Are Aging OECD Welfare States on the Path to the Politics of Gerontocracy? Evidence from 18 Democracies, 1980-2002

Abstract

In the fifteen years since 1990, the average OECD median voter age has increased three times faster than in the preceding thirty years. We use panel data covering the years 1980-2002 to investigate the effects of population aging on both the program size and the benefit generosity of public pensions in 18 OECD countries. Population aging appears to lead to the cutting of smaller slices out of larger cakes: it increases overall pension program spending but decreases the generosity of individual benefits. Controlling for political, institutional and time-period effects, we find that public pension efforts are significantly mediated by welfare regime type and have more fully adopted a retrenchment logic since the late 1980s. It is the politics of double fiscal and electoral straitjackets, not gerontocracy, which reigns supreme in public pension spending today. Population aging is undoubtedly accelerating in most OECD democracies. But contrary to alarmist political economy predictions, these democracies are not yet dominated by a new distributive politics of elderly power.

Full text

Electronic copy available at: http://ssrn.com/abstract=1225672
Forthcoming, Journal of Public Policy, Vol. 29 (2009)
Are Aging OECD Welfare States on the Path to the
Politics of Gerontocracy?
Evidence from 18 Democracies, 1980-2002
Markus Tepe
1
and Pieter Vanhuysse
2
3
Abstract In the fifteen years since 1990, the average OECD median voter age has
increased three times faster than in the preceding thirty years. We use panel data covering
the years 1980-2002 to investigate the effects of population aging on both the program
size and the benefit generosity of public pensions in 18 OECD countries. Population
aging appears to lead to the cutting of smaller slices out of larger cakes: it increases
overall pension program spending but decreases the generosity of individual benefits.
Controlling for political, institutional and time-period effects, we find that public pension
efforts are significantly mediated by welfare regime type and have more fully adopted a
retrenchment logic since the late 1980s. It is the politics of double fiscal and electoral
straitjackets, not gerontocracy, which reigns supreme in public pension spending today.
Population aging is undoubtedly accelerating in most OECD democracies. But contrary
to alarmist political economy predictions, these democracies are not yet dominated by a
new distributive politics of elderly power
.
Keywords: Generational Conflict; Public Policy; Pension Politics; Population Aging;
Welfare Retrenchment; Median Voter Models
JEL Classification Numbers: H55, D72
1
School of Business & Economics, Free University Berlin, Garystr.21, 14195 Berlin,
Germany, Telephone: (+49)3083857184, Email: markus.tepe@wiwiss.fu-berlin.de.
2
Corresponding author: School of Political Sciences and Faculty of Education,
University of Haifa, Mount Carmel 31905, Haifa, Israel. Telephone (+972) 547682974.
Email: pieterv@construct.haifa.ac.il
3
Equal authors, listed alphabetically.

Electronic copy available at: http://ssrn.com/abstract=1225672
1
MOTIVATION
In the last twenty-five years or so, the political, economic, and social consequences of
aging OECD societies have risen to the top of the agenda of policy communities and
social scientists alike. Fertility rates have hit post-war lows and elderly cohorts live
longer but typically do not retire later (Castles 2004; Ebbinghaus 2006). As a result, pay-
as-you-go public pension systems in maturing welfare states have come increasingly
under strain, especially in advanced capitalist democracies (Immergut et al. 2007; Myles
and Pierson 2001), but also in emerging democracies in Europe (Vanhuysse 2001; 2006).
International institutions such as the IMF (2004), the OECD (1996) and the World Bank
(1994) have devoted major policy documents to addressing the perceived old age crisis.
In electoral-numerical terms, the political balance between different age cohorts has
shifted in favor of the elderly in most rich democracies. Theoretically at least, the
possibility exists that future distributive conflict over state resources will be played out
less along lines of class, skill, or ideology, and more along generational lines, with
growing elderly age cohorts increasingly gaining the upper hand over shrinking younger
age cohorts. Starting with Preston (1984), Pampel and Williamson (1985), and Heclo
(1988), this has led many political scientists and sociologists to consider the likelihood
that, along a number of electoral and public policy dimensions, liberal democracies are
increasingly likely to reflect inter-generational conflict or elderly power (e.g. Busemeyer
et al. 2008; Castles 2004; Esping-Andersen and Sarasa 2002; Lynch 2006; Pampel 1994;
Pampel and Williamson 1989; Sabbagh and Vanhuysse 2008).
Economists and political economists have been particularly prominent in painting
a gloomy picture of the redistributive consequences of demographic change
(Browning
1975; Mulligan and Sala-i-Martin 1999; 2003; Persson and Tabellini 2000). As

2
populations age, median voters get older and the pro-public-pension coalition gains
political clout. The closer the elderly come to obtaining an electoral majority, the more
pension politics are locked-in. Systemic reforms or even benefit reductions then become
much harder, heralding a possible beginning of gerontocracy politics. Denouncing the
'deficit delusion' and 'fiscal child abuse' of US policymakers, Kotlikoff and Burns (2004)
predict that a 'coming generational storm' will have arrived by 2030. Discussing
Germany, Sinn and Uebelmesser (2002:157) conclude that the beginning of the next
decade will be 'Germany's last chance for a partial transition to a funded pension system.
Thereafter, the country will effectively be a gerontocracy.' Reviewing the social-
economic consequences of demographic change worldwide, the IMF (2004:165) lists the
year in which voters aged 50 and older will comprise at least 50.1 percent of all voters,
after adjusting for the notoriously higher turnout rates of elderly voters. Under the
ominous heading 'The Last Train for Pension Reform Departs In…,' the IMF declares
that countries such as Denmark, Greece, Italy, Norway, Portugal, and Sweden have until
2019 to reform their pensions. The US, Germany and France are said to be able to 'board
the last train' until 2015, while Finland and Switzerland will have practically entered the
age of gerontocracy by 2012.
Prior empirical research has indeed found that the proportion of elderly citizens in
the population tends to increase overall pension expenditure. But since population aging
increases the 'objective' need for pension spending, a hypothetical social planner,
benevolent dictator, or "philosopher-king" model would also predict more overall
pension spending. What makes theories of gerontocracy noteworthy is their prediction
that population aging significantly affects the generosity of individual pensions. The
more specific question thus regards not program size but benefit generosity. Are pensions

3
per elderly person higher, the higher the weight of elderly voters? In this article we
therefore compare the effect of population aging on both the size and the generosity of
public pensions in mature welfare states, using two alternative ways to capture the effect
of the latter variable. To do so, we investigate 18 OECD countries between 1980 and
2002. The article is organized as follows. In section 2 we describe and discuss two
competing political economy hypotheses on the population ageing effect. Focusing on
their empirical implications, both will be compared to the benchmark model of a
philosopher-king. Sections 3 and 4 present our refined hypotheses, the data, and the
estimation strategy. In the fifth section we present the main findings on the effect of
population ageing on pension effort. We then explore the role of political and institutional
determinants in influencing the ageing effect. The last section concludes.
POPULATION AGING AND PUBLIC PENSIONS: THEORY AND EVIDENCE
Theoretical political economy models in the tradition of Browning (1975) stem from
median voter assumptions and illustrate how re-election seeking politicians and self-
interested voters shape pension policy. Each voter is assumed to vote for the pension
system that promises him/her the largest lifetime utility. Within such frameworks,
population aging can have two opposite effects on pension benefits. First, the “elderly
power” hypothesis suggests that a larger share of elderly people in society will increase
individual pensions (Persson and Tabellini 2000; IMF 2004; Mulligan and Sala-i-Martin
1999, 2003; Sinn and Uebelmesser 2002; for reviews see Galasso 2006; Galasso and
Profeta 2007). Since the elderly internalize only the benefits and not the costs of higher
pension contribution rates, they will tend to vote for more generous pension benefits for
themselves. So, too, might a growing share of under-65s who sense they will soon be

4
pensioners. As population aging increases the electoral clout of pensioners and soon-to-
be pensioners, it is predicted to lead to "elderly power."
By contrast, the “fiscal leakage” hypothesis, similarly drawing on the median
voter framework, reaches very different conclusions. It holds that low fertility rates and
longer life expectancies decrease the expected profitability of pay-as-you-go systems to
currently working voters, thereby inducing them to favor smaller current pensions (Razin
et al. 2002; Breyer and Stolte 2001; Razin and Sadka 2007). This effect might be
triggered if current contributors lose their trust in the implicit generational contract which
underlies pay-as-you-go systems (Sabbagh and Vanhuysse 2008). The young are willing
to pay taxes for the currently retired only because they expect the young in the
succeeding generation to do the same for them. Awareness of population aging may lead
current workers to expect smaller future pension benefits than those they currently pay to
the currently old. The fiscal leakage hypothesis therefore predicts that as long as the
voting bloc of the retired does not represent an absolute majority, population aging will
actually reduce pension benefits.
Both sets of median voter models can be criticized on similar grounds. On the
supply side of pension politics, they tend to ignore the degree to which aggregate public
expenditures, driven as they are by path dependence effects and long-term spending
commitments, are an instance of 'inheritance before choice' (Rose 1990). Governments
cannot generally change welfare state programs at will. Rather, they inherit fiscal
commitments of past governments, grounded in laws and institutional structures (Rose
1990; Pierson 2004). This is particularly true for public pension schemes, where
entitlements are defined by prior contributions which give pensions the status of vested
rights (Scharpf 2000). On the demand side of pension politics, the implicit assumption of

5
median voter models that current pension incomes are the main determinant of
pensioners' political behaviour 'exogenizes away' a wide array of well-established
influences. While voter volatility has been on the rise in Western Europe in recent
decades, the elderly are among the groups least likely to act as floating voters, as their
party identifications tend to be more firmly established than those of younger generations
(Butler and Stokes 1974; Goerres 2009). And while elderly voters record comparatively
high voting turnout levels, they do not tend to vote as a homogenous constituency, single-
mindedly focused on the real value of their pensions (Goerres 2009). Instead of seeking
to maximize pension incomes, older people may actually follow Simon-type satisficing
approaches to guard against the threats of advancing age, by trying to maintain a portfolio
of resources that includes welfare benefits but also social capital and emotional capital
(Rose 2008a,b). In fact, pension incomes do not even appear to be the core concern of
elderly voters with falling consumption demands. It is health, not pensions, which tends
to be elderly persons' biggest priority (Rose 2008a,b).
Having noted these inherent limitations in median voter models, in this article we
will mainly focus on the empirical support for competing theoretical predictions
regarding the effect of population aging on public policy. What, indeed, is the current
state of empirical knowledge? Table 1 summarizes the existing evidence. Note that we do
not include here the many studies on general welfare state effort, as measured, for
instance, by total social spending as a percentage of GDP. Nor do we include recent
studies of age bias, or generational bias, in public spending (e.g. Pampel 1994; Esping-
Andersen and Sarasa 2002; Lynch 2006). Thus far, the evidence for the "fiscal leakage"
hypothesis has been scarce and less than convincing. Razin et al. (2002) show that the old
age dependency ratio has a negative effect on the labor tax and on total social transfers.

