ANRV377-PL12-06 ARI 13 April 2009 12:43

ANNUAL
REVIEWS Further Social Policy in Developing
Click here for quick links to
Annual Reviews content online,
including: Countries
• Other articles in this volume
• Top cited articles
• Top downloaded articles Isabela Mares1 and Matthew E. Carnes2
• Our comprehensive search
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

1
Department of Political Science, Columbia University, New York, New York 1007;
email: [email protected]
2
Department of Government, Georgetown University, Washington, District of Columbia
20057; email: [email protected]
by University of Sussex on 06/29/12. For personal use only.

Annu. Rev. Polit. Sci. 2009. 12:93–113 Key Words

The Annual Review of Political Science is online at social protection, welfare state, reform
polisci.annualreviews.org

This article’s doi: Abstract

10.1146/annurev.polisci.12.071207.093504
Drawing on recent work and data on social protection in the develop-
Copyright  c 2009 by Annual Reviews. ing world, this essay evaluates the current state of the art and suggests
All rights reserved
several important new lines of research. We first examine the historical
1094-2939/09/0615-0093$20.00 origin and evolution of social protection systems in developing coun-
tries, arguing that insufficient attention has been paid to the authoritar-
ian roots of developing nations’ social policy. As a preliminary effort to
remedy this shortcoming in the literature, we offer a political logic for
the observed variation in the character of institutions of social policy es-
tablished by nondemocratic regimes. Next, we explore recent research
examining linkages between models of economic development and wel-
fare regimes in developing countries. Finally, we turn to the study of the
political determinants of the social policy reforms that occurred in the
final decades of the twentieth century, arguing that variation in reform
across policy areas has been more complex than is generally appreci-
ated in the literature. To explain this variation, we develop a theory that
identifies the political coalitions supporting different policy outcomes.

93
 ANRV377-PL12-06 ARI 13 April 2009 12:43

INTRODUCTION Latin American countries, governments have

Our understanding of the development of poli- reneged on multiple occasions on many of the
cies of social protection is uneven across the social policy promises enshrined in their leg-
regions of the world. We know a great deal islation. In both regions, highly uneven social
about the causes and distributional implica- protection resulted because governments either
tions of social policies in developed countries lacked the administrative capacity to enforce
(Esping-Andersen 1990, Huber & Stephens contributions to social insurance or deliberately
2001). By contrast, our grasp of the variation in chose to manipulate social insurance to benefit
the design and economic consequences of so- particular sectors.
cial policies in developing economies is sketchy A second shortcoming of the existing
and preliminary. Although the study of social research is the absence of systematic cross-
policies in countries outside the OECD is a national and temporal data on the implementa-
rapidly growing area of research (Wong 2003; tion and design of social policies. This has hin-
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

Weyland 2005, 2007; Segura-Ubiergo 2007; dered our understanding of the distributional
Brooks 2008; Haggard & Kaufman 2008), po- impact of social policies and their effects on eco-
litical science as a field is far from understanding nomic and labor market outcomes, such as the
by University of Sussex on 06/29/12. For personal use only.

the variation in the character of social protec- level of employment, overall labor force par-
tion and the political factors that have caused ticipation rates, and economic growth (Mares
these outcomes. 2007a). Consider for example the relationship
A first striking indicator of the incomplete- between the level of social protection and the
ness of our knowledge is the absence of ba- size of the informal economy. Since the influ-
sic descriptive information regarding the cross- ential paper of Harris & Tadaro (1970), social
national and temporal variation in the level of security contributions and other labor market
social protection. In many developing coun- regulations have been argued to be the main
tries, the introduction of social policy legis- cause of labor market “dualism” in many devel-
lation occurred in the early or middle twen- oping countries, segmenting employment be-
tieth century, yet we still lack reliable data tween secure, higher-paid jobs in the formal
on the implementation of these policies, the sector and precarious, lower-paid informal jobs.
scope of coverage, or the levels of benefits. In High social security contributions and above-
some cases, social policy legislation provided market-clearing wages were theorized to force
genuine protection to many citizens for vari- workers to queue for jobs in the formal sec-
ous employment-related risks. In other cases, tor and subsequently shunt many workers into
however, such measures were no more than the informal economy. However, a growing
promises that were not fulfilled or that were amount of empirical evidence raises significant
implemented in a highly discretionary fash- doubts about the explanatory power of this re-
ion. For example, according to data on “offi- ceived wisdom. For example, over the past two
cial” levels of social protection recorded by the decades, lack of enforcement of minimum-wage
Social Security Administration of the United legislation in many Latin American countries
States, 40 sub-Saharan countries had policies of has resulted in a tremendous downward flex-
old-age and disability insurance in place, with ibility of wages in the formal sector. At the
“promised” levels of social protection roughly same time, many social insurance programs
comparable to that found in economies with have been dismantled or have experienced se-
much higher levels of economic development vere cutbacks. Yet despite these changes, we see
(Social Security Administration 1999). Yet this no decline in the size of the informal economy
information on social policy commitments is across the region. By 2005, the best estimates
clearly contradicted by case studies of the im- show that 50% of all salaried workers in Latin
plementation of these policies. Similarly, in America continued to work informally (Galiani

94 Mares · Carnes
 ANRV377-PL12-06 ARI 13 April 2009 12:43

& Weinschelbaum 2006). These results call into countries that covers 14–18 OECD economies.
question the expected positive relationship be- In this limited universe of cases, the theo-
tween social insurance expansion and informal- retical variables that are considered important
ity, leaving us with no convincing alternative determinants of the trajectory of social pol-
theory of the factors that cause variation in the icy development—such as the strength of left-
size of the informal sector. wing partisanship, the organizational density of
Third, many developing countries have ex- business and labor, and the type of electoral
perienced tremendous social policy change system—are highly correlated with each other.
in recent decades. Countries as diverse as As a result, many quantitative studies of the de-
Thailand and Bolivia have enacted a wide range terminants of social protection in advanced in-
of social policy reforms—modifying the level of dustrialized countries leave out important ex-
insurance coverage, the eligibility conditions, planatory variables and are unable to test all
and the mix between private and publicly pro- relevant competing theories against each other.
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

vided benefits received by their citizens. How- Broadening our sample size by including de-
ever, the direction of these changes is still in- veloping countries allows us to overcome prob-
sufficiently understood (Mares 2007b). Have lems of multi-collinearity and to include a larger
by University of Sussex on 06/29/12. For personal use only.

welfare states around the world been driven number of potential explanatory variables in
by the competitive concerns of globalization to our estimations. A study of policies of social
adopt “residualistic” models, characterized by protection with a broader empirical scope can
narrow social insurance coverage and a strong have important theoretical externalities.
reliance on private-type institutions of social Drawing on recent work and data on so-
protection? Do we find significant differences cial protection in the developing world, this
across policy areas in the direction and mag- article evaluates the current state of the art
nitude of policy change? Have some particular and suggests several important new lines of re-
subsystems of the welfare state been more vul- search. We first examine the historical origin
nerable to retrenchment than others? Although and evolution of social protection systems in
the study of the recent politics of welfare state developing countries, arguing that insufficient
adjustment in developing countries has been a attention has been paid to the authoritarian
rapidly growing area of research in recent years, roots of developing nations’ social policy. As
many puzzles remain. We have an insufficient a preliminary effort to remedy this shortcom-
understanding of the variation in the distribu- ing in the literature, we offer a political logic
tional implications of policy changes enacted in for the observed variation in the character of
recent years. Existing studies do not adequately institutions of social policy established by non-
explain the strong divergence across policy ar- democratic regimes. Next, we explore recent
eas (often within the same country); we need research examining linkages between strategies
an explanation that is grounded in a coherent of economic development and welfare regimes
micro-logic and can account not only for the in developing countries. This represents a novel
preferences of political and bureaucratic elites agenda of research, paralleling the latest re-
but also for the preferences of organized groups search on social policy in advanced industrial-
and mass publics. ized societies. However, it requires better speci-
Fourth, the empirical bias that has char- fication of its micro-foundational logic. Finally,
acterized the existing literature on compara- we turn to the study of the political determi-
tive systems of social protection has limited the nants of the social policy reforms that occurred
efforts of scholars to test competing explana- in the final decades of the twentieth century,
tions against each other. Most of the explana- arguing that variation in reform across policy
tions that seek to account for cross-national dif- areas has been more complex than is gener-
ferences in the level of social protection have ally appreciated in the literature. To explain this
only been tested in a narrow subsample of variation, we develop a theory that identifies the

www.annualreviews.org • Social Policy in Developing Countries 95

 ANRV377-PL12-06 ARI 13 April 2009 12:43

political coalitions supporting different policy tensive social policy commitments. The logic
outcomes. is that competition among political candidates
will lead to social policy expansion to reach new
groups of voters. Likewise, democratic free-
THE HISTORICAL ORIGIN dom permits interest group organization and
OF SOCIAL POLICIES IN pressure for the increase of social spending
DEVELOPING COUNTRIES: (Haggard & Kaufman 2008). Although some
THE UNEXPLORED POLITICAL disagreement exists (Mulligan et al. 2003), the
LOGIC OF AUTHORITARIAN quantitative literature seems to support this
GOVERNMENTS contention that democracies spend more than
The first generation of research examining the nondemocracies on particular social programs
variation in social policy has explored the im- (Przeworski et al. 2000, Lake & Baum 2001,
pact of underlying structural variables—such as Avelino et al. 2005).
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

the level of economic development or the level Within democracies, we now know a great
and speed of industrialization—on the level deal about patterns of social spending and pol-
of social spending. For economists, the mech- icy, especially in developed economies. For ex-
by University of Sussex on 06/29/12. For personal use only.

