Is the Informal Sector Politically Different? (Null) Answers from Latin America

World Development 102 (2018) 170–182

Contents lists available at ScienceDirect

World Development journal homepage: www.elsevier.com/locate/worlddev

Is the Informal Sector Politically Different? (Null) Answers from Latin America Andy Baker, Vania Ximena Velasco-Guachalla University of Colorado Boulder, United States

a r t i c l e

i n f o

Article history: Accepted 20 September 2017

Key words: Latin America informal sector mass political behavior elections voting

s u m m a r y Scholars have produced a limited understanding of the effect of informal labor status on a worker’s political attitudes and behavior. We present descriptive evidence on the micropolitical correlates of informality using direct measures of the concept in public opinion surveys from 18 Latin American countries. We test three scholarly impressions of informal workers—that they are less politically engaged, more rightleaning, and more favorable toward noncontributory social programs than formal-sector workers. These are grounded in a dualist conception of labor markets that views the formal and informal sectors as having little overlap. We find minimal evidence for these impressions and argue that recent empirical findings consistent with a revisionist view of informality better account for our null results. According to this view, informal and formal labor markets are highly integrated, which, we argue, melds together the economic interests and political preferences of individuals in both sectors. We also provide evidence that casts doubt on alternative explanations that would attribute our null results to the timing of our surveys, to arational sources of political behavior, or to measurement error. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction High rates of economic informality are virtually a defining characteristic of macroeconomies, product markets, and labor in less developed countries. Despite this, social science has produced a limited understanding of the effect of informality status on a person’s political attitudes and behavior. Many scholarly views of micropolitics in the informal sector are impressionistic, while survey-based studies of mass political behavior rarely distinguish informal from formal workers. In practice, scholars have tended to rely on broad indicators of wealth, income, or class (Mainwaring, Torcal, & Somma, 2015) and on labels such as ‘‘popular sectors,” ‘‘low-skilled sectors,” or ‘‘the urban poor” (Collier & Handlin, 2009). All of these skirt the defining essence of informality: the lack of state presence. We present descriptive evidence on the micropolitical correlates of informality using direct measures of the concept in surveys from 18 Latin American countries. We test three scholarly impressions of informal workers, all of which are grounded in a dualist conception of labor markets that views the formal and informal sectors as having little overlap. One is that informal workers are less politically engaged than formal-sector workers, the second is that they are more right-leaning in vote choice and issue attitudes, and the third is that they are more favorable toward noncontributory social programs. We find little evidence for these impressions. https://doi.org/10.1016/j.worlddev.2017.09.014 0305-750X/Ó 2017 Elsevier Ltd. All rights reserved.

We argue that recent empirical findings consistent with a revisionist view of informality better account for our null results. The informal and formal sectors are much more integrated than the dualist view holds, as evidenced by relatively frequent worker transitions between the two and by pooling within households of formal and informal earnings. This integration melds together the economic interests and political preferences of individuals in both sectors. We also provide evidence to dismiss alternative explanations that would attribute our null results to the timing of our surveys, to arational sources of political behavior, or to measurement error. 2. The dualist view of informality and its micropolitical implications Scholarly impressions of the informal sector, especially in political science, are inspired by the dualist view of labor markets. Grounded in de Soto’s (1989) canonical treatment, it considers labor markets in developing countries to be deeply segmented between the formal and informal sectors (Harris & Todaro, 1970). Informal workers are seen to be almost permanently shut out of formality by high regulatory barriers and a lack of available opportunities. They treat unregistered work as a last resort while queuing for better prospects in the formal sector that rarely transpire. For example, Portes and Hoffman betray this view of rigidity by referring to social classes, of which the ‘‘informal proletariat” is

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

one in their framework, as ‘‘discrete and durable categories of the population” (Portes & Hoffman, 2003, p. 42). Similarly, Banerjee and Duflo refer to South Asia’s many sole proprietors of unregistered firms as ‘‘reluctant entrepreneurs,” on the premise that these proprietors prefer steady employment in the formal sector but rarely achieve it (Banerjee & Duflo, 2011, p. 205). The dualist viewpoint does not itself yield ready-made expectations about mass political behavior in the informal sector, but the impressions offered by scholars of politics are generally consistent with this view of labor-market immobility and near-permanent exclusion from formality. In the 1960s, many scholars saw the new urban poor—most of them migrants from rural areas who moved to squatter settlements and worked informal jobs—to be a potential source of radical political activism (Huntington, 1968; Soares, 1967), but this view was soon countered with a portrait that largely prevails to this day (Nelson, 1970). The most frequently made, long-held, and well-known argument, which we call the undermobilization hypothesis, is that informal workers are less likely to be collectively organized and to participate in politics. Because they tend to be either wage earners in small firms or self-employed sole proprietors, informal workers are socially atomized and relatively lacking in immediate, visible common interests (Blofield, 2011, p.8; Kurtz, 2004; Dix, 1989). This heterogeneity in social experience and in policy demands creates high barriers to collective action:  ‘‘Small artisan and service shops, vending, and domestic service are inherently difficult to organize. . . . Even social and religious organizations are weak among the urban poor” (Nelson, 1970, pp. 405–6).  ‘‘informal and micro-enterprise sectors . . . are notoriously difficult to organize, as workers’ economic activities leave them widely dispersed, disconnected, and unregulated” (Roberts, 2002, p. 22).  ‘‘. . . informal sector presence . . . pacif[ies] marginalized populations” (Milner & Rudra, 2015, p. 669).1 Another frequently made argument, which we call the rightleaning hypothesis, is that informal workers are less likely to vote for the political left than formal workers. Scholars do not always flesh this argument out as clearly as they do the undermobilization hypothesis, but we see two purported mechanisms for this alleged voting pattern. The first stems in part from the undermobilization hypothesis itself: union affiliation links formal workers into the left’s primary organized constituency, while informal workers are less likely to be collectively organized and thus less receptive to the ideological and class-based political appeals made by leaders of the political left (Roberts, 2002, p. 22). Untethered by class- and group-based sympathies, informal workers are more susceptible to nonprogrammatic and ideologically ambiguous appeals, often from clientelistic elites (Cameron, 1991; Oxhorn, 1998):  ‘‘Clientelistic linkages are better suited than unions to organize—and win votes among—the fragmented and heterogenous strata of urban unemployed, self-employed, and informal sector workers generated by deindustrialization” (Levitsky, 2003, p. 140).

1 The undermobilization hypothesis also underlies Ansell and Samuels’ (2014, p. 42) model of regime change, and it finds resonance in OECD contexts. David Rueda (2005) makes an analogous argument about the distinction between workers with secure, full-time employment (‘‘insiders”) and those with more flexible and part-time arrangements (‘‘outsiders”): ‘‘outsiders tend to be less politically active and electorally relevant . . . than insiders” (p. 62).

171

 ‘‘Weaker union organization [in Latin America than in Western Europe], weaker civil society, high levels of clientelism, weaker social capital . . . all weaken the counter-hegemonic position of progressive forces and thus translate into a less hospitable climate of public opinion for left political parties” (Huber & Stephens, 2012, p. 39).  ‘‘For many low-income groups, the consequence of labour market flexibilization and the collapse of prior patterns of corporatist incorporation has been social fragmentation, and this has limited the capacity for political participation. . . . It then becomes individually rational to vote for a vote-buying party or candidate over one with a programme of redistribution. Thus the political system is ‘unanchored’ from the poor and allowed to drift further to the centre or centre-right” (Schneider & Soskice, 2009, p. 45). The second mechanism is that informal workers possess more capitalist-friendly values. Since many informal-sector workers are small-time entrepreneurs and even employers, they may be natural critics of Latin America’s interventionist and exclusionary states, an assertion that underlies de Soto’s provocative claim that ‘‘the constituency of capitalism has always been poor people that are outside the system”2 (de Soto, 1989; Weyland, 1996). Similarly, in the 1970s and 1980s, various scholars explained the surprising lack of socialism and radicalism among the new informal sector’s core—rural-born workers who moved to cities—in terms of a ‘‘migrant ethic” (Portes, 1971, p. 713). The ethic promoted individualist values that eschewed class appeals and structuralist explanations for their economic plights:  ‘‘Wealthier and better-educated people may think in terms of governmental policies and their effect on economic conditions, but the poor and uneducated [in Third World cities] are less likely to blame the authorities for general economic difficulties” (Nelson, 1970, p. 404).  ‘‘. . .urban migrants . . . tend to see their present and future in terms of individual, rather than class or group, mobility. Their demands tend to center on acquiring a bit of land on which to construct a dwelling and on such amenities as sewers and transportation for their barrios, rather than on grievances against a factory boss, much less against the capitalist system itself. Such is not the kind of social situation in which class solidarity thrives” (Dix, 1989, p. 32). We label the third impression the noncontributory preference hypothesis, which holds that informal workers favor Latin America’s relatively new noncontributory social assistance policies over its traditional contributory social insurance ones. This newer scholarly impression is rooted in the fact that social policy benefits lie at the core of the informality/formality divide. Built up incrementally over the course of the 20th century, Latin American welfare states have largely been grounded in Bismarckian, corporatist principles of work-based, contributory social insurance (Esping-Anderson, 1990): the government administers payroll taxes on formalsector work and grants benefits only to active and former contributors and their families (Haggard & Kaufman, 2008; Rudra, 2008). Informal workers, by definition, are labor ‘‘outsiders” that do not benefit from (and often even partially fund through consumption taxation) these payroll-funded social programs. Thus, they should prefer means-tested and universalistic social assistance, such as the conditional cash transfer and minimum pension programs that have been implemented in many countries in recent years (Garay,