6
However, they apply standard OLS procedures on a sample of 13 countries over 28 years
without taking into account problems of autocorrelation and non-stationarity. Employing
event history analysis, Hicks and Zorn (2005) show that the share of elderly increases the
likelihood of cuts in welfare spending. But these authors subsequently published an errata
list acknowledging that, upon closer scrutiny, none of their main findings are correct or
can be reproduced (Hicks and Zorn 2007).
Table 1
In contrast, Castles (2004) finds that older populations in 1960 reduce social spending
cutbacks. Moreover, Pampel and Williamson (1989: 71, column 4), Lindert (1996),
Breyer and Craig (1997), Tabellini (2000), Disney (2007) and Shelton (2008) all find that
the proportion of elderly people in society increases public pensions or tax and
contribution rates. However, this result is strong only when the size of public pensions is
measured as pension expenditure per GDP or GNP (see also Galasso and Profeta 2002).
When the dependent variable is pension expenditure per elderly person as a share of GDP
per capita, as in Pampel and Williamson (1985; 1989: 71, column 9), the proportion of
elderly turns out to have only a weak effect. When the dependent variable is pension
expenditure per elderly person, as in Breyer and Craig (1997) and Mulligan and Sala-i-
Martin (1999), the proportion of elderly has no significant effect. Empirical studies in
political science and political sociology, while much more sensitive to institutional and
contextual variables, generally reach similar neo-pluralist conclusions to political
economy models regarding effect of population aging. Especially in the USA, it is
argued, the swelling of the ranks of pensioner interest groups has been accompanied by a

7
rise of lobbying power and ultimately helps to explain benefit levels and growth rates
(Pampel and Williamson 1985, 1989; also Huber and Stephens 2001, Pierson 1994).
On the whole, the existing evidence on the effect of population aging on the
overall size of pension programs tilts in the direction of a positive effect. This is
consistent with the "elderly power" thesis, though only roughly so. Importantly, the
theoretically more specific question of whether and how aging affects the generosity of
individual pension benefits remains largely unsettled. Moreover, in order to discover the
public policy effects of population aging, it is clearly desirable to investigate the relevant
time period. Between 1960 and 1990, the average OECD median voter age increased by
0.7 years. But in half the subsequent time span (1990 to 2005), the median voter age
actually increased by three times as much - 2.1 years (own computations from United
Nations 2006). Yet, Castles (2004) excepted, none of the research reviewed in Table 1
uses data later than 1994. Instead, we use data from a sample of 18 OECD countries
covering a maximum time span from 1980 to 2002. The sample includes Australia,
Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Italy, Japan, the
Netherlands, New Zealand, Norway, Sweden, Switzerland, the United Kingdom and the
United States. Covering both the decade before and the decade after the time when
median voter age started accelerating should allow us to better estimate the changing
effect of population aging on pension effort.
REFINED HYPOTHESES
The finding that demographic composition is an important determinant for the size of
public pensions as measured by expenditure per GDP and pension contribution rates does
not discriminate well against alternative explanations, since these variables do not

8
precisely capture the predictions of the political economy models of population aging.
Both major competing hypotheses imply that a larger share of elderly will alter the
generosity and not just the absolute size of public pensions. For instance, Persson and
Tabellini (2000:130) specifically predict “that pensions per retiree will be higher, the
higher the weight of old voters, as this shifts the median-voter equilibrium toward a more
generous pension system”. From a theoretical political science perspective, the case for
studying benefit generosity, rather than merely program spending, is equally strong.
Pointing out that “it is difficult to imagine that anyone struggled for spending per se,”
Esping-Andersen (1990:21) criticized macro-social spending studies for their inability to
indicate much about the impact of welfare programs on the well being of individual
citizens or households. One core reason, eminently applicable to pensions and population
aging, is that changes in macro-social needs can mask real cuts in individual benefits. As
many observers have noted, whenever the percentage growth of welfare program
dependents exceeds the percentage per capita reduction in welfare benefits, aggregate
spending data will misleadingly indicate higher welfare effort even despite benefit cuts
(see also Allan and Scruggs 2004:498; Korpi and Palme 2003; Rose 1984, 1985; Scruggs
and Allan 2006,). Given knowledge of total pension expenditure per GDP, pension effort
varies with GDP. For instance, a recession will make pensions macro-spending grow
faster as a percentage of GDP, even without any real budget changes (Rose and Peters
1978). Similarly, in the context of a growing service-sector economy, Baumol cost-
disease implies that productivity growth in services lags behind that in the manufacturing
(Esping-Andersen 1999; Pierson 2001). This means that a constant rate of welfare
spending per GDP over time would actually imply lower levels of welfare services being
provided to citizens.

9
Our study therefore differs in a fundamental point from prior empirical research:
in the dependent variable. We aim to study both the size and the generosity of public
pensions. In an effort to reproduce prior empirical findings on the effect of aging on the
size of pensions spending and to assess the performance our statistical model, we test
Hypothesis 1:
• (H1) Positive size effect: Societies with a larger share of elderly citizens devote more
resources to public pension programs overall (the "philosopher king," "benevolent
dictator" or "social planner" argument).
In addition, to test the predictions of the two competing generic prediction on the effect
of population aging on pension benefit generosity, we test the following two hypotheses:
• (H2) Positive benefit effect: The generosity of individual public pensions increases with
a larger share of elderly people in society (the "elderly power" argument).
• (H3) Negative benefit effect: The generosity of individual public pensions decreases
with a larger share of elderly people in society (the "fiscal leakage" argument)
DATA AND METHOD
In addition to testing more refined hypotheses, our dataset better allows us to assess the
validity of alternative theories of gerontocracy, which have predictions for both time
series and cross country data. A cross-country setup including developing and
industrialized countries, as in Perotti (1996) or Tabellini (2000), is unable to capture the
time effect of aging in mature welfare states. Appendix Table 1 presents the definition
and source of the variables employed in the regression analysis. Appendix Table 2 shows
the summary statistics.

10
Dependent variables
Our empirical analysis employs three alternative dependent variables. Cash pension
expenditure as a percentage of GDP is used to test Hypothesis 1 (data were taken from
OECD 2007). To test Hypotheses 2 and 3 and to check the robustness of our findings, we
employ two different measures. Following Breyer and Craig (1997) and Mulligan and
Sala-i-Martin (1999), pension expenditure per elderly person is used as our first proxy for
pension benefit generosity. This variable is computed by dividing cash public pension
expenditure by the number of individuals aged 65 and older (OECD 2006 Labor Force
Statistics). Second, as an alternative measure for public pension generosity, we also use
the pension generosity scores calculated by Scruggs (2005). These take into account the
average minimum and standard pension replacement rates, the number of years needed to
qualify for public pensions receipt, and the pension take-up rates. The minimum pension
replacement rate is calculated as the ratio of net public pension paid to a person with no
work history at retirement, while the standard pension replacement rate measures the
ratio of net public pension paid to a person earning the average productive worker (APW)
wage in each year of their working career upon retirement. (see also Scruggs and Allan
2006: 62). Table 2 compares the temporal development between 1980 and 2002 of our
three dependent variables and of the Old Age Dependency Ratio (ODR, defined as the
number of people aged 65+ divided by the number of people aged 15-64).
Table 2

11
Clearly, there has been significant variation in our sample. Compare, for instance, the
widely varying levels of public pension spending per GDP and pension generosity scores
of similarly 'aging' countries such as Switzerland and Austria. Conversely, countries such
as Sweden and New Zealand exhibit roughly similar levels of spending per GDP and
pension generosity scores despite wide gaps in their ODR. Generally, conservative-
corporatist welfare states such as Austria, Italy, France and Germany stand out with
respect to pension spending per GDP and, the latter country excepted, score high also on
pension generosity scores (more on this below). Overall, only three countries in our
sample have reduced pension expenditure per GDP over time, but reductions in both our
measures of pension benefit generosity took place much more frequently. Mean levels of
macro-pension effort have increased between 1980 and 2002, but have remained stable
for the two measures of benefit generosity. Similarly, mean values for standard deviation,
range, and coefficient of variation have increased or stayed stable for pension program
size, but they have decreased or stayed stable for both measures of pension benefit
generosity.
Independent variables
Consistent with Breyer and Craig (1997) we include controls in all our models for GDP
per capita (log), real GDP growth rates, long-term interest rates and inflation rates to
capture some of the economic foundations of public pension programs. Another
important fundamental relates to labor markets. Increasing labor market participation
rates among as many working-aged citizens as possible is a straightforward way of
improving the fiscal foundations of public pensions systems even in a context of
population aging (Barr 2001; Castles 2004). We therefore include in Models 5 and 6 of

