1

Including Outsiders in Latin America

1.1 Introduction
Throughout the twentieth century, much of the population of Latin
America lacked access to health care services, stable income, and pen-
sions. Although states introduced social protections for workers in the
formal sector (those with labor contracts), workers outside the formal
labor market and their dependents, whom I  call “outsiders,” remained
unprotected or underserved by social policy. Outsiders include the urban
informal sector – the self-employed, street vendors, and employees hired
off the books – as well as rural workers and the unemployed. In the last
decade of the twentieth century, outsiders represented between 40  per-
cent and 80 percent of the population in the middle-income countries of
the region, and a large share of them lived in poverty.1
Two macro-level transformations that occurred in the last decades of
the twentieth century – the adoption of democratic regimes and economic
liberalization – raised contradictory expectations about the likelihood
that Latin American states would extend social protections to outsiders.
Democracies institutionalized electoral participation and opened chan-
nels for the expression of interests and demands, which seemed to augur
well for initiatives to reduce the welfare gap. Yet the debt crisis of the
early 1980s and the implementation of market-oriented reforms gave rise
to a period of state retrenchment marked by the remarkable spread of
pension privatization in the 1990s and the extension of small-scale, tar-
geted, and often clientelistic benefits to the very poor. In light of these

1
Estimates with government data (see Appendix 1). See also Portes and Hoffman (2003:
49, 53).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 2 Social Policy Expansion in Latin America

changes, a broad academic consensus maintained that despite the wide-

spread adoption of democracy, Latin America had entered a new era of
market expansion and limited state involvement in social protection.2
Contrary to this picture of state retreat, a dramatic expansion of social
policy for outsiders took shape in several middle-income countries of
Latin America during the 1990s and 2000s, when governments began
to extend pensions, cash transfers, and health care services to millions of
previously unprotected outsiders. Consider these examples. Following a
constitutional reform in the late 1980s, Brazil adopted a universal health
care system as well as broad-reaching pension programs for outsiders,
both of which began implementation in the early 1990s. Cash transfers
initiated in 2001 reached 11 million low-income households by 2006. In
the 2000s, Mexico, which had been historically characterized by modest
social expenditures, launched a health insurance system for outsiders,
cash transfers for 5 million children, and pensions for outsiders that by
2010 reached a similar share of people 65 and older than the preexist-
ing program for formal workers.3 With some exceptions, by 2010, sev-
eral middle-income countries in the region had expanded cash transfers,
pensions, and health care services for at least 35 percent of the outsider
population.4
As demonstrated in Figure 1.1, the magnitude of social policy change is
striking. By 2010 pensions in select countries reached between 48 percent
and 100 percent of outsiders aged sixty-five and older, and transfers were
provided to school-age children on a massive scale.5 At the same time,
these benefits are costly. Although cash transfers are able to reach many
individuals with relatively smaller investments, representing between 0.2
percent and about 1 percent of the GDP – comparable to family allow-
ances for formal-sector workers – health care services and some pension
programs demand more significant investments.

2
See, for example, Haggard and Kaufman (2008); Kaufman and Segura-Ubiergo (2001);
Rudra (2002); on pension reform, see Brooks (2001; 2009); Huber and Stephens (2000);
Kay (1998); Madrid (2002; 2003); Mesa-Lago (1994); on clientelism, see Cornelius et
al. (1994); Dresser (1994); Kurtz (2004a); Magaloni (2006); Roberts (1995); Weyland
(1996a).
3
Calculated with data on formal-sector pension coverage for people 65 and older in 2011
and benefits for outsiders in 2010 from SEDESOL.
4
In South America, exceptions include Paraguay, Peru, and Venezuela. The latter two cases
are discussed in Chapter 8. As discussed in Chapter 2, I use the threshold of 35 percent to
operationalize expansion.
5
Author’s estimate with government data of pensions, cash transfers and population. For
more information and sources, see Appendix 1.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
3

Including Outsiders in Latin America 3

97%

79%
74% 72%

55% 58%
48%
36%
30% 27%
25%
10%
0% 0% 0% 0%

Pensions Transfers Pensions Transfers Pensions Transfers Pensions Transfers

Argentina Brazil Mexico Chile

Before (c. 1990, Brazil 1980) After (c. 2010)

Figure 1.1 Share of outsider seniors and children with benefits before and after
expansion, selected countries.
Note: Outsider seniors 65+ (60+ Brazil, 1980).
Source: Author’s calculations with government sources (see Appendix 1).

These social policy innovations for outsiders are puzzling for a num-
ber of reasons. First, not only did social policy expansion take place at
a time of state retreat, but these benefits also reached the most vulner-
able and disempowered sectors of the population. This outcome runs
counter to the widely held assumption that outsiders lack the capacity to
exert political influence and attain meaningful policy responses in Latin
America’s nascent democracies. According to the literature, outsiders
face formidable obstacles to collective action because they have hetero-
geneous interests stemming from their diverse and often individualistic
economic activities (e.g., working as street vendors), which limit their
ability to coordinate around common goals and demands and develop
organizations to represent their interests (see Cross 1998; Kurtz 2004a).
At the same time, if organizations form among outsiders, these are seen
as having scarce resources. This prevents these organizations from hav-
ing a meaningful influence on state policy and often leads them to suc-
cumb to co-optation and clientelism.6 Furthermore, scholars argue that
in the context of scarcity produced by the debt crisis and market reforms,
powerful insiders prioritized the protection of their own benefits over

6
See particularly Kurtz (2004a). For a critique and discussion of this literature, see Arce
and Bellinger (2007); essays in Collier and Handlin (2009); Garay (2007); Holland and
Palmer-Rubin (2015).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 4 Social Policy Expansion in Latin America

the establishment of insider–outsider coalitions that could have improved

outsiders’ political organization (see Etchemendy 2011; Oxhorn 1998;
Weyland 1996a).
Social policy expansion is also intriguing because it involves nondiscre-
tionary benefits. Existing research often characterizes most middle-income
countries of Latin America as having patronage-based bureaucracies and
clientelistic parties,7 both of which hinder the creation of nondiscretion-
ary policies for the most vulnerable sectors of the population (Kitschelt
and Wilkinson 2007; Rothstein 2011), as such benefits are expected to
loosen voters from clientelistic arrangements, and thus undermine the
power of clientelist machines. Yet, as we see in the following chapters,
political parties considered exemplars of clientelism, such as the Peronist
or Justicialista Party (PJ) in Argentina, sponsored nondiscretionary social
policy innovations for millions of low-income outsiders, revealing more
complex relationships between political parties and poor voters.8
Finally, new social policies for outsiders display remarkable cross-
national variation. Two distinct models of social policy, which I call inclu-
sive and restrictive, can be identified. Inclusive policies provide relatively
generous benefits to all or a large pool of outsiders and tend to involve
some level of social participation in policy implementation. Restrictive
policies, by contrast, provide smaller benefits to a more limited pool of
outsiders and are implemented in a nonparticipatory way. Although there
is variation within each of these broad categories, sharp contrasts distin-
guish the two models.
This book seeks to explain the circumstances under which incumbents
in Latin America extended large-scale, nondiscretionary social policies
to outsiders, the most vulnerable and disempowered sector of the popu-
lation, and why we observe remarkable differences in the policy mod-
els that have taken shape. More specifically, why have some incumbents
embarked on the expansion of nondiscretionary social policies for out-
siders while others have not? Why have some governments created more
generous, broad-reaching policies than others? Why do some allow social
organizations and movements to participate in policy implementation
while others reach out to beneficiaries in a top-down manner?
To address these questions, this study draws on a comparative his-
torical analysis of social policy development in three areas that have

