A review of international experience

Riester Pensions in Germany: 
Design, Dynamics, Targeting Success, 
and Crowding-In
Axel Börsch-Supan, Michela Coppola, and Anette Reil-Held 
Riester pensions are heavily subsidized voluntary private pension schemes designed to fill the 
emerging pension gap in Germany. After a slow start and several design changes, the plans 
took off very quickly. Saving incentives have been effective in reaching households with 
children; they have been somewhat less successful in attracting low-income earners. There is 
no evidence that the pensions have crowded out other saving: aggregate national saving has 
increased since their introduction.
D
emographic change poses major problems for public pay-as-you-go pension schemes 
around the world, forcing many to reduce their generosity. In Germany, reductions in 
benefits took place in several steps, starting in 1992 with the indexation of benefits to net 
rather than gross earnings and the introduction of semiactuarial adjustments in case of 
early retirement and extending to the introduction of the “sustainability factor” in 2004, 
which indexes benefits to the inverse of the system dependency ratio (that is, the ratio of 
contributors to beneficiaries). It is estimated that in 2040, the average benefits-to-earnings 
ratio in Germany will be about 18 percentage points lower than in 2010.
These reforms will result in a large and potentially increasing gap in old-age income 
relative to current benefit levels. Recognizing this, the governments of Germany and 
many other countries have accompanied public pension reform with efforts to strengthen 
funded second-pillar (occupational) and third-pillar (private savings) pensions. In order 
to accelerate the uptake of such private pensions, countries have adopted two competing 
strategies: making such supplementary pensions mandatory (as the Netherlands and Swe-
den have done) or heavily subsidizing savings for private pensions, typically by matching 
contributions, providing tax credits, and/or permitting contributions to supplementary 
pensions to be excluded when calculating income for tax purposes. Germany has taken the 
incentive-oriented approach, with its Riester pensions (named after Walter Riester, former 
secretary of labor and social security), which were legislated as part of the 2001 pension 
reform.
1
This research was prepared for a World Bank workshop on matching defined contribution schemes 
held in Washington, D.C., in June 2010. The authors are grateful for comments by Frank Eich, 
Richard Hinz, Robert Holzmann, and David Tuesta Cardenas and for financial support by the 
Research Institute for Policies on Pension and Aging and the German Science Foundation, which 
generously financed the creation of the SAVE panel data set and research based on these data.

82 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Riester pensions are effectively state-subsidized voluntary individual saving arrange-
ments with an associated (largely annuitized) pay-out plan. They are designed to enable all 
individuals receiving a public pension or a similar state-provided benefit to fill the emerg-
ing pension gap through individual savings to an extent that at least offsets the reduction 
in public pensions. As low-income households and households with children are assumed 
to have the greatest difficulty in offsetting these reductions, large tax credits are targeted 
to these households. A more modest tax deduction for contribution applies to everyone.
This chapter focuses on public acceptance of the Riester pensions and the dynamics 
of take-up over time. It highlights the distributional and cost aspects: who joined the new 
pension scheme, how costly is the scheme, who pays for it, and who is likely to be able 
to fill the emerging pension income gap. The chapter also investigates whether the new 
private pension plans displace saving for other purposes.
Riester pensions and their matching subsidies were introduced in a step-wise fashion 
from 2002 through 2008, allowing the effects of subsidies on uptake rates to be identi-
fied. Data come from SAVE, a new panel data set on households’ saving and asset choices, 
sociodemographic characteristics, and the psychological determinants of saving and old-
age provision behavior. The SAVE data capture the period 2001–10. They offer a unique 
opportunity to investigate old-age provision based on very recent data with a broad set of 
explanatory variables.
The issues investigated contribute to the discussion of the impact of retirement sav-
ing incentives in various countries. In the United States, Venti and Wise (1990) on one 
side and Gale and Scholz (1994) on the other sparked a controversial discussion about 
the efficacy of individual retirement accounts (IRAs) as saving devices (see Skinner and 
Hubbard 1996 for an early review). Disney, Emmerson, and Wakefield (2001) provide 
a helpful review tailored to the situation in the United Kingdom. These controversies 
have accompanied pension reform in almost all developed countries, generating interest 
in cross-national analyses of retirement saving behavior under different tax and subsidy 
regimes (see Börsch-Supan 2003, 2004). This chapter adds the recent German experience 
to this discussion and updates the paper by Börsch-Supan, Reil-Held, and Schunk (2008).
The chapter is structured around the themes of design, cost, and the effectiveness of 
the new system. The first section describes the key initial design features of Riester pensions 
in Germany. The second section documents the uptake of Riester pensions between 2001 
and 2010 and the design adaptations that contributed to the uptake dynamics. The third 
section summarizes the costs of the Riester scheme and the changes in retirement saving 
associated with them. The following three sections address the effectiveness of the scheme. 
Section 4 provides descriptive statistics on the dynamics of enrollment in Riester pension 
plans by socioeconomic characteristics. Section 5 reports circumstantial evidence on the 
macroeconomic impact of Riester pensions, in particular whether they have crowded out 
other saving. Section 6 shows how the pension income gap is likely to be filled, the ulti-
mate aim of the scheme. The last section provides some concluding observations.
Initi  al Design of Riester Pensions
Riester pensions are designed to fill the gap that will be created by the scheduled reduc-
tion of public pension benefits in response to the pressures of population aging. In order 

