Immergut 2010
Immergut 2010
Immergut 2010
POLITICAL
INSTITUTIONS
ELLEN M. IMMERGUT
INTRODUCTION
As all of the entries in this Handbook testify, welfare state comparisons aim to
answer particular puzzles. Thus, a specific question determines the range of cases
chosen, the time-frame of the study, and the welfare state outcomes that will serve as
the dependent variable. So, too, is it the case with states and political institutions.
How they are defined and measured, what one chooses as the independent and
dependent variables of a study, and the methodologies that one wishes to use, all
stem from the particular questions about the welfare state that are being posed. This
entry sketches the changing research questions that have motivated research on the
impact, of state structures and political institutions on welfare states over the last
decades. It will pay particular attention to how researchers operationalize political
institutions and state structures in both quantitative and qualitative analyses.
It begins with the state-centred approach, moves on to the new institutionalism,
and ends with new directions in institutionalist research. Indeed, as we will see, there
has been a movement from broad, comparative-historical studies of state structures
to a more narrow focus on the impact of particular political institutions. The future
potential of this line of research, however, may well depend upon a reopening and
broadening of research on states and political institutions to consider the transna-
tional, comparative-historical questions with which this strand of research on the
welfare state began.
228 ELLEN M . IMMERGUT
Contemporary theories about the role played by state structures and political
institutions in the emergence, growth and (possible) decline of welfare states cannot
be understood without reference to the larger debates that prompted interest in the
study of the welfare state in the post-war period. Starting in the late 1970s, interest in
the role of the state in the development of welfare states was connected to a general
resurgence of interest in the state, which was a response both to pluralist theories of
politics and political development, and to Marxist theories of the state. The pluralist
view of politics assumed that (in an open political system), interest groups would form
spontaneously as citizens felt disturbed by recurring problems, and that these groups
would pressure government to enact policies to solve these problems. Responding to
continued interest group struggle, governments would gradually improve these poli-
cies, such that all of the activities of government could be explained in terms of a
balance of interest group pressures—including pressures from unorganized or poten-
tial' interests. For the welfare state, the implication was that social policies were simply
a response to social need, as political rights could be used to pressure governments to
respond to social imperatives. In a surprising twist on this argument, however,
convergence theory argued that all governments—whether democratic or authoritari-
an—would need to address these functionalist needs, and that, therefore, welfare state
spending could be predicted from the resources available for social spending (size of
Gross National Product) and social need (per cent of the population over 65 years of :
age), as well as the age of the social security system (Wilensky 1975). Thus, the pluralist
view, in both its interest group and functionalist variants saw social need and economic:
development as the determinant of welfare state generosity
With the resurgence of interest in Marxist theory at the end of the 1960s and early-
1970s, a second type of functionalism emerged. Marxist functionalism posited that
the capitalist state served the function of supporting the capitalist system. However*
exactly how actors within the state were to be regarded—as representatives of the
capitalists or as neutral managers—was the subject of heated debate, as well as the
process by which civil servants and politicians would know7 which policies to
introduce to serve the aims of capital. The welfare state was an important part of
this debate, as welfare state policies could be interpreted either as proof that demo-
cratic politics could remedy the inequalities of capitalism or as a sign that capitalism
was capable of surface reforms that nevertheless did not threaten the basis of the
capitalist system itself. Indeed, some authors viewed the welfare state as a product of
cross-class coalitions that rationalized capitalism in the short term, but would
eventually undermine both the conditions for capitalist accumulation and legitima-
tion (Block 1984; Gough 1979; Offe 1984).
One response to these bold declarations of both pluralist and Marxist functional-
ism was a debate on 'Do politics matter?', which focused on the role of left parties and
partisan politics, as well as corporatist patterns of interest-group intermediation
(Castles 1982a; Schmidt 1996; Wilensky 1981; see also Chapter 14 above), factors
POLITICAL INSTITUTIONS 229
whose impact is more apparent if the range of comparison is reduced to the 'rich
democracies' (Wilensky 2002). A second response was the development of 'conflict
theory', a Weberian alternative to both Marxist and pluralist views o f politics, which
saw the development of states and social policies as the product of political contesta-
tion, including contestation by social classes. In contrast to the debate with conver-
gence theory, which took the form o f quantitative, large-N tests and assumed
continuous relationships between causal variables, conflict theory focused on the
importance of specific national or regional trajectories, and on the role of historically
particular configurations of states and social structures in explaining both political
development and the emergence of the welfare state. Comparative case studies were
critical to refuting the view of welfare states as a response to democracy and/or social
need, as in authoritarian nations social rights were introduced as an alternative to
political rights (Flora and Alber 1981).
