A review of international experience

Introduction and 
Conceptual Issues

3
CHAPTER 1
Early Lessons from Country Experience 
with Matching Contribution Schemes
Robert Holzmann, Richard Hinz, and David Tuesta
Matching defined contribution schemes are gaining popularity in both rich and poor coun-
tries as a promising means to reduce gaps in the participation in formal pension systems. 
Matching contributions by employers, the government, or both to defined contribution 
schemes are used alone or jointly with other interventions to motivate participation in pen-
sion schemes. Although it remains far too early to develop firm conclusions or policy guid-
ance, this chapter provides an overview of the currently available evidence that is presented 
in this volume and offers preliminary observations about the potential use of this design. 
This experience, mostly derived from higher-income countries that is now being supple-
mented with some early experience from other settings, suggests that matching is moder-
ately effective in increasing program participation but not generally measurably effective in 
raising contributions and thus benefit levels. Other interventions—which are increasingly 
guided by lessons from behavioral economics and finance—may prove to be more effective 
and typically cost much less, which may help explain some of the differences in outcomes 
across countries. It is not yet clear how transferrable the experience in higher-income envi-
ronments will be to other settings; considerable further evaluation is needed before any firm 
conclusions can be reached.
Addressing the Coverage Gap
Achieving broad pension coverage and adequate levels of income protection in old age 
remains an elusive goal for nearly every country. In general, coverage has advanced with 
development and growth in income, and there is a strong relationship between the level 
of per capita income and participation in formal pension systems. However, there are 
considerable differences among countries at similar levels of development in pension cov-
erage and in the way in which participation in pension systems has evolved in different 
settings. The differences in experiences and outcomes indicate that context, the design of 
the system, and the path of its development play a central role in the dynamics of pension 
coverage and benefit levels.
In the vast majority of countries, less than half the working population is currently 
covered by a formal pension scheme. In low-income and developing countries, the num-
ber of working-age adults participating in a pension scheme is very often less than 1 in 
This chapter was presented at review meetings and seminars in Madrid and Washington, D.C., and 
has profited from written comments by internal and external reviewers, in particular by Will Price, 
Rafael Rofman, and Sylvester Schieber. The authors are grateful for the comments and suggestions 
received but are ultimately responsible for any gaps or errors. The views expressed are their own and 
do not necessarily reflect those of the institutions with which they are affiliated.

4 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
10. Many middle-income countries have seen coverage rates decline in recent decades 
despite the expansion of coverage mandates and efforts to reform their systems to establish 
stronger individual incentives (Rofman and Oliveri 2012). In higher-income countries, 
coverage levels under mandated schemes remain high, but fiscal pressures caused by the 
generous benefits provided to early cohorts and exacerbated by rapidly falling fertility 
rates are now imposing the need for reductions in future benefit levels. Such reductions 
will require a substantial increase in supplementary retirement savings if income replace-
ment rates are to be maintained.
All of these factors have brought the imperative for the extension of coverage to 
the forefront of the pension policy debate. An earlier World Bank publication investi-
gated the role of social pensions and other transfers to increase retirement income support 
across countries of different development levels (Holzmann, Robalino, and Takayama 
2009). This volume extends the consideration of how to provide adequate income to 
elderly populations by focusing on the potential role of matching contributions to induce 
broader participation in pensions and other retirement saving to help close the coverage 
and adequacy gap.
THE POTENTIAL PROMISE OF MATCHING CONTRIBUTION SYSTEMS
An approach that has been adopted in a growing number of high-income countries is the 
provision of matching contributions. These contributions provide more tangible incen-
tives for individuals to participate in pension funds than the more traditional approach 
of mandating participation and providing preferential tax treatment, especially for low-
income groups and individuals who may not participate in the formal labor force and 
therefore receive no advantage from tax-based incentives.
In principle, matching contributions may be provided for public programs or by the 
sponsors of private occupational plans and could be associated with either defined con-
tribution or defined benefit systems. In practice, there are numerous cases of both public 
and private schemes utilizing the design; however, nearly all current examples are associ-
ated with various types of individual retirement savings accounts. This volume, which 
reviews experience to date and seeks to derive some initial observations and policy lessons, 
focuses on what is here termed matching defined contribution (MDC) schemes.
In all of these systems, the matching design feature has the common goal of increas-
ing system participation and saving levels. Four characteristics are common to MDC 
schemes examined here: individual accounts, defined contribution, direct contribution 
from sponsor to complement individual contributions, and prefunding of benefits. The 
prevalence of defined contribution systems is likely a reflection of two factors. First, 
although in theory it is possible to incorporate a contribution match in a defined benefit 
system, the linkages between the match and the benefit received are complex and less 
transparent. More importantly, the populations these arrangements are seeking to reach, 
especially lower-income groups or those in developing countries, may have irregular earn-
ing patterns over their life cycle or be predominantly engaged in the informal sector. The 
systems reviewed vary considerably in the structure and level of the matching contribu-
tion. Although the majority of schemes provide ex ante matches, some provide ex post 
matches.

