A review of international experience

lated fiscal expenditure. The idea behind this comparison is that on a net basis 
(aggregate new savings less aggregate cost of the subsidy), the match should 
increase the pension wealth of the elderly and the capital stock on which future 
benefits are paid by more than the overall cost of the subsidy. Such a calculation 
would take account of compensating or strengthening effects. Ex ante projec-
tions with an appropriate overlapping generations model would offer first indica-
tions of effects; for the more relevant ex post evaluation, the data and estimation 
requirements need to be developed.
Comparison to Ex Post Subsidy
Are matching contributions (ex ante subsidies) for voluntary or mandated schemes less 
expensive than noncontributory or subsidized benefit levels (ex post subsidies)? Measur-
ing fiscal efficiency requires considering the likely cost and comparing it with alternatives. 
Doing so yields the following considerations:
 
• A demogrant provides everyone with a minimum transfer in old age, regardless 
of individual circumstances. This approach is very effective in distributive terms, 
but not fiscally efficient because the leakages are high as many receiving the ben-
efit do not need it to maintain the level of consumption in old age that such a 
benefit is intended to achieve. The same minimum income support for needy 
elderly can be generated at much lower cost when targeted transfers are provided 
in the form of general social assistance or categorical social pensions (Grosh and 
Leite 2009). Of course, targeting, however well done, will lead to inclusion and 
exclusion errors.
 
• Compared to both demogrants and targeted ex post benefits, an MDC scheme 
can be constructed that is fiscally more efficient as long as individuals do some 
additional saving and targeting works effectively. If individuals are not induced 
by the match to increase their saving, for whatever reason, then matching leads to 
distributively inferior results. This has implications for the lowest-income groups 
where saving capacity for old age is limited. If individuals do some saving but ex 
ante targeting does not work well, it can lead to either fiscally less efficient out-
comes than ex post targeting or distributionally less efficient outcomes or both. 

22 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
This will depend on the size of the inclusion and exclusion errors and the reaction 
of individuals to the subsidy provided. Unfortunately, the relevant experience and 
data are not available to undertake such a comparison.
None of the countries reviewed appears to have established a comprehensive set 
of outcome measures or undertaken estimates of the fiscal effectiveness of the match-
ing design in relation to alternatives. No studies have been undertaken to provide com-
parative measures even at the conceptual level of the cost-effectiveness of key parameters. 
Such a study should be relatively simple at the individual level; however, at the macro-
level, the comparison would have to make a number of heroic assumptions to allow for 
comparability.
Preliminary 
Policy
 Conclusions and Next Steps
Because few MDC schemes have been subjected to a rigorous impact evaluation, policy 
lessons are by necessity very preliminary. Based on the evidence presented in this volume, 
a few conclusions and suggestions for future analysis can nevertheless be drawn:
1. Empirical evidence, collected largely from high-income countries, suggests that 
MDC schemes raise participation in pension systems moderately and have at best 
modest (and in some instances ambiguous) effects on contribution levels of par-
ticipating individuals. Both findings are consistent with theoretical predictions. 
The persistence of any of these effects—and ultimately their influence on life-
time wealth accumulation and the provision of retirement income—will require 
much longer-term assessment. The net impact of matching on individual wealth 
and macroeconomic savings levels is empirically even more difficult to assess 
and subject to conflicting views. The weight of the limited evidence to date for 
high-income and perhaps some high-middle-income countries is that MDCs will 
make a helpful but insufficient contribution to solving the challenges of pension 
coverage, income maintenance, and informality of labor. They may, however, be 
effective for special groups that are difficult to reach by other means.
2. It is unclear to what extent the results for high-income countries will translate 
into the context of lower- and middle-income countries. The groups for which 
there is experience in high-income countries may not be representative of the 
broader population, nor are they likely to share characteristics and behavior pat-
terns with target populations in low- and middle-income settings, which are 
likely to be characterized by many more constraints in their everyday life and 
much larger shocks for which social risk management instruments are not avail-
able. As a result, the incentives by MDC programs to enhance retirement savings 
may translate less well or at least differently. Since few programs of this kind 
existed until recently in developing economies, and none has been subject to 
rigorous impact evaluation, not much is yet known about how the experience in 
other settings will translate; caution is therefore advised.
3. The threshold and other parameters of the match seem to have as much, or even 
a greater, effect as the overall level of the match on behavior and saving outcomes 