anism linking higher economic development ample, left-wing, labor-associated governments

and higher levels of social spending goes back are more likely to undertake greater social
to Wagner’s observation that spending on so- spending (Esping-Andersen 1990, Huber &
cial transfers rises as incomes rise. This pos- Stephens 2001); yet, these effects are more
itive income elasticity of government trans- pronounced in the formative years of labor-
fers is expected to account both for the steady associated parties than in the later, reform
growth of social spending and for the observed period—in which these parties have often pro-
cross-national variation. Political scientists, by tected their core constituencies of “insiders” at
contrast, emphasize the new needs and vulner- the cost of the broader working class (Rueda
abilities in the economy created by the pro- 2005). On the other side of the political spec-
cess of industrialization. Early empirical work trum, employers and associated parties on the
on OECD countries found a positive correla- right are not univocally opposed to social
tion between the level of industrialization or policy expansion. Under certain conditions—
economic development and aggregate social such as increased economic competition or
spending (Wilensky 1975). However, the rela- the presence of industrial risks in the produc-
tionship between economic development and tion process—firms may support government-
the size of the government or the level of gov- administered social protection schemes and
ernment revenues does not prove to be ro- thus enter into cross-class alliances with work-
bust in a broader sample that includes devel- ers (Mares 2003, 2005a; Swenson 2002).
oping countries (Adsera & Boix 2002, Mares However, one significant weakness of this
2005a). Economic development does not seem literature is its binary distinction between
to explain much of the variation in spend- democracy and nondemocracy. This blunt di-
ing among developing countries. The short- chotomy misses the important variation in so-
comings of these variables point to the need cial spending and social policy design within
to consider political determinants of social regime types. More fundamentally, even when
policy. variation in regime types appears to account for
Among the latter, perhaps the most stud- differences in the growth of social policy ex-
ied question in the literature has been the re- penditures, it cannot explain the political ori-
lationship between regime type and the size of gin of social insurance programs. As Esping-
the government. Typically, this work has argued Andersen (1990, p. 15) pointedly remarks, “the
that democracies, owing to their need to cater thesis that democracy leads to larger welfare
to a broad electoral base, will have more ex- states confronts the historical oddity that the

96 Mares · Carnes
 ANRV377-PL12-06 ARI 13 April 2009 12:43

Table 1 Regime type at time of first adoption of social insurance policies, by region∗
Region Regime type Old age Disability Sickness Unemployment
Latin America authoritarian 12 16 13 0
democratic 2 3 1 3
Advanced industrialized nations authoritarian 7 8 7 2
democratic 15 14 13 19
Eastern Europe authoritarian 9 7 8 2
democratic 0 2 1 2
Africa authoritarian 3 4 3 1
democratic 0 0 0 0
Middle East authoritarian 5 8 3 –
democratic 1 0 0 –
Asia authoritarian 4 9 2 –
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

democratic 2 2 3 –
TOTAL authoritarian 40 52 36 5
by University of Sussex on 06/29/12. For personal use only.

democratic 20 21 18 24

∗
Sources: Flora & Heidenheimer 1981; Mares 2005a; U.S. Social Security Administration, Social Security Programs
Throughout the World, various years.

first major welfare state initiatives occurred under autocracies. We find the opposite pat-
prior to democracy and were powerfully moti- tern when we move to different regions of the
vated to arrest its realization.” In the European world. Disability insurance, to cover workplace
context, nondemocratic regimes—such as those injuries that could leave workers incapacitated,
of Bismarck in Germany and von Taaffe in was often the first policy adopted and is the
Austria—pioneered social insurance legislation, most prevalent policy cross-nationally. How-
and the expansion of social programs often oc- ever, of the 73 countries included in this sample,
curred under conditions of limited suffrage. >70% were autocracies at the time of adoption.
One quantitative analysis of the evolution of so- And when the advanced industrialized nations
cial spending in Europe during the last decades are removed from the total, fully 88% of de-
of the nineteenth century and the first two veloping nations adopted disability insurance
decades of the twentieth century suggests that under authoritarian rule. Both old-age insur-
variables capturing differences in the type of ance and sickness insurance show similar pat-
democracy can account for only very small dif- terns, with twice as many autocratic nations in-
ferences in social spending (Lindert 1994). stituting these programs as democracies. The
In the case of developing countries, regime only departure from the pattern of authoritar-
type even more strikingly fails to account for ian first-adoption has been unemployment in-
differences in the origin of social programs. surance. Far more limited cross-nationally, un-
Here, the overwhelming number of social in- employment insurance is largely restricted to
surance programs were initially adopted by the advanced industrialized nations, and thus
nondemocratic governments. Table 1 presents was adopted primarily under democracies.
data on regime type at the time of the adoption Nondemocratic regimes have adopted so-
of different social policies. Several regional pat- cial insurance programs that vary dramatically
terns stand out. In Europe and North America, in the scope of insurance coverage, the ag-
the majority of policies were adopted by democ- gregate level of benefits, and the composition
racies. Only one third of the earliest measures of those benefits. In some cases, authoritarian
(disability, old age, and sickness) were instituted governments have designed policies to serve

www.annualreviews.org • Social Policy in Developing Countries 97

 ANRV377-PL12-06 ARI 13 April 2009 12:43

extremely narrow political clienteles and have Turning our attention to the relationship
been able to maintain this narrowness of cover- between autocratic regimes and their popula-
age despite rapid economic change and grow- tions has significant predictive power regarding
ing affluence. For example, prior to democrati- the design of, and spending on, social welfare
zation in Taiwan, the major old-age insurance policies. In doing so, we build on recent work
program—the Labor Insurance Act—provided that has sought to better specify the political
employment-related benefits to <10% of the economy of autocracy, noting the dilemma au-
labor force. In contrast, other authoritarian thoritarian rulers face in remaining in power
regimes have chosen to enact universalistic over a population whose support can never be
health care and education systems. For exam- completely trusted (Wintrobe 1998). In partic-
ple, in Brazil in 1971, the military government ular, the group that maintains the autocrat in
of Emilio Medici presided over a dramatic ex- power, which can also credibly mount a chal-
pansion of social insurance in the countryside lenge to her rule, has become central to “selec-
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

(Malloy 1979). torate” theories of political survival (Bueno de

Theories linking regime type and social pol- Mesquita 2003). Organizations strong enough
icy thus need to be grounded in an explana- to put an autocrat in power must also possess
by University of Sussex on 06/29/12. For personal use only.

tion of the internal political dynamics of these the resources to end her rule. Three distinct
regimes. Unfortunately, prevailing theories of political equilibria exist as solutions to the “au-
autocracies offer very limited guidance in this tocrat’s dilemma,” the challenge of preserving
effort. Much of the theoretical work on autoc- her rule in the face of her launching organiza-
racies tends to be taxonomical; it has generated tion (Haber 2007).
a long list of possible nondemocratic regime In a first scenario, the leader attempts to
types (O’Donnell 1979, Linz 2000). These in- eradicate the power of the launching organi-
clude “neo-sultanistic, neo-patrimonial, per- zation through a strategy premised on terror,
sonal, bureaucratic authoritarian, military, in- torture, and purges. If the dictator succeeds,
clusive military, exclusive military, single party, the outcome is one of unconstrained power
dominant (or hegemonic) party, semiauthor- and discretion. Under these conditions, how-
itarian, autocratic, and totalitarian—to name ever, the dictator will be unable to commit her-
some of the most commonly used categories self not to expropriate predatorily all returns
of authoritarian government” (Haber 2007, from economic activity. The consequence is a
p. 694). This attention to the diverse institu- lack of investment in these regimes, which will
tional shapes of autocracies has proven helpful depress economic activities and tax revenues.
in understanding the specifics of particular The implications for social policy are relatively
cases, but it provides less traction in cross- straightforward: One should see little or no so-
national comparisons and cannot account for cial policy legislation enacted by these regimes.
the variation in policy output of nondemocratic Much like the classic “stationary bandit” with
regimes. To illustrate this point, consider the a short time horizon, they extract rents to the
broad variation in the social policy legislation maximum extent possible and target spending
adopted by military regimes. Some of these on their repressive apparatus (Olson 1993).
regimes have adopted policies that favor ur- The second potential strategy pursued by
ban labor, whereas others have taken a harsh authoritarian leaders in their conflict with com-
and repressive stance against industrial work- peting organizations is one of collusion. In this
ers. Still other military regimes have introduced equilibrium, the dictator seeks to prevent coups
universalistic social insurance. In short, “mili- by providing leaders of the launching organi-
tary regimes”—one critical category of prevail- zation with a stream of rents in selected eco-
ing taxonomies of autocratic regimes—is not nomic sectors, and by simultaneously restrict-
particularly helpful in explaining variation in ing the level of economic competition in the
social policy outcomes. nation. Measures such as barriers to entry, the