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
5
The use of individual accounts with defined contributions as the underlying struc-
ture provides direct linkages between density and level of contributions to align individual 
incentives and provides transparency in the value of benefits. Directly matching contribu-
tions provides an immediate and easily understandable value proposition to prospective 
entrants to the system. Funding of the accumulated contributions and returns with finan-
cial institutions should offer credibility, portability, and appropriate returns. The expected 
saving incentives created by MDCs are an alternative or complement to other potential 
incentives such as preferential tax treatment and the presentation of choices in a way 
intended to influence behavior (what Thaler and Sunstein have termed “nudging”), as 
well as efforts to create a more conducive old-age saving environment such as financial 
education and straightforward advocacy. MDC schemes are also attractive because they 
define and constrain future fiscal exposure. They should increase coverage but not encour-
age informality. If individuals are sufficiently incentivized by the match, the accumulated 
amounts at retirement may reduce or even eliminate the need for basic benefits. Inducing 
funded supplemental coverage will facilitate reductions in mandated public systems for 
higher-income groups. Ideally, coverage should increase and the level of labor informal-
ity decrease at lower fiscal costs than expanding noncontributory systems or subsidizing 
earnings-based defined benefit programs.
POLICY QUESTIONS
Addressing the coverage gap raises a range of interrelated policy questions that vary by 
setting. In low- and middle-income countries, the primary challenge is to expand pension 
coverage beyond civil servants and the small proportion of the workforce employed in the 
formal sector. When much of the labor force has no fixed employer, or is self-employed, 
the traditional method of expanding coverage through wage-based mandated contribu-
tions is not a viable option. In these countries, establishing pension systems with incen-
tives for participation that are attractive to low-income and young people with no prior 
experience with social insurance and saving systems imposes enormous challenges. These 
groups nearly always struggle to meet short-term needs and require liquidity in any sav-
ings that they are able to accrue in order to manage a variety of risks. Effective solutions 
must not impose disincentives for participation in the formal sector through high payroll 
tax contributions or create adverse redistribution through tax-based subsidies that provide 
value only to the highest-income groups.
In other middle-income settings, the challenge is to maintain the coverage rates 
achieved in earlier decades in the face of increasing informality of the workforce due to 
transitions from centrally planned economies or changes in labor patterns resulting from 
economic development and competition in a global economy. In middle-income and 
most higher-income countries, the imperative is also to establish retirement income and 
saving systems that are able to supplement the diminished capacity of earlier earnings-
based public systems in order to provide adequate income replacement for future cohorts.
The social policy concern is accentuated by low, stagnating, and at times falling coverage 
of old-age pensions and other social programs—a stark contrast with expectations that 
emerging economies would follow the same path as the current high-income economies, 
achieving coverage expansion in step with income growth.

