Skip to main content

Backing out of private pension provision: lessons from Germany


Financing pensions in the EU is a challenge. Many EU countries introduced private pension schemes to compensate declining public pension levels due to reforms made necessary by demographic change. In 2001, Germany introduced the Riester pension. Ten years after introduction the prevalence rate of this voluntary private pension scheme approximates 37 %. However, numerous criticisms raise doubts that the market for Riester products is transparent. Using the 2010 German SAVE survey, this article investigates terminated and dormant Riester contracts on a household level for the first time. Respectively 14.5 and 12.5 % of households who own or have owned a Riester contract terminated or set dormant their contract. We find that around 45 % of terminated or dormant Riester contracts are caused at least partly by product-related reasons, which is significantly higher than for endowment life insurance contracts. The uptake of a new contract after a termination is more likely if a termination is product-related. Nevertheless, after a termination 73 % of households do not sign a new contract, which can have serious long-term consequences for old-age income. Households with low income, low financial wealth or low pension literacy are more likely to have terminated or dormant contracts. Low income and low financial wealth households also have the lowest prevalence rate of Riester contracts and are at higher risk of old-age poverty.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3


  1. See Barr and Diamond (2008) for a detailed evaluation of recent pension reforms. The 2012 Ageing Report of the European Commission and the Economic Policy Committee (2012) contains details on each pension system of the EU-27 member states.

  2. To receive the full state subsidy, the individual has to save 4% of his or her gross income, at least €60 each year. The subsidy for each individual is €154/year. Furthermore, a child born before 1.1.2008 increases the subsidies by €185/year for each child or €300/year if the child is born in 2008 or later. Individuals below 25 years receive a one time bonus of €200 when they sign a Riester contract. Contributions up to €2100/year can be deducted from taxable income taking already received subsidies into account.

  3. Homestead” funds allow using the accumulated capital of the Riester contract to finance the purchase or construction of privately used real estate.

  4. The estimate of Geyer (2011) is based on the German Socio-Economic Panel (SOEP). 37.4 million are directly eligible and 3.9 million indirectly eligible. The estimates are higher compared to the one of Fasshauer and Toutaoui (2009).

  5. Currently the introduction of a uniform information leaflet as suggested by the Ministry of Finance is discussed in the media (e.g. Reiche 2011).

  6. Until 2008 the termination rate of life insurance contracts includes contracts for which contributions are reduced or set to zero. From 2008 on, these contracts are no longer included.

  7. See Blank (2011, pp. 111–112) for a detailed overview of the distribution of Riester contracts over household characteristics.

  8. Fang and Kung (2012) investigate why policyholders lapse their life insurance. The study is based on the Health and Retirement Study (HRS). They observe whether an individual currently has a life insurance, whether the individual obtained the policy since the last wave or whether the individual lapsed a policy. They do not directly observe the reason why a life insurance has been lapsed. Based on a dynamic discrete choice model of life insurance decisions they investigate which factors influence the lapsation of life insurances. When policy holders are young, their results suggest that a large fraction of lapsations is caused by idiosyncratic shocks. Idiosyncratic shocks will become less important and income, health or bequest motive shocks will gain relevance if individuals age.

  9. Since introduction in 2001 until June 2010, the BaFin certified more than 5000 old-age provision products and more than 2000 certifications were changed (BaFin 2011). Since 1.7.2010 certification is provided by the Bundeszentralamt für Steuern (BZSt).

  10. For more information on the SAVE survey, visit the SAVE section of the MEA homepage

  11. For the regression analyses no weights are used (Deaton 1997, p. 70).

  12. Individuals receiving a disability pension, who are eligible for Riester subsidies since 2008, cannot be separated in SAVE from individuals receiving an old-age pension, who are not eligible. The induced bias should be very small since disability pensioners reflect a very small fraction of pensioners.

  13. Households who answered categories 1–3 of question 73 (Ziegelmeyer and Nick 2012, Appendix B) are also included even if households were not eligible in all years from 2005–2010. The measure of eligibility covers 2006–2010 and for some households even 2005, but it is not perfect since it does not consider occupational changes between the survey years or eligibility before 2005/2006.

