Social Policy Expansion and Retrenchment in Latin America: Causal Paths to Successful Reform

Abstract

The literature on social policy expansion and retrenchment in Latin America is vast, but scholars differ in how they explain outcomes, arriving at different conclusions about the role of democracy, left parties, favorable economic conditions, and social movements. What can welfare state developments since the end of the commodity boom teach us about the theoretical power of these arguments? This paper engages this question, seeking to explain recent incidents of successful social policy reform in 10 presidential administrations in Argentina, Brazil, Chile, and Uruguay. Using a combination of crisp-set qualitative comparative analysis (csQCA) and case studies, we identify multiple paths toward successful social policy expansion and retrenchment. The QCA results highlight two key findings. First, social policy expansion was generally carried out by programmatic parties (often, though not always, from the left). Second, retrenchment is most likely when nonprogrammatic right parties face fiscal constraints. The case studies affirm these findings and show that differences in electoral competition, social movement pressure, and policy legacies shape the contexts in which expansion and retrenchment is feasible. The results provide new insight into social policy reform, underscoring the relevance of complex forms of causality.

Introduction

During the first decade of the 21^st century, Latin American states expanded social welfare policies, improving program coverage, increasing benefit size, and in some cases, boosting the quality of public services. This ‘expansionary wave’ marked an abrupt change from the trends of the 1980s and ‘90s, when policy makers eliminated programs, privatized benefits, and cut spending.¹ The literature on these two periods is vast, but scholars differ in how they explain expansion and retrenchment. While many agree that democracy and electoral competition facilitated expansion during the early 2000s (Altman and Castiglioni, 2020; Castiglioni, 2018; Garay, 2016; Huber and Stephens, 2012a; Pribble, 2013), there is less agreement about how democracy influenced retrenchment just years earlier (Brooks, 2009; Haggard and Kaufman, 2008; Madrid, 2003). The literature also differs in how it views the effects of left parties, policy legacies, and social movements, for both social policy expansion and retrenchment (Altman and Castiglioni, 2019; Ciccia and Guzmán-Concha, 2021; Cruz-Martínez, 2021; Ewig, 2016; Fairfield and Garay, 2017; Garay, 2016; Huber and Stephens, 2012a; Pribble, 2013). One area of consensus relates to the effect of economic conditions. Scholars agree that the debt crisis and the adoption of austerity policies influenced retrenchment in the 1980s and ‘90s (Brooks, 2009; Madrid, 2003), whereas favorable conditions in the early 2000s facilitated expansion (Holland and Schneider, 2017).

By 2011, the commodity boom of the early 2000s had begun to fade,² leading some analysts to wonder if Latin American states would halt expansion or perhaps even cut social benefits. A string of electoral victories by right-wing parties fortified this concern, raising the question of whether conservative actors would retrench the social policy gains achieved during the expansionary phase. How has Latin American social policy changed in the wake of the commodity boom? What can welfare state developments since that period teach us about the theoretical power of existing arguments? This paper engages these questions, seeking to characterize and explain recent reforms and probe lingering debates in the literature about the determinants of social policy expansion and retrenchment in Latin America.

Changes to Latin American social policies have provoked a rich and diverse body of scholarship that embraces methodological pluralism, but scholars remain divided over the question of what factors explain expansion and retrenchment. The sources of these disagreements are manifold but the fact that much of the literature on this period relies on small-N, comparative historical analysis and case studies, may explain some of the divergence. Authors deploying these methods varied in which cases they selected and what policy sectors they analyzed. Studies also differed in how authors conceptualized and measured social policy reform, with some focusing on changes to coverage and/or spending (e.g. Huber and Stephens, 2012a; Barrientos, 2018), while others adopted multidimensional measures (Garay, 2016; Pribble, 2013).

Yet another source of disagreement may be rooted in the nature of causal processes that drive social policy expansion and retrenchment. Most studies of the retrenchment and expansionary phases understand independent variables as having the same effect across the countries and policies under study. We argue for the possibility that causation may be more complex, involving equifinality (i.e. the possibility that there are different paths to the same outcome), conjunctural causation (i.e. when causal conditions do not operate on their own, but in combination with other factors), and asymmetry (i.e. that the causes of expansion are not the same as those that explain retrenchment). To probe for complex causality, we use Qualitative Comparative Analysis (QCA). In particular, we analyze the data using crisp-set QCA (csQCA), originally developed by Ragin (1987), and the Stata package “fuzzy”, developed by Longest and Vaisey (2008). We complement the csQCA with case studies of the main causal paths that explain expansion and retrenchment.