12
Tables 3, 4, and 5 two labor-market related variables: Female Participation Rate, defined
as the female share in the active labor force, and Elderly Participation Rate, defined as
share of 55plussers in the active labor force. Both variables would be predicted to have a
negative effect on our three variables of pension effort to the extent that higher shares of
each group that actually participate in the labor market may create less of a 'welfare
dependency culture'. Alternatively, both participation rates could be predicted to have a
positive effect, as they enhance the social security revenues and therefore the fiscal
foundations for pension spending.
Our independent variable of main interest, however, is Old Age Dependency Ratio
(ODR).
1
ODR admittedly does not capture the population up to 15 years of age, which is
also an economically dependent demographic group (Castles 2004:120; 124). But since
children do not vote, ODR is a better measure for political analysis, as it is a proxy for the
relative electoral strength of different generations of (potential) voters. ODR is in our
view a better measure than the median voter age that lies at the heart of most theoretical
political economy models, as it more directly captures the relative numerical balance
between elderly and working-age voters and, indirectly, the financial (im)balance of
pensions systems and welfare states. Temporal changes in ODR lag behind those in
median voter age, with the main growth acceleration still to be expected around 2012
(United Nations 2006). ODR, lastly, is also a conservative measure in the sense that it
includes aging voters of working age, who may increasingly identify their self-interest
with that of pensioners as they approach pension age (Van Parijs 1998; IMF 2004). We
attempt to capture the latter effect through the 55-plus ratio. Lastly, the brunt of
population aging still lies in the near future in most OECD democracies, when current
median voters will have actually reached pension age. We attempt to capture expected

13
future population aging, a proxy for these impending demographic problems, by means of
ODR Projection Difference, defined as the difference between the United Nations World
Population Prospect (2006) projection of a country's ODR twenty years hence and its
ODR at present.
Estimation strategy
In line with prior empirical studies by Lindert (1996), Breyer and Craig (1997) and
Disney (2007), our annual observations have been grouped into eight-year averages, as
we are primarily interested in structural changes rather than annual fluctuation. The
variables for each country are calculated as averages of, respectively, the periods 1980-
1986, 1987-1995 and 1996-2002. This makes a sample of up to 54 observations (T=3,
N=18).
2
The use of these eight-year periods is a way for taking into account the fact that
electoral and social changes often need time to materialize into public policy pension
efforts. Each period represents two average legislation periods, which should be long
enough for voters to influence the course of pension politics. The statistical model takes
the following form:
y
i,t
= β
0
+ β
1
(x
i,t
) + β
2
(z
i,t
) +γ
i
+ ε
i,t
where y
i,t
denotes pension expenditure per GDP, pension generosity scores or real pension
benefit per elderly in country i at time t. The variable x denotes the set of control
variables (log GDP per capita, real GDP growth, long-term interest rate and inflation
rate) derived from the philosopher-king model. The variable z denotes the independent

14
variable of main interest. The datasets panel structure suggests using the random or fixed
effects estimator.
One key question regards whether the unit effects should be treated as random or
fixed. The random effect estimator is heavily influenced by cross-sectional variance and
depends on the assumption that unobserved heterogeneity is mean independent of the
causal variable (Halaby 2004: 511). This assumption would be defensible under
randomized assignment but not in a sample consisting of 18 OECD countries, where each
unit is having a distinct set of social security institutions. The fixed effects estimator,
which exploits within unit variation as a mean of purging unit heterogeneity, offers to
dispense the random effects assumption and still obtains unbiased and consistent
estimates when unit effects are arbitrarily correlated with explanatory variables (Halaby
2004: 516). Nielsen and Andersen (1995: 686) argue that the fixed effect estimator can be
interpreted as “throwing away” all between unit variations in the data. This is true,
however it protects against biased and inconsistent parameter estimates, since the
possible efficiency advantage of the random effects estimator depends on the random
effects assumption.
Without plausible theoretical grounds or empirical evidence for the random
effects assumption, bias and consistency considerations alone would lead to a fixed
effects model (Halaby 2004: 521). With non-experimental data, the fixed effects
estimator is nearly always preferable to the random effect estimator (Allison 1994: 181;
Nickell 1981: 1418). To test the prediction of political economy models of pensions, the
fixed effects (or within country) estimator is particularly apt, since these models specify
how electoral outcomes change pension spending and generosity within and not between
countries. Political economists often acknowledge this. Breyer and Craig (1997: 717),

15
who rely on the random effect estimator, nevertheless concede that their 'prior belief was
that within-country estimates are more reliable tests'. Nickell (1981: 1418) similarly
argues that “if one takes the view that, in any particular model, the individual effects are
likely to be correlated with all the observed exogenous variables, then one is led
inexorably to the fixed effects model.” Below, Tables 3 to 5 (on main demographic
effects) and 6 to 8 (on political and institutional effects) investigate each time the same
models with respect to three different pension effort dependent variables: program size,
generosity scores, and spending per elderly.
EMPIRICAL ANALYSIS: BASELINE EFFECTS OF POPULATION AGING
Table 3 indicates that Interest rate has a negative effect and Inflation rate a positive but
non-significant effect on pension program size, consistent with Breyer and Craig (1997).
GDP levels have a positive effect on program size, and GDP growth rate a negative
effect. The finding on growth confirms those by Disney (2007) and Allan and Scruggs
(2004) and may have two possible explanations. First, since GDP is used as a
denominator for pension expenditure, GDP growth decreases pension expenditure per
GDP if pension expenditure remains unchanged. Second, following the formal argument
by Breyer and Craig (1997), pension expenditure per GDP represents a proxy for the
pension contribution rate. In this respect, GDP growth may reduce the relative burden of
social insurance contributions on labor income (Disney 2007: 547). Using the alternative
dependent variables in Tables 4 and 5 does not significantly alter the general pattern for
Interest rate, Inflation rate, and GDP levels and growth rates. To check for the influence
of public-financial constraints, we also introduced general government debt rates per
GDP and government budget deficits per GDP as control variables in all 9 Models for all

16
three dependent variables, but both variables were generally non-significant (results not
shown).
Table 3
Table 3 also gives results for our test of the positive size effect (or philosopher-king)
hypothesis (H1). The effect of ODR is positive and statistically significant, although this
is somewhat sensitive to model specification. Cross-sectionally, countries with a larger
share of elderly thus appear to devote more resources towards pension programs. This
finding is consistent with prior findings on older datasets regarding pension spending per
GDP (e.g. Breyer and Craig 1997; Pampel and Williamson 1985; 1989; Tabellini 2000).
Interestingly, investigating the sense of expected demographic worsening appears to
indicate a potential sense of future-orientation among current-day policymakers (Model
3). The more the ODR is expected to increase over the next twenty years, the more the
size of public pensions today is likely to be cut; an effect that is significant at the one-
percent level. Surprisingly perhaps, enlarging the imputed electoral pensioner
constituency by including elderly workers again changes the picture (Model 4). The
effect size for 55plus Ratio is close to zero and the effect is not significant.
Once we consider two different measures of pension benefit generosity rather than
program size as our dependent variable, the main effect of population aging turns out to
be rather different. With the fixed effect specification, the coefficients in Model 2, 5 and
6 for ODR and in Model 4 for 55plus Ratio have no statistically significant effect on
pension generosity scores (Table 4) and a significant but negative effect on real pension
expenditure per elderly (Table 5). The effect of the expected future demographic aging

17
(Model 3) is insignificant as regards spending per elderly, but significant, large, and
negative as regards the pension generosity score.
Table 4
By way of rough preliminary conclusion, population aging today (though not in the near
future) appears to simultaneously increase program size (the positive size hypothesis, H1)
and to reduce the generosity of benefits going to the elderly today (the negative benefit
hypothesis, H3), at least as measured by real spending per elderly person. While this
finding needs to be further refined, these 'bare effects' of population aging do hold true
after controlling for the well-known argument that maturing welfare states have now
entered a new era of post-Golden-Age retrenchment politics (Huber and Stephens 2001;
Pierson 1994; 2001; Vanhuysse 2001; Weaver 1986). To incorporate the stylized
observation in Figure 1 that OECD democracies have on the whole experienced
accelerated median voter aging since the early 1990s, we have included dummies for the
two most recent eight-year periods, 1988-1995, and 1996-2002 in all Models in Tables 3-
5. The emerging picture indicates increased retrenchment in more recent years, as
compared to the early-to-mid-1980s. That is, both the middle period and the most recent
period effects are negative for every single model that includes ODR, and for all three
dependent variables. This 'retrenchment politics' picture is further corroborated by the
observation that when it comes to pension program size and generosity scores, the size of
this negative period dummy effect is at least as large, and most often significantly larger,
in the case of our most recent period (1996-2002) as compared to our middle period
(1988-1995).

18
Table 5
Taken together, these observations on the demographic and time period effects appear to
provide further empirical footing for the thesis that the political economy of pensions has
become a different, more cutbacks-driven, ballgame in the 1990s than it was in preceding
decades. The increasing political and budgetary pressures that accompanied the faster
growth of elderly cohorts appear to have required stronger overall financial commitments
while simultaneously necessitating cuts in pension benefits, or, at a minimum, the end to
the rise in benefit generosity. At least among the EU member states in our sample, the
adoption of the Maastricht criteria in 1991 may have served as an independent additional
source of pressure on pre-existing budget deficits. In other words, pension politics
appears to have caught up with the dynamics of retrenchment that have transformed
welfare state politics more generally since the 1980s. Population aging simply appears to
have driven home more urgently to governments, or at least finance ministers, the basic
fiscal fundamentals of budgetary constraint (Rose 1984; 1985). Total yearly expenditure
on pensions is a function of the number of pensioners alive in a given year times the unit
cost per pensioner. Since the number of pensioners is currently increasing due to the post-
war baby boom and increasing life expectancy, higher macro-level pension spending
inevitably follows, unless pension unit cost growth is reduced. One way of interpreting
our findings in tables 3 to 5 is to argue that both developments are now occurring
simultaneously, leading OECD governments, as it were, to cut smaller (micro-)slices out
of larger (macro-)cakes.

