European Journal of Population 11:371-398, 1995.
@ 1995 Kluwer Academic Publishers. Printed in the Netherlands.
3 71
Adjustments of Public Pension Schemes in Twelve
Industrialized Countries: Possible Answers to
Population Ageing
J E A N - P I E R R E G O N N O T ~, C H R I S T O P H E R P R I N Z 2 a n d
NICO KEILMAN 3
1Regional Adviser for Population and Development, Economic Commission for Europe, Palais des
Nations, Geneva 10, Switzerland; 2European Centre for Social Welfare Policy and Research,
Berggasse 17, A-1090 Vienna, Austria; 3Statistics Norway, P.O. Box 8131 Dep., N-0033 Oslo,
Norway
Received 24 January 1995; accepted 8 June 1995
Abstract. This article analyses the impact of population dynamics on future public pension expenditure in twelve industrialized countries. Whereas previous studies have mainly emphasized ageing
effects, this study looks into the consequences of changing marital status structures as well. Old
age pensions, disability, and survivor's pensions are investigated. Various sets of demographic and
pension scenarios are formulated for the projections, dealing with changes in demographic, labour
force and pension system variables in the future. The analyses show that there can be no adequate
demographic response to rising pension costs caused by population ageing at the horizon 2030.
Neither an increase in fertility nor an inflow of migrants can rejuvenate national populations, unless
fertility and/or migration reach unrealisticallyhigh levels. Instead, substantial reductions of the public
pension burden have to be sought in socioeconomic measures.
Gonnot, J.P., Prinz, C. et Keilman, N., 1995. Ajustement des syst6mes de retraite publics dans douze
pays industrialisds. Rdponses possibles au vieillissement de la population.
R~sum~. Cet article analyse l'impact des dynamiques de population sur les ddpenses de retraite
venir dans douze pays industrialis6s. Alors que des 6tudes ant6rieures avaient surtout insist6 sur les
effets du vieillissement,Cet article examine les cons6quences des changements dans les structures par
6tat matrimonial. Les retraites pour les figes avanc6s, l'invalidit6 et les retraites du conjoint survivant
sont explor6s. Diff6rents ensembles de scdnarios d6mographiques et de retraite sont formulds par
des projections, portant sur les changements ddmographiques, les changements dans la population
active et dans les syst6mes de retraite qui peuvent intervenir dans le futur. Les analyses montrent
qu'il n'y a pas de r6ponse d6mographique addquate au cofits croissants des retraites, caus6s par le
vieillissement ddmographique g l'horizon de l'an 2030. Ni un accroissement de la f6condit6 ni un
afflux de migrants ne peuvent rajeunir les populations nationales, ~ moins que la f6condit6 et/ou la
migration n'atteignent de tr~s hauts niveaux peu rdalistes. Au contraire, des r6ductions substantielles
du poids des retraites publiques doivent atre recherch6es dans des mesures socio-dconomiques.
1. Introduction
T h e c h a n g i n g d e m o g r a p h i c structure o f p o p u l a t i o n s i n i n d u s t r i a l i z e d c o u n t r i e s has
a n u m b e r o f s o c i o e c o n o m i c c o n s e q u e n c e s , a m o n g w h i c h that for social e x p e n d i t u r e
372
JEAN-PIERRE GONNOT, CHRISTOPHER PRINZ AND NICO KEILMAN
has attracted most attention in recent years. An important factor here is the marked
age-dependency of social expenditure. Although an ageing population may have
implications for aspects such as economic growth, aggregate demand, the labour
force, and income distribution, the problem of resources necessary in the future for
maintaining current standards of living among the elderly in developed countries,
in particular in the long term, has probably most often been an issue of concern to
policy makers and scientists (Myers, 1994). In 1985, almost 9 per cent of GDP in
OECD countries was spent on public pensions (OECD, 1988a), and the substantial
rise in both the number and proportion of the elderly in most industrialized countries
is likely to have a major upward effect on this share. Indeed, whereas OECD
projections indicate a twenty per cent rise in social expenditure spending during
the period 1980-2040, the growth in public pension expenditures alone may be as
large as almost 80 per cent (OECD, 1988b).
In this paper we investigate what the impact of dynamics in living arrangements
and age structures might be on future public pensions expenditures in industrialized
countries. The paper summarizes the main results of the international comparative
project "Social security, household and family dynamics in aging societies", which
was recently completed at the International Institute for Applied Systems Analysis (IIASA) in Laxenburg, Austria. The project includes various demographic and
pension scenarios for pension costs until the year 2050 for 12 countries: Austria,
Canada, Czechoslovakia, ~Finland, France, Germany, 1 Hungary, Italy, the Netherlands, Norway, Poland, and Sweden. It extends earlier comparative studies (cf.
Heller et al., 1986; Holzmann, 1987) into several directions.
i. It facilitates the investigation of the impact of marital status on social security,
in addition to fertility, mortality and international migration. Marital status is
not only important for widow's pensions, but also for old-age public pensions.
The reason is that entitlements, in particular those for females, are dependent
on marital status in many countries, through the intermediate effect of work
histories (numbers of years worked).
2. Future trajectories of demographic and pension variables are analysed for
various scenarios, bothTor demographic and socioeconomic variables.
3. It includes a number of countries from Central and Eastern Europe, in addition
to western countries.
4. Pension benefits are endogenous, and pension projections are country-based,
instead of scheme-based. In a number of countries, several schemes exist for
various segments of the labour force. These schemes were collapsed into one
average scheme for each country, so as to facilitate national projections, instead
of scheme-based projections.
The present paper summarizes only the main findings of the project. An extensive text has recently appeared (Gonnot et al., 1995), and the interested reader
is referred to that volume for details not included here. The book also includes
extended case studies for Austria, the Netherlands and Sweden, dealing with a
ADJUSTMENTS OF PUBLIC PENSION SCHEMES
373
parity-dependent pension system, the impact of household dynamics, and the consequences of economic growth for public pensions, respectively.
2.
M e t h o d
Two sub-models have been used to investigate the problem. A demographic model
projects the population broken down by age, sex, and marital status. The model
is of the multidimensional (multistate) cohort-component type (Keilman, 1985).
A pension model (adapted after Malabouche, 1987) determines the benefits from
and the contributions to state pensions. First, a further breakdown by work status
(whether or not in the labour force) is applied to the population for each combination
of age, sex and marital status. Next, pension contributions are related to wages and
work status, and benefits are determined by the work history and demographic
parameters of the individual. The pension model assumes that total benefits are
found as the product of four factors: annual benefit rate, average number of years
insured, average gross salary and population retired. Average number of years
insured are found on the basis of a cohort rearrangement of age-specific labour
force participation rates. The model covers the following three arrangements: old
age pensions, survivor pensions for widows over 55, and disability pensions.
Two types of demographic input data were used for the projections in each
country:
the structure of the population broken down by sex, five-year age group,
and four marital statuses (never-married, currently married, divorced and
widow(er)) as of 1985;
- age- and sex specific rates for fertility, mortality, marriage formation and
dissolution, and (for some countries) international migration, applying to the
period 1980-1984.
In addition to these demographic input data we used the following socioeconomic data (annual figures, as of 1985):
- labour force participation rates broken down by age, sex and (for women)
marital status;
- average gross salary by sex;
average pension benefit rate;
average contribution rate.