4.  RIESTER PENSIONS IN GERMANY: DESIGN, DYNAMICS, TARGETING SUCCESS, AND CROWDING-IN 
83
to achieve this goal, the 2001 German Retirement Savings Act (Altersvermögensgesetz) 
introduced a comprehensive regime of saving incentives for certified pension products 
that have become known as Riester pensions. These new retirement saving arrangements 
are eligible for two kinds of subsidies: a match of the participant’s contribution and the 
deduction of all contributions from income for tax purposes. The initial certification reg-
ulations and subsidy mechanisms were rather complex. They were significantly simplified 
in 2005.
CRITERIA FOR “CERTIFIABLE” PRODUCTS
Initially, private pension plans were eligible for subsidies if they fulfilled 11 criteria stipu-
lated by the Certification of Retirement Pension Contracts Act. These criteria included, 
among others, the following requirements:
 
• Savers must make regular contributions.
 
• Providers must guarantee that the nominal investment return in each calendar 
year be positive.
 
• Benefits must be disbursed as certain types of lifelong annuities.
 
• Lump-sum payments at retirement must not exceed 20 percent of the accumu-
lated wealth.
 
• Administrative and marketing costs must be spread over at least 10 years.
 
• All providers must be registered with a supervisory board. 
Most Riester products were provided by the insurance industry, but banks and 
investment funds also offered Riester products.
ELIGIBILITY
Not all households are eligible for the subsidies. The eligibility criteria are complex, 
although the intention was that everyone affected by the reduction in public pension 
benefits should be eligible for private pension subsidies. The group of eligible beneficia-
ries includes employees paying mandatory social insurance contributions, recipients of 
wage compensation benefits (such as unemployment benefits, child-raising benefits, and 
so forth), self-employed people who are mandatory members of the public pension sys-
tem, farmers, and tenured civil servants. Spouses of eligible individuals are also entitled to 
receive subsidies (“indirect entitlements”), provided they enroll in a separate pension plan 
of their own. 
SUBSIDIES
As a result of government budget constraints, the generosity of the subsidy scheme was 
phased in starting in 2002 and ending in 2008, with a doubling of the applicable param-
eters in each step (table 4.1). Column 1 shows the percentage of gross earnings that must 
be contributed to the plan in order to qualify for the full subsidy. The full subsidy (col-
umns 2–4) has three components. The first is a basic subsidy (column 2) that is a match 
of the participant’s own contribution. This basic subsidy is means tested for low- and 
middle-income households. Individuals eligible for this basic subsidy pay their savings 

84 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
into a certified pension plan; in the initial design, they had to file an application form to 
receive the subsidy each year. The plan provider receives the matching basic subsidy from 
the government and credits it to the participant’s account as part of the total contribu-
tion. The second component is an additional subsidy for each child (column 3). The third 
component is the tax deductibility of contributions to Riester pension plans, which are 
treated as special expenses up to a maximum amount (column 4). This amount was fixed 
in nominal terms at €2,100 from 2008 onward; unless raised over time, this component 
will eventually be eroded.
2
The matching contribution subsidy is reduced proportionally if contributions to the 
Riester plan are lower than the maximum indicated in column 1 of table 4.1. As the sub-
sidy itself is counted as part of that contribution, some mathematical skills were required 
to compute the exact contribution required to receive the full subsidy—another feature 
that reduced the transparency of the design. The minimum own contribution depended 
on the number of children and the year in which the individual had enrolled in the Riester 
plan.
The complexity of the design made the substantial size of the subsidies less appar-
ent for many people. Low-income participants received a relatively large subsidy, thanks 
to the matching basic subsidy; higher-income participants accrued additional subsidies 
from the tax deductions, thanks to the progressive nature of income tax rates. As a result, 
the profile of subsidies was slightly U-shaped (figure 4.1). Overall, subsidies averaged 
about 45 percent of contributions, ranging from 24 percent to 90 percent, depending on 
income and number of children. At the mean earnings (€42,000), the statutory subsidy 
rate was 39 percent.
OWNER-OCCUPIED HOUSING
Pension savings typically must accumulate until retirement, after which they are disbursed 
in some type of annuity. The 2001 law permitted withdrawals of €10,000–€50,000 from 
the accumulated capital, which had to be paid back into the pension plan in monthly 
installments by the age of 65 (if not, the subsidies received had to be paid back). The 
law was changed in 2008 to permit withdrawals of up to 100 percent of the accumu-
lated capital to purchase owner-occupied housing without having to pay back the amount 
TABLE 4.1  Statutory incentives for supplementary pension provision, 2002–08
Year
Maximum contribution 
(% of gross earnings)
(1)
Basic subsidy 
(€/year)
(2)
Child subsidy
(€/year)
(3)
Maximum tax deduction
(€/year)
(4)
2002
1
38
46
525
2004/05
2
76
92
1,050
2006
3
114
138
1,575
2008
4
154
185
a
2,100
SOURCE: Authors’ compilation.
A. The child subsidy is €300 for children born after 2007.

