SEVEN GENERAL OBSERVATIONS

These books offer six general lessons of broad interest to scholars of party politics and one key point of disagreement.

1. Across the three regions of the world covered in this review essay, institutionalized democratic party systems are the exception rather than the norm. Moreover, in Latin America, from the 1980s until this decade, there was more movement towards party system deinstitutionalization and party collapse than towards institutionalization.

The persistence of cases of weak institutionalization runs counter to some expectations in the literature on the advanced industrial democracies. Aldrich’s (Reference Aldrich1995) classic book suggests that the incentives for politicians to invest in party building in an era of mass enfranchisement are powerful. Whereas Lipset and Rokkan (Reference Lipset and Rokkan1967) famously emphasized the freezing of party systems, most party systems in the third and fourth waves of democratization have experienced high volatility, rapid change, and weak parties (Mainwaring and Zoco Reference Mainwaring and Zoco2007; Mainwaring et al. Reference Mainwaring, Gervasoni and España-Nájera2016; Pop-Eleches Reference Pop-Eleches2010).

‘Why No Party-Building in Peru?’ by Steven Levitsky and Mauricio Zavaleta (in Levitsky et al. Reference Levitsky, Loxton, Van Dyke and Domínguez2016) offers a fascinating analysis of an extreme case of party weakness. In many third wave democracies and semi-democracies, low party system institutionalization is in part a product of bad governance. One might expect parties of presidents who have presided over robust economic growth and sharp declines in poverty to use these successes to cultivate loyalties among citizens and to build an organization. The Peruvian case defies these expectations. Peruvian politicians have adopted party-less strategies and mechanisms to gain electoral success. They are free agents with few attachments to party labels.

Schattschneider (Reference Schattschneider1942: 1) famously wrote that ‘Political parties created modern democracy and modern democracy is unthinkable save in terms of the parties.’ Levitsky and Zavaleta, however, argue that Peru is a democracy without parties. Although I believe that it overstates the case to say that there are no parties in Peru, the evidence of party weakness is unassailable. Persistent party weakness in the context of a democracy raises questions about Schattschneider’s conventional wisdom: must parties still create modern democracy, and is democracy still unthinkable without parties? Still, consistent with the conventional assumption, Levitsky and Zavaleta note some dysfunctional consequences of extreme party weakness, such as hampered accountability and representation.

2. Notwithstanding the prevalence of cases of high volatility, some cases in the third and fourth waves of democratization have developed institutionalized party systems and solid parties. Variance across country cases and individual parties is great.

The primary question for Riedl (Reference Riedl2014) and Roberts (Reference Roberts2014) revolves around understanding these sharp differences in levels of institutionalization. Lupu (Reference Lupu2016) and Levitsky et al. (Reference Levitsky, Loxton, Van Dyke and Domínguez2016) focus on parties rather than party systems, but they also analyse variance across cases.

Levitsky, Loxton and Van Dyke’s introduction (in Levitsky et al. Reference Levitsky, Loxton, Van Dyke and Domínguez2016) strikes a nice balance between highlighting the challenges to party building in third wave democracies in Latin America and observing that some successful new parties have emerged. Some African and Asian countries also have fairly institutionalized party systems (for example, Taiwan, as Tun-jen Cheng and Yung-ming Hsu show in their chapter in Hicken and Kuhonta Reference Hicken and Kuhonta2015). In short, the obstacles to party building in third and fourth wave democracies are powerful but not insurmountable.

3. Democratic longevity does not suffice to ensure party system institutionalization. In an excellent article, Converse (Reference Converse1969) argued that democratic longevity was crucial for building partisan identification, which in turn is a micro-foundation for system-level stability: if large numbers of voters are attached to a party, the percentage of floating voters is necessarily small, limiting system-level instability. The idea that party systems would stabilize over time as politicians invest in building organizations and voters can increasingly use party labels as information short-cuts makes intuitive sense. However, on average, there has not been a trend over time towards greater institutionalization in Latin America, Africa and Asia (Mainwaring et al. Reference Mainwaring, Gervasoni and España-Nájera2016).

In many countries, bad governing performance and corruption have generated increasing citizen cynicism towards parties rather than strengthening bonds between voters and parties. Politicians have created ‘party substitutes’ (Hale Reference Hale2006; Levitsky and Zavaleta’s chapter in Levitsky et al. Reference Levitsky, Loxton, Van Dyke and Domínguez2016), that is, alternative vehicles that do some of what parties traditionally did and enable them to achieve their goals. For candidates for high-level executive posts, television is a particularly important party substitute, enabling candidates to connect directly with voters.

4. In Latin America (though apparently not in Africa or Asia), sharp programmatic differences create stronger party ‘brands’, to use Lupu’s terminology. In turn, stronger party brands facilitate party system institutionalization. Conversely, as Lupu (Reference Lupu2016), Morgan (Reference Morgan2011) and Seawright (Reference Seawright2012) have argued, programmatic convergence puts parties at risk for brand dilution. Coupled with bad governing performance, brand dilution can lead to party collapse.

In their introduction, Levitsky, Loxton and Van Dyke (Reference Levitsky, Loxton, Van Dyke and Domínguez2016) argue that ‘extraordinary conflict’ is the most fertile breeding ground for major new parties. Sharp conflict such as ‘social revolution, civil war, authoritarian repression, and sustained popular mobilization’, fosters stronger brands and attachments and creates stronger incentives for building an organization. Conversely, the muffled policy disputes during the short-lived Washington consensus frequently gave rise to brand erosion.