Conflict theory challenged pluralism by pointing out that pluralist politics played
out on an uneven playing field, one structured in particular by the independent and
powerful role played by states. From their engagement with Marxism, conflict theorists
developed a series of class-centred views on the politics of the welfare state, but, at the
same time, also an appreciation o f the relative autonomy of the state. Coalitions
between classes and political parties or between workers and employers in different
economic sectors—i.e. class-based interests and class-based politics, also called the
'democratic class struggle'—were important for explaining differences in the develop-
ment of the welfare states (Esping- Andersen 1990; Korpi 1983; Stephens 1979). Once in
place, however, these policies affected the mobilization of class-based interests by
providing class actors with political and ideational resources, such as a belief in
entitlement to social rights (Esping-Andersen and Korpi 1984). At the same time,
however, studies of capitalist states rediscovered the 'fiscal interests' of the state so
stressed by Schumpeter (Myles 1984) and the role of a relatively autonomous state in
stabilizing market conditions for open economies (Cameron 1978; Stephens 1979).
Thus, this engagement with class theory also produced an interest in state structures
and actors, in sequences of political development, as well as in the feedback effects of
public policies.
Perhaps nowhere was this view more directly and clearly stated than in Skocpol's
introduction to the book, Bringing the State Back In (1985). Here, Skocpol pointed to
a range of work in political science and sociology that was influenced by Max Weber
and Otto H i n t z e s view of states as organizations controlling territory and engaged in
transnational conflicts and relationships:
socio-political orders and the transnational relations within which they must
maneuver for survival and advantage in relation to other states. (1985: 8)
Thus, one could view states as actors and as structures peopled with officials whose
interests, ideas, and capacities were dependent upon the organization and interna-
tional position of the state. Moreover, states did not just assert 'relative' autonomy,
but (as in the Tocquevillian view) influenced the organization, goals, and resources of
social actors, whether these be viewed as classes, groups, or strata. As an example
of the more concrete implications of this perspective, Skocpol turned to the work of
Heclo (1974), which claimed that pension policy in Britain and Sweden had been
shaped by civil servants 'puzzling' over problems of income security in old age, and
as such by 'autonomous' state actors, rather than as a result of interest-group
bargaining or class conflict more broadly understood.
O n a more structural level, Huntington's (1971) analysis of how the particular
pattern of political development in the United States had left in place a 'Tudor' M jte
of courts and parties rather than a centralized, bureaucratic state along Weberian
lines provides a good example of how state structure can influence policy develop-
ment. As a consequence of this earlier phase in political development, Rooso Jr
could only count on sufficient state capacities for massive government intervention
in the agricultural area, and this is where his 'New Deal5 of the 1930s began.
The particular pattern for political development had repercussions as well for the
development of political parties, which jumped into the breach of weak bureaucracy
to create a spoils system, bending American social policy in the late nineteenth
century to the will of electoral machines in the area of civil war pensions (Skocpol
1992). Consequently, this policy legacy worked as a brake on further development of
the United States welfare state during the interwar period, a time when many
European states introduced social insurance on a widespread basis (Orloff and
Skocpol 1984). State structures and policy legacies also affected the development of
intellectual ideas about policy, the formation of interest group demands, and the
development of policy coalitions, for example, in the spread of Keynesian ideas and
policies during the Great Depression (Weir and Skocpol 1985).
March and Olsen (1984) incorporated the 'state-centred' approach into their water-
shed article pointing to the development of a 'new institutionalism' in political
science. This was the beginning of a shift in emphasis away from the state to one
focusing more on political institutions, and as such, a shift away from a broader
sociological analysis of the state to a narrower concern with the impact of democratic
POLITICAL INSTITUTIONS 231
political institutions on public policies. At the same time, in moving away from the
state qua state, institutionalist perspectives on politics focused more strongly on
questions of agency, identities, and interpretation than had the previous 'state-
centred' approach (Immergut 2005).
Political Configurations
Typologies provide us with a useful heuristic for identifying the key features of a
political system, and to connect these to particular political patterns. However, they
are highly vulnerable to the misspecification of cases, and are inherently static, as full-
scale constitutional change occurs only rarely. Consequently, some scholars have
argued that one should view political institutions as dynamic 'political configura-
tions', which change with political majorities and political preferences. Tsebelis (1995)
understands political configurations in terms of 'veto-players'. For Tsebelis, a veto
player is any political actor that must agree for legislation to be passed. Veto players
may be partisan or institutional. Partisan veto players are the parties in government.
