3 Class, gender and inequality
The role of welfare states
Walter Korpi
Inequality is an issue which has always fascinated and energized human beings, in
their roles of citizens as well as of social scientists. As Norberto Bobbio (1996) has
pointed out, in Western societies views on inequality can serve as a prism separating
the political left and the political right. Social scientists have traditionally
studied inequality in terms of the distribution of material and other advantages
among socio-economic classes or strata. In these distributive conflicts, issues
related to the relative roles to be played by markets and democratic politics have
been in the forefront. Here social scientists have seen political measures, especially
those associated with the welfare state, as being of major relevance for the patterning
and degree of inequality. While the mainstream focus in politics as well as
in research thus has been on inequality in terms of class, there have long been
undercurrents focusing on inequality in terms of gender. A couple of decades ago,
these undercurrents came to the surface in the social sciences, where they generated
considerable turbulence. Also in the context of gender inequality, however,
the role of welfare states has been a central issue.
The turbulence once generated by issues related to gender inequality would
now appear to be calming down. There is an emerging consensus that gender as
well as class are major bases for inequalities and that both must be included in the
analysis without excluding the other (see, for example, Acker 1989; Lister 1997;
Williams 1995). Against this background this chapter attempts to integrate gender
and class into an analysis of different dimensions of inequality and examines the
ways in which these two factors interact with different types of welfare states in
distributive processes.1 In such an effort we face at least three major challenges.
First, we have to tackle the old question: Inequality of what? Is it here enough
to look at the distribution of income and material standards of living, or should we
also attempt to include aspects of agency? In that case, what areas of agency are
important for gender inequalities? Second, we have to develop typologies of welfare
states, which can help us to handle and to make sense of the puzzling mosaic
of differences between countries that we observe. An important question here is
if we can find typologies relevant for gender inequality as well as for class inequality.
Third, we have to analyze the relations between gender inequality and class
inequality. To what extent do they have similar driving forces? Do they show similar
developments over time? I will here discuss these questions and bring in
empirical data from comparative analyses of inequalities in 18 OECD countries.
These countries are Australia, Austria, Belgium, Canada, Denmark, France,
Finland, Germany, Ireland, Italy, Japan, Netherlands, New Zealand, Norway,
Sweden, Switzerland, the United Kingdom, and the United States.2
Inequality of what?
Inequality in terms of material standards of living has been central for class analysis
but has also been important in analysis of gender inequality. In this context a
large number of studies have focused on the intra-family distribution of standards
of living.3 These studies indicate that in the affluent Western countries some significant
inequalities are likely to remain within nuclear families and that these
inequalities are exacerbated in the case of marriage dissolutions, thereby disadvantaging
women in terms of material conditions of life. Outside the nuclear
family, we also find some significant differences between men and women in
these respects.
Yet it would appear that analyses of gender inequality in modern Western societies
cannot be limited to standards of living and material achievements but must
also consider inequalities with respect to agency. As Amartya Sen (1992) argues,
when assessing individual well-being or the goodness of a social order, we must
consider not only manifest but also latent or potential aspects of a person’s wellbeing.
Sen assumes that freedom to choose is an important component of
well-being, and he defines freedom in terms of “alternative sets of accomplishments
that we have the power to achieve” (1992: 40). In the context of gender
inequality, it is therefore important to consider freedom to choose, that is the
range of alternative achievements among which women and men actually have
the capability to choose and which thus define their real opportunities to achieve
well-being.4
Feminist scholars have long highlighted different expressions of gendered
agency inequalities with respect to civil and political rights, inequalities which
historically have been important in what now are affluent Western countries.5
Major parts of these inequalities were associated with the institution of coverture,
whereby women lost basic parts of their civil rights when entering marriage.
Furthermore, women’s access to higher education was limited, and marriage bars
prevented them from entering the civil service and some of the professions. In
many countries women received the right to vote later than men. In several
European countries considerable parts of these legally enforced gender inequalities
in civil and political rights remained until well after the end of the Second World
War.
But at the beginning of the new century, what dimensions of agency can now
be assumed to remain most important for gender inequality in the rich Western
countries? Without claiming full coverage, I will here concentrate on three arenas
in which gendered agency inequalities can be expected to be of central importance,
namely democratic politics, higher education, and labor force participation.
Most of us would probably agree that these three arenas are worthy of study in
Class, gender and inequality 53
their own right, even if they do not exhaust areas in which gendered agency
inequality can be of importance.
Gendered agency inequality in politics and education
Gendered agency inequality in the arena of politics can occur at different levels;
at the basic level, for example, in terms of differences in the right to vote, and at
elite levels in terms of differences between men and women in representation in
parliaments and governments. We will here concentrate on elite level gender
inequality. As an indicator we will use a simple index of the equality gap, showing
the percentage point difference between the share of women and the share of men
in parliaments (weighted average of both chambers where relevant).6 In terms of
representation in legislatures, the average equality gap among our 18 countries was
–90 percentage points in 1950, indicating that women made up only 5 percent of
the membership in legislatures (Figure 3.1). Two decades later, in 1970, the average
equality gap remained largely unchanged. Thereafter, however, this gap tends
to decrease and ends up –50 in 1998, when on average 25 percent of parliamentary
members were women.
Behind the average level of the equality gap in 1998 we find major differences
among our 18 countries (Figure 3.2). At this time, the equality gap was greatest in
France and Japan (–82) with only 9 percent women in the legislature, followed by
Italy, the United States and Ireland. The equality gap was lowest in Sweden
(–14), with 43 percent women in the parliament, followed by Denmark, Norway,
54 Walter Korpi
1950 1970 1990 1998
0
–10
–20
–30
–40
–50
–60
–70
–80
–90
–100
–50
–64
–90–90
Genderequalitygap
Figure 3.1 Gender equality gap with respect to representation in legislatures, 1950–1998
(average percentage point difference women–men in 18 countries).