19
A further caveat might qualify these preliminary conclusions. As we have spelled
out above, an important additional driver of spending effort in public pension systems is
their underlying labor market foundations. In many countries, the working-age population
has shrunk in recent decades because current youth cohorts study much longer rather than
going out to work in their late teens. At the other end of the career cycle, elderly workers
have been sent on early retirement schemes, sometimes on a massive scale, especially in
continental Western Europe (Ebbinghaus 2006) and Eastern Europe (Vanhuysse 2006).
Conversely, female labor force participation rates have been on the rise in many post-
industrial economies (Esping-Andersen 1999). Models 5 and 6 in Tables 3 to 5 therefore
investigate the effect of the labor market participation rates of, respectively, elderly
workers (of both genders) and working-age women.
Once we control for ODR, period effects, and macro-economic variables, female
participation has a positive effect on all three pension effort variables, though it is large
and significant only in the case of pension generosity scores (Table 4). This provides
tentative support for the interpretation that higher participation rates improve the fiscal
scope for pension effort. But interestingly, higher levels of labor market participation
among elderly workers reduces pension effort on all three variables, and the effect is
significant for both program size (Table 3) and spending per elderly (Table 5). This
appears to contradict the argument that workers, qua workers, tend increasingly to reason
like pensioners as they approach pensionable age (IMF 2004; Van Parijs 1998). If
anything, it indicates that elderly workers, when participating, might also be prone to
fiscal leakage reasoning (Razin et al.2002; Razin and Sadka 2007). Additionally, larger
shares of elderly workers actually participating in the labor market may contribute to a

20
'work culture,' reducing popular support among this group for high pension outlays for
current pensioners.
THE ROLE OF POLITICS AND INSTITUTIONS
To refine the above insights and to check for the role of political and institutional effects,
we have computed a further set of three regression models in Table 6 on the same three
dependent variables. We focus on the effect on pension effort of the particular strength
and constitution of governing coalitions in Model 7. The coalition variable is based on
the number of parties in government and their parliamentary status. It ranges from strong
(lowest value signifies one single party government in which one party takes all
government seats) to weak (highest value signifies a caretaker government not intended
to undertake any kind of serious policymaking). Models 8 and 9, in turn, test the effect of
welfare regime type as defined by, respectively, Lynch (2006) and Esping-Andersen
(1990). The rationale for these variables is familiar. Extensive bodies of literature have
documented the important role played by the particular model of welfare regime in
shaping social spending (e.g. Castles 2004; Esping-Andersen 1990; 1999; Korpi and
Palme 2003; Myles and Pierson 2001). On the other hand, contemporary theories of
welfare state politics provide less clear-cut hypotheses on the role of government
coalition strength. Scholars of blame avoidance often argue that retrenchment efforts will
become less transparent, hence less risky, if blame is shared by bringing many political
players on board (Weaver 1986; Pierson 1994; 2001). A larger number of participating
parties is then predicted to increase the potential for pension retrenchment. The veto
player approach (Bonoli 2001; Huber and Stephens 2001; Tsebelis 1999) argues the
opposite. It assumes that parties are more or less cohesive entities representing the

21
interests of specific groups. A larger number of political parties in government decreases
the possibility to achieve substantial policy reform, especially cutbacks, as coalition
partners need to compromise and protect a larger number of interest groups.
Results show that the strength of the governing coalition matters mainly for
pension program size. Interestingly, weaker coalitions appear better able to reduce
program size. This lends support to the blame avoidance thesis over the veto player
thesis. Note also that our period dummies in these Models 7 again support a
'retrenchment politics' interpretation. Both the middle period and the most recent period
effects are systematically negative for all three dependent variables. In addition,
regarding program size and generosity scores, the size of the period dummy is much
larger in the case of the most recent period 1996-2002 as compared to the middle period.
Table 6
Model 8 in Table 6 represents a first attempt to investigate the effect of welfare regime
type. Including time-invariant variables for the regime affiliation into the statistical model
now requires applying the between-effects estimator, since the random effect assumption
has to be rejected on both theoretical and methodological grounds. The between-effects
estimator draws solely on the cross-sectional information in the data using period
averages. Instead of applying the wrong model (random effects) for the wrong reasons
(time invariant variables), we have therefore decided to use a between-effect estimator
providing unbiased estimation coefficients.
3
Lynch's (2006) influential historical-
institutional theory explains varying degrees of public policy old-age-bias according to
two critical junctures: the early twentieth-century choice between citizenship-based and

22
occupational ways of structuring welfare provision, and the choice after World War II
between programmatic and particularistic modes of electoral competition. These
junctures, Lynch argues, produced a continuum of old-age spending bias among
contemporary OECD democracies. Here we focus only on both extremes of this
continuum, which coincide with the institutional-structural divide between "Beveridgean"
regimes with universal or means-tested benefits and "Bismarckian" regimes with income-
related contributory benefits. Model 8 thus includes only Lynch's (2006: 57) nine most
citizenship-based (or Beveridgean) welfare regimes and her five regimes classified as
most occupational (or Bismarckian): Italy, Austria, Belgium, the USA and Japan (no
comparable data were available on Greece and Spain). Controlling for macro-economic
variables and ODR, both these regimes types turn out to have a negative though not
significant effect on pension generosity scores and a positive, non-significant effect on
real benefits per elderly. Turning to program size as a share of GDP, the independent
effect of citizenship-based welfare regimes is significant and negative, indicating that the
institutional set-up of this first set of Beveridgean welfare regimes can be conducive to
overall pension program cutbacks. The effect of occupational regimes, while also
negative, is much smaller and non-significant.
However, Lynch's (2006: 57) citizen-based welfare regimes are really a motley
crew composed of Nordic welfare regimes (Sweden, Denmark, Norway, Finland) and
Anglo-Saxon regimes (the UK, Ireland, Canada, Australia and New Zealand). These, of
course, comprise two very distinct models of welfare capitalism that differ markedly on a
wide range of dimensions, including, but not limited to, their degrees of de-
commodification, post-tax-and-transfer redistribution and poverty alleviation and,
importantly for our purposes, the relative generosity and size of their public pension

23
systems. To scrutinize the effect of welfare regime type on pension effort in greater
detail, Model 9 therefore further subdivides only Lynch's nine citizen-based regimes,
according to Esping-Andersen's (1990) classic typology into liberal and social-
democratic regimes. This provides more specific information regarding pension
retrenchment differences between these two Beveridgean regime subsets. Applying these
regimes dummies, Model 9 indicates that social-democratic regimes also pursue pension
program size cutbacks (Kitschelt 2001). Regarding real benefits per elderly, the positive
association with citizen-based regimes from Model 8 now appears, in Model 9, to be
near-zero in the liberal subset, but still positive in the social-democratic subset. In similar
vein, Model 8's finding of a small negative relationship between citizenship-based
regimes and generosity scores is revealed in Model 9 to have been composed of a much
larger positive effect in the case of social-democratic regimes and an equally large
negative effect for liberal regimes. In other words, unpacking Lynch's citizenship-based
regimes indicates that while both citizenship-based regime types have cut overall pension
program size, benefit generosity has been cut or stabilized in the liberal cases but has
actually been expanded in the social-democratic cases.
Overall, these regression analyses indicate that pension spending efforts in public
pension systems are strongly but not solely driven by macroeconomic trends and fiscal
needs. In addition, institutional differences as expressed in welfare regime design appear
to be a continuing core factor behind pension effort, independently of population ageing
as such. Even in our sample of fast-aging OECD democracies, larger elderly voting
weights do not translate automatically into expansion of either program size or benefit
generosity. This interpretation is in line with Castles's (2004: 133-134) pointer that the
expansion of OECD pension spending between 1980 and 1998 have been less strongly

24
related to population aging than to pension system coverage and generosity – variables
that reflect the institutional set-up of welfare regimes. On the other hand, as the politics
of welfare retrenchment are not the mirror image of welfare expansion, ratchet effects are
likely to be present. That is, the larger electoral clout of elderly voters may more strongly
constrain governments' scope for pension cutbacks. As Castles (2004: 133-137) notes,
between 1980 and 1998, OECD social expenditure retrenchment has been least extensive
where populations have been aging most rapidly and where the initial elderly
constituency was the largest to start with. Conversely, the fact that liberal welfare
regimes are younger demographically and leaner substantively (Table 2) may have
allowed them to further reduce already lower levels of pension program size and pension
benefit generosity (Table 6, Model 9).
In other words, both social needs and political-institutional structure are likely to
mediate the electoral power-pension spending nexus, and to shape the policy effects of
population aging. The role of welfare regime type is, in addition, likely to reflect distinct
labor-market-welfare-state constellations. For instance, in the social-democratic regimes,
macro-level pension program cuts have gone hand in hand with increases in generosity
scores and in real benefits in these already generous pension systems. This is a joint result
of much better labor market participation levels, which provide better fiscal foundations
for public pensions systems, and of higher levels of investment in human capital through
education spending and labor market spending (Iversen and Stephens 2008; Vanhuysse
2008).
Consider, by contrast, occupational (or conservative-corporatist) welfare regimes,
which cope with high old age dependency rates of a magnitude comparable to those in
Denmark and Norway. These regimes stand out in terms of all three measures of pension