Some of these observed data may not be very representative for the past, or for
the decades to come. Therefore, in order to deal with the uncertainty connected
to future demographic and socioeconomic developments, several scenarios were
designed. These are not aimed at producing realistic forecasts, but rather at investigating the range of possible demographic and socioeconomic changes, and to
study their impact in terms of ageing and changes in the marital composition of the
elderly, and of work history. Four demographic scenarios were designed, common
for all countries. The changes included in three of these four scenarios gradually
take place over the period 1985-2005.
374
JEAN-PIERRE GONNOT, CHRISTOPHER PRINZ AND NICO KEILMAN
~'~a~
~
aaa~
O
ggRg8
t I~
I T ~ o
~ , ~ = = ~ ~ 8
•-x. ~ . ~ . , a . ' , R .
o..,.x o , x .
O
=
~ oo
0
8~
TM
~
~ ~.
o
0
0
Z
~
~ F ~ , ~
,,-,
,,..,
o
,,.,o
c'q,!"-I
~,1
~1
~
~",,1~1
',"-~JL'a
°
0
,,,
ADJUSTMENTSOF PUBLIC PENSION SCHEMES
375
1) A constant rates scenario with rates remaining constant at their 1980-84 level
which shows how much change is already embodied in the age and marital-status
structure of the population, and serves as a basis for comparison.
2) A replacement fertility scenario which assumes that fertility will gradually
reach replacement level in the year 2005. For ten countries, in which fertility
was below replacement in 1985, this implies a high fertility scenario whereas it
corresponds to constant and decreasing fertility in Czechoslovakia and Poland,
respectively from the levels current in 1985.
3) A low mortality scenario under which age-specific mortality rates are
decreased by 30% for women and 45% for men between 1985 and 2005. In terms
of life expectancy at birth, it is roughly equivalent to an increase of 8-10 years for
males and 4--5 years for females, so that the sex differentials are approximately
reduced by one-half.
4) A Western low rates scenario which combines the most extreme demographic
rates observed in the period 1980-84 in Western Europe: West German fertility
(1.28 children per woman), Swedish marriage and divorce (one-third never married,
mean age at first marriage of 28 for women and 30 for men, one-third of all
marriages ending in divorce), Swiss mortality (a life expectancy at birth of 74
years for men and 81 years for women).
International migration is set to zero in,all four scenarios.
For nine countries, country-specific scenarios were also prepared, and these
are summarized in Table 1. Country-specific scenarios can be seen as "realistic"
scenarios, at least for the first few decades of the projection period. Based on
assumptions which are similar or close to those of the medium variant of the
official national projections (as of 1990), they are aimed at depicting a probable
future.
For six countries, the national scenarios also provide an assessment of the
demographic impact of international migration. In Austria and Canada, the net
migration is assumed to be twice the current level. In Germany, it is assumed that
the country receives 2.1 million Germans living in Eastern Europe between 1985
and 2000 in addition to the 520,000 annual migrants reported during the period
1985-1990. A second scenario is based on the assumption of an additional 200,000
net migrants annually beyond the year 1990. In France, net migration is assumed
to remain at its estimated current level in 1985. In the Netherlands a decrease is
expected from 25,000 annual net immigrants in 1985 to 20,000 in 1995. In Sweden
the decline in the net number of immigrants is assumed to be steeper, from 29,000
in 1988 to 10,000 in 2010.
Both national scenarios prepared for France as well as national scenario II for
Sweden also permit the assessment of the consequence of a continuous long-term
decline in mortality.
Among the socioeconomic scenarios which were used in the project there is
one with increased female labour force participation, a raise in the mean age
at retirement, and one with a mixed capital-funded/pay-as-you-go-system. These
376
JEAN-PIERREGONNOT,CHRISTOPHERPRINZAND NICO KEILMAN
Table 2. Countries broken down by changes in old-age dependency ratio for the period 1985-2030,
various scenarios (percentage points)
Constant
rates
Replacement
fertility
Low morality
Westernlow
rates
National
30+
Germany
Netherlands
Canada
25-30
Finland
Italy
Austria
Germany
Canada
Netherlands
Finland
20-25 Netherlands
Canada
Germany
France
Norway
Sweden
Italy
Austria
Hungary
Netherlands
Canada
Germany I
France I
Poland
France
Norway
Austria
Germany II
France II
Sweden
Czechoslovakia
Hungary 1I
Czechoslovakia
Sweden II
15-20 Finland
Italy
Canada
Netherlands
10--15 Austria
France
Norway
Sweden
Germany
Finland
Italy
France
Poland
Hungary
Austria
Norway
Sweden
Poland
Hungary
Czechoslovakia
5-10
Poland
Hungary
0-5
Poland
Hungary I
Sweden I
Czechoslovakia Czechoslovakia
scenarios are presented as they are introduced in section 4, together with a number
of other adjustments to the pension system.
3. Projection results for four demographic scenarios - present pension
systems
3.1. AGEING IS CERTAIN
The scenarios' impact on ageing is summarized in Tables 2 and 3. Table 2 shows
countries according to changes in old-age dependency ratio (OADR) over the
period 1985-2030 for each o f the scenarios, and the O A D R values themselves are
ADJUSTMENTSOF PUBLICPENSIONSCHEMES
377
Table 3. Old Age Dependency ratio, 1985 and 2030, various scenarios
1985
2030
Constant
rates
Replacement
fertility
Low
mortality
Western
low rates
Nat.I
Nat.II
Cut-off age 65 (population 65+/population 15-64)
Austria
Canadaa
Czechoslovakia
Finland
21.2
15.7
17.1
18.4
35.6
37.6
20.9
35.1
32.0
35.2
20.9
32.5
46.1
47.3
28.2
46.0
44.7
43.4
33.8
43.4
39.6
36.5
29.4
France
Germany
Hungary
Italy
19.7
21.1
18.7
19.0
31.6
41.1
23.8
34.7
30.4
36.0
22.7
31.4
40.2
53.2
32.9
45.0
39.1
49.0
39.4
41.2
40.0
41.5
23.8
Netherlands
Norway
Poland
Sweden
17.6
24.5
14.5
27.0
40.6
35.0
21.2
37.8
36.3
32.3
22.1
35.1
51.3
44.8
28.2
48.8
43.7
39.9
34.6
42.0
38.5
24.2
35.8
36.1
38.4
30.2
38.9
Cut-off age 60 (population 60+/population 15-59)
a
Austria
Canada a
Czechoslovakia
Finland
32.1
23.7
26.2
27.7
55.4
53.0
30.6
50.8
48.9
49.2
31.0
46.6
68.5
64.5
38.7
64.0
67.8
61.3
47.6
62.2
58.6
51.0
41.7
France
Germany
Hungary
Italy
30.0
31.3
30.0
30.0
46.2
66.5
35.5
54.2
44.3
56.1
33.7
48.2
56.5
82.3
46.4
66.8
57.0
75.6
55.5
63.2
56.5
67.2
35.5
Netherlands
Norway
Poland
Sweden
26.3
36.1
23.1
39.3
61.6
53.2
29.8
55.3
54.2
48.4
31.2
50.7
74.7
65.1
38.0
68.4
66.2
60.6
47.1
61.8
57.5
34.3
51.7
51.4
61.4
43.2
55.5
2031
presented in Table 3. 2 Regarding the four c o m m o n scenarios, four main conclusions
arise from these tables. First, ageing is certain for all countries and scenarios.