  14. The termination rate of the BaFin could be corrected further downwards if instead of the stock at the beginning of 2003, the inflows in 2001 and 2002 were known. It would be corrected upwards if the outflows in 2002 were known (2001 can be neglected). However, the change should be minimal since those figures missing are close to the introduction of Riester contracts.

  15. For a comparison of the number of terminated and dormant contracts between pension insurance funds and other types of Riester contracts see Ziegelmeyer and Nick (2012), Sect. 4.1).

  16. If the category “other reason” was specified in the text field, reasons which could be assigned to another category were grouped into the corresponding category.

  17. If an individual does not want to be personally responsible for a dormant or terminated contract, the individual is more likely to select product related reasons. The negative effects of income or financial wealth on dormant and terminated contracts due to personal circumstances should be biased downwards and the negative influence of income and financial wealth on product related dormant or terminated contracts should increase. Thus, if no biases are present, it will be easier to test hypotheses 2a and 2b.

  18. Another reason explaining this difference could be the longer experience of consumers and providers with endowment life insurance contracts. In addition, the media had already discussed the ill-advice and poor quality of the Riester products when the survey was conducted. This could have caused more respondents to report bad-advice or poor quality in connection with Riester products.

  19. Sum of deposits held in savings accounts, building savings contracts, bonds, stocks, stock mutual and real estate funds, and other financial assets. Old-age provision assets are excluded due to partial endogeneity to the dependent variable.

  20. Since we do not know who owns the contract in the household, we take the characteristics of the reference person, who should be the most financially knowledgeable person in the household. Measurement error may bias the results downwards.

  21. Basic education (d) if household head had 9–10 years of education; high education (d) if household head had 13–14 years of education (ref.); undergraduate edu. (d) if household head had 16–17 years of education; graduate education (d) if household head had 18–19 years of education.

  22. The article contains additional information on these measures as well as their validation.

  23. A high rate of terminations could also be seen as a sign of a transparent market. However, terminations or interchanges of contracts usually carry costs. Thus, there are incentives to pick the correct contract. One could argue that the problem is not the transparency of the market but the cost structure of contracts, e.g. the high initial fees of Riester contracts, which are normally spread over the first 5 years.

  24. Contracts are similar to cash value life insurances without risk component. § 1 Abs. 4 Satz 1 und 2 Versicherungsaufsichtsgesetz (VAG): “Als Kapitalisierungsgeschäfte (Anlage Teil A Nr. 23) gelten Geschäfte, bei denen unter Anwendung eines mathematischen Verfahrens die im voraus festgesetzten einmaligen oder wiederkehrenden Prämien und die übernommenen Verpflichtungen nach Dauer und Höhe festgelegt sind.


  • Achen CH (1986) The statistical analysis of quasi-experiments. University of California Press, Berkeley

    Google Scholar 

  • Agarwal S, Driscoll JC, Gabaix X, Laibson D, Dell’Ariccia G, Pence KM (2009) The age of reason: financial decisions over the life cycle and implications for regulation. Brookings papers on economic activity fall 2:51–118

    Google Scholar 

  • Barr N, Diamond P (2008) Pension reforms: principles and policy choices. Oxford University Press, Oxford

    Google Scholar 

  • Benjamin DJ, Brown SA, Shapiro JM (2006) Who is ‘Behavioral’? Cognitive ability and anomalous preferences. Levine’s working paper archive

  • Bilias Y, Georgarakos D, Haliassos M (2010) Portfolio inertia and stock market fluctuations. J Money Credit Bank 42:715–742

    Article  Google Scholar 

  • Blank F (2011) Die Riester-Rente: Überblick zum Stand der Forschung und sozialpolitische Bewertung nach zehn Jahren. Sozialer Fortschritt 6(2011):109–115