25
effort (Table 2). This may be partly a result of the institutional ways in which these
regimes create a strong perception of pensions as acquired rights among voters, in
addition to the highly patronage-based modes of political competition in countries such as
Belgium, Austria, and Italy. This is consistent with Scharpf's (2000) suggestion that in
conservative welfare systems, the contributory-financed pension entitlements tend to
acquire a quasi-legal status of property rights, leading pensioners to expect, and fight for,
protection against retrospective cuts. Pensions are seen as a vested entitlement, and
contributors expect to draw something out of the system once they retire. In addition,
conservative regimes uniquely combine a high level of state involvement in providing
generous earnings-related pensions, a particularly strong spending bias in favor of elderly
age cohorts, and high levels of early labor market exit (e.g. Ebbinghaus 2006; Esping-
Andersen 1999). Levels of early and disability retirement in conservative-corporatist
countries such as Germany, Italy, France and Belgium have been eclipsed in recent
decades only by those in post-communist Hungary and Poland (Vanhuysse 2006).
Similarly, conservative regimes occupied five of the top six spots amidst a population of
21 OECD countries on Lynch's (2006) ranking of pro-elderly public policy bias (see also
Esping-Andersen and Sarasa 2002). As Castles (2004: 127) reports, pension programs
have kept expanding between 1980 and 1998 in the OECD due to causal factors other
than population ageing, most dramatically where generous early retirement, low levels of
labor market participation and high pension benefit generosity have created a "triple
whammy" spending drive.
The mediating effect of political institutions in ageing democracies is still far from
being settled either theoretically or methodologically (Kittel and Winner 2005). Future
comparative research on the institutional foundations of welfare state policies would

26
benefit from larger cross-sectional pension effort data not currently available, or from
international survey data to introduce institutional pension measures into a pooled
analysis. Other effects of population aging on public policy remain to be explored in
future research. This article has focused exclusively on a single welfare program, albeit
the one that constitutes the largest share of social spending budgets widely across the
OECD. To more fully understand the multiple effects of population ageing, future
research should disaggregate government spending further into different programmes
other than pensions, in order to take into account the functional relations between welfare
programs (Rose 1984, 1985). Since growth in public finance is constrained by growth in
revenues and by growth of other government programs, changes in either of these factors
can tighten or loosen pension spending constraints (Rose and Karran 1987). In times of
growing fiscal austerity, retrenchment requirements at the macro level can thus lead
governments to try and reduce spending growth on either pensions or competing
programs, or, alternatively, to increase revenues (Rose and Peters 1978). At the micro
level, fiscal retrenchment strategies are more likely to lead governments to reduce
upward adjustments of pension benefits that take into account inflation and/or economic
growth, rather than reducing absolute nominal money values of pension benefits.
DISCUSSION
Studying the political economy of social security between in 1960, 1970s, and 1980s,
Breyer and Craig (1997:721) concluded that public pensions are the “kudzu of
government programs,” in that they keep expanding every decade. Including more recent
data in the analysis, our study has come to different and more refined conclusions.
Particularly since the late 1980s, population aging has indeed been associated with

27
increased macro-level pension spending in 18 OECD democracies. However, our
conclusion is not just that the positive size effect or "philosopher-king" hypothesis (H1)
cannot be rejected. We also found that populating aging has been associated with
decreasing pension benefit generosity or, at least, a halt to increasing generosity since the
late 1980s. Thus we can also reject the theoretically more specific hypothesis of a
positive benefit effect (H2), which is increasingly put forward, or implicitly assumed, by
the political economy literature and by international financial institutions. The argument
that "elderly power" is likely to lead to more generous pension benefits cannot, thus far,
be corroborated empirically. If anything, we have found tentative support for the less
frequently proposed negative benefit effect or "fiscal leakage" hypothesis (H3) in the
form of the negative effect of population aging on real pension expenditures per elderly
person, though not on pension generosity scores.
The weak support for elderly power predictions by median voter models may not
come as much of a surprise to empirical researchers doing more richly contextualized
analysis. At the macro level, median voter models ignore the degree to which aggregate
public expenditures such as pensions are driven by path dependence and long-term
spending commitments that make it difficult for governments to change spending effort at
will (Rose 1990). At the micro level, the assumption that current pension incomes are
pensioners' dominant concern ignores large bodies of attitudinal and behavioral research,
indicating, among other things, that pensioners rarely behave like floating voters (Butler
and Stokes 1974) and do not tend to vote as a homogenous single-issue constituency
(Goerres 2009). Recent studies indicate that elderly voters actually care less about the
real value of their pensions than about health issues (Rose 2008a,b). Future research
could therefore usefully analyse the effects of population aging on health care spending.

28
Health spending not only disproportionately benefits elderly voters; it also tends to be
supported by non-elderly voters, who can benefit in both present and future life cycle
stages. Moreover, unit cost growth in health spending may rise faster than that in pension
spending. For both reasons, health care macro-spending growth might well exceed
pension spending growth in the face of population aging. Lastly, health spending may
have important second-order effects for subsequent population aging, to the degree that it
affects longevity.
In sum, we have argued that it is the logic of retrenchment politics and double
fiscal-electoral straitjackets, not gerontocracy, which reigns supreme in the political
economy of pensions today. More than in previous decades, contemporary public pension
politics is driven by tighter macro-fiscal constraints, caused by the increasing budgetary
pressures accompanying the growth of elderly cohorts. Larger pensioner populations
have indeed led to larger overall spending commitments. At the same time, the
concomitant budgetary pressures appear to have led to lower real pension expenditures
per elderly person, as if governments were forced to cut smaller slices out of larger cakes.
This interpretation is compatible with the very substantial amount of pension reform and
retrenchment activity that has been observed across the OECD in the 1990s and the
present decade (e.g. Castles 2004; Huber and Stephens 2001; Immergut et al. 2007;
Pierson 2001; Lindbom 2007). Future work integrating also the revenue side of pension
spending (Rose and Karran 1987; Swank and Steinmo 2002) as well as human capital
investment policies (Iversen and Stephens 2008; Vanhuysse 2008) promises to shed
further light on the political dynamics of population aging. Our findings also point to the
need to pry open further black boxes regarding the political tactics employed, the
concrete measures used, and the coalitions involved, in program-specific retrenchment.

29
Another conclusion points to the continuing primacy of comparative politics. The
dialectical pressures arising from aging electorates and budgetary constraints are
mediated by institutions. Partisan effects may be on the wane in contemporary welfare
state politics.
4
But political institutions still matter crucially. In particular, the design and
organizational set-up of welfare state regimes, and their interaction with labor markets,
shapes the extent to which numerical pressures are translated into policy outcomes. These
variables, too, receive comparatively little attention in median voter models, yet they
crucially mediate the effect of absolute voter numbers on public pension effort. Looking
towards future developments, growing pensioner constituencies may continue to reshape
the distributional and electoral logics of welfare state politics, forcing those governments
in most dire financial straits to mitigate the severity of benefit cuts by an increased
reliance on obfuscation measures, grandfathering clauses, blame buffering, pension
indexation and benefit formula changes, and other hidden retrenchment measures. Such
tactics certainly appear to have formed an essential part of the pension reforms that have
been pursued across Europe since 1990. There is no doubt, then, that demographic
pressures will continue to exert severe pressures on many OECD public pension systems
in the decades ahead. But importantly, fast-aging welfare states are not yet at present
driven by the political logic of gerontocracy.

30
ENDNOTES
1
While we follow common practice by studying the ratio of 65plussers to working-age
people on pension effort, the reported effect of this measure on our second dependent
variable (pension spending per elderly) may be negatively biased, as independent and
dependent variables incorporate the number of 65plussers in, respectively, the
numerator and the denominator. This provides a further rationale for using pension
replacement rates as an alternative measure of pension generosity.
2
While choosing eight-year periods for averaging is somewhat arbitrary, it is consistent
with Breyer and Craig (1997), who use four ten-year periods, and Disney (2007), who
uses three ten-year periods. Lindert (1997) uses five four-year periods.
3
Note, however, that findings for the between-effect estimator specification should not
be overvalued since the statistical model suffers from a small number of observations
(N=18). This concerns particularly Table 8 (real pension expenditure per elderly) where
negative values for the adjusted R-squared already indicate that the overall statistical
model is not significant.
4
We ran models similar to those in Table 6 to explore the effect of partisan politics,
measured as the share of cabinet seats belonging to the dominant leftwing, respectively
rightwing, party. Findings (available on demand) indicate that the effect of both
ideological formations in power, though negative, is consistently small and non-
significant, for all three dependent variables.

