Abstract
In young democracies with weak parties, there is some evidence that partisan identification may shift in response to short-term government performance. The massive protests that erupted in Brazil in June 2013 sharply increased the salience of, and public attention to, poor government performance and took most observers by surprise. They were also widely depicted as nonpartisan or even antipartisan. We use two well-timed surveys to examine the effects of the protests on mass partisanship. We find that the protests led to increased nonpartisanship and decreased attachment to the governing Partido dos Trabalhadores (Workers’ Party, PT) among the public as a whole. We also show that small leftist parties were more broadly represented among protesters than has been previously recognized.
Introduction
On 6 June 2013 on Avenida Paulista, one of the main thoroughfares in the city of São Paulo, a group of around 2,000 people organized by the Movimento Passe Livre (Free Fare Movement, MPL) gathered to protest the 7 percent hike in bus and metros fares that had just been put into effect in the city. The protests slowly grew in size over the next week until 13 June, when harsh police repression of the protesters galvanized public opinion in favor of the protesters and greatly increased their number. On 17 June an estimated 200,000 protesters took to the streets in cities across Brazil (Benson and Levine 2013). And on 20 June the number of people participating in the demonstrations surpassed one million (Prada and Marcello 2013).
What began as localized protests motivated by a transit fare hike turned into a national movement against increases in the cost of living, substandard provision of social services, corruption, police brutality, and more. The reach of the protests was massive; according to our survey conducted in July, only 3 percent of Brazilians were not aware of them. At their height, the protests were certainly larger than those in 1992 at the time of the impeachment and resignation of President Fernando Collor de Mello and arguably larger than those surrounding the 1985 transition to democracy (Moseley and Layton 2013; Saad-Filho 2013).
While protest participation and awareness were extremely high, the possible medium- to long-term impact of the protests on Brazilian politics remains unclear. In contrast to 1984 and 1992, neither the regime itself nor the legitimacy of Brazil's popularly elected president (Dilma Rousseff) was seriously questioned. Although the president's approval ratings fell precipitously over the course of the protests and, as of this writing, have only partly recovered, she remains the most popular candidate in polling in advance of the 2014 election.
Although the protests seem unlikely to upend the Brazilian political order entirely, we provide evidence from a unique set of surveys that the protests may have medium-term consequences for one particular aspect of political life: partisan sympathy. While most of the literature from long-standing democracies treats partisanship as the “unmoved mover” of other political views and behaviors, growing evidence from younger democracies with historically weak parties suggests that citizens’ partisan identities are more malleable in these contexts. 1 In fact, recent studies from Brazil provide evidence that Brazilian partisanship is responsive to government performance in the short term (Baker et al. 2010; Weitz-Shapiro and Winters 2014).
For the classic statement of partisanship as an “unmoved mover,” see Campbell et al. (1960). For reviews of the debate about the malleability of partisanship, see Johnston (2006) and Lavine, Johnston, and Steenbergen (2012). For some evidence of the growing strength of partisanship in the Brazilian context, see Samuels (2006) and Samuels and Zucco (2012).
Two features of the June 2013 protests make them a particularly interesting lens through which to examine partisanship in Brazil. First, although sparked by an increase in bus fares, the protests quickly took on a much broader character that took politicians, journalists, and other observers by surprise. While scholars have hypothesized that poor performance can decrease partisan support and attachment (Baker et al. 2010; Chong et al. 2013), the long time horizon over which voters typically observe these changes makes it difficult to isolate the possible effects of these transformations. Given the surprise nature of the protests, they can be understood as an exogenous shock that suddenly increased the salience of poor performance for voters. Second, the protests were widely portrayed and understood as nonpartisan or even antipartisan. This should have increased the probability that Brazilian citizens would view the failings of government highlighted by protesters as a failure of parties writ large.