2 http://www.pbs.org/wgbh/commandingheights/shared/minitextlo/int_hernandodesoto.html.

172

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

2016; Levy, 2008, chap. 4).3 By contrast, formal-sector workers, as labor ‘‘insiders,” should prefer investment in the traditional contributory schemes4:  ‘‘Workers in formal sector, dependent jobs show greater support for contributory insurance policies, since these provide benefits that stand in direct relationship to their contributions (and restrict benefits to contributors). . . In contrast, independents are more likely to support non-contributory social policies . . . Recognizing the difficulty they face in making regular insurance contributions out of their wages or informal income, they prefer more bare-bones programmes that are financed through general tax revenues” (Carnes & Mares, 2014, p. 706; see also Carnes & Mares, 2016, pp. 1650–1652).  ‘‘. . .growth in the informal sector . . . creates new constituents for noncontributory social spending” (Dion, 2010, 46; see also Garay, 2016).  ‘‘. . . those who work in the formal sector, with payroll taxes and benefits, typically oppose providing ‘free’ non-contributory benefits to those who work for low pay in the informal sector. . .” (Willmore, 2014, p.8). In sum, scholarly impressions are that, relative to formal-sector workers, informal workers are politically and socially disengaged and right-leaning when they do vote (Altamirano & Wibbels, 2012). They may also be more pro-market in their attitudes or, potentially, more favorable toward means-tested social programs.

3. The revisionist view of informality and its micropolitical implications These impressions have a strong hold in political science, but the dualist view of labor markets upon which they are based may be inaccurate. Recent research on informality has put forth what we call a ‘‘revisionist” vision that finds the borders between formal and informal work to be porous on at least two grounds. First, intertemporal shifts between the two sectors occur with high frequency: ‘‘the labor market is characterized by substantial mobility of workers between the formal and the informal sectors” (Levy, 2008, p. 42; also IDB, 2004 and Maloney, 1999). Table 1 provides some of the main evidence for this: occupational transition probabilities gathered from employment and household panel surveys in three countries. The probability that an informal wage earner transitions to formal wage earning in any given year is incredibly high, ranging from .21 (Argentina) to .45 (Brazil). The probability that a formal wage earner transitions to informal wage earning is lower at about 10% in all three countries; however, considering this is the probability in any given year, the odds of any single transition over a life cycle or even a single decade are very high. In fact, economists have pinpointed a life cycle pattern that is common across national contexts: youth enter the labor market as informal wage earners, secure formal wage-earning posts in 3 Moreover, informality may be partially endogenous to means-tested programs— as some workers avoid formalizing to continue receiving benefits (Garganta & Gasparini, 2015)—which would further reinforce this divide in policy attitudes. 4 The noncontributory preference hypothesis need not be contradictory with the right-leaning one. Indeed, some see the former as a third possible mechanism underlying the latter. Because of their tight representational linkages to organized labor (Hunter, 2010, pp. 91–2, 151), leftist leaders often promoted ever more generous Bismarckian benefits for insiders to the neglect of, and sometimes in outright opposition to, progressive social policies that would boost well-being in the informal sector (Haggard & Kaufman, 2008, p. 197; Pribble, 2013, p. 28). In reality, however, parties from across the ideological spectrum, many of them leftist, have introduced noncontributory programs, so the relationship to vote choice would vary by country (Brooks, 2015; Huber, Mustillo, & Stephens, 2008). For this reason, we categorize the noncontributory preference hypothesis separately, and we discuss the party-of-implementation issue more below.

Table 1 Annual probability of transitioning between occupational statuses

Formal wage earner ? Informal wage earner Informal wage earner ? Formal wage earner Formal wage earner ? Informal self-employed Informal self-employed ? Formal wage earner

Argentina

Brazil

Mexico

0.09 0.21 0.02 0.03

0.08 0.45 0.02 0.08

0.10 0.39 0.04 0.14

Source: Perry et al., 2007, pp. 54–55; Gapminder.org to calculate weights for male and female share of the workforce. Note: Argentina data collected semiannually from 1993 to 2001, Brazil data collected monthly from 1982 to 2001, and Mexico data collected quarterly from 1987 to 2004.

middle age, and transition to informal self-employment in their senior years (Perry et al., 2007, p. 53). For many, this is because— counter to the reluctant entrepreneurs claim—informality is a lifestyle or even an economic choice. Second, intra-household integration of formality and informality is common. In many countries, formal sector workers secure health insurance and other benefits for their entire family through their paycheck contributions, so one spouse often chooses informal employment to avoid double-paying for benefits (Maloney, 1999, p. 287; Perry et al., 2007, p. 194). Because of this incentive structure, Galiani and Weinschelbaum find that, across Latin America, ‘‘spouses, ceteris paribus, are more likely to work informally if the head of household works formally” (2012, p. 834). In Mexico, they estimate that more than 35% of working married couples with primary to secondary education are discordant pairs, meaning one spouse is informal and the other formal.5 All told, this integration on multiple grounds minimizes the difference between informal and formal communities as well as the extent to which political behavior and preferences are rooted in one’s present work situation. Because of the intertemporal mobility, the policy interests and partisan preferences of informal and formal workers are likely to be far less distinct than what is assumed by the dualistic approach: ‘‘. . . the frequent movement of workers between the formal and informal sectors attenuates the insider/outsider cleavage” (Schneider, 2013, p. 108). For example, many informal workers may not object to generous formalsector benefits because they realistically aspire to formality in the future, while many formal workers may support meanstested or universalistic programs because they anticipate a potential spell of informality (Carnes & Mares, 2014, 2016). Because of the household pooling of income and benefits, the distinction to any one member between formal and informal structures is muted. In the end, we expect political differences between formal and informal workers to be small to nonexistent. 4. Data and variables Almost all previous studies of the micropolitical consequences of informality are based on anecdotal impressions or, at best, surveys in which the user treats class or income as a proxy for informality. In general, there has been a long history of indifference toward precisely measuring workers’ informality/formality status in public opinion surveys about politics.6 Fortunately, this has recently changed in some circles, affording us the opportunity to present evidence based on nationally representative samples. We conduct secondary analyses of Latin American Public Opinion Project (LAPOP) data in 18 countries as well as analyses of original questions administered in two elections studies: Brazilian Electoral Panel Study 2014 (BEPS 2014; Ames et al., 2014) and the Argentine Panel 5 Among working couples with incomplete primary or at least some tertiary education, the rates are lower, around 25% (p. 827). 6 See Holland, 2017 (chapter 2) and Singer, 2016 for recent exceptions.

