When Argentina redemocratized in 1983, its established political parties returned to contest elections. The Radical Civic Union (UCR), founded in 1891, emerged as a party of middle-class sharecroppers and small farmers threatened by the influx of European immigrants (Persello 2007).With the rise of Peronism in the 1940s, Argentina's oldest mass party formed the basis of the middle- and upper-class opposition to Juan Domingo Perón.

The Peronist party – officially called the Justicialist Party (PJ) – was the electoral vehicle founded by Perón in the 1940s. Proscribed on and off from openly competing in elections in the intervening decades, the party drew its electoral support from the rural poor and urban working classes (Lupu and Stokes 2009), developing corporatist and patronage ties to the Argentine labor movement. Its elites often relied on nationalist rhetoric promoting statist economic development, a “revolution in production,” and redistributive social justice (McGuire 1997).

To be sure, both the established Argentine parties were heterogeneous and, at times, inconsistent in these positions. Both also suffered various internal conflicts that sometimes fractured the party. For some scholars of Argentina, the distinction between the parties was one of rhetoric rather than ideology (Ostiguy 2009).

Rhetoric, ideology, and class appeal often went hand in hand, however. Whereas Radicals appealed to the upwardly mobile middle classes by promoting labor flexibility and strengthening democratic institutions (Gibson 1996), Peronism appealed to popular sectors with corporatism and nationalist rhetoric (Mora y Araujo 1995). In September 1986, nearly 50 percent of survey respondents named the UCR as “the party most bound to privileged sectors”; only 8 percent named the PJ.

One-quarter of the established political parties in Latin America have broken down since the mid-1990s. These were not fleeting organizations: some were more than a century old. Many had survived economic booms and busts, repressive dictators, guerrilla insurgencies, even revolutions. What brought these once powerful political organizations to their knees so quickly? Why did voters abandon the parties they had once been so attached to – parties that some had risked jail, torture, or even death to defend just a few years earlier?

Answering these questions requires knowing something about how citizens form political attitudes and make voting decisions. It also requires that we understand the incentives of party elites, and how their actions affect mass behavior. Explanations of party decline, particularly in studies of Latin America, typically focus on elite strategies and structural constraints. But party breakdown is fundamentally about the attitudes and choices of individual voters. If we want to understand why established parties break down, our macro-level explanations must start with a model of how individual citizens relate to parties.

To understand party breakdown in Latin America, we must understand how party brands influence voters' choices. When a party's brand is clear, some voters form strong attachments to it, attachments so strong that voters may ignore the party's actual performance in office. But at times, parties become indistinguishable and their brands blur. When this happens, voters detach from parties and start to look at the party's performance more seriously when making voting decisions. Parties become increasingly susceptible to voters' short-term retrospective evaluations. When established parties both dilute their brands and perform poorly in office, voters defect en masse from established parties.

Party brands are an important heuristic for voters. They may pay little attention to politics and lack the time or resources to evaluate each candidate's record or proposals in detail.

In the early 1980s, many Latin American voters identified with established political parties. But during much of the 1990s, Latin Americans gradually became less attached to these parties. Attachments with both of Argentina's oldest parties declined over the course of the 1990s and early 2000s. In Venezuela, voters felt less and less affinity with that country's main parties during the 1990s. Declines in mass partisanship made these established parties susceptible to voters' short-term retrospective evaluations and breakdown.

My branding model of partisanship posited that these declines in mass partisanship resulted from the parties diluting their brands. Case studies showed that during the 1990s, political elites in Argentina and Venezuela implemented policies that were inconsistent with the traditional positions of their party, provoked intraparty conflicts, and formed strange-bedfellow alliances with traditional rivals. I argued that these behaviors diluted the brands of their parties and showed that they were indeed associated with eroding voter attachments in the aggregate.

Still, we cannot know for sure from these aggregate relationships whether my individual-level branding model is at work. Most studies of mass partisanship implicitly assume that party behavior has no effect whatsoever on individual party attachments. So there is in fact good reason to be skeptical. We need to test directly whether party behavior informs partisanship at the individual level.