98 Mares · Carnes
 ANRV377-PL12-06 ARI 13 April 2009 12:43

selective allocation of import permits, and the tutions to benefit their friends and buy off their
requirement of licenses or concessions to enter potential adversaries.
particular economic activities confer an ongo- The Nationalists’ social policy legislation
ing privileged status on members of the launch- was introduced by presidential decree—rather
ing organization (Haber 2007). To be effective, than an act of the legislature—on March 1,
the dictator must also bind her own hands and 1950, the day of Chiang Kai-shek’s politi-
those of her rivals and private economic agents, cal inauguration. The centerpiece of the leg-
limiting them to a pattern of distribution of islation was the Labor Insurance Act. Cov-
rents that increases the costs of their defec- erage was limited to employees of central
tion from this arrangement. In other words, the and local government-owned enterprises and
dictator sets up institutions that alleviate rivals’ blue-collar workers in private enterprises with
fear of arbitrary expropriation and also guaran- more than ten employees. In the same year,
tee them a steady stream of rents (Haber et al. the government introduced soldiers’ insurance.
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

2003). In short, the government instituted a system

The strategy of collusion has distinct impli- that benefited its most likely challengers—
cations for the design of social policies. Con- fellow Nationalists (the launching organiza-
by University of Sussex on 06/29/12. For personal use only.

sistent with a set of economic policies premised tion), workers with the potential to organize
on the creation of monopolies, we expect to see (and the holders of private capital for whom
the establishment of “restrictive” social policies, they worked), and the military.
characterized by narrow coverage and generous A third distinct political equilibrium found
benefits. Thus, the beneficiaries of social policy in authoritarian governments is characterized
largesse will be workers in the state-owned or by organizational proliferation. In this case, the
private-sector industries on which the regime is dictator encourages the creation of competing
also conferring trade protection and other priv- organizations in order to raise the costs of col-
ileges. High levels of benefits ensure the loy- lective action by the launching organization.
alty of these workers to their leaders and their We see this logic at work in a variety of cases that
employers, reducing strikes and other labor ac- have been categorized as distinct regime types
tivism and encouraging wage moderation. Fi- by previous studies: military governments that
nancing for these generous programs is made encourage the creation of political parties or
possible by the monopoly rents these industries one-party Communist regimes that foster po-
enjoy. litical competition among rival organizations,
An example of this sort of collusive social such as the secret police and the military. Haber
policy occurred under the “immigrant regime” invokes Mexico under the PRI as the paradig-
of the Nationalist government in Taiwan. Po- matic case of this political strategy. As Haber
litically, this regime relied on the support (2007, p. 703) argues, “the key to the PRI’s suc-
of “mainland Chinese” who had immigrated cess was that it sat at the center of a network of
with Chiang Kai-shek to Taiwan during the organizations which were designed to align the
civil war. The institutions and policies put in incentives of a wide variety of organized cor-
place by Chiang Kai-shek contained strong fa- porate groups, making it extremely difficult for
voritism toward this group and an implicit bias any of them to mount a credible challenge to
against native Taiwanese, denying them partic- the party’s monopoly on power.”
ipation in the national power structure (Tsai & In sharp contrast to the collusive autocracies
Chang 1985), through a strategy often referred or to dictatorships that pursue strategies of po-
to as “internal colonization” (Hechter 1975, litical terror, regimes premised on this logic of
Grabowski 1988). Thus, even though the main- organizational multiplication confer economic
land Chinese constituted a minority in number, rights to broader segments of the population.
they enjoyed disproportionate power and were Although their resulting property rights sys-
able to set up the political and economic insti- tem is more extensive, it is also more uneven,

www.annualreviews.org • Social Policy in Developing Countries 99

 ANRV377-PL12-06 ARI 13 April 2009 12:43

as groups that have been granted some pri- cies. One expects these autocracies to enact so-
vate property rights can easily usurp the rights cial policies with two (seemingly contradictory)
of unprotected individuals. These regimes of- policy characteristics. On the one hand, the so-
ten produce a chaotic mix of economic policies cial policies will be characterized by high levels
granting selective benefits to a variety of groups. of institutional fragmentation. This is associ-
In the past century, authoritarian governments ated with pronounced inequities in the level of
that employed organizational proliferation fre- social policies. They will, most likely, resemble
quently pursued economic strategies involving the “conservative welfare regimes” of advanced
import substitution, targeting tariff policies and industrialized democracies, where we find pol-
government spending to benefit key groups of icy privileges granted to civil servants and a
local producers and workers, and thus buying large number of occupational groups (Esping-
their political support and diffusing potential Andersen 1990). On the other hand, the level
tensions. of social policy coverage found in these regimes
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

This characterization of the logic of orga- will be much broader than the coverage under
nizational proliferation pursued by the PRI in collusive autocracies.
Mexico resonates with the characteristic poli- The political development of social pro-
by University of Sussex on 06/29/12. For personal use only.

cies pursued by “state corporatist” regimes tection in many Latin American countries
(Schmitter 1979). In this strategy, the state provides ample examples of this logic. Under
organizes society into constituent units that the military junta that came to power in 1943
are “limited in number” by government re- in Argentina, efforts were made to develop
striction, “singular” because competitors have new ties to labor through a number of state-
been eradicated, “compulsory” through legal administered social policy measures, including
or other institutional constraints, “noncompet- pensions, public health and social assistance,
itive” owing to state intervention, “hierarchi- housing finance, and unemployment insurance
cally ordered” under state management, “func- (Godio 2000). Yet the mix of benefits varied
tionally differentiated” through organizations greatly. Wages were significantly lower for
such as industrial unions, recognized by the government workers and health and education
state, with a “representational monopoly” on workers, but these sectors were compensated
members of their group, and with state “con- through even stronger job protections (Carnes
trols on leadership selection and interest ar- 2008). Equally importantly, the 1945 Law of
ticulation” (Schmitter 1979, pp. 20–21). The Professional Associations assured labor union
organizational proliferation model seeks to co- bargaining rights and recognized monopoly
opt existing competitors by incorporating them unions in each industrial sector; however, this
into the system of benefits as under collusive au- measure also limited the independence of labor
thoritarianism, but it also seeks to cultivate new and restricted its right to strike (Collier &
organizations. These new organizations dilute Collier 2002). In Brazil, too, the harder-line
the power of elites from the launching orga- military regime that emerged in 1964 first
nization, and they create overlapping responsi- sought the most radical kind of organizational
bilities and restraints on defection. Indeed, the proliferation—establishing an “official” oppo-
total pool of beneficiaries grows substantially sition party—the Democratic Movement of
such that even if benefits are uneven, mounting Brazil (MDB). When this proved impossible to
a credible threat against the government may manage, the government sought to fragment it
be difficult. Rents in various forms reach more into smaller parties. On the social policy front,
members of the population. it extended social security entitlements to rural
The social policy profile of autocracies pur- workers in 1971, and in 1974 gave pensions
suing a political strategy premised on organi- to older urban workers who had not made
zational proliferation will differ dramatically full payments into the social security system.
from the social policy profile of other autocra- Both were clear efforts to win the allegiance

100 Mares · Carnes

 ANRV377-PL12-06 ARI 13 April 2009 12:43

of alternative political bases. However, the and the scant attention it has received in the ex-
benefits to these rural and informal workers isting literature, we contend that future studies
lagged far behind those enjoyed by industrial must deepen our theoretical understanding of
and civil service workers (Haggard & Kaufman the political (especially authoritarian) dynamics
2008). Our third example of organizational driving social policy development. We have of-
proliferation comes from Mexico (Haber fered a preliminary outline of such an approach,
2007). Seeking to consolidate control over a but in-depth case studies are needed to test its
nation wracked by the regional and economic logic.
rivalries of the 1910–1920 Mexican Revolu-
tion, Plutarco Elı́as Calles and his successors
designed their new party to gather the disparate VARIETIES OF DEVELOPMENTAL
strands of the population—with sections for CAPITALISM AND VARIETIES OF
the military, peasants, industrial workers, and WELFARE REGIMES
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

“popular” middle classes (Skidmore & Smith The explanation outlined above attributes di-
2005). Although the Instituto Mexicano de vergence in social policy outcomes to differ-
Seguro Social (IMSS) was not created until the ent levels of political competition within auto-
by University of Sussex on 06/29/12. For personal use only.