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

Election Study 2015 (APES 2015; Lupu, Gervasoni, Oliveros, & Schiumerini, 2015). We convey the basics here, but our informality status measures are described in greater detail in part I of the Supplemental information. In 2006 and 2008, LAPOP ran batteries in most countries that enable measurement of informal work using the most useful operational definition of the term. Designers of economic household surveys have implemented different measures of the concept, but the most widely accepted is now the ‘‘legalistic/social protection” definition (Holland, 2017; Perry et al., 2007), which defines workers as informal ‘‘if their employment relationship is not subject to standard labour legislation, taxation, social protection, or entitlement to certain employment benefits” (ILO, 2002, p. 126).7 Strictly speaking, the legalistic and social protection criteria are conceptually distinct. Under the legalistic criterion, employment is informal if not backed by a government-recognized and legally compliant labor contract, and microentrepreneurship is informal if one’s firm is not registered and tax compliant. Under the social protection criterion, work that does not carry state-administered social benefits is informal. In practice, there is a heavy degree of overlap between these two criteria in Latin America. Because of the Bismarckian tradition, a formal employment contract brings with it enrollment in the state-run social security system, which includes retirement pensions, health insurance, and (depending on the country) other benefits such as unemployment insurance and a housing allowance. With all of this in mind, the LAPOP batteries allow us to define the three primary classes of workers that correspond to those used by the World Bank (Perry et al., 2007). We code formal wage earners as those enrolled in a payroll-based benefit plan. LAPOP in 2006 asked whether respondents had seguro social, a phrase that translates as ‘‘social security” but refers in Latin America to the formal-sector benefits plans. In 2008, LAPOP asked if respondents had social security or health insurance through their employer, with some variation (reported in the Supplemental information) by country. Informal wage earners are respondents who answered ‘‘no” to this question and who said they were ‘‘salaried” in the public or private sector. Finally, informal self-employed workers are self-declared ‘‘own-account workers” without payroll-based benefits. It would be ideal to ask if the firms of these individuals, most of them microentrepreneurs, are licensed and registered with proper authorities. LAPOP did not contain such a question, but ancillary analyses we conducted showed that virtually all of the microfirms run by respondents in this category are informal.8 In one set of analyses, we combine the two types of informal respondents into a single informal workers group, and we report the effects of this category relative to the omitted baseline category of formal wage earners. This approach tests the characterizations of informality writ large made by previous scholars. We also report a second set of results from specifications that break informality into its two constituent parts,9 still treating formal wage earners as the omitted baseline. For various reasons, informal entrepreneurs may 7 Our measure improves on two other means of defining the informal sector that have been discredited. The ‘‘self-employment” definition used by some political scientists counts only microentrepreneurs as informal (Altamirano & Wibbels, 2012; Berens, 2015), thereby misclassifying the informal wage earners that actually outnumber microentrepreneurs in most Latin American countries (Perry et al., 2007). The ‘‘productive” definition, which Gasparini and Tornarolli (2009) consider to be ‘‘theoretically weak” (p. 20), tallies small firms and unskilled workers as informal, regardless of whether their work is registered and monitored by the state. 8 In particular, LAPOP in 2006 also asked respondents to report the size of the firm in which they worked. Ninety-two percent of own-account workers reported firms of size four (employees) or less, and we know from household surveys that almost 90% of these small firms are unregistered (Perry et al., 2007, p. 25). Similarly, in the APES 2015 battery that we designed, only 18% of own-account workers said their firm was registered with authorities, and perhaps even this number is exaggerated by social desirability bias. 9 This option is not available in BEPS 2014.

173

be hypothesized to act differently from informal wage earners in the political sphere, so we do break these out at the cost of some statistical power. Also, we always include separate dummy variables for non-EAP or unemployed10 citizens and for formal business owners and managers. Given its relative novelty and its centrality to our argument, we conduct two measurement validity checks on the LAPOP measure of informality. These are reported in part II of the Supplemental information. They show that our measure has a high degree of convergent validity (i.e., it is highly correlated at the country level with household survey measures of informality) and criterion-related validity (i.e., it follows known patterns with respect to age). In the Argentina and Brazil single-country studies, the means of classifying workers is different. In Argentina, the battery is one we designed and wish to see employed more frequently in public opinion surveys. (For this reason, this battery is shown first in the Supplemental information.) Wage-earning employees were asked if they made payroll contributions to a pension plan, so this question bifurcates formal from informal wage earners. Own account workers were asked if their activities were registered with the proper tax authority (Administración Federal de Ingresos Públicos), so this question bifurcates informal self-employed from formal business owners. In Brazil, formal wage earners are defined as those with a carteira assinada (signed workers booklet that is proof of private sector contract and social security benefits) or an estatutário (civil service contract). Throughout, we control for four factors that are largely causally prior to informality status: Education, gender (Woman), Age, and Urbanicity of residence. To avoid post-treatment bias, we keep the models trim and do not control for variables that are potentially endogenous to informality, such as wealth, income, or ideology. To test the various assertions and hypotheses mentioned above, we walk through a set of models with different dependent variables, starting first with measures of political participation, organizational engagement, and partisanship to test the undermobilization hypothesis. We then use two different measures of vote choice to test the right-leaning hypothesis, and we follow this with an analysis of some economic issue items to assess the effects of informality on policy attitudes. (Wordings and codes for all variables are in part III of Supplemental information.) 5. Results We present the most important results graphically, although all coefficients are reported in tables in part IV of the Supplemental information. For each dependent variable, we report two figures: one plots the coefficients and 95% confidence intervals for the informal workers dummy variable, and the other plots the coefficients and 95% confidence intervals for the informal wage earners dummy variables and the informal self-employed workers dummy variables separately. For each of these specifications, we estimated one model per country and, with hierarchical modeling techniques, a single regionwide model. (We also run these regionwide models using matching techniques, and these results are reported in part V of the Supplemental information.) Both country and regionwide models pool the two survey years. Although it should go without saying, we wish to stress that we will interpret substantive, and not just statistical, significance. (a) Tests of the undermobilization hypothesis To see if informal workers are less politically mobilized and socially organized than formal workers, we run models with four 10 We combine non-EAP and unemployed out of necessity. LAPOP 2006 lacks a good measure of unemployment.

174

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

different dependent variables. The first is a Political participation index, which is an additive index used by Boulding and Holzner (2016) of several binary items asking respondents if they engaged in different forms of political participation (e.g., protesting, campaigning). We limit this to non-voting forms of participation so that we can treat voting as a case apart. Second, we run a separate set of models with Vote turnout as the dependent variable using a measure of whether the respondent self-reported as having voted in the most recent presidential election. We treat voting separately because of its singular importance and relatively high frequency but also because many Latin American countries have compulsory voting; one might expect turnout behavior in these countries to be less stratified by social categories than are other forms of participation. The third is Partisanship, which is a binary measure of whether the respondent reports having a partisan identity. The final is an Organizational participation variable (again from Boulding & Holzner, 2016), an ordinal measure of how frequently respondents attended non-religious group meetings.11 Figure 1 reports negative binomial coefficients for the political participation index models. Figure 1 exemplifies the graphical scheme we use throughout. In the top panel, there is one coefficient per country (that for the informal workers category), and these are presented with their 95% confidence intervals as thin black lines and with three-letter ISO 3166 country-code labels on the x-axis. The regionwide coefficient, estimated with a hierarchical model, is depicted with its 95% confidence interval as a thick black line and ‘‘Region” label on the x-axis. In the lower panel, there are two coefficients per country (those for informal wage earners (‘‘we”) and those for informal self-employed workers (‘‘se”)), and then two for the region as a whole. In both figures, the omitted baseline category is formal wage earners; in other words, coefficients are effects relative to formal wage earning. The results reported in Figure 1 do not support the undermobilization hypothesis. In the top panel, informality has a negative effect in 10 of 18 countries, but it is statistically significant in only three of these ten. The confidence interval for the region as a whole is decidedly centered at zero. Parsing out informal workers into wage earners and self-employed yields only three statistically significant negative differences. By comparison, the effect (shown in part IV of the Supplemental information) of education regionwide is statistically significant and, moving from 5th to 95th percentile, yields a change of two-fifths of a standard deviation. Figure 2 repeats the exercise for the binary measure of voter turnout. The x-axis labels are suffixed by ‘‘(C)” in countries that had compulsory voting with sanctions at the time.12 The effects of informality show slightly more signs of life for this turnout dependent variable. Informal worker status has a negative and statistically distinguishable effect on voter turnout in 5 of 18 countries, and the regionwide coefficient is also negative and statistically significant. However, the substantive effects are small. Formality boosts the (regionwide) propensity to vote by just 2.3 percentage points. This is about .05 of a standard deviation and just a 10% decrease over the baseline probability (23.7% in the sample) of self-reported abstention. By comparison, the effect of not being in the EAP is five times as great at 11.5 percentage points. Parsing by the two types of informal workers (bottom panel) sheds light on a few more negative and statistically significant relationships—informal wage earners vote at a lower rate than self-employed ones. Again, however, statistical significance occurs in just a minority (10 of 36) of instances, and substantive effects are small. Finally, compulsory voting does not moderate the effect of informality. 11 The precise list of group types is neighborhood, community, professional, and women’s. 12 Compulsory voting data are from Payne, Zovato, Carrillo Flórez, and Allamand Zavala (2003, 55).