We take advantage of unique data collected just before and after the June protests to examine the relationship between the demonstrations and partisan attachment, both for the small group of protesters and the mass public. 2 First, we describe the partisan makeup of the protesters themselves. We show that individuals who reported participating in the protests were far less likely to express attachment to the governing Partido dos Trabalhadores (Worker's Party, PT) when compared with other respondents, but that protesters report relatively high rates of affinity with smaller leftist parties. These findings complicate the existing narrative about the antipartisan nature of the protests. Second, we show that the protests were followed by a marked rise in nonpartisanship among the general population. Most of this increase in nonpartisanship can be explained by a decrease in support for the PT. These shifts in partisanship were particularly concentrated in the urban epicenters of the demonstrations, São Paulo and Rio de Janeiro. Finally, we show that in the city and state of Rio de Janeiro, the incumbent Partido do Movimento Democrático Brasileiro (Brazilian Democratic Movement Party, PMDB) – another major political party in Brazil – suffered a significant decrease in partisan identification in the aftermath of the protests. These patterns in the data provide important evidence that discontent expressed through protest can affect mass public opinion more generally. They also point to the continued fragility of partisan identity in Brazil and provide further evidence that partisan attachment in Brazil is quite sensitive to short-term political performance.
Both surveys were run as part of IBOPE's monthly nationwide omnibus survey. The first survey was enumerated over the period 16–20 May 2013, and the second over the period 11–15 July 2013. For both surveys, IBOPE sampled 140 cities using a probability-proportional-to-size (PPS) method within 25 strata defined by 25 of Brazil's 27 states – the survey rotates on a monthly basis among three small states in the North. Then census tracts were selected using PPS with stratification across zones of major metropolitan areas. Enumerators recruited individual respondents using a quota sampling scheme designed to produce a sample representative of the national population in terms of age, gender, and socioeconomic category.
Not-So-Nonpartisan Protests?
The June 2013 protests have been thus far understood as largely nonpartisan or even antipartisan. Over the course of the protests, observers noted the widely expressed desire to go to the streets “sem partido” (“without political parties”) (Mische 2013). Protests were rife with placards expressing antipartisan sentiments, among them: “No party represents me!” and “We don't have a party. We are Brazil!” 3 At times, these sentiments veered into violence, with supporters of political parties harassed by other protesters and their party flags burned (Mische 2013; Saad-Filho 2013; Tatagiba and Blikstad 2013).
“Nenhum partido me representa” and “Não temos partido. Nós somos Brasil!” (Mische 2013).
Our July 2013 survey asked directly about protest participation and thus allowed for comparisons between protesters and nonprotesters on a number of dimensions. Three percent of our total sample, or 64 respondents, reported having attended or participated in one of the demonstrations that took place in June. Focusing on partisanship, Table 1 compares the protesters in our sample to the remainder of the sample. 4 Our data show clear evidence of the muted role of the PT in the protests. Relative to the general population, protest participants were significantly less likely to claim an affinity with the PT: only 13 percent of protesters reported feeling close to the PT, compared to 23 percent of the remaining respondents. This decreased affinity with the PT, however, did not translate into a highly apartisan group of protesters. Instead, lower levels of attachment to the PT were partially compensated by unusually high rates of affinity with two small parties further to the left on the ideological spectrum than the PT: the Partido Verde (Green Party, PV) and, especially, the Partido Socialismo e Liberdade (Socialism and Freedom Party, PSOL). 5 As a consequence, the rate of nonpartisanship among self-reported protesters was only slightly lower than among the remainder of respondents, and this difference is not statistically significant. These results suggest that while media portrayals of the protests as being disengaged from the PT were correct, partisan sentiment overall was more widespread among protesters than has been understood. Furthermore, although many academics and left-wing activists have expressed concern that the PT's weakness in the protests created space for reactionary or right wing groups to dominate the protests, our findings suggest that left-wing partisan groups had a much stronger presence.
Partisanship among Protesters and Nonprotesters
Note: From July 2013 survey conducted by IBOPE on behalf of the authors. Rows in bold indicate differences that are statistically significant at the p < 0.10 level.
Source: Authors’ own calculation.
Protesters in our sample were also more likely to be male, young, and have attended at least some college. Results available upon request.