7
On bureaucracies, see Calvo and Murillo (2004) and Luna and Mardones (2014). On
party systems, see Hagopian (2014); Kitschelt et al. (2010); Mainwaring (1999); Roberts
(2014). On clientelistic linkages, see Luna (2014).
8
Levitsky (2003); for the recent period, see Etchemendy and Garay (2011).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
5

Including Outsiders in Latin America 5

exhibited a marked social policy divide separating insiders from outsid-

ers  – pensions, income support, and health care  – in four of the most
industrialized countries of Latin America  – Argentina, Brazil, Mexico,
and Chile – since democratization in the 1980s and 1990s. In an effort
to better understand the circumstances under which expansion occurred,
this study examines all democratic administrations in these countries
within this period, some of which did embark on expanding social policy
while others did not. Furthermore, it assesses the leverage of the analyti-
cal argument through a longitudinal analysis of these same cases since
the establishment of benefits for insiders in the first half of the twentieth
century, and conducts a broader comparison with four middle-income
countries in Latin America and beyond, discussing the applicability of the
argument to Peru, Venezuela, Uruguay, and South Africa.
As presented later in this chapter and laid out in Chapter 2, this study
finds that expansion occurred in democratic regimes with (a) high elec-
toral competition for outsiders and/or (b) large-scale social mobilization
from below. Each of these two dynamics compelled incumbents to expand
nondiscretionary social policy and temper the existing divide separating
insiders and outsiders. At the same time, contrasting models of social
protection resulted from the negotiations involved in policy design, the
preferences of those engaged in the design process, and their institutional
power. Restrictive models were built when conservatives had strong insti-
tutional power and social movements were not involved in policy design;
inclusive models were adopted when social movements demanding pol-
icy change participated in negotiations around expansion, either because
they had propelled that process in the first place, or because they could
influence policy design through an allied party in government. To preview
the outcomes documented and analyzed in this book, social policy expan-
sion in Argentina and Brazil produced inclusive social policy, while in
Chile and Mexico, a restrictive model took shape.

1.2 Why Study Social Policy Expansion

and Policy Models?
Mapping and understanding these social policy innovations is critical
for comparativists interested in the sources of welfare development and
variation in social policy models in developing countries, and in their
political and welfare effects. The cases discussed in this book illuminate
the challenges of extending benefits in societies with deep insider-outsider
divides, where some sectors are protected and others are not.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 6 Social Policy Expansion in Latin America

The adoption of broad-reaching benefits for outsiders has been largely

unanticipated and initially overlooked by scholars of social policy in
the region. In fact the comparative literature has emphasized obstacles
towards expansion, stressing in particular financial impediments to
social policy adoption.9 Yet, as analyzed in this book, financial consid-
erations are not the critical factor determining adoption of social policy
innovations in middle-income countries. Governments have embarked
on expansion under particular circumstances, and then employed dif-
ferent strategies to raise the necessary resources for implementation. At
the same time, even if some programs may seem relatively inexpensive,
expansion has often taken place across different policy areas requiring
substantial investments.
Understanding these expansions is important because they target out-
sider populations who have received much less attention within the social
policy  – and political science  – literature despite their numerical rele-
vance in a region characterized by massive labor informality, segmented
labor markets, and unemployment. Most of the comparative social
policy literature has instead focused on the well-protected labor force
and on social programs for insiders. Even the health care initiatives that
reached outsiders and were created before the third wave of democracy
have remained largely unaddressed by comparativists, with the exception
of James McGuire’s seminal work on the evolution of infant mortality
rates (2010). The focus on formal-sector programs has overshadowed
important aspects of the historical evolution of social policy in the region
within the comparative literature. Understanding the conditions of social
policy expansion for outsiders may shed new light on broader political
dynamics that require deeper exploration. This book seeks to contribute
to this pursuit.
A focus on the features of the new social programs also helps advance
our understanding of social policy dynamics. Unlike the literature on the
welfare state in industrial democracies, which has paid significant atten-
tion to the characteristics of social programs and explored the political
underpinnings of variation in social policy models, research on programs
for low-income populations in Latin America (e.g., conditional cash
transfers) has tended to treat these programs as a homogenous group,
thereby hiding significant variation that remains unexplained. This is par-
ticularly the case also in studies of clientelism that have focused primarily

9
Haggard and Kaufaman’s book on Latin America, East Asia, and Eastern Europe reaches
this conclusion (2008).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
7

Including Outsiders in Latin America 7

on the method of distribution of social programs, rather than also exam-

ining the different types of benefits that have been extended, and in stud-
ies of social expenditure, which do not separate investments across policy
areas with precision, identify what kinds of programs are funded, and
indicate whether beneficiaries are insiders, outsiders, or both.10 Knowing
the features of different policies, how consistent they are across policy
areas, and how they interact with benefits for insiders is fundamental
for informing policy making and advancing our understanding of the
politics of social policy. More generally, learning more about these ben-
efits for outsiders will allow us to better understand the shape of welfare
systems in the region, which include programs for insiders and outsiders,
and to comprehend the ways in which these benefits interact. These are
fundamental concerns for those seeking to improve welfare and labor
markets in developing countries, and they are critical aspects of social
policy dynamics that connect with themes of inequality, redistribution,
and government responsiveness.
More broadly, the expansion of social programs for outsiders has
important political and normative implications. First, these benefits
have created new and stronger connections between the state and citi-
zens who previously lacked access to many programs. At least in the lan-
guage in which social programs are framed, these benefits are presented
as linking the fate and prosperity of the larger political community to
individual improvement, access to services, and transfers for the most
vulnerable – and often marginal – populations. At the same time, some
of these initiatives – as discussed in this book – have included participa-
tory arrangements for implementation, which have opened up spaces of
interaction between beneficiaries and the state, providing new oppor-
tunities for citizens to have a “voice” and participate in policy mak-
ing. Secondly, the adoption of some of these programs has contributed
to reducing inequality, improving infant mortality rates, and increas-
ing school enrollment and completion.11 These are all historically dif-
ficult accomplishments for developing countries. At the same time, these
expansions have mainly benefited women, who have been generally less
able to access social security protections in the formal labor market,
and who now constitute the main beneficiaries of pensions for outsiders
and transfers for low-income households in several countries. How this