4.  RIESTER PENSIONS IN GERMANY: DESIGN, DYNAMICS, TARGETING SUCCESS, AND CROWDING-IN 
85
withdrawn. In addition, loan agreements (Darlehensverträge) and building society saving 
contracts (Bausparverträge) can be certified as Riester products receiving the subsidies 
described above. To date, such housing-related products do not constitute a significant 
share of Riester products.
Uptake Dynamics and Amended Design
About 1.4 million Riester pension plans were taken up in the first year after the introduc-
tion of incentives (figure 4.2).
3
 After a period of initial enthusiasm, however, demand for 
Riester pensions flattened in 2003 and 2004. This lackluster growth and widespread criti-
cism of the complex eligibility and subsidy design led to a simplification of the design in 
2005 in an effort to improve acceptance of the new system by households and providers.
The changes introduced in 2005 did not affect the subsidies described in table 4.1 
and figure 4.1. They included the following:
1. The application procedure was simplified by replacing it with a one-time perma-
nent subsidy application. Savers eligible for subsidies were allowed to authorize 
their pension provider to submit a subsidy application on their behalf every year.
2. The number of certification criteria was reduced from 11 to 5.
3. The share of Riester pensions to be annuitized was reduced from 80 percent to 
70 percent (that is, 30 percent of the accumulated capital could be received as a 
lump sum).
FIGURE 4.1  Subsidy as percentage of total (own plus government matching) contribution
0
10
20
30
40
50
60
70
80
90
100
10
15
20
25
30
35
40
45
50
55
60
65
70
75
annual earnings (€, thousands)
single, no children
single, one child
married, one income, two children
%
SOURCE: Deutsche Bundesbank 2002.
NOTE: Share of government assistance (direct subsidy/tax advantage) in total savings for supplementary old-age provision.

86 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
4. The participant’s minimum own contribution was standardized to €60 a year. 
5. The transparency of pension saving products and providers was improved. More 
information on investment options, the structure of the portfolio, and the risk 
parameters was required. Providers were required to introduce a standardized cal-
culation facilitating comparisons of alternative products.
6. The period over which acquisition and marketing costs could be spread was 
reduced from 10 to 5 years, making selling Riester pensions more attractive for 
providers. Although there were no formal limits for such charges, the certification 
process imposed some soft restrictions on charges that could be interpreted as 
usurious.
7. These changes were accompanied by a modest government-sponsored public 
relations campaign. Main advertising efforts, however, were left to providers of 
Riester products.
Demand for Riester pensions rose significantly in 2005 after these changes came 
into effect (figure 4.2). About 900,000 new policies were signed in the last quarter of 
2005 alone—about four times as many as during all of 2004. This upward trend contin-
ued throughout 2008. By the end of 2008, more than 12 million pension plans eligible 
for subsidy support had been taken up.
After 2008, the rate of increase flattened. This trend may have been the result of 
three factors: the maximum contribution rate was reached in 2008; some sociodemo-
graphic strata may have reached saturation levels; and the financial crisis, which resulted 
in very low or negative returns, may have discouraged new households from starting a 
FIGURE 4.2  Development of Riester pensions (million contracts)
millions
0 
2 
4 
6 
8 
10 
12 
14 
16 
2001 
Q2 2002 
Q4 2002 
Q2 2003 
Q4 2003 
Q2 2004 
Q4 2004 
Q2 2005 
Q4 2005 
Q2 2006 
Q4 2006 
Q2 2007 
Q4 2007 
Q2 2008 
Q4 2008 
Q2 2009 
Q4 2009 
Q2 2010 
Q4 2010 
Q2 201
1 
Q4 201
1 
bank savings contracts 
home-annuity contracts (Wohn-Riester) 
investment fund contracts 
insurance contracts 
SOURCE: Authors’ calculations based on data from quarterly press releases of the Federal Ministry for Labour and Social 
Affairs (www.bmas.de).
NOTE: Q = quarter.