Figure 1. The impact of informality on political participation: evidence from 18 Latin American countries.

Figure 3 looks at partisanship as a dependent variable, testing the hypothesis that informal workers have more volatile voter preferences and are less anchored to political institutions. The top panel shows that this is borne out in just three countries. The regionwide coefficient shows a negative and statistically significant effect, but again the size of this effect is small (2.5 percentage points, or 1/17th of a standard deviation). Only a few more statistically significant negative effects come to light when parsing informality into its two constituent parts with, again, informal wage earners a touch less likely to be partisans than the informally self-employed. Finally, Figure 4 shows results for the organizational participation dependent variable. The effect of informality is negligible here. It is virtually zero even when statistical power is high in the regionwide model, and it is only rarely distinguishable from zero—regardless of how informality is measured—in the countrylevel models. We also conduct similar analyses using the election studies from Argentina (APES 2015) and Brazil (BEPS 2014). Although obviously much narrower in geographic scope, these present the advantage of being true election studies with respondents asked about voting behavior that occurred (at most) a few weeks earlier. Aside from constructing a binary turnout variable for both countries, we also construct dependent variables of partisanship and (in Argentina) participation in organizations. For Brazil, we exploit our panel data to measure Vote stability, which is whether respondents changed (=0) or did not change (=1) their vote intentions between candidates during the last month of the campaign. Preference volatility during campaigns potentially betrays the

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

Figure 2. The impact of informality on voter turnout: evidence from 18 Latin American countries.

weakness of socioeconomic roles in anchoring individuals to parties and platforms. Table 2 reports the results, and, across the six regressions, we reject only one null hypothesis about the effects of informality. The one statistically significant effect (on turnout in Argentina) is about 1/6th of a standard deviation. (b) Tests of the right-leaning hypothesis To gauge voter preferences in LAPOP, we first use Ideology of vote choice. Unfortunately, LAPOP data collection usually does not occur amidst election campaigns, so we must rely on survey questions that ask respondents how they voted in the most recent presidential election. Given the imperfections of human recall, we limit the analysis to LAPOP country years in which an election occurred at most two years previously. Answers were converted to the left M right ideology scores of the candidate with the WiesehomeierBenoit dataset, which used an expert survey to score Latin American parties on a 1 (left) to 20 (right) ideological dimension (Wiesehomeier & Benoit, 2009). Non-voters and other unvalenced responses, such as ‘‘can’t remember,” are dropped. The top panel of Figure 5 shows informality to have, regionwide, no statistically significant effect on the ideological direction of vote choice. In individual election years, its effect is statistically significant in only 3 of 12 instances. In two of these (Brazil 2006, Mexico 2006), informal-sector workers were actually more leftist, contra the right-leaning hypothesis. Only in one instance (Nicaragua 2006) were informals more right-leaning. The bottom panel

175

Figure 3. The impact of informality on partisanship: evidence from 18 Latin American countries.

reveals no additional statistically significant relationships. (We also run these as nominal choice models, the substantive results of which are identical. Coefficients are reported in part VI of the Supplemental information.) We also conducted similar analyses using the election studies from Argentina 2015 and Brazil 2014, again taking advantage of surveys that occurred during or just after campaigns. Both races were largely three-candidate affairs, with each offering at least one left-of-center and one right-of-center option. In Argentina, Daniel Scioli of the incumbent Peronist faction Frente para la Victoria was the clear leftist option, and eventual winner Mauricio Macri of the Propuesta Republicana party was the clear rightist option. Third-place finisher Sergio Massa belonged to a more centrist Peronist faction, so the election posed a straightforward leftist, centrist, and rightist option. In Brazil, incumbent and eventual winner Dilma Rousseff of the leftist Partido dos Trabalhadores faced Aécio Neves of the center-right Partido da Social Democracia Brasileiro as her main challenger. The third-place finisher, Marina Silva, was also a left-of-center candidacy, although WiesehomeierBenoit scores her nominal party (Partido Socialista Brasileiro) slightly to the right of Rousseff’s. Table 3 reports estimates from four different statistical models: one ordinary least squares (OLS) regression per country that treats as the dependent variable the ideological score of the respondent’s vote choice (similar to the models underlying Figure 5 above) and one multinomial logit (MNL) per country that treats the leftmost candidate (Scioli and Rousseff, respectively) as the base category

176

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

those used to generate the market policy attitudes index used above, of individual policies (Krosnick, 1991)). In Argentina, we asked respondents which program they wanted the government to raise spending on: Programa Familias13 (a conditional cash transfer program) or Administración Nacional de la Seguridad Social (ANSES, the formal sector social security program): Now let’s talk about social policies. Do you want the government to raise benefits for Programa Familias, which is given to the poor, or do you want the government to raise benefits for ANSES, which benefits only formal sector workers? Or do you not want the government to raise social spending? (1) Programa familias (33%). (2) ANSES (41%). (3) Not raise social spending (27%). In Brazil, we asked respondents which kind of program they preferred—means-tested programs like Bolsa Família (the main conditional cash transfer program) or contributory programs like the Institute Nacional de Seguro Social (INSS, the formal sector social security program): Now let’s talk about social policies. Some types of programs, like Bolsa Família, use resources from taxes paid by everybody to benefit some people of low income. Other types of programs, like the INSS, use resources from taxes paid by those with a signed workers booklet and benefit only those who pay. Which type of policy do you prefer? (1) Policies that benefits some with taxes paid by everyone, like the Bolsa Familia. (43%). (2) Policies that benefit those who pay, like the INSS. (30%). (3) Both or indifference or neither [Not read] (27%).

Figure 4. The impact of informality on participation in organizations: evidence from 18 Latin American countries.

of the dependent variable. The informality coefficient shows signs of life in the nominal choice model for Argentina, where informal workers were (about 10 percentage points) more likely to vote for Scioli than Macri. In Brazil, however, the estimated effect of informality on the main Neves versus Rousseff cleavage is virtually nil. One purported mechanism underlying the right-leaning hypothesis is that informal workers should have economic policy attitudes that are more individualistic and pro-market. To test for this, we used the factor scores generated from the first dimension of a factor analysis conducted on four different variables. Each variable asked respondents to report their beliefs (on a seven-point Likert scale) about the proper role of the state in the economy. Responses to all four showed a strong loading on a single dimension, and each respondent’s scoring on this dimension is the Market policy attitudes index, which ranges from statist (low values) to promarket (high values). Figure 6 shows there to be virtually no effect of informality on attitudes toward market policies. (c) Tests of the noncontributory preference hypothesis We designed survey questions that forced respondents to choose between this tradeoff of formal-sector versus meanstested social welfare programs and placed them on the BEPS 2014 and the APES 2015. Although scholars have studied social policy preferences in Latin America, they are generally not in the habit of forcing respondents to make tradeoffs among potentially competing policy options and priorities. (This wording technique is known to be superior to the more commonly used ratings, like

We estimated multinomial logits for each of these. We also run two other models using as dependent variables survey measures of support for certain social policies. In Argentina, we use an ordinal measure of Support for the Asignación Universal por Hijo (AUH), a monthly allowance paid to informal workers with children. Informality is an eligibility criterion for AUH, so one would surely expect informal workers to support it more enthusiastically. In Brazil, we create a Means-test index, which is an index created from five binary questions asking for approval or disapproval of different anti-poverty social policies. Higher values indicate more support for means-tested policies. Table 4 reports the results. The most important hypothesis tests are found in the first column and first row of the two MNL models. These coefficients indicate whether informal workers tilt toward means-tested programs when posed a direct tradeoff between them and formal-sector social security schemes. In both countries, we fail to reject the null. Informality status has no statistically significant effect despite a particularly large N in Argentina. Somewhat oddly, informal workers were more likely to support Programa Familias over no rise in social spending, but not over rises in ANSES spending. Also, informality status has no effect on support for the AUH in Argentina—this despite the fact that only informal workers are eligible for it. Informality shows signs of life in explaining attitudes toward the battery of means-tested programs in Brazil. Here the effect is statistically significant and about onefifth of a standard deviation in size.