The PV was founded in 1986 and currently holds 15 seats in the Chamber of Deputies. Marina Silva, a PT defector who had served as President Lula da Silva's environment minister, received nearly 20 percent of the first round presidential vote in 2010 as the PV candidate. The PSOL was founded in 2004 by dissident PT members who had voted against Lula's proposed pension reform. Founder Heloisa Helena placed third in the first round of the 2006 presidential election.
Protest Awareness and Partisanship
Although only a small proportion of Brazilians participated in the June 2013 protests, nearly all of our respondents were aware of the protests. Table 2 shows that three-quarters of our respondents at least read or heard about the demonstrations on the news and that almost half of our respondents discussed the demonstrations with their family and friends. 6 Thus, even with relatively limited direct participation, it is clear that the protests had the character of a national event – noticed by nearly everyone and therefore likely to be salient for the formation of opinions about the government and political parties.
Awareness of Protests
Note: From July 2013 survey conducted by IBOPE on behalf of the authors.
Source: Authors’ own calculation.
Seventeen percent of respondents reported being aware of the protests but “not paying much attention,” while only three percent reported not being aware of them.
In Table 3, we report respondents’ answers to a question asking them to name the three most important motivations for the demonstrations. A majority of people named the catalyzing event, the fare increases, as one of the three reasons. Notably, however, almost as many people cited the delivery of health and education services, and corruption also ranked very highly. Somewhat surprisingly, police violence and repression – two variables that have figured prominently in most accounts explaining the rapid spread of the protests – were mentioned by only 8 percent of respondents. This suggests that by the time of our survey in early July, the demonstrations were broadly understood through the lens of the quality of government service delivery and the problem of corruption, rather than the more specific circumstances responsible for the start and spread of the protests.
Perceived Reasons for the Protests
Note: From July 2013 survey conducted by IBOPE on behalf of the authors.
Source: Authors’ own calculation.
In order to assess the effects of the protests on partisan attachment, we compared expressions of partisan identification measured just before the protests to those measured just after. Although not a panel study, comparability was enhanced by the fact that we compared data from two surveys conducted by the same polling firm using the same survey methodology. In Table 4, we show the rising rates of nonpartisanship between May and July. In the May survey, 52 percent of our respondents said that they did not identify with any political party. After the outbreak of the demonstrations in June, this level had risen to 57 percent. The rise in nonpartisanship was particularly pronounced in the two epicenters of protest, São Paulo and Rio de Janeiro. In the city of São Paulo, nonpartisanship increased from 53 to 61 percent; in the city of Rio de Janeiro, from 30 to 43 percent.
Rates of Nonpartisanship
Note: From May and July 2013 surveys conducted by IBOPE on behalf of the authors. Rows in bold indicate differences that are statistically significant at the p < 0.10 level. Standard errors are based on the Welch correction for unequal variances.
Source: Authors’ own calculation.
Much of this rise in nonpartisanship is associated with voters defecting from the governing PT. In Table 5 we show that PT partisanship dropped nationwide by four percentage points, from 26 to 22 percent, between May and July. The drop in PT support was again particularly pronounced in São Paulo and Rio de Janeiro, where the share of PT partisans shrank from 32 percent to 18 percent and 34 percent to 16 percent, respectively.
Rates of PT Partisanship
Note: From May and July 2013 surveys conducted by IBOPE on behalf of the authors. According to a standard t-test, all differences are statistically significant at the p < 0.05 level except for the small cities indicator.
Source: Authors’ own calculation.
No single opposition party appeared to benefit from the general populace's rising disenchantment with the PT. The Partido da Social Democracia Brasileira (Brazilian Social Democratic Party, PSDB) – the PT's strongest competitor in presidential races – saw their levels of support increase by only one percentage point from 4.3 percent in May to 5.4 percent in July (p < 0.11). Brazil's other major party, the PMDB, which garnered higher levels of identification than the PT in the first half of the 1990s, actually saw its proportion of supporters fall from 7.4 to 5.5 percent (p < 0.02); as we explain below, much of this decline was linked to the defection of PMDB supporters in Rio de Janeiro.
