DEMOCRACY AT WORK

Like its practice, democracy is a rich, muddied, and highly contested concept. Many democratic theorists highlight the central role of contestation, participation, and citizenship as core principles (Dahl Reference Dahl1971; Marshall Reference Marshall1950; Pateman Reference Pateman2012).Footnote ³ Dryzek reminds us that democracy is “dynamic and open-ended” (Dryzek Reference Dryzek2000, 29). We capture the complexity of democratic politics by illuminating how multiple features of democracy contribute to well-being. Much of the empirical literature evaluating democracy's effects focuses on how elections influence government performance and citizens’ well-being (Avelino et al. Reference Avelino, Brown and Hunter2005; Boix Reference Boix2001; Gerring et al. Reference Gerring, Knutsen, Skaaning, Teorell, Coppedge and Staffin2015). We acknowledge elections’ role in aggregating preferences and promoting accountability in decision-making, but we focus our theoretical and empirical attention on ongoing forms of citizen participation, inclusive social policies, and state reform as crucial factors for citizens’ well-being.

We begin with the role of citizenship in the democratic project because of its centrality in democratic theory (Dahl Reference Dahl1971; Marshall Reference Marshall1950; Somers Reference Somers2008). Our understanding of democratic citizenship stems from Marshall's three complementary dimensions: civil, political, and social rights.Footnote ⁴ Ensuring access to these rights is central to democratic politics because citizens must have unimpaired opportunities to formulate preferences, engage in individual and collective action, and participate in deliberative processes in pursuit of their interests (Avritzer Reference Avritzer2002; Dahl Reference Dahl1971; Dryzek Reference Dryzek2000; Fung and Wright Reference Fung and Wright2003). Marshall's seminal work on citizenship highlights wide variation in citizens’ ability to effectively use these rights; variation occurs across the three dimensions he identifies, across different social groups, and over time. Meaningful access to political rights includes citizens’ ability to engage in political competition, contestation, and formal electoral processes. Meaningful access to civil rights protects citizens’ liberty from excessive state intrusion. Beyond civil and political citizenship lies social citizenship, which involves “the right to a modicum of economic welfare and security to the right to share to the full in the social heritage and to live the life of a civilized being according to the standards prevailing in society” (Marshall Reference Marshall1950, 8). For countries transitioning to democracy, the extension of full citizenship must overcome decades or even centuries of practices that have left vulnerable groups subject to semifeudal and authoritarian social relations (O'Donnell Reference O'Donnell1998).

How do democracies construct meaningful citizenship? Democratic institutions create opportunities for citizens to gather information, organize, and advocate for their interests. Electoral politics is one possible democratic avenue because it permits citizens to select candidates and parties who they believe will represent their interests (Ferraz and Finan Reference Ferraz and Finan2011; Gerring et al. Reference Gerring, Knutsen, Skaaning, Teorell, Coppedge and Staffin2015; Huber and Stephens Reference Huber and Stephens2012). Representative government provides citizens with accountability mechanisms to encourage elected officials to enact inclusionary policies and programs. Some researchers find that electing ideologically driven parties, especially leftist parties, is likely to result in expanded political access (Rueschemeyer et al. Reference Rueschemeyer, Stephens and Stephens1992) and produce more progressive outcomes related to poverty, inequality, and social spending (Sandbrook et al. Reference Sandbrook, Edelman, Heller and Teichman2007).

Participation, including ongoing dialogue and deliberation among government officials, individuals, and organized actors, represents another democratic avenue toward citizenship. These forms of participation occur in multiple venues such as public hearings, policy conferences, and street demonstrations at both subnational and national levels (Dryzek Reference Dryzek2000; Snyder Reference Snyder2001; Tarrow Reference Tarrow1998). Social accountability expands when citizens and civil society organizations (CSOs) can publicly deliberate with government officials, propose alternative policies, and monitor public goods provision (Fox Reference Fox2015; Putnam et al. Reference Putnam, Leonardi and Nanetti1994; Smulovitz and Perruzzoti Reference Peruzzotti and Smulovitz2000).