Finland and the Netherlands. The gender composition of governments partly
follows the patterning of legislatures but with somewhat higher equality gaps.
The social democratic cabinets in Norway in 1994 and in Sweden in 1998 were
the first ones with a full gender balance.
The above differences in gendered political representation between countries
reflect a range of factors, significant among which are electoral systems and the
relative strength of different political tendencies. Here party lists and proportional
elections appear to favor women relative to single member majority
elections. Given the level of voter preference for men, single member districts
tend to activate this preference both among voters and among parties in the
nominating process. Even with only a relatively small preference for men among
voters, voting for individual candidates may thus generate social processes similar
to those described by Schelling (1978) in the context of racial segregation,
processes which can result in large imbalances in the sex-ratio among elected representatives.
Voting for party lists may allow gender preferences to be expressed in
a more balanced way. Left parties appear to have been more inclined to elect
female candidates than have other political tendencies (Norris 1987). This is
somewhat surprising given the traditional tendency for women to vote for confessional
parties to a larger extent than men (Tingsten 1937).
Class, gender and inequality 55
0
–10
–20
–30
–40
–50
–60
–70
–80
–90
Genderequalitygap
Japan
UnitedStates
Belgium
Switzerland
Austria
Germany
Netherlands
Norway
Sweden
–14
–26
–28
–32
– 36
– 42
– 44
– 48
–50
–54
–60
–64
– 68
–72
–76
–80
–82–82
France
Italy
Ireland
UK
Canada
Australia
NewZealand
Finland
Denmark
Figure 3.2 Gender equality gap with respect to representation in legislatures in 18 countries,
1998.
As noted above, in several countries women’s access to higher education was for
a long time limited. During the period after the Second World War, however, in
all of our countries in successive cohorts the proportion with university degrees
has increased faster among women than among men. Taking the equality gap as
the percentage point difference between the proportion of women and the proportion
of men with a university degree we find that in 1994, among women
55–64 years (born in the 1930s) this distance was –17 percentage points while it
was only –3 percentage points among those 25–34 years (born in the 1960s)
(Figure 3.3). Noticeable gender differences are found primarily in Switzerland,
Belgium, the United Kingdom, Austria and Germany. In the cohorts from the
1960s, women had on the average a slightly higher cumulative number of years of
schooling than did men. Among those with a recent Ph.D. degree, however,
women are still a minority and we also find gender differences in terms of more or
less prestigious educational specialization.
Labor force participation and agency related gendered policy
institutions
In analyses of gendered agency inequality, the role of labor force participation has
long been central. Labor force participation is thus a key issue in the vigorous
debates on gender inequality in terms of the distinction between paid vs. unpaid
work as well as in the debate on strategies to decrease gender inequality captured
by the horns of the Wollstonecraft’s Dilemma (Pateman 1988: 252). Here we must
recognize that behind the distinction between paid and unpaid work looms the
fact that the labor force is an arena for the major socio-economic processes of
stratification and distribution in modern societies, processes where agency is crucial
and which are located outside the family. Individuals who are excluded from
56 Walter Korpi
0
–2
–4
–6
–8
–10
–12
–14
–16
–18
–20
–17
–3
Genderequalitygap
56– 64 years 25–34 years
Figure 3.3 Gender equality gap with respect to percentage of university graduates by age in
1994 (average of 17 countries).
participation in the labor force, among them traditionally many women, are disfavored
not only in terms of material standards of living but also in terms of social
rights. Participation in the labor force is also likely to affect a person’s self-perception
and identity in ways which influence her capabilities and freedom in many
different areas of life. Furthermore such participation is very relevant for interaction
patterns and bargaining positions within the nuclear family. It can therefore be
argued that in contemporary Western societies, gender differences in terms of
labor force participation constitute a crucial area of gendered agency inequality.
Being outside the labor force can be seen as a major indicator of agency poverty.
In this context an important question is to what extent different types of welfare
states shape patterns of gendered agency inequalities in terms of labor force
participation. To discuss these questions we need a gender-relevant typology of
welfare states. We all know Gösta Esping-Andersen’s (1990) seminal three-fold
distinction between “liberal,” “conservative,” and “social democratic” welfare
state regimes. The conceptualization of these broad regime types has proved very
fruitful. By opening up new and imaginative perspectives in the comparative
study of welfare states, it has stimulated much research. However, this as well as
other mainstream typologies have proved problematic in the context of gender
relevant analyses, especially when it comes to labor force participation (Lewis
1992; Orloff 1993; Hobson 1991; Sainsbury 1996; Shaver 1989).
Esping-Andersen’s typology spans a wide range of factors, from causal forces
over aspects of policy programs to policy outcomes, and has therefore been useful
primarily for descriptive purposes. To improve the analytic power of a typology it
would appear fruitful to focus on welfare state institutions as intervening variables
between, on the one hand, driving forces, and, on the other hand, policy outcomes.
I will here outline a typology of gender policy aspects of welfare state
institutions likely to be of relevance for gendered agency inequality in terms of
labor force participation. In this context it must however be remembered that
while the typologies we construct can be of heuristic value in helping us to bring
out the main contours of the vast empirical mosaic we can observe, such typologies
are always great simplifications of the outcomes of complicated historical
processes and can only be expected to show a partial fit with reality.
The typology of gender policies proposed here has two dimensions; the degree
to which policy institutions give general support to the nuclear family and the
degree to which they support a dual earner family model.7 These two dimensions
need not be uncorrelated with each other, but can be used to indicate the relative
weight of different types of public policies likely to be of relevance for gendered
agency inequality in terms of labor force participation.
