – in other words, significant guarantees on vital interests (part 4 in the definition) were not delivered either by elites-in-power or elites-in-opposition after the pact was signed – the case is coded as a failed attempt at negotiations (Figure 1).

Figure 1. Outcomes of attempts at pacts.

The level of democracy is operationalized by the Liberal Democracy index of V-Dem (Pemstein et al., 2017) that ranges from 0 to 1. In QCA analysis, it is treated as a fuzzy variable and calibrated accordingly. A .10 increase in this democracy index 3 years after the pact is concluded is coded as partial inclusion, whereas a .15 increase is understood as full inclusion (i.e., democratization). Every country in the data set that passed a .15 threshold has later completely democratized. Some of those with a .10 increase in the first 3 years may have later backslid such as Ukraine, which experienced backsliding in 2010 after democratization in 2004. However, these backslides resulted from an explicit event in which a leader perverted already established democratic practices and were not directly connected with the earlier pacted transition.

Conditions and calibration

Data have been collected on the organizational capacity of parties, unions, and the Catholic Church. Established parties are operationalized by their age based on the Political Parties of The World encyclopedia (Day and Degenhardt, 1980, 1984; Day et al., 1996; Szajkowski, 2005; Sagar, 2009). The fuzzy set approach is applied using the following thresholds: parties which are younger than 2 years are calibrated at full exclusion, since they are at their movement stage (Della Porta and Diani, 2011), were created as the regime started to open up, or its foundations began to fracture; parties older than 5 years are calibrated as partial inclusion, because they are expected to reach the organization stage; finally, parties which are almost as old as the spell of an authoritarian leader (10 years) are calibrated as fully included.

Membership and age of the trade unions that participated in pacted transitions are based on the Trade Unions of The World encyclopedia (Harper, 1987, 2005; Blackburn, 2016). The presence of the Catholic Church is coded based on the case material.

The regime type variable is based on the GWF data set (Geddes et al., 2014). Using several assumptions and pursuing QCA-friendliness, I dichotomized it into two categories – party/military and personalistic/hybrid. Since military and party dictatorships are known to affect the mode of regime change in the same way (ibid.), I have merged these categories. I collapsed the ‘monarchy’ type of regime with the ‘personalist’ category, although, of course, they have different types of legitimacy. Four regimes were coded as ‘democracy’ in the GWF data set – namely, Albania (in 1997), Ukraine (2004 and 2014), and Kenya (2008). I believe that those are competitive authoritarian regimes because although they had facade elections, none were regular, free, and fair (i.e., conducted on an even playing field). Competitive authoritarian, or ‘hybrid’, regimes tend to have super-presidential constitutions, which leads to the inherently high level of patronalism – that is, ‘the personalized’ (as opposed to impersonal, institutionalized, and organizational) ‘exchange of concrete rewards and punishments’ (Hale, 2015, p. 22) and low state capacity; therefore, those are lumped together with the personalistic category.

Data collection procedure

The cases where at least the negotiations were successful were collected in four steps. First, potential cases were identified on the basis of existing databases – Stradiotto and Guo (2014), GWF dataset (Geddes et al., 2014), Treisman (2017, 2020), and ‘What Makes or Breaks National Dialogues?’ report (Paffenholz et al., 2017). Second, each case was individually checked with the existing literature to see whether it fits the criteria of pacted transition used in this paper. An attempt at a pact should answer several requirements to be included: the subject of negotiations is democratization, not a civil war or independency; both elites-in-power and elites-in-opposition should be present in the negotiations; the negotiations are not commanded by a foreign power; they take place inside the country; and, finally, they have an explicit character to an extent that the counterparts are present in the same room.

The attempts at pacted transitions in my database vary to what extent the incumbent and the opposition control the situation, whether the incumbent and the opposition are represented in negotiations, what is discussed, how the foreign powers have intervened, and by the scale of violence and contention. The Supplementary Material contains a detailed discussion on the thresholds within which the variation was tolerated.