Institutional veto players are any institutions, such as second chambers, that must
agree to legislation. However, not all veto players are equal. Amongst the partisan
veto players, only the two with most diverging—i.e. most distant—positions on an
issue are relevant. The reason is that if the two most distant political parties can come
to agreement on a piece of legislation, all 'intermediate' parties should also agree to a
given proposal. Similarly, an institutional veto player is only relevant to the extent
that it has a different position from the government, otherwise, it is considered to be
'absorbed'. For example, if the United States Presidency, House, and Senate are all
controlled by one party, we do not expect these three 'potential' institutional veto
players to block one another's decisions, and hence, in this configuration there are no
relevant institutional veto players. Veto players analysis aims to explain differences in
the abilities of governments to pass specific laws, and thus to predict the likelihood of
policy change.
Similarly, Immergut (1992I?) focuses on institutional veto points created from the •
interaction of constitutional rules and political results. Veto points are essentially the •
institutional veto players seen from a different angle. As in the case of institutional
veto players, what counts is whether a political arena is constitutionally qualified to
veto a proposal, and whether the majority in this area differs from that of the
government. Thus, veto points analysis also makes a distinction between 'potential'
and relevant veto points. Potential veto points are second chambers, constitutional
courts, and referenda. In addition, when governments lack a parliamentary majority,
the first chamber of the parliament is a veto point. (For an overview of veto points
and veto players in selected O E C D countries, see Table 15.1.) As the table makes clear,
the combination of electoral results and constitutional rules in creating veto points
means that the number and locations of veto points in a particular political system
change as the political configuration changes. Moreover, veto points by themselves
do not veto anything; instead, they are used by opponents of legislation to block the
legislation. Thus, veto points analysis is dynamic and depends upon political majo-
rities and political preferences, including those of interest groups.
Several studies of the development of the welfare state have shown that constitu-
tional veto points matter for interests that wish to block greater governmental
intervention in health and other social policy areas. At the same time, however,
these authors point to other variables that must be considered as well, such as the
Table 15.1 Veto points arid Veto players in selected OECD countries ( 1 9 8 0 - 2 0 0 5 )
Country Potential Institutional institutional Veto Typical Governments Institutional Veto Partisan Veto Max. Policy
Veto Points 3 Points After Points Players Distance between
Absorption 0 Govern men ta I Pa rtiesc
Australia Bicameralism High Senate, High Court Single or Two-Party 1-2 1-2 1.5 0-14.5 3.5
Court Federalism Majority
Referendum
Austria Constitutional Court Constitutional Court Single or Multi-Party 0-1 0.56 1-2 1.85 0-36.3 21.4
Federalism Majority
Belgium Bicameralism None Multi-Party Surplus 0 0 4-6 4.58 11.5-55.6 26.3
Federalism Majority
Canada Supreme Court Joint Decision Making d mainly Single-Party 0-2 1.6 1 1 0 0
Federalism S M D s Majority
Denmark Referendum Parliament Multi-Party Minority 0-1 0.94 1-4 2.64 0-50.5 17.4
Finland President (pre-1999} President (pre-1999), Multi-Party Surplus 0-2 1.29 3-5 4.12 .9-79.0 39.5
Parliament Majority
France President President, Single and Multi-Party 1-2 1.34 1-3 2.07 0-32.9 14.1
Bicameralism Constitutional Court Majority
Constitutional Court
Referendum S M D ' s
Germany Bicameralism Second Chamber Multi-Party Majority 0-2 1.70 2-3 2.56 1.3-28.3 17.3
Constitutional Court Constitutional Court
Federalism
(continued)
Table 15.1 Continued
Country Potential Institutional Institutional Veto Typical Governments Institutional Veto Partisan Veto Max. Policy
Veto Points 3 Points After Points Players Distance between
Absorption 13 Governmental Parties 0
Italy Bicameralism Chamber of Deputies, Multi-Party Minority/ 0-2 0.24 1-8 5.12 0-41,5 20.7
Senate Majority
Japan Bicameralism House of Until 1993 Single- 0-1 0.14 1-7 1.95 0-29.3 5.32
Representatives Party Majority, then
mainly Multi-Party
Majority
First Chamber, Second Multi-Party Majority 0-2 0.03 2-3 2.43 2.8-32.7 18.7
Netherlands Bicameralism
Chamber (Minority)
House of Pre-1993 Single-Party 0-1 0.38 1-2 1.30 0-29.63 4.79
New Zealand Pre-1993 S M D s
Representatives Majority, then Single-
or Multi-Party
Minority
Portugal President President, Parliament, Single or Multi-Party 1-3 1.86 0-3 1.44 0-15.0 2.3
Constitutional Court Constitutional Court:; Majority (Minority)
Referendum
Spain Constitutional Court Lower House of Single-Party Majority/ 0-1 0.45 1 1 0 0
Parliament Minority
Sweden None except Parliament Single-Party Minority 0-1 0.94 1-4 1.52 0-44.4 7.8
Parliament or Multi-Party
Minority/Majority
Country Potential institutional Institutional Veto Typical G o v e r n m e n t s Institutional V e t o Partisan Veto M a x . Policy
Veto Points3 Points After Points Players Distance between
Absorption5 Governmental Parties0
Sources: Country chapters and country data found in: Immergut et al. 2 0 0 7 ; Klingemann et al. 2 0 0 6 ; Obinger et al. 2005.