10
Haggard and Kaufman (2008); Huber, Mustillo, and Stephens (2008); Rudra (2002);
Rudra and Haggard (2005); Segura-Ubiergo (2007).
11
See Appendix 1 for further reference.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 8 Social Policy Expansion in Latin America

connection between welfare and women unfolds and what its ramifica-
tions are across different models of social policy for outsiders constitutes
a fundamental topic of research, as gender inequality has been associ-
ated with many pernicious social dynamics worldwide.
A key political question, then, is why outsiders matter in some coun-
tries, to the point of becoming the target of large-scale policy innova-
tions. The next section discusses alternative explanations for social policy
expansion, and the following sections present the argument advanced in
this book.

1.3 Alternative Explanations

Three different arguments have emerged within popular discourse and in
academic debates as potential explanations to account for incumbents’
decisions to launch social policies for outsiders and for variation in the
resulting policies. The first account focuses on economic change and
emphasizes the abundance of agricultural and mineral revenues in the
2000s to explain social policy innovations and the amount of benefits
distributed. The second highlights the arrival of left-wing coalitions to
power, while the final argument references the diffusion of policy models
to explain the increased popularity of social programs for outsiders and
the models adopted.

Economic Abundance and the Commodity Boom

At first glance, the timing of recent social policy innovations suggests
that increased state revenue from the agricultural and mineral commod-
ity boom of the early twenty-first century has provided incumbents with
the resources needed for social policy expansion, fueling or making pos-
sible the adoption of large-scale social programs across the region.12
According to this view, fiscal constraints in the 1980s and 1990s pre-
cluded politicians from expanding, but in the 2000s the rise in commod-
ity prices and the GDP growth it propelled allowed commodity exporters
to extend social policy to the excluded. Expanding the logic of this argu-
ment, one should then expect the amount of revenue from the boom or
the share of resources controlled by the state to be related to the social

12
Contrasts in average regional rates of economic growth are not particularly marked
between the 1990s, when Latin America grew on average 2.6 percent, and the 2000s,
when the region’s average growth rate was 3.2 percent (GDP rates of growth from
ECLAC; CEPALSTAT).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
9

Including Outsiders in Latin America 9

policy model adopted – restrictive or inclusive – with more generous and

broad-reaching benefits made possible by higher export-led growth.
Several studies have argued that export-led growth has allowed left-
wing governments in the region to implement their redistributive agendas
(see Campello 2015; Levisky and Roberts 2011; Weyland 2011). Without
denying the importance of the commodity boom and the broader eco-
nomic context in which it unfolded, proving its centrality to social policy
expansion and the resulting models of social policy adopted faces major
empirical and theoretical challenges.
Empirically, a close examination reveals that the occurrence and the
timing of expansion – understood as the creation of broad-reaching
nondiscretionary programs in selected policy areas – do not coincide
neatly with the timing of the boom, as governments adopted new benefits
both before and during the boom or did not expand nondiscretionary
benefits across selected social policy areas despite growing export-led
revenue. At the same time, countries that are not primarily commodity
exporters did embark on expansion, as illustrated by Mexico.13 Second,
GDP growth does not explain cross-national variation in the magnitude
of new benefits. In other words, incumbents with larger commodity
earnings are not necessarily more enthusiastic providers of benefits for
outsiders.
With respect to the timing and occurrence of expansion, incumbents
adopted new social policies for outsiders at times of both economic strain
and abundance since the late 1980s (see Figure 1.2). For example, facing
a severe economic crisis in 1998, Ecuadoran president Jamil Mahuad
(1998–2000) established a massive transfer program reaching 1 million
outsiders to contain the political and social ramifications of his fiscal
adjustment program (personal communication, Mahuad; Banco Central
de Ecuador 2010). By the time Ecuador achieved economic growth –
fueled in part by rising oil prices – after a dramatic drop in GDP in 1999,
40 percent of the country’s households were already receiving income
transfers on a regular basis. In Brazil, social policy expansion began in the
late 1980s, propelled by coalitions of social movements and labor unions
that actively mobilized during the democratic transition. Implementation

13
Data from Cepalstat (www.cepal.org, accessed March, 2015), shows that in 2010,
Mexico’s primary exports represented 24 percent of total exports while in the middle-
income economies of South America, primary exports represented between 62.9 percent
and 95.7 percent of total exports. Between 2002 and 2013, the average share of primary
exports was 23.4 percent in Mexico and it ranged from 55.3 to 92.5 percent in the coun-
tries of South America.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 10 Social Policy Expansion in Latin America

10% Arg_P
Arg_HC
Ur_HC+ P
Chi_HC Ecu_HC Ecu_HC
Bol_P+IS
5% Bra_HC Bol_P Mex_HC Chi_P
Ecu_IS Mex_P Bol_IS
GDP Change

Bra_P Mex_IS
Bra_IS Bra_IS Arg_IS Ec_P
0% Uru_IS Uru_IS
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
–5%
Bra–P

–10% Arg_IS

–15%
Year of Expansion
Figure 1.2 Timing of expansion and GDP change, selected countries, 1992–2010.
Note: IS: Income Support (cash transfer programs); P: Pensions; HC (health care).
Sources:  Author’s estimates of timing of expansion, GDP from World Bank
Development Indicators and ECLAC.

started at a time of economic hardship and inflation in the early 1990s,

which anteceded the successful stabilization accomplished by Itamar
Franco’s finance minister, Fernando Henrique Cardoso, in 1994, well
before the commodities boom.
Other episodes of expansion coincide with buoyant international mar-
kets and favorable macroeconomic conditions, such as the extension of
pension benefits in Argentina beginning in 2006. Still other policy inno-
vations, such as Mexico’s policy expansion between 2001 and 2007, took
shape at a time of modest growth and in the absence of windfalls from
agricultural, oil, or mineral exports.
Finally, the expansion of nondiscretionary, large-scale benefits did not
take shape consistently across selected policy areas in Peru, Paraguay,
or Venezuela, three countries that benefited from the commodity boom
(see Figure 1.3). Even though Peru experienced a massive increase in
GDP driven by mineral exports, it only expanded health services gradu-
ally, achieving some meaningful coverage by 2010 (see Cameron 2011).
In Venezuela under Hugo Chávez (1998–2012), an undeniable case of
export boom and state control of oil revenues and social policy inno-
vations did not include broad-reaching, nondiscretionary transfers for
children or seniors.14
The level of GDP growth is not a good predictor of the model adopted.
Focusing on the four countries under examination, we see that Brazil
adopted and implemented inclusive health care and pension benefits at

14
I discuss the cases of Peru and Venezuela in Chapter 8.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
11

Including Outsiders in Latin America 11

20

15

10
% GDP Growth

0
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014
–5

–10
Year

Paraguay Peru Venezuela

Figure 1.3 Economic growth, Paraguay, Peru, and Venezuela, 1990–2014.