    Article  Google Scholar 

  • Boeri T, Börsch-Supan A, Tabellini G (2000) Der Sozialstaat in Europa: Die Reformbereitschaft der Bürger: Eine Umfrage in vier Ländern. Dt. Inst. für Altersvorsorge, Köln

    Google Scholar 

  • Börsch-Supan A, Gasche M (2010) Kann die Riester-Rente die Rentenlücke in der gesetzlichen Rente schließen? MEA discussion paper 201-10, MEA Mannheim

  • Börsch-Supan A, Coppola M, Essig L, Eymann A, Schunk D (2009) The German SAVE study. Design and results. MEA Study No. 6, MEA Mannheim

  • Bucher-Koenen T (2011) Financial literacy, Riester pensions, and other private old age provision in Germany. MEA discussion paper 250-11, MEA Mannheim

  • Bucher-Koenen T, Koenen J (2011) Do Smarter consumers get better advice? An analytical framework and evidence from German Private Pensions. Mimeo Max-Planck-Institute for Social Law and Social Policy Munich(MPISOC-MEA) and University of Bonn

  • Bucher-Koenen T, Ziegelmeyer M (2011) Who lost the most? Financial literacy, cognitive abilities, and the financial crisis. Working paper series 1299, European Central Bank

  • Bundesanstalt für Finanzdienstleistungsaufsicht (eds) (2011) ’10 Jahresbericht der Bundesanstalt für Finanzdienstleistungsaufsicht. Bonn und Frankfurt am Main, Mai 2011

  • Bundesministerium der Justiz (eds) (2006) Verordnung über die Berichterstattung von Versicherungsunternehmen gegenüber der Bundesanstalt für Finanzdienstleistungsaufsicht (Versicherungsberichterstattungs-Verordnung–BerVersV)

  • Bundesministerium für Arbeit und Soziales (eds) (2011) Riester-Rente: 15 Millionen-Grenze fast erreicht. Pressemitteilung,, Last access: 18.11.2011

  • Bundesministerium für Arbeit und Soziales (eds) (2012) Mehr als 15 Millionen Riester-Verträge Pressemitteilung,, Last access: 26.6.2012

  • Bundesministerium für Justiz (eds) § 1 Abs. 4 Satz 1 und 2 Versicherungsaufsichtsgesetz (VAG)

  • Calvet LE, Campbell JY, Sodini P (2009) Fight or Flight? Portfolio rebalancing by individual investors. Quart J Econ 124(1):301–348

    Article  Google Scholar 

  • Coppola M, Gasche M (2011) Die Riester-Förderung: Das unbekannte Wesen. MEA discussion paper 244-11, MEA Mannheim

  • Coppola M, Reil-Held A (2009) Staatlich geförderte private Altersvorsorge: eine Bestandsaufnahme basierend auf SAVE 2003–2008. MEA discussion paper 195-09, MEA Mannheim

  • Deaton A (1997) The analysis of household surveys: a microeconometric approach to development policy, 3rd edn. Published for the World Bank by Johns Hopkins University Press, Baltimore

  • Deutscher Bundestag (eds) (2001) Gesetz zur Reform der gesetzlichen Rentenversicherung und zur Förderung eines kapitalgedeckten Altersvorsorgevermögens (Altersvermögensgesetz–AVmG). Berlin, Bundesgesetzblatt (BGBl) Teil I, Nr. 31, 1310–1343

  • European Commission and the Economic Policy Committee (eds) (2012) The 2012 ageing report: economic and budgetary projections for the EU-27 Member States (2010–2060). European Economy No 2/2012, ISBN 978-92-79-22850-6

  • Fang H, Kung E (2012) Why do life insurance policyholders lapse? The roles of income, health and bequest motive shocks. NBER working paper no. 17899

  • Fasshauer S, Toutaoui N (2009) Die Anzahl des förderberechtigten Personenkreises der Riester-Rente: eine Annäherung. Deutsche Rentenversicherung 6(2009):478–486