The paper presents an analysis of all major successful social policy reforms (24 in total) in Argentina, Brazil, Chile, and Uruguay between 2010 and 2019.³ Drawing on the existing literature, we identify six key factors that scholars agree were important in shaping variation in the scope and character of welfare state expansion and retrenchment: electoral competition, party ideology, party’s linkage mechanism, social mobilization, economic recession, and foreign debt.⁴ We then code these conditions, along with all of the major successful policy reforms enacted by ten presidential administrations. The selection of reforms includes all instances of successful expansion and retrenchment. The administrations include Argentine Presidents Cristina Kirchner (2011-2015 term) and Mauricio Macri (2015-2019), Brazilian Presidents Dilma Rousseff (2011-2016), Michael Temer (2015-2018), and Jair Bolsonaro (2019-2023), Chilean Presidents Sebastián Piñera (2010-2014 and 2018-2021) and Michelle Bachelet (2014-2018), and Uruguayan Presidents José Mujica (2010-2015) and Tabaré Vázquez (2015-2019). The analysis identifies five minimized paths toward social policy expansion and three paths toward retrenchment.⁵ After presenting the findings, we discuss each of those paths, reflecting on key reforms included in each track and underscoring the ways that different combinations of causal conditions produce expansion and retrenchment.

The results point to two key findings. First, most expansionary reforms after the commodity boom have been advanced by programmatic parties, often from the left, in the absence of recession. Second, recent retrenchment has most often occurred during nonprogrammatic right party rule and in settings of fiscal constraints. Policy legacies are a necessary factor in every single case of reform, but because the condition does not vary, we exclude it from the csQCA and instead explore its effect in the case studies, along with the role of social mobilization and electoral competition.

The paper’s findings are important, as they shed light on several unresolved debates in the literature – namely, whether left and right governments differ in their approach to social policy formation and change, whether resources alone guarantee expansion and prevent retrenchment, and whether electoral competition provides an impetus to expand social welfare programs, among others. The causal paths identified in this paper may only apply to the specific countries included in the study and to the 2010-2019 time period. Argentina, Brazil, Chile, and Uruguay all have extensive welfare states. The countries also benefitted from the commodity boom, engaging in notable social policy expansion during the first 12 years of the 21^st century. The end of the bonanza, therefore, likely placed unique pressure on this set of cases. While these scope conditions may limit the generalizability of our results, one aspect of the study has broad theoretical relevance: the finding that social policy reform involves complex forms of causation. Specifically, this paper shows that different paths can lead to the same social policy outcome and that no causal condition on its own can determine expansion and retrenchment. This is an important finding for future research on social policy reform in Latin America and beyond.

Determinants of social policy expansion and retrenchment in Latin America

This paper analyzes instances of social policy expansion and retrenchment. To date, much of the research on welfare state change has separated the processes rather than including both within a single analysis. This is because there is agreement in the literature that the politics of retrenchment and expansion are different (Pierson, 1994). In both strands of the literature, however, authors tend to focus on similar causal factors: the role of economic growth, external pressure, democratic competition, partisan ideology, policy legacies, and social mobilization. Since QCA analysis allows for equifinality and causal asymmetry, we are able to analyze the processes of expansion and retrenchmet together.⁶

Studies of welfare state retrenchment have found that while Latin American countries in the 1980s and 1990s tended to engage in deep reforms, including privatization and program cuts, European welfare states generally witnessed more covert forms of retrenchment. Scholars argued that outright retrenchment was made possible in Latin America because of the region’s severe economic (debt) crisis, the tenuous character of democracy, the fragmentation of civil society, and the weakness of left parties (Brooks, 2009; Huber and Stephens, 2012a; Madrid, 2003). Conversely, European countries witnessed more limited retrenchment (Hacker, 2004; Pierson, 1994). Scholars argue this resulted from the prevalence of institutional veto points that protected against social policy roll-backs and policy legacies that made such reforms unpopular (Huber and Stephens, 2001; Pierson, 1994, 2001).