Sources: GDP from World Bank Development Indicators.

times of low GDP growth (late 1980s–90s), while Bolsa Família, a

famous cash transfer that was not costlier than prior innovations, was
adopted in 2003 before the commodity boom that benefited Brazil begin-
ning in 2004. Argentina also adopted the inclusive model with expan-
sions occurring at times of both severe economic decline (2002) and
growth (2005–7). On the other hand, in Chile, during times of economic
growth bolstered by growing copper exports, incumbents did not adopt
an inclusive model but rather a restrictive one (2003–4, 2008), as they
did in Mexico, where incumbents faced overall modest growth (2001–3,
2006).
At a theoretical level, a critical weakness of an explanation of expan-
sion based on the commodity boom is that new economic resources in
and of themselves do not provide a straightforward explanation for why
incumbents would want to devote part of those resources for social pro-
grams for outsiders. Undoubtedly states that benefited from the commod-
ity boom had more room to maneuver financially (see Campello 2015).
Yet governments faced competing demands for funding from various
economic and social actors. In the context of economic growth, power-
ful businesses and producers often demanded funding for infrastructure,
transportation, energy projects or subsidies while labor unions expected
better salaries and pension benefits for insiders, which had deteriorated
in several countries in the 1980s and 1990s (see Etchemendy and Collier

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 12 Social Policy Expansion in Latin America

2007). In cases in which the state did not control revenues from commod-
ities directly but captured them via export taxes or royalties, businesses
and producers also pressured to avoid taxation (see Fairfield 2015).
Understanding why governments paid so much attention to outsiders
beginning in the 1990s, in the context of state retreat, economic crisis, or
high levels of growth, requires an explanation that goes beyond the avail-
ability of increased resources. Resources in and of themselves do not tell
us why governments were willing to reach out to outsiders, historically
the poorest and most politically vulnerable sectors of the population, and
why they chose to allocate resources through social policy transfers and
services instead of prioritizing other initiatives such as food subsidies or
employment-creation programs. This book argues that specific political
factors are fundamental to understanding governments’ social policy
decisions and the incentives that led incumbents – regardless of having
windfall revenue or not – to embark on expansion.

Partisan Politics: The Left Turn

Another potential explanation for social policy adoption emphasizes the
rise of left-wing governments in Latin America since the early 2000s. This
“left turn” by which “nearly two-thirds of Latin Americans lived under
some form of left-leaning national government” (Levitsky and Roberts
2011: 1), seems to lend credence to a theory of expansion based on left-
party power.15
In their most recent book, Huber and Stephens argue that in twenty-
first-century Latin America, “the longevity of democracy and the strength
of parties to the left of center” have produced a departure from the historic
trend of poverty and inequality and help explain differences in redistribu-
tive outcomes across countries (2012: 240). Although this book shares
the view that partisanship matters for the social policy model adopted, it
challenges the centrality of left-party strength in the executive branch or
in congress in accounting for why large-scale social policies for outsiders
were adopted in the first place. First, political parties in Latin America
face a highly divided social structure and left and left-populist parties
have generally prioritized the interests of labor union allies over those
of outsiders (see, for example, Collier and Collier 1991; Levitsky 2003).
The preference of left parties toward labor unions and insiders is also
found in Western Europe (see Rueda 2005, 2007; Lindavall and Rueda

15
See Levitsky and Roberts (2011); Weyland, Madrid, and Hunter (2010).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
13

Including Outsiders in Latin America 13

2013), where unionized workers are more numerous. In Latin America,

left and left-populist governments only launched policies for outsiders
under specific circumstances. The goal of this book is to identify these
circumstances and explain when and why left parties embarked on large-
scale expansion.
Second, the episodes of expansion analyzed herein demonstrate that
incumbents on both the left and the right of the political spectrum
launched significant policy innovations for outsiders. A focus on the left
as the driver of social policy change leaves unanswered the question of
why right-wing incumbents pursued expansion as well. For example,
the creation of massive income transfers and health insurance occurred
under the right-leaning presidency of Vicente Fox (2000–6) of Mexico’s
National Action Party (PAN), at a time when the left held only a small
share of seats in Congress. Likewise, some episodes of policy adoption
in Brazil took place under the administrations of José Sarney (1985–90)
and Itamar Franco (1992–5), neither of whom belonged to a popular-
based party or faced more than a minor left-wing party presence in the
Constituent Assembly and in Congress, the bodies that effectively passed
some of these innovations into law.
As shown in subsequent chapters, partisanship does help account
for different preferences regarding policy design that may affect policy
outcomes, as left or center-left politicians typically favor broader ben-
efits, and center-right or conservative parties prefer more modest ones
(see also Huber and Stephens 2012; Pribble 2013). However, left party
strength, understood as the arrival of left-wing candidates to office or the
growth of the share of left-party seats in Congress, does not in and of
itself explain the adoption of broad-reaching social programs for outsid-
ers in the cases under investigation.

Diffusion of Policy Models
A final potential explanation for the expansion of social policy and the
models adopted is based on the diffusion of policy models. The scholarly
literature on market reforms does emphasize diffusion as an important
factor contributing to the adoption of social policy change, particularly
privatization, in Latin America’s nascent democracies. As discussed by
Weyland (2006; 2004), theories of diffusion contend that policy change
results from the spread of policy principles (such as universality) or policy
models (such as social security) in temporal waves across geographically
proximate countries (see Weyland 2006:  19–21). Despite background