1940s and was restricted to a small number of cratic regimes. However, a number of scholars
beneficiaries, it would be gradually expanded argue either that regime type is epiphenome-
to larger segments of the urban working class. nal or that its causal effect is strongly medi-
Benefits were not even, though, as state work- ated by other variables (Haggard & Kaufman
ers received the far more generous assistance 2008). In this related line of research, a number
of their own social security agency (ISSSTE). of studies are seeking to relate cross-national
The conditions under which authoritarian differences in social insurance programs to dif-
governments choose a strategy based on orga- ferences in the economic strategies pursued by
nizational co-optation versus proliferation re- various countries (Wibbels & Ahlquist 2007,
main uncertain. It would seem that the number Haggard & Kaufman 2008). This promising
of critical supporters that make up the launch- line of research is characterized by an intense
ing organization—roughly analogous to the search for the “first-order” causal variables that
“selectorate” in the terminology of Bueno de explain the initial adoption of a developmental
Mesquita et al.—is central to the choice of strat- model of capitalism.
egy. Where membership in the launching or- The theoretical antecedent of these studies
ganization is more restricted, and where other is the scholarship stressing the causal linkages
potential challengers are limited in number, it between “production regimes” and “welfare
is easier to set up institutions to favor the focal regimes” in advanced industrialized economies
elites. Conversely, when the autocrat is brought (Kitschelt et al. 1999, Hall & Soskice 2001,
to power by a broad coalition of interests, or in a Ebbinghaus & Manow 2001). These studies
civil conflict with many disparate parties and in- have identified systematic cross-national diver-
terests, then organizational proliferation will be gence in the organization of financial and in-
preferred. This allows the autocrat to cultivate dustrial relations systems among advanced in-
ties to each of the potential challengers, frag- dustrialized countries, distinguishing between
menting them and frustrating their efforts to coordinated and liberal market economies. In
collectively bring down the regime. Finally, the coordinated market economies, a dense net-
initial political ties and the ideological goals of work of business organizations as well as the
the launching organization constrain the policy provision of patient capital by firms facilitates
choices of the dictator, leading to policies with high levels of investment in firm- or industry-
distinct distributional biases. specific skills. The presence of these institutions
Given the large number of cases in which so- of nonmarket coordination affects the social
cial policy adoption occurred under autocracy, policy preferences of employers and workers in

www.annualreviews.org • Social Policy in Developing Countries 101

 ANRV377-PL12-06 ARI 13 April 2009 12:43

these economies and the formation of cross- (Wibbels & Ahlquist 2007). Capitalists were
class alliances in support of social policies that supportive of this redistribution through social
protect investment in skills (Estevez Abe et al. security programs because it ensured continued
2001; Mares 2001, 2003). These include insur- demand for their domestically produced goods.
ance policies with earnings-related benefits that In contrast, in large, labor-abundant, and
raise the reservation wage of high-skilled work- low-inequality economies, ISI policies were
ers and lower their willingness to take up jobs staved off through the efforts of relatively
that do not correspond to their qualifications. stronger rural sectors who would not tolerate
By contrast, in liberal market economies, firms’ the urban bias of ISI. In these cases, “export-led
incentives to undertake long-term investments industrialization” (ELI) was chosen. Capitalists
in the skills of their workers are lower because required two features in the labor market—a
of weak employers’ associations and fluid capi- skilled workforce and low labor costs—in or-
tal markets. In this context, cross-class alliances der to remain internationally competitive. They
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

among employers and employees are less likely favored a “human capital” social policy regime
to form, and business is more likely to press for emphasizing education and health care and with
the deregulation of labor markets and for re- little or no redistributive social policy (Wibbels
by University of Sussex on 06/29/12. For personal use only.

ductions in social policy benefits (Wood 2001). & Ahlquist 2007). In these cases, workers have
More recently, attention has turned to the not felt the need to call for more redistribu-
ways that nations use social policies to shape or tion, and collective action has been complicated
create a labor force adequate to their “particu- by the relatively large and homogenous labor
lar development projects” (Wibbels & Ahlquist force.
2007). According to this logic, the choice of so- Finally, a “mixed” social spending regime
cial policies is not driven by partisan politics or combines “modest levels of insurance and hu-
by the level of democracy or even by regime man capital spending” (Wibbels & Ahlquist
type, but by four structural conditions: domes- 2007). These regimes have occurred in coun-
tic market size, relative abundance or scarcity of tries with large domestic markets and abun-
labor, asset inequality, and the openness of the dant labor, but in which ISI developed ow-
international economy. The resulting policies ing to high levels of inequality. Nevertheless,
tend to cluster on three main “regimes” that the abundance of labor made it impossible for
emphasize redistributive social security insur- workers to collectively demand high insurance
ance, human capital development, or a mix of spending, leading to only moderate redistribu-
the two. tion and the adoption of some human capital
In countries marked by large domestic spending. Similar mixed outcomes may also oc-
markets, scarce labor, and high inequality, cur in smaller countries with mineral wealth.
inward-oriented “import-substitution industri- This elegant and parsimonious theory
alization” (ISI) development strategies predom- presents a novel explanation of the determi-
inated as a response to the Great Depression. nants of social spending in developing coun-
Capitalists sought to create a labor force capa- tries. Several aspects of this theory, however,
ble of domestic industrial production, as well as require further analysis. A critical assumption
a consumer base capable of absorbing their out- is that all leaders facing similar structural con-
put. To do so, they needed to invest in skill de- ditions will happen upon similar development
velopment among workers, and they looked to strategies with similar implications for social
social policy as a way to retain workers by insur- policy spending. However, we have suggested
ing them against the risks of injury or sickness. above that authoritarian regimes may have po-
Urban industrial workers came to possess a high litical motives, unrelated to structural economic
level of bargaining power in these labor-scarce conditions, that determine their patterns of so-
economies and were able to effectively organize cial spending. A political answer based on the
to lobby for “insurance-based social regimes” electoral needs of the government in power

102 Mares · Carnes

 ANRV377-PL12-06 ARI 13 April 2009 12:43

underwriting the ISI policies with subsidies and downplay the variation in social spending that
trade protections needs to be developed. occurred over time, especially in the most re-
In addition, the causal connection between cent decades. The examples of South Korea,
some of the underlying structural variables and Tunisia, and Mexico, which are cited as pre-
a developmental regime is often unclear in this senting stable social security spending, actually
theory. On the one hand, many of the struc- show increases of 100% or more (and decreases
tural factors that are taken as given and exoge- of 50% or more) in particular policy areas in
nous might be in fact endogenous to the politi- the period of 1970–1982. Thus, although de-
cal process and social policy regime adopted. In scribing spending regimes is helpful, this anal-
particular, levels of inequality and labor mar- ysis may miss some of the most important and
ket stratification would seem to be both causes politically contentious modifications that have
and consequences of the educational system happened over time.
and of social policy more generally. In addi- An alternative structural approach empha-
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

tion, many countries show significant variation sizes the importance of “critical alignments”
over time in their levels of labor scarcity. Fi- that occurred in the mid-twentieth century in
nally, several countries experienced significant different regions of the world, and contends
by University of Sussex on 06/29/12. For personal use only.

changes in their developmental model. Korea that distinct logics are needed to explain the dif-
shifted from an ISI model to ELI in the mid- ferent processes of social policy adoption and
1950s (Chibber 2003). Many Latin American reform in developing countries (Haggard &
countries pursued the ISI strategies for only Kaufman 2008).
brief periods of time. Models relying on time- Analyzing spending levels in each of the
invariant structural variables cannot account for three principal social policy areas—pensions,
this contingency in the choice of developmental health, and education—the critical-alignments
models. approach proposes three regional social welfare
The preferences and incentives of politi- types for Latin America, Asia, and Eastern
cal actors over different policy choices also Europe (Haggard & Kaufman 2008). Com-
are inadequately specified. Consider the choice paring social programs in 1980, Haggard &
whether or not to invest in educational poli- Kaufman describe how “social security spend-
cies. It is unclear why both workers and em- ing in Latin America dwarfs that of East Asia’s
ployers under ISI do not prefer spending on minimalist welfare states,” whereas Eastern
education, recognizing the advantages of such European nations consistently outspend (often
a choice in the long term. Why do workers not by a factor of two) the Latin American states. In
call for more broad-based educational efforts? health care, Latin America tends to emphasize
Why would employers prefer to bear the cost curative health care while Asia spends propor-
of educating workers in basic skills when they tionally more on basic health care. In education,
could outsource education to the government? although spending levels are not radically dif-
Further, recognizing the gains that could be ferent between the two regions, Latin America
made if efficiency were achieved and markets devoted more spending to secondary educa-
opened, why would capitalists “vote” against tion, and Asia showed superior performance in
universalistic education that would make work- student retention through fifth grade.
ers interchangeable and lower the wage rate? Critical realignments are argued to have
A political answer, based on the concentration provided the genesis of the distinct social wel-
of rents for industrial workers and the electoral fare policy regimes in the mid-twentieth cen-
needs of the government in power underwriting tury in each region, through which the chang-
the ISI policies with subsidies and trade protec- ing composition of the ruling elite led to new
tions, needs to be developed. “incentives to co-opt, control or repress urban
A final difficulty with this approach is that labor unions and political movements repre-
it emphasizes policy continuity and tends to senting the working class and the peasantry”

www.annualreviews.org • Social Policy in Developing Countries 103

 ANRV377-PL12-06 ARI 13 April 2009 12:43

(Haggard & Kaufman 2008, p. 20). At this time, position. Similarly, in Latin America, unions
the Cold War produced very different results that had been privileged under earlier poli-
in Eastern Europe, where Communist dictator- cies opposed both expansion of coverage (which
ships took control and made workers central to would dilute their benefits) and retrenchment
their political strategy, and Asia, where viru- in the programs that targeted them.
lently anti-Communist leaders severely limited Although this complex structural analysis is
the emergence of left parties and weakened la- a significant advance over much of the existing
bor. In Latin America, reformist leaders came literature, the causal explanation it ultimately
from the military and middle classes and sought presents is nearly as complex as the reality it
to win electoral support from the urban work- describes. The line between theory and history
ing class by extending employment-related so- is blurred to the point where it is unclear which
cial insurance. Each of these ruling coalitions variables are more important than others.
played a crucial role in setting up each nation’s Further, the emphasis on regional clusters
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

development strategy—whether ISI or export- may obscure more than it reveals. On the one
led growth—which reinforced and supported hand, the “cross-regional” differences are not
the coalitions. Where large urban labor forces as stark as they are portrayed. Consider the
by University of Sussex on 06/29/12. For personal use only.