Along related lines, Roberts (Reference Roberts2014) argues that in the aftermath of neoliberal economic reforms in Latin America, party systems that featured clear programmatic differentiation during the neoliberal period were more stable than those in which programmatic differentiation was minimal.Footnote ¹ In Brazil, Chile, El Salvador, Mexico, Nicaragua and Uruguay, a left-of-centre party consistently opposed neoliberal reform. These cases had sharp programmatic differences across parties; Roberts calls them ‘aligned’ party systems.Footnote ² If voters became disgruntled with mediocre economic performance, they could turn to alternatives within the system. These systems did not collapse.

In contrast, in Argentina, Bolivia, Costa Rica, Ecuador and Venezuela, erstwhile left-of-centre parties undertook neoliberal reform. Programmatic differentiation eroded, leading to what Roberts calls ‘programmatic de-alignment’. These were ‘bait and switch’ cases: presidents promised one set of policies and implemented another (Stokes Reference Stokes2001). The countries that ended the neoliberal juncture without programmatically structured party systems were vulnerable to subsequent partial or full decomposition.

Finally, in Colombia, Guatemala, Honduras, Panama and Paraguay, a conservative party led neoliberal reform, and no major party opposed the changes. Here, too, lack of programmatic differentiation coupled with mediocre or poor governing performance led voters to turn away from the system and towards new options.

Three excellent chapters in Levitsky et al. add further evidence that strong party brands are good for party building in Latin America. Noam Lupu’s ‘Building Party Brands in Argentina and Brazil’ argues that to be successful, new parties must build partisans. He further argues that a clear brand is a necessary condition for building partisans. Party consistency and clear programmatic differentiation from their main rivals facilitate the development of strong party brands. These attributes helped the Partido dos Trabalhadores (PT – Workers’ Party) in Brazil build a clear brand and a strong base of party identifiers. Conversely, in Argentina, Frente País Solidario (FREPASO – Front for a Country in Solidarity) quickly diluted its brand, leading to a collapse of partisanship.

Although Lupu’s argument is persuasive, Brazil’s Partido da Social Democracia Brasileira (PSDB – Social Democracy Party) is a contrary case of a new party that was electorally successful without creating many partisans. Since its creation in 1988, it has won the presidency twice (1994 and 1998) and come second in the last four elections (2002–14), and it has regularly captured powerful state governorships. Yet it has always had relatively few party identifiers, consistently below 7 per cent of survey respondents, raising the question of how parties achieve extended success despite having few partisans.

David Samuels and Cesar Zucco’s chapter, ‘Party Building in Brazil: The Rise of the PT in Perspective’ (in Levitsky et al. Reference Levitsky, Loxton, Van Dyke and Domínguez2016), focuses on the PT’s success in building partisans during the 2000s (PT partisans have declined precipitously in the last several years). It qualifies Lupu’s argument that clear programmatic differentiation is essential for branding. By 2010 about 25 per cent of Brazilian voters were PT identifiers. Contrary to what one might expect based on Lupu’s chapter, PT identifiers proliferated even as the party greatly moderated after winning the presidency in 2002. Samuels and Zucco argue that this was due to the institutionalization of grassroots participation, combined with programmatic centralization. The PT retained a strong brand despite changes that seem inimical to branding – especially diminished programmatic differentiation with respect to other parties and forging governing alliances with conservative parties. The deep decline in PT partisans since 2012, however, suggests that Lupu is also correct: the combination of brand dilution and bad governing performance is often fatal for partisanship.

In her chapter, ‘Insurgent Successor Parties: Scaling Down to Build a Party after War’, Alisha C. Holland (in Levitsky et al. Reference Levitsky, Loxton, Van Dyke and Domínguez2016) examines reasons for wildly different success of insurgent movements that transform themselves into political parties after war. She adds to the evidence that a sharper programmatic identity can be good for building parties. By contrasting the cases of the Frente Farabundo Martí para la Liberación Nacional (FMLN – Farabundo Martí National Liberation Front) in El Salvador and the Alianza Democrática M-19 (AD M-19 – M-19 Democratic Alliance) in Colombia, she argues that insurgent successor parties are more likely to fare well if they do not dilute their brand name by veering sharply to the centre. Created as a revolutionary guerrilla front in 1980, the FMLN transformed itself into a political party in 1992 and became one of the most electorally successful leftist parties in Latin American history. In part, it enjoyed success because, when it traded bullets for ballots, it preserved its image as a leftist party and eschewed opportunistic alliances. The former Colombian guerrilla organization AD M-19 fared well in its first two elections, in 1990 and 1991. However, it has subsequently disappeared, a victim in part of brand dilution by moving to the centre.

Morgan (Reference Morgan2011) and Seawright (Reference Seawright2012) agree that brand dilution makes parties vulnerable to erosion and collapse. Both argued that the programmatic convergence between the former governing parties (Acción Democrática (AD – Democratic Action) and COPEI, the Christian Democratic Independent Political Electoral Organization Committee) was a key ingredient in party system collapse in Venezuela.