THREE PATHWAYS TO WELL-BEING

We move beyond elections to connect democracy to well-being through three main pathways. Participatory institutions serve as one pathway by promoting deliberative decision making that forges new relationships—among citizens, CSOs, and public officials—establishing the basis for investment in public goods that poor citizens need. Participatory institutions are state-sanctioned institutional processes that devolve decision-making authority to venues that incorporate both citizen and government officials (Abers and Keck Reference Abers and Keck2013; Avritzer Reference Avritzer2002; Baiocchi et al. Reference Baiocchi, Heller and Silva2011; Cornwall and Coelho Reference Cornwall and Coelho2007; Fung and Wright Reference Fung and Wright2003; Wampler Reference Wampler2015). The proliferation of these types of incremental policy-making bodies now allows citizens to focus narrowly on policy implementation and more broadly on public goods provisioning.

The richness—and messiness—of democratic politics is directly related to the multiple venues for competition over scarce resources. Political contestation and competition permit citizens, organized groups, and elected governments to expand access to civil, political, and social rights through the establishment of new institutions and social policies (Marshall Reference Marshall1950; Sen Reference Sen1999; Somers Reference Somers2008; Yashar Reference Yashar2005). More recent scholarship on participation and deliberation acknowledges a broader range of processes and institutions, including extensive experimentation with new democratic institutions and increased emphasis on transparency and open government (Fox Reference Fox2015; Pateman Reference Pateman2012). Such approaches allow citizens and government officials to overcome democratic deficits associated with democratic regimes (e.g., low participation, minimal voter knowledge, unresponsive bureaucracies). In this way, citizen engagement in democratic politics and expansion of social rights are reinforcing.

Social policies designed to empower citizens represent our second pathway to well-being. Development, conceived through a human capabilities approach, emphasizes “the crucial role of social opportunities to expand the realm of human agency and freedom, both as an end in itself and as a means of further expansion of freedom” (Drèze and Sen 2002, 8). Research on older democracies reveals how citizen engagement and expanded social provisioning work together to create policy feedback effects that restructure subsequent politics (Pierson Reference Pierson1993). Skocpol reveals how, in the U.S., the post–Civil War expansion of benefits for veterans and widows, known as “mothers’ pensions,” structured maternalist social welfare in the progressive era and beyond (Skocpol Reference Skocpol1995). Importantly, she reveals how policy was linked to women's voluntary organizations that took shape prior to the expansion of women's suffrage. Mettler's work on the GI Bill demonstrates how social policies create feelings of inclusion to construct citizens (Mettler Reference Mettler2007). Social policy can thus create a sense of belonging that is vital for full citizenship. Within less-developed countries, social policy can be a pathway for the deepening of citizenship in its fullest sense (Hunter and Sugiyama Reference Hunter and Sugiyama2014).

Governments in the developing world grappled with dual pressures, to expand social inclusion while facing fiscal constraints associated with neoliberal economic reforms (Huber Reference Huber and Esping-Andersen1996). Rather than expand social welfare through universalist principles, new democracies have focused on inclusion through targeted programs, aiming for greater efficiency and effectiveness (Teichman Reference Teichman2004). Conditional cash transfer (CCT) programs best exemplify this approach. These programs target resources to the poor while also inducing behavioral changes thought to promote human development (Fiszbein and Schady Reference Fiszbein and Schady2009). CCTs have spread throughout the developing world since the late 1990s (Sugiyama Reference Sugiyama and Hunter2013). Social policy research reveals that governments need not spend large sums of money to achieve marked improvements in poverty relief, education, and health. Policy design and progressive investment in areas that affect the poor, such as primary school and preventive health care, are more important than absolute spending levels (McGuire Reference McGuire2010; Sugiyama Reference Sugiyama and Hunter2013). This is particularly true for developing countries that have historically prioritized expensive services such as hospitals and universities, rather than more basic services that the poor depend upon.

State capacity in democracies is our third pathway to well-being. Much contemporary scholarship on the developmental state is rooted in Sen's work on fostering human capabilities as a means to promote productivity, the foundation for economic growth (Evans and Heller Reference Evans and Heller2015). A human capabilities approach broadens the scope of action to all citizens and highlights the state's ability to deliver public goods (Sen Reference Sen1999), rather than using collaboration with a narrow group of potential industrialists to incentivize industrial investment (Kohli Reference Kohli2004). Democratic institutions support a human capabilities approach in several ways, including the expansion of civil society through the protection of civil and political rights, a broadening of access points into the state, and the expansion of social rights (Evans and Heller Reference Evans and Heller2015; Ostrom Reference Ostrom1996; Sen Reference Sen1999). Low levels of state capacity (e.g., poorly trained personnel, poor access to telecommunications and equipment) can make it difficult to execute even well-designed public policy. The state's capacity to engage citizens and deliver policy reforms is therefore intricately tied to a larger democratic process of inclusion and participation.