According to most reports at the time of the protests, the PT was the target of much of the protesters’ ire – not surprising given its control of the presidency and the mayoralty of São Paulo, where the protests started. However, the other main locus of the protests was the city of Rio de Janeiro, a historical PMDB stronghold with a PMDB mayor and governor. Among the issues of special importance to Rio were the city and state's poor planning for the 2016 Olympic Games, as well as the cost overruns that have beset Brazil's 2014 World Cup preparations. Both the mayor and governor were roundly condemned for their repression of the protests. The events surrounding the protests were clearly damaging for their party. Between May and July 2013 support for the PMDB fell from 8.6 to 2.9 percent in the city (p < 0.15) and from 8.3 to 3 percent in the state (p < 0.04).
Major Political Events, Partisan Change and Partisan Futures in Brazil
The demonstrations that began in June 2013 in Brazil promoted a wide variety of short-term policy changes, including the rescinding of bus fare increases in São Paulo and Rio de Janeiro. At the same time, the extent to which the protests will have lasting political consequences remains unclear. After meeting with protest leaders at the end of June, President Rousseff announced a “national pact” to invest more in city transportation, health and education; to make corruption a felony; and to create a constituent assembly to consider additional political reform (The Economist 2013). As of November 2013, however, many of these initiatives have been slowed or stymied in the legislature.
In this paper we point to a heretofore unexamined consequence of the protests: partisan change among Brazilian citizens. In contrast to contemporary understandings of partisanship as a social identity that is not responsive to short-term political events, we provide evidence of a case where mass partisanship shifted in response to a particular political event. The unexpected nature of the protests, which took observers of all types by surprise, allow us to consider them an exogenous shock and attribute changes in partisanship to the protests and the events immediately surrounding them. Although relatively few Brazilians participated in the protests, information about them was widely disseminated by the media and through (virtual and real) social networks, leaving only a small portion of Brazilians unaware of them. By comparing partisanship levels in May with those in July, we found that the June protests and the events surrounding them led to a five percentage point increase in the share of Brazilians declining to identify with any political party. The governing PT was hit the hardest, suffering a four percentage point decrease in affiliation. 7 These shifts were particularly pronounced in the two largest cities in Brazil, São Paulo and Rio de Janeiro. This short-term shift in partisanship is consistent with other previous work that shows the responsiveness of partisanship to local political events in developing democracies (Baker et al. 2010; Chong et al. 2013).
Other public opinion data collected during this period corroborates our findings. For example, Datafolha is a large survey firm that runs regular public opinion polls. In early June, before the protests, 23 percent of its respondents identified with the PT and 55 percent identified with no party. Their data, too, show a sharp rise in nonpartisanship immediately after the protests. Data they collected in October 2013 and February 2014 show these levels have not returned to pre-protest levels: 18 percent report identifying with the PT in both October and February and 60 percent and 66 percent report no party preference in October and February, respectively. “Avaliação Dilma Rousseff e intenção de voto presidente,” 6–7 June 2013, “Avaliação da presidente Dilma Rousseff,” 11 October 2013, “Avaliação da presidente Dilma Rousseff,” 19–20 February 2014, all at <www.datafolha.com.br> (12 March 2014).
Our data also show that no other party has benefited from the loss of support for the PT. This makes it possible that the PT will eventually recover some of its erstwhile supporters. The PT has lost, and then recovered, supporters in the past – most notably in the aftermath of the mensalão scandal, which revealed the PT's reliance on payments to congressional deputies in exchange for legislative support. Insofar as PT supporters are “bounded partisans” (Baker et al. 2010), it is possible that some citizens will return to PT identities in the run-up to the 2014 elections. On the other hand, we observe some regional trends here that may speak to lasting change in the nature of the PT's coalition. Since the election of Luiz Inácio Lula da Silva as president in 2002, the PT has steadily gained support in the country's North and Northeast, while losing support in its traditional strongholds of the South and Southeast. The effects of the June 2013 protests in depressing PT support in the major urban centers of Rio and São Paulo may further reinforce this shift.