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 14 Social Policy Expansion in Latin America

differences, policy blueprints disseminated across borders influence gov-

ernments into adopting similar policies, producing policy convergence.
Models typically spread through two mechanisms: (a) a strong interna-
tional actor that pressures or provides strong incentives for governments
to adopt a particular policy blueprint; or (b) policy makers’ decisions to
emulate policy models they find prestigious, appropriate, or legitimate to
solve a specific problem.16
As subsequent chapters show in greater detail, diffusion provides an
inadequate account for social policy expansion in the cases under inves-
tigation, for three main reasons: (a) the lack of evidence that the mecha-
nisms of diffusion – imposition or incentives granted by a powerful actor,
and emulation – play significant roles in policy adoption, (b) the absence
of policy convergence across countries, and relatedly, (c)  the challenge
posed to the diffusion argument by cases of non-adoption despite expan-
sion in other cases.
Powerful international actors with the capacity to pressure or provide
strong incentives for countries to adopt new policy models or principles
did not play key roles in expansion. The World Bank (and associated
agencies) was a key player in pension privatization and social develop-
ment debates in the region (Brooks 2009, 2001; Madrid 2003; Weyland
2004, 2006), yet the process tracing for this project showed that multi-
lateral agencies have not been relevant actors behind recent expansions
in Argentina, Brazil, Chile and Mexico. Acknowledging that countries
such as Brazil and Mexico, rather than international agencies, were the
first movers in the design of conditional cash transfers for outsiders, an
interviewed World Bank official suggested that the agency “has trailed
behind governments in the region.”17
Concerning emulation, scholars have often suggested that diffusion
is facilitated by the existence of a coherent, simple model (see Weyland
2004, 2006). In the 1990s, there was no consistent single model of health
care services (Kaufman and Nelson 2004; Weyland 2006: 7), pensions, or
even cash transfers for outsiders that multilateral institutions promoted
or sought to impose on developing countries. At the same time, early ini-
tiatives such as Brazil’s social assistance pensions or the universal health
system (SUS) were not a source of inspiration for policy makers in the
other countries under investigation. Proponents of social programs gen-
erally paid attention to benefits for insiders within their own countries
16
On these mechanisms, see the classic literature on diffusion: Heclo (1974); Meyer and
Rowan (1977).
17
Author’s interview, World Bank office, Buenos Aires.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
15

Including Outsiders in Latin America 15

only, or to multiple initiatives rather than to a particular model of social

policy for outsiders.
Regarding the policy outcome, this study found no evidence of conver-
gence around a common model of expansion. There is remarkable variation
in the scope of coverage, benefit levels, and the form of implementation
of social programs across cases. Funding mechanisms also diverge sharply
across countries, with some programs being tax-financed and others funded
by cross-subsidies from social security benefits. New income support pro-
grams, which have become highly visible interventions cross-nationally, are
a good example; they generally employ similar tools in the form of cash
transfers. Yet these cash transfer programs also vary significantly across crit-
ical features of social policy that are relevant to this study. Some transfers
are conditional on school attendance while others are not; some are tem-
porary while others are permanent; and, as detailed in subsequent chapters,
these programs vary in scope, benefit level, and funding mechanism.18 Aside
from the specific tool used, these programmatic features are important defi-
nitional elements of social policy models, and they imply different welfare
and political effects. The absence of convergence in these key policy features
undermines the explanatory power of a theory based on diffusion.
Finally, theories of diffusion have a hard time explaining instances of
non-adoption. Why did Venezuela not adopt broad-reaching transfers
despite political rhetoric favoring the poor and despite available eco-
nomic resources to launch large-scale policies? Given that technocrats
constitute important “carriers” of best practices and models lauded by
epistemic (knowledge-based) communities and are critical sources of
expertise and advice for governments and parties (Murillo 2009: 35–6;
Weyland 2006: 46–7),19 why did Peru not launch large-scale social pol-
icy expansion in the twenty-first century despite high levels of economic
growth and the power of technocrats within the national government
(Dargent 2011)?
With respect to policy principles, it is difficult to attribute expansion-
ary decisions to the spread of a particular principle such as universality
or fairness. In the cases analyzed in this book, debates and consensus
about fairness, universality, and the need to provide social policy benefits
to the excluded existed among policy experts and some policy makers

18
On variation in income transfers, see Garay (2016).
19
On the classic literature on diffusion and emulation, see Heclo (1974); Meyer and
Rowan (1977); on sociological institutionalism, see Thelen (1999). For diffusion in Latin
America, see Collier and Messick (1975).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 16 Social Policy Expansion in Latin America

long before the expansion of social policies occurred, without resulting

in policy adoption.20
More generally, ideas do matter for policy making. Several studies
have addressed the way in which ideas favoring innovation influence the
policy process (Hall 1992, 1997; Weir 1992). Empirical cases in this book
demonstrate that program features are informed by ideas advocated by
social movements or technocrats and experts linked to political parties.
In some cases, these ideas are influenced by existing policy arrangements
for insiders, which inspire the demands of social movements who often
press for benefits similar to those received by these sectors. Ideas may
also be imported through technical assistance, technocrats’ training, or
emulation. However, though these factors are important for informing
the availability and technical soundness of specific initiatives, they have
not been sufficient on their own to drive policy makers to embark on
costly policy innovations or determine why broader or leaner benefits,
or participatory rather than direct state implementation, were adopted.

1.4 Outsiders: Political Relevance

and Social Policy Neglect
The comparative political economy literature on Western Europe defines
outsiders as workers who  – unlike insiders  – lack secure employment
(Esping-Andersen 1999; Rueda 2007: 20). In Latin America, throughout
most of the twentieth century, outsiders not only lacked secure, formal
employment, but they and their dependents were also excluded from social
security protections (e.g., health insurance and pensions) that were histori-
cally and systematically extended in the region to formal workers only.21
Outsiders represent a large share of the workforce in Latin America, includ-
ing urban informal workers (e.g., street vendors or the self-employed),
employees hired off the books, rural workers, and the unemployed.
The size and composition of the outsider population has evolved since
the first half of the twentieth century, when labor and social legislation
was first established. As Chapter  3 shows, several categories of work-
ers were formalized between the 1920s and 1940s with the creation of
labor regulations and social security (or social insurance) systems funded
mainly with payroll contributions by employers and employees. If we
measure outsiders as the share of the total population that is not affiliated

20
See Chapter 3 for further reference.
21
See Chapter 3 for full discussion.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
17

Including Outsiders in Latin America 17

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
1950s–1960s 1970s 1980s 1990s 2000s 2010s

Argentina Brazil Chile Mexico

Figure 1.4 Outsiders as a share of total population, selected countries, 1950s–

2010.
Note:  In Chile and Brazil, the share of outsider workers out of economically
active population; Argentina and Mexico, the share of outsider population out
of total population.
Source:  Author’s estimates with government sources and secondary literature
(Appendix 1).

to social security, we see that this share declined at different rates with
industrialization and the growth of public-sector employment between
the 1950s and 1980s. By the early 1980s or 1990s, it reached a plateau,
making up between 40 percent and 60 percent of the total population –
including both workers and their dependents – in the cases under inves-
tigation (Figure  1.4). The size of the outsider population did grow at
different moments during crises, but usually on a temporary basis. Within
the workforce, informal and rural workers are the groups that generally
fall into the ranks of the unemployed during economic shocks, recessions,
and times of high inflation.22
At the same time that the relative size of the outsider population
stagnated – growing during crises – the political relevance of outsiders
increased significantly in the last quarter of the twentieth century.
Beginning in the late 1970s, Latin America witnessed the longest, broadest,
and deepest wave of democratization in the region’s history, which
afforded outsiders political weight in different ways. First, compared to
previous episodes of democratic rule, outsiders gained often unprecedented