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
23
in high-income countries. Similar empirical evidence is not yet available for low- 
and middle-income countries. In view of the many more binding constraints 
in developing countries, complex matching rates and other parameters may be 
required to achieve the envisaged outcomes. A key design challenge in these set-
tings will be to balance this with a need for simplicity and transparency. At least 
equally important for participation and saving effects are interventions directed 
toward overcoming behavioral and other limitations, in particular through the 
provision of information and financial education, a choice architecture consis-
tent with the observed inertia in behavior, and social marketing, public service 
announcements, and other types of advocacy. Key considerations in designing 
such programs include the following:
 
• Information about the subsidized scheme and its operation is critical in creat-
ing awareness, which seems to be an important factor in changing behavior (in 
general and for retirement income saving in particular).
 
• Many (but not all) experts believe that the ability to understand the offered 
saving products and to apply that knowledge is a critical factor for participa-
tion in social risk management programs (whether mandated or voluntary, 
unsubsidized or matched).
 
• The setting of defaults and other forms of choice architecture are increas-
ingly recognized as key determinants of outcomes. Policy makers need to take 
advantage of lessons from behavioral economics about harnessing the power 
of inertia (in particular through automatic enrollment with limited opt-outs), 
simplifying administrative processes, and setting parameters.
 
• Advocacy and education efforts—such as seminars, public service advertising, 
social marketing, publicly sponsored retirement savings information websites, 
and entertainment education—may be very important in promoting MDC 
schemes.
4. Individuals at all income levels evade participation in mandated schemes for a 
variety of reasons, including ineffective design choices, poor alignment of incen-
tives with the environmental context, and lack of trust in public and financial 
institutions. Addressing these issues may go a long way toward increasing par-
ticipation. These efforts should be undertaken before matching schemes are 
introduced.
5. The effect that pension-related matching incentives may have in changing 
employment patterns or reducing the level of informality in developing and tran-
sition economies remains wholly unknown. Thus far, there is little evidence that 
these schemes are sufficient to change individual decisions. This remains a dif-
ficult issue to effectively evaluate, but one that merits investigation despite the 
methodological challenges.
6. Examples and a framework are lacking for a comprehensive cost-benefit analysis 
for MDC designs to compare them with potential alternatives. Rigorous impact 
evaluation focused on microeconomic assessment is necessary to guide program 

24 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
design. This needs to be complemented by a more comprehensive welfare analy-
sis, including cost-benefit evaluation of alternative approaches. Such analysis 
should address areas not covered in the research to date including consideration 
of the following:
 
• A pure saving instrument (and pooling risks over time by individuals) or a 
traditional risk management instrument (pooling risks across individuals) in 
relation to matching designs
 
• Differentiation of MDC outcomes in relation to various income groups; 
theoretical considerations and empirical indications suggest that matching to 
improve retirement saving is not likely to be effective for the poorest in society, 
and initial evidence suggests large substitution effects among higher-income 
groups
 
• Gender analysis of MDC, addressing differences in labor force participation, 
longevity, and other factors likely to create different outcomes
 
• The longer-term role of matching designs and possible development of exit 
strategies for their use as a transitional device to generate participation as econ-
omies meet thresholds where coverage expansions become feasible.
7. Finally, it will be important to consider the possible future role and new designs 
of MDC schemes under different social policy scenarios, such as the following:
 
• The marginalization or even the demise of pay-as-you-go earnings-based (Bis-
marckian) public pensions in countries with a high level of informality
 
• A further reduction of the mandatory defined contribution pension systems in 
countries with low informality but also facing demographically driven reduc-
tions in the generosity of public earnings-related and basic schemes
 
• The need for more portability of social benefits and the portability of the sub-
sidy element of MDCs across borders.
Notes
1.  All modern monitoring and evaluation books apply a similar logic, although they may use 
different terminology. For recent highly readable publications on monitoring and evalua-
tion, see Gertler and others (2010); Khandker, Koolwal, and Samad (2009); and Leeuw and 
Vaessen (2009). Also see the websites of the Massachusetts Institute of Technology’s Poverty 
Action Lab (http://www.povertyactionlab.org/), the World Bank (http://www.worldbank.org/
oed/ecd), and the World Bank’s Strategic Impact Evaluation Fund (http://go.worldbank.org/
X81HJAZSG0).
2. See 
http://www.direct.gov.uk/en/Pensionsandretirementplanning/Companyandpersonalpen-
sions/WorkplacePensions/DG_200722 for on overview of the government’s plans and time 
schedule.