Data

Among the forty-five attempts at making a pact that I have collected, twenty-nine were successful and led to democratization and nine failed (meaning that the negotiations collapsed). In the remaining seven cases, while negotiations turned out to be successful, democratization did not follow (see Table A1). None of the countries with failed negotiation attempts experienced full democratization afterward. In the analysis of democratization, the cases of failed attempts at pacts are lumped together with and treated as those of non-democratization – therefore, I analyze sixteen cases of non-democratization.

Political parties, in my sample of forty-five negotiations, participate in every pact with only two exceptions. Trade unions are represented in sixteen negotiations, while the Catholic Church is present in thirteen cases (see Table A1).

Analysis and findings

Methods: QCA and regression analysis

This paper uses fuzzy-set qualitative comparative analysis (fsQCA) because the logic of my argument implies equifinality: there might be different constellations of organizational features that give the opposition a high organizational capacity and lead to similar outcomes. More than that, my hypotheses appeal to the necessity and sufficiency terms.

In my research, the QCA analysis reveals the sufficient and necessary conditions that determine the two outcomes: (1) further democratization and (2) failure of negotiations. The analysis shows that the presence of a trade union during the negotiations is sufficient for democratization, while the absence of a trade union in negotiations is necessary for both failed negotiations and non-democratization. In addition, party and military dictatorships with established opposition parties, or with the Catholic Church involved, also negotiate successful pacted transitions even without trade unions.

In the Online Appendix, the main finding of the QCA analysis is investigated further with t-tests and regression models. The t-test shows that the effect caused by trade unions is not a mere result of a higher level of economic development. Countries that have trade unions in negotiations do not have significantly higher levels of GDP per capita. The regression analysis reveals a very robust linear relationship between the share of the population with trade union membership and the increase in the level of democracy that follows in the 3 years after the transition.

All the analyses were performed using R (version 4.0.2) and RStudio (version 1.3.1056) software with the QCA, SetMethods, and venn packages for the QCA analysis (Dusa, 2019, 2020; Oana et al., 2020). For the statistical analysis, the arm, aod, and car packages were used (Fox and Weisberg, 2019; Gelman and Su, 2020).

QCA analysis

The QCA analysis is organized as follows. I first present the analysis of necessity for failed negotiation attempts or successful negotiations (independently of whether they lead to democratization or not). Second, I analyze the necessity for an occurrence of democratization after successful pacts – that is, cases where negotiations went successfully. Third, I present the analysis of sufficiency for the joint variable that merges the (non-)occurrence of democratization and the (non-)failure of negotiations.

Analysis of necessity: the failure of negotiations

Necessity implies that whenever the outcome is present, the condition is also present. Necessity (or at least necessity that is not tautological) occurs very rarely in reality. Therefore, most of the conditions illustrated in the full table of necessity in the supplementary materials (see Tables A2 and A3) are not necessary. Conventionally, conditions that pass the consistency threshold of .9, which implies that fewer than in 10 percent of cases the condition is present, but the outcome is absent – can be stated as necessary (Schneider and Wagemann, 2012). Since this occurs very rarely in reality, only those conditions that have a high consistency score are discussed in the text. Unlike in conventional quantitative methods, one can still claim the necessity even with lower consistency scores if one provides a justification as to why the outliers seem to be irrelevant (ibid.)

Only one condition is necessary for the attempt at negotiations to fail (see Table A2), even though it is trivial. The absence of a trade union in the room is fully necessary for the collapse of negotiations, with a consistency score of 1. A statement of sufficiency can be inferred from this, which suggests that whenever a trade union is present, the attempt at negotiations succeeds. The coverage score for this condition is .47. This means that the number of cases with no trade union in the room is larger than the number of cases with failed attempts at pacts, even though a trade union is absent whenever there is a failed attempt. This indicated the trivialness of the condition (Schneider and Wagemann, 2012, pp. 145–147). In line with that, RoN is another parameter that indicates the trivialness of necessity ⁵ (Schneider and Wagemann, 2012, pp. 145–147) and is also quite low at .47.