13 The Programa Familias name was used by mistake. At the time of the survey, programa familias por la inclusion social was a defunct predecessor to the conditional cash transfer program that was active, Asignación Universal por Hijo (AUH). In our defense, familias was described to respondents as a program for the poor. More importantly, the effect of informality is no different even when respondents were asked about AUH. (See Table 4 below).

177

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182 Table 2 The impact of informality on turnout, partisanship, participation in organizations, and vote stability: Argentina 2015 and Brazil 2014 presidential elections Argentina 2015

Informal workers Formal wage earners Formal business owner Non-EAP Education Woman Age Urbanicity Constant N

Brazil 2014 Turnout

Partisanship

Participation in orgs.

.399* (.190) Omitted baseline .146 (.343) .440* (.185) .232* (.044) .018 (.114) .007 (.004) .169 (.164) 1.306 (.352) 1,274

.013 (.121) Omitted baseline .325 (.208) .118 (.110) .002 (.032) .110 (.087) .006* (.003) .178 (.144) .465 (.254) 1,106

.099 (.115) Omitted baseline .272 (.181) .110 (.105) .147* (.035) .096 (.091) .000 (.003) .051 (.153)

Informal workers Formal wage earners

Partisanship

Vote stability

.034 (.086) Omitted baseline

.155 (.169) Omitted baseline

.074 (.175) .068* (.029) .243 (.132) .006 (.005) .367* (.172) .320 (.424) 1,159

.040 (.085) .023 (.013) .217* (.066) .010* (.002) .109 (.104) 1.009 (.209) 3,004

.103 (.169) -.010 (.029) .048 (.135) .014* (.005) .325 (.200) .730 (.446) 701

Formal business owner Non-EAP Education Woman Age Urbanicity Constant

1,864

Turnout .134 (.182) Omitted baseline

N

Note: * = p < .05. Entries are probit coefficients and standard errors in parentheses. Source: BEPS 2014 and APES 2015.

differences in the commonly hypothesized directions that, nonetheless, are of minor substantive importance. They are certainly not of a magnitude that would support what some scholars attribute to them, such as the difference between a center-right leaning political system and a center-left leaning one (Huber & Stephens, 2012; Schneider & Soskice, 2009). In the vast majority of instances, we find no statistically significant differences whatsoever. Our conclusion is that scholars have overblown the political differences between the formal and informal sectors.

6. Explanations for the null findings

Figure 5. The impact of informality on ideology of vote choice: evidence from 12 Latin American elections.

We take this collection of findings as evidence that the gap in political attitudes and behaviors between formal and informal sector workers is quite small. At most, we find statistically significant

The reason, we think, lies in the insights of the revisionist model, which has found the formal and informal sectors to be more integrated than was long thought. Of course, our results thus far have not shown this directly, and in fact demonstrating the counterfactual—that a truly dualistic labor market would produce sharp political differences between formal and informal workers—is difficult since the reality in Latin America is the revisionist one. For example, the social policy preferences of most workers are surely shaped by the fact that they estimate the chance of a future switch to or from informality to be relatively high (Carnes & Mares, 2016). Still, we can think of at least two possible counterfactual tests. One is to treat individuals who do not cross the threshold between informality and formality as more akin to how individuals would be in a dualistic labor market than individuals who do cross the threshold. Because of its panel structure, APES 2015 does provide some ability to differentiate among Persistent informal workers (who were informal in both panel survey waves), Persistent formal workers (who were formal in both waves), and Switchers (who switched between informality and formality between waves). Table 5 reestimates the MNL model of support for Programa Familias (from Table 4) with these three categories as independent variables. When tallied only for persistent informals, the coefficient on informality is now more than twice its size in Table 4, and the effect relative to switchers is statistically significant. (With respect to persistent formal workers, the effect is almost so at p = .074). To be clear, we see these results as suggestive at best because of the relatively short time frame, the small N (because we must collapse over wave), and the lack of a measure of expectations about future job changes.

178

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

Table 3 The impact of informality on vote choice: Argentina 2015 and Brazil 2014 presidential elections Argentina 2015

Brazil 2014 OLS

Informal workers Formal wage earners Formal business owner Non-EAP Education Woman Age Urbanicity Constant N

.578 (.355) Omitted baseline .464 (.641) .404 (.330) .415* (.095) .111 (.260) .005 (.008) .072 (.431) 10.718 (.778) 1,147

Multinomial Logit

OLS

Macri Scioli

Massa Scioli

.427* (.201) Omitted baseline .332 (.362) .291 (.186) .282* (.056) .126 (.151) .003 (.005) .071 (.271) .855 (.460) 1,102

.370 (.331) Omitted baseline .385 (.590) .033 (.307) .133 (.098) .354 (.250) .009 (.008) .427 (.345) 1.495 (.716)

Informal workers Formal workers

Non-EAP Education Woman Age Urbanicity Constant N

Multinomial Logit Neves Rousseff

Silva Rousseff

.063 (.325) Omitted baseline

.016 (.247) Omitted baseline

.279 (.269) Omitted baseline

.145 (.328) .159* (.054) .120 (.255) .015 (.009) .589 (.339) 6.060 (.800) 1,010

.133 (.257) .130* (.043) .082 (.198) .011 (.007) .629* (.304) 2.716 (.665) 985

.115 (.261) .054 (.045) .091 (.218) .002 (.008) .691 (.400) 1.798 (.682)

Note: * = p < .05. Source: BEPS 2014 and APES 2015.

A second counterfactual could lie in observing countries with more genuinely dualistic labor markets. For example, switches between informality and formality are rare in India, where 80– 85% of the labor force is informal (Schneider & Enste, 2013, pp. 58–9). Again, surveys that contain measures of both informality status and political attitudes are rare, yet cutting edge studies do find informal workers in India to be more clientelistic and promarket than are formal workers (Milner & Rudra, 2017; Wibbels, 2017). Although further research is needed, we think alternative explanations for our null results are less convincing. We consider three. (a) Timing

Figure 6. The impact of informality on market policy attitudes: evidence from 18 Latin American countries.

Our results may not support the conventional wisdom about informality because the impressions of informals as undermobilized and right-leaning are a product of an earlier era in Latin American politics. The results presented above are from 2006 or later, whereas some of the scholarly impressions about informality are based on observations made before (and in some cases decades before) 2006. Although rarely (if ever) tested with survey data, these impressions may have been empirically accurate at the time, with important political, economic, and social changes in the new millennium whittling away at the previous political tendencies of informal-sector workers. One such change could be the rise of new forms of collective action and political participation. Scholars have alleged a proliferation, since democratization, of urban associations of low-income individuals articulating a variety of demands and featuring minimal linkages to unions, parties or the state (Collier & Handlin, 2009). Similarly, new social movements have organized around gender, sexual, and ethnic identities (Alvarez, 1990; Van Cott, 2005). Urban political protest has also been common, spurred on by spontaneous outrage and organizations that have little to do with the labor movement (Boulding, 2014). Finally, the new noncontributory welfare programs have themselves encouraged collective political action (Garay, 2007; Handlin, 2013). All told, the

179

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182 Table 4 The impact of informality on social policy attitudes: Argentina 2015 and Brazil 2014 presidential elections Argentina 2015

Brazil 2014 Ordered probit Support for AUH

Informal workers

Constant

.069 (.104) Omitted baseline .061 (.162) .048 (.091) .053* (.027) .104 (.074) .003 (.002) .082 (.120) NA

N

1,119

Formal wage earners Formal business owner Non-EAP Education Woman Age Urbanicity

Multinomial Logit Programa Famılias ANSES

Programa Famılias Neither

.306 (.164) Omitted baseline .673* (.317) .272 (.155) .175* (.048) .011 (.119) .013* (.004) 1.264* (.198) 1.660 (.365) 2,336

.442* (.181) Omitted baseline .260 (.363) .208 (.168) .346* (.052) .003 (.134) .015* (.004) 1.027* (.211) 2.430 (.386)

Informal workers Formal workers

Non-EAP Education Woman Age Urbanicity Constant N

OLS

Multinomial Logit

Means-test index

Bolsa Famılia INSS

Bolsa Famılia Both

.214* (.074) Omitted baseline

.238 (.246) Omitted baseline

.271 (.238) Omitted baseline

.094 (.072) .120* (.011) .042 (.054) .013* (.002) .337* (.056) .930 (.153) 3,058

.370 (.246) .080* (.039) .423* (.204) .025* (.007) .732* (.370) 3.123 (.607) 1,071

.591* (.235) .032 (.042) .003 (.186) .012 (.007) .213 (.267) 1.076 (.622)

*

Note: = p < .05. Source: BEPS 2014 and APES 2015.