If sharp programmatic differences foster party building, then the increased polarization that has characterized most Latin American party systems since the emergence of the new left from 1998 on should be propitious for party building and for system institutionalization. The experiences of Hugo Chávez’s party in Venezuela, the Partido Socialista Unido de Venezuela (PSUV – United Socialist Party of Venezuela), and Evo Morales’s party in Bolivia, the Movimiento al Socialismo (MAS – Movement Toward Socialism), support this hypothesis. Even though the PSUV suffered huge losses in the December 2015 congressional elections, it fared remarkably well for a party that was presiding over triple-digit inflation, shortages of a vast number of products including many foods, and one of the highest homicide rates in the world.

These arguments about the advantages of sharp programmatic differences modify Sartori’s (Reference Sartori1976) influential work on the perils of polarization. Although I agree with these recent works about the advantages of clear programmatic differences, it would be a mistake to conclude that they are an unalloyed asset for democracy. Sharp programmatic differences combined with steadfast normative commitments to democracy tend to be good for democracy, but if some radical (left or right) parties prefer non-democratic outcomes, high polarization can imperil democracy. On a smaller scale, greater polarization in the US has hindered congressional capacity to get things done.

5. Party competition in Africa and Asia does not revolve primarily around different programmatic positions, contrary to Downs’s (Reference Downs1957) assumption. This is a critical difference between the books on Africa and Asia, on the one hand, and the three on Latin America, on the other. In Latin America, clientelism and personalism are pervasive, but intense programmatic competition characterizes many party systems. The weakness of programmatic competition in Africa and Asia means that system institutionalization must be constructed on different pillars: clientelism, personalism and perhaps ethnic bonds.

6. In all three regions, many major parties in post-transition democracies previously supported authoritarian regimes, and they have sometimes helped stimulate party system institutionalization under democracy. This is a major theme in Hicken and Kuhonta’s (Reference Hicken and Kuhonta2015) introduction to their volume, in Riedl (Reference Riedl2014), and also in James Loxton’s chapter, ‘Authoritarian Successor Parties and the New Right in Latin America’, in Levitsky et al. (Reference Levitsky, Loxton, Van Dyke and Domínguez2016). Loxton analyses ‘authoritarian successor parties’ – ‘parties that emerge from authoritarian regimes, but that operate after transitions to democracy’. He addresses why all successful new conservative parties in Latin America have been authoritarian successor parties. These parties have built-in assets (an ‘authoritarian inheritance’) that help them compete in democratic regimes – clientele networks, financial resources, a source of cohesion, territorial organization and a brand name. He illustrates his argument with the cases of Alianza Republicana Nacionalista (ARENA – Nationalist Republican Alliance) in El Salvador and the Unión Demócrata Independiente (UDI – Independent Democratic Union) in Chile. Both were created under dictatorships by high-level regime sympathizers, and they enjoyed electoral success after transitions to democracy.

Riedl (Reference Riedl2014) argues that the institutionalized democratic party systems among the four African cases she examined in depth had their genesis in the antecedent authoritarian ruling party. Likewise, in their introduction to Party System Institutionalization in Asia, Hicken and Kuhonta (Reference Hicken and Kuhonta2015: 15–16) argue that ‘to get highly institutionalized party systems in democratic contexts, it is useful to have had some form of an authoritarian party in power at an earlier point in time’. Advancing a claim that Riedl makes for sub-Saharan Africa, Hicken and Kuhonta (Reference Hicken and Kuhonta2015: 16) argue that the highly institutionalized authoritarian parties in Taiwan (until the 1990s), Singapore and Malaysia helped trigger the emergence of more cohesive and solid opposition parties.

The institutionalized post-authoritarian party system in Taiwan illustrates the point, as the chapter by Cheng and Hsu in Party System Institutionalization in Asia (Hicken and Kuhonta Reference Hicken and Kuhonta2015) shows. The KMT (Kuomintang) was the ruling party in mainland China from 1928 to 1949 and in Taiwan from 1949 until 2000. It successfully transitioned from an authoritarian ruling party to one that competes under democratic conditions, forming the backbone of an institutionalized democratic party system. Consistent with Hicken and Kuhonta’s argument, antecedent authoritarian regimes that failed to build a solid party (such as those in the Philippines and Thailand) gave rise to democracies with weakly institutionalized party systems. Governing parties in these countries lacked the resources for party building that the KMT enjoyed.

7. These books offer a mixture of some agreement and some disagreement about causes of party system institutionalization, party building and party collapse. The three books on Latin America share a broad agreement about the benefits for party system institutionalization (or party-building) of strong party brands or sharp programmatic differentiation. The books on Africa and Asia offer divergent perspectives. For Riedl, democratic party system institutionalization hinges on decisions made by incumbent authoritarian elites prior to incipient political liberalization. Hicken and Kuhonta’s main argument about the Asian cases is that democratic party system institutionalization is facilitated by the institutionalization of an authoritarian party during the antecedent dictatorship. Although these books have advanced understanding of why party systems institutionalize or erode, there is ample space for additional research on this question.

PARTY BRANDS IN CRISIS

Noam Lupu’s (Reference Lupu2016) book is a tour de force. It is innovative; it takes on an important new research question; the scholarship is first rate. The argument is clear and parsimonious. It evinces deep knowledge of the relevant literatures and it goes beyond them in important ways. It moves effectively between the individual voter and party levels.

Previous scholars including Morgan (Reference Morgan2011), Seawright (Reference Seawright2012) and Tanaka (Reference Tanaka1998) have written important books on party system collapse. Lupu focuses on party collapse; the unit of analysis is individual parties. It is the first work I know of to address this issue. Party collapse can have far-reaching effects on democratic politics, and the phenomenon has been surprisingly common in Latin America in the last two decades. Lupu’s argument is also relevant for understanding party system erosion or collapse.