The dimension of general family support reflects to what degree public policies
help to sustain the nuclear family while presuming that, or being neutral to
whether or not, wives have the primary responsibility for caring and reproductive
work within the family and only have a marginal labor force attachment. As
empirical indicators for this dimension we can look at three indicators: (1) the
level of cash child allowances to minor children; (2) the level of family tax
benefits to minor children and to an economically non-active spouse; and (3) the
Class, gender and inequality 57
level of public daycare services for the somewhat older children (from three years
up to school age).
While the first two indicators above are self-explanatory in this context, the third
one is more problematic. This is because in many countries presently available statistical
information does not allow us to distinguish between full-time and part-time
day care. In most continental European countries, daycare centers have thus traditionally
been organized primarily for the somewhat older pre-school children and on
a part-time basis, intended only to complement caring work within the family. Thus
they often offer only half-day care, are closed on Wednesdays and do not provide
lunches for the children, presuming the presence of a mother at home. In other countries,
for example the Nordic ones, daycare is typically provided on a full-day,
full-week basis and is intended to enable mothers’ continuous work careers. At present
it is however not possible to find comparative data reflecting this difference.
The dimension of dual earner support reflects the extent to which public policy
institutions encourage women’s continuous labor force participation; enable parents,
men as well as women, to combine parenthood with paid work, and attempt
to redistribute caring work within the family. For this dimension we use four indicators,
measuring the extent to which public policies provide (1) daycare services
for the youngest children (0–2 years); (2) paid maternity leave; (3) public home
help to the elderly; and (4) paid paternity leave. The first three indicators reflect
policy efforts to enable wives and mothers to carry on a continuous work career
while the last one indicates efforts to redistribute caring work within the family.8
Because of statistical problems with now available basic data, the above empirical
indicators are partly unreliable. We will here therefore treat them as ordinal
variables and rank our countries from 1 to 18 in terms of both these dimensions;
giving rank 1 to the country with the highest level of support provisions. These
rankings in the two-dimensional space are shown in Figure 3.4. It appears that we
can here distinguish between three different gender policy models. The figure
shows that seven of our countries have chosen to provide very little of general
family support as well as of dual earner support, having ranks 12–18 on both
these dimensions. The dominant gender policy chosen in these countries can
thus be described as a Market oriented model; that is, their policy is to allow market
forces to dominate the shaping of gender relations by leaving individuals to find
private solutions within the context of their market resources and family relations.
These seven countries are New Zealand, the United States, Switzerland,
Australia, Japan, Canada and the United Kingdom. The remaining eleven countries
provide some degree of general family support as well as of dual earner
support. Among them, however, only Sweden, Denmark, Finland and Norway,
can be described as having a dual earner support model. The remaining seven countries;
France, Belgium, Germany, Italy, Austria, the Netherlands and Ireland, can
then be said to have a General family support model. It can here be noted that in
terms of these rankings, Norway is relatively similar to France. Among the countries
with a general family support model, France thus has the most developed dual
earner support policies, while among countries with the dual earner model,
Norway has the strongest emphasis on general family support policies.
58 Walter Korpi
Political tendencies, institutions and inequality
In the development of gender policy models many different actors have been
involved, among them churches and women’s movements acting outside political
parties or in close association among them. In many countries women’s movements
have been important in changing public attitudes and values and in
bringing up gender issues to the top of the political agenda. Yet in the countries
considered here women have not formed feminist political parties; in recent
decades they have instead chosen to attempt to affect the policies of already
established parties from the outside or from the inside.
The development of gender policy models is quite clearly related to the strength
and longevity of the presence in governments of the main political tendencies in
our countries. In this context the most important differences among parties can be
roughly described in terms of two dimensions; on the one hand, the left-right
dimension centering around the relative role of politics and markets in distributive
processes, and, on the other hand, what perhaps could be called a confessional
or moral dimension, focusing especially on the role of the family.
Class, gender and inequality 59
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Rank general family support
Rankdualearnersupport
General family
support model
Market oriented
model
Dual earner
support model
NZ
US
CH
AU
JP
UK
CA
IR
AT
IT
NL
DE
BE
FR
NO
FI
SE
DK
Figure 3.4 Ranks of countries by two dimensions of gender policies and gender policy
models, 1985–1990.
During the period after the Second World War, we can thus here distinguish
between secular conservative-centrist parties, confessional parties, and left
parties.9
Secular conservative-centrist parties have generally tended to be wary about
introducing policies impinging on market distributive processes, whereas left parties
have been inclined to introduce policies to modify market distribution to
decrease inequality and poverty. In this context, European confessional parties
have often attempted to occupy some kind of a middle ground, especially in terms
of counteracting poverty. The confessional or moral dimension is a much less clear
one. Of relevance here is that this dimension reflects positions taken with respect
to the role of the traditional family form. Confessional parties have seen the
preservation of the traditional family as one of their most important goals. While
a concern with pro-natalism has also made some secular parties, especially conservative-centrist
ones, to occasionally support similar policies, on the whole the
preservation of the traditional family form has played a lesser role among the secular
parties.
During the period after the Second World War, in the countries of continental
Europe confessional parties have been of major importance and they
have tended to favor the traditional family and the general family support; a
model now found in Austria, Belgium, France, Germany, the Netherlands and
Italy but also in Ireland with a very strong position for the Catholic Church.
Although having a relatively strong confessional party presence, in this context
Switzerland constitutes an exception with a market oriented gender policy
model, something probably reflecting that its constitution tends to inhibit political
action on the national level.10 Apart from Ireland, in the English-speaking
countries secular conservative-centrist parties have dominated. With their general
propensity to avoid meddling with market forces, they have come to favor
the market oriented gender policy model. The traditional left parties have been
strongest in the four Nordic countries, and in recent decades they have come to
support the dual earner model. In this context women’s movements have been
significant.