    Google Scholar 

  • Feigl M, Jaroszek L, Leinert J, Tiffe A, Westerheide P (2010) Transparenz von privaten Riester und Basisrentenprodukten: Abschlussbericht zum Projekt Nr. 7/09. Zentrum für Europäische Wirtschaftsforschung (ZEW)

  • Gasche M, Ziegelmeyer M (2010) Hat die Finanz- und Wirtschaftskrise Verbreitung und Volumen der Riester-Rente beeinflusst? Wirtschaftsdienst 90(4):255–261

    Article  Google Scholar 

  • German Insurance Association (eds) (2007) Geschäftsentwicklung 2006, Die deutsche Lebensversicherung in Zahlen

  • German Insurance Association (eds) (2008) Geschäftsentwicklung 2007, Die deutsche Lebensversicherung in Zahlen

  • German Insurance Association (eds) (2009) Geschäftsentwicklung 2008, Die deutsche Lebensversicherung in Zahlen

  • German Insurance Association (eds) (2010) Geschäftsentwicklung 2009, Die deutsche Lebensversicherung in Zahlen

  • German Insurance Association (eds) (2011) Die deutsche Lebensversicherung in Zahlen 2010/2011

  • Geyer J (2011) Riester-Rente: Rezept gegen Altersarmut? Wochenbericht des DIW Berlin Nr.47/2011

  • Geyer J, Steiner V (2009) Zahl der Riester-Renten steigt sprunghaft an: aber Geringverdiener halten sich noch zurück. DIW Wochenbericht 32:534–541

    Google Scholar 

  • Hagen K, Kleinlein A (2011) Zehn Jahre Riester-Rente: Kein Grund zum Feiern. Wochenbericht des DIW Berlin Nr.47/2011

  • Hagen K, Reisch LA (2010) Riesterrente: Politik ohne Marktbeobachtung. Wochenbericht des DIW Berlin Nr.8/2010

  • Handelsblatt (eds) (2011) Riester-Wechsler riskieren einen Teil ihrer Rente.,3421918.html, Last access: 4.11.2011

  • Heckman JJ (1976) The common structure of statistical models of truncation, sample selection and limited dependent variables and a simple estimator of such models. Ann Econ Soc Meas 5:475–492

    Google Scholar 

  • Heckman JJ (1979) Sample selection bias as a specification error. Econometrica 47:153–161

    Article  Google Scholar 

  • Hoyer N (2011) Schlecht versorgt mit Riester-Rente. Wirtschaftswoche, 17.10.2011., Last access: 4.11.2011

  • Inderst R, Ottaviani M (2009) Misselling through agents. Am Econ Rev 99:883–908

    Article  Google Scholar 

  • Kahneman D, Knetsch JL, Thaler RH (1991) Anomalies: the endowment effect, Loss Aversion, and Status Quo Bias. J Econ Perspect 5(1):193–206

    Article  Google Scholar 

  • Kimball MS, Shumway T (2010) Investor sophistication and the home bias, diversification, and employer stock puzzles. University of Michigan working paper

  • Lusardi A, Mitchell O (2007a) Baby boomer retirement security: the roles of planning, financial literacy, and housing wealth. J Monet Econ 54:205–224

    Article  Google Scholar 

  • Lusardi A, Mitchell O (2007b) Financial literacy and retirement preparedness: evidence and implications for financial education. Bus Econ 42(1):35–44

    Article  Google Scholar 

  • Lusardi A, Mitchell O (2008) Planning and financial literacy: how do women fare? Am Econ Rev 98(2):413–417

    Article  Google Scholar 

  • Maki DM (2004) Private pension and public policy. Brooking Institution Press, chapter Financial Education and Private Pension, pp 126–145

  • Reiche L (2011) Finanzminister zieht bei Riester die Reißleine. Manager Magazin, 22.06.2011.,2828,769719,00.html, Last access 24.2.2012

  • Rubin DB (1987) Multiple imputation for nonresponse in surveys. Wiley, New York

    Book  Google Scholar 

  • Rubin DB (1996) Multiple imputation after 18 + years. J Am Stat Assoc 91(434):473–489