Lupu argues that it takes both a sharp drop in party identification and economic crisis or some other poor governing performance for a party to collapse. He skilfully uses a variety of methods – cross-national quantitative data, case studies of two parties each in Argentina and Venezuela, a survey experiment – to test the argument. He builds on US literature on party identification, but also modifies it in rich ways. Lupu begins with an outstanding discussion of party brands (Chapter 2) and partisanship. Strong party brands foster deep attachments from citizens. Parties can dilute their brands by sharp programmatic shifts (‘party inconsistency’) or by converging programmatically towards other parties. If parties dilute their brands, partisanship often erodes, making governing parties more dependent on performance in office. Thus, party behaviour, especially consistency and programmatic distinctiveness, affects voter attachments.

A governing party that enjoys strong partisan attachments can withstand poor performance in office. Parties that dilute their brand and preside over economic crises are vulnerable to collapse. In Latin America in the 1980s and 1990s, many erstwhile left-of-centre governing parties faced this toxic combination and collapsed.

Chapter 3 provides solid quantitative evidence for 105 Latin American presidential elections from 1978 to 2007 that the combination of brand dilution and bad government economic performance makes governing parties vulnerable to collapse. In contrast, bad economic performance by itself did not increase the likelihood of party collapse. The qualitative case studies in Chapters 4 and 5 provide in-depth evidence for these points. Argentina’s Unión Cívica Radical (UCR – Radical Party) presided over a severe economic downturn, including hyperinflation (3080 per cent) in 1989, when it governed from 1983 to 1989. Because it still commanded a large contingent of party identifiers, even an economic disaster of this magnitude did not crush the party electorally; its presidential candidate won 37 per cent of the vote in 1989. But when the party diluted its brand in the 1990s, UCR partisanship plummeted, depriving the party of a reliable core of voters. Following another economic disaster under a UCR president (1999–2001), the party was doomed to electoral oblivion.

Decreased party identification without bad government performance did not produce party collapse. Argentina’s Peronist Partido Justicialista (PJ – Justicialist Party) engaged in brand dilution in the 1990s, leading to a decline in partisanship. But because economic results under PJ presidents were respectable, the party was able to continue to win the presidency. Likewise, in Venezuela (Chapter 5), governing parties survived lacklustre economic performance when they had a stable of partisans, but brand dilution in the 1990s greatly diminished partisans. The country’s two largest parties of the 1958–88 era collapsed in 1998 under the combination of brand dilution and bad economic performance.

In Chapter 6, a survey experiment tests whether treatment effects that emphasized the programmatic differences or blurring among parties would affect respondents’ propensity to identify with a party. The theory predicts that treatment effects that call attention to programmatic differences should increase the likelihood that respondents identify with a party, and vice versa. The results support the theory. Moreover, younger and less informed voters should on average have less crystallized prior opinions and preferences and hence should be more likely to be affected by the treatment. On this score, the evidence is mixed; only three of eight treatment effects were statistically significant (at p<0.10) – all in the hypothesized direction.

Chapter 7 shows that, consistent with the theory, more polarized party systems tend to foster more partisanship. Moreover, consistent with theory, individuals who perceive greater polarization in the party system are more likely to be partisan; these individuals have a sharper perception of the party’s brand. Using panel surveys conducted in the US in 1965, 1973, 1982 and 1997, Lupu shows that individuals who perceived high polarization between the parties were more likely to become more partisan over a long time span.

Although Party Brands in Crisis is an exceptional achievement, I question a few arguments. For example, there is a mismatch between the coding rules for party breakdown and the cases that Lupu includes in this category. Lupu defines established parties as ‘those that in two consecutive elections either receive a plurality of the (presidential) vote, or attract no less than one-third of the winning vote share’ (Lupu Reference Lupu2016: 43). The operational criterion for a breakdown is that ‘from one election to the next, an established party ceases to meet the criteria for being considered established’ (Lupu Reference Lupu2016: 44). Lupu counts 11 cases of party breakdowns for the 1978–2007 period, but many others also seem to qualify: the Conservative Party in Colombia in 2002; the Liberal Party in Colombia in 2006; the Bolivian Conciencia de Patria (CONDEPA – Conscience of the Fatherland) in 2002; the Izquierda Democrática (Democratic Left) of Ecuador in 1992, and so on.

Lupu’s evidence for the cases he analyses in depth is powerful, but it travels less well to most cases of party breakdown in Latin America. Only one other case among his 11, the breakdown of the Colorados in Uruguay in 2004, appears to meet the conditions for breakdown as clearly as the three cases in the in-depth qualitative chapters. Three of his 11 cases of breakdown did not involve parties that governed when they broke down (the Movimiento de la Izquierda Revolucionaria (MIR – Movement of the Revolutionary Left) in Bolivia; the Partido de Avanzada Nacional (PAN – National Advancement Party) in Guatemala; and the Alianza Popular Revolucionaria Americana (APRA – American Popular Revolutionary Alliance) in Peru). The theoretical argument does not apply to them. The other two Bolivian parties that broke down after governing do not fit the story entirely; economic performance was not bad when they broke down in 2002 and 2005, respectively. The Costa Rican Partido de Unidad Social Cristiana (PUSC – Social Christian Unity Party) broke down in 2006 after presiding over one of the most robust periods of economic growth in that country’s history. The Partido Reformista Social Cristiano (PRSC – Social Christian Reformist Party) of the Dominican Republic broke down in 1996 after presiding over two years of fairly rapid growth and moderate inflation.