Among large third-wave democracies, decentralization was thought to address gaps in state capacity. Local governance would render local authorities more responsive to voters, corruption more visible and thus easier to control, resulting in improvements in local service provision (Faguet Reference Faguet2004; Grindle Reference Grindle2007, 7–8). Yet, decentralization has not lived up to expectations in many settings (Gibson Reference Gibson2013; Giraudy Reference Giraudy2013) and entrenched clientelism often distorts governing in ways that perpetuate poverty and harm the poor (Diaz-Cayeros et al. Reference Diaz-Cayeros, Estevez and Magaloni2016; Weyland Reference Weyland1996). Subnational authoritarian enclaves continue to hinder governance and impede the construction of meaningful citizenship in many federal democracies.

We argue that democracies’ ability to direct local state capacity toward service provision is integral to promoting well-being. Local governments deliver nationally guaranteed services, which results in complex coordination processes. For example, these local governments may not prioritize services for some citizens or may even use state capacity to harm them. Citizens’ and nationally elected governments’ ability to harness state capacity and use it to deliver services is a crucial third pathway to improving well-being—especially in new democracies where the local state faces myriad problems with coordination, oversight, and accountability.

The emergence of participatory institutions, innovative social programs, as well as the need to harness local state capacity to deliver services all highlight the complexity of democratic practice. Despite this complexity, political scientists have devoted the lion's share of their attention to the role of national-level elections to explain democratic achievements in well-being (Gerring et al. Reference Gerring, Knutsen, Skaaning, Teorell, Coppedge and Staffin2015; McGuire Reference McGuire2010). Most of the democratic canon begins with the fundamental role of “free and fair” elections; without them democratic politics is absent (Dahl Reference Dahl1971). From an empirical standpoint, election data are publicly available, measurable, and comparable, which enables cross-national research. Unfortunately, a focus on elections obscures the diverse ways that public contestation and participation help to construct citizenship and promote citizens’ ongoing involvement in quotidian democratic activities. This focus also has the potential to hang too much on one representative institution, suggesting that if competitive elections do not produce better outcomes for citizens, democracy itself may be at fault.

THE BRAZILIAN DEMOCRATIC CONTEXT

Brazil's experience with democracy offers a rich theoretical basis to examine a more expansive conceptualization of democracy. Brazilian democracy emerged after a lengthy political opening following military rule (1964–1985). The country resurrected its previous political institutions and federal system while also strengthening local subnational autonomy through decentralization. Observers cautioned that the country's political institutions would create challenges to governability; problems included presidentialism, a malapportioned Congress, ideologically inchoate parties, a weakly institutionalized party system, and open list proportional representation (Ames Reference Ames2001; Lamounier and Meneguello Reference Lamounier and Meneguello1986; Mainwaring Reference Mainwaring1999; Power Reference Power2010). Further, Brazil's institutional design was thought to promote clientelistic politics across the federal system (Abrucio Reference Abrucio, Fernando2005; Hagopian Reference Hagopian1996; Samuels Reference Samuels2003). Despite these shortcomings, the electorate has experienced competitive municipal, state, and national elections. Between 1994 and 2014, presidential elections stabilized into regular competitive patterns between two parties, the center-right Brazilian Social Democratic Party (PSDB) and the center-left Workers’ Party (PT).

The democratic transition also forced elected officials to confront pent-up social demands amid serious economic constraints. The national government adopted neoliberal economic reforms based on fiscal austerity due to anemic economic growth and hyperinflation in the late 1980s and early 1990s. Policymakers were starved of revenue they needed to address a complex mosaic of historic social exclusion. The military dictatorship (1964–1985) benefitted from a relatively strong economy from the mid-1960s to the late 1970s, but public investments in education and health fell behind those of other nations experiencing similar economic growth.Footnote ⁵ By 1991, the Brazilian government classified about 45% of the population as poor and 20% as indigent (Ipeadata 2016). Brazil was also one of the most unequal countries in the world, with a Gini coefficient of 0.61 in 1991 (World Bank 2016). Other social indicators, such as life expectancy and infant mortality rates, were also relatively poor given the country's middle-income status (McGuire Reference McGuire2010). These macroindicators obscure uneven regional economic development—geographic pockets of deep poverty resulted from concentrated industrial development in the south and southeast, while the north and northeast were left behind. The historic exclusion of Afro-Brazilians and women has also produced gendered and racial features to socioeconomic exclusion and rights violations.Footnote ⁶