    Article  Google Scholar 

  • Sartori AE (2003) An estimator for some binary-outcome selection models without exclusion restrictions. Polit Anal 11(2):111–138

    Article  Google Scholar 

  • Schunk D (2008) A Markov chain Monte Carlo algorithm for multiple imputation in large surveys. Adv Stat Anal 92(1):101–114

    Article  Google Scholar 

  • Van Rooij M, Lusardi A, Alessie R (2011) Financial literacy and stock market participation. J Financ Econ 101(2):449–472

    Article  Google Scholar 

  • Ziegelmeyer M (2009) Documentation of the logical imputation using the panel structure of the 2003–2008 German SAVE Survey. MEA Discussion Paper 173-09, MEA Mannheim

  • Ziegelmeyer M (2013) Illuminate the unknown: evaluation of imputation procedures based on the SAVE Survey. Adv Stat Anal 97(1):49–76

    Article  Google Scholar 

  • Ziegelmeyer M, Nick J (2012) Backing out of private pension provision: lessons from Germany. Banque centrale du Luxembourg working paper 74

Download references


We thank Martin Gasche (MEA) and two unknown reviewers for very helpful support. Thomas Lueg (GDV), Heinz-Josef Nüssgens (BMAS), and Stefan Sieren (BaFin) provided many useful explanations and clarifications. Thanks as well to Muriel Bouchet (BCL), Paolo Guarda (BCL), Bettina Lamla (MEA) and Armin Rick (University of Chicago) for their helpful comments. We are grateful to participants at the Annual Meeting of the Austrian Economic Association (Vienna 2012), and the SAVEPHF Conference (Munich 2012) as well as seminar participants at MEA and BCL for many helpful questions and suggestions. We thank the German Science Foundation (DFG) which has generously financed the creation of the SAVE panel dataset. The views expressed in this paper are personal views of the authors and do not necessarily reflect those of the Banque centrale du Luxembourg or the Eurosystem.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Michael Ziegelmeyer.


Appendix 1: Inflows and outflows of pension insurance Riester contracts according to the BaFin

The main difference between figures provided by the GDV and the BaFin might be that the BaFin collects all contracts from companies according to the “Gesetz über die Zertifizierung von Altersvorsorge- und Basisrentenverträgen” (AltZerG). This means that not only Riester contracts are included in the statistic, but also so called “Basis-Renten” contracts. Moreover, only the number of collective insurances is known to the BaFin. Not known is how many individual contracts are included in a collective insurance. Furthermore, pension insurance Riester contracts which are carried out as “KapitalisierungsgeschäftFootnote 24 are not separately recorded by the BaFin. Finally, the BaFin does not capture smaller insurance companies, which are under supervision of the federal states. The numbers provided in Table 9 can be seen only as a rough approximation of the number of Riester contracts.

Table 9 Stock and flows of pension insurance Riester contracts over different categories (BaFin)

The total number of outflows over all years is almost equal for both GDV (Table 1) and BaFin (Table 9). However, the yearly figures deviate strongly. Whereas a more or less steady increase of outflows can be observed for the numbers provided by the GDV, the total number of outflows as reported by the BaFin is already relatively high in the early years. This is partly based on so called remaining cash outs for the years 2003 and 2005, which are defined as transfers in consequence of a change in the type of insurance, the amount insured or the contribution within a technical change of contracts (Bundesministerium der Justiz, BerVersV, Anlage 2, Abschnitt A). A request why remaining cash outs are so large in some years could not be answered by an expert of the BaFin. Outflows due to death or occupational disability are only relevant for 1.9 % of the total outflows over all years. The category “Ablauf” plays a role in only 0.6 % and premature cash outs in 5.2 % of cases. Premature cash outs are defined as the reduction of the amount insured or the contribution as long as they are neither of the category termination nor the result of a technical contract change (Bundesministerium der Justiz, BerVersV, Anlage 2, Abschnitt A). Over all years, the category termination is the largest one with 73.6 % or 1.755 million terminations.