Lupu’s argument about the importance of a clear party brand for building partisanship is convincing, but there are exceptions. The US in the mid-twentieth century combined strong partisanship, limited brand differentiation at the national level, and deep intraparty differences.

Finally, in my view, the jury is still out on what survey questions about partisanship tap in some Latin American countries. The 2010 Brazilian national election panel survey showed high individual-level volatility in party identification over the course of just a few months. Of 275 individuals who reported a party identification in Wave 1 of the survey who were also interviewed in Wave 2, only 135 expressed the same party identification. Mexican panel surveys have also exhibited high individual-level instability in partisanship over a short time. If individual-level short-term instability is rampant, what are we to make of the existing measures of the concept?

Despite these minor reservations, Party Brands in Crisis is an excellent book, with outstanding analysis and conceptual and theoretical innovations.

AUTHORITARIAN ORIGINS OF DEMOCRATIC PARTY SYSTEMS IN AFRICA

In her agenda-setting book, Riedl (Reference Riedl2014) explains variance in the level of institutionalization of democratic party systems.Footnote ³ Her answer unfolds over three path-dependent stages. First (Chapter 4), the antecedent authoritarian incumbents faced a choice of how to rule. They could either incorporate local elites (Ghana and Senegal) or they could suppress them and replace them with state-sponsored organizations (Benin and Zambia). Both strategies could enable the authoritarian incumbents to govern, but they produced very different consequences for their ability to survive competitive elections and remain powerful contenders in a democratic or semi-democratic regime.

In a second stage (Chapter 5), as pressures to liberalize authoritarian rule emerged, the rulers who governed by incorporating local elites enjoyed strong support in society and were therefore able to control transitions to more open political regimes. The support of local elites enabled the dictators to win local elections. In contrast, the rulers who employed state substitution (Benin and Zambia) faced the defection of local elites. They had less support in society and were less able to control the transitions.

In the third stage (Chapter 6), controlled transitions led to more orderly, stable party systems in Ghana and Senegal. The authoritarian leaders imposed their preferred electoral systems so as to help preserve their advantage. Conversely, more open transitions in which the authoritarian incumbents were soundly defeated led to more volatile party systems in Benin and Zambia. Thus, the distal key to understanding the stability of ‘democratic’ party systems is the political strategy (incorporation versus state substitution) chosen by dictators at an early stage.

The country cases are well chosen because Benin and Senegal, and Ghana and Zambia, shared several important historical features but ended up with divergent levels of party system institutionalization. Riedl offers a wealth of new data and convincingly explores and refutes alternative explanations for variance in institutionalization (Chapter 3). The book is theoretically, conceptually and empirically sophisticated. Riedl tests the entire causal chain with her four case studies (Chapters 4 to 6), and the third stage of her argument with a statistical analysis for 22 African countries (Reidl Reference Riedl2014: 62, Table 3.1). The case studies largely support her arguments. The statistical analysis (although based on a very low N) supports the claim that the authoritarian incumbent’s electoral success in the ‘founding election’ is correlated with subsequent party system institutionalization. The research design is commendable, and she makes good use of interviews.

Other aspects of her book are less convincing or are puzzling for non-Africanists. Riedl’s analysis of her four main cases is convincing; the evidence to support generalizing her arguments about the first two stages beyond these four cases to other sub-Saharan African countries is less so. Riedl claims that dictators’ initial choice of ruling strategy was largely idiosyncratic, but she does not demonstrate this point. Some dictators are better positioned than others to elicit the support of different local elites and hence would be more likely to choose incorporation. For example, a dictator might be more likely to pursue incorporation in places where he is a member of the same ethnic group (or religion) as the local elites, and more likely to suppress local elites and pursue state substitution where the local elites come from a different ethnic or religious group.

More analysis of the degree to which electoral competition was fair and the regime was democratic during the period she considers ‘democratic’ would be useful. Levitsky and Way (Reference Levitsky and Way2010) regard all four of her cases as competitive authoritarian regimes throughout the 1990s, which, according to Riedl, is after the regime transitions. For some cases, then, her argument might not be about institutionalization of democratic party systems, but rather about the capacity of authoritarian elites to forestall a transition.

This observation suggests a minor tweaking of Riedl’s explanation: variance in party system institutionalization could result largely from differences in the competitive playing fields. Authoritarian incumbents should generally fare better in elections that are less fair and free, producing greater system stability precisely because competition was less fair. Freedom House scores lend initial plausibility to this hypothesis. The combined Freedom House scores for the two cases of high volatility (5 for both Benin in 1991 and Zambia in 1991) were much more democratic than the scores for the two cases of high institutionalization, Senegal in 1981 (8) and Ghana in 1992 (10) at the time of the transitions to multiparty elections.

In Authoritarian Origins, the authoritarian incumbents chose between incorporation of local elites and state substitution uniformly across the country’s territory. This argument might be correct in these four cases, but many authoritarian rulers incorporate some local elites (perhaps those of their ethnicity, religion or nationality) and brutally suppress others. Iraq and Syria today are tragic examples. Riedl’s analysis assumes that local elites are homogeneous enough across the country’s territory that the authoritarian rulers can adopt a uniform strategy towards them, but this homogeneity is the exception rather than the norm in countries with deep ethnic, religious and national divides. This raises a question about why (and in what ways) local elites were homogeneous enough that the authoritarian incumbents could choose a uniform strategy towards them.