31
REFERENCES
Allan, J. and Scruggs, L. (2004), ‘Political partisanship and welfare state reform in
advanced industrial societies’, American Journal of Political Science 48(3): 496–
512.
Allison P. (1994), ‘Using panel data to estimate the effects of events’, Sociological
Methods and Research 23: 174–99.
Anderson K. and Lynch, J. (2007), ‘Reconsidering seniority bias - Aging, internal
institutions, and union support for pension reform’, Comparative Politics, 39(2):
189–208.
Armingeon, K., Leimgruber, P., Beyeler, M. and Menegale, S. (2007), Comparative
Political Dataset 1960-2004, Institute of Political Science, University of Berne.
Barr, N (2001) The Welfare State as Piggy Bank, Oxford: Oxford University Press.
Blome, A., Keck, W. (2008), 'Young and Old in the Welfare State – How Social Policies
Influence Intergenerational Imbalances,' paper presented at the Conference on
Politics and Public Policies in Ageing Societies, Max Planck Institute for the
Study of Societies, Cologne, 24-25 July.
Bonoli, G. (2001). 'Political Institutions, Veto Points, and the Process of Welfare State
Adaptation,' in Paul Pierson (ed.). The New Politics of the Welfare State. Oxford
and New York: Oxford University Press.
Breyer, F. and Stolte, K. (2001) ‘Demographic change, endogenous labor supply and the
political feasibility of pension reform’, Journal of Population Economics 14: 409–
424.
Breyer, F. and Craig, B. (1997) ‘Voting on social security: Evidence from OECD
countries’, European Journal of Political Economy 13: 705–724.
Browning, E. (1975), ‘Why the social insurance budget is too large in a Democracy,’
Economic Inquiry 13: 373–388.
Busemeyer, M.R., Goerres, A., Weschle, S. (2008) Demands for Redistributive Policies
in an Era of Demographic Aging, Cologne: MPIfG Working Paper 08/3.
Butler, D. and Stokes, D. (1974), Political Change in Britain: The Evolution of Electoral
Choice. London: Macmillan.
Castles, F.G. (2004), The Future of the Welfare State: Crisis Myths and Crisis Realities.
Oxford: Oxford University Press.

32
Disney, R. (2007), ‘Population ageing and the size of the welfare state: Is there a puzzle
to explain?’, European Journal of Political Economy 23(2): 542–553.
Ebbinghaus, B. (2006), Reforming Early Retirement in Europe, Japan and the USA.
Oxford: Oxford University Press.
Esping-Andersen, G. (1990), The Three Worlds of Welfare Capitalism. Cambridge,
United Kingdom: Polity Press.
Esping-Andersen, G. (1999) Social Foundations of Postindustrial Economies. New York:
Oxford University Press.
Esping-Andersen, G. and Sarasa, S. (2002), ‘The generational conflict reconsidered’,
Journal of European Social Policy, 12(1): 5–21.
Galasso, V. (2006), The Political Future of Social Security in Aging Societies, Boston:
MIT Press.
Galasso, V. and Profeta, P. (2002), ‘The political economy of social security: A survey’,
European Journal of Political Economy 18: 1–29.
Galasso, V. and Profeta, P. (2007), ‘How does ageing affect the welfare state?’,
European Journal of Political Economy 23: 554–563.
Goerres, A. (forthcoming 2009). The Political Participation of Older People in Ageing
Europe. Basingstoke: Palgrave Macmillan.
Halaby, C. (2004), ‘Panel Models in Sociological Research: Theory into Practice’,
Annual Review of Sociology 30: 507–544.
Hicks, A. and Zorn, C. (2005), ‘Economic Globalization, the Macro Economy, and
Reversals of Welfare Expansion in Affluent Democracies, 1978-1994',
International Organization 59: 631–662.
Hicks, A. and Zorn, C. (2007), "Errata, Revisions, and Data: ‘Economic Globalization,
the Macro Economy, and Reversals of Welfare Expansion in Affluent
Democracies, 1978–94’, posted online by the authors at:
http://www.sociology.emory.edu/ahicks/IO
Huber, E. and Stephens, J (2001) Development and Crisis of the Welfare State. Chicago:
University of Chicago Press.
IMF (2004) World Economic Outlook. Washington DC: International Monetary Fund.
Immergut, E. Anderson, K., and Schulze, I. (eds.) (2007) The Handbook of West
European Pension Politics. Oxford: Oxford University Press.

33
Iversen, T., Stephens, J.D. (2008), 'Partisan Politics, the Welfare State, and Three Worlds
of Human Capital Formation,' Comparative Political Studies 41(4-5): 600-637.
Kitschelt, H. (2001) 'Partisan competition and welfare state retrenchment: When do
politicians choose unpopular policies?' in P. Pierson (ed), The New Politics of the
Welfare State. Oxford: Oxford University Press, pp. 265-302.
Kittel, B. and Obinger, H. (2003) ‘Political parties, institutions, and the dynamics of
social expenditure in times of austerity’, Journal of European Public Policy
10(1):20–45.
Kittel, B. and Winner, H. (2005) ‘How reliable is pooled analysis in political economy?’
European Journal of Political Research 44:269–293.
Korpi W. and Palme, J. (2003) 'New politics and class politics in the context of austerity
and globalization: Welfare state regress in 18 countries, 1975-95, American
Political Science Review 97(3):425-446.
Kotlikoff, L.J. and Burns, S. (2004) The Coming Generational Storm. Cambridge: MIT
Press.
Lindbom, A. (2007) 'Obfuscating retrenchment: Swedish welfare policy in the 1990s,'
Journal of Public Policy, 27(2): 129-150.
Lindert, P. (1996) ‘What limits social spending?’ Explorations in Economic History 33:
1–34.
Lynch, J. (2006) Age and the Welfare State, Cambridge, Cambridge University Press.
Mulligan, C., and Sala-i-Martin, X. (2003), ‘Social Security, Retirement, and the Single-
Mindedness of the Electorate’, Cambridge: NBER Working Paper 9691.
Mulligan, C. and Sala-i-Martin, X. (1999), ‘Gerontocracy, retirement and social security’,
Cambridge: NBER Working Paper 7119.
Myles, J. and Pierson, P. (2001) ‘The comparative political economy of pension reform'
in P. Pierson (ed.), The New Politics of the Welfare State. Oxford: Oxford
University Press: 305–333.
Nickell S. (1981)‚ ‘Biases in dynamic models with fixed effects’, Econometrica 49:
1417–26.
Nielsen F and Alderson A (1995) Income inequality, development, and dualism: results
from an unbalanced cross-national panel. American Sociological Review, 60:
674–701.

34
OECD. (1996) Ageing in OECD Countries: A Critical Policy Challenge. Paris.
OECD (2006) Economic Outlook Database, Paris.
OECD (2007a). Social Expenditure Database 1980-2003, Paris.
OECD (2007b). Labor Force Statistics, Paris.
OECD (2007c). Pensions at a Glance, Paris.
Pampel, F. (1994) 'Population Aging, Class Context, and Age Inequality in Public
Spending,' American Journal of Sociology 100(1):153-195.
Pampel, F. and Williamson, J.B. (1985) ‘Age Structure, Politics, and Cross-National
Patterns of Public Pension Expenditures,’ American Sociological Review, 50 (6),
782-798.
Pampel, F. and Williamson, J.B. (1989) Age, Class, Politics and the Welfare State.
Cambridge: Cambridge University Press.
Perotti, R. (1996), ‘Growth, Income Distribution, and Democracy: What the Data Says’,
Journal of Economic Growth 1: 149–187.
Persson, T. and Tabellini, G. (2000), Political Economics, Cambridge, MIT Press.
Pierson, P. (1994) Dismantling the Welfare State? Cambridge University Press,
Cambridge.
Pierson, P., ed. (2001) The New Politics of the Welfare State. Oxford: Oxford University
Press.
Preston, S. (1984) 'Children and the elderly: Divergent paths for America's dependents.'
Demography 21:435-457.
Razin, A. and Sadka, E. (2007) ‘Aging population: The complex effect of fiscal leakages
on the politico-economic equilibrium’, European Journal of Political Economy
23(2): 564–575.
Razin, A., Sadka, E. and Swagel, P. (2002) ‘The aging population and the size of the
welfare state’, Journal of Political Economy 110(4): 900–918.
Rose, R. (1984), Understanding Big Government. The Program Approach, London:
Sage.
Rose, R. (1985), 'The Programme Approach to the Growth of Government,' British
Journal of Political Science 15(1): 1-28. '
Rose, R. (1990), 'Inheritance before Choice in Public Policy,' Journal of Theoretical
Politics 2(3): 263-291.

35
Rose, R. (2008a), 'Relating multiple resources to multiple needs of older people,' paper
presented at the Conference on Politics and Public Policies in Ageing Societies,
Max Planck Institute for the Study of Societies, Cologne, 24-25 July.
Rose, R. (2008b), 'Older Europeans in the aftermath of transformation: A bottom-up
perspective,' paper presented at the joint AARP-European Centre for Social
Welfare in Vienna Meeting on Re-Inventing Retirement: reshaping Health and
Financial Security, 22-24 October.
Rose, R. and Karran, T. (1987), Taxation by political inertia. Financing the growth of
government in Britain, London: Allan & Unwin.
Rose, R, and Peters, G. (1978) Can Government Go Bankrupt? New York: Basic Books.
Sabbagh, C., and Vanhuysse, P. (2008), 'Intergenerational Justice Perceptions and the
Role of Welfare Regimes: A Comparative Analysis of University Students,' paper
presented at the International Sociological Association (ISA) conference,
Barcelona, 5-8 September.
Scharpf, F. (2000), ‘The Viability of Advanced Welfare States in the International
Economy: Vulnerabilities and Options’, Journal of European Public Policy 7(2):
190–228.
Scruggs, L. (2005) Welfare State Entitlements Data Set, Version 1.1.
Scruggs, L. (2006) Welfare-state decommodification in 18 OECD countries: a replication
and revision, Journal of European Social Policy, 16(1): 55-72.
Shelton, C. (2008), ‘The ageing population and the size of the welfare state: Is there a
puzzle?’ Journal of Public Economics 92: 647-651.
Sinn, H.-W. and Uebelmesser, S. (2002), ‘Pensions and the path to gerontocracy in
Germany’, European Journal of Political Economy 19: 153–158.
Swank, D., and Steinmo, S. (2002), 'The New Political Economy of Taxation in
Advanced Capitalist Democracies,' American Journal of Political Science
46(3):642-655.
Tabellini, G. (2000) ‘A positive theory of social security’, Scandinavian Journal of
Economics 102(3): 523–545.
Tsebelis, G. (1999), 'Veto players and law production in parliamentary democracies: An
empirical analysis,' American Political Science Review 93: 591-608.
United Nations (2006) World Population Prospect: Population Database, New York.