4.  RIESTER PENSIONS IN GERMANY: DESIGN, DYNAMICS, TARGETING SUCCESS, AND CROWDING-IN 
87
Riester pension. Although the increase in the number of contracts slowed after 2008, the 
number of contracts continued to increase, at a linear rate. By the end of 2009, about 
40 percent of households potentially eligible for subsidies held at least one Riester plan 
(Coppola and Gasche 2011).
The dynamic development of the Riester pensions occurred in parallel with the 
uptake of occupational pensions. In contrast, the uptake of other unsubsidized private 
retirement saving arrangements was essentially flat (figure 4.3).
FIGURE 4.3  Coverage by private and occupational pensions, 2003–10
7% 
19% 
26% 
31% 
33% 
38% 
13% 
19% 
23% 
21% 
23% 
25% 
27% 
11% 
15% 
23% 
16% 
15% 
15% 
15% 
0
5
10
15
20
25
30
35
40
45
2002
 2004 
 2005 
 2006
 2007
 2008
 2009
Riester pension 
occupational pension 
private pension 
% coverage
13% 
SOURCE: Authors’ calculations based on data from the SAVE panel.
NOTE: Brackets show 95 percent confi dence intervals. Figures are weighted.
In 2006, Riester pensions overtook occupational pensions as the main instrument 
for funded pension provision in Germany. Largely as a result of this development, the 
share of households with no private supplemental pensions decreased from 73 percent in 
2001 to about 45 percent in 2009. By 2010, about a quarter of households had at least 
two saving instruments for their old-age provision (figure 4.4).
Fiscal Costs and Savings
Riester pensions create direct fiscal costs, in the form of the means-tested basic and child 
subsidies and the foregone tax revenue that results from the exclusion of the allowable con-
tribution levels. In 2010, these costs reached €3.5 billion, about 80 percent of which are 
attributable to the means-tested direct subsidies (figure 4.5). In comparison, costs for pub-
lic pension benefits (old-age, disability, and survivor pensions) were €225 billion in 2010.

88 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
FIGURE 4.4  Percentage of households with private and occupational pension instruments, 
2003–10
73% 
63% 
53% 
54% 
49% 
49% 
45% 
14% 
20% 
19% 
23% 
28% 
27% 
30% 
10% 
13% 
19% 
19% 
18% 
18% 
20% 
2% 
4% 
9% 
4% 
5% 
5% 
5% 
0
10
20
30
40
50
60
70
80
2002 2004
2005
2006
2007
2008
2009
no old-age provision 
1 instrument 
2 instruments 
3 instruments 
% of households
SOURCE: Authors’ calculations based on data from the SAVE panel.
NOTE: Brackets show 95 percent confi dence intervals. Figures are weighted.
FIGURE 4.5  Fiscal costs of Riester pensions, 2003–10
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
2003 2004 2005 2006 2007 2008 2009 2010 
direct benefit 
total costs 
tax deduction 
€, thousands
SOURCE: Deutsche Rentenversicherung 2011.