Thus, for example, within the Swedish Social Democratic Party women’s
movements have long worked within the party to affect gender-relevant policies
(Hobson and Lindholm 1997). An example of efficient working outside parties
was given in the run-up to the 1998 parliamentary elections, when a network of
feminists threatened to field a feminist party if the established parties did not
increase their nominations of women. This threat was ultimately withdrawn but
resulted in the highest proportion so far of women in the parliament and in a government
with an equal representation of women and men. In this context it can
be noted that Norway, where a confessional party has had a more influential government
position than those in other protestant countries, has a combination
of gender policies which is relatively similar to the one in France, which is the
continental European country where the confessional party impact has been
weakest and where secular conservative-centrist parties have dominated postwar
governments.
60 Walter Korpi
Policy models and gendered agency poverty
As argued above, in modern Western societies being outside the labor force can be
seen as an indicator of agency poverty. To what extent are different gender policy
models associated with differences in terms of this indicator of agency poverty? We
must here limit the discussion to one empirical measure; the female–male differences
in percentages outside the labor force in the 25–54 years age category,
taken as an average for 1983 and 1990.11
According to this measure the range of variation in the overall gendered differences
in agency poverty rates among our 18 countries is very large, with almost
a 50 percentage point difference between Ireland and Sweden (Figure 3.5). We are
here primarily interested in the size of the overall differences among countries
with different gender policy institutions rather than in variations within these
models. As expected, we tend to find the highest levels of gendered agency
poverty in countries with the general family support model, the lowest ones in
countries with the dual earner model, while countries with the market oriented
model tend to fall in between.
Because of the multitude of factors affecting gender differences in labor force
Class, gender and inequality 61
60
50
40
30
20
10
0
Ireland
Netherlands
Germany
France
Australia
UK
USA
Norway
Finland
GendergapWomen–Men(%)
General family support
Market oriented
Dual earner
support
55
48
43
36
32
(30)
26
36
34
(30)
25 24
23 23
17
9
7 6
Italy
Belgium
Austria
Japan
Switzerland
NewZealand
Canada
Denmark
Sweden
Figure 3.5 Gender equality gaps in agency poverty by gender policy model (indicated by
the percentage point differences in proportions outside the labor force, 25–54
years, averages 1983 and 1990).
participation we cannot expect more than a partial fit between our policy models
and observed gender differences. Thus Japan and Australia with market oriented
gender policy models have higher gender differences than expected. In Japan,
gender differences probably to some extent reflect cultural traditions and the still
relatively high frequency of three-generation families. During the period after
the Second World War, Australia has to a large extent satisfied its labor demand
through immigration, something which may have lessened pressures on women to
enter the labor force. In this context we can also note that the gender difference
is relatively low in France in spite of its general family support model, while this
difference is comparatively high in Norway with a dual earner model. A hypothesis
in this context is that this pattern reflects the relatively weak government
position of the confessional parties in France but again, the comparatively strong
position of the confessional party in Norway.
A class-related welfare state typology and income inequality
For the analysis of the role of welfare states in the context of class inequality, we
need a different typology. Assuming again that welfare state institutions provide
a fruitful basis for analytically oriented typologies, we will here focus on the major
social insurance programs, primarily old age pensions and sickness insurance, the
development of which has often been associated with major societal conflicts and
which are of great importance for the economic security of most citizens. Again it
must be remembered that typologies can never be expected to show a perfect fit
with existing realities.
In this context we can use a typology keyed to three different aspects of social
insurance institutions: (1) criteria for benefit eligibility; (2) principles used for
determining benefit levels; and (3) structures for governing social insurance institutions
(Korpi and Palme 1998). Eligibility can be determined in terms of proven
need, contributions, labor force participation, belongingness to a specific occupational
category, and citizenship (residence). Principles for benefit levels result
in a continuum running from minimum or flat-rate benefits to benefits which are
markedly related to previous income. In terms of structures of governance it is
fruitful to make a distinction between programs run by elected representatives for
employers and employees and programs governed in other forms. On the bases of
these criteria, among the social insurance institutions found in old age pensions
and sickness insurance in our 18 countries during the past century, five different
ideal-typical institutional forms can be distinguished. These five ideal-typical
institutions are the targeted, voluntary state-subsidized, state corporatist, basic
security, and encompassing models, and they have emerged roughly in the order
stated here. The principle characteristics of these models are outlined in
Figure 3.6.
The targeted model originates in the old poor laws and relies on a means test
before granting benefits at a minimum level (indicated in the figure by horizontal
lines). Targeted programs have been important in many countries but are now
dominant only in Australia, where however the means test has been gradually
62 Walter Korpi
relaxed so as to exclude the rich rather than to include only the poor. The voluntary
state-subsidized model with a multitude of separate mutual-benefit associations
has historically been important in sickness as well as unemployment insurance but
has not proved practicable in the context of old age pensions. It is therefore no
longer dominant in any of our countries. The state corporatist model, introduced by
Bismarck in Germany in the late nineteenth century, broke with the poor laws as
well as with voluntarism by being based on compulsory insurance. In principal,
this model is limited to the economically active part of the population; it provides
separate programs for selected occupational categories, such as industrial workers,
agricultural workers, salaried employees, self-employed, and farmers, each program
with different conditions and entitlements. In this model benefits are clearly
earnings-related (indicated in the figure by vertical lines). Unlike the other four
models, state corporatist social insurance institutions are governed by elected
representatives for employers and employees. Catholic confessional parties have
traditionally favored the state corporatist model. Pension programs of this type are
now found in Austria, Belgium, France, Germany, and Italy with a somewhat differing
variant in Japan.