In Riedl’s account, the authoritarian incumbents stuck to their initial strategy of incorporation or substitution. But some regimes switch courses, so one wonders why the dictators in Senegal, Benin, Ghana and Zambia stuck to their initial strategy. The Brazilian military dictatorship of 1964–85 initially pursued a path of state substitution, but it quickly realized that the support of local political elites could help undergird the regime. As a result, in 1965, it created a party, Aliança Renovadora Nacional (ARENA – National Renewal Alliance), that allowed supportive civilian politicians to rule at the local and state levels.

Another question that Riedl’s book poses is whether and why there was so much path dependence across the three stages of her argument. The authoritarian rulers’ strategy towards local elites aligns closely with how much support they had in society when they began the transition, and in turn this variable strongly predicts subsequent party system institutionalization. Riedl (Reference Riedl2014: xiv–xv) characterizes some democratic African party systems as having ‘persistently high institutionalization’ while others have ‘more fragmentation and weak institutionalization’. She argues that founding elections have great sticking power on subsequent institutionalization.

The Latin American reality has been starkly different. There was no clear path dependence between dictators’ strategy, on the one hand, and their level of support in society around the time of the transition, or between their support at the time of the transition and subsequent party system institutionalization. If our beginning point is the strategy that dictators employed towards parties (rather than local elites in Riedl’s analysis),Footnote ⁴ some authoritarian regimes that suppressed parties ended up almost entirely bereft of electoral support (Argentina 1966–73 and 1976–83, and Uruguay 1973–84), consistent with Riedl’s argument. In other cases, however, ‘authoritarian successor parties’ (Loxton’s concept) associated with military dictatorships that suppressed parties (Bolivia 1971–8, Chile 1973–90) won ample electoral support in free and fair elections. Some authoritarian successor parties were electorally successful long after transitions to democracy (e.g. Mexico’s Partido Revolucionario Institucional (PRI – Institutional Revolutionary Party)), but other authoritarian successor parties faded (Brazil). Some country cases moved from high to low institutionalization over time (Argentina, Colombia and Venezuela), while a few moved in the opposite direction (Brazil and El Salvador). Founding elections and modes of transitions to democracy did not have much explanatory power past the short term. Likewise, in the minimally democratic post-Soviet cases, early institutionalization did not necessarily stick (Pop-Eleches Reference Pop-Eleches2010). These cross-regional differences pose questions about the mechanisms by which the initial institutionalization might get reproduced more in Africa than in Latin America or the post-communist countries.

CHANGING COURSE IN LATIN AMERICA

Kenneth Roberts’s (Reference Roberts2014) book examines the evolution of Latin American party systems over a long time, from the era of import substitution industrialization or ISI (the 1930s until roughly 1980) to 2010. It is essential reading for scholars who study Latin American party systems. It evinces great scope and ambition and theoretical boldness and offers innovative arguments.

Roberts uses a critical junctures approach, asserting that the introduction of neoliberal economic policies profoundly reshaped patterns of political representation. In his analysis, the great mover of party system change is change in economic models. He delineates three time periods according to economic models: the period of import substitution industrialization (from the 1930s to the 1980s in most countries); the era of neoliberal policies (roughly 1980 to the second half of the 1990s); and the aftermath, a ‘reactive sequence’ characterized by the deep questioning of neoliberalism. His primary dependent variable is the stability or volatility of party systems. In the book’s second and third time periods, the level of programmatic structuring in party systems and the nature of the left turn are also dependent variables.

Roberts classifies the pre-1980s party systems as labour mobilizing or elitist. The former had two characteristics: ‘the emergence in the ISI era of a new, mass-based populist or leftist party . . . that . . . exercise[d] political leadership of the trade union movement and [was] a serious contender for national political power through electoral means’ and moderately high trade union density. Party systems that did not meet both characteristics were ‘elitist’. Roberts posits that once they formed early in the ISI era, party systems remained in one category or the other.

This static categorization is problematic. By my reckoning, seven of the 16 country cases shifted from one category of party system to the other during the era of import substitution industrialization. The characteristics that mark a party system as labour mobilizing or elitist changed profoundly in these countries. For six of these seven cases (all but Bolivia), in minor asides, Roberts acknowledges that his static categorization overlooks important change – but his theoretical argument rests on it.

Roberts includes Nicaragua as a labour-mobilizing party system because of the 1979 Sandinista revolution, but until 1979 this was not the case. The union movement was very weak, and there was no electorally significant left-of-centre party. He classifies the Uruguayan party system as elitist during the entire import substitution industrialization era, but from 1971 on, with the creation of the leftist Frente Amplio (FA – Broad Front) as a major electoral contender, this categorization is questionable. Roberts regards Mexico and Venezuela as having labour-mobilizing systems, but both became labour demobilizing long before the end of the import substitution industrialization era. The classification of Brazil as a labour-mobilizing party system is sensible from 1946 until the 1964 military coup, especially in the final years preceding the coup, but Brazil had a repressive anti-labour military dictatorship and a restricted party system from 1965 until 1980. It is debatable whether Bolivia’s Movimiento Nacionalista Revolucionario (MNR – Revolutionary Nationalist Movement) ever met the criterion of exercising ‘political leadership of the trade union movement’ (Roberts Reference Roberts2014: 71). For most of the post-1945 period, electorally minor Marxist parties led the labour movement. The classification as labour mobilizing is defensible for the 1950s, when the revolutionary MNR governments won labour’s support, but when they turned to stabilization policies in the early 1960s, the MNR/labour alliance shattered. Peru met the criteria for a labour-mobilizing party system from the time of the creation of APRA in the 1930s until the 1950s, but APRA turned to the centre-right in the 1950s and 1960s and eventually lost support in the labour movement. By the early 1970s, the Peruvian Communist Party and other Marxist parties that were electorally marginal controlled the most important labour organization, the Peruvian General Labour Confederation.