6 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
When initially confronted with this coverage challenge, policy makers believed that 
reforms designed to establish better links between contributions and benefits in mandated 
and earnings-related schemes, often by creating individual and funded accounts, would 
overcome these problems. To date, limited success has been achieved, however, and in 
some cases coverage has declined following such reforms.
Closely related to these social policy issues are broader economic concerns over the 
high and often rising level of informality. This growth is perceived as hampering eco-
nomic development because workers in the informal sector are considered less prone to 
learn, innovate, and use productivity-enhancing technologies. Applying basic coverage 
options—such as universal or means-test benefits for the elderly—to take care of informal 
sector workers may prove counterproductive, because such options reduce the incentive to 
become formal while increasing the pressure on formal sector workers to become informal 
as their tax burden increases. A number of recent studies indicate that this may be happen-
ing in countries across Latin America (see, for example, Aterido, Hallward-Driemeier, and 
Pagés 2011; Levy 2008; and Ribe, Robalino, and Walker 2012) although at present the 
empirical evidence remains tentative.
In all settings, fiscal concerns arise from the cost of coverage expansion through 
noncontributory basic schemes, the need to control the fiscal costs of traditional national 
earnings-based defined benefit systems, and the effects of sustained informality on pro-
ductivity and public revenues.
All three issues are potentially linked in a downward spiral in which (1) coverage 
concerns lead to the introduction or strengthening of basic provisions, which (2) con-
strain employment and increase informality, (3) increasing pressure to leave the formal 
sector, (4) worsening the fiscal position of the public pension scheme, (5) leading to ben-
efit cuts, which in turn increase the need for better basic benefits. 
Against this background, policy options in low- and middle-income countries are 
limited and largely untested. The most obvious and direct approach is to establish stron-
ger incentives for participation in formal pension and saving schemes. The key challenge 
is to develop a design that will motivate lower-income groups—which require power-
ful, immediate, and readily understood incentives to overcome their inherent consump-
tion preference and liquidity constraints—to direct their limited resources to retirement 
income. Such a system must also be attractive to informal sector workers, many of whom 
may be of moderate or even higher-income levels, while not increasing the incentives for 
workers to leave the formal sector.
PURPOSE AND ORGANIZATION OF THIS VOLUME
Several high-income countries, most notably Germany, New Zealand, and the United 
States, have adopted the MDC design to complement benefit levels under public schemes. 
Other countries, including Japan and the United Kingdom, have recently initiated MDC 
schemes to raise savings earmarked for retirement income. This has fostered interest in the 
design in a variety of other settings. Emerging economies in Asia (China, India, Thailand) 
and Latin America (Chile, Colombia, Mexico, Peru) have implemented or are consid-
ering implementing MDC-based schemes to encourage participation in voluntary and 
sometimes mandated schemes by individuals who would otherwise have no coverage at 
all. Such matching incentives may not in principle be restricted to defined contribution 

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
7
schemes, but examples of their use in defined benefit systems are rare (only the case of a 
new system in Republic of Korea is included in this volume) but worthwhile to review. 
Despite the growing experience, there is little consolidated knowledge to provide 
evidence-based policy guidance about the role and limits of MDCs for expanding retire-
ment savings, about best practice in the design and implementation of MDC programs, 
or about the interaction of MDC policies with other interventions such as financial lit-
eracy as complementary (or substitute) approaches. Nearly all of the experience with these 
arrangements and associated research has come from higher-income countries. Applica-
bility to the vastly different circumstances in middle-income and developing countries 
remains uncertain.
Against the backdrop of pilots in emerging and developed economies, the Social 
Protection Unit of the World Bank’s Human Development Network, in cooperation with 
the Research Institute for Policies on Pension and Aging (RIPPA) of Japan, organized a con-
ference in June 2011 to provide a forum for sharing information and analyzing experience 
from around the world with this emerging design. The Spanish Bank BBVA, which man-
ages pension funds in a number of Latin American countries, participated in the confer-
ence and subsequently joined with the World Bank and RIPPA to provide the resources 
to supplement and organize the material presented at the conference and produce this 
publication.
This volume—based on the presentations at the conference—provides overviews 
and analyses intended to inform the ongoing design and use of MDC schemes. The pub-
lication thus presents a first stocktaking of country experiences and some limited obser-
vations about the potential role and effectiveness of the design that can be derived from 
experience to date. It is not an effort to formulate or articulate a World Bank policy 
position or to provide any specific direction to countries considering the design. The evi-
dence is far too limited to support such an effort, and as noted above, there is insufficient 
evidence to assess the transferability of the more extensive experience from higher-income 
countries to other settings. Not considered at the conference and in the volume are similar 
matching design approaches to increased coverage under health care or other social insur-
ance programs. There is, however, considerable value in consolidating the knowledge that 
can be gleaned from the wide range of experiments with the design, considering what les-
sons are beginning to emerge and how these can inform future initiatives and a research 
agenda; these issues are summarized in this introductory chapter.
The book contains four parts. Part I provides an overview of the more general issues 
and experience in expanding supplementary pension coverage in Organisation for Eco-
nomic Co-operation and Development (OECD) countries to establish a broader frame-
work of the challenges of coverage expansion. Part II reviews experience with MDCs 
in high-income countries. Part III describes early efforts in lower- and middle-income 
settings. Part IV provides an overview of lessons from the emerging field of behavioral 
economics and reviews key issues in the enabling environment and the main parameters 
likely to be relevant in establishing an MDC in a developing country context. 
The remainder of this introductory chapter is organized as follows. The next section 
offers a conceptual framework for the objectives, intervention, mechanisms, and modali-
ties of MDCs. The following sections provide country examples, extract some tentative 
lessons from their experience, and draw preliminary policy conclusions.