36
Vanhuysse, P. (2001) 'The political economy of pensions: Western theories, Eastern
facts,' Journal of European Public Policy 8(5):853-861.
Vanhuysse, P. (2006) Divide and Pacify: Strategic Social Policies and Political Protests
in Post-Communist Democracies. Budapest-New York: Central European
University Press.
Vanhuysse, Pieter (2008), 'Review Article: The New Political Economy of Skill
Formation,' Public Administration Review, 68(5): 955-959.
Van Parijs, P. (1998), ‘The disfranchisement of the elderly, and other attempts to secure
intergenerational justice’, Philosophy and Public Affairs 27(4):292–333.
Weaver, K.R. (1986), 'The Politics of Blame Avoidance,' Journal of Public Policy 6(4),
371-398.
World Bank (1994), Averting the Old Age Crisis. Oxford, Oxford University Press.
World Bank (2006), World Development Indicators, Washington.

37
ACKNOWLEDGEMENTS
We gratefully acknowledge funding from the Haifa Centre for German and European
Studies (HCGES). For helpful comments, we thank Friedrich Breyer, Irwin Collier,
Achim Goerres, Olli Kangas, Bernhard Kittel, Bernd Marin, Frances Rosenbluth, this
journal's Editor and anonymous reviewers, and seminar participants at the ECPR 2007
General Conference in Pisa, the July 2008 Workshop on the Politics of Aging at the Max
Planck Institute in Cologne, and the European Centre for Social Welfare Policy and
Research in Vienna. For providing excellent research environments during the writing of
this paper, we thank the Leitner Program in International and Comparative Political
Economy at Yale University, the DFG Graduate School Research on Organizational
Paths at the Free University Berlin, the Max Planck Institute, and the European Centre in
Vienna.

38
Table 1. Existing empirical evidence on the effect of population ageing on public pensions
Author(s) N Period Dependent Variable Demographic variables
Pampel and Williamson
(1989)
18 1950-1980
(7 time points)
Old age pension expenditure per GNP Prop65+ (+)
Pampel and Williamson
(1989)
18 1950-1980
(7 time points)
Old age pension expenditure per elderly/ GNP
per capita
Prop65+ (+)
Lindert (1996) 19 1960-1981
(4 yr averages)
Pension expenditure per GDP Prop65+ (+)
Breyer, Craig (1997) 20 1960-1990
(10 yr averages)
Pension benefits per GNP Median voter age (+)
Tabellini (2000) 40 1978-1982
(cross sectional)
Social security contributions as % of GNP Prop65+ (+)
Razin, Sadka, Swagel
(2002)
13 1965-1992
(annual)
Labor taxes as % of GDP Dependency ratio (1 minus labor
force divided by pop) (-)
Castles (2004) 21 1980-1998 Old age cash benefits (% of GDP) Prop65+ in 1960 (+) Non sign.
Castles (2004, Ch. 6) 21 1980-1998 Old age cash benefits (% of GDP) Change in Prop65+ (+)
Castles (2004, Ch. 6) 21 1980-1998 Old age cash benefits (% of GDP) Change in Prop65+/(GDP/cap)
(+)
Castles (2004, Ch. 6) 21 1980-1998 Retrenchment in 12-component measure of
social expenditure
Prop65 in 1960 (- = fewer cuts)
Hicks, Zorn (2005) 18 1978-1994 Social expenditure as % of GDP retrenchment
events
Prop65 (+)
Hicks, Zorn (2007
Errata)
18 1978-1994 Social expenditure as % of GDP retrenchment
events
Prop65 (non sig. effect)
Disney (2007) 21 1975-1995 (10 yr
averages)
Labor taxes as % of GDP Old age dependency ratio (+)
Shelton (2008) 13 1965-1992
(annual)
Labor taxes as % of GDP Old age dependency ratio (+)

39
Table 2. Changes between 1980 and 2002: old age dependency rates and three measures of pension effort
Country Old age dependency ratio
a
Pension exp. per GDP
b
Pension generosity scores
c
Real pen. exp. per elderly
d
1980 2002 1980 2002 1980 2002 1980 2002
Australia 14.8 18.9 3.1 3.2 9.1 8.1 9.7 9.7
Austria 24 22.8 9.8 12.4 11.8 13.1 9.8 10
Belgium 21.7 25.9 5.9 6.9 11.4 12.4 9.1 9.2
Canada 13.9 18.5 2.8 4 10.3 12.7 9.8 9.8
Denmark 22 22.2 4.6 5.3 13.5 13.5 13.4 13.4
Finland 17.7 22.8 4.7 4.7 14.7 12.6 9.2 9
France 21.9 25 7.5 10 13.6 12.6 9.4 9.7
Germany 23.4 27 9.8 10.9 7.6 6.7 9.6 9.7
Ireland 18.2 16.4 4 2.5 7.8 10.4 8.9 8.8
Italy 19.3 26.9 7.2 11.2 8.7 15.3 9.5 9.4
Japan 13.5 27.6 2.9 6.8 8.1 9.8 19.3 19.1
Netherlands 17.4 20.2 5.4 4.6 13.3 13 9.3 9.1
New Zealand
15.4 18.1 6.8 4.5 15.5 15.8 10.8 10.1
Norway 23.4 22.8 4 4.9 13.9 15 13.4 13.7
Sweden 25.4 26.5 6.6 6.8 17 11.4 13.6 13.8
Switzerland 21.4 23.7 5.5 6.4 7.4 6.4 10.9 10.5
UK 23.3 24.3 3.8 5.3 7.5 9.3 8 8.2
USA 17.1 18.6 5.1 5.4 11.7 11.4 9.8 9.6
Mean 19.7 22.7 5.5 6.4 11.3 11.6 10.8 10.7
Stand. Dev. 3.8 3.5 2.1 2.9 3.1 2.7 2.7 2.7
Range 11.9 11.1 7 9.9 9.6 9.4 11.4 10.9
Coef. Var. 0.2 0.2 0.4 0.4 0.3 0.2 0.3 0.3
Note: a=Old age dependency ratio measured as the share of elderly (65+) as a percentage of the working age population (15-64) (OECD 2007b),
b=Cash pension benefits as a percentage of GDP (OECD 2007a), c=Generosity of public pension benefits in percentage points (Scruggs 2005),
d=Real cash pension expenditure in $ US at constant prices (2000) per working age population (15-64) log transformed (OECD 2007a)

40
Table 3. Determinants of pension expenditure per GDP
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Random
Fixed Fixed Fixed Fixed Fixed
GDP per capita (log) 0.79 0.97 -0.55 0.72 1.8 1.81
[1.4] [1.7] [1.4] [1.7] [1.5] [1.7]
GDP growth -0.43*** -0.45*** -0.39*** -0.48*** -0.51*** -0.50***
[0.1] [0.1] [0.1] [0.1] [0.1] [0.1]
Interest rate -0.16** -0.16* -0.16** -0.18* -0.19* -0.21**
[0.07] [0.08] [0.07] [0.09] [0.09] [0.1]
Inflation rate 0.12 0.12 0.19** 0.12 0.093 0.072
[0.08] [0.09] [0.09] [0.09] [0.06] [0.1]
ODR 0.13* 0.093 0.088* 0.12*
[0.07] [0.07] [0.05] [0.07]
1988-1995 -0.026 -0.074 1.46* 0.11 -0.87 -0.93
[0.6] [0.8] [0.8] [0.8] [0.7] [0.8]
1996-2002 -0.38 -0.44 2.47* -0.21 -1.18 -1.8
[0.9] [1.3] [1.4] [1.3] [1.2] [1.4]
ODR 20 year dif. -0.16***
[0.05]
55plus ratio 0.031
[0.03]
Elderly part. -0.13***
[0.03]
Female part. 0.0006
[0.03]
Observations 54 54 54 54 52 52
Number of id 18 18 18 18 18 18
R-squared . 0.54 0.66 0.54 0.79 0.61
Adj. R
-squared . 0.47 0.6 0.46 0.75 0.54
Note: N=18 countries, T= averages for 3 periods (1980-1987, 1988-1995, 1996-2003) Robust
standard errors in brackets, *** p<0.01, ** p<0.05, * p<0.1 levels of significance, constant
included but not reported.

41
Table 4. Determinants of pension generosity scores
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Random
Fixed Fixed Fixed Fixed Fixed
GDP per capita (log) 2.68 5.93 3.52 5.62 6.95* 5.18
[2.3] [3.6] [3.0] [3.5] [4.0] [3.1]
GDP growth -0.11 -0.28 -0.21 -0.37* -0.35 -0.53**
[0.2] [0.2] [0.2] [0.2] [0.2] [0.2]
Interest rate 0.12 -0.0035 -0.01 -0.045 -0.056 -0.18
[0.2] [0.1] [0.1] [0.1] [0.1] [0.1]
Inflation rate 0.04 0.047 0.17* 0.062 0.0052 -0.096
[0.1] [0.1] [0.09] [0.1] [0.1] [0.1]
ODR 0.16 0.19 0.21 0.17
[0.1] [0.2] [0.2] [0.1]
1988-1995 -0.86 -2.54 -0.029 -2.23 -3.53* -4.20**
[1.1] [1.7] [1.3] [1.6] [2.1] [1.7]
1996-2002 -1.76 -4.74 0.024 -4.31 -6.10* -7.35**
[1.9] [2.9] [2.2] [2.7] [3.5] [2.8]
ODR 20 year dif. -0.26***
[0.08]
55plus ratio 0.045
[0.06]
Elderly part. -0.052
[0.06]
Female part. 0.16***
[0.06]
Observations 54 54 54 54 52 52
Number of id 18 18 18 18 18 18
R-squared . 0.24 0.4 0.21 0.32 0.46
Adj. R
-squared . 0.13 0.31 0.093 0.19 0.36
Note: N=18 countries, T= averages for 3 periods (1980-1987, 1988-1995, 1996-2003) Robust
standard errors in brackets, *** p<0.01, ** p<0.05, * p<0.1 levels of significance, constant
included but not reported.