were co-opted or integrated into the state in differences in the relative emphasis on educa-
Latin America and Eastern Europe, ISI policies tional spending, which, according to this expla-
were chosen, further cementing the industrial nation, differs dramatically between East Asia
workforce’s privileged place in the economy. In and Latin America. There is no difference in
Asia, as many of the nations emerged from colo- the average spending on education as a percent-
nial rule in the postwar period, Cold War con- age of GDP between Latin America and East
cerns and U.S. support encouraged export-led Asia—both are 3.2%—and the difference in av-
development strategies. erage education spending as a percentage of to-
A separate logic is proposed to explain tal government spending is only 2.3% (17.5%
the reform process of social welfare policy in of total spending in Asia versus 15.2% in Latin
the three regions during the post-1980 “third America). Neither difference is statistically sig-
wave” of democracy. Here the causal impor- nificant ( p-values of 0.6 and 0.55, respectively).
tance of two main independent variables is Indeed, very few of the regional differences in
stressed. First, economic conditions in the welfare state spending are statistically signifi-
1980s set up a path of either expansion or con- cant. Morever, the “regional dummies” invoked
traction. Where economic growth and struc- by this analysis conceal too much within-region
tural conditions were positive (East Asia), social variation. This within-region difference is per-
policy programs tended to expand in size and haps starker in East Asia than in Latin America
coverage. Where these conditions were nega- (Ramesh 2005). Explanations for cross-national
tive (Latin America and Eastern Europe), the variation within regional clusters at these crit-
return to democracy was marked by a contrac- ical moments invoke a long list of potential
tion of social policy commitments. Second, the explanatory variables, which includes regime
reforms of the 1980s and 1990s are argued to type, strength of the labor movement, eco-
build on the earlier political legacy that so- nomic structure, strength of populist parties,
cial policy had created in each region. Thus, in and so on. The analysis does not specify their
East Asia, where social policy had been limited, relative importance nor the levels and critical
the expansion of coverage was politically pop- values at which changes in these variables ac-
ular, drawing in new supporters to democratic count for cross-national variation within a re-
governments increasingly interested in cultivat- gional cluster.
ing an electoral base. Conversely, in Eastern A final shortcoming of the structural
Europe, where social policy commitments were “production regime” approach, shared with
broad, efforts at reduction met widespread op- their theoretical counterpart in the literature

104 Mares · Carnes

 ANRV377-PL12-06 ARI 13 April 2009 12:43

focusing on advanced industrialized societies, most dramatic examples are the introduction of
is their assumption of strong policy coherence universalistic health insurance by Taiwan and
within countries. In the following section, we Korea in 1994 and 1997, respectively (Wong
show that evolution of social policies in recent 2003). Explanations of these changes focus on
decades is characterized by tremendous varia- two consequences of democratization. First,
tion across policy areas even within the same the incentives of politicians changed follow-
country. ing transitions to democracy (Wong 2004). In
the new context of heightened electoral com-
petition, these studies argue, politicians have
THE POLITICS OF SOCIAL appealed to voters by offering social policy
POLICY REFORMS promises, but they failed to internalize the long-
During recent decades, social policies through- term fiscal implications of programmatic ex-
out the world have undergone significant pansion. Because politicians from all political
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

modification. Change and transformation parties have found this strategy attractive, par-
are common to both developed and devel- tisan differences carry little explanatory weight
oping countries. However, the magnitude of in accounting for variation in outcomes. An-
by University of Sussex on 06/29/12. For personal use only.

change experienced by advanced industrialized other consequence of democratization was the

economies is dwarfed by the thorough-going mobilization of new social groups, such as non-
policy changes that occurred in Latin America, governmental organizations (NGOs), which
Asia, and Eastern Europe in the 1990s. heightened political awareness of the strong in-
Two kinds of policy changes have received equities in policy coverage and swayed political
significant attention. The first is the wave of opinion in favor of the introduction of univer-
pension privatization that has swept the world. salistic programs (H.J. Kwon 2002, S. Kwon
By 2005, 24 countries had adopted a private pil- 2002).
lar. The speed of privatization and the magni- Current research investigating these
tude of policy change have varied dramatically changes—the privatization of pensions and the
across countries. Some countries, such as Chile introduction of universalistic social policies—
and Mexico, have entirely abolished the pre- has generated insights that complement or
existing public pillar. Others, such as Uruguay, challenge the findings regarding the politics of
have adopted a mixed system, where workers’ social policy reforms in advanced industrialized
pensions are financed by a combination of the economies. One important set of theoretical
public and private systems. Still other coun- implications concerns the role of partisanship
tries have enacted a parallel privatization, giv- in affecting the direction of social policy
ing their citizens a choice between the public reforms. The introduction of universalistic
and the private pillar. In explaining the adop- social policies in Korea and Taiwan challenges
tion of these policies, some studies have empha- the proposition advanced by power-resource
sized the impact of the competitive and fiscal scholars that strong and encompassing trade
pressures of globalization—most notably that unions and social democratic parties are a
of capital market liberalization (Madrid 2003, necessary institutional precondition of uni-
Brooks 2008). Other studies have focused on versalism (Esping-Andersen & Korpi 1984).
policy diffusion through the imitation of for- Studies of pension privatization in Latin
eign models, noting how cognitive biases and America also reveal that left-wing parties are
heuristic shortcuts lead policy makers to fol- not less likely than right-wing parties to adopt
low the example set by their neighbors’ reform pension privatization, although some studies
(Weyland 2005, 2007). find a small partisan effect on the magnitude
The second set of changes that has received of the reforms (Brooks 2008). Reformulating
attention in the literature is the expansion of the “Nixon goes to China hypothesis,” other
social protection in East Asian economies. The scholars have suggested that left-wing parties

www.annualreviews.org • Social Policy in Developing Countries 105

 ANRV377-PL12-06 ARI 13 April 2009 12:43

are more likely to adopt radical reforms, that larger programs are stickier is not borne out
owing to their superior ability to communi- by the data: Pension programs, the largest social
cate to their constituents the advantages of policy program in most developing countries,
these reforms and the absence of alternatives have been the most vulnerable to retrench-
(Cukierman & Tommasi 1998, Tommasi 2002). ment. These developments point to the need to
Nevertheless, the importance of partisanship identify the conditions under which dramatic,
as an explanatory variable continues to remain “path-departing” policy change is possible even
contested. In recent research, Huber and in the presence of programs with large polit-
others argue that—conditional on the age ical constituencies (Ebbinghaus 2005, Brooks
of democracies—the political strength of 2008).
left-wing parties contributes to differences in The emphasis on these two broad changes—
patterns of spending across Latin America, the privatization of pensions and the intro-
leading to lower levels of inequality (Huber duction of universalistic health insurance—has
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

et al. 2006). obscured other important transformations in

Recent social policy reforms in developing social policies that have taken place in many de-
countries also present challenges to explana- veloping countries. Figure 1 maps examples of
by University of Sussex on 06/29/12. For personal use only.

tions that have stressed the constraining effects policy changes in developing countries in a two-
of previous policies and institutions (Pierson dimensional policy space. The horizontal axis
1994, 2001). Scholars of policy reforms in ad- captures variation in the “socialization of risks”
vanced industrialized societies have emphasized across different social policies. Policies taking
that existing policies “exhibit increasing returns low values along this axis are characterized by
and self-reinforcement: each step along a path low levels of risk pooling. These are private-
produces consequences which make that path type policies where the level of benefits is linked
more attractive in the next round and raises the either to the actuarial risk facing the individual
cost of shifting to an alternative path” (Myles or to the returns made by investments in per-
& Pierson 2001, p. 312). An important impli- sonal, private accounts. The common feature of
cation of this argument is that the “stickiness” both policies is the individualization of risk, or
of existing policies is higher for larger pro- in Baldwin’s definition, the “fact that individuals
grams and for programs with a longer politi- face risk alone and not as part of a broader com-
cal history. As Pierson argues, a policy of long munity of risks” (Baldwin 1990, p. 3). Policies
duration and large scope will have a large polit- taking high values on this axis are characterized
ical constituency supporting it, which will con- by a high diffusion of risks across covered oc-
strain efforts of politicians to change the status cupational groups. The vertical axis measures
quo. The wave of pension privatization that has the redistribution undertaken by the particular
swept the world casts some doubt on the ex- social policy, in other words, the “propoor bias”
planatory power of this theory. The prediction of the particular policy. Policies taking high val-
ues along this axis involve higher redistribution
from high-income to low-income groups. By
High