However, sharp differences between countries and a marked dispersion can be
noted. Canada, G e r m a n y and the Netherlands are top-ranking countries and will
see their O A D R doubling and possibly tripling, reaching a level situated between
35 and 55 elderly per 100 persons aged 1 5 - 6 4 in 2030, while at the other end of the
spectrum, Czechoslovakia, Hungary, Poland and Sweden will experience a more
limited ageing. Second, an increase in fertility will be o f little help to prevent ageing
378
JEAN-PIERREGONNOT,CHRISTOPHERPRINZANDNICOKEILMAN
Table 4. Populationaged 60 and over. Constantrates scenario
1985
2000
Absolute (1000s) Index(1985 = 100)
Austria
Canada
Czechoslovakia
Finland
France
Germany
Hungary
Italy
Netherlands
Norway
Poland
Sweden
2015
2 0 3 0 2050
1,508
3,823
2,536
861
102
124
101
112
110
172
123
143
129
206
126
148
105
184
133
128
9,973
16,050
1,930
10,607
113
110
102
120
133
113
111
126
150
125
104
142
141
92
99
122
2,434
882
5,212
1,930
118
95
120
97
154
113
147
113
184
130
159
115
163
116
172
100
until 2030 - between minus 1 and minus 5 points in the OADR as compared with
the constant rates scenario - unless fertility jumps far beyond replacement level.
Third, further mortality improvements would lead to dramatic levels of ageing,
especially for Western countries: plus about one-third in the OADR under the low
mortality scenario as compared with the constant rates scenario, but it is impossible
to tell if this is an upper limit. Fourth, ageing in East-European countries could be
as strong as in West-European countries, if fertility and mortality drop substantially
and simultaneously as exemplified by the Western low rates scenario.
Table 3 shows that the timing and the extent of ageing also depends on the
definition of our ageing indicators. If, for the calculation of the OADRs, age 60
is taken as the cut-off age, instead of age 65, ageing looks more severe in all
countries, and differences between countries become larger. However, the order of
countries regarding the level of and the increase in the OADR remains largely the
same. Differences in the order also vary over time. By 2030, ageing in Austria and
particularly in Germany comes out more strongly when age 60 is taken as cut-off
age, while the opposite conclusion holds for Canada. These differences are largely
a consequence of the size of the population aged 60-64 in 2030 (birth cohorts
1966-70) in relation to the size of the older cohorts. In general, though, the choice
of the cut-off age is of little relevance and should rather depend on the aims of the
study. Since the focus in this paper is on pensions, and since average retirement
age in European countries is closer to 60 than to 65, the elderly will be defined as
all persons aged 60 or older in the remaining sections.
Results from national scenarios mainly reflect the assumption made with respect
to mortality. When moderate mortality changes are assumed, results are very close
to those of the constant rates scenario. This is the case for Canada, Germany,
ADJUSTMENTS OF PUBLIC PENSION SCHEMES
379
Hungary I, the Netherlands, Poland and Sweden. When mortality is expected to
decline substantially - France II and Hungary II - results are similar to those of
the low mortality scenario. Finally, when a strong decrease is assumed for both
fertility and mortality, as for Czechoslovakia, results are analogous to those of the
Western low rates scenario.
A second important finding arising from two special migration scenarios, which
were computed for Austria and Canada, is that the impact of immigration on ageing
is limited. Assuming the same annual numbers of immigrants as in the respective
national scenarios (i.e. 15,000 immigrants with a mean age of about 25 years for
Austria, and 148,000 immigrants with a mean age of about 31 years for Canada),
together with the constant rates scenario for the other demographic components,
results in a decrease in the OADR of 3 points for Canada and 4 points for Austria.
Calculations for Austria also showed that the effect of immigration on the OADR
depends very much on the age pattern of migrants (results not shown here). If
immigrants to Austria would have an age pattern similar to that of immigrants to
Canada - i.e an mean age of about 31 years instead of 25 years - it would require an
annual net number of migrants of 28,000 instead of the expected 15,000 to lower
the OADR by 5 points by the year 2030.
Table 4 shows that a great variety of changes in the size of the elderly population
will be associated with ageing. In the three top ranking countries for ageing Canada, the Netherlands and Germany - the size of the elderly would respectively
double, increase by 84 per cent and by 25 per cent between 1985 and 2030 under
the constant rates scenario. Hungary would experience limited ageing and hardly
any increase in the size of its elderly population. Poland would experience limited
ageing while its elderly population would increase by a substantial 59 per cent.
This apparent contradiction can be explained by a growth in the population of
working ages which is only a little under that in the elderly population. Sweden
indicates both limited ageing and a limited growth of the elderly population. If
only a moderate decline in mortality occurs in the future (as assumed under the
Western low rates scenario), the maximum growth of the elderly population will
take place in most countries between 2000 and 2015 (figures not shown here).
But in case there is a strong decline in mortality (the low mortality scenario) the
elderly population will continue to grow rapidly beyond 2015. Over the period
1985-2030, the average growth rate of the population aged 60 and over would
lie between about 0.5 and 1.5 per cent (depending on the country) under constant
mortality, and between about 1 to 2 per cent under the low mortality scenario.
3.2. MARITAL STATUS STRUCTURES OF THE ELDERLY IN THE FUTURE
The current marital composition of the population aged 60 and over mostly reflects
the sex differential in mortality. It is characterized by a strong contrast between
the high proportion of males who are married and the much lower proportion for
females married. Given our interest in pension issues, the focus of this section will
380
JEAN-PIERREGONNOT,CHRISTOPHERPRINZAND NICO KEILMAN
Table 5. Marital composition of the population aged 60 and over, 1985 and 2030 (percentage, total equals 100 for each sex)
Males
S
M
Austria 1985
Austria 2030
Constant rates
Low mortality
Western low rates
National
Canada 1985
Canada 2030
Constant rates
Low mortality
Western low rates
National
Czechoslovakia 1985
Czechoslovakia 2030
Constant rates
Low mortality
Western low rates
National
Finland 1985
Finland 2030
Constant rates
Low mortality
Western low rates
France 1985
France 2030
Constant rates
Low mortality
Western low rates
National I
National II
Germany 1990
Germany 2030
Constant rates
Low mortality
Western low rates
National I
National II
D
W
Females
S
M
Sex ratioa
D
W
6
77
3
13
11
35
5
49
57
16
15
18
16
65
65
55
62
7
7
12
8
12
13
15
14
14
13
14
13
36
44
37
40
11
11
16
11
40
32
33
37
74
88
78
81
8
79
3
11
8
49
3
41
77
12
12
16
11
65
65
56
68
9
10
14
9
14
14
14
12
9
9
11
8
40
47
35
44
12
13
16
12
40
32
38
37
76
89
75
77
4
78
4
15
5
39
5
52
68
9
9
10
12
66
65
57
60
8
8
17
11
17
18
16
17
5
5
5
6
32
42
34
37
11
11
19
14
53
42
42
44
70
86
77
81
9
74
5
13
13
36
6
44
60
24
22
24
52
51
49
12
12
14
13
14
14
17
16
16
29
36
31
15
15
18
40
33
36
69
84
77
8
76
3
13
9
42
4
46
69
16
15
17
15
15
63
63
57
65
65
7
7
13
7
7
13
15
13
12
12
12
11
13
11
11
36
44
35
40
41
10
10
15
10
10
42
34
37
38
38
72
85
77
76
77
4
78
3
15
8
38
5
49
59
21
20
23
19
18
59
59
53
60
61
7
7
13
8
8
13
15
12
13
13
13
12
13
10
10
36
44
34
39
39
10
10
16
10
11
41
34
37
41
40
78
92
79
80
80
3 81
ADJUSTMENTSOF PUBLIC PENSION SCHEMES
Table 5. Continued
Males
S
M
Hungary 1985
Hungary 2030
Constant rates
Low mortality
Western low rates
National I
National II
Italy 1985
Italy 2030
Constant rates
Low mortality
Western low rates
Netherlands 1985
Netherlands 2030
Constant rates
Low mortality
Western low rates
National
Norway 1985
Norway 2030
Constant rates
Low mortality
Western low rates
Poland 1985
Poland 2030
Constant rates
Low mortality
Western low rates
National I
Sweden 1985
Sweden 2030
Constant rates
Low mortality
Western low rates
National I
National II
D
W
Females
S
M
Sexr~io
D
W
4
77
3
16
6
37
5
52
68
11
10
12
12
12
64
64
56
61
58
9
9
17
12
11
16
18
16
16
19
4
4
5
4
4
28
38
32
30
28
14
14
19
16
15
54
44
44
50
53
68
85
76
68
77
7
79
0
14
12
42
0
45
72
11
11
15
73
71
61
1
1
10
15
17
14
8
8
10
42
51
38
2
2
10
48
39
42
75
89
77
6
78
3
13
10
46
4
40
74
19
17
20
21
61
61
55
55
9
9
12
12
11
12
13
12
12
12
i3
13
38
45
36
37
13
13
16
15
37
30
35
35
76
88
78
80
11
73
3
12
12
46
4
39
77
22
20
23
54
54
51
12
12
14
12
14
12
14
14
15
35
42
34
14
14
16
37
31
35
78
91
80
3
82
2
13
7
40
3
51
65
9
9
11
9
71
71
58
74
4
4
14
4
16
17
16
13
5
5
6
5
33
44
36
38
8
8
16
8
55
43
43
49
68
85
78
68
12
69
7
12
10
45
7
37
79
33
31
33
32
36
42
42
41
43
40
15
15
16
15
14
11
13
11
11
10
24
23
24
23
28
27
33
27
30
28
18
18
17
18
17
31
27
31
30
28
79
91
77
81
81
a) Number of men per 100 women.