When it comes to social policy expansion, scholars also cite the importance of economic factors in facilitating change. Specifically, the literature points to the importance of growth (McGuire, 2010), trade openness (Cameron, 1978; Katzenstein, 1985), revenue-collection capabilities (Cruz-Martínez, 2021), and the effects of neoliberal reforms (Ewig and Kay, 2011). Holland and Schneider (2017) contend that increased revenue generated by export earnings during the commodity boom facilitated (easy) expansion. Similarly, Murillo, Oliveros, and Vaishnav (2011) note that positive terms of trade meant that Latin American governments were not forced to seek external credit to finance policy. This allowed politicians to pursue domestic priorities, including those that fell on the left of the ideological spectrum.

Studies of welfare state expansion also point to the importance of political factors, including democracy, institutional design, party ideology, and party character. A great deal of research on social policy change points to the importance of democracy and electoral competition (Garay, 2016; Haggard and Kaufman, 2008; Huber and Stephens, 2012a; McGuire, 2010; Pribble, 2013; Segura-Ubiergo, 2007), though authors differ in how they conceptualize the effects of regime type and competitiveness. Huber and Stephens (2012a) highlight the ways that democratic politics empower working class and marginalized voters. Garay (2016) focuses on electoral competition, noting that where parties must compete for the support of labor market outsiders, expansion is more likely.

Some studies of Latin American social policy expansion find that the election of left leaning presidents and the growing power of left parties in the legislature played a decisive role in the growth of social programs (Huber and Stephens, 2012a). Pribble (2013) focuses on the impact of parties, but argues that it is important to consider ideology in combination with other aspects of party organization – namely, the predominant linkage structure and the strength of elite-voter ties – a concept that the author terms ‘party character.’ Ponce de León (2021) also focuses on party linkage strategies, showing that programmatic commitments shift the quality of social policy legislation that is adopted.

Other studies find that right-wing parties also implemented redistributive social policies, while left-wing governments sometimes carried out reforms that benefit labor market insiders at the expense of those living in extreme poverty (Altman and Castiglioni, 2019; Fairfield and Garay, 2017; Garay, 2016). Ewig (2016) argues that, beginning in the late 1990s, Latin America witnessed a convergence towards greater equity in social policy, regardless of government ideology. A similar trend is identified by De La O (2015), who notes that CCTs spread across the entire region, independent of partisan ideology.

Virtually all studies of social policy expansion and retrenchment in Latin America find that policy legacies are an important determinant of reform outcomes (Arza, 2012; Brooks, 2009; Niedzwiecki and Pribble, 2017; Castiglioni, 2018; Dion, 2010; Ewig and Kay, 2011; Huber and Stephens, 2012a; Martínez-Franzoni and Sánchez-Ancochea, 2016; Pribble, 2013; Vommaro and Gené, 2022). Policy legacies matter because they generate incentives (or disincentives) for reform, by structuring the distribution of power among actors, shaping public opinion, and generating problems that need fixing.

Some studies of social policy expansion also stress the importance of mobilization and social movements (Altman and Castiglioni, 2020; Ciccia and Guzmán-Concha, 2021). Garay (2016), for example, finds that mobilization from below prompted inclusive expansion, whereas expansion that came solely in response to electoral pressure was more restricted. In the case of Bolivia, Anria and Niedzwiecki (Anria and Niedzwiecki, 2016; Niedzwiecki and Anria, 2019) show that unions and grassroots organizations coordinated pressure in the streets to secure the passage of a universal pension reform in 2008.

Clearly, then, while rich in detail and concept development, the literature on Latin American social policy development has arrived at divergent conclusions about the causes of social policy transformation. We believe that part of this disagreement relates to the complex nature of the causal processes that drive successful policy change.

Method, Case Selection, Outcomes and Causal Conditions

Crisp-Set Qualitative Comparative Analysis (csQCA)

Our analysis uses crisp-set QCA based on set-theory, originally developed by Ragin (1987). QCA is the most appropriate method for this paper for several reasons. First, it allows for equifinality or the possibility that there are different paths to the same outcome (Schneider and Wagemann, 2012, p. 5). QCA is well-equipped for testing contrasting theories on social policy reform. Second, the method allows for the possibility of conjunctural causation, a form of causality characterized by variables that do not operate on their own, but in combination with other factors (Schneider and Wagemann, 2012, p. 6; Vis, 2012, p. 172). Third, we assume that the causes that explain expansion are not the same as the ones that account for retrenchment, and this lack of symmetry makes QCA the most appropriate method for analyzing both expansion and retrenchment in a unified framework (Schneider and Wagemann, 2012, p. 6). Finally, the mid-size of our sample (24 country-policies) is too large for in-depth qualitative analysis and too small for regression analysis, but adequate for QCA (Ragin, 2000, p. 25).