Analysis of necessity: democratization

The necessity analysis that tests the condition of democratization (see Table A3) shows that for democratization not to happen, the trade unions should be absent. The absence of a trade union in the room is fully necessary for the successful negotiations that did not lead to democratization with the consistency score of .96, which means that in all cases where a trade union is present, the initial success of negotiations is also followed by full democratization. The coverage score for this condition is .56, which means that the number of cases without a trade union in the room is twice larger than the number of cases without democratization. Therefore, this condition is also somewhat trivial.

Two conditions – lacking the Catholic Church in negotiations and transitioning from a personalistic dictatorship – are not considered as necessary. The consistency of necessity for the absence of the Catholic Church scores high at .93. However, this statement is inconsistent with the two cases of failed negotiations – Congo and Ivory Coast where the Catholic Church participated in negotiations and played a very important role. The second condition does not pass the threshold of .77, which means that the statement of necessity is wrong in a quarter of the cases. As many as nine countries coded as personalistic dictatorships still had successful negotiations and at least limited democratization. ⁶ Therefore, there are no grounds to accept the necessity of the two conditions. The other conditions do not appear to be necessary for democratization.

Analysis of sufficiency: a joint fuzzy-set test of the success or failure of negotiations and subsequent democratization

The two outcomes (i.e., the success of negotiations and further democratization) are two consecutive stages of the same process – an attempt at pacted transition. As the analysis of necessity has shown, they are affected by similar conditions in the same direction. Therefore, the two outcomes are merged in the section with sufficiency analysis to avoid tautology in the first place (the two models are mostly similar if conducted on two separate outcomes). When I test for democratization (occurring as a result of a pact) alone, I thus make no distinction between the cases of successful negotiations but no democratization and the cases of failed negotiations. Creating the combined variable allows me to ‘partially include’ cases with successful negotiations and no democratization that would be coded as ‘full exclusion’ otherwise. On top of that, the number of negative cases that are failed negotiation attempts (9 cases) and successful negotiations without democratization (7 cases) is limited (see the cases in Table A1). Since the negative outcome cases are used in the analysis of sufficiency to calculate consistency scores, and my data are skewed toward positive outcomes, merging different sorts of negative outcomes in one model while partially accounting for the difference solves this problem.

To undertake a sufficiency analysis, I first create and then logically minimize the truth table (TT) rows (see Table 1). Each TT row is a possible combination of conditions, and the cases fall into TT rows depending on their membership in conditions. Since I have four conditions, sixteen combinations are possible. Each combination row has one outcome – either 0 or 1. These combinations of conditions can be minimized (for instance, by omitting redundant conditions) to tease out the comprehensible pathways. Each row has a consistency score – namely, the percent of case members that fit the outcome. During the minimization, a consistency threshold needs to be set, which is to say that we have to accept a certain percent of outlying cases before we can claim that belonging to a certain combination of conditions can still be accepted as sufficient for an outcome. I stick to the lowest conventional threshold of .75 (Schneider and Wagemann, 2012), which implies that I accept up to one-quarter of outliers. The borderline is indicated in a table. Therefore, cases below the line have more than a quarter of failed attempts or several successful negotiations that did not lead to democratization.

Table 1 The truth table for the combined fuzzy outcome of both success of negotiations and democratization

To reach a short, comprehensible result, a truth table should be minimized. In the minimization process, the different combinations of conditions that lead to positive outcomes are compared to tease out the solution that ‘is expressed in a more parsimonious yet logically equivalent manner’ (Schneider and Wagemann, 2012, p. 9). In this process, an analysis should deal with logical remainders – the theoretically possible combinations of conditions (or TT rows) that have no matching cases in the empirical world and whose outcomes are therefore unknown. However, in the minimization process, one can make simplifying assumptions on their outcome – or refrain from doing so.