a
382
JEAN-PIERRE GONNOT, CHRISTOPHER PRINZ AND NICO KEILMAN
40-
30-
20-
[ ] Constant
rates
10-
%
[ ] Replacement
fertility
l
-10-
-20.
-30-
-40.
[ ] Western low
rates
[ ] National
o
_
,~
~
'~
~-
=
o
Fig. 1. Percentage changes in the labour force between 1985 and 2030, four demographic
scenarios.
be on widows and widowers. Further details on marital status of the elderly can be
found in Gonnot et al., (1995, pp. 61-67).
Figures in Table 5 indicate that the current pattern will change substantially. A
main future demographic trend is the strong increase in the proportion of single
and divorced old men. As very limited changes in the proportion of widowers
among males aged 60 and over are expected - between 10 and 15 per cent under
all scenarios - the increase in the proportion of single and divorced will result in a
sharp drop in the proportion of males married. 3
Trends in the marital composition of the female elderly population are more
complex to picture. The magnitude and, sometimes, the direction of change in
the proportion of widows are somewhat uncertain. Under constant conditions, the
percentage of widows would significantly decrease in Austria, Germany, Sweden,
France, and the Netherlands. It will increase by 4 points in Poland while little
change would be observed in other countries. Under the low mortality scenario, the
percentage of widows would decrease in all countries by 6 to 18 percentage points
between 1985 and 2030. Under the Westem low rates scenario, the percentage
of widows would also fall in all countries. However, the decline would be more
limited than under the low mortality scenario. As a consequence of the uncertainty
ADJUSTMENTS OF PUBLIC PENSION SCHEMES
383
regarding trends in the percentage of widows, changes in the percentage of older
women who are married appear to be rather unpredictable for all countries but
Sweden, where a strong fall is to be expected.
3.3. THELABOURFORCE
When studying the consequences of ageing processes for the labour force, its
age composition is an important variable. All other factors remaining equal, an
ageing labour force implies less rapid progression in the hierarchy for younger
workers, slower adaptation to new technologies, and, possibly, lower productivity.
When it comes to public pension issues however, the age composition is not
without relevance, but we can focus on the size of the labour force. In the different
demographic scenarios it was assumed that activity rates by age, sex and marital
status will remain constant at their 1985 level. (In one of the other scenarios an
increase in female labour force participation was assumed, see section 4.1) Figure
1, which displays changes in the size of the labour force between 1985 and 2030
shows much variety in both direction and size among the different countries.
Under constant demographic conditions, the size of the labour force in 2030 would
increase by 23 per cent in Poland and 13 per cent in Czechoslovakia compared
with 1985, it would remain constant in France, and decrease in all other countries
by between 4 per cent in Canada up to a dramatic 37 per cent in Germany. If
fertility progressively returns to replacement level by 2005, it would only make a
difference of about 5 to 10 percentage points in the size of the labour force by the
year 2030 (a longer time horizon, e.g. 2050 would have resulted in a larger effect
of high fertility). Gains from a further decline in mortality would also be limited to
a few percentage points, because death rates are relatively low at working ages. As
compared with the constant rates scenario, the Western low rates scenario implies
a drop by 5-8 percentage points in Finland, France and Hungary, and of 12-13
points in Czechoslovakia and Poland. This contrasts with low fertility countries
- Austria, Germany, The Netherlands and Italy - where no additional changes
are observed. Largest differences with the constant rates scenario are observed in
national scenarios assuming a net inflow of migrants - plus 48 percentage points in
Austria, plus 35 in Canada, 28 in France, 17 in Sweden and 16 in the Netherlands.
3.4. THE PENSION BOOM
Several demographic and social developments have been discussed in the preceding
sections which are likely to result in a substantial increase in pension expenditures.
Reasons why pension expenditure will increase are the following. First, the number
of elderly people will increase. Second, among the female elderly, a growing
number will qualify for pension and therefore claim for it. Third, the average
female pension will be markedly higher in most of the countries. However, this
increase will be offset to some extent by a general decrease in the average male
384
JEAN-PIERRE GONNOT, CHRISTOPHER PRINZ AND NICO KEILMAN
Table 6. Indices of the growth in real pension expenditure (1985 = 100)
Scenario
2000
2015
2030
2050
Austria
Constant rates
Low mortality
Western low rates
National
105
110
108
101
115
135
128
126
128
157
145
155
105
140
124
153
Canada
Constant rates
Low mortality
Western low rates
National
127
133
129
147
172
201
179
212
212
262
222
303
190
256
202
326
Czechoslovakia
Constant rates
Low mortality
Western low rates
National
108
114
116
110
133
162
164
146
141
185
187
170
146
198
199
192
Finland
Constant rates
Low mortality
Western low rates
116
122
120
143
171
157
153
200
174
131
184
152
France
Constant rates
Low mortality
Western low rates
National II
115
120
118
119
138
162
148
143
160
198
175
176
152
200
169
209
Germany 1)
Constant rates
Low mortality
Western low rates
115
120
118
119
141
130
125
157
139
89
123
101
Hungary
Constant rates
Low mortality
Western low rates
National I
110
117
120
243
127
156
161
279
121
163
170
266
112
158
163
247
Italy
Constant rates
Low mortality
Western low rates
118
123
120
126
149
135
143
177
155
123
163
136
Netherlands
Constant rates
Low mortality
Western low rates
National
120
123
120
121
157
175
159
163
183
215
188
197
155
198
162
181
Norway
Constant rates
Low mortality
Western low rates
104
111
106
132
160
138
160
206
169
147
205
157
ADJUSTMENTSOF PUBLICPENSIONSCHEMES
385
Table 6. Continued
Scenario
2000
2015
2030
2050
Poland
Constant rates
Low mortality
Western low rates
National
115
121
122
115
145
172
174
145
155
198
199
155
166
218
217
164
Sweden
Constant rates
Low mortality
Western low rates
National
148
156
149
152
168
201
171
178
172
221
175
198
151
206
154
197
1 All social security related calculations for Germany in this table and subsequent ones
apply to the FRG only. Re-unification of the two Germanies made the old GDR pension
system obsolete, and adding FRG and GDR results (as was done for the demographic
part) would give irrelevant results.