Appendix 2

See Table 10.

Table 10 Missing rates of used Riester questions

Appendix 3

See Table 11.

Table 11 Summary statistics of the 2010 SAVE survey

Appendix 4: Reasons for terminated endowment life insurance contracts

Since 2005, 12.6 % of the owners of endowment life insurance contracts terminated their contract at least once (Table 12). Figure 4 shows the reasons for all terminations since 2005. Terminations due to the need of money for other issues are outstanding with almost 50 %. Households seem less reluctant to liquidate an endowment life insurance compared to a Riester contract when there is need for money. The reasons might be that a household loses all the subsidies if the money is not transferred to a new Riester contract. The loss of subsidies might be worse compared to only receiving the repurchase value of an endowment life insurance. Terminations due to inappropriate products are more important compared to terminations due to poor advice. The same pattern is observed for Riester contracts, only on a higher level (Fig. 2).

Table 12 Termination of endowment life insurance contracts
Fig. 4
figure 4

Reasons for terminated endowment life insurance contracts. Source own calculation based on SAVE 2010, data is weighted and multiply imputed; multiple answers possible

One might be concerned that the group of households having at least once owned a Riester contract and the group of households having at least once owned an endowment life insurance (2005–2010) are systematically different, and thus, it might be inappropriate to compare the termination rates between the two groups. Of the restricted sample of 1,432 households 27.2 % had at least once owned a Riester contract and an endowment life insurance (2005–2010). For households having at least once owned both, the termination rate for endowment life insurances (Riester contracts) is 26 % (41 %) due to product-related reasons. The fraction is reduced to 23 % (35 %) if product-related reasons are the only reason. The difference between Riester contracts and endowment life insurances stays large. One reason for the lower termination rate due to product-related reasons might be that both providers and consumers have more experience with endowment life insurances.

Appendix 5: Pension literacy in the 2009 SAVE survey

Whereas the first question (question 69) asks for an institutional detail on the financing of the statutory pension system, the second question (70) asks how the contributions are used. Both questions should measure pension literacy in a compact way. Table 13 summarizes the answers given to question 69. Overall, 38 % did not provide any answer to the question about the contribution rate to the statutory pension system. In 2009, the total contribution rate of employees and employers was 19.9 % of the income subject to social-security contributions. Around 27 % provided answers between 19 and 20 %, which is relatively close to the true value and can be considered as correct even if they did not provide the exact value. Extending the range of correct values to +/− 1 percentage points or +/− 2 percentage points does not change the results in the multivariate analysis. The other households estimate a contribution rate which is sometimes relatively far away from the true value.

Table 13 Pension literacy: contribution rate to statutory pension system

Table 14 provides an overview of the answers given to question 70. Around 48 % provide the correct answer that current contributions to the statutory pension system are exclusively used for today’s retirees as usually the case in a pay-as-you-go pension system. The fraction of correct answers almost doubles compared to question 69.

Table 14 Pension literacy: for what are current contributions to the statutory pension system used?

The pension literacy index is equal to zero if both questions are not correctly answered. It is equal to one if one of the two questions is correctly answered and equal to two if both questions are correctly answered. Table 15 shows the pension literacy index used in the multivariate analysis. Around 41 % of all household heads were unable to provide any correct answer. 44 % answer one question correctly and 15 % answer two questions correctly. As a robustness test a dummy variable is created which takes the value one if at least one question is correctly answered. The effects of the multivariate analysis remain unchanged.

Table 15 Pension literacy index

Appendix 6

See Table 16.

Table 16 Probit: determinants of terminated or dormant Riester contracts

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Ziegelmeyer, M., Nick, J. Backing out of private pension provision: lessons from Germany. Empirica 40, 505–539 (2013).

Download citation

  • Published:

  • Issue Date:

  • DOI:


  • Private pension
  • Riester
  • Termination
  • Financial literacy
  • SAVE

JEL Classification

  • D12
  • D91
  • D14
  • J26