There are three ways of minimizing truth tables. The complex solution makes no simplifying assumptions on where logical remainders lead at all. This leads to the most precise yet possibly cumbersome result. An intermediate solution is based on manually set directional expectations that generally derive from theories and the analysis of necessity. Finally, in a parsimonious solution, simplifying assumptions are automatically made to lead to the shortest result. In my case, the parsimonious way of minimization is equivalent to the intermediate one. Both simplifying assumptions made by the computer simulation and my directional expectations are always set to 1. In other words, each of my conditions is theoretically expected to lead to the positive outcome, successful pact and democratization, while the same simplifying assumption is needed for the most parsimonious outcome.

In the graphical form, the truth table is presented in Figure 2. Each oval-shaped figure indicates the presence of a particular condition – the presence of a trade union, the Catholic Church, party regime, and established opposition party. Each constellation of the conditions leads to a specific outcome. The outcomes are indicated with different fills. Vertical stripes stand for successful negotiations plus further democratization. No fill (white) is failed negotiations and no democratization. Finally, question marks indicate that the data for this constellation is absent.

Figure 2. Venn diagram of constellations of various conditions that affect failed, successful negotiations and democratization as a result of the pact.

The whole condition of the presence of a trade union is filled with vertical lines, which implies that democratization has happened in all of the cases where it was present in negotiations. However, cases with trade unions are always located in the constellations, including other factors contributing to successful negotiations. It is never the case that trade unions are present alone – the upper-left part of the shape corresponding to trade unions indicates a question mark. However, a second constellation where democratization almost always occurred can be seen outside of the trade union oval; in the party dictatorship, it comes together with established parties or the Church.

The solution formula for the fuzzy truth table in the analysis of sufficiency (see Table 2) suggests that democratization follows in two general ways. First, countries democratize whenever trade unions are present during the negotiations. Second, they democratize when the initial regime is a party dictatorship (rather than personalistic) combined with one of the two favorable factors: established opposition parties or the Catholic Church. The sufficiency scores are as high as .95, .98, and .92, meaning virtually no cases deviate from this explanation. The coverage of this model (a close analog of R² in regression analysis) is .69, which means that the model explains 69% of the cases where the outcome – democratization – has occurred, whereas 31% remain unexplained. See Table 3 for the covered and unexplained cases.

Table 2 The parsimonious solution for a truth table of the combined fuzzy outcome of both success of negotiations and a democratic outcome, covS – coverage, covU – unique coverage, overlapping cases with trade unions and party regimes are marked with italics

Table 3 Summary of the final solution for a truth table of the combined fuzzy outcome of both success of negotiations and a democratic outcome

In other words, the model shows two pathways toward successful pacted transition: through the trade unions (I) and from party dictatorships (II). These two pathways are marked with numbers in Figure 2 and Table 2. They are not mutually exclusive, and do, in fact, overlap in eleven cases. Obviously, trade unions are more likely to emerge and sustain in a less restrictive, more institutionalized, and bureaucratized context (i.e., a party regime).

Nevertheless, just having a trade union in negotiations, even if deriving from a personalistic dictatorship, is sufficient for successful democratization through the pact. Neither the trade union nor party dictatorship ever occurs alone. In most cases, the trade unions occur together with the constellation of almost every other condition possible. Speaking of the regime, party dictatorships in my database are always accompanied by at least one more favorable condition, which is not surprising as these regimes provide a less restrictive context and allow for the opposition. Military dictatorships only result in successful negotiations and democratization when the other conditions are present: an established political party, a Church, or a trade union. The two military dictatorships that did not have any of these three conditions – Myanmar in 1988, and Taiwan in 1990, and the one that had an established opposition party – Egypt in 2013 – did not manage to negotiate or democratize.

The Supplementary Material provides two points for consideration. First, the quantitative analysis reaches the results similar to those of the QCA analysis while controlling for the effects of economic development. It also demonstrates that the larger trade unions are, the larger ‘increase in the level of democracy that follows. Additionally, the Supplementary Material showcases a vignette of Tunisia‘s recent transition in the years 2011–2013. It demonstrates the role played by its trade union UGTT.