22
On the informal economy, see Beccaria and Maurizio (2003); Schijman and Dorna
(2012); Portes and Hoffman (2003).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 18 Social Policy Expansion in Latin America

access to the vote, made possible by the elimination of prior restrictions

on suffrage in countries in which these restrictions existed (e.g., literacy
requirements). Second, democratic politics also created a more favorable
environment for political organization and participation in protests, and
raised the cost for incumbents to repress social movements.23 These two
conditions made the outsider population electorally more attractive and
consequential for party leaders, and created incentives for social activism
and organization among outsiders to emerge and endure.
Following the democratic transitions in each country, outsiders dis-
played two fundamental features that under particular circumstances
incentivized governments to reach out to them with large-scale social
policies: political relevance and social policy neglect. The argument intro-
duced next and fully developed in Chapter  2 explains why, when, and
how within this environment incumbents decided to “include outsiders,”
that is, to provide them with nondiscretionary benefits across social pol-
icy areas characterized by deep divides separating insiders from outsiders.

1.5 The Argument in Brief

In explaining the two questions of why, after decades of neglect, some
incumbents expanded nondiscretionary provisions to outsiders while oth-
ers did not, and why there is remarkable variation in the shape of social
policies that were adopted, this study provides an analytical framework
that emphasizes the role of factors likely to emerge within democratic
politics. Regarding the first question, I argue that incumbents expanded
social policy when faced with high levels of electoral competition for the
vote of outsiders and/or with large-scale social mobilization by coali-
tions of social movements and labor unions that pushed for social policy
expansion through the use of protest, institutional channels, or alliances
with the governing party. In the face of these pressures, incumbents con-
sidered social policy expansion (a) a powerful instrument to elicit outsid-
ers’ electoral support when a credible challenger threatened to defeat the
incumbent party by courting outsider voters, and (b) necessary to miti-
gate intense social mobilization. When democracies lack the incentives
created by these two conditions – electoral competition for outsiders and
social mobilization from below – incumbents are less likely to embark on
the expansion of large-scale nondiscretionary social benefits.

23
See, for example, Houtzager (1998), Huneeus (2000), and Novaes (1991), and references
in Chapter 3.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
19

Including Outsiders in Latin America 19

Dynamics of expansion therefore feature two politically driven

processes: one “from above,” motivated by electoral competition for
outsider voters that credibly threatens incumbents’ continuity, as exem-
plified by the episodes of expansion in Chile and Mexico in the late
1990s and 2000s and Brazil in the 2000s; and one “from below,” pro-
pelled by social mobilization, characteristic of Argentina in the 2000s
as well as the episodes of policy expansion in Brazil in the late 1980s
and early 1990s. Absent these conditions, large-scale nondiscretionary
social policy expansion is less likely to take place, despite the fact that
outsiders make up a large electorate and are relatively excluded from
social policy.
Regarding the second question of why there is striking variation in
the policy models adopted (considering the scope of coverage, benefit
levels, and the presence of participatory or state-centric implementation),
I argue that negotiations over policy design, considering those involved
in negotiations, their preferences, and their institutional power, yielded
distinct models of social policy. Schematically, incumbents may negotiate
policy design either (a) exclusively with the parties in Congress, or (b)
with social movements in addition to, or instead of, parties in Congress.
In the presence of social movements pushing for expansion, it is likely
that incumbents negotiate with them or respond strategically to their
demands.
When expansion is negotiated among parties in Congress, those
negotiations are likely to produce policy outcomes that accommodate
the preferences of different parties regarding social policy. Parties’ influ-
ence in turn is shaped by their institutional power. When conservative
and center-right parties are strong (that is, parties that represent higher-
income sectors, as discussed in Chapter 2), resultant policies are generally
restrictive; in consonance with the preferences of elites, and often with
these parties’ programs as well, they provide relatively small or moder-
ate coverage through direct state implementation and do not protect the
full outsider population. In Mexico and Chile, the processes of social
policy design occurred in the context of high electoral competition (with-
out social mobilization from below); they included negotiations among
congressional parties and were shaped by the balance of partisan power.
Given the power of conservative parties in both cases, these negotiations
resulted in restrictive models.
If incumbents negotiate expansion with social movements or respond
strategically to these movements’ demands, policy outcomes tend to
result in inclusive benefits, with broad coverage and participatory

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 20 Social Policy Expansion in Latin America

implementation. Large-scale coalitions of social movements and union

allies, which, as discussed previously, were the coalitions driving incum-
bents to launch expansion in some cases, generally demand both broad
benefits – similar to those received by low-income insiders – and partici-
pation in policy implementation.
Social movements generally gain access to the policy design process
when they have catalyzed expansion through protest or pressure in insti-
tutional arenas (e.g., via lobbying or public opinion campaigns), or when
they are allied to the governing coalition and exercise influence over
policy choices from within the government (even if they are not able to
propel on their own the initial decision to launch massive benefits). As
long as powerful social movement coalitions that push for expansion
are involved in the process of design, resultant policies are more likely
to take on inclusive features. In Argentina, social movements and labor
allies have engaged in social policy making by pushing their social policy
proposals through protest and – sometimes also – through institutional
channels and negotiating with policy makers over policy design. In Brazil,
social movements and labor allies primarily used institutional channels
to advance their social policy agenda for expansion and/or to influence
policy design. As a result, inclusive policies were adopted in these two
country cases.

1.6 Research Design, Cases, and Data

This study provides the first comprehensive analysis of social policy
expansion in four of the largest and most industrialized economies of
Latin America – Argentina, Brazil, Mexico, and Chile – in three policy
areas: pensions, health care, and income support since the beginning of
the democratic transitions in the 1980s until 2010. To identify the factors
and mechanisms driving social policy change, I have conducted process
tracing of social policy making across all different administrations in each
case, covering a total of twenty-one administrations, analyzing instances
of successful social policy expansion as well as failed attempts to create
large-scale nondiscretionary benefits for outsiders.24 The empirical analy-
sis is based on original measures of outsiders and social benefits and new
data sets of electoral dynamics and social mobilization constructed for
this project, as well as archival research and extensive interviews with key
informants, as detailed later.