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
25
References
Aterido, Reyes, Mary Hallward-Driemeier, and Carmen Pagés. 2011. “Does Expanding Health 
Insurance beyond Formal-Sector Workers Encourage Informality? Measuring the Impact of 
Mexico’s Seguro Popular.” IZA Discussion Paper 5996, Institute for the Study of Labor, Bonn.
Gertler, Paul J., Sebastian Martinez, Patrick Premand, Laura B. Rawlings, and Christel M. J. Ver-
meersch. 2010. Impact Evaluation in Practice. Washington, DC: World Bank.
Grosh, Margaret, and Phillippe Leite. 2009. “Defining Eligibility for Social Pensions: A View from 
a Social Assistance Perspective.” In Closing the Coverage Gap: The Role of Social Pensions and 
Other Retirement Income Transfers, ed. R. Holzmann, D. A. Robalino, and N. Takayama, 161–
86. Washington, DC: World Bank.
Holzmann, R., D. A. Robalino, and N. Takayama. 2009. Closing the Coverage Gap: The Role of 
Social Pensions and Other Retirement Income Transfers. Washington, DC: World Bank.
Khandker, S. R., G. B. Koolwal, and H. Samad. 2009. Handbook on Quantitative Methods of Pro-
gram Evaluation. Washington, DC: World Bank.
Leeuw, F., and J. Vaessen, 2009. Impact Evaluations and Development: NONIE Guidance on Impact 
Evaluation. Washington, DC: World Bank.
Levy, Santiago. 2008. Good Intentions, Bad Outcomes. Washington, DC: Brookings Institution.
Ribe, Helena, David Robalino, and Ian Walker. 2012. From Right to Reality: Incentives, Labor Mar-
kets, and the Challenge of Achieving Universal Social Protection in Latin America and the Carib-
bean. Latin American Development Forum Series. Washington, DC: World Bank.
Rofman, Rafel, and Maria Laura Oliveri. 2012. “La cobertura de los sistemas previsionales en 
América Latina: conceptos e indicadores.” (English version in preparation.) Series de Docu-
mentos de Trabajo sobre Políticas Sociales No. 7, World Bank, Buenos Aires.
Thaler, Richard H., and Cass R. Sunstein. 2009. Nudge: Improving Decisions about Health, Wealth, 
and Happiness. New York: Penguin Books.

27
CHAPTER 2
Policies to Encourage 
Private Pension Savings: 
Evidence from OECD Countries
Edward Whitehouse
Private pensions provide about 20 percent of retirement income on average in Organisa-
tion for Economic Co-operation and Development (OECD) countries. They are manda-
tory or quasi-mandatory in 14 of the 34 OECD countries; in the other 20, an average 
of almost 30 percent of the working-age population has either a personal pension plan or 
an employer-provided plan. In most countries, the share of retirement income provided by 
private pensions has been increasing for at least two decades. The trend is likely to continue, 
thanks to the introduction of compulsory private pensions and the fact that more private 
retirement savings are needed to fill the pension gap resulting from lower public benefits in 
the future. Tax benefits and matching contributions are used to motivate individuals to save 
for retirement. Other policy options, such as automatic enrollment, should be considered as 
well, although rigorous evidence beyond a few countries is needed to determine how effective 
automatic enrollment is in extending coverage of private pensions.
M
andatory pension benefits, especially public pensions, will be much lower for workers 
entering the labor market today than they were for their parents and grandparents. 
Voluntary private provision for old age will therefore be necessary to maintain living stan-
dards into retirement. Indeed, many of the reforms to public pensions have been predi-
cated on the assumption that voluntary retirement savings will increase.
In some countries, such as Canada, Japan, the United Kingdom, and the United 
States, voluntary savings have long been necessary. Such savings are a new phenomenon in 
other countries, such as France and Germany. Moreover, the need to save for old age now 
encompasses more of the population, including groups such as low earners, who have not 
traditionally made active retirement saving decisions.
Some data from the Organisation for Economic Co-operation and Development 
(OECD) suggest that coverage of and contributions to retirement savings plans are 
adequate. Others imply that there may be substantial gaps. This inconclusive evidence 
Andrew Reilly of the Organisation for Economic Co-operation and Development helped prepare 
this chapter. Participants at the conference on the Potential for Matching Defined Contribution 
(MDC) Design Features in Pension Systems to Increase Coverage in Low- and Middle-Income 
Countries, organized by the World Bank and the Research Institute for Policies on Pension and 
Aging of Japan, provided useful comments. In particular, the author is grateful to Richard Hinz, 
Robert Holzmann, Robert Palacios, John Piggott, Sylvester Schieber, and Noriyuki Takayama.