Table 7. Ratio of growth of pension expenditures to growth of population aged 60 and
over 1985-2050. Constant rates scenario
1985/2000
2000/2015
2015/2030
2030/2050
Austria
Canada
Czechoslovakia
Finland
0.40
0.89
0.13
0.75
0.80
1.07
0.88
1.15
1.46
0.85
0.38
0.50
1.04
1.00
1.40
0.91
France
Germany
Hungary
Italy
0.87
0.67
0.20
1.11
0.87
0.75
0.53
0.75
0.77
2.00
1.17
0.94
1.13
0.92
0.56
1.00
0.90
- 1.251
1.33
0.97
0.64
0.90
0.80
1.15
0.61
1.20
0.50
0.75
1.08
1.18
0.71
Netherlands
Norway
Poland
Sweden
--0.061
1 Minus sign indicates negative growth rate for population aged 60 and over.
pension. Fourth, in several countries, the average pension will increase as the
pension s y s t e m b e c o m e s m o r e and, eventually, fully mature.
Table 6 reports indices o f the growth in real total pension expenditures as
projected under the constant rates, low mortality, Western low rates and national
scenarios for the 12 countries included in the present study. In the absence o f
migration, pension costs are projected to grow till around 2030 in m o s t countries,
exceptions being C z e c h o s l o v a k i a and Poland where the upward trend will continue b e y o n d that date. Highest pension expenditures are observed under the low
mortality scenario in Western countries, under the Western low rates scenario in
386
JEAN-PIERRE GONNOT, CHRISTOPHER PRINZ AND NICO KEILMAN
Czechoslovakia and Poland and under national scenario II in Hungary. Accordingly, pension expenditures would increase by about 60 per cent in Austria and
the former FRG, by between about 80 to 100 per cent in Czechoslovakia, Finland
and Italy. Pension expenditures would more than double in France, Hungary, the
Netherlands, Norway, Poland and Sweden, and more than triple in Canada.
Differences in indices between the constant rates scenario and the low mortality
or the Western low rates scenarios show that a further improvement in mortality
will add another 20 to 50 percentage points to the rise in pension expenditures.
The model used for preparing those pension projections does not allow a direct
assessment of the respective contribution of the different demographic and non
demographic factors to the increase in pension expenditures. This is due to the
fact that in this model, contrary to more simple pension models, average pension
is a dependent variable. However, it is possible to obtain a crude estimate of the
impact of the increase in the size of the elderly population on pension expenditures
by comparing the growth of pension expenditure to that of the population aged
60 and over. Figures in Table 7 show a great deal of variety in this ratio across
countries and over the period 1985-2000. Between 1985 and 2000, factors other
than the growth of the aged population will play a major role in the growth of
pension expenditures in Austria, Czechoslovakia, Hungary, Norway and Sweden.
In Sweden, where the pension system will reach full maturity during this period,
a very rapid growth of pension expenditures is expected - between 2.6 and 3.0
per cent annually - while the aged population will slightly decrease. On the other
hand, in Poland and possibly in Italy, pension expenditures will grow slower than
the elderly population. Between 2000 and 2015, it seems that ratios are more
homogenous and that the rapid growth of the elderly population will drive the
rise in pension expenditures. Beyond 2015, figures indicate again much diversity.
The picture becomes even more complex when other scenarios are considered.
However, a main conclusion is that the growth of the elderly population is only
one of the elements at stake and does not always play a major role.
4. Adjustments of public pension schemes
The burden of the demographic fate is impressive and inevitably leads to the
question: is it possible to preserve the actual performance of state pension systems,
and which reforms should be undertaken in order to do so? 4
Within the framework of a pay-as-you-go system financial disequilibria can
be solved by either decreasing total and/or individual benefits or increasing total
and/or individual contributions. Five possible measures were selected and simulated among the numerous existing proposals: an increase in female participation
in the labour force, a rise in age at retirement, a burden-sharing between people
who are economically active/retired people, an increase in contribution rate and the
creation of a complementary capital-funded pension scheme. These five proposals
are those most frequently put forward and have a special long-term relevance as
387
ADJUSTMENTS OF PUBLIC PENSION SCHEMES
40.
30.
20-
10.
%
o.
-10 •
-20 •
-30 •
• Constant
rates
[ ] High activity
-40
Fig. 2. Percentage changes in the labour force between 1985 and 2030, two labour market
scenarios.
they are believed to be the ones which are likely to have the greatest quantitative
impact. In addition, effects of increasing productivity were investigated.
4.1. INCREASINGWOMEN'S ECONOMIC ACTIVITY
In order to assess the impact of an increase in labour force participation of women
on pension systems, a high activity scenario was designed. It assumes that by the
year 2005, both males and females would follow the highest overall pattern of
activity observed in 1985 among the countries included in this study, namely that
of the former GDR. 5 Figure 2 shows the changes in the size of the labour force
over the period 1985-2030, under constant demographic conditions, alternatively
assuming constant and high activity rates.
Compared with the constant rates scenario, an increase in female activity either
strongly limits the decrease or substantially adds to the growth in size of the labour
force: plus 30 percentage points in Italy and the Netherlands, plus 20 in Austria and
France, and still around 15 in other non-Scandinavian countries. It is important to
note that it is much more than that gained under the replacement fertility scenario,
see Figure 1. On the other hand, an increase in female activity would also raise,
sometimes dramatically, women's average benefit entitlement: by 68% in Italy, 34%
388
JEAN-PIERRE GONNOT, CHRISTOPHERPRINZ AND NICO KEILMAN
Table 8. Change in average benefits per beneficiary. High activity and late retirement
scenarios versus constant rates scenario, 2030. Countries with pure earnings-related
pension
Austria
Czechoslovakia
France
Hungary
Italy
Poland
"High activity" scenario
"Late retirement" scenario
Females
Males
Females
26
-3
34
12
68
0
14
14
11
13
12
0
11
18
11
16
19
0
120-
100-
80,
60.
%
40-
20-
•
Constant
rates
[ ] Late
retirement
-20
Fig. 3. Percentage increase in the number of retirees between 1985 and 2030, two pension
scenarios.
in France, 26% in Austria and 12% in Hungary (see Table 8). In Czechoslovakia,
a slight decrease is indicated due to a lower participation of women at higher age
in the former GDR than in Czechoslovakia. In Poland, benefits would not change
because maximum benefits corresponds to a work record of 25 years.
Overall, this scenario would limit benefit-cuts by about 10 percentage points
with a maximum of 15 points in the Netherlands, figures which are above the results
389
ADJUSTMENTS OF PUBLIC PENSION SCHEMES
Table 9. Contribution rates: actual, and as required to balance the pension
account, 1985 and 2030. Low mortality scenario
Actual rate
Austria
Canada
Finland
France
Germany a
Italy
Netherlands
Sweden
Norway
Czechoslovakia
Hungary
Poland
22.7
3.6
18.7
14.0
18.7
24.2
17.7
19.5
Balance rate
1985
2030
32.0
10.6
17.6
17.7
20.7
30.6
12.6
19.5
16.1
16.0
22.1
19.0
63.1
27.9
40.9
34.0
48.0
62.5
31.7
50.0
34.3
25.0
39.4
29.1
a 1990.
of the replacement fertility scenario except for Austria. In addition, in a context
of high female economic activity, in the long run general provisions for survivor's
pensions do not seem any longer necessary. Waiving those provisions would give
additional substantial help to the state pension system.
4.2. RAISING AGE AT RETIREMENT
An often proposed solution to the pension problem is to raise the age at retirement.
In the present scenario, we assume that age at retirement is progressively raised
to 65 years on average for both males and females. 6 - still under constant demographic conditions - and that no survivors pensions are available under age 60. 7
As illustrated by Figure 3, this would strongly limit the increase in the number
of pensioners (to less than 25%) in countries with low age at retirement. In some
countries with high age at retirement (Canada, France, Finland) the gain is often
still substantial but would always not prevent a "pensioner boom".
An increase in age at retirement lengthens the average working-life and consequently results in increased benefit entitlement in countries with partially or
purely earnings-related schemes: between +11 and +14 percentage points for men
andbetween +11 and +19 percentage points for women (see Table 8). In Austria,
France and Italy the increase in benefits entitlement would be much less than under
the high activity scenario while, again, in Poland no difference would be observed
because of the 25-years rule. The positive impact on average benefits is much more
substantial than in the high activity scenario in a majority of countries. No deterioration in average benefits would be experienced in Czechoslovakia, Hungary and
Poland. Cuts in benefits will be limited to less than 10% in Austria and less than
390
JEAN-PIERREGONNOT,CHRISTOPHERPRINZ AND NICO KEILMAN
Table 10. Average benefits as a percentage of average net salary. Balanced
contributionrate. Constantrates scenario
1985
2000
2015
2030
2050
Austria
Finland
France
74
69
45
70
71
47
80
78
51
99
84
56
111
84
57
Germany
54
63
69
87
83
Italy
Netherlands
Sweden
70
31
53
75
32
88
82
35
98
105
39
106
112
40
108
25% in France, Italy and the Netherlands. Still, cuts would amount to about 35%
in Finland and Germany, and to about 45% in Canada.
4.3. INCREASING CONTRIBUTION RATE
Actual payroll tax rates, if obtainable, as well as balance rates for the countries in
1985 and in 2030 under the low mortality scenario are presented in Table 9. Only in
Finland and Sweden were the pension funds balanced in 1985. Small disequilibria
already exist in Austria, France, Germany and Italy. In Canada where the payroll
tax rate is extremely low - 3.6% - contributions fortunately do not represent the
main source of income for state pensions. Although these figures probably closely
reflect real situations, they should not be interpreted as an exact measure of the
present financial position of national state pension funds because of the many
simplifying assumptions on which the modelling is based. Under the low mortality
scenario - the worst case - the state pension fund would be balanced in 2030 in
Canada if the contribution rate were set to 28% of the gross earnings, in France and
the Netherlands to about 32-34% of gross earnings. Such levels of contribution
are just a few points higher than the highest actual rates (24% in Italy) and might
still be socially or economically affordable. On the contrary, this would not be the
case in the other countries of this sample: 41% in Finland, 48% in Germany, 50%
in Sweden, and about 63% in Austria and Italy.
Similar calculations can becarried out for Norway and Central Eastern European
countries 8 in order to assess at which level to set the payroll tax rate in order to
balance contributions and benefits when ageing is at a maximum. In Norway, the
balance contribution rate is estimated to be about 16% in 1985 and 34% in 2030
under the low mortality scenario. In order to fully fund national pension schemes
in 1985, contribution rates should have been set to 16% in former Czechoslovakia,
19% in Poland and 22% in Hungary in 1985 and respectively to 25%, 29% and 39%
in 2030 under the low mortality scenario. Under the Western low rates scenario,
contribution rates would have to be increased by an additional 5 percentage points
(figures not shown in Table 9). Compared with Western countries, Czechoslovakia
391
ADJUSTMENTSOF PUBLICPENSION SCHEMES
Table 11. Contributionrates with and without burden sharing (ratios benefits/gross salary
and benefits/net salary constant at their 1985 level). Constant rates scenario
1985
Austria
France
Germanya
Italy
22.7
14.0
18.7
24.2
2030
No burden sharing
Burden sharing
Difference
47.3
26.9
37.0
53.4
35.4
22.6
27.7
40.9
11.9
4.3
9.3
12.4
a 1990.
and Poland belong in the lower comer while figures for Hungary indicate a more
difficult situation.
4.4. BURDEN-SHARING BETWEEN ACTIVES AND RETIREES
In all calculations, it was assumed that benefits are indexed for wage increase. In
practice, this is achieved by linking benefits to gross income. However, under such
a system, net income decreases if the contribution rate goes up while gross income
and, consequently, benefits increase. This implies that the whole additional burden
is with the actives. In a context of ageing, this might result in a situation where
benefits are higher than net income. Under the constant rates scenario, this would
be the case beyond 2030 in Austria, Italy and Sweden (see Table 10). In order to
distribute the pension burden more equitably in the future, it has been proposed to
link benefits to net income.
Results of the simulation for four countries are presented in Table 11 and
indicate that burden sharing makes a substantial difference in terms of contribution
paid by the actives. By the year 2030, and under the constant rates scenario, the
contribution rate required to keep the ratio benefits/income constant at its 1985
level would be lower by about 4 percentage points in France and by 9 to 12 points
in Austria, Germany and Italy if benefits were to be linked to net income instead
of gross income.
4.5. CAPITAL-FUNDED VERSUS PAY-AS-YOU-GO SYSTEMS
Work by Samuelson (1958) and Aaron (1966) has shown that a pay-as-you-go
pension system provides every generation of workers with higher benefits than
a capital-funded system if the sum of the rates of growth of the population and
of the real income per worker exceeds the rate of interest. This conclusion was
reached in the context of a neo-classical growth model which assumed monotonic
change in the size of the cohort. In addition, the comparison refers to a pure mature
pay-as-you-go system versus a pure mature capital-funded system, and does not
consider whether a mixed system would perform better. The second limitation
has been addressed by several authors whose works have confirmed the general
JEAN-PIERREGONNOT,CHRISTOPHERPRINZAND NICOKEILMAN
392
Table 12. Contribution rates, mixed pension scheme. (Pay-as-You-Go + Complementary
Capital-Funded) 2030. Low mortality scenario
Pay-as-You-Go + Complementary
capital-funded (rate of interest)
Difference with pure
Pay-As-You-Go
1%
2%
1%
2%
Males
Austria
Canada
Finland
France
Germany
Italy
Netherlands
Sweden
39.0
10.1
29.1
25.9
29.8
41.2
39.2
30.5
34.9
8.4
26.6
22.9
26.9
36.9
37.8
27.6
-24.1
- 17.8
- 11.8
-8.1
-28.7
-21.3
7.5
- 19.5
-28.2
- 19.5
- 14.3
- 11.1
-31.6
-25.6
6.1
-22.4
Widowedfemales
Austria
Canada
Finland
France
Germany
Italy
Netherlands
Sweden
43.3
13.4
31.9
29.3
31.3
46.8
51.7
34.9
39.0
11.0
28.6
26.0
28.8
42.5
48.0
31.1
-19.8
-14.5
- 9.0
-4.7
-27.2
- 15.7
11.2
-15.1
-24.1
-16.9
- 12.3
-8.0
- 29.7
-20.0
8.0
-18.9
validity of the "Aaron conditions" with qualifications. Breyer (1988), for instance,
demonstrated that in the process of transition f r o m a p a y - a s - y o u - g o system to a
capital-funded system, the first generation of pensioners incurred a loss o f benefits.
While these analyses constitute valuable contributions to the political e c o n o m y
of pension systems, they do not provide any real clue as to whether pensioners
would be better off under a p a y - a s - y o u - g o system or under a m i x e d system at
the peak of ageing in the year 2030. The answer to this question can only be
given by a simulation which takes into account the concrete time f r a m e as well as
the actual d e m o g r a p h i c characteristics of the present and future retired and active
populations.
A capital-funded pension scheme is similar to a life-insurance except annuities are paid instead of a lump sum. One saves m o n e y during activity and gets
back interest and principal after retirement. Such a scheme is advantageous if the
p r e m i u m paid to the insurance c o m p a n y in order to keep benefits constant at their
present level is less than the corresponding increase in contribution under the payas-you-go system. Calculations have been carried out for Western countries in the
following way. First, we estimated the average contribution which would need
ADJUSTMENTSOF PUBLICPENSIONSCHEMES
393
Table 13. Percentage of total benefits paid by the pay-as-you-go system
Austria
France
Germany
Italy
2000
2015
2030
2050
93
89
80
85
72
67
60
67
51
52
41
49
47
49
39
44
Table 14. Average Annual percentage increase in productivity necessary for maintaining constant the average net salary and contribution rate in the period 1985-2030.
Low mortality scenario
Austria
Canada
Czechoslovakia
Finland
France
Germany
Hungary
Italy
Netherlands
Norway
Poland
Sweden
Net salary
Contribution rate
1.4
0.5
0.3
0.7
0.5
0.9
0.6
1.4
0.5
0.5
0.3
1.1
1.5
2.2
1.0
1.9
1.4
2.0
1.3
1.6
2.1
1.7
1.0
2.1
to be paid by the actives in 2030 under a pure pay-as-go-system in order for the
pensioners to receive the average pension paid in 1985. Second, the average annuities to be paid in 2030 by the insurance company were simply calculated as the
difference between the average pension paid in 1985 and the average pension which
could be paid in 2030 under a pure pay-as-you-go system if the contribution rate
would be kept constant at its present level. Third, the annual premium necessary to
generate those annuities was estimated from annuities tables assuming real interest
rates o f 1 and 2 per cent. It was also assumed that every active starts contributing in
1990. Results are shown in Table 12 for males and widowed females who usually
have a work-record not too different from the average female work-record. In all
countries with the exception o f the Netherlands, the only country with a purely
universal pension system, the mixed system - pay-as-you-go plus complementary
capital-funded s c h e m e - performs much better than the pure pay-as-you-go system.
Assuming a rate of interest o f 2% would allow to keep the contribution below or
close to the 30% mark in Finland, France, the f o r m e r FRG, and Sweden.
In Canada, it would bring the contribution rate down to a low level - about
10%. On the other hand, it does not really bring a solution to Austria and Italy
394
JEAN-PIERRE GONNOT, CHRISTOPHER PRINZ AND NICO KEILMAN
where it still requires a contribution rate between 35 and 40% depending on the
assumptions. Another negative aspect of this system is that it adds to sex inequalities
as the returns depend on the number of years contributed but not linearly as in the
pay-as-you-go system. This makes the system less attractive for people with low
work-records, especially women.9 Alternatively, it provides a strong incentive to
join or to remain in the labour force. This system is therefore consistent with
policies aimed at increasing activity. However, a mixed system calls for a number
of remarks. First, the system can only bring a solution to the future decline of
the return of the pay-as-you-go system if it is implemented immediately. This also
implies that, in the short and medium term, the sum of contributions paid to both the
pay-as-you-go scheme and the capital-funded scheme will exceed the contribution
needed to balance the pay-as-you-go scheme. Indeed, in most countries, pensioners
would get more prior to 2030 from the existing pay-as-you-go system than from a
mixed system as more detailed calculations for Austria show. Table 13 also shows
that, by the year 2030 and under the low mortality scenario, half of the benefits
would be paid by the private scheme in Austria, France and Italy and 40% in the
former FRG. Therefore, although the capital-funded scheme is initially viewed as
a secondary scheme, the logic of capital accumulation and the deterioration of the
return from the pay-as-you-go scheme would result in a mixed system in which
each scheme plays an equal role.
4.6, Is
PRODUCTIVITY THE MIRACLE SOLUTION?
In the absence of reforms of the pension system and of changes in the working
and retirement behaviours, pensions could be kept constant in the future providing
the contribution rate is adjusted upward. This would result in a decrease in the
real average income of the economically active population. However, it has been
suggested that this would not necessarily be the case if productivity increases. Using
the results presented in section 4.3, one can easily estimate at what rate productivity
should grow in order to maintain average income constant in real terms. 10 Results
of calculations for the low mortality scenario are shown in Table 14. In Canada,
Czechoslovakia, France, the Netherlands, Norway and Poland this would require
that productivity grows at a rate of between 0.3 to 0.5 per cent annually. Obviously
this is a mild requirement by historical standards. The figure for Finland is only
slightly higher (0.7 per cent), while the situation is somewhat different in Germany
and Sweden where growth rates of 0.9 and 1.1 per cent, respectively, would be
necessary. Austria and Italy are far worse off: 1.4 points of growth would have to
be devoted to income maintenance.
However, maintaining work income constant in real terms does not mean that
the contribution rate would not increase dramatically. If one would like to keep the
contribution rate constant, requirements on productivity would be much stronger.
According to figures in the second column of Table 14, which are based on results
presented in section 3.4, productivity should grow at rates between 1 and 2.2
ADJUSTMENTS OF PUBLIC PENSION SCHEMES
395
per cent. Consequently, there is little doubt that a relatively small increase in
productivity will allow to prevent a decline in real work income in many countries.
On the other hand, this would not solve the problem of the deterioration of the
performance of public pension schemes.
5. Conclusions
In most of the countries included in this project, there can be no adequate demographic response to population ageing up to the year 2030. Neither a baby boom nor
an inflow of migrants can rejuvenate national populations (unless fertility and/or
migration reach unrealistically high levels). The reason is that in many Western
countries, the movement of the large cohorts born in the 1950s and early 1960s
into older age results in a very large ageing momentum. Therefore, the overall conclusion of this paper is that demographic variables are of limited help in relieving
the burden of future public pensions expenditures. Hence substantial reductions of
the public pension burden have to be sought in socioeconomic measures, and not
in adjusting demographic conditions.
When public pension systems remain unchanged, expenditures will increase
sharply in the future. This would imply strong reductions in benefits in case contributions by the working age population and other aspects of the pension systems
would be left unchanged. Our simulations indicate that any policy which aims
at balancing the system by just increasing the contribution rate, and no other
measures, will most probably fail in most of the countries under study, because the
contribution levels involved are unrealistically high.
Rising female labour force participation rates to levels comparable with those
in the former GDR or Finland will give a clear improvement in countries where
relatively few women work, or where entitlements depend only weakly on work
history. Although it would not be sufficient to solve the pension problem, an important secondary effect is a strong reduction of the male/female gap in entitlements.
A mean age at retirement of 65 would solve the pension problem in Czechoslovakia and Poland under any scenario, ll and bring great relief to the system in
Austria, France, Hungary, Italy, and the Netherlands.
Another possible policy reform is "burden-sharing", i.e. to link pension benefits
to net income, and not to gross income. Burden sharing would avoid a situation in
which benefits that are indexed to gross wages go up just because the contribution
rate grows, for instance as a consequence of increasing pension expenditures. Our
computations indicate that this might bring substantial relief in Austria and Italy.
An increasing share of supplementary capital-funded pensions (sometimes
labelled as "reserve funded pensions", or "private pensions") will most probably
be part of the solution to the pension burden in the future. In nearly all countries
included in the present project (the Netherlands is the exception), a mixed system,
aiming at total individual pension benefits at their 1985 level, would perform much
better than a pure pay-as-you-go system. However, it would involve substantial
396
JEAN-PIERRE GONNOT, CHRISTOPHER PRINZ AND NICO KEILMAN
transition costs and it would still require a contribution rate as high as 35-40 per
cent in Italy and Austria.
None of the different pension policy measures described above in itself will
probably be able to counteract the implications of population ageing, but a combination of them might be the answer. Since each of the policy measures has different
effects in different countries, the combinations required will also be very different.
The specific choice has to be made in the context of the social and cultural environment of each country. The international comparative study did not include a pension
scenario based on a bundle of policy measures, exactly because different bundles
are needed in different countries. But within the framework of the project, Ott et
al. (1991) investigated a combination of policies when they carried out the country
case study of the unified Germany. They find that a no change policy (unchanged
demographic parameters and pension system) would lead to a balancing contribution rate of 37 per cent in 2030, up from 19 per cent in 1985. Alternatively, the
balancing pension level would be only 56 per cent of the 1985-pension level. But
when three different policy measures are combined the situation changes drastically. A new pension scheme based on (i) burden sharing (linking the pension to
net income instead of to gross income), (ii) raising the average age at retirement
by six years, and (iii) deduction of other income of survivors from their dependents' benefits would bring the balancing contribution rate down to 23-25 per cent
in 2030 (depending on the demographic and labour market scenario), compared
to the 37 per cent level under the no change policy. The balancing pension level
would be between 92 and 95 per cent of today's benefits, instead of 56 per cent. The
authors conclude that such a three-fold adaptation of the German pension system
appears feasible, especially if future economic growth is moderate or strong. These
thoughts on economic growth concur with the suggestion put forward in section
4.6 that a rise in productivity may be one of the many factors which together might
bring relief to the expected pressure on the system of public pensions.
Acknowledgements
This project has been carried out at the International Institute for Applied Systems
Analysis (IIASA) in Austria, in co-operation with the Netherlands Interdisciplinary
Demographic Institute. In various stages of the project, the research team was
supported by national researchers in the 12 participating countries: Tommy Bengtsson, Helge Brnnborg, Thomas Btittner, Tomas Fiala, Peter Findl, Ewa Fratczak,
Heinz Galler, Richard Gisser, Henk Heeren, Jerzy Holzer, Evert van Imhoff, Janina
Jozwiak, Irena Kotowska, Agneta Kruse, Jacques Ledent, Jarl Lindgren, Notburga
Ott, Antonella Pinelli, Jean-Louis Rallu, Alessandra De Rose, Vladimir Roubicek,
Kalman Szabo, and Gabriella Vukovich. Their help is gratefully acknowledged.
Two anonymous referees provided us with valuable comments. This is the revised
version of a paper presented at the EAPS/BiB Conference on "Population-relevant
policies in Europe", Wiesbaden, Germany, 25-29 October 1993.
ADJUSTMENTS OF PUBLIC PENSION SCHEMES
397
Notes
1 The project was planned in 1988, and the majority of the analyses were carried out before the
two Germanies were unified (1990) and hence also before the Czech Republic and Slovakia came
into existence (1993). Thus we have only results for Czechoslovakia as a whole. Furthermore, the
calculations were carried out for the former FRG and GDR separately, and the results were combined.
2 In Table 2, the OADR is defined as the number of persons aged 65 and over per 100 persons aged
15-64. Table 3 gives the OADR based on this definition, and an alternative one with the elderly
defined as persons aged 60 and over.
3 Under all scenarios, mortality is assumed to be independent of marital status. Since death risks for
single and divorced males are markedly higher than those for married males, this could, every other
thing being equal, moderate somewhat the rise in the proportion single. In general, the fact that we
assumed mortality to be independent of marital status has only limited consequences, as illustrated
in Gonnot et al. (1995, pp. 213-215). The reason is that mortality differences by marital status are
large at ages with low death rates, and rather small for the elderly, for whom death rates are high.
4 Taking actual performances of pension systems which are exceptional in many respects - large
labour force versus small number of retirees, high female economic activity versus low work-record
for women retired - as a point of reference for comparison is probably questionable but unavoidable
from a social or policy point of view.
5 Female labour force participation rates in the former GDR might be considered as artificially high
due to hidden unemployment. However, it should be noted that rates observed in Finland in 1989 are
even higher.
6 T h e m e a n age at retirement is some 5 years lower than the legal retirement age in many countries.
This is caused by disability pensions and early retirement schemes.
7 Activity rates for the age-groups 55-59 and 60-64 were assumed to have a corresponding upward
shift.
8 In Norway a n d Central/Eastern European countries, no separate pensions-specific payroll tax rate
is in operation. Instead, pensions are financed from general government revenues.
9 However, our calculations underestimate returns from a savings-type pension scheme for women
as it was assumed that the years worked by women were consecutive.
10 This assumes that all productivity gain goes to public pensions. But one may argue that other
sectors, such as the health system, the care sector, or the labour market, are likely to take their share
as well. Such productivity gains that go to other sectors are not considered here.
11 These countries would face problems after 2050.
References
Aaron, H.J., (1966). 'The social insurance paradox'. Canadian Journal of Economics and Political
Science 33: 371-374.
Breyer, E, (1988). 'On the intergenerational Pareto efficiency of pay-as-you-go financed pension
systems'. Diskussionsbeitrag 139. Freie Universit~it Hagen.
Gonnot, J.-P., Keilman, N. and Prinz, C., (1995). Social Security, Family and Household in Ageing
Societies. Kluwer Academic Publishers, Dordrecht.
Heller, RS., Hemming, R. and Kohnert, P.W., (1986). 'Aging and social expenditure in major industrialised countries, 1985-2025.' IMF Occasional Paper no. 47. (International Monetary Fund,
Washington DC.
Holzmann, R., (1987). 'Ageing and social security costs'. European Journal of Population 3:411-437.
Keilman, N.W., (1985). 'Nuptiality models and the two-sex problem in national population forecasts'.
European Journal of Population 1: 207-235.
Malabouche, G., (1987). 'L'6volution ~tlong terme du syst~me de retraites: une nouvelle m6thode de
projection'. Population 1: 9-38.
Myers, G., (1994). 'Population growth and age structure: Implications and policy responses', in:
Proceedings European Population Conference, March 1993, Volume 1. (United Nations, New
York and Geneva): 207-270.
OECD, (1988a). 'The future of social protection'. Social Policy Studies no. 6 (OECD, Paris).
398
JEAN-PIERREGONNOT, CHRISTOPHERPRINZ AND NICO KEILMAN
OECD, (1988b). 'Ageing populations: The social policy implications'. (OECD, Paris).
Ott, N., Th. Bt~ttner and Galler, H.P., (1991). 'Demographic changes and their implications on some
aspects of social security in the unified Germany: German case study'. Collaborative Paper no.
CP-91-002 (lntemational Institute for Applied Systems Analysis, Laxenburg).
Samuelson, P.A., (1958). 'An exact consumption-loan model of interest with or without the social
contrivance of money'. Journal of Political Economy 66: 467-482.