24
On process tracing, see Bennett and Checkel (2015) and Seawright and Collier (2004).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
21

Including Outsiders in Latin America 21

Comparative Approach and Cases

Recent comparative studies of social policy in Latin America have focused
generally on a single social policy across countries, on the evolution of
social policy in a single country, or on large-N cross-national compari-
sons of aggregate spending.25 To better address the questions raised in
this book, I  conducted a small-N comparative study of four countries
across several administrations and three policy areas in each country.
Given the absence of substantial research on this topic, this combina-
tion of longitudinal within-case and cross-country perspectives enables a
comprehensive assessment of different theoretical insights and of poten-
tial explanatory factors underlying policy change, and allows for sub-
stantial theory development.
Within each country case, I  employ process-tracing analysis of the
policy-making process for each episode of policy expansion as well as
for instances of unsuccessful attempts to expand social policy. Careful
process tracing of policy making enables both identifying the factors and
the specific causal mechanisms shaping expansion and assessing alterna-
tive explanations by presenting detailed evidence on their role in policy
processes. Process tracing also facilitates the generation of credible and
detailed information on social policy, which is notably absent in large-N
studies of social spending in Latin America.26
This comparative design involving cross-country and within-country
comparisons across different administrations and policy areas also allows
for an exploration of the effects of alternative national-level and policy-
specific conditions that may be driving observed outcomes. Moreover,
a longitudinal analysis explicitly takes into account temporal dynamics
that affect expansion – especially time lags between policy adoption and
implementation – and avoids “excessive causal proximity” in the analysis
of processes that may unfold over longer periods of time (see Grzymala-
Busse 2002: 16).

25
On single-country studies, see Díaz-Cayeros, Estévez, and Magaloni (forthcoming). On
large-N analysis using social expenditure see Huber, Mustillo, and Stephens (2008);
Rudra (2002); Rudra and Haggard (2005); Segura-Ubiergo (2007).
26
Several studies of social expenditure use IMF expenditure data that a) only capture
direct national-level spending and thus severely underestimate social service expendi-
ture in federal systems, and that b) group spending by theme into two categories  –
social services, and social security and welfare – which prevents students from observing
variations across individual policy areas. See, for example, Carnes and Mares (2009);
Haggard and Kaufman (2008); Kaufman and Segura-Ubiergo (2001); Segura-Ubiergo
(2007).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 22 Social Policy Expansion in Latin America

Main Cases: Argentina, Brazil, Chile, and Mexico

This case selection maximizes variation in potential explanatory factors
while holding constant certain background conditions. In terms of simi-
larities, Argentina, Brazil, Chile, and Mexico are among the most industri-
alized countries in Latin America, and together with Colombia, Uruguay,
and Venezuela, they have the highest income per capita in the region.27
In the 1990s, all four countries adopted market-oriented reforms associ-
ated with neoliberal ideas of social protection.28 These four countries,
moreover, had experienced extensive periods of authoritarianism and/
or limited democratic competition since the 1920s, when social security
systems for formal-sector workers were first adopted in some of these
countries. Finally, following the creation of labor market regulations and
social benefits for insiders, labor movements in all four countries were
among the most powerful in the region (see Collier and Collier 1991).
Despite these similarities, there is theoretically relevant variation in
potential explanatory factors across cases and over time throughout the
period under examination, from the 1980s through 2010. First, the cases
differ in terms of the presence of social movement coalitions (SMCs) –
alliances of social movements and labor unions demanding social pol-
icy for outsiders – in the policy areas under investigation. These SMCs
emerged in Brazil and Argentina but not in Chile and Mexico. Second,
the composition and level of institutionalization of party systems vary
across countries and over time. Though still fluid, since the third wave of
democracy and until 2010, the party systems in Chile and Mexico, and
increasingly in Brazil, became more institutionalized, with rather stable
partisan coalition options at the national level, whereas the party system
in Argentina experienced a partial collapse in the early 2000s (see Lupu
2014; Roberts 2014). Third, partisan competition for the votes of outsid-
ers took place in Chile and Mexico since 1999 and 2000, respectively,
but occurred in Argentina and Brazil only in 1999 and 2002, respectively.
Economic conditions, such as financial stability, unemployment, and
the existence of a major economic crisis or windfall economic resources,
also vary across cases and over time, which allows for assessment of their
weight on decisions to expand. There are contradictory expectations in

27
The most industrialized countries in the region are Argentina, Brazil, Chile, Mexico,
Uruguay, and Venezuela. These are also the countries with the highest GDP per capita,
and the ones that have developed systems of social protection for insiders since the first
half of the twentieth century. At the same time, they comprise close to 75 percent of the
population in the region (see CEPAL 2009).
28
See Brooks (2001); Madrid (2003); Weyland (2004).

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
23

Including Outsiders in Latin America 23

the literature regarding these economic indicators. While economic crises

may affect a government’s ability to fund new benefits, they may also
produce more need and demand. As shown in the chapters that follow,
when combined with electoral competition or mobilization, crises may
intensify the level of mobilization or increase the chances of a challenger
seeking to win office.
Furthermore, variation in the size of the outsider population across
the four cases provides a two-pair comparison displaying relatively lower
and higher levels of exclusion. Between 1990 and 2010, Argentina and
Chile had outsider populations of similar – and relatively smaller – sizes,
ranging from 40 percent to less than 50 percent of the total population.
Mexico and Brazil, by contrast, had relatively larger outsider popula-
tions, ranging from more than 50 percent to 60 percent of the total popu-
lation, an important share of which resided in rural areas. This two-pair
comparison permits assessment of the effect of structural conditions on
social policy change. Did the countries with smaller outsider populations
build more generous and broad-reaching benefits? As we see in the fol-
lowing chapters, resulting social policy models do not map onto preex-
isting structural and socioeconomic characteristics, with Argentina and
Brazil, and Mexico and Chile, respectively, building systems more similar
to each other.

Broader Comparisons
In Chapters 3 and 8, the main findings of this book are further assessed
against two additional comparison sets of cases: (a) a longitudinal analy-
sis of the four cases under investigation since the establishment of ben-
efits for insiders in the 1920s and 1940s, and (b) a broader cross-country
comparison within and beyond Latin America.
Longitudinal Comparison. In Chapter 3, I examine the period that starts
with the creation of benefits for insiders in the 1920s and 1940s and
that ends with the recent adoption of large policies for outsiders. This
comparison has two main goals. First, I  seek to assess the explanatory
power of this study’s framework against a period in which most ben-
efits were extended to insiders and outsiders were largely neglected.
Second, this longitudinal comparison also permits the analysis of varia-
tion in social policy provisions incumbents extended to outsiders under
nondemocratic political regimes and different economic circumstances.
Specifically, I investigate why incumbents in some authoritarian regimes
provided some benefits to outsiders while others did not introduce social
policy innovations, and I analyze the specific features of those provisions

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 24 Social Policy Expansion in Latin America