Table 5 The impact of stability in informality status on social policy attitudes: Argentina 2015 Multinomial Logit

Persistent informal workers Persistent formal workers Switchers Non-EAP Education Woman Age Urbanicity Constant N

Programa Famılias ANSES

Programa Famılias Neither

.855* (.424) .170 (.444) Omitted baseline .819* (.370) .196* (.083) .007 (.210) .009 (.006) 1.440* (.347) 1.215 (.649) 697

.091 (.461) .242 (.395) Omitted baseline .228 (.379) .112 (.077) .282 (.228) .006 (.007) 1.040 (.623) 2.430 (.386)

Note: * = p < .05. Source: APES 2015.

formal/informal civic engagement gap may have disappeared in Latin America because, when it comes to spurring collective political activity, organized labor is no longer the only game in town. Change to the welfare state may have also reshaped policy attitudes and voting cleavages. One interpretation is that, since many left-of-center governments introduced or expanded the new means-tested programs, informal workers finally had an immediate programmatic reason to vote for the left (Huber & Stephens, 2012, pp. 237, 247). For example, in Brazil and Venezuela, these social programs may have created such a realignment that the poor and informal are now disproportionately supportive of the left (Handlin, 2013; Hunter & Power, 2007; Zucco, 2008; Mainwaring et al., 2015). Although seemingly grounded in recent trends and scholarly findings, evidence for this explanation of the null findings is weak.

In other words, we suspect that the alleged political gaps between the informal and formal sectors have always been rather minimal. Unfortunately, the time series data needed to track associational memberships and voting behavior over the decades in Latin America are unavailable, yet one cannot help but notice that the language about an alleged increase in associational life among the urban, informal poor echoes similar arguments made four decades ago. Most famously, Janice Perlman (1976) debunked the ‘‘myth of marginality” about Brazil’s favelas by saying, among other things, that they ‘‘have an extremely rich associational life” and documenting their relatively high rates of political participation (p. 132). Similarly, de Soto (1989) argued in the 1980s that associations sprang up in urban slums precisely in response to informality—as a means to increase compliance with informal contracts (p. 167). (See also Thornton, 2000 and Stokes, 1995.) As for differences in political leanings, in recent years it has not always been the left that introduced noncontributory programs. In fact, most scholars find the governing party’s ideology to be a weak predictor of whether a country adopts means-tested or universalistic social programs (Brooks, 2015; Haggard & Kaufman, 2008; Huber et al., 2008). Our results from Argentina and Brazil are particularly telling; these are two most-likely cases for finding an effect of informality on social policy attitudes and voting behavior. In Argentina, only informal workers are eligible for AUH, and in Brazil, only the poor are eligible for Bolsa Família. However, recall from Table 4 that informality had no effect on support for either, even when posed as a tradeoff with strictly formal sector programs. This seems to undermine the claim that new social policies have closed a pre-existing gap or opened a new gap in attitudes and voting behavior between formal and informal workers. As an even more direct empirical test, we consult a different survey that was conducted in one country before the nationwide implementation by the left of a major means-tested program. The first wave of the Brazilian four-city survey (Baker, 2009) occurred in spring of 1999. This was soon after the October 1998 re-election of center-right president Fernando Henrique Cardoso (FHC), who defeated leftist and second-place finisher Luiz Inácio Lula da Silva (Lula) and third-place finisher and centrist Ciro Gomes. Lula eventually won the presidency in 2002 and subse-

180

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

Table 6 The impact of informality on vote choice and economic policy attitudes: Brazil 1999 Vote choice OLS

Informal worker Formal worker Non-EAP Education Woman Age Constant N

.409 (.286) Omitted baseline .046 (.261) .092 (.072) .057 (.199) .124 (.288) 10.799 (1.205) 800

Market policy attitudes Multinomial Logit

OLS

FHC Lula

Gomes Lula

.385 (.271) Omitted baseline .045 (.232) .119 (.062) .063 (.191) .183 (.263) .262 (1.141) 574

.656 (.580) Omitted baseline .278 (.519) .052 (.123) .864* (.415) 1.218* (.578) 7.595 (2.549)

.093 (.108) Omitted baseline .069 (.094) .075* (.023) .283* (.075) .114 (.097) 2.106 (.449) 800

Note: * = p < .05. All models also include city fixed effects that are not reported. Source: Four-city (Belém, Porto Alegre, Recife, São Paulo) Brazil survey, 1999 wave.

quently scaled up the Bolsa Família conditional cash transfer program, a move that, to many observers, rebalanced his constituency toward lower class voters (Hunter & Power, 2007; Zucco, 2008). Fortunately, this 1999 survey, which did contain the rare measure of formality (‘‘Do you have a signed workers card?”), gets before this. We ran three models, all presented in Table 6. As dependent variables for two models of vote choice, we use respondents’ reports of how they voted in the 1998 presidential election, which occurred about 6 months prior (scaled from left to right using Wiesehomeier-Benoit for an OLS model and nominally for an MNL model). The third dependent variable is a multi-item index of support for market-friendly policies and trends (privatization, international trade, and foreign investment), scaled so that low values reflect unfavorable opinions and high values reflect favorable ones. Table 6 shows there to be no statistically discernable effect of informality on vote choice or issue attitudes. All told, the counterargument that informals used to be more right-leaning does not pass scrutiny.

(b) Absence of self-interest A second possibility for the null findings is that scholarly expectations of the effects of informality are excessively structuralist, which is tantamount to saying that they overestimate the impact of self-interest on mass political behavior. Social scientists and other observers are long on theories that attribute mass public opinion and voting behavior to individuals’ self-interest, and most of the hypotheses inspired by the dualist vision assume that informal workers act politically the way they do because of their economic interests. Yet corroborating evidence is often surprisingly short (Dion & Birchfield, 2010). Numerous studies of political psychology and behavior in the U.S. find economic self-interest to have at best a minor effect (Sears, Lau, Tyler, & Allen, 1980). Many scholars of Latin American voting behavior are skeptical that citizens vote according to policy opinions (i.e., positional issue voting) or that parties are even programmatic enough to make that possible (Kitschelt, Hawkins, Luna, Rosas, & Zechmeister, 2010; McCann & Lawson, 2003). In fact, Latin American voters change their candidate preferences with a high degree of frequency during campaigns, seemingly belying the notion that votes are anchored to economic structures and interests (Greene, 2011).

We find evidence, however, that casts doubt on the notion that we can attribute the null findings to the arational forces that drive policy attitudes and voting behavior. In particular, wealth—excluded from our models yet fundamental to policy and selfinterest—matters for voting behavior and for policy attitudes. For example, in Argentina and Brazil, poor voters are far more likely to support means-tested programs than are wealthy individuals (Meltzer & Richard, 1981). If we add Household wealth to the models in Table 4 (see part VII of Supplemental information for these results), there is roughly a 20 percentage point difference in (and doubling of) the probability of supporting the means-tested programs (Bolsa Família or Programa Familia) between those at the 5th and 95th percentile of wealth. The latter are instead supportive of state spending on formal-sector programs. Moreover, there are corresponding wealth-based cleavages in electoral choice in these two countries (Zucco, 2008). (See Table SI.13 in Supplemental information). At least on this front, self-interest appears to be alive and well, and, in separate research, scholars have found Latin American elections to contain ample instances of positional issue voting (Baker & Greene, 2011). It is just that apparent and immediate self-interest is not alive and well with respect to one’s current formality status.