28 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
suggests no grounds for complacency among policy makers. Fortunately, governments 
throughout the OECD are designing and implementing policies to encourage private 
pension saving.
This chapter is organized as follows. It begins by presenting evidence on the growing 
role of private pensions in providing retirement income in OECD countries. It then looks 
at the coverage of private pensions, distinguishing between voluntary and mandatory 
schemes. The third section examines financial incentives for voluntary private pension 
savings (tax incentives and matched contributions). The fourth section looks at mandat-
ing contributions to private pensions and soft compulsion. The last section summarizes 
the chapter’s main findings.
The Growing Role of Private Pensions
Private pensions have been playing a greater role in providing incomes for some time. 
Between 1990 and 2007, aggregate expenditure on private pension benefits grew 23 per-
cent faster than national income, rising from 1.3 percent to 1.6 percent of gross domestic 
product (GDP). Over the same period, public pension spending increased 15 percent 
faster than GDP and accounted for 7.0 percent of GDP in 2007.
The share of private pensions in total benefits in 23 OECD countries grew mod-
erately between 1990 and 2007, from 17.5 percent to 21.5 percent, with only 4 coun-
tries registering declines (figure 2.1). In 2007, private pensions provided about half of all 
benefits in Switzerland (through mandatory private pensions), the Netherlands (through 
quasi-mandatory schemes), and Canada and the United Kingdom (through widespread 
voluntary private pension provision). In Iceland, more than 60 percent of pension spend-
ing was private (through mandatory private plans). As a share of GDP, the largest private 
expenditures were in Switzerland (6.0 percent) and the Netherlands (5.2 percent), fol-
lowed by the United Kingdom and the United States, at about 4.5 percent each. 
In addition to the aggregate-level analysis of figure 2.1, it is useful to look at micro-
level data. Figure 2.2 shows the pattern of income sources from household survey data. 
For people over age 65, income comes from three sources: public transfers (mainly public 
pensions), work (employment and self-employment), and capital (mainly private pension 
benefits).
In the 27 OECD countries shown, public transfers make up an average of 60 per-
cent of older people’s incomes. People over age 65 are most reliant on the state for their 
incomes in France and Hungary, where 85 percent of their incomes come from public 
transfers. The state provides at least three-quarters of old-age income in Austria, Belgium, 
the Czech Republic, Luxembourg, Poland, and the Slovak Republic. At the other end 
of the spectrum, public transfers represent just 15 percent of average old-age income in 
Finland (this figure is low, however, because it includes mandatory occupational plans 
as capital income, whereas the national accounts and OECD Pensions at a Glance stud-
ies treat these schemes as part of the public sector). The share of old-age income derived 
from public transfers is also very low in the Republic of Korea, where the public pension 
scheme was established only in 1988 (and old-age pensions paid only as of 2008). Public 
transfers also provide less than half of old-age income in Australia, Canada, Japan, the 
Netherlands, Switzerland,
1
 the United Kingdom, and the United States.

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
29
FIGURE 2.1  Expenditure on private pension benefits as percentage of total pension expenditure, in 
selected OECD countries, 1990 and 2007
% of total pension expenditure 
1990
0  10 20 30 40 50 60 
France 
Finland 
Greece 
Czech Republic 
Austria 
Portugal 
Germany 
Slovak Republic 
Italy 
Norway 
Chile 
Ireland 
OECD
Sweden 
Denmark 
Belgium 
Australia 
United States 
United Kingdom 
Switzerland 
Canada 
Netherlands 
Iceland 
2007
SOURCE: OECD 2011. 
FIGURE 2.2  Sources of income of people over 65 in selected OECD countries, mid-2000s
0 25 50 75 
100 
Finland 
Korea, Rep. 
Mexico 
United States 
Australia 
Turkey 
Canada 
Switzerland 
Japan 
Netherlands 
United Kingdom 
Ireland 
Denmark 
Iceland 
Norway 
New Zealand 
Portugal 
Greece 
Sweden 
Spain 
Italy 
Germany 
Czech Republic 
Poland 
Luxembourg 
Austria 
Belgium 
Slovak Republic 
Hungary 
France 
public transfers 
work 
capital 
% of total income
SOURCES: OECD 2008, 2009.