To sum up, this analysis shows that the same variable – the presence of a trade union in negotiations – is sufficient both for the success of negotiations and for the subsequent democratization. Its absence, therefore, is a necessary condition for the collapse of negotiations and democratization. But, at the same time, it is not the only pathway, and many party or military dictatorships that have resilient churches or established opposition parties (or both) also democratize through a pact.

Concluding discussion: organized labor and successful pacts

The main and the most robust result of the analysis is that the presence of a trade union is sufficient both for democratization and for the success of negotiations, while the lack of a trade union is necessary for the failure of negotiations. At the same time, the QCA analysis has shown, it is still possible to achieve democratization through a pacted transition without a trade union being involved in negotiations, but then there should be several favorable factors combined.

There is no significant difference in the economic development between the countries which have trade unions and those which do not. Therefore, the expectations of modernization theories do not help us explain the effect of trade unions. As I demonstrated in a vignette case study of Tunisia’s revolution, this effect is most likely caused by their organizational capacity. They help ensure credible commitment and provide a credible threat that both make politicians follow their agreements on the earlier stages and ensure democratization later in the first years after the collapse of an old regime.

Given that trade unions in my cases were durable enough to survive the autocratic regime and significant enough to be represented during the negotiations, they also persist after the autocratic collapse. Their existence is therefore way more robust and less dependent on leadership than that of political parties. It is especially relevant for the period after the collapse of autocracy when the parties suddenly become exposed to free and fair elections and a drastic change of agenda. Therefore, as the trade union’s existence after the regime change is very stable and its capacity to start strikes is strong, it helps ensure that the counterparts in negotiations do not eschew their promises right after the negotiations. As they stay as a mediator for a longer period, it helps prevent any power from becoming dominant and starting to change the rules of the game in its favor.

Collier and Mahoney (1997) argued that the working class played a crucial and underestimated role in undermining the authoritarian regimes and the success of democratization. My analysis tests this argument with QCA and statistics and proves that the participation of a trade union, the organized labor in class terms, in a pacted transition has a direct and incredibly strong effect. The results presented in this paper strengthen the thesis of the ‘proletarianization’ hypothesis that suggests a direct connection between the strength of a working class and democratization (Rueschemeyer et al., 1992; Collier, 1999, p. 10). My analysis finds a robust linear connection between a share of the population that is involved in a trade union that participates in negotiations and a subsequent increase in democracy score. Alternatively, this democratizing effect caused by a trade union can have structural, societal origins that go beyond the process of negotiations – the countries where the trade unions are large might have a higher level of what Putnam et al. (1994) call social capital – but even the proxy indicators of this variable are not available for my sample, therefore any testing becomes impossible.

In addition, the results of this analysis support the argument that trade unions never act alone as drivers of democratization – as I have shown (see Figure 2), their presence always goes hand in hand with other organizations of opposition (Valenzuela, 1989, p. 445; Collier, 1999).

This analysis, first, enhanced the argument of Geddes et al. (2018) that party regimes are likely to make pacts and, second, has shown an alternative pathway towards a pacted transition. Several countries underwent successful attempts at negotiations (Sudan in 1985) and democratized through a pact (Peru in 1980, Madagascar in 1992, Mali in 1991, Benin in 1991) despite not being party regimes. Those cases made successful pacted transitions thanks to trade unions or the Catholic Church.

This paper speaks against the original ‘elitist’ understanding of pacted transitions (O‘Donnell et al., 1986; Rosendorff, 2001) that treated negotiations as a kind of exclusive elite exercise that can hardly be affected from below. In line with Collier (1999), it rather proves that, paradoxically, these exclusive elite pacts are the most successful and democratizing when the representatives of the lower class are participating. Finally, this analysis has shown that a successful, democratizing pacted transition demands much more prerequisites than just a fortunate will of political elites: namely the opposition organizations need to be very strong.