Civil society expanded greatly during redemocratization to advance democratic inclusion. Social movements mobilized to secure greater access to public goods, participate in policymaking venues, and broaden citizenship (Avritzer Reference Avritzer2002). Movements, including the sanitaristas (public health activists), students, women, and Afro-Brazilians, advocated for the expansion of social rights at the Constitutional Assembly (1987–8). The 1988 Constitution enshrined new social rights to housing, employment, health care, and education as well as political rights that would permit the development of a broader democratic architecture. Many new rights were aspirational as the government struggled to deliver public goods under fiscal austerity (Huber Reference Huber and Esping-Andersen1996, 171–2).

Limiting our study to Brazil leverages the benefits that come from a subnational, single-country study, which holds national institutions and electoral politics constant (King et al. Reference King, Keohane and Verba1994; Snyder Reference Snyder2001). Brazilian municipalities (N = 5570) are responsible for delivering many services and there is remarkable variation in local experiences with participatory institutions, coverage of new social programs, and local administrative performance—meaning that local quality of life and individuals’ potential to develop agency also vary. Wide municipal variation also makes our research applicable across broader contexts because some less affluent, rural municipalities face challenges resembling those in less affluent countries around the world, whereas more affluent municipalities more closely resemble those in older, wealthier democracies. Brazilian federalism renders municipalities independent and politically autonomous units, representing ideal laboratories for examining the role of democratic mechanisms on human development outcomes. But municipalities also rely on the federal government for financial transfers, thus allowing the federal government significant opportunities to induce municipalities to adopt new policies and institutions.

DEMOCRACY AT WORK IN BRAZIL

Brazil has experienced high rates of infant mortality like other developing and middle-income countries at the start of democratization (World Health Organization 2016). Importantly, these rates have fallen much farther in some cities and regions relative to others and subnational governments have taken different approaches to experimenting with social programs (Brazilian Ministry of Health 2016). Scholars and practitioners still do not know the extent to which new democratic institutions, social programs, expansion of state capacity, and democratic elections are responsible for Brazil's reductions in infant mortality, especially in the context of rapid economic growth that should also reduce infant mortality. Explaining variation in infant mortality is essential for understanding well-being in a broader sense. There is now a global consensus on the vital importance of infant mortality for development (Nussbaum Reference Nussbaum2011; Sen Reference Sen1999; United Nations Development Programme (UNDP) 2015). Infant mortality is a strong proxy for well-being because surviving beyond one's first year represents a necessary condition for pursuing one's chosen life (McGuire Reference McGuire2010; Nussbaum Reference Nussbaum2011). Brazil's government collects municipal-level data on new democratic institutions, new social programs, budget management, elections, and infant mortality that help us connect these isolated areas. Brazil thus offers a unique opportunity to test connections between participatory institutions, policy, and managerial aspects of democracy on local well-being at a level of breadth and depth that has never been achieved previously.

RESEARCH DESIGN AND CASE SELECTION

We draw on an original dataset covering Brazil's 5,570 municipalities to evaluate connections between local participatory institutions, federal social programs, and infant mortality. As discussed above, we focus on infant mortality as a proxy for well-being because surviving birth represents a necessary condition for comfort, health, and happiness while pursuing one's chosen life (McGuire Reference McGuire2010; Nussbaum Reference Nussbaum2011). Moreover, infant mortality is a useful proxy for testing our arguments because it can change quickly due to both policy intervention and neglect (Aquino et al. Reference Aquino, de Oliviera and Bareto2009). If participation and social programs are related to well-being, it should be apparent with infant mortality.

Our data on local democracy, social policies, state capacity, and economic growth represents one of the largest datasets on subnational policies in the developing world and the only one aligning key local aspects of participation, social programs, and administrative capacity with local outcomes. Our data covers all Brazilian municipalities from 2006 to 2013, which translates to models with up to 28,618 municipal-year observations.