I posited that as major parties converge, voters become unable to distinguish one party brand from another, weakening partisan identities. In other words, when rival parties signal convergence in salient ways – say, by forming an alliance – the difference between their brands shrinks, making it less likely that voters will feel a comparative fit with one over the other. In Argentina, bipartisan legislative support, repeated attempts to form coalition governments, and policy convergence between the PJ and UCR were associated with eroding attachments with both parties. When AD and COPEI similarly converged, entered into coalition, and cooperated to forestall electoral losses, Venezuelan voters' affinities with the parties similarly weakened. Thus, I expect party convergence to weaken partisanship, whereas divergence strengthens it.

In order to test the effect of party convergence on partisanship, I embedded an experiment in a 2009 survey of voters in Argentina.

Between 1958 and 1993, Venezuela's two major parties, Democratic Action (AD) and the Independent Political Electoral Organizing Committee (COPEI), together drew an average of 78 percent of the vote in national elections. But by 1998, a mere 3 percent of Venezuelans cast their ballots for these parties. After Bolivia transitioned to democracy in 1980, the three parties that dominated politics – the rightist Nationalist and Democratic Action (ADN), centrist Revolutionary Nationalist Movement (MNR), and center-left Revolutionary Left Movement (MIR) – together received an average of 67 percent of the vote. But in 2002, ADN attracted only 3 percent of the votes, and neither it nor the MIR even fielded a presidential candidate in the 2005 election.

The dramatic and sudden declines in the staying power of established political parties – parties with an extended history of being competitive for national office – is one of the most puzzling features of Latin American democratic politics since the ThirdWave of democratization. Between 1978 and 2007, onequarter of the region's established parties broke down, meaning that they suddenly became uncompetitive for national executive office. These breakdowns entailed an average drop of nearly 80 percent in the share of the party's vote from one election to the next. Parties that had only recently been major competitors were relegated to an average vote share of merely 6 percent. Yet these very parties, some more than a century old, had survived economic booms and busts, authoritarian repression, guerrilla insurgencies, and revolutionary movements.

Why, then, have many broken down in recent decades? Traditional theories of party politics fall flat in explaining party breakdown. Scholars of political parties expect party systems to form around enduring social cleavages or the political struggles that surround their emergence. Proponents of spatial models of party competition expect parties to consistently match voter preferences. Neither tradition can explain why established parties break down suddenly and decisively.

Scholars of comparative politics often attribute party breakdowns instead to poor performance by incumbent parties. Corruption scandals or poor economic stewardship, they argue, cause voters to reject the incumbent party en masse, leading the party to break down. But bad performance is far more widespread than party breakdown, and established parties have survived some major economic crises. In the 1980s in Peru, for instance, President Alan García's economic policies led to some of the worst hyperinflation in world history, peaking in 1989 at 12,378 percent.

In Latin America, party convergence is associated with weakening partisan attachments. And that relationship appears to be causal: when survey respondents were given information that signaled party convergence, they became less likely to say that they identify with a party. However, the recent dramatic policy switches and strange-bedfellow alliances in Latin America are rare in other parts of the world. Perhaps we are observing a contingent set of relationships that will not obtain elsewhere. In other settings, does party polarization clarify party brands and strengthen voter partisanship?

There are good reasons to think not: party polarization has long been seen as a problem for democracy. Studies link the polarization of the Democratic and Republican Parties in the United States to legislative gridlock, elite incivility, income inequality, and voter disengagement (e.g., Hetherington 2009; Layman et al. 2006; McCarty et al. 2006). Scholars also blame party polarization for democratic breakdowns in Germany's Weimar Republic, France's Fourth Republic, and Chile prior to 1973 (Sani and Sartori 1983; Sartori 1976; Valenzuela 1978; but see Bermeo 2003). And across a broad range of countries, polarization also seems to contribute to corruption, bad economic performance, and smaller government (Brown et al. 2011; Frye 2002; Lindqvist and Östling 2010).

But we still know little about the relationship between party polarization and individual attitudes like voter partisanship. Research in the United States highlights the coinciding rise in party polarization and resurgent voter partisanship (Abramowitz and Saunders 2008; Bartels 2000; Brewer 2005; Hetherington 2001; Levendusky 2009). But these studies primarily describe trends within a single case. They leave unexamined the causal connection between party polarization and individual partisanship. A handful of comparative studies find a similar association between party polarization and rates of partisanship in aggregate terms (Berglund et al. 2006; Holmberg 1994; Schmitt 2009; Schmitt and Holmberg 1995). But these studies rely on simple correlations from a very small sample of Western European countries. Do these correlations generalize beyond this limited sample? Do voters actually register party polarization and change their perceptions of their parties?