We use crisp-set QCA as opposed to fuzzy-set QCA due to its ease of interpretation and, more importantly, because the outcome and causal conditions in our study lend themselves to dichotomization. This denotes either membership or non-membership in a set, as opposed to gradations in their set membership (Schneider and Wagemann, 2012).⁷

Case selection

We analyze social policy reforms undertaken during Latin America’s post-commodity boom era (2010-2019) in Argentina, Brazil, Chile, and Uruguay. We choose this specific time period to analyze a context in which resource constraints force politicians to prioritize initiatives, and potentially engage in both expansion and retrenchment. We select these four countries because they are widely considered, along with Costa Rica, to have the most developed social policies in the region (Barba Solano, 2007; Filgueira, 1998, 2007; Huber and Stephens, 2012a; Martínez-Franzoni, 2008; Mesa-Lago, 1978).⁸ Countries with mature welfare states face different opportunities for expansion and pressures for retrenchment than countries with more limited social protection. Thus, limiting our analysis to advanced social protection systems allows for a meaningful comparison that avoids concept stretching. All four countries experienced the commodity boom, but the cases also exhibit variation on key causal conditions, including economic growth, partisan ideology, and party character. Tables 1 and 2 summarize these differences. Taken together, the case selection strategy delimits our scope conditions to countries that exhibit the most developed social policies in the region, specifically South American countries.

Table 1. Post-Commodity Boom Presidential Administrations (2011–2019)

Table 2. Years of Negative Real GDP Growth (2002–2018)

Source: World Bank (2020).

Theoretical expectations

The outcome and causal conditions in our analysis are coded as having two possible outcomes (presence=1 or absence=0). Table A.1 in the Appendix shows this coding.

Outcome: successful social policy reform

Our outcome of interest is successful social policy reform. Social policies are state-funded programs that seek to improve the health, education, and well-being of residents by addressing social risks. We include reforms to health systems, pensions, child and elder care, education, family leave, and conditional cash transfers. Successful social policy reforms can take the value of either expansion (1) or retrenchment (0). We analyze all major successful social policy reforms (24 in total), excluding reforms that were discussed but never submitted to Congress, as well as bills that died in committee, or any situation resulting in status quo policy continuity. The coding of policy reforms is based on a close reading of the government documents (laws and official websites), as well as secondary literature.

Our operationalization of expansion and retrenchment follows the literature on Latin American social policy that has moved away from spending to more precise measures.⁹ We include three indicators: coverage, eligibility criteria, and benefit generosity. Expansion of social policy occurs when a policy reform increases coverage, eases eligibility criteria, improves access, and/or ameliorates benefit generosity in real terms. Retrenchment of social policy involves the opposite: a decrease in coverage, tightening of eligibility criteria, a reduction in access, and/or a decline in benefit generosity in real terms.¹⁰ To measure the existence of expansion or retrenchment, we analyze data on changes to the share of people who are covered by a policy, changes to eligibility criteria, changes in access, and changes to the value of the transfer relative to inflation.

The decision to dichotomize the outcome as expansion (1) and retrenchment (0) deserves particular attention. We recognize that the absence of expansion entails not only retrenchment, but also maintenance of status quo policy. However, maintenance of status quo would mask two conceptually different processes. In some instances, status quo maintenance is the result of failed reform (i.e. elites attempted to alter the policy, but were blocked), while in other cases the status quo is maintained because no reform is attempted. The literature has shown that failed reform and the absence of attempted reform are theoretically distinct processes, driven by different causal factors, and should therefore not be analyzed together (Brooks, 2009; Madrid, 2003; Pribble, 2013). Without extensive field research and in-depth interviews with politicians, one would not be able to differentiate between the two types of status quo maintenance and would, therefore, risk conflating failed reforms and reforms that were never attempted. We, therefore, focus our analysis exclusively on successful reforms – that is, instances in which elites attempted to alter the policy and were able to do so. In those cases, the reform outcome can take one of two forms: expansion and retrenchment.