N o t e s ; a This column lists only the institutions mentioned by Huber, Ragin, and Stephens 1993, and does not include the first chamber of parliaments, as they are only a veto point
w h e n governments lack a majority in this chamber. A s Huber, Ragin, and Stephen's coding of the veto points includes single member districts, this institutional feature is included here
as weil, although perhaps the electoral system should be considered as part of the system of political competition rather than as a veto point. Largely following Huber, Ragin Et
Stephen's coding decisions, referenda are listed only when regularly used to veto laws, which is also in line with the discussion in Tsebelis 1995. W h e n governments can control the
calling of referenda, however, they are absorbed, thus leaving Switzerland as the only country with the referendum as an active veto point, in line with Huber, Ragin and Stephen's
coding as well- In place of Huber, Ragin, and Stephen's later work including judicial review, we mention only courts that actually regularly veto laws, and thus focus more directly on
constitutional courts than judicial review in general.
b
This column is based on Immergut et al. 2007, which considers the majorities for each political configuration in order to determine whether a veto point is open or closed at a
particular point in time. The procedures here are the same as used by Tsebelis 1995, except for the consideration of first chambers of parliament as veto points under minority
governments. The partisan veto players are coded following Tsebelis 1995.
c
Difference between the two most distant parties in government; based on the Manifesto Research Group's coding of the left-right positions o f parties (Klingemann etal. 2006).
d
See Canada chapter in Obinger et al. 2 0 0 5 .
236 ELLEN M . IMMERGUT
type of political parties (Maioni 1998; Obinger 1998), feedback effects over time
(Hacker 1998), and to diffusion effects (Brooks 2009), all of which will be discussed
in more depth below. Moreover, the impact of veto points has varied over time and
across policy areas, indicating the importance of historical contingency and other
more complex patterns of contextual causality (Tuohy 1999). Geographic context
may also be important. Studies of Latin American pension politics show support for
the veto points approach: pension privatization has been more difficult where veto
points facilitate interest group or partisan opposition to privatization (Brooks 2002;
Kay 1999). However, in Western Europe, politicians have actually been successful in
introducing pension reforms in systems with strong veto points and many veto
players (Bonoli 2000a; Immergut et al. 2007). One reason for this may be that
when governments cannot force change through the political decision-making pro-
cess, they must generate societal consensus for reforms, which m a y b e more effective
than a pattern of top-down decision making followed by public protests (Baccaro
2002; Natali and Rhodes 2004; Schludi 2005). Consensus generation may be more
important in mature democracies, however; hence the greater applicability of veto
points theory to Western Europe in the interwar and immediate post-war periods,
and to Latin America today.
Multivariate Analysis
Quantitative efforts to test the impact of state structures and political institutions on
welfare state outcomes have been highly influenced by the work of Huber, Ragin, and
Stephens (Huber et al. 1993). Previous efforts to operationalize state struclure
had been limited to state administrative capacities, defined as state centralization,
the age of social security programmes, the budgetary weight of government person-
nel, and traditions of bureaucratic patrimonialism (Hicks and Swank 1992; Pampel
and Williamson 1989; Wilensky 1981,1975). Instead, Huber et al. focus on the impact
of institutions on political decision making, and therefore on the ability of societ.1!
actors to press their claims upon the state. Drawing both on the distinction
between majoritarian and consensus democracy, as well as veto points analysis,
they combine institutional rigidities providing barriers to government action (feder-
alism, presidentialism, bicameralism, referenda) or parliamentary representation
(single-member districts) into an index of constitutional structure. Controlling (01
partisanship, they find that these institutional impediments do indeed slow the
growth of social expenditures in the expansion phase of the welfare state, but e\ett
only a moderate impact on retrenchment efforts (Huber and Stephens 2001a).