3 4 contrast, in policies taking low values along this

Targeting Universalism
Redistribution

axis, little or no redistribution from the rich to

the poor is attempted.
2 1 Existing research has predominantly fo-
Privatization Contributory insurance cused on two cells of the policy space. Private
Low

old-age insurance policies occupy the lower

Low High left-hand corner of the policy space. Although
Socialization of risks these policies shift a significant part of the risk
Figure 1 to individuals, the benefit structure also favors
Socialization of risks. higher-income earners. The other policies that

106 Mares · Carnes

 ANRV377-PL12-06 ARI 13 April 2009 12:43

have been studied are situated in the upper social policy space in Figure 1, are also distinct
right-hand corner of the policy space. Univer- from the other social policy alternatives dis-
salistic social policies in Asia are characterized cussed above. In these policies, insurance con-
not only by high levels of coverage but also tributions are not tied to the incidence of risk
by high levels of redistribution across income faced by the individual. As a result, risk is so-
groups. According to estimates of the Korean cialized, and the redistribution of risks is higher
Health Ministry, the ratio between health ben- than in private social policies. The structure
efits and contributions for individuals in the of benefits is regressive, preserving (and some-
lower 10% of the income distribution was 6.45. times amplifying) market-based inequalities of
By contrast, it was 0.27 for the highest 10% of income. Korean pension reforms are an exam-
the population (Chun 2006). Studies of the dis- ple of a policy situated in this quadrant of the
tributional implications of the Taiwanese health policy space (Moon 2001).
care sector reveal a similar pattern (Cheng The literature’s exclusive exploration of
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

2004, 2005). quadrants 2 and 4 of the policy space has had

But, as Figure 1 illustrates, the other two important consequences. Significant variation
cells of this policy space are not empty in the in the design and distributive implications of
by University of Sussex on 06/29/12. For personal use only.

way that the literature might suggest. The vast social policies has not been adequately ex-
majority of health care reforms introduced by plored. While many countries in Latin America
Latin American countries occupy the upper have privatized their old-age insurance policies,
left-hand corner of this policy space. Many of they have also introduced health care policies
these countries have introduced reforms that that explicitly provide benefits to the poor.
attempt to increase the access of low-income Studies of social policy reform in Asia also fail
groups to health services (Mesa-Lago 2007). to account for variation across policy areas. We
Although means-testing limits the scope of see the introduction of universalistic policies in
beneficiaries of these policies (resulting in one subsystem of the welfare state, health care,
low values on the coverage dimension), these whereas the development of old-age insurance
policies take high values on the redistributive policies has followed a dramatically different
dimension of the social policy space. Consider trajectory. While Korea has introduced Bis-
the following examples. Beginning in the marckian policies, Taiwan has recently priva-
mid-1990s, the Brazilian health ministry in- tized its pension system. Existing explanations
troduced policies that guarantee basic services focusing on regional dummies or even country-
for lower-income groups (Weyland 2005, p. specific attributes cannot adequately account
231). In 1993, El Salvador introduced similar for this variation.
policies subsidizing health benefits for the Mares (2007b) proposes a political expla-
poor. Peru introduced the Programa de Salud nation of these policy outcomes that explores
Basica para Todos in 1994, which increased the sources of variation in individual prefer-
basic health care in rural areas. A related policy, ences over these policy alternatives and the
the Programa de Fortalecimiento de Servicios strategic alliances formed in support of, or op-
de Salud provided subsidies for the poor. In position to, various programs. In this model,
Colombia, the introduction of health insurance wage earners in the formal sector are the piv-
subsidies for poorer sectors led to an increase otal political group. These employees can ei-
in the access to health benefits of the poor ther choose to form a political coalition with
from 4% in 1993 to 40% in 1997 (Weyland lower-income groups (and thus support poli-
2005). In Mexico, President Zedillo introduced cies with a propoor bias) or they can ally with
Progresa, a program providing health subsidies higher-income groups (supporting programs
to low-income groups, in 1995 (Gertler 2007). with no effort to redistribute income). The
Contributory social insurance policies, lo- model also seeks to account for conditions un-
cated in the lower right-hand quadrant of the der which these coalitions support policies that

www.annualreviews.org • Social Policy in Developing Countries 107

 ANRV377-PL12-06 ARI 13 April 2009 12:43

take different values along the dimension mea- insecurity of wage earners in the formal sec-
suring the socialization of risks and favor either tor contributes to the formation of coalitions
“private-type” social policies that involve no re- between this group and the poor in many de-
distribution of risks across occupations or poli- veloping countries.
cies characterized by broad coverage and a high Differences in the extractive capacity of the
interoccupational redistribution of risks. state explain variation in the location of the
The preferences for social policies of the piv- preferences of pivotal voters along the horizon-
otal group are affected by several underlying tal axis of the social policy space and the choice
economic characteristics. These include the rel- between private-type social policies and poli-
ative level of income and the relative risk profile cies involving greater redistribution of risks. In
with respect to the risk insured by the program this model, workers in the pivotal sector choose
(old age, sickness, disability, or unemployment). among different policies by comparing the costs
The relative salience of risk versus income de- of these programs (their post-tax wages) against
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

pends on the overall level of inequality and the the expected future stream of benefits. In the
distribution of risks in the economy. All things case of a private program, those benefits are the
equal, if the level of inequality is high, the bene- result of returns from private funds. By contrast,
by University of Sussex on 06/29/12. For personal use only.

fits from risk redistribution for high-risk groups in the case of a policy administered by the state,
might be outweighed by their disutility in terms future streams of benefits are a function of the
of lower income. By contrast, if the level of capacity of the state to enforce social insurance
inequality is low, one expects the distribution legislation. If workers in the formal sector assess
of risks to be a stronger predictor of the dis- that the ability of the state to enforce existing
tributional conflicts over a new policy. This social insurance legislation is low, they will sup-
analysis thus reconciles two competing theo- port private-type social policies. Weaknesses in
retical claims regarding the politics of redistri- state capacity erode the potential gains these
bution, which have emphasized either income- workers can reap from a broad redistribution
based (Melzer-Richard) or risk-based conflict, of risks. The importance of this variable differs
showing that the relative importance of each across policy areas: Its impact is larger for poli-
(income and risk) depends on the underlying cies with a longer lag time between the payment
distribution of these variables in the economy. of the contributions and the receipt of social in-
In addition to these variables, an impor- surance benefits (such as pensions) and smaller
tant individual-level determinant of social pol- in policies where this lag is shorter. The impli-
icy preferences is the economic security of the cation of this argument is that the fiscal capacity
pivotal group—in other words, its expectation of the state affects the size of the political coali-
of steady, uninterrupted employment. This as- tions favoring different levels of redistribution.
sumption is the basis of the complex redistribu- Finally, rules pertaining to the design of dif-
tional mechanisms worked out in contributory ferent programs also affect the strategies pur-
insurance policies. An increase in the economic sued by the pivotal sector and, hence, the com-
insecurity of the pivotal group—due to exoge- position of the political coalitions supporting
nous changes such as tertiarization, an increase different policies. Consider the consequences of
in labor market flexibility, or informalization— the “policy externalities” of different programs
has important implications for its social pol- (Roemer 2006). (Policies such as immunization
icy preferences. On the one hand, the expected programs have large externalities—they affect
utility derived from all forms of contributory the health status of all citizens.) An increase in
insurance declines. On the other hand, their the magnitude of positive policy externalities
willingness to support policies with a higher is likely to affect the probability of a coalition
propoor bias increases because they expect between middle classes and the poor. A sec-
to rely on residual policies more frequently. ond relevant feature of policy design is the lag
In other words, an increase in the economic time between the payment of contributions and

108 Mares · Carnes

 ANRV377-PL12-06 ARI 13 April 2009 12:43

the receipt of social policy benefits. We expect

High
3 4
pivotal groups to discount more strongly the

Distribution of risks and income

Targeting Universalism
promised social policy benefits in policies with M+L M+L

Policy feedbacks
a high lag time and favor more limited social-
ization of risks and income redistribution.

&
To summarize the above discussion, the
2 1
four outcomes identified in the policy space Privatization Contributory Insurance
in Figure 1 are distinct political equilib- M+U M+U
ria supported by different political coalitions.