Causal conditions

The first causal condition we include in our analysis is party ideology. Levitsky and Roberts (2011) define parties of the left as those that focus on reducing social and economic inequalities and enhancing the political participation of marginalized groups. Parties of the right, by contrast, perceive such inequalities as natural, maintain a more hierarchical vision of social relations, and place very little emphasis on redistribution. We therefore expect left parties to be more inclined to expand social policy and right parties to be more open to retrenchment. We use Huber and Stephens’ (2012b) coding of parties, with the exception that we code Frente para la Victoria in Argentina as left and not “other.” We updated the series for parties not coded in the dataset.

In addition to the ideological orientation of the president, we code whether the executive’s party relies predominantly on a programmatic linkage mechanism to appeal to voters. Linkage mechanisms define the nature of the relationship between voters and a party. The literature on linkage techniques reveals three general strategies: programmatic, clientelist, and interest incorporation (Kitschelt and Wilkinson, 2007; Morgan, 2011; Roberts, 2002). We reduce these three approaches to a dichotomous coding of programmatic and non-programmatic strategies. We recognize that parties often rely on a mix of linkage strategies, but we focus on the predominant approach. We expect that parties that employ programmatic linkages will remain more wedded to their party’s ideological program and more susceptible to electoral pressure. We, therefore, expect the combination of programmatic linkage and left ideology to increase the likelihood of expansion, whereas non-programmatic parties may prove more flexible. When it comes to retrenchment, we expect parties of the right to be more likely to roll back benefits than left parties, but that such reforms will be less likely for programmatic right parties, since they must compete on the basis of their record and retrenchment is generally unpopular with voters. Our coding of programmatic parties is drawn from Pribble (2013) for Argentina, Chile, and Uruguay and we code the Workers’ Party in Brazil as programmatic and Temer’s and Bolsonaro’s parties as non-programmatic. Table 1 summarizes the ideological position of the ten administrations and the predominant linkage mechanism deployed by each party across the four countries included in the analysis.

The third causal condition included in our analysis is a measure of social mobilization, which we code based on a close reading of news coverage for each social policy reform. Building on Garay (2016) and Ciccia and Guzmán-Concha (2021), we expect bottom-up pressure to promote social policy expansion. For retrenchment, we expect social mobilization to limit the likelihood of program cuts and roll-backs (Niedzwiecki, 2014). In cases of expansion, we code a reform as involving bottom-up pressure when social movements were actively organizing in support of the project or when there were protests leading up to the introduction of a policy initiative. For example, Chile’s 2015 university reform, which guaranteed free access for roughly 130,000 students, came on the heels of massive protests. We, therefore, code that reform as involving bottom-up pressure. For retrenching reforms, we code a case as exhibiting bottom-up pressure if social movements organized to defend the existing program against retrenchment. One example is Argentina in 2017, when then-President Mauricio Macri altered the formula for income transfer adjustments sparking massive protests against this retrenchment policy (Di Costa, 2018; Vommaro and Gené, 2022).

Fourth, we expect that in a context of economic recession, retrenchment is more likely. We use data from the World Bank and code any previous country-year in which the real growth rate was zero or negative as experiencing recession, while all positive-growth country years were coded as having no recession. Dichotomizing GDP growth can be problematic, but the literature suggests that moments of economic contraction are substantively different from periods of expansion, no matter how meager growth might be. Moreover, as illustrated in Table 2, our cases differ substantially in terms of this condition.