The basic security model includes, in principle, all citizens (residents), and insures
them within one and the same program. It is thus based on universalism but provides
typically only flat-rate benefits, thus a relatively low safety net for all (indicated by
Class, gender and inequality 63
(a) Targeted (b) Voluntary subsidized
(c) Corporatist (d) Basic security
(e) Encompassing
Figure 3.6 Ideal-typical models of social insurance institutions.
horizontal lines). In accordance with principles expressed by William Beveridge,
this is to leave room for high-income earners to protect their usual standard of
living via various forms of private insurance. The basic security model is now found
in Britain, Canada, Denmark, Ireland, Netherlands, New Zealand, Switzerland,
and the United States.12 The encompassing model combines the Beveridgian principle
of universalism with the Bismarckian principle of clearly earnings-related
benefits. This model was first introduced in Sweden in the 1950s, when the universal
flat-rate “People’s Pension” program was complemented with the clearly
earnings-related occupational pension program for all economically active individuals.
This model of social insurance is now found also in Finland and Norway.
These different models of social insurance institutions receive their socio-political
significance partly because they organize citizens into differing configurations
in terms of the distribution of risks and resources. Therefore they are likely to
influence the formation of citizens’ interests, identities and patterns of collective
action. Of socio-political relevance also is that they involve different degrees of
political interventions into market distributive processes, something reflected in
the relative size of budgets available for redistribution. Furthermore these models
can be seen as associated with different types of redistributive strategies. These
redistributive strategies can be characterized in terms of the degree to which benefits
are targeted to population categories with relatively low incomes, that is what
proportion of monies spent goes to the poor.
By having a high degree of low-income targeting, the targeted model relies on
a Robin Hood type redistributive strategy of financing benefits to the poor via
general taxation. Thereby it tends to drive a wedge between the short-term economic
interests of the poor and the rest of the population, which has to finance
redistribution but gets very little in material returns. As a result this model tends
to produce relatively small redistributive budgets. By limiting itself to a low safety
net, the basic security model follows a redistributive strategy of simple egalitarianism
in giving the same benefits to all without regard to previous income.
Thereby it becomes inefficient for safeguarding accustomed standards of living for
middle and high-income earners. In the long run the better off therefore tend to
develop various forms of private insurance and come to rely on these additional
programs for the major part of their economic security, while mainly manual
workers remain dependent on the public programs. In this way the material interests
of the economically better off tend to diverge from those of manual workers,
also here tending to result in relatively limited redistributive budgets.
Because of its separate programs for different occupational categories, the state
corporatist model tends to segment the population according to socio-economic
status and can thus be said to follow a strategy of redistribution among relative
equals. Since programs are clearly earnings-related, however, this model tends to
result in relatively large redistributive budgets.13 By including all citizens within
the same insurance programs and by providing the economically active with
clearly earnings-related benefits, the redistributive strategy of the encompassing
model can be said to follow the Matthew principle of giving more to those who
already have much. However, since it provides all citizens with relatively good
64 Walter Korpi
protection for accustomed standards of living, the encompassing model tends to
limit or to crowed out the various forms of private insurance. Therefore it tends to
produce relatively large redistributive budgets.
The total redistributive effect of social insurance institutions can be expected
to depend not only on the degree to which monies spent go to low-income citizens
but also on the total amount of money to be spent, that is, on the size of
redistributive budgets. The relationship between these two factors is likely to be
of a multiplicative nature. Furthermore, we can expect a trade-off so that the more
highly targeted a program is, the smaller is the redistributive budget likely to be.
On the basis of these considerations, we can expect that the lowest degree of redistribution
and therefore the highest levels of inequality are likely to be
associated with the basic security and the targeted models. The highest degree of
redistribution and the lowest levels of inequality are expected to be associated
with the encompassing model. The corporatist model is expected to generate
middling levels of inequality. It must of course here be remembered that in this
context, social insurance institutions provide only one set of intervening variables
in the distributive processes generating income inequality; here also many other
significant factors are likely to operate.14
To provide something of a test for the hypotheses sketched above, we can use
data from the Luxembourg Income Study (LIS) to analyze the distribution of
household income in 15 of our 18 countries in the 1980s. We have here computed
three measures on income inequality and five measures on levels of poverty in different
demographic categories.15 In Table 3.1, countries are grouped by type of
Class, gender and inequality 65
Table 3.1 Average rank of countries in terms of eight indicators of inequality and poverty,
by type of social insurance institutions
Country Type of social Average
(Year of data set) insurance institutions rank*
United States (1986) Basic security 1
United Kingdom (1989) Basic security 2
Canada (1987) Basic security 3
Ireland (1987 Basic security 4
Australia (1989) Targeted 5
Switzerland (1982) Basic security 8
Denmark (1987) Basic security 10
Netherlands (1987) Basic security 11
Italy (1986) State corporatist 6
France (1989) State corporatist 7
Germany (1984) State corporatist 9
Belgium (1988) State corporatist 12
Norway (1986) Encompassing 13
Sweden (1987) Encompassing 14
Finland (1987) Encompassing 15
* Gini total population; Gini persons 25–59 years; P90/P10 total population, and poverty measures for
total population; persons 25–59 years; persons 65+ years; children, and lone mothers.
social insurance institutions, and the table also gives their rank in terms of the
average rank on the above eight indicators of income inequality and poverty. On
the whole the expectations sketched above are borne out.
Of the eight countries with the basic security or targeted model, six end up
among the highest ranks in terms of inequality (United States, Britain, Canada,
Ireland, Australia and Switzerland). The three countries with the encompassing
model (Norway, Sweden and Finland) are all found among the three lowest ranks.
In the middle of the distribution we do however find some exceptions. Thus
Denmark and the Netherlands with basic security models have levels of inequality
comparable to the lowest among the state corporatist ones. Among the four
countries with the state corporatist model, Germany and Belgium have middling
ranks as expected while Italy and France have higher levels of inequality than
expected.