30 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
In the East Asian OECD countries, employment and self-employment provide a 
very large proportion of income for people over age 65: 44 percent in Japan and 59 per-
cent in Korea. Income from work accounts for at least a quarter of old-age income in the 
Czech Republic, Greece, Iceland, Portugal, Spain, and the United States. In some of these 
countries, the large share of work income probably reflects the fact that many people have 
not had full contribution histories in the public pension scheme and therefore keep work-
ing to fill these gaps. In Iceland and the United States, for example, the retirement age is 
typically set at 65, but many people continue to work longer, and some choose to delay 
receiving their pension benefits to increase their value. In contrast, income from work 
(employment and self-employment) accounts for less than 10 percent of older people’s 
incomes in France, the Netherlands, and Sweden.
Income from capital—mainly in the form of private pensions—is the most impor-
tant source of old-age income in Australia, Canada, Denmark, the Netherlands, the 
United Kingdom, and the United States (Finland apart, for the reasons described above). 
In these countries, capital income accounts for about 30 percent or more of older people’s 
incomes.
Figure 2.2 shows average values. The composition of income varies widely across the 
income distribution: poorer older people derive their income almost exclusively from pub-
lic transfers; private pensions and other capital income play a significant part only among 
richer pensioners (see Disney and Whitehouse 2001, 2003; Förster and Mira d’Ercole 
2005). The fact that the role of private pensions and capital in retirement income has been 
growing may exacerbate growing income inequality in old age.
The analysis of aggregate financial flows from public and private pensions and 
income distribution data is backward looking, in the sense that it reflects the past design 
of pension systems. A significant development in many countries around the world has 
been the introduction of mandatory private pensions. In most cases—in Chile, Estonia, 
Mexico, Poland, the Slovak Republic, and Sweden—these pensions are a substitute for all 
or part of public, earnings-related schemes. In Australia, Norway, and Switzerland, com-
pulsory private pensions were added to existing public provision.
Figure 2.3 shows the structure of the projected retirement income package for a 
worker entering the labor market in 2008. Data are presented for 21 OECD countries 
where private pensions are either mandatory or voluntary but coverage is high. The hori-
zontal axis shows weighted-average pension wealth, defined as the net present value of 
the lifetime flow of pension benefits; it is shown as a multiple of economywide average 
earnings (the weights reflect the national earnings distribution). Iceland, where the total 
pension is worth 19 times average earnings, shows the largest overall figure. This approach 
is designed to capture differences in the structure of the pension package with individual 
earnings. In many countries, lower-income workers will be more reliant on public benefits 
than richer people.
Countries are ranked in figure 2.3 by the role of public benefits in the overall pack-
age, with the largest public share at the top. In the Czech Republic and New Zealand, 
for example, contributions to voluntary private pensions typically represent a small share 
of earnings.
2
 In Chile, where the formal pension system covers about 60 percent of the 
workforce, retirees receive more than 80 percent of their income from private pensions. 

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
31
In Mexico, where the formal pension system covers less than half the workforce, retirees 
receive about 70 percent of their income from private pensions.
One of the major drivers of the shift to private pension provision is the reduction in 
public pension benefits. Figure 2.4 shows the change in overall lifetime pensions (includ-
ing both public and mandatory private benefits) that resulted from pension reforms 
adopted since the early 1990s. It compares simulations of pension entitlements for work-
ers spending a full career under the prereform and postreform rules. The average reduction 
in benefits is 22 percent, with Finland, France, and Sweden cutting benefits by slightly 
less than this figure and the Slovak Republic and Italy cutting benefits by rather more. In 
some cases, reductions in benefits have been accompanied by increases in both coverage 
of and contributions to private pensions. But in many cases, there does not appear to have 
been a reaction to fill the pension gap.
FIGURE 2.3  Contribution of public and private components to simulated lifetime benefits in 21 
OECD countries, 2008
0 5 10 
15 
20 
Chile 
Mexico 
Iceland 
Israel 
Australia 
Netherlands 
Denmark 
Slovak Republic 
United Kingdom 
Ireland 
Sweden 
United States 
Estonia 
Canada 
Switzerland 
Poland 
Belgium 
Germany 
Norway 
New Zealand 
Czech Republic 
public mandatory voluntary