We base our strategy on previous efforts to explain variation in infant mortality and that of other health outcomes. Both Rasella et al. (Reference Rasella, Aquino, Santos, Paes-Sousa and Barreto2013) and Macinko et al. (Reference Macinko, Guanais and de Souza2006) use conditional negative binomial models with municipal fixed effects to test hypotheses connecting Brazil's Bolsa Família and PSF to infant mortality at the municipal level. We follow suit in our study; negative binomial regression models resolve several statistical challenges in estimating models of infant mortality. Specifically, negative binomial regressions provide improved estimation in cases where count and rate outcome data is widely dispersed—as in our case, where the unconditional mean of infant mortality is much smaller than its variance (Cameron and Trivedi Reference Cameron and Trivedi2009; Hilbe Reference Hilbe2007). We then use panel data models with municipal fixed effects to account for correlations between unobserved, time-invariant characteristics of the panel and our independent variables.Footnote ¹¹

ROBUSTNESS CHECKS

We perform a variety of tests to assess the robustness of our results.Footnote ¹⁹ First, our results are robust to different lags of the dependent variables, model specification, and exclusion of outliers. For example, models that use per capita municipal spending or per capita municipal GDP instead of per capita municipal health care spending also produce results that are broadly similar to those in Tables 1 and 2. Geographic dummy variables are sometimes significant determinants of infant mortality, especially the North and the Northeast compared to the South. However, the central explanatory variables retain their significance, magnitude, and direction in models with geographic dummies, too. Next, we present the results of models excluding the Bolsa Família management variable, which limits our data coverage to 2006–13 in our primary models. Dropping this variable extends our coverage back to 2000 and increases the number of observations in each model. The results for our central remaining variables are broadly similar to those in our primary models.

Next, we address the potential for nonlinear relationships between our central explanatory variables and infant mortality. Simple measures to improve medical, education, and sanitation services may quickly reduce infant mortality in municipalities where it is very high. However, these services may already exist in municipalities with low infant mortality rates and new efforts targeting infant mortality further may produce diminishing marginal returns. We split our sample into municipalities with low, medium, and high levels of infant mortality to address this possibility and present new models of infant mortality in Table I(i) of the Appendix. The results for our central explanatory variables do vary some across each tercile of infant mortality. We find the most support for our arguments in municipalities with low and medium levels of infant mortality. The results for the political variables in our models also vary across terciles. For instance, having a mayor from the PT is systematically associated with lower levels of infant mortality among municipalities in the lowest tercile (lowest infant mortality rates) and competitive mayoral elections are associated with low infant mortality rates in the highest tercile. These results suggest that having a partisan champion that promotes health care, education, and sanitation for the poor is important in many municipalities in our dataset. It also suggests that political competition influences well-being under many circumstances.

We account for endogeniety in our models in several different ways. Specifically, it is possible that previous levels of infant mortality influence municipalities’ future choices surrounding health care spending and service provision. For example, a municipality struggling with infant mortality might have adopted health care-related councils and increased health care spending to address this problem. Macinko et al. (Reference Macinko, Guanais and de Souza2006) address a similar endogeneity issue surrounding infant mortality and federal social programs through instrumental variable regression using the mayor's party as an instrument for Bolsa Família and PSF coverage. However, previous scholarship connects the mayor's party directly to infant mortality rates in Brazil (Touchton and Wampler Reference Touchton and Wampler2014), which suggests that this variable violates the exclusion restriction for instrumental variables and would therefore be inappropriate to include as an instrument in our models (Wooldridge Reference Wooldridge2013). Instead, we use Arellano-Bond dynamic panel models to account for this potential concern.Footnote ²⁰ These models use the “system” generalized method of moments (GMM) with one lag of the dependent variable. We then use the policy council variables, Bolsa Família and PSF coverage, Bolsa Família management, per capita health care spending, and low-income wages as instruments, beginning with the second lag and going back as far in time as the data exists for each variable. The direction of the coefficients and the general levels of statistical significance are all similar to those in the primary models. We also use dummy variables for each year to ensure the assumption of no correlation across units holds. The results in Table I(j) of the Appendix thus provide supporting evidence for Table 1.

An additional, important consideration is that certain municipalities might be predisposed to target infant mortality more than other municipalities for some unobserved reason. These municipalities might then promote participatory governance and strong management of federal programs as means to an infant mortality-reducing end, along with many other unobserved programs or policies. Any relationships between participatory governance, state capacity and infant mortality could therefore only reflect municipal predispositions, as opposed to any impact from the specific institutions or programs. We find little evidence to support such a scenario; there are only low correlations between municipalities with voluntary policy councils related to health care, social program coverage, and management of these social programs.Footnote ²¹ This suggests that municipalities do not tend to excel in all three areas simultaneously and that a commitment to participatory governance, social program coverage, and social program management do not stem from an unobserved penchant for reducing infant mortality. Results for propensity score matching corroborates this argument for policy councils (Table I(k)) and provides support for their independent role in improving well-being.