42
Table 5. Determinants of real pension expenditure per elderly (log)
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Random
Fixed Fixed Fixed Fixed Fixed
GDP per capita (log) 0.2 0.16 0.12 0.29 0.034 -0.012
[0.5] [0.4] [0.5] [0.4] [0.3] [0.4]
GDP growth -0.074***
-0.073***
-0.037 -0.065** -0.069** -0.073**
[0.03] [0.03] [0.03] [0.03] [0.03] [0.03]
Interest rate -0.038* -0.038 -0.030* -0.029 -0.027 -0.034
[0.02] [0.02] [0.02] [0.02] [0.02] [0.02]
Inflation rate 0.027 0.028 0.026 0.024 0.034* 0.028
[0.02] [0.02] [0.02] [0.02] [0.02] [0.02]
ODR -0.039 -0.039** -0.047***
-0.043***
[0.02] [0.01] [0.01] [0.01]
1988-1995 -0.37* -0.35** -0.37* -0.43*** -0.24 -0.26
[0.2] [0.2] [0.2] [0.2] [0.2] [0.2]
1996-2002 -0.34 -0.31 -0.35 -0.41 -0.044 -0.16
[0.3] [0.2] [0.4] [0.3] [0.3] [0.3]
ODR 20 year dif. -0.0018
[0.01]
55plus ratio -0.015**
[0.007]
Elderly part. -0.020*
[0.010]
Female part. 0.0045
[0.009]
Observations 54 54 54 54 52 52
Number of id 18 18 18 18 18 18
R-squared . 0.78 0.74 0.78 0.81 0.77
Adj. R
-squared . 0.75 0.7 0.74 0.78 0.72
Note: N=18 countries, T= averages for 3 periods (1980-1987, 1988-1995, 1996-2003) Robust
standard errors in brackets, *** p<0.01, ** p<0.05, * p<0.1 levels of significance, constant
included but not reported.

43
Table 6. Politics and institutions in pension expenditure per GDP
Pension exp. per GDP Pension generosity scores Real pen. exp. per elderly
Model 7 Model 8 Model 9 Model 7 Model 8 Model 9 Model 7 Model 8 Model 9
Fixed Between Between Fixed Between Between Fixed Between Between
GDP per capita (log) 1.76 -3.44 -3.73 5.68 -1.81 -4.35 0.3 -0.084 -1.96
[1.8] [3.9] [3.9] [3.6] [5.9] [4.9] [0.4] [6.1] [6.3]
GDP growth -0.48*** -0.79 -0.79 -0.28 -0.71 -0.72 -0.077** -0.45 -0.27
[0.1] [0.6] [0.6] [0.2] [1.0] [0.8] [0.03] [1.0] [1.0]
Interest rate -0.16* 0.057 -0.0031 -0.0034 2.01** 1.51** -0.038* -0.53 -0.39
[0.08] [0.5] [0.5] [0.1] [0.8] [0.7] [0.02] [0.9] [0.8]
Inflation rate 0.073 0.56 0.65 0.06 -1.65 -0.89 0.02 -0.55 -0.5
[0.08] [0.7] [0.7] [0.1] [1.1] [0.9] [0.02] [1.1] [1.2]
ODR 0.098 0.26 0.22 0.19 0.018 -0.32 -0.039** 0.14 -0.088
[0.07] [0.2] [0.2] [0.2] [0.3] [0.3] [0.02] [0.3] [0.3]
1988-1995 -0.58 -2.38 -0.43**
[1.0] [1.7] [0.2]
1996-2002 -1.04 -4.55 -0.41
[1.5] [2.9] [0.3]
Gov. coalition type -0.43* 0.13 -0.071
[0.2] [0.4] [0.06]
Citizenship-based sys
-3.18* -0.35 3.67
[1.4] [2.2] [2.3]
Occupational systems
-0.061 -0.42 2.57
[1.3] [2.0] [2.0]
Liberal -3.43** -2.63 0.0004
[1.4] [1.8] [2.3]
Social democratic -2.88* 2.12 3.24
[1.3] [1.6] [2.1]
Observations 54 54 54 54 54 54 54 54 54
Number of id 18 18 18 18 18 18 18 18 18
R-squared 0.6 0.69 0.69 0.25 0.5 0.67 0.8 0.36 0.35
Adj. R
-squared 0.53 0.48 0.48 0.11 0.15 0.44 0.76 -0.087 -0.1
Note: N=18 countries, T= averages for 3 periods (1980-1987, 1988-1995, 1996-2003) Robust standard errors in brackets, *** p<0.01, ** p<0.05, *
p<0.1 levels of significance, constant included but not reported.

44
Appendix Table 1. Definition and source of variables
Variable Definition Source
Pension expenditure per
GDP
Cash pension benefits as a percentage of
GDP
OECD Social Expenditure
Database (2007)
Pension generosity
scores
Index composed of public pension minimum
and standard replacement rates, minimum
qualifying period, and pensions take-up rate
Scruggs (2005) Welfare
State Entitlements Dataset
Real pension benefits
per
elderly
Real cash pension expenditure in $ US at
constant prices (2000) per working age
population (15-64) log transformed
OECD Social Expenditure
Database (2007), WDI
Log GDP per capita Log of real Gross Domestic Product per
capita
Penn World Tables (2007)
Real growth rate Real Gross Domestic Product growth World Bank (2007) World
Development Indicators
Interest rate Long-term interest rate Armingeon et al. (2007)
Comparative Political
Dataset
Inflation Consumer price index growth World Bank (2007) World
Development Indicators
Old age dependency
ratio
Old age dependency ratio measured as the
share of the elderly (65+) as a percentage of
the working age population (15-64)
OECD (2007) Labor Force
Statistics
ODR 20 year dif. Projection for ODR in year t+20 minus ODR
in year t
United Nations (2006) World
Population Prospect
ODR 55 plus Old age dependency ratio measured as the
share of the elderly (55+) as a percentage of
the working age population (15-54)
OECD (2007) Labor Force
Statistics
Elderly part. Elderly (55-64) labor force participation rate
(percentage)
OECD (2007) Labor Force
Statistic
Female part. Female labor force participation rate
(percentage)
OECD (2007) Labor Force
Statistic
Gov. coal. type Classification: (1) single party majority
government (2) minimal winning coalition (3)
surplus coalition (4) single party minority
government (5) multi party minority
government
Woldendorp, Keman and
Budge (1998), Armingeon et
al. (2007)
Citizenship-based Citizenship-based welfare regimes with
universal and means-tested systems (SWE,
DNK, NOR, FIN, UK, IRE, CAN, AUS, NZL)
Lynch (2006: 57)
Occupational systems Occupational welfare regimes with
particularistic competition (ITA, BEL, AUT,
USA, JPN )
Lynch (2006: 57)
Liberal welfare regime AUS, CAN, IRE, NZL, UK Esping-Andersen (1990)
Social dem. welfare
regime
DNK, NOR, FIN, SWE Esping-Andersen (1990)

45
Appendix Table 2. Summary statistics
Variable Obs. Mean Std. Dev. Min. Max.
Pension exp. per GDP 54 6.20 2.43 2.30 12.11
Pension gen. Scores 54 12.13 2.94 6.26 18.49
Real pen. exp. per elderly
54 10.68 2.60 8.01 19.32
GDP per capita (log) 54 9.77 0.33 8.89 10.39
GDP growth 54 2.56 1.20 0.65 8.74
Interest rate 54 8.46 3.25 1.87 15.57
Inflation 54 4.27 3.02 -0.02 13.31
ODR 54 21.01 3.31 14.62 27.60
ODR 20 year difference 54 7.14 5.09 -2.16 21.69
ODR 55 plus 54 42.80 6.38 31.55 54.56
Elderly participation rate 52 48.24 13.01 23.02 69.69
Female participation rate 52 61.91 10.68 38.15 80.04
Gov. coalition type 54 2.55 1.17 1 5
Citizenship-based reg. 54 0.50 0.50 0 1
Occupational reg. 54 0.28 0.45 0 1
Liberal welfare state 54 0.28 0.45 0 1
Social dem. welfare state
54 0.22 0.42 0 1

46
About the authors:
Markus Tepe is a graduate of Exeter University and the Westfälische-Wilhelms
University Münster and is currently a Research Fellow at the Collaborative Research
Center 597 "Transformations of the State" at the University of Oldenburg. He has been a
member of the German Research Foundation's Doctoral Research Program on Path
Dependence at the School of Business and Economics at the Free University Berlin. His
research centers on political economy, public policy, generational conflict, comparative
politics methodology, and comparative welfare state research and has appeared in
Public
Choice
.
Pieter Vanhuysse received his PhD from the London School of Economics. A former
Fellow of the Institute for Advanced Study at Collegium Budapest, he is currently
Lecturer at the University of Haifa. His work centres on the comparative politics of social
policies, education, human capital, and democratic transitions, and has appeared in
journals such as
Public Choice,
Journal of Social Policy, Political Studies, Journal of
European Public Policy, Politics, Europe-Asia Studies, International Journal of
Sociology and Social Policy, International Journal of Social Welfare,
and
Public
Administration Review
. Dr Vanhuysse's book
Divide and Pacify: Strategic Social
Policies and Political Protests in Post-Communist Democracies
(Central European
University Press) was nominated for The American Sociological Association's Award for
Distinguished Contribution to Scholarship, Section on Political Sociology, for 2006.