Fifth, and relatedly, international pressure can shape domestic decisions on social policy. Since international pressure can produce incentives for retrenchment and/or expansion, we do not have clear expectations regarding the direction of causality. In the 1980s and 1990s, multilateral loans came with conditionalities that led to retrenchment. By contrast, in the 2000s, international credit helped fund expansion. We expect international pressure will be strongest in countries that exhibit high levels of debt. This is because those countries may need to seek financial assistance to fund new policy initiatives. Additionally, high debt countries face pressure to appease investors’ concerns about deficits, which could result in efforts to cut back on social programs. Building on the work of Blofield, Pribble, and Giambruno (2022), we use the World Bank’s measure of debt as a share of GDP to assess fiscal constraints. We code countries with a debt/GDP ratio of 50 percent or more as facing international pressure (1), and 0 otherwise. We arrived at the 50 percent threshold inductively, by comparing the debt levels of 15 Latin American countries across three time points. Figures 1-3 summarize the data, making clear that the majority of countries in the region have debt levels well below 50 percent of GDP. Those states that do reach “high” levels of debt are precisely the countries that we know have faced international pressure – namely, Argentina from 2015 on, Brazil throughout the time period, El Salvador throughout the time period, and Costa Rica and Ecuador for the final years of the series. In the cases of Argentina, Costa Rica, Ecuador, and El Salvador, the era of high debt coincided with greater IMF involvement in the economy.¹¹ The coding also adequately captures the freedom from international pressure enjoyed by Colombia, Peru, Mexico, Chile, and Uruguay during this period. For this reason, we contend that the 50 percent threshold makes theoretical and substantive sense.

Figure 1. Debt as a Share of GDP (2010).

Figure 2. Debt as a Share of GDP (2015).

Figure 3. Debt as a Share of GDP (2019).

Finally, we code for the presence or absence of strong electoral competition. Scholars vary in how they define and operationalize competition. Conceptually, competitive systems are those in which parties must work hard to win the support of voters. In such settings, we expect that parties will be encouraged to support broader social programs that benefit a large sector of voters and respond to popular demands, rather than narrowly targeted benefits. For this reason, we expect parties in a competitive environment to favor social policy expansion. Moreover, since parties in competitive systems must avoid alienating large portions of the electorate, we expect them to avoid actions that are unpopular, such as retrenching social programs.

We measure competition as the share of the vote won by the second leading candidate in national legislative elections or first-round executive elections. We contend that if the second leading candidate/party wins at least 25 percent of the vote, the presidential election can be considered competitive. This is because that candidate will enjoy a broad enough base that they can reasonably build upon and challenge the front runner, ensuring competitive dynamics. Drawing on our knowledge of the four cases, we found that this measure was most effective at capturing the dynamics and evolution of Argentina, Brazil, Chile, and Uruguay’s political systems. We compared this operationalization with alternative measures, such as difference in vote share between the first and second-place candidates (Dash and Sacchidananda Mukherjee, 2015; Pribble, 2015), margin of victory (Alfano and Baraldi, 2015), and electoral volatility (Immergut and Abou-Chadi, 2014), but found that our indicator has better internal validity. All four of our cases are multi-party presidential democracies that require second-round elections if a candidate does not win an absolute majority of the vote (Brazil, Chile, and Uruguay) or, in the case of Argentina, 45 percent of the vote or 40 percent with at least 10 points more than the runner-up. Moreover, the cases exhibit divergence regarding the presence of pre-electoral coalitions. While some countries have strong coalitions that run a unique candidate, others tend to see many candidates in the first round. This makes it hard to interpret the meaning of margins of victory in the first round. Moreover, volatility can serve as a proxy for a weakly rooted (Roberts and Wibbels, 1999) or weakly institutionalized (Mainwaring, 2018) party system; so volatility-based measures of competition would cast Brazil as very competitive and Uruguay much less so, thereby mis-classifying the cases.

Discussion of results

We analyze our data using the Stata package “fuzzy” developed by Longest and Vaisey (2008). The truth table used for the csQCA, which presents the coding of the outcome and causal conditions in our analysis, is included in Appendix A1. Our coding of policy reforms reveals that Argentina, Brazil, Chile, and Uruguay all engaged in some policy expansion. Retrenchment, by contrast, happened only in Argentina, Brazil, and Chile.

Table 3 presents the csQCA results that reveal eight paths toward expansion and five towards retrenchment, each with 100 percent Y-consistency. One factor that jumps out of this first level of analysis is that only parties of the right (in the presence or absence of other factors) retrenched social policy. By contrast, the analysis shows that, in the presence and absence of other causal conditions, parties of both the left and right have pursued expansion.

Table 3. Complete csQCA Results Y-CONSISTENCY vs All Other Y-CONSISTENCY

Note: L=Left, l=Right; B=Bottom-up Pressure, b=∼Bottom-up Pressure; P=Programmatic Parties, p=∼Programmatic Parties, C=Competitive Elections, c=∼Competitive Elections; D=Debt 50% or more; d=Debt less than 50%; R: Recession; r: ∼Recession.