On the whole, however, these results point to what can be called “the paradox
of redistribution” (Korpi and Palme 1998). The targeted model, giving only to the
poor, and the basic security model, giving equally much to all, thus tend to produce
higher levels of inequality than the encompassing model giving more to
those who already have much. This paradox is to a large extent accounted for by
the fact that in the targeted and basic security models, middle and high income
earners must attempt to safeguard their accustomed standards of living by private
or occupational insurance or by savings, something which tends to produce even
higher levels of inequality than those found in the income-related benefits of the
encompassing and state corporatist models. By tending to “crowd out” such solutions,
the markedly earning-related public programs tend to result in
comparatively low levels of inequality.
It is here tempting to advance hypotheses for future research that may help to
account for what appear as unexpected results above. Thus in Italy, the wellknown
regional differences between the North and the South may contribute to
increase income inequalities. Unlike Germany and Belgium where Christian
democratic parties have dominated during the postwar period, in spite of its state
corporatist model French distributive policies have been largely directed by secular
conservative-centrist parties with less concern for avoiding poverty than
have traditional confessional parties. In the Netherlands with a basic security
model, however, confessional parties as well as the social democratic party have
been relatively strong, and in Denmark the social democratic party has been
stronger than its counterparts in the other countries with a basic security model.
Such differences in political constellations may help to account for relatively low
levels of inequality in those countries with unexpected levels of inequality.
Types of welfare states and patterns of class and gender
inequalities
The above discussion indicates that to account for the role of welfare states in
gender inequality and class inequality, we would appear to need two separate
typologies reflecting the nature of different aspects of welfare state institutions. In
66 Walter Korpi
combination these two typologies can serve to describe major constellations of
welfare state institutions. As noted above each of the three types of political tendencies,
which have dominated governments in our 18 countries during the
decades after the Second World War, can be expected to have favored specific
constellations of welfare state institutions. With their stress on the primacy of
market processes, secular conservative-centrist parties can be expected to have
opted for the targeted or basic security models of social insurance institutions as
well as the market oriented gender policy model. Confessional parties are likely to
have thrown their support behind corporatist social insurance institutions and the
general family support model.
Left parties, which were latecomers into the social policy making arena, found
themselves in the context of welfare state institutions largely erected by the other
political tendencies and these institutions turned out to be difficult to change,
especially in countries where left parties were relatively weak. The strategies
which the left parties chose have therefore come to differ greatly among our
countries, much more so than the strategies adopted by confessional parties and
secular conservative-centrist parties. It is thus hardly possible to talk here about a
“social democratic model” of welfare state institutions in the sense that it has been
adhered to by most social democratic parties. In countries where left parties have
had a long-term hold on governments, we would however expect them to have
gradually come to support encompassing social insurance institutions and dual
earner gender policies. Again it must however be remembered that the types of
welfare state institutions we now can observe are the products of the interplay
between many different actors, often working at cross-purposes, over long historical
periods, and with overlaid effects.
When we combine our typologies of social insurance institutions with gender
policy institutions, we find a targeted/market oriented constellation of institutions
in Australia and basic security/market oriented constellations in Canada,
Switzerland, Britain, the United States and New Zealand (Table 3.2). With one
exception, in these countries the dominance of conservative-centrist secular parties
has been marked. The exception here is Switzerland with a relatively strong
presence of confessional parties. However, as noted above, the characteristics of
the Swiss constitution have contributed to drastically limiting the role of policy
making on the national level. The constellation of state corporatist/general family
support institutions is found in Austria, Belgium, Germany, Italy, which all are
countries with a strong presence of Christian democratic parties. Also in France
we find a similar constellation, although its Christian democratic parties have
been considerably weaker and faded out in the 1960s. Here, the state corporatist
social insurance institutions were introduced before the First World War and in
the period between the two world wars and have been largely retained, but as
noted above, in the general family support institutions have been complemented
with a significant dose of dual earner support policies. In Finland, Norway, and
Sweden, three countries where social democratic parties have been strong, we find
the encompassing/dual earner constellation.
An unexpected combination of basic security/general family support institutions
Class, gender and inequality 67
is found in Ireland and the Netherlands. Catholic Ireland received its basic security
social insurance institutions while it was part of Britain and in the 1930s
efforts failed to change them in a state corporatist direction. In the Netherlands,
however, the social democratic party with its allied trade unions managed to
change its pensions insurance institutions from the state corporatist to basic security
one.16 Denmark has a basic security/dual earner combination of institutions. As
in the other Nordic countries, during the postwar decades the Danish social
democrats attempted to move from basic security to encompassing institutions but
these efforts failed. In Japan we find an unusual combination of state
corporatist/market oriented institutions. Its social insurance institutions became
constructed according to a variant of the Bismarckian model during the Meiji
Restoration. With cultural and religious traditions differing from those in Europe,
however, it gender policies have developed in a different direction.
As discussed above, in terms of class inequality we would expect to find the relatively
highest levels in countries with the targeted or basic security model and
the lowest ones in the encompassing countries with the corporatist countries
somewhere in between. In terms of gendered agency inequality, the relatively
highest inequalities are expected in countries with the general family support
model, the lowest ones in those with the dual earner model and middling levels in
those with market oriented models. In Table 3.2 we have information on 35 of
these outcomes, and 27 of them are in accordance with the above predictions.