(c) Measurement error We must finally consider the bugbear that plagues all nullfinding claims: poor measurement. Measurement error attenuates effect sizes, so our substantively small results could arise from inaccurate classifications of respondents’ worker status. Moreover, some have argued that informal work is more of a continuum than a binary distinction (ILO, 2013). Recall, however, that our LAPOP measures have high convergent and criterion-validity. Moreover, the possession of a signed workers booklet—as asked in Brazil—carries a straightforward, one-to-one correspondence with the most standard definitions of formality, and whether one has a signed booklet or not is of great salience in Brazilian workers’ lives: ‘‘The existence of [the carteira de trabalho] permits an easy empirical separation of workers with formal labor contracts that must comply with the labor laws from workers with informal labor contracts that are not under this legislation” (Barros & Corseuil, 2004, 329). Finally, our null results are backed by other scholars with valid measures. Holland (2017, Table 2.2) uses an ordinal measure

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

of informality in Colombia and finds, as we do, that informal workers are not more favorable toward tax-and-transfer redistribution. (See also Berens, 2015). We do encourage future study and refinement of informality measures in public opinion surveys, but we doubt that changes in measurement will dramatically increase the estimated effects of informality. 7. Summary and conclusion Scholars of Latin American politics have long been interested in the structural, socioeconomic roots of citizens’ political preferences and behaviors (O’Donnell, 1978). One of the long-held impressions in this research is that informal workers are more socially atomized, less participatory, and more right-leaning than their formal sector (and thus sometimes unionized) counterparts. A newer impression holds that they have different social policy preferences. In this paper, we present survey evidence on a wide range of political behaviors and find these impressions to be largely false. On average, informal sector workers are slightly less partisan and less likely to vote, but the gap in behavior is small. In terms of community involvement, non-voting forms of political participation, vote choice, and market-policy attitudes, no gap exists. In Argentina and Brazil, informal workers are sometimes more likely to favor means-tested over payroll-based social programs. However, this evidence is mixed at best, which is especially surprising given how payroll-based programs discriminate (by definition) against informals. We thus cast doubt on these longstanding impressions of political attitudes and behavior among informal workers. Moreover, in grounding our findings in the revisionist view of informality, we also cast doubt on the stableness of some socioeconomic roles and structural designations in Latin America. To be sure, social mobility in the region is low, making wealth distinctions slow to change (Azevedo & Bouillon, 2009). However, rates of job churning are high (IDB, 2004), with some of this churn occurring between formality and informality. As a result, occupations and jobs provide relatively unstable roles and interests, thereby weakening the extent to which they shape political thought and activity (Baker, 2009). Finally, with this study we also issue a clarion call for scholars to make informality/formal status a standard measure in politically oriented public opinion surveys, particularly those conducted in less developed countries. This may seem contradictory, since we find its effects on a host of important political attitudes and behaviors to be minimal in Latin America. Yet our findings do not at all preclude the possibility that informal workers, even amidst the prospect of job churning, are attuned to political issues or even behaviors not covered here and not part of the traditional left– right spectrum (e.g., Holland, 2017). Acknowledgment We thank Pulapre Balakrishnan, Carew Boulding, Jeronimo Carballo, Stephen Chaudoin, Timothy Hellwig, Alisha Holland, Keith Maskus, and Stephanie Rickard for comments on earlier drafts. We also thank the Latin American Public Opinion Project (LAPOP) and its major supporters (the United States Agency for International Development, the United Nations Development Program, the Inter-American Development Bank, and Vanderbilt University) for making its data available (www.LapopSurveys.org). Appendix A. Supplemental information Supplemental information associated with this article can be found at https://doi.org/10.1016/j.worlddev.2017.09.014.

181

References Altamirano, Melina, and Wibbels, Erik (2012). ‘‘Economic Informality and Electoral Democracy”. Paper presented at the 2012 Annual Meeting of the Midwest Political Science Association, Chicago, IL. Alvarez, Sonia (1990). Engendering democracy in Brazil: Women’s movements in transition politics. Princeton: Princeton University Press. Ames, Barry, Huberts, Alyssa, Machado, Fabiana, Renno, Lucio R., Samuels, David, & Smith, Amy Erica (2014). Brazilian Electoral Panel Study: 2014 Results. Washington, DC: IADB Working Paper. Ansell, Ben W., & Samuels, David J. (2014). Inequality and democratization: An elitecompetition approach. New York: Cambridge University Press. Azevedo, Viviane, and Bouillon, César P. (2009). ‘‘Social Mobility in Latin America: A Review of Existing Evidence”. Research Department Working Papers. Interamerican Development Bank: Washington, DC. Baker, Andy (2009). The market and the masses in Latin America: Policy reform and consumption in liberalizing economies. New York: Cambridge University Press. Baker, Andy, & Greene, Kenneth F. (2011). The Latin American left’s mandate: Free market policies and issue voting in new democracies. World Politics, 63(1), 43–77. Banerjee, Abhijit V., & Duflo, Esther (2011). Poor economics: A radical rethinking of the way to right global poverty. New York: Public Affairs. Barros, Ricardo Paes, & Corseuil, Carlos Henrique (2004). The impact of regulations on Brazilian labor market performance. In James J. Heckman & Carmen Pagés (Eds.), Law and employment: Lessons from Latin America and the Caribbean (pp. 273–350). Chicago: University of Chicago Press. Berens, Sarah (2015). Preferences on redistribution in fragmented labor markets in Latin America and the Caribbean. Journal of Politics in Latin America, 7(3), 117–156. Blofield, Merike (2011). ‘‘Introduction: Inequality and Politics in Latin America”. In: Merike Blofield and Juan Pablo Luna, eds., The Great Gap: Inequality and the Politics of Redistribution in Latin America. University Park, PA: Pennsylvania State University Press, pp. 1–18. Boulding, Carew (2014). NGOs, political protest, and civil society. New York: Cambridge University Press. Boulding, Carew, and Holzner, Claudio (2016). ‘‘Community Organizations and Strategic Mobilization: Explaining the Puzzle of Poor People’s Participation in Latin America”. University of Colorado, Working Paper. Brooks, Sarah M. (2015). Social protection for the poorest: The adoption of antipoverty cash transfer programs in the Global South. Politics and Society, 43 (4), 551–582. Cameron, Maxwell. A. (1991). Political parties and the worker-employer cleavage: The impact of the informal sector on voting in Lima, Peru. Bulletin of Latin American Research, 10(3), 293–313. Carnes, Matthew E., & Mares, Isabela (2014). Coalitional realignment and the adoption of non-contributory social insurance. Socio-Economic Review, 12(4), 695–722. Carnes, Matthew E., & Mares, Isabela (2016). Redefining Who’s ‘In’ and Who’s ‘Out’: Explaining preferences for redistribution in Bolivia. The Journal of Development Studies, 52(11), 1647–1664. Collier, Ruth Berins, & Handlin, Samuel (2009). Reorganizing popular politics: participation and the new interest regime in Latin America. University Park, PA: Penn State University Press. de Soto, Hernando (1989). The other path: The invisible revolution in the third world. New York: Harper and Row. Dion, Michelle L. (2010). Workers and welfare: Comparative institutional change in twentieth-century Mexico. Pittsburgh: University of Pittsburgh Press. Dion, Michelle, & Birchfield, Vicki (2010). Economic development, income inequality, and preferences for redistribution. International Studies Quarterly, 54(2), 315–334. Dix, Robert H. (1989). Cleavage structures and party systems in Latin America. Comparative Politics, 22(1), 23–37. Esping-Anderson, Gøsta (1990). The three worlds of welfare capitalism. Princeton: Princeton University Press. Garay, Candelaria (2007). Social policy and collective action: Unemployed workers, community associations and protest in Argentina. Politics & Society, 35(2), 301–328. Garay, Candelaria (2016). Social policy expansion in Latin America. New York: Cambridge University Press. Garganta, Santiago, & Gasparini, Leonardo (2015). The impact of a social program on labor informality: The case of AUH in Argentina. Journal of Development Economics, 115, 99–110. Gasparini, Leonardo, & Tornarolli, Leopoldo (2009). Labor informality in Latin America and the Caribbean: Patterns and trends from household survey microdata. Desarrollo y Sociedad, 63, 13–80. Greene, Kenneth (2011). Campaign persuasion and nascent partisanship in Mexico’s new democracy. American Journal of Political Science, 55(2), 398–416. Haggard, Stephan, & Kaufman, Robert R. (2008). Development, democracy, and welfare states: Latin America, East Asia, and Eastern Europe. Princeton: Princeton University Press. Handlin, Samuel. (2013). Social protection and the politicization of class cleavages during Latin America’s left turn. Comparative Political Studies, 46(12), 1582–1609. Harris, John R., & Todaro, Michael P. (1970). Migration, unemployment, and development: A two-sector analysis. American Economic Review, 60(1), 126–142.