Schmidt, too, has investigated the impact of institutional veto points and partisan-
ship on social expenditures (even expanding the list of institutional veto playei s to
include central banks and membership in the EU) and finds them to be significant ¡is
brakes to both welfare state expansion and retrenchment (2002a). Some scholars
include unions and employers' associations as a 'veto point,' as their role in social
insurance administration may place them in a position to veto policies (Ebbinghaus
POLITICAL INSTITUTIONS 237
party politics, coalitions, and feedbacks between previous policies and current political
interests—in short, on political context. In addition, political and interest-group
strategies can be found that bypass federal constraints on welfare state expansion or
blockages to welfare state reform (Hacker 2002; Maioni 1998; Obinger et al. 2005a).
Finally, federalism (again in conjunction with specific party systems) may not only
have an impact on the welfare state, but on the constellations of interests and their
organization, thus affecting the politics of the welfare state by changing the constella-
tion of political actors (Martin and Swank 2008).
Electoral Systems
Although included in earlier quantitative studies of the emergence and generosity of
welfare states (Huber et al. 1993), electoral systems are re-emerging as a focus of
analysis. Cusack, Estévez-Abe, Iversen, and Soskice have all pointed to the link
between proportional representation, coordinated market economies and social:
policies (Cusack et al. 2007; Estévez-Abe et al. 2001). They argue that proportional
representation allows interests favouring a generous social wage, incentives for
human capital formation, and substantial employment protection to achieve better
political representation. Critical to their argument is that proportional electoral
systems allow the representation of more interests, and that therefore alliances
between middle class and working class voters favouring redistribution on terms;
favourable to the middle class emerge more readily than in majoritarian systems.
Similarly, Lynch (2006) argues that the ability of policymakers to modernize systems
of occupational welfare to include all citizens, and to cover risks of the entire life-
cycle more equally, depends upon the rules of the electoral game—whether politi-
cians make programmatic or clientelistic appeals to voters—and on programme
structures (fragmented or consolidated). Kersbergen and Manow combine the ana-
lysis of electoral systems with social cleavages to explain the development of different ;
types of welfare states (van Kersbergen and Manow 2009). Proportional representa-
tion combined with the rural-urban cleavage in the Scandinavian countries made ;,
agricultural interests central as a coalition partner, and, hence, were responsible for
the universal welfare states of the so-called 'social democratic 5 world of welfare;
Proportional representation combined with the state-church cleavage in continental
Europe produced the Christian democratic or 'conservative 5 welfare state, with its
emphasis on subsidiarity and status-conserving benefits. Majoritarian electoral sys-
tems, by contrast, resulted in less redistribution, and hence, 'liberal' welfare states.
European welfare states with their distinctions between workers, salaried employees,
and civil servants (Kocka 1981). Newer work has turned to the impact of welfare states
in demarcating 'self and 'other' in issues of race and ethnicity, as well as gender
(Lieberman 2005; Naumann 2005). Thus, welfare state politics and policies do not
just affect the organization of interests and their expression in politics, but can even
reframe societal categories and reconfigure the categories of political conflict.
CONCLUSIONS
From this survey of the impact of state structures and political institutions on the
welfare state, we see that institutions have had a significant impact on welfare state
structures and outcomes. Institutional veto points have provided leverage over
political decision making to actors opposed to expansion of social rights for various
reasons, and are hence associated with more restricted welfare states and lower levels
of government expenditures. By allowing interests opposed to government interven
tion with opportunities for blocking policies, veto points have had a significant
impact on the public-private mix in social provision. Partisan veto players, on the
other hand, have proved to be more significant for the politics of welfare state
retrenchment, notably not by blocking legislation, but instead by helping to generate
societal consensus for policy change.
Institutional effects are interaction effects, however, and thus cannot be predicted
without attention to the preferences of political actors. Moreover, as institutions and::
political decisions made within those institutions affect the constellation of
organized actors, the preferences of citizens, and the capacities of states, institutional-:
effects must be studied over time and with sensitivity to the fact that such effects-
change over time. In addition, institutional analysis must consider the problem of
'endogeneity'. Institutions emerge in specific social contexts. Therefore, one cannot
consider the impact of institutions on welfare state outcomes without also consider-
ing whether particular institutions were chosen with those welfare state outcomes in
mind. Finally, recent research demonstrates that geographic dispersion may be as
important as the time dimension. New research on the welfare state is thus headed
for greater cross-regional comparison and the engagement with transnational
patterns and processes with which the state-centred approach began.