(e.g., scope, benefit levels, nondiscretionary access) and why they were
adopted.
Cross-country Comparison. The second broader comparison involves
three additional middle-income countries in the region – Peru, Uruguay,
and Venezuela – as well as South Africa, a middle-income country with
high levels of economic inequality, which has been analyzed in compara-
tive political economy and public policy studies together with middle-
income countries of Latin America (see Lieberman 2003, 2009; Seidman
1994). These four cases display broad variation across both explanatory
factors and outcomes. Venezuela and Peru allow us to assess social policy
dynamics in cases without significant expansion of large-scale nondiscre-
tionary benefits, and without a strong presence of the key factors pro-
pelling incumbents to expand. Uruguay in turn exemplifies dynamics of
expansion propelled by electoral competition for outsiders and exhibits
intertemporal variation in the process of policy design, with conserva-
tives dominating policy making in the early expansions and social move-
ments engaging in the process of expansion through an allied party at a
later stage. Finally, South Africa provides an excellent case for assessing
the applicability of the argument in a context of no electoral competi-
tion for outsider voters. In this particular case, social movements allied
to a dominant party played a prominent role in prompting adoption and
shaping policy in specific areas.

Data Sources and Analysis

This book’s comparative analysis of social policy processes relied on
original data from multiple sources gathered for this project. I collected
social policy data from policy documents and public archives in all four
of the main country cases, and used them to create new measures of social
policy. I  conducted interviews with approximately 265 key informants.
These included leaders of social movements, labor unions and rural orga-
nizations, policy makers who were directly involved in policy expansion
(including secretaries, ministers, and former presidents), legislators, party
leaders, politicians and candidates involved in presidential campaigns,
representatives of employers’ associations, pension funds, and private
providers in the health care sector. These interviews helped to recon-
struct the process of policy expansion and policy design, its underlying
motivations, and the goals and policy preferences of the different actors
involved. Transcripts of congressional sessions, party documents and cam-
paign platforms, as well as petitions and documents produced by social

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
25

Including Outsiders in Latin America 25

movements and congresses of labor confederations, further contributed

to understanding the motivations, demands, and expectations regarding
expansion, the coalitions or alliances they formed to achieve or oppose
expansion, and the negotiations over the shape of social programs.
In addition, I constructed a database of episodes of social policy mak-
ing based on content analysis of newspaper articles referring to poli-
cies for outsiders from 1989 through 2011 in Argentina, 1987 through
2006 in Brazil, 2002 through 2008 in Chile, and 1988 through 2007 in
Mexico. For each country, I surveyed at least one national newspaper:
Clarín in Argentina, El Mercurio in Chile, and La Jornada in Mexico.
For Brazil, I worked with an index built by the Library of the Brazilian
Senate with seven national newspapers.29 For the cases of Mexico and
Argentina, I also consulted Reforma and La Nación, respectively, for
parts of the period under investigation. Finally, I created a Data Set of
Protest for Argentina that maps the evolution of social mobilization for
social benefits, which was led primarily by unemployed workers and their
labor union allies from 1996 through 2010. This data set includes close
to 2,000 protest events, as well as information on these events concern-
ing their duration, location, participants, demands, state responses, and
violence.
Collecting data systematically from different sources afforded the
opportunity for triangulating information, a fundamental aspect of pro-
cess tracing in qualitative research. The database of policy-making events
further helped inform the examination of government documents and
the conducting of interviews, and ultimately produced more detailed and
accurate chronologies of social policy processes – another critical aspect
of process tracing.
To characterize the evolution of electoral competition for outsiders,
I used available quantitative data, especially electoral surveys (from sur-
vey firms and academic institutions), as well as ecological data. With
poll data I identified outsider and insider voters and mapped their voting
patterns in Brazil, Chile, and Mexico since the first democratic election
(comparable national data were not available for Argentina). To comple-
ment these measures, I constructed measures of electoral competition for
outsiders for each selected country since the first democratic presidential
elections until about 2010–11, identifying districts in which outsiders

29
Folha de São Paulo, O Estado de São Paulo, Jornal do Brasil, Jornal da Tarde, Jornal de
Brasília, Correio Brasiliense, and O Globo. Some articles from other newspapers have
also been used.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
 26 Social Policy Expansion in Latin America

are the majority of the population, and measuring levels of competition

in presidential elections in those districts. For some countries and years
I  relied on existing election data sets constructed by other researchers,
while for others, I used government documents and archival research to
construct new data sets of elections (see Appendix 3 for further informa-
tion on these data sets).

1.7 Structure of this Book

This book is organized into eight chapters. Chapter 2 presents this book’s
analytical argument, which accounts for the expansion of social policy
for outsiders and for the different models these policies took. It intro-
duces the conditions that lead incumbents to expand nondiscretionary
social policy and the factors that shape policy design along restrictive
and inclusive models. It also considers the strategies incumbents use to
overcome opponents and to obtain funding for new programs. Chapter 3
adopts a longer historical perspective, drawing on the analytical frame-
work to understand why large-scale, nondiscretionary social programs
were not expanded to outsiders prior to the expansions documented here.
It further analyzes the kinds of policies created for outsiders in previous
decades and identifies the conditions under which governments: (a) did
not expand any benefits for outsiders, (b) expanded large, discretionary
(often temporary) benefits, or (c) created small benefits for outsiders. The
time period begins with the creation of the first large-scale social programs
for the formal workforce in the 1920s and 1940s through the 1980s or
1990s, depending on when large-scale expansion began in each country.
In this chapter, I demonstrate that governments provided little social pro-
tection for outsiders during this period because of the few instances of
democracies that had mobilization from below or electoral competition
for outsiders. The two cases in which social protection was extended to
outsiders – the inauguration of health care services in Argentina in the
late 1940s and in Chile in the 1960s – occurred in the context of electoral
competition for outsiders.
Chapters 4 through 7 draw on this analytical framework to explain
social policy expansion and the adoption of inclusive social policy for
outsiders in Argentina and Brazil, and restrictive social policy in Mexico
and Chile. Through an in-depth analysis of social policy making in each
administration since democratization, I  show that incumbents, irre-
spective of partisan affiliation, were likely to expand social policy in
response to high electoral competition for outsiders and/or mobilization

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001
27

Including Outsiders in Latin America 27

from below pressing for social benefits. I further show that governments
adopted different models of social policy depending on whether incum-
bents negotiated policy design with the congressional opposition or
whether they (also) responded to social mobilization and granted social
movements access to the process of policy design. The concluding chapter
extends the argument to a broader comparison of middle-income coun-
tries and assesses the social and political effects and theoretical implica-
tions of these social policy transformations.

Downloaded from https:/www.cambridge.org/core. Columbia University Libraries, on 25 Jun 2017 at 20:33:49, subject to the Cambridge
Core terms of use, available at https:/www.cambridge.org/core/terms. https://doi.org/10.1017/9781316585405.001