8 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Objectives, Interventions, Mechanisms, and Modalities
Assessing the effectiveness of a policy intervention starts with consideration of the objec-
tives, a clear understanding of how the core elements of the intervention are defined, and 
how it expects to achieve the desired outcomes. Such a “theory of change” is critical to 
assess the effectiveness of any intervention and is at the core of monitoring and evaluation 
efforts.
1
SETTING OBJECTIVES 
Policy discussions in countries that have or are planning to introduce MDC systems sug-
gest three primary objectives: expanding pension coverage, reducing informality, and 
increasing fiscal efficiency. Considering these requires some degree of conjecture as coun-
tries are rarely explicit about the objectives of a policy intervention and even less specific 
regarding how to measure outcomes. Furthermore, the political discussion is often over-
loaded with secondary objectives driven by group interests or political imperatives.
Expanding Coverage
The increase in basic or supplementary benefit coverage is probably the primary objective 
in most countries that have or are planning to introduce MDC-type pension systems. 
This objective is particularly relevant for low- and middle-income countries, where most 
people are not afforded even basic coverage for old-age income and health care.
Supplementary coverage is additional coverage for people who participate in man-
dated systems (as workers or beneficiaries) but whose benefit levels are considered inad-
equate. Providing such coverage is particularly relevant in high-income countries, where 
the large majority of the population is covered under earnings-related schemes or everyone 
is covered under universal schemes that only partially replace their income in old age.
One measure of the success of the basic benefit coverage is the number of people 
who receive a benefit at or above a minimum level (typically the poverty line). For sup-
plementary benefits, success in coverage expansion may be measured by the number of 
people whose benefits rise above a threshold of a specified percentage or amount. In this 
case, however, the benefit increase may be caused by shifts in savings from unsubsidized to 
subsidized forms without an increase in net wealth.
Obviously, the success of the program rarely depends entirely on the financial incen-
tives provided but will also be affected by government and operational capacities in addi-
tion to other factors.
Reducing Informality 
Reducing the incentives to participate in the informal sector (that is, evading the costs of 
participating in mandated social insurance schemes) is another important objective of an 
MDC system, particularly in middle-income countries. This objective may be conceptual-
ized as establishing conditions that will encourage individuals to make contributions and 
acquire rights for the first time or contribute more to an MDC scheme. Success can be 
measured in terms of the number of registered participants as well as by changes in the 
contribution density of participants. MDCs should also not create incentives to reduce 
the level of pension savings by those already contributing.

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
9
In developing countries, mandated participation in pension systems is often condi-
tional on the number of employees in a firm; workers in firms below a certain size may 
be exempt from mandatory contributions. In other instances, workers may not be able to 
contribute because the firm is not formally registered or licensed. In such cases, an MDC 
alone may not be the appropriate intervention.
Increasing Fiscal Efficiency
MDCs create fiscal costs—either directly, through the matching contributions by govern-
ments, or indirectly, through preferential tax treatment of individuals or their employers. 
A simple measure of success would be the increase in the coverage rate or adequacy target 
per currency unit of public expenditure for matching and/or foregone tax revenue result-
ing from the increased savings that is excluded from taxation.
Another fiscal element that is often explicitly part of the design of an MDC scheme 
is the cost saving through reduced transfers to the elderly in the form of universal or 
means-tested benefits. These kinds of ex post transfers may include the costs for minimum 
benefits in an earnings-related scheme. Very optimistically, one could imagine a take-up of 
low-level matching contributions that largely eliminates the need for these transfers, with 
the saved fiscal costs well exceeding the new fiscal costs of the matching payments. Very 
pessimistically, one could imagine that MDCs merely shift savings from unsubsidized to 
subsidized forms—or, even worse, that most people substitute public funds for individual 
savings, with take-up concentrated largely in the upper-income strata.
Other Objectives
In a system with progressive income tax rates, favorable tax treatment of pension contribu-
tion disproportionately favors people with higher incomes. By providing a subsidy that 
is directly proportional to the contribution, an MDC scheme may not suffer from this 
shortcoming unless, again, the take-up is concentrated among high-income earners.
MDCs may also be linked with the objective of facilitating the transition to a fully 
or partially funded system by inducing individual contributions to support a funded or 
capitalized reserve from which benefits will be paid.
DEFINING THE INTERVENTION: CORE ELEMENTS OF MATCHING DEFINED 
CONTRIBUTION SCHEMES
The following features define an MDC scheme: 
 
• Individual account. Accounting records should be maintained to clearly distin-
guish individual contribution and retirement saving outcomes.
 