Political changes such as military dictatorships that suppressed parties and/or created new ones, revolution (in Nicaragua), elite pacts that subordinated the labour movement (Venezuela), the emergence of new leftist parties (Uruguay), and a sharp turn away from leftist, nationalist policies by major parties that had mobilized labour (Mexico, Peru and Venezuela) led to profound changes in party systems long before the transition from import substitution industrialization to neoliberal policies. The book largely neglects these important political causes of party system change in favour of focusing on the neoliberal period as a critical juncture. These profound party system changes anteceded the transition to neoliberalism by decades in Venezuela, Mexico and Brazil. If many party systems experienced profound changes during the era of import substitution industrialization, and if many factors beyond economic models deeply influenced their development, there is no compelling theoretical reason to anchor the analysis of party system change and stability primarily in economic models.

In Chapters 5 (the core chapter) and 7 and 8 (country case studies), Roberts explains the divergent outcomes in party system stability during the era of neoliberal restructuring. The profound reorientation in economic models and deep economic crisis in most countries led to major change in most party systems. He argues that labour-mobilizing systems were more likely than elitist systems to experience massive disruptions during this period. At first glance, the empirical evidence supports this argument. From 1980 until 2000, the eight party systems classified as labour mobilizing had a mean electoral volatility of 32.4 in presidential elections and 27.1 in lower chamber (or unicameral) elections, compared with 18.4 and 19.0 for the eight elitist party systems (Roberts Reference Roberts2014: 109, Table 5.5).

However, this argument is a product of the problematic coding of the party systems on the eve of neoliberal policies. If we reclassified the seven cases that seem miscoded in 1979 (Bolivia, Brazil, Mexico, Nicaragua, Peru, Uruguay and Venezuela), the relationship between the kind of pre-1980s party system and electoral instability during the 1980–2000 period is the opposite of what Roberts reports: a country mean of 18.8 and 11.8 for presidential and lower chamber elections, respectively, for the three labour-mobilizing cases (Argentina, Chile and Uruguay) compared with 27.0 and 25.6 for the 13 elitist party systems.

The causal logic that Roberts (2014: 93–102) uses to connect labour-mobilizing party systems with deeper turmoil in the neoliberal era and elitist party systems with greater stability is also questionable. He invokes two causal mechanisms. First, the labour-mobilizing cases experienced deeper economic crises because they were embedded in more statist economies. They faced ‘a deeper set of adjustment burdens’ because they had ‘advanced further with state-led development’ (Roberts Reference Roberts2014: 93). In turn, deeper economic crises provoked greater disruption in party systems.

The claim that the economic crises were on average more intense in the labour-mobilizing cases is correct if we accept Roberts’s sorting of the cases, but it is worth considering alternative hypotheses for why some of these cases faced exceptionally difficult adjustment burdens. It is not clear that it was primarily because they had advanced further with state-led development. For example, the US-funded war against Nicaragua’s revolutionary Sandinista government (1979–90) and ill-advised economic policies contributed greatly to the severe economic distress in the late 1980s. Nicaragua was not a case of deep statism or a strong import substitution industrialization coalition prior to the 1979 revolution, so it does not conform to Roberts’s theoretical reasoning. More generally, it is not obvious that Latin American economies characterized by greater statist intervention necessarily bore deeper adjustment burdens in moments of economic downturn. Deductively, it could go either way; it might be easier to cut state spending where it is greater to begin with.

Roberts’s second causal mechanism about why the labour-mobilizing cases experienced greater party system volatility under neoliberalism is that they were built on programmatic linkages between popular sectors and parties. These linkages were disrupted by neoliberal restructuring more than the linkages in the elitist party systems (Roberts Reference Roberts2014: 93). In these cases, stabilization attempts ‘generated fierce political resistance and distributive conflicts’ (Roberts Reference Roberts2014: 93).

Roberts does not offer systematic empirical support for the argument that labour-mobilizing cases generated more political resistance than the elitist systems and that this political resistance was an important factor in the worse economic performance of the former cases. Indeed, he undermines this assertion when he writes that ‘levels of social protest fell sharply in Latin America during the late 1980s and early 1990s, the peak years of structural adjustment’ (Roberts Reference Roberts2014: 124). In fact, in several countries, successful stabilization attempts in the wake of hyperinflation generated strong popular support (Weyland Reference Weyland2002).

It is not clear that economic crises were deeper in systems classified as labour mobilizing primarily because of political resistance and distributive conflicts. For example, Argentina’s hyperinflationary crisis in the late 1980s resulted more from the disastrous economic legacy of the military dictatorship of 1976–83 and ill-advised heterodox economic policies undertaken by the Alfonsín government (1983–9) than from ‘fierce political resistance and distributive conflicts’. The latter characterized the 1983–9 and 1999–2002 periods, but they were not the primary causes of the severe crises. The hyperinflationary crisis in Peru (1988–90) under Alan García resulted more from ill-advised bank nationalizations and other misguided economic policies, terrorism and the debt crisis inherited from the military dictatorship of 1968–80 than from popular resistance. Overall, the empirical relationship between labour-mobilizing party systems and more severe economic crises is not clear, and the causal mechanisms Roberts invokes to explain this gap in severity (if there is one) are not convincing.