68 Walter Korpi
Table 3.2 Combinations of institutional models of social insurance and gender policies and
of inequality with respect to class and gender in 18 countries, 1985–1990
Institutional models of
Country Social insurance/Gender policy Class Gender
inequality inequality
Canada Basic security/Market oriented High Medium
Switzerland Basic security/Market oriented High (Medium)
United Kingdom Basic security/Market oriented High Medium
United States Basic security/Market oriented High Medium
New Zealand Basic security/Market oriented (High) Medium
Australia Targeted/Market oriented High High*
Ireland Basic security/General family support High High
Netherlands Basic security/General family support Medium* High
Denmark Basic security/Dual earner Medium* Low
Belgium State corporatist/General family support Medium High
Germany State corporatist/General family support Medium High
Italy State corporatist/General family support High* High
France State corporatist/General family support High* Medium*
Austria State corporatist/General family support (Medium) (Medium*)
Japan State corporatist/Market oriented – High*
Finland Encompassing/Dual earner Low Low
Norway Encompassing/Dual earner Low Low
Sweden Encompassing/Dual earner Low Low
* Unexpected level of inequality, given type of institutions.
It can be noted here that it is only the encompassing/dual earner constellation
of institutions which tends to generate relatively low class inequality as well as low
gender inequality. In the other institutional constellations, we can expect to find
asymmetrical patterns of gender and class inequalities. Thus the combinations of
targeted or basic security social insurance institutions with market oriented gender
policy institutions tend to generate high class inequality but medium level gender
inequality, while the state corporatist/general family support constellation tends to
yield the opposite pattern. As a result we find relatively low correlations between
class inequality and gendered agency inequalities in our countries, something
attesting to the fruitfulness of developing separate institutional typologies in these
two policy areas. In this context it can be remembered that also in terms of gendered
agency inequalities in areas of political representation, tertiary education,
and labor force participation, we tend to find low levels of “crystallization.” Thus,
for example, Ireland has the highest gender gap in labor force participation and a
relatively high one in terms of political representation but has extinguished the
gender gap in access to university education.
A long-term perspective
It is instructive to look at the relationships between class inequality and gender
inequality in a long-term perspective. While it is here not possible to compare
levels of inequalities in these two respects, it is possible to crudely sketch trends in
the development of inequalities in each of the separate dimensions. Fifty years ago
T. H. Marshall (1950) talked about the “modern drive towards social equality,”
stating that “the modern drive towards social equality is the latest phase of an evolution
of citizenship which has been in continuous progress for some 250 years.”
As is well-known, Marshall saw this evolution as containing the gradual development
of the civil, political and social components of citizenship. Although
Marshall did not consider gender inequality, it appears that with the gradual
emergence of citizenship rights, for long periods inequality with respect to gender
as well as class decreased in roughly parallel although not identical ways. What
does the situation look like half a century later?
During the period after the Second World War, governments in several
Western countries have made strong efforts to decrease class inequality as well as
gender inequality. In many respects, however, class inequalities appear to have
been considerably more resistant to reform and reduction than have gender
inequalities. These differences in reform resistance can be illustrated in different
areas. With respect to life expectancy, we know that in Western Europe and
North America, women now have considerably longer life expectancy than have
men. But in the 1990s in these rich countries, among men, the risk of premature
death is clearly greater for manual workers than for men in higher socio-economic
classes. As shown above, the female disadvantage in access to university
education has markedly decreased. In all of these countries, however, marked
class inequalities in access to higher education remain.
As indicated above, the decreases in class inequality and in gender inequality
Class, gender and inequality 69
long followed largely parallel paths. During the last quarter of the twentieth century,
however, these paths have come to diverge. As shown above, the decrease in
gender inequality appears to have accelerated since the early 1970s, while since
the early 1980s inequalities in disposable household income have markedly
increased in these countries. In this context it is also important to consider longterm
changes in levels of unemployment, an often overlooked indicator of class
inequality. As is well-known, unemployment disproportionally hits lower socioeconomic
categories. Therefore the disappearance of recurring periods of mass
unemployment characterizing the industrialized countries up to the Second World
War and the arrival in most of our countries of full employment in the decades
after 1945 can be seen as a major decrease in class inequality. In the same perspective,
however, the return of mass unemployment after the 1970s, especially in
the West European countries, is a major reversal of the trend of declining class
inequalities.
Those who appreciate equality can rejoice at the tendencies towards decreasing
gender inequalities which we can now observe. However, very much remains to be
done and since political and economic factors are relevant here, reversals are
always possible. Yet, those who appreciate equality should be concerned about the
marked tendencies towards increasing class inequalities starting during the last
quarter of the twentieth century. For social scientists, it is important to attempt to
understand the diverging trends in these two areas. These changes indicate that
the development of citizenship is unlikely to be one of evolution. Processes of distributive
conflict are at work here.
Notes
1 Race, ethnicity, and immigration are examples of other important factors of relevance
for inequality in many countries (see Woodward and Kohli 2001) but cannot be discussed
here.
2 These countries have had uninterrupted political democracy after the Second World
War and have a population of at least one million. The data presented here come primarily
from the Social Citizenship Indicator Program (SCIP) which now is in the process
of being built up at the Swedish Institute for Social Research, Stockholm University.
3 For example, Cantillon and Nolan 1998; McLanahan, Sorensen and Watson 1989;
Millar and Glendinning 1989; Pahl 1989.
4 Needless to say, freedom to choose is of relevance also in the area of class inequality,
but where differences in standards of living still remain central.
5 For example, Lewis 1986; O’Connor 1996; Orloff 1993; Pateman 1988; Vogel 1991.
6 Thus Equality Gap = % Women–% Men.
7 For a detailed description of the empirical indicators used in constructing this typology,
see Korpi 2000.
8 In the measure of dual earner support, the indicators of daycare for the 0–2 year olds
and for the character of maternity leave are given a weight twice that of the other two
indicators. The three indicators for general family support are all given the same
weight.
9 In addition to these three tendencies there have of course also been other types of parties,
such as the “green” ones, but in terms of government positions they have been of
less importance during the period considered here.