To better synthesize the results of our analysis, we use logical minimization, meaning that when two paths are identical except in one causal condition and lead to the same result, then that causal condition can be eliminated (Ragin, 2017, p. 32). Table 4 presents the minimized results.

Table 4. csQCA Logical Minimization

Note: upper-case denotes presence of a condition and lower-case the absence of that condition.

The minimized analysis reveals five combinations of factors that produce social policy expansion and three paths that lead to retrenchment. These five paths explain 20 of our 24 cases. In the table, all-caps terms represent the presence of the condition, while lowercase terms represent the absence of the condition.

Expansion

Table 4 identifies five separate paths toward expansion, and three (Paths 1, 2, and 5) that account for the vast majority (12) of our cases of expansion. For the sake of simplicity, we focus on these three paths. The first path (#1) is one of expansion under programmatic left parties, no recession, and high indebtedness, but an absence of mobilization. This path includes Frente Amplio Presidents José (Pepe) Mujica and Tabaré Vázquez in Uruguay and Workers’ Party President Dilma Rouseff in Brazil. The reforms range in sectoral focus from healthcare in Brazil and Uruguay to the creation of a new care system in Uruguay; to cash transfer policy in Brazil and maternity leave in Uruguay.

The second expansionary path (#2) accounts for four of the reforms and also includes programmatic left parties and no recession, but with the differences that these are cases of low indebtedness and high electoral competitiveness. These are expansionary reforms undertaken by the administrations of Michelle Bachelet in Chile, including education, university, healthcare, and a cash transfer.

Taken together, the first two paths account for 9 of the 14 expansionary reforms and while there is a combination of both transfers and services, services constitute the bulk of the reforms (six of the nine initiatives). We believe this fact is important, as expanding some services tends to be ‘harder’ than expanding noncontributory transfers (such as CCTs), due to the high cost and to the complicated nature of the political coalition required to push reforms forward (Holland and Schneider, 2017). In addition, the implementation of new services tends to require stronger state capacity and more spending than the provision of new cash transfers (Arza et al., 2022). It is, therefore, interesting to identify the conditions under which such ‘hard’ reforms are possible. Our findings suggest that they are most common when there is no recession and in the presence of a programmatic left in power.

A good example of this expansionary path is Michelle Bachelet’s push to improve Chile’s public healthcare system. In 2014, following a competitive presidential election, President Bachelet committed her administration to building 20 new hospitals; starting construction on 20 additional hospitals; and establishing contracts for 20 more hospitals. By the end of her term, four years later, the president had managed to uphold that promise. The new infrastructure was spread across the country’s territory and in several cases the new hospitals replaced health centers that were built in the 1960s and ‘70s (Rehbein, 2018). The project, therefore, constitutes a remarkable expansion to Chile’s public health infrastructure, increasing service capacity and quality. Policy legacies also played a role in Chile’s healthcare expansion. Roughly 70 percent of Chileans rely on the public sector for care, so these reforms are potentially very popular. While the public system is large, it has suffered from infrastructural decay, long wait-lines, and quality gaps. These weaknesses became increasingly clear following the 2004 passage of the AUGE healthcare reform, which guarantees treatment for a set of illnesses inside pre-determined timelines. In the wake of the reform, the state often struggled to meet the AUGE commitments, motivating frustration among citizens and generating budget pressure as the state was forced to outsource care to private actors. These legacies helped motivate President Bachelet to engage in the 2014 reform.

The third main path (#5) that we highlight here is one of expansion under a programmatic right party. President Mauricio Macri’s administration in Argentina expanded access to CCTs and to early childhood education, and improved the benefit generosity of contributory pensions. These three reforms took place after a highly competitive presidential election and in the absence of recession, yet with high levels of debt. Case evidence shows that the role of competition is important here and underscores the way that highly competitive environments in combination with programmatic parties can incentivize expansion, even from parties of the right and even in the presence of high debt.

Taken together, the five expansionary paths in Table 3 show that the absence of economic recession is an important determinant of policy expansion. Indeed, recession is absent in all cases of expansion. In the absence of recession, revenue grows, which may facilitate expansion by reducing dependence on external credit or because politicians can improve programs without having to introduce new taxes. This fact is especially important for countries with high levels of debt. The finding suggests that future research must take economic growth seriously and incorporate it systematically into analyses of policy reform. Studies of the early 2000s often treated the economic bonanza as a background factor rather than a key causal condition (Arza et al., 2022).