• Defined contributions. Benefits are solely based on the accrued value of contri-
butions and earnings on accumulated assets. 
 
• Sponsor. Employers, the government, or other sponsoring entities make direct 
financial contributions to encourage individuals’ participation.
 
• Own contributions. Regular and own contributions by individuals are expected, 
although sometimes after an initial “kick-start” contribution.

10 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
 
• Funding. In principle, schemes can be funded or unfunded (using a derived cred-
iting rate and notional accounts). In practice, there are no examples of matching 
being incorporated into unfunded retirement savings schemes.
 
• Mandatory or voluntary. Mandatory schemes typically focus on extension of 
coverage. Voluntary schemes are typically employer based or designed to supple-
ment other systems with broad coverage but relatively low benefit levels.
Other known design elements of MDCs are complementary and potentially sub-
stitutive to matching. As with matching, these intend to encourage the participation of 
individuals in retirement saving programs. The main examples include the following: 
 
• Tax preferences for contributions and/or benefits under a comprehensive income 
tax approach that aims to eliminate distortions of taxation on savings and move 
toward a consumption-type structure (exempting contributions and interest from 
taxation but taxing the disbursement, or taxing contributions but leaving interest 
earnings and disbursement tax free)
 
• Nudging or choice architecture (Thaler and Sunstein 2009) to motivate indi-
viduals to participate in savings plans; specific mechanisms include automatic 
enrollment, default contribution levels, investment options, and thresholds for 
matching levels
 
• Financial education and related interventions to create the enabling environment 
for individuals to learn about the importance of planning and retirement saving 
and offer them support to acquire the requisite skills, attitudes, and behaviors.
ENVISAGED MECHANISM: THE THEORY OF CHANGE
Identifying the mechanism through which outcomes will be achieved begins with an anal-
ysis of the problems to be solved and the rationale for government intervention. Public 
interventions are traditionally undertaken for two key reasons: to correct market failures 
and to redistribute income. Market failures are often linked to asymmetric information, 
which leads to poorly functioning (or nonexistent) markets. Government interventions 
are intended to substitute for or improve market outcomes. Redistribution is undertaken 
to correct perceived failures in the way markets generate and distribute income. To these 
motivations for public intervention, a third has been added in recent years: to correct 
the behavioral limitations of individuals. A fourth reason may be to correct government 
failure itself—by, for example, redesigning social insurance programs that do not achieve 
sufficient coverage due to poor design or implementation.
Individuals may rationally evade government social insurance programs for a variety 
of reasons, including the high costs associated with participation in the formal sector, 
liquidity preferences, the poor fit between a program and individual preferences, and lack 
of trust. These factors can be conceptualized as increasing the discount rate individuals 
apply to pension schemes. An MDC scheme seeks to increase the internal rate of return of 
a pension scheme in order to increase participation, contributions, or both.
Individuals may fail to employ market-based social risk management instruments 
to address long-term contingencies, such as old age, for a variety of reasons, including 
the absence of appropriate instruments, the inability to plan ahead, and the existence 

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
11
of more immediate shocks for which risk management instruments are not available. In 
the absence of an additional external motivation, these factors may make it rational for 
some people to focus on the short rather than the long term. Under such circumstances, 
matching may actually distort rational individual decisions and divert resources from, say, 
human capital accumulation to retirement saving.
Whether contributions grow as a result of the provision of a match is a priori unde-
termined. A saving subsidy in the form of a match potentially creates both a substitution 
effect (which makes current consumption more expensive and hence increases current sav-
ing) and an income effect (which increases the demand for current and future consump-
tion and hence decreases current saving). MDC interventions with multitier structures 
have even more complicated effects on individuals, which depend on the position of the 
individual before and after the introduction of or change to the intervention. Some con-
figurations are predicted to clearly encourage or discourage saving; the effect of others is 
indeterminate. The predictions are broadly borne out in the country lessons discussed in 
the chapters.
DESIGN AND ADMINISTRATIVE MODALITIES
The main design features of MDC schemes include the following: 
 
• Matching rates typically range from 25 percent to 100 percent but can reach 
300 percent and more. For supplementary MDCs offered by employers, there 
may be complex structures with lower match rates for higher contribution levels 
or no match for a first tier before a declining match applies. For basic matches 
offered by governments, flat-rate and multitiered matching contributions (which 
address informal sector workers, for whom earnings cannot be easily established) 
are more typical.
 