Low
Figure 2 presents these coalitions, showing
how their interaction with the level of state ca- Low High
pacity produces each of the distinct outcomes. Extractive capacity of the state
They are “self-sustaining” in the sense that ac-
Figure 2
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

tors supporting them have no incentives to de-

Extractive capacity of the state.
fect. Contributory insurance policies (quadrant
1) are supported by a coalition of middle and up-
the formation of different political coalitions,
by University of Sussex on 06/29/12. For personal use only.

per classes in conditions of high state extractive

the theory sketched here provides more explicit
capacity. These Bismarckian policies—once the
micro-foundations for macro-level outcomes.
modal social policy adopted by developed and
As such, it complements earlier studies exam-
developing countries alike—are rare in recent
ining social policy reforms in developing coun-
decades. Changes in economic conditions, such
tries, which have focused primarily on the po-
as the decline in the importance of full-time
litical incentives or ideational beliefs of elites. It
employment, make these policies less attractive
also highlights the effect of seemingly small dif-
to urban workers than alternatives where ac-
ferences in institutional design on the political
cess to benefits is not contingent on an unin-
incentives of actors. The observation that “pol-
terrupted stream of contributions. In contrast
icy feedbacks” have different consequences for
to contributory insurance programs, universal-
policy outcomes was initially developed by his-
istic policies (quadrant 4) rely on a coalition of
torical institutionalist scholars (Skocpol 1992,
the middle classes and the poor in economies
Pierson 1994). This research shows that policy
with high extractive capacity. In these environ-
feedbacks have lasting political consequences.
ments, middle-class workers are likely to accept
On the one hand, they create incentives for
a policy outcome that redistributes to the poor
pivotal groups to defect from the policy sta-
if their economic prospects are secure and if the
tus quo. On the other hand, they generate self-
poor have a better risk profile, which thus cre-
enforcing political equilibria around new poli-
ates gains from redistribution of risks (Mares
cies and contribute to the formation of different
2003). In countries with weak extractive capac-
political alliances. This suggests that the study
ity, we find either coalitions between the middle
of the recent dynamics of social policy reform
class and the poor in support of means-tested
in developing countries is an area of research
programs (quadrant 3) or coalitions between
where insights of historical and rational choice
the middle class and the rich in favor of pri-
institutionalism can be fruitfully integrated.
vate social policies (quadrant 2). The difference
between these two outcomes can be attributed
to differences in either (a) the risk profile or
economic insecurity of the pivotal group or CONCLUSION
(b) policy feedbacks, such as lag times or pol- We have written elsewhere of the need for
icy externalities, between programs. “research on social policy to become more
By modeling individual preferences for dif- unequivocally comparative in its orientation”
ferent policies and the conditions leading to (Carnes & Mares 2007, p. 882), and here we

www.annualreviews.org • Social Policy in Developing Countries 109

 ANRV377-PL12-06 ARI 13 April 2009 12:43

have pointed out some of the insights gained macro-level regularities. A third direction is
from studying the social insurance programs the study of recent developments in social
of developing countries. This has allowed us policies. We have demonstrated the existence
to observe greater variation in social policy of significant variation across policy areas in
outcomes—both across countries and regions, the trajectories and distributional implications
and across policy areas within individual coun- of social policies in many developing countries,
try cases—as well as a wider range of values on and we have proposed an explanation for these
the independent variables that are most studied outcomes that identifies the different political
in the literature. coalitions supporting these policies. Their size
In this article, we have explored three and their political composition are, in turn,
ongoing areas of research on the social policies affected by broader institutional factors and by
of developing countries. The first concerns the underlying distribution of income and risk.
the political origin of social programs. In The study of social policy in develop-
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

many developing countries, social policies ing countries presents one of the most excit-
have originated under authoritarian regimes. ing research agendas in comparative political
The large variation in the scope of these economy. Better data and more sophisticated
by University of Sussex on 06/29/12. For personal use only.

programs and the level of benefits is, so far, research methods now allow us to untangle as-
unaccounted for by existing studies. We have pects of policy design and distributional con-
proposed a political explanation for these sequences that could not have been detected
policies that is based on the strategies pursued earlier. Yet many very basic questions are still
by authoritarian leaders in their struggle with in need of systematic analysis and explanation.
the political organizations that brought them This growing line of study promises not only
to power. A second exciting area of research to give insights regarding a new class of cases—
explores the linkages between developmental developing countries—but also to enrich our
strategies pursued by various countries and understanding of important aspects of social
their social policy regimes. In examining this policy around the world. In the current world
recent literature, we have argued that a better context, the ability to adapt economically is
specification of the social policy preferences of crucial. Policies of risk reduction and allevia-
relevant political actors is needed to provide tion have never been more central to political
a stronger micro-foundation for the observed life.

DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this review.

LITERATURE CITED
Adsera A, Boix C. 2002. Trade, democracy and the size of the public sector: the political underpinnings of
openness. Int. Org. 56(2):229–62
Avelino G, Brown D, Hunter W. 2005. The effects of capital mobility, trade openness and democracy on
social spending in Latin America 1980–1999. Am. J. Polit. Sci. 49(3):625–41
Baldwin P. 1990. The Politics of Social Solidarity: Class Bases of the European Welfare State. New York: Cambridge
Univ. Press
Brooks S. 2008. Social Protection and the Market in Latin America: The Transformation of Social Security Institutions.
New York: Cambridge Univ. Press
Bueno de Mesquita B. 2003. The Logic of Political Survival. Cambridge, MA: MIT Press
Carnes ME. 2008. The politics of labor market regulation in Latin America. Ph.D. thesis, Dep. Polit. Sci., Stanford
Univ.

110 Mares · Carnes

 ANRV377-PL12-06 ARI 13 April 2009 12:43

Carnes ME, Mares I. 2007. The welfare state in global perspective. In Handbook of Comparative Politics, ed. C
Boix, S Stokes, pp. 868–85. New York: Oxford Univ. Press
Cheng P. 2004. Social health insurance in Taiwan: current status and perspectives. Work. Pap. Dep. Financ. Aff.,
Bur. Natl. Health Ins., Taipei, Taiwan
Cheng P. 2005. Assessing redistributional effects of new financing mechanisms proposed for Taiwan’s national health
insurance. Presented at World Congr., 5th, Int. Health Econ. Assoc.
Chibber V. 2003. Locked in Place: State-building and Late Industrialization in India. Princeton, NJ: Princeton
Univ. Press
Chun C-B. 2006. Distributional implications of Korean health care system. Work. Pap. Health Ins. Res. Cent.,
Seoul, Korea
Collier RB, Collier D. 2002. Shaping the Political Arena: Critical Junctures, the Labor Movement, and Regime
Dynamics in Latin America. Notre Dame, IN: Univ. Notre Dame Press
Cukierman A, Tommasi M. 1998. When does it take a Nixon to go to China? Am. Econ. Rev. 88(1):180–97
Ebbinghaus B. 2005. Can path dependence explain institutional change? Two approaches to welfare state reform.
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

MPIfG Disc. Pap. No. 05/02. Max Planck Inst. Stud. Societies, Cologne
Ebbinghaus B, Manow P. 2001. Comparing Welfare Capitalism: Social Policy and Political Economy in Europe,
Japan and the USA. London: Routledge
Esping-Andersen G. 1990. The Three Worlds of Welfare Capitalism. Princeton, NJ: Princeton Univ. Press
by University of Sussex on 06/29/12. For personal use only.

Esping-Andersen G, Korpi W. 1984. Social policy as class politics in post-war capitalism: Scandinavia, Austria
and Germany. In Order and Conflict in Contemporary Capitalism, ed. J Goldthorpe, pp. 179–208. New York:
Oxford Univ. Press
Estevez-Abe M, Soskice D, Iversen T. 2001. Social protection and the formation of skills: a reinterpretation
of the welfare state. See Hall & Soskice 2001, pp. 145–83
Flora P, Heidenheimer A. 1981. The Development of Welfare States in Europe and America. New Brunswick, NJ:
Transaction Books
Galiani S, Weinschelbaum F. 2006. Modeling informality formally: households and firms. Unpublished manuscript,
Washington Univ., St. Louis
Gertler P. 2007. Child development: What have we learned from Oportunidades? Presented at Econ. Dep., Columbia
Univ.
Godio J. 2000. Historia del movimiento obrero argentino: 1870–2000. Buenos Aires: Corregidor. 2 vols.
Grabowski R. 1988. Taiwanese economic development: an alternative interpretation. Dev. Change 19(1):53–67
Haber S. 2007. Authoritarian government. In The Oxford Handbook of Political Economy, ed. B Weingast, D
Wittmans, pp. 688–711. New York: Oxford Univ. Press
Haber S, Razo A, Maurer N. 2003. The Politics of Property Rights: Political Instability, Credible Commitments and
Economic Growth in Mexico 1976–1929. New York: Cambridge Univ. Press
Haggard S, Kaufman R. 2008. Democracy, Development, and Welfare States: Latin America, East Asia and Eastern
Europe. Princeton, NJ: Princeton Univ. Press
Hall PA, Soskice DW. 2001. Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. New
York: Oxford Univ. Press
Harris J, Todaro M. 1970. Migration, unemployment and employment: a two-sector analysis. Am. Econ. Rev.
60:126–42
Hechter M. 1975. Internal Colonialism: The Celtic Fringe in British National Development, 1536–1966. London:
Routledge & Kegan Paul
Huber E, Pribble J, Nielsen F, Stephens J. 2006. Politics and inequality in Latin America and the Caribbean.
Am. Sociol. Rev. 71:943–63
Huber E, Stephens JD. 2001. Development and Crisis of the Welfare State: Parties and Policies in Global Markets.
Chicago: Univ. Chicago Press
Kitschelt H, Lange P, Marks G, Stephens JD. 1999. Continuity and Change in Contemporary Capitalism. New
York: Oxford Univ. Press
Kwon H-J. 2002. Advocacy coalitions and the politics of welfare in Korea. Policy Polit. 31(1):69–82
Kwon S. 2002. Achieving health insurance for all: lessons from the republic of Korea. Work. Pap. ESS1. Soc. Secur.
Policy and Dev. Branch, Int. Labor Off., Seoul, Korea

www.annualreviews.org • Social Policy in Developing Countries 111

 ANRV377-PL12-06 ARI 13 April 2009 12:43

Lake D, Baum M. 2001. The invisible hand of democracy: political control and the provision of public services.
Comp. Polit. Stud. 34(6):587–621
Lindert P. 1994. The rise of social spending 1880–1930. Explor. Econ. Hist. 31:1–36
Linz J. 2000. Authoritarian and Totalitarian Regimes. Boulder, CO: Lynne Rienner
Madrid RL. 2003. Retiring the State: The Politics of Pension Privatization in Latin America and Beyond. Stanford,
CA: Stanford Univ. Press
Malloy JM. 1979. The Politics of Social Security in Brazil. Pittsburgh: Univ. Pittsburgh Press
Mares I. 2001. Firms and the welfare state: When, why and how does social policy matter to employers? See
Hall & Soskice 2001, pp. 184–212
Mares I. 2003. The Politics of Social Risk: Business and Welfare State Development. New York: Cambridge Univ.
Press
Mares I. 2005a. Social protection around the world: external insecurity, state capacity and domestic political
cleavages. Comp. Polit. Stud. 38(6):623–51
Mares I. 2007a. The economic consequences of the welfare state. Int. Soc. Secur. Rev. 60(2–3):65–81
Mares I. 2007b. The great divergence in social protection. Presented at Comp. Polit. Sem., Princeton Univ.
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