Retrenchment

The minimized results in Table 4 reveal that there are three paths toward retrenchment (Paths 6-8) – the first under a programmatic right party and the last two under nonprogrammatic right parties. All three paths involve the presence of high levels of debt, suggesting that fiscal pressure is a motivating factor for retrenchment. We expect that this works both by limiting access to funding, but also by providing a framing to help justify/explain potentially unpopular cuts to the electorate.

The last two paths toward retrenchment (#7 and #8) include reforms by non-programmatic, right-wing Presidents Michel Temer and Jair Bolsonaro, in a context of high indebtedness. The case of Bolsonaro’s 2019 pension reform (#7) suggests that even in settings where there is no recession, non-programmatic parties of the right may still undertake retrenchment. This is especially true in the presence of policy legacies that facilitate the reforms. Brazil’s pension retrenchment was presented as an initiative aimed at addressing the country’s growing deficit; a key policy legacy of the existing system. The reform was expected to save the treasury around U$D195 billion in the next decade (McGeever and Marcello, 2019). Bolsonaro, therefore, used the deficit to justify and frame this reform. The reform applies to workers in the public and private systems. It sets a minimum retirement age, increases the minimum years of required contribution, and increases the amount of transfer with years of contribution. By linking individual pensions more closely to years of contribution, the reform prevents higher-paid workers from retiring earlier, but it also affects those who have interrupted careers in the formal labor market, especially women. Sátyro (2021, pp. 331–332) explains that people who enter the formal labor market at 16 years old or younger would need 49 years of work to receive a full pension, while the life expectancy in Brazil’s poorest regions is 67 years old.

Conclusion

The results of this analysis provide new insight into the politics of successful social policy reform in South America’s advanced welfare states, as well as into lingering debates about the determinants of social policy expansion and retrenchment. Using csQCA, we find that there are multiple paths toward successful social policy expansion and retrenchment. One of the central findings is that the most common path toward expansion involved the absence of recession and the presence of programmatic parties, especially of the left. Furthermore, the results of our analysis provide evidence that in the post-commodity boom era, retrenchment is feasible, especially in settings where policy legacies, most notably large budget deficits, allow politicians to frame reform as necessary. While retrenchment was only carried out by parties of the right, it appears to be most common among non-programmatic parties. Such parties can more easily shift competition away from issues of policy toward charismatic or anti-system appeals. Moreover, retrenchment was always carried out in the presence of high debt levels, suggesting that cuts are more likely as fiscal constraints (and international pressure) tightens.

The findings presented in this paper have methodological, theoretical, and practical importance. In methodological terms, the results suggest that QCA is an especially useful approach for analyzing social policy expansion and retrenchment, due to the likelihood of complex forms of causality. In this paper we identify multiple paths that lead to both expansion and retrenchment, we find that no one causal condition on its own causes change, and that the causal processes of expansion and retrenchment are different across cases and sectors.

The results of our analysis shed light on lingering debates in the literature about the role of economic resources, electoral competition, and political parties in determining social policy outcomes in Latin America. Specifically, the analysis suggests that rather than producing ‘contradictory’ findings, existing research has successfully identified elements of the multiple paths that lead to social policy expansion and retrenchment in the region, but that work remains to theorize the impact of these conditions in combination with one another. For example, we find that ideology seems to influence social policy reform, but the effect happens in combination with the party’s predominant linkage mechanism and the availability of resources. Future research could explore whether the causal paths inductively identified in this paper travel to other countries and time periods.

The Covid-19 pandemic had devastating social and economic effects in the region, exposing inequalities and generating hardship and new demands for state intervention. Simultaneously, the pandemic crippled local economies, making it difficult for countries to respond with increased spending and new programs in answer to growing demands. The findings presented in this paper provide scholars with a fresh new perspective about how countries might navigate the post-pandemic social crisis. The analysis suggests that we can expect countries headed by programmatic left parties, in competitive settings, where economic growth picks back up, to be the best prepared to respond the crisis. By contrast, right-wing governments, especially those headed by non-programmatic parties and carrying a high debt burden, may be slow to respond and even use the occasion to roll-back social programs. These factors could generate vastly different trajectories of recovery and well-being in the wake of the pandemic.