At the individual level, the causal link between party polarization and voter partisanship also remains both unclear and untested. Studies of the United States and Western Europe emphasize the salience of party competition. Following Carmines and Stimson (1989), they expect voter partisanship to increase as parties become more and more salient in political discourse.

Why have a quarter of Latin America's established parties collapsed in recent years? Perhaps they were simply unlucky, as many scholars suggest: they governed at times of economic crisis and simply fell afoul of public opinion. Or perhaps their downfall was an unintended consequence of the economic and institutional reforms they themselves passed, as others argue.

Both answers are partially right. The international economic environment of the 1980s and 1990s severely limited the policy options of leaders around the world, and frequent international economic crises also weakened many Latin American incumbents. But running for reelection on the heels of an economic crisis was not enough to cause an established party's total breakdown. Although many suffered serious electoral losses, they often lived to fight another day. Similarly, economic reforms in and of themselves did not always hurt incumbent parties electorally. When the reforms succeeded, incumbents were often rewarded at the polls.

It is the way economic reform was often pursued that made parties susceptible to collapse. During the 1980s and 1990s, leaders across the region implemented policies that were inconsistent with their traditional party brand, provoked intraparty conflicts, and formed strange-bedfellow alliances with historic rivals. These behaviors diluted the brands of their parties, eroding voters' attachments to them.Without the assured support of partisans, governing parties became more susceptible to negative valence evaluations and, when confronted with new economic crises, broke down.

Empirically testing this proposition across the region requires abstracting away from the specifics of the argument. Figures on voter partisanship are simply not available for most Latin American democracies for most of this time period. Nor do we have sufficient data on voters' evaluations of their leaders' economic performance. Moreover, reliable cross-national data on party politics in the region are far from perfect. So any region-wide analysis of these hypotheses can only be considered suggestive. Nevertheless, a comparative analysis of the whole region can tell us whether behaviors that dilute party brands and economic crises generally correlate with party breakdown. On its own, finding such a relationship would not compellingly establish the validity of my argument, but it would be a suggestive first step in the process of hypothesis-testing.

A region-wide comparative analysis can also help us determine whether the explanations put forward by prior studies generalize across the region. Do parties become more susceptible to breakdown in the wake of political decentralization?

Venezuelan democracy reemerged in 1958 following the ouster of military dictator Marcos Pérez Jiménez. Leaders of the country's major political parties returned from exile to contest the first election. The oldest among them, Democratic Action (AD), was also the most powerful. AD had the broadest popular appeal and geographic penetration (Ellner 1982; Martz 1966). The party always saw itself as a multiclass organization even as it developed close ties with organized labor and, at times, the rural peasantry. Party elites often characterized AD as social-democratic; indeed, one of its leaders, Carlos Andrés Pérez, was a three-term vice president of the Socialist International, the global umbrella organization for socialist and social-democratic parties.

For the following four decades, AD's primary rival was the Independent Political Electoral Organizing Committee (COPEI), a party founded by Catholic critics of AD's first governments. COPEI's support was based in the conservative and religious Andean states (Álvarez 2004; Combellas Lares 1985; Herman 1980), the same region that produced the country's military dictators. Supporters considered COPEI a Christian democratic party, and one of its leaders, Eduardo Fernández, served as president of the Christian Democrat International.

Initially, AD and COPEI committed themselves to representative democracy and some informal collaboration in the Pact of Punto Fijo. But that collaboration faded over time, and the two parties came to control stable segments of the electorate. Between 1973 and 1988, AD and COPEI attracted an average of 90 percent of the vote in presidential elections (Figure 5.1).

The remarkable stability of AD and COPEI made Venezuela an exception among Latin American democracies. Some observers credited the parties with buttressing Venezuelan democracy even as democracies across the region broke down in the 1970s. Others criticized their stability and control over organized civil society for undermining political competition and representation (Coppedge 1994; Gil Yepes 1978; Karl 1997; Levine 1973).

Both camps were stunned by the collapse of these established parties in 1998.