• Thresholds and ceilings on the contribution base (linked to a multiple of average 
income) or overall limits on matched amounts direct subsidies to lower-income 
groups and limit fiscal costs.
 
• Eligibility conditions for matching include very specific characteristics of 
the beneficiary (income level, age, family status, number and age of children, 
employment status, company size, level of formality) to focus the match on a 
target population.
 
• Matching contributions provide an ex ante transfer linked with ex post transfer 
(minimum pension) that is conditional on a required amount of prior contribu-
tions (length or level) by the beneficiary or family members.
 
• Withdrawals can be made to purchase a first home or to buffer periods of unem-
ployment or other contingencies.
 
• In addition to the match, consumption-type tax treatment is provided for sav-
ings, subject only to income tax at time of receipt (similar to other pension sav-
ings plans).
 
• Payout modalities include lump sums, phased withdrawals, mandated immedi-
ate or deferred annuitization, and annuitization defaults (for example, minimum 
annuity levels, deferred annuities).

12 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Administrative modalities include the following:
 
• Contribution collection by employer, financial institution, social security institu-
tion, or local “aggregators” (nongovernmental organizations or others that collect 
payments from contributors)
 
• Recordkeeping and client communication by financial institution or regional or 
national social security fund
 
• Asset management by pension/health funds, specialized asset manager, or regional 
or national social security institution
 
• Benefit disbursement by financial institution or regional or national social secu-
rity institution.
Country Examples
MDC schemes reflect diverse policy objectives at different stages of the pension sys-
tem development. Common features can be organized by the main coverage extension 
scenarios:
 
• Supplementing universal (basic or means-tested) benefits 
 
• Supplementing low or reduced earnings-related benefits
 
• Expanding coverage within the mandated social insurance scheme
 
• Expanding coverage outside the mandated social insurance scheme (universal 
approach)
 
• Expanding coverage outside the mandated social insurance scheme (sector- or 
group- specific approach).
This section illustrates these coverage objectives with country examples detailed in 
later chapters in this volume to distill lessons and to raise policy and research questions.
SUPPLEMENTING UNIVERSAL (BASIC OR MEANS-TESTED) BENEFITS
Two countries have sought to use MDC schemes to supplement universal benefits or 
limit mandated provisions: New Zealand (which introduced such a plan in 2007) and the 
United Kingdom (which had intended to introduce such a plan in 2010 but canceled the 
program following the election of a new government).
New Zealand is one of the few countries with an old-age program in which everyone 
receives a pension (this type of program is known as a demogrant). Every resident of New 
Zealand age 65 and older who has lived in the country for at least 10 years receives a flat-
rate pension based on a certain percentage of the average wage. Provision for old age above 
this basic and taxable benefit was left to voluntary savings, creating issues of adequacy of 
retirement income.
After discussions and a defeated referendum on introducing mandated earnings-
related benefits to eventually replace the demogrant, the government introduced an 
MDC-type scheme, the design of which was, to a significant degree, informed by the 
emerging field of behavioral economics; this design is both complex and comprehensive. 