Mesa-Lago C. 2007. Social security in Latin America. Latin Am. Res. Rev. 42(2):181–201
Moon H. 2001. The Korean pension system: current state and tasks ahead. Work. Pap. 2001–3, Korea Dev. Inst.,
Seoul, Korea
by University of Sussex on 06/29/12. For personal use only.

Mulligan C, Gil R, Sala-I-Martin X. 2003. Do democracies have different public policies than non-
democracies? J. Econ. Persp. 18(1):51–74
Myles J, Pierson P. 2001. The comparative political economy of pension reform. In The New Politics of the
Welfare State, ed. P Pierson, pp. 410–56. New York: Oxford Univ. Press
O’Donnell G. 1979. Modernization and Bureaucratic Authoritarianism: Studies in South American Politics. Berke-
ley: Inst. Int. Stud.
Olson M. 1993. Dictatorship, democracy, and development. Am. Polit. Sci. Rev. 87(3):567–76
Pierson P. 1994. Dismantling the Welfare State? Reagan, Thatcher, and the Politics of Retrenchment. New York:
Cambridge Univ. Press
Pierson P. 2001. Coping with permanent austerity: welfare state restructuring in affluent democracies. In The
New Politics of the Welfare State, ed. P Pierson, pp. 410–56. New York: Oxford Univ. Press
Przeworski A, Alvarez M, Cheibub JA, Limongi F, et al. 2000. Democracy and Development: Political Institutions
and Well-being in the World, 1950–1990. Cambridge Studies in the Theory of Democracy. New York: Cambridge
Univ. Press
Ramesh M. 2005. Welfare and capitalist regimes in the Asian NIEs. Presented at workshop Varieties of Capitalism
in Asia: The State, Corporate Governance and Social Policy, Korea Univ., Seoul, Sep. 9–10
Roemer J. 2006. Democracy, Education and Equality. New York: Cambridge Univ. Press
Rueda D. 2005. Insider-outsider politics in industrialized democracies: the challenge to social democratic
parties. Am. Polit. Sci. Rev. 99(1):61–74
Schmitter P. 1979. Still the century of corporatism? In Trends Towards Corporatist Intermediation, ed. P Schmitter,
G Lehmbruch, pp. 43–94. Beverly Hills, CA: Sage
Segura-Ubiergo A. 2007. The Political Economy of the Welfare State in Latin America: Globalization, Democracy,
and Development. Cambridge/New York: Cambridge Univ. Press
Skidmore TE, Smith PH. 2005. Modern Latin America. New York: Oxford Univ. Press. 6th ed.
Skocpol T. 1992. Protecting Soldiers and Mothers. Cambridge, MA: Harvard Univ. Press
Social Security Administration. 1999. Social Security Programs Around the World. Washington, DC: Soc. Secur.
Admin.
Swenson P. 2002. Capitalists against Markets: The Making of Labor Markets and Welfare States in the United States
and Sweden. Oxford/New York: Oxford Univ. Press
Tommasi M. 2002. Crisis, political institutions and policy reform: it’s not the policy, it’s the polity, stupid! Presented
at Annu. World Bank Conf. Dev. Econ.
Tsai W, Chang L. 1985. Politics, ideology and social welfare programs: a critical evaluation of social welfare
legislation in Taiwan. Taiwan J. Sociol. 17:225–39
Weyland K. 2005. Theories of policy diffusion: lessons from Latin American pension reform. World Polit.
57(2):264–98

112 Mares · Carnes

 ANRV377-PL12-06 ARI 13 April 2009 12:43

Weyland K. 2007. Bounded Rationality and Policy Diffusion: Social Sector Reform in Latin America. Princeton, NJ:
Princeton Univ. Press
Wibbels E, Ahlquist J. 2007. Labor markets, development strategies and social spending. Unpublished manuscript.
Available at http://students.washington.edu/jsa5/WibbelsAhlquist2007.pdf
Wilensky HL. 1975. The Welfare State and Equality: Structural and Ideological Roots of Public Expenditures.
Berkeley/Los Angeles: Univ. Calif. Press
Wintrobe R. 1998. The Political Economy of Dictatorship. New York: Cambridge Univ. Press
Wong J. 2003. Healthy Democracies: Welfare Policies in Taiwan and South Korea. Ithaca, NY: Cornell Univ. Press
Wood S. 2001. Business, government and patterns of labor market policy in Britain and the Federal Republic
of Germany. See Hall & Soskice 2001, pp. 247–74
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org
by University of Sussex on 06/29/12. For personal use only.

www.annualreviews.org • Social Policy in Developing Countries 113

 AR377-FM ARI 7 April 2009 11:25

Annual Review of
Political Science

Contents Volume 12, 2009

A Conversation with Robert A. Dahl

Robert A. Dahl and Margaret Levi p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

Neorepublicanism: A Normative and Institutional Research Program

Frank Lovett and Philip Pettit p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p11
by University of Sussex on 06/29/12. For personal use only.

Domestic Terrorism: The Hidden Side of Political Violence

Ignacio Sánchez-Cuenca and Luis de la Calle p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p31
Women in Parliaments: Descriptive and Substantive Representation
Lena Wängnerud p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p51
Self-Government in Our Times
Adam Przeworski p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p71
Social Policy in Developing Countries
Isabela Mares and Matthew E. Carnes p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p93
Variation in Institutional Strength
Steven Levitsky and María Victoria Murillo p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 115
Quality of Government: What You Get
Sören Holmberg, Bo Rothstein, and Naghmeh Nasiritousi p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 135
Democratization and Economic Globalization
Helen V. Milner and Bumba Mukherjee p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 163
Has the Study of Global Politics Found Religion?
Daniel Philpott p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 183
Redistricting: Reading Between the Lines
Raymond La Raja p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 203
Does Efficiency Shape the Territorial Structure of Government?
Liesbet Hooghe and Gary Marks p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 225
Bargaining Failures and Civil War
Barbara F. Walter p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 243
Hobbesian Hierarchy: The Political Economy of Political
Organization
David A. Lake p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 263

v
 AR377-FM ARI 7 April 2009 11:25

Negative Campaigning
Richard R. Lau and Ivy Brown Rovner p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 285
The Institutional Origins of Inequality in Sub-Saharan Africa
Nicolas van de Walle p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 307
Riots
Steven I. Wilkinson p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 329
Regimes and the Rule of Law: Judicial Independence in Comparative
Perspective
Gretchen Helmke and Frances Rosenbluth p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 345
Field Experiments and the Political Economy of Development
Annu. Rev. Polit. Sci. 2009.12:93-113. Downloaded from www.annualreviews.org

Macartan Humphreys and Jeremy M. Weinstein p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 367

Laboratory Experiments in Political Economy
by University of Sussex on 06/29/12. For personal use only.

Thomas R. Palfrey p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 379

Field Experiments on Political Behavior and Collective Action
Eline A. de Rooij, Donald P. Green, and Alan S. Gerber p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 389
Experiments on Racial Priming in Political Campaigns
Vincent L. Hutchings and Ashley E. Jardina p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 397
Elections Under Authoritarianism
Jennifer Gandhi and Ellen Lust-Okar p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 403
On Assessing the Political Effects of Racial Prejudice
Leonie Huddy and Stanley Feldman p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 423
A “Second Coming”? The Return of German Political Theory
Dana Villa p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 449
Group Membership, Group Identity, and Group Consciousness:
Measures of Racial Identity in American Politics?
Paula D. McClain, Jessica D. Johnson Carew, Eugene Walton, Jr.,
and Candis S. Watts p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 471
Opiates for the Matches: Matching Methods for Causal Inference
Jasjeet Sekhon p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 487

Indexes

Cumulative Index of Contributing Authors, Volumes 8–12 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 509

Cumulative Index of Chapter Titles, Volumes 8–12 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 511

Errata

An online log of corrections to Annual Review of Political Science articles may be found
at http://polisci.annualreviews.org/

vi Contents