Chapters 6 (the core analysis) and 9 (case studies) cover the third historical part of the book, the aftermath of neoliberal reform. They attempt to explain variance in party system stability and in programmatic alignment. This is the strongest part of the volume, with sharper clarity about the causal argument, important original claims and solid empirical evidence. As I discussed earlier, Roberts’s primary argument is that party systems in which the main left-of-centre party was not tainted by brand dilution during the neoliberal period were more stable in the aftermath.

Roberts’s analysis of the implications for programmatic alignment or dealignment circa 1998 for the left turn in Latin America that began that year is equally insightful. In much of Latin America, leftist parties and populist leaders assumed power from 1999 on, with Hugo Chávez in Venezuela leading the charge. But the kind of left that came to power was very different in the programmatically aligned and dealigned cases. In the latter, a more radical, anti-system left sometimes emerged. The prior party system offered no credible alternative to neoliberalism, so disenchanted voters turned to radical outsiders. In contrast, in countries in which a left-of-centre party had consistently opposed neoliberal policies, the system offered a credible alternative. Voters turned to these alternatives in Brazil in 2002 (the PT), Chile in 2001 (the Socialists), El Salvador in 2009 (the FMLN), Nicaragua in 2006 (the Frente Sandinista de Liberación Nacional (FSLN – Sandinista National Liberation Front)) and Uruguay in 2004 (the FA).

Roberts’s attempt to categorize and explain differences in Latin American party systems across a wide sweep of time has many strengths. At the same time, this ambitious book could have been improved with more rigorous coding of cases, measures that consistently matched the arguments, and empirical measures of the programmatic alignment of party systems. It could also have been improved with multivariate testing of some arguments and further consideration of alternative explanations, some of which I have briefly suggested. Clear, explicit and consistent coding of when the era of import substitution industrialization, the neoliberal juncture and the reactive sequence began and ended in different countries, along with measurements of party system change that match the beginning and end of these periods, would also have helped. Although the case studies convey Roberts’s deep knowledge of Latin America, they do not consistently show the causal mechanisms that he invokes.

Ultimately, I am sceptical that the shift from import substitution industrialization to neoliberalism and the subsequent challenge to neoliberalism offer a compelling and parsimonious explanation for the stability or volatility and programmatic nature of Latin American party systems. Profound political change including regime change, dictatorships that suppress parties and sometimes create new ones, revolutions, and radical programmatic shifts in parties that are not tightly connected to regional trends in economic models can also prompt radical party system change. Major changes in ideological winds and corruption scandals (Seawright Reference Seawright2012) can also have a profound effect on parties and party systems.

THE EDITED VOLUMES

Spatial constraints and the diversity of arguments in the two edited volumes allow for only a very short overview. Levitsky et al.’s Challenges of Party Building in Latin America (Reference Levitsky, Loxton, Van Dyke and Domínguez2016) is an excellent if uneven volume on an important subject. It analyses challenges and successes of party building in the post-1978 wave of democratization in Latin America. It enhances knowledge about conditions for successful party building in an era of great challenges to that enterprise. The authors include many distinguished senior scholars and some excellent younger ones. Many arguments in the volume about party building are relevant for understanding party system institutionalization. The unit of analysis (parties versus party system) is different, but party building and system institutionalization are linked; systems with solid parties are generally institutionalized, and vice versa.

Hicken and Kuhonta’s Party System Institutionalization in Asia (Reference Hicken and Kuhonta2015) makes a valuable contribution to the broader comparative literature and brings the Asian cases squarely into the broader comparative debate. The book fruitfully brings together the analysis of party system institutionalization under democratic, semi-democratic and authoritarian regimes in 12 Asian countries, including all of the largest. All of the chapters except the editors’ introduction and my conclusion are single-country studies. They provide clear accounts of party system institutionalization, or in the cases of single-party regimes, of party institutionalization.

Like Riedl (Reference Riedl2014), Hicken and Kuhonta emphasize path dependence in the formation of post-authoritarian party systems. They temper this emphasis with an awareness that ‘some cases may track differently’ (Hicken and Kuhonta Reference Hicken and Kuhonta2015: 13). Although their arguments about path dependence are suggestive, the book does not develop them in as much detail as Riedl does.

One source of creative contention within Hicken and Kuhonta’s book revolves around the degree to which social scientists can use the same indicators to measure party system institutionalization under both democratic and authoritarian regimes. In their introduction, Hicken and Kuhonta argue on behalf of using more or less the same measures to assess institutionalization in both types of regimes. In the conclusion, I argue that political scientists should distinguish between different kinds of regimes and party systems before they analyse party system institutionalization. Party systems in democratic regimes are a different breed from hegemonic party systems or party-state systems (such as the one dominated by the Chinese Communist Party). Indicators of institutionalization under democracy do not necessarily imply institutionalization under authoritarianism. For example, electoral volatility is a good indicator of party system institutionalization under democracy. Under authoritarian rule, low electoral volatility could express a stability of electoral patterns that stems from an uneven playing field.