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
13
The KiwiSaver program, described in chapter 5, utilizes several tiers of matching-type 
incentives. It provides a flat contribution (known as a kick-start) on the opening of an 
account, a tax credit (actually a pure subsidy) for contributions, and mandates match-
ing contributions by employers (2 percent of payroll, to increase to 3 percent by 2013). 
Employees can choose to contribute 2 percent, 4 percent, or 8 percent of their earnings. 
It combines these subsidies with an auto-enrollment feature in which new employees are 
automatically enrolled in the system but afforded the ability to opt out after a specified 
period. It allows preretirement withdrawals for several purposes, most notably the pur-
chase of a first home.
The United Kingdom has had a Beveridge-inspired basic pension scheme with flat-
rate contributions and benefits since the 1940s. The country has a long history of trying 
to improve saving outcomes of all types—from short term (precautionary) saving to very 
long-term pension and retirement income saving—in order to complement basic pension 
plan provisions. The Saving Gateway program, described in chapter 6, planned to use 
matching contributions to increase the savings of people with low incomes. Programs—
tested through limited-scale pilot efforts in 2002–04 and 2005–07—experimented 
with different matching rates, contribution limits, eligibility rules, and recruitment 
mechanisms.
In 2008, the Labour government then in power decided to roll out Saving Gateway 
nationally to up to 8 million people, or 20 percent of the population between the ages 
of 16 and 65. In May 2010, the newly elected Conservative–Liberal Democrat coalition 
government canceled the scheme as part of its broader program of fiscal retrenchment. 
However, with the National Employment Savings Trust (NEST), the U.K. government 
is aiming to increase coverage of the privately funded pension system from 2012 by using 
auto-enrollment together with matching as another example of nudging behavior. A 
minimum contribution requirement is gradually being introduced, ultimately reaching 
4 percent of wages by individuals, 3 percent from employers and 1 percent from the 
government in tax relief. Employees and employers are free to contribute more, and some 
employers will match higher contributions. Auto-enrollment is being used because such 
employer matching has been in existence for many years, and many employees did not 
join schemes even when very generous matches were offered by employers.
2
SUPPLEMENTING LOW OR REDUCED EARNINGS-RELATED BENEFITS
A number of MDC schemes are related to low levels of income replacement or reforms 
that reduced the generosity of public earnings-related schemes, primarily in high-income 
OECD countries. There are wide variations in design and operation across countries. 
A number of high-income countries have introduced compensatory supplementary 
and funded pensions with more modest amounts of direct fiscal support through flat-rate 
and similar subsidies (beyond consumption-type tax treatment). To provide some back-
ground on the overall issue of providing supplementary coverage, chapter 2 reviews the 
experience among OECD member countries with policies that encourage private pension 
saving.
The 401(k) plans used in the United States (named after a section of the tax code 
that authorized the particular type of tax-preferred saving arrangement), examined in 
chapter 3, are probably the most important and most investigated MDC scheme in the 

14 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
world. These plans emerged in the early 1980s as part of the employer-sponsored pen-
sion system that supplements the relatively low (on average, less than 40 percent) income 
replacement rates of the mandatory public social security system. These were part of the 
transition in voluntary employer pension coverage from defined benefit to defined contri-
bution plans. The underlying arrangement enables workers to determine the level of pre-
tax contributions (technically known as salary deferrals) to a defined contribution plan. 
To ensure that the value of this tax preference did not disproportionately favor individuals 
with higher incomes (who not only had a greater ability to save but who, because of the 
progressive income tax system, obtained a larger value from the tax deferral), rules were 
established that limited the amount higher-income workers could contribute; this amount 
is linked to the overall average share of income deferred by all participants. This rule led 
employers to create a wide range of matching contribution designs to induce higher con-
tributions from low- and average-income workers to enable the higher-paid to take full 
advantage of potential tax preferences. The widespread use of 401(k) plans and the diver-
sity of matching arrangements have been the subject of extensive study and are the source 
of much of the knowledge about the behavioral effects of various matching designs that 
are discussed in chapter 15. 
In 2001, Germany introduced another important variant of the supplemental 
arrangement that is known as the Riester pension (in recognition of the former minister of 
labor and social security who was a main proponent of the initiative) after having imposed 
a significant prospective reduction in the value of public pension system benefits. The 
Riester pension plans that were introduced in stages from 2002 to 2008 involve a means-
tested match of contributions from the government, an additional per child subsidy, a tax 
preference on contributions up to a maximum level, and an associated (largely annuitized) 
payout plan. The development of and experience with this system is discussed in chap-
ter 4. This MDC scheme is heavily subsidized by the budget and at times substitutes for 
corporate pensions. After a slow start and several design changes, including simplifica-
tions, in 2005, Riester pension 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.
In more recent years, Japan has, like the United States, sought to expand its occu-
pational pension system to include defined contribution plans as it attempts to address 
dramatic demographic changes that will constrain the public social security system and 
thereby relieve some of the pressures on traditional corporate defined benefit plans. The 
Japanese pension system and the new defined contribution plans are discussed in chap-
ter 7. In an interesting variation, matching was introduced in 2011 with a design in which 
the employee is permitted to match the employer’s contribution to the defined contribu-
tion plan—reversing the typical arrangement. Thus far, however, there is no evidence that 
this design will meet with any more success than other defined contribution plans, which 
have had only very limited acceptance, thus indicating the importance of context and 
incentives in matching arrangements.
EXPANDING COVERAGE WITHIN THE MANDATED SOCIAL INSURANCE SCHEME
Social insurance schemes in middle-income countries have difficulty expanding cover-
age to low-income groups. Some vulnerable groups, such as youth, are difficult to reach, 