182

A. Baker, V.X. Velasco-Guachalla / World Development 102 (2018) 170–182

Holland, Alisha. (2017). Forbearance as redistribution: The politics of informal welfare in Latin America. New York: Cambridge University Press. Huber, Evelyne, Mustillo, Thomas, & Stephens, John D. (2008). Politics and social spending in Latin America. Journal of Politics, 70(2), 420–436. Huber, Evelyne, & Stephens, John D. (2012). Democracy and the Left: Social Policy and Inequality in Latin America. Chicago: University of Chicago Press. Hunter, Wendy. (2010). The transformation of the workers’ party in Brazil, 1989–2009. New York: Cambridge University Press. Hunter, Wendy, & Power, Timothy J. (2007). Rewarding Lula: Executive power, social policy, and the Brazilian elections of 2006. Latin American Politics and Society, 49(1), 1–30. Huntington, Samuel P. (1968). Political order in changing societies. New Haven: Yale University Press. Inter-American Development Bank [IDB] (2004). Good jobs wanted: Labor markets in Latin America: 2004 economic and social progress report. Washington, D.C.: InterAmerican Development Bank. International Labour Organization [ILO] (2002). Decent Work and the Informal Economy. Geneva: International Labour Office. International Labour Organization [ILO] (2013). Measuring informality: A statistical manual on the informal sector and informal employment. Geneva: ILO. Kitschelt, Herbert, Hawkins, Kirk A., Luna, Juan Pablo, Rosas, Guillermo, & Zechmeister, Elizabeth J. (2010). Latin American Party Systems. New York: Cambridge University Press. Krosnick, Jon. (1991). Response strategies for coping with the cognitive demands of attitude measures in surveys. Applied Cognitive Psychology, 5(3), 213–236. Kurtz, Marcus (2004). The dilemmas of democracy in the open economy. World Politics, 56(2), 262–302. Levitsky, Steven (2003). Transforming labor-based parties in Latin America: Argentine Peronism in comparative perspective. New York: Cambridge University Press. Levy, Santiago (2008). Good intentions, bad outcomes: Social policy, informality, and economic growth in Mexico. Washington, D.C.: Brookings Institution Press. Lupu, Noam, Gervasoni, Carlos, Oliveros, Virginia, and Schiumerini, Luis (2015). Argentine Panel Election Study. Mainwaring, Scott, Torcal, Mariano, & Somma, Nicolás M. (2015). The left and the mobilization of class voting in Latin America. In Ryan Carlin, Matthew Singer, & Elizabeth Zechmeister (Eds.), The Latin American Voter (pp. 69–98). Ann Arbor, MI: University of Michigan Press. Maloney, William F. (1999). Does informality imply segmentation in urban labor markets? Evidence from sectoral transitions in Mexico. World Bank Economic Review, 13(2), 275–302. McCann, James A., & Lawson, Chappell H. (2003). An electorate adrift? Public opinion and the quality of democracy in Mexico. Latin American Research Review, 38(3), 60–81. Meltzer, Allan H., & Richard, Scott F. (1981). A rational theory of the size of government. Journal of Political Economy, 89(5), 914–927. Milner, Helen V., and Rudra, Nita (2017). ‘‘Globalization and the Political Benefits of the Informal Economy”. Paper presented at the Annual Meeting of the American Political Science Association, Philadelphia, PA. Milner, Helen, & Rudra, Nita (2015). Globalization and the political benefits of the informal economy. International Studies Review, 17(4), 664–669. Nelson, Joan. (1970). The urban poor: Disruption of political integration in third world cities? World Politics, 22(3), 393–414. O’Donnell, Guillermo (1978). State and alliances in Argentina, 1956–1976. Journal of Development Studies, 15(1), 3–33. Oxhorn, Philip (1998). The social foundations of Latin America’s recurrent populism: Problems of popular sector class formation and collective action. Journal of Historical Sociology, 11(2), 212–246. Payne, J. Mark, Zovatto G., Daniel, Carrillo Flórez, Fernando, & Allamand Zavala, Andrés (2003). La Política Importa: Democracia y Desarrollo en América Latina. Washington, D.C.: Inter-American Development Bank.

Perlman, Janice E. (1976). The myth of marginality: Urban poverty and politics in Rio de Janeiro. Berkeley, CA: University of California Press. Perry, Guillermo E., Maloney, William F., Arias, Omar S., Fajnzylber, Pablo, Mason, Andrew D., & Saavedra-Chanduvi, Jaime. (2007). Informality: Exit and exclusion. Washington, D.C.: World Bank Publications. Portes, Alejandro (1971). Urbanization and politics in Latin America. Social Science Quarterly, 52(3), 697–720. Portes, Alejandro, & Hoffman, Kelly (2003). Latin American class structures: Their composition and change during the neoliberal era. Latin American Research Review, 38(1), 41–82. Pribble, Jennifer (2013). Welfare and party politics in Latin America. New York: Cambridge University Press. Roberts, Kenneth M. (2002). Social inequalities without class cleavages in Latin America’s neoliberal era. Studies in Comparative International Development, 36 (4), 3–33. Rudra, Nita (2008). Globalization and the race to the bottom in developing countries: Who really gets hurt? New York: Cambridge University Press. Rueda, David. (2005). Insider-outsider Politics in Industrialized Democracies: The challenge to social democratic parties. American Political Science Review, 99(1), 61–74. Schneider, Ben Ross. (2013). Hierachical capitalism in Latin America. New York: Cambridge University Press. Schneider, Friedrich, & Enste, Dominik H. (2013). The shadow economy: An international survey (second ed.). New York: Cambridge University Press. Schneider, Ben Ross, & Soskice, David (2009). Inequality in developed countries and Latin America: Coordinated, liberal, and hierarchical systems. Economy and Society, 38(1), 17–52. Sears, David, Lau, Richard R., Tyler, Tom R., & Allen, Harris M. (1980). Self-interest vs. symbolic politics in policy attitudes and presidential voting. American Political Science Review, 74(3), 670–684. Singer, Matthew M. (2016). Informal sector work and evaluations of the incumbent: The electoral effect of vulnerability on economic voting. Latin American Politics and Society, 58(2), 49–73. Soares, Glaucio. (1967). The politics of uneven development: The case of Brazil. In Seymour M. Lipset & Stein Rokkan (Eds.), Party systems and voter alignments: Cross-national perspectives. New York: The Free Press of Glencoe. Stokes, Susan C. (1995). Cultures in Conflict: Social Movements and the State in Peru. Berkeley: CA: University of California Press. Thornton, Douglas. S. (2000). Political attitudes and participation of informal and formal sector workers in Mexico. Comparative Political Studies, 33(10), 1279–1309. Van Cott, Donna Lee (2005). From movements to parties in Latin America: The evolution of ethnic parties. New York: Cambridge University Press. Weyland, Kurt. (1996). Neopopulism and neoliberalism in Latin America: Unexpected affinities. Studies in Comparative International Development, 31(3), 3–31. Wibbels, Erik (2017). ‘‘Informal Work, Risk and Clientelism: Evidence from 145 Slums in India”. Paper presented at the Annual Meeting of the American Political Science Association, Philadelphia, PA. Wiesehomeier, Nina., & Benoit, Kenneth (2009). Presidents, parties and policy competition. Journal of Politics, 71(4), 1435–1447. Willmore, Larry (2014). ‘‘Towards Universal Pension Coverage in Mexico: Briefing Number 13”. Pension Watch: Briefings on Social Protection in Older Age. London: Help Age International. Zucco, Cesar Jr, (2008). The president’s new constituency: Lula and the pragmatic vote in Brazil’s 2006 presidential elections. Journal of Latin American Studies, 40 (1), 29–49.