10 See Immergut 1992; Huber et al. 1993.
70 Walter Korpi
11 For an analysis based on a larger number of indicators, see Korpi 2000.
12 In Britain, Ireland and the United States, benefit eligibility is in principle based on
contributions but in the other countries on citizenship or residence. In Britain,
Canada, Ireland, Switzerland, and the United States, pension programs are to some
extent earnings-related but because of low ceilings for benefits, the degree of earningsrelatedness
is much lower than in the encompassing countries.
13 In some countries state corporatist programs have excluded the top income earners,
enticing them to look for private insurance.
14 These other factors include, inter alia, industrial relations systems, wage setting procedures,
and demographic factors.
15 The inequality measures include Gini coefficients for the total population and for the
25–59 year category as well as the P90/P10 ratio for the total population. Defining
poverty as the proportion below 50 percent of median income, poverty levels are computed
for the total population, the prime working age population (25–59 years), the
elderly (65+ years), minor children and lone mothers. In comparing households of different
sizes, household income is weighted by the size of family and to account for
economies of scale, by giving different weights to the “first” and other family members.
The equivalence scale is one used by the OECD and gives a weight of 1.0 to the first
adult, 0.7 to the second adult and 0.5 to each additional person, regardless of age. To
avoid statistical problems associated with differing practices for including teenagers in
the data, we have excluded households with a head below 20 years of age. Countries
have then been ranked on each of the seven indicators and the average ranks are given
in Table 3.1. Here disposable income refers to net cash income after direct taxes and
social security contributions as well as after public transfers.
16 In health insurance, however, state corporatist institutions were retained.
References
Acker, J. (1989) “The problem with patriarchy”, Sociology 23: 235–40.
Bobbio, N. (1996) Left and Right. The Significance of a Political Distinction, Cambridge:
Polity Press.
Cantillon, S. and Nolan, B. (1998) “Are married women more deprived than their husbands?”,
Journal of Social Policy 27: 151–71.
Esping-Andersen, G. (1990) The Three Worlds of Welfare Capitalism, Cambridge: Polity
Press.
Hobson, B. (1991) “Economic dependency and women’s social citizenship: Some thoughts
on Esping-Andersen’s welfare state regimes”, paper presented at the Conference on
Gender, Citizenship and Social Policy, Social Science History Association Meeting in
New Orleans.
Hobson, B. and Lindholm, M. (1997) “Collective identities, women’s power resources, the
making of welfare states”, Theory and Society 26: 475–508.
Huber, E., Ragin, C., and Stephens, J. D. (1993) “Social democracy, Christian democracy,
constitutional structure, and the welfare state”, American Journal of Sociology 99: 711–49.
Immergut, E. (1992) Health Politics. Interests and Institutions in Western Europe, Cambridge:
Cambridge University Press.
Korpi, W. (2000) “Faces of inequality: Gender, class and patterns of inequalities in different
types of welfare states”, Social Politics: International Studies in Gender, State and
Society, 7.
Korpi, W. and Palme, J. (1998) “The paradox of redistribution and the strategy of equality:
Welfare state institutions, inequality and poverty in the Western countries”, American
Sociological Review 63: 661–87.
Class, gender and inequality 71
Lewis, J. (1992) “Gender and the development of welfare regimes”, Journal of European
Social Policy 2: 159–73.
—— (1986) Labour and Love. Women’s Experience of Home and Family, 1850–1940, Oxford:
Blackwell.
Lister, R. (1997) Citizenship. Feminist Perspectives, London: Macmillan.
Marshall, T. H. (1950) “Citizenship and social class”, in Marshall, T. H. (ed.) Citizenship
and Social Class and Other Essays, Cambridge: Cambridge University Press: 1–85.
McLanahan, S., Sorensen, A., and Watson, D. (1989) “Sex differences in poverty,
1950–1980”, Signs: Journal of Women in Culture and Society 15: 102–22.
Millar, J. and Glendinning, C. (1989) “Gender and poverty”, Journal of Social Policy 18:
363–81.
Norris, P. (1987) Politics and Sexual Equality: The Comparative Position of Women in Western
Democracies, Boulder: Rienner & Wheatsheaf.
O’Connor, J. S. (1996) “From women in the welfare state to gendering welfare state
regimes”, Current Sociology 44: 1–124.
Orloff, A. S. (1993) “Gender and the social rights in citizenship: The comparative analysis
of gender relations and welfare states”, American Sociological Review 58: 303–28.
Pahl, J. (1989) Money and Marriage, New York: St Martin’s Press.
Pateman, C. (1988) “The patriarchal welfare state”, in Gutman, A. (ed.) Democracy and the
Welfare State, Princeton: Princeton University Press: 231–60.
Sainsbury, D. (1996) Gender Equality and Welfare States, Cambridge: Cambridge University
Press.
Schelling, T. C. (1978) Micromotives and Macrobehavior, New York: Norton.
Sen, A. (1992) Inequality Re-examined, Oxford: Oxford University Press.
Shaver, S. (1989) “Gender, class and the welfare state: The case of income security in
Australia”, Feminist Review 32: 90–110.
Tingsten, H. (1937) Political Behaviour. Studies in Election Statistics, London: P. S. King &
Son.
Vogel, U. (1991) “Is citizenship gender-specific?”, in Vogel, U. and Moran, M. (eds) The
Frontiers of Citizenship, London: Macmillan.
Williams, F. (1995) “Race/ethnicity, gender, and class in welfare states: A framework for
comparative analysis”, Social Politics: International Studies in Gender, State and Society 2:
127–59.
Woodward, A. and Kohli, M. (eds) (2001) Inclusions/Exclusions in European Societies,
London: Routledge.
72 Walter Korpi