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
15
but their early integration into a social insurance scheme is important for later behavior. 
Everywhere in the world, self-employed workers exhibit low participation and contribu-
tion efforts, particularly in rural areas.
A number of countries are attempting to encourage participation by groups that are 
difficult to integrate by offering matching incentives within the mandated social insurance 
scheme. The match is financed by the budget or through redistributed contributions.
Korea’s social insurance pension scheme, discussed in chapter 8, was established 
in 1988 and made universal in 1999. It covers all working-age (age 18–60) adults who 
make or are exempted from making contributions. Contribution delinquency remains 
an issue, creating concerns for future pension adequacy. To strengthen participation 
of farmers and fishers, the government has, since 1995, when the national pension 
extended its compulsory coverage to all rural residents, offered contribution subsidies 
of 50 percent of the total contribution (with a cap). This subsidy is scheduled to end 
in 2013. Although the Korean pension scheme is of the defined benefit type and does 
not squarely fall into the MDC definition, it offers one of the few matching schemes in 
middle-/high-income countries that have been evaluated. As the subsidy does not apply 
to other self-employed groups beyond farmers and fishers, it offers a natural experi-
ment for testing its effectiveness. It is found to have had a moderate effect in increasing 
participation in the system by individuals who would otherwise be expected to evade 
making contributions.
In Colombia—which is examined along with Mexico and Peru in chapter 10—at 
least three MDC schemes have been established to encourage voluntary contributions. 
Two are already in operation, one as part of the funded individual account system and 
one as part of the alternative unfunded defined benefit scheme. Individuals must choose 
between the two schemes. In both cases, the match provides minimum income guarantees 
for retirees and is financed by contribution income.
Mexico has had at least two matching-type schemes in operation since the mandated 
individual account pension scheme was established in the late 1990s. The first scheme 
targets low-wage workers by providing a flat-rate “social contribution” to all participants 
below an income ceiling (introduced in 2009) equal to 5.5 percent of the presumed mini-
mum wage for each day of work. A second scheme, for civil servants, was introduced with 
the 2007 reform that moved their pensions toward a funded defined contribution scheme. 
The match—a government match of Mex$3.25 for each Mex$1 of employee contribu-
tion, with a ceiling of 2.0 percent of the contribution base for the employee and 6.5 per-
cent for the employer—should increase contributions.
Peru’s MDC scheme, originally legislated in 2008, was designed to promote cov-
erage for workers in small and microenterprises while enhancing competitiveness and 
encouraging participation in the formal labor market. The matching component of 
this law, the Welfare Pension System (Sistema de Pensiones Sociales), has recently been 
included in the 2012 reform of the private pension system. This scheme is focused only 
on microenterprise employees (those working in firms with no more than 10 employ-
ees) and is mandatory for people under age 40 and earning less than 1.5 times the legal 
minimum wage. Both the contribution rate and the government matching are to be 
defined during the year. It is likely that the government will finally implement this 
scheme.

16 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Chile, examined in chapter 9, introduced two youth employment subsidy schemes, 
with the objective of promoting formal youth employment through incentives for both 
the supply of and demand for labor. The schemes’ introduction, in 2008, occurred around 
the time of a major pension reform that introduced ex post subsidies—in particular, guar-
anteed old-age income through the solidarity pillar of the pension system. The Subsi-
dio Previsional a los Trabajadores Jóvenes (SPTJ) scheme provides an explicit subsidy for 
social security contributions. A first component (introduced in October 2008) amounts 
to 50 percent of social security contributions at minimum wage. This, however, is paid 
to the employer to provide a subsidy for the cost of hiring younger workers while also 
providing an incentive for contributions to the social security system. A second compo-
nent (introduced in July 2011) provides a matching payment to the worker of the same 
amount to subsidize contributions.
EXPANDING COVERAGE OUTSIDE THE MANDATED SOCIAL INSURANCE SCHEME 
(UNIVERSAL APPROACH)
In most low- and many middle-income countries, the majority of workers work in the 
informal sector. Integration of the workers into the formal sector pension scheme is 
unlikely to be realistic in the near term. A few countries, including India and Thailand, 
do offer voluntary coverage outside the mandated social insurance scheme and provide a 
government match to induce participation.
In India, discussed in chapter 12, less than 10 percent of the population works in 
the formal sector—and much of the formal sector employment is in the public sector. 
To address a looming pension problem in an existing defined benefit system for cen-
tral government workers, the reform of 2004 introduced a funded defined contribution 
scheme for new entrants to the civil service and unbundled recordkeeping and asset man-
agement. With an administrative structure established through this reform, the New Pen-
sion Scheme provided the potential infrastructure for an MDC scheme for all