4.  RIESTER PENSIONS IN GERMANY: DESIGN, DYNAMICS, TARGETING SUCCESS, AND CROWDING-IN 
89
The fiscal costs shown in figure 4.5 are the actual payment of subsidies or foregone 
income tax revenue resulting from the new system. As Riester subsidies can be claimed 
up to two years after paying into the account and tax losses typically occur the year after 
income generation, it is not easy to match induced Riester savings with the subsidies paid 
out the same year. In addition to a potential mismatch between years are the deeper ques-
tions of causality and substitution, which are addressed later in this chapter.
Stolz and Rieckhoff (2008, 2009, 2010, 2011) attempt to matching savings and 
subsidies (table 4.2). Their calculations, however, include only the direct subsidies, not 
the tax deductions. Total savings in the accounts are composed of the matching subsidy 
and contributions by individuals on which the subsidy is calculated. Assuming that these 
overall savings would not have been made without the incentive of the subsidies, €1 of 
subsidies (not including the value of the income tax exclusion) induced an additional €2.2 
savings for old age. Under these strong assumptions, this figure could be interpreted as a 
measure of the cost-effectiveness of the subsidy.
TABLE 4.2  Direct subsidies and associated savings in Riester plans, 2005–08
Year
Direct subsidies
(€, thousands)
Total savings
(€, thousands)
Own contribution
(€, thousands)
Effectiveness
(%)
2005
521,917 1,762,749 1,240,832 
2.4
2006
1,134,339 3,635,886 2,501,547 
2.2
2007
1,445,688 4,834,565 3,388,877 
2.3
2008
2,543,300 7,815,500 5,272,200 
2.1
SOURCES: Stolz and Rieckhoff 2008, 2009, 2010, 2011 based on data from the Central Subsidies Agency.
Gerber and Zwick (2010) link data from the Central Subsidies Agency (ZfA) with 
income tax data in order to include the effects of tax deductions. They compute a broader 
measure of subsidy rates that can be converted into a similar measure of cost-effectiveness. 
Because the linkage solves some but creates other time matching problems, the results are 
not fully comparable with the results shown in table 4.2. The cost-effectiveness measure 
of €1.9 of new savings for each €1 of subsidies across all households is similar to the €2.2 
figure Stolz and Rieckhoff estimate. The estimate is substantially higher for households 
with children (€2.4) than for childless households (€1.1).
Targeting Success
Another dimension of program effectiveness is the ability of the subsidy to reach the tar-
geted population. Riester pensions were specifically designed to help low-income earners 
and families with children—two segments of the German population in which saving rates 
are typically low—build up savings for retirement. The Central Subsidies Agency provides 
quarterly uptake figures that are associated with a limited number of socioeconomic char-
acteristics. Analyses by Stolz and Rieckhoff (2008, 2011) show that individuals with low 

90 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
labor income, women, families, and employees in the former German Democratic Repub-
lic are well represented among subsidy recipients.
4
 These data are based on subsidized 
individuals, not households, however; they do not reveal total household income, only 
those income components that are subject to means testing.
The analysis of targeting reported here used the SAVE data set (Börsch-Supan and 
others 2010 provide a detailed description of the data set; Essig 2005 and Schunk 2006 
provide additional details). The panel was surveyed in 2001, 2003, and then annually 
from 2005 through 2010. SAVE provides data at the level of the individual respondent 
and the respondent’s spouse (hereafter referred to as the household). In 2006, the total 
sample size was about 3,500 households. In this chapter, the analysis is restricted to house-
holds that have not yet retired.
5
 All descriptive statistics are weighted based on the income 
and age distribution of the German microcensus.
6
One would expect to find age-related differences, because the replacement rate of the 
public pension system will slowly decline for future cohorts as a result of the recent reforms. 
Younger generations are thus more strongly affected than older generations. In addition, 
younger generations have a longer period over which to accrue the subsidies of the new sys-
tem, and they are more likely to have dependent children and thus be eligible for the higher 
level of subsidies. One would therefore expect higher participation rates among younger 
households than among older ones. This is indeed what figure 4.6 indicates.
UPTAKE AMONG HOUSEHOLDS WITH CHILDREN
Figure 4.7 shows the strong positive relationship between the number of children and 
the proportion of households with a Riester pension plan. As Riester subsidies increase 
linearly with the number of children, it is not surprising that there is strong demand for 
Riester products among parents with more than two children.
About 60 percent of households with two or more children—more than twice as 
many as among childless households—had a Riester pension plan in 2010. Uptake was 
strongest among households with more than three children. Figure 4.8 provides signifi-
cant evidence of the effectiveness of the system in enrolling families with children. This 
success can be attributed to the large child subsidy, as there was no comparable increase in 
other private pension instruments.
UPTAKE AMONG LOW-INCOME HOUSEHOLDS 
Reaching low-income households has proven to be more difficult than eliciting partici-
pation by families with children. The proportion of households holding private pension 
instruments increases with disposable household income (figure 4.9). This pattern is most 
apparent for occupational pensions. Only about 19 percent of households in the lowest 
income bracket have taken up Riester pension plans. In contrast, almost half of house-
holds in the top two income brackets report participating in the system.
Given the substantial challenges in reaching low-income groups with any type of 
retirement savings, it is a measure of some success of the system that low-income house-
holds hold more Riester pensions than any other kind of private pension. As a result, 
Riester pensions are more equally distributed by income than occupational pensions or 
unsubsidized private pension schemes.

4.  RIESTER PENSIONS IN GERMANY: DESIGN, DYNAMICS, TARGETING SUCCESS, AND CROWDING-IN 
91
FIGURE 4.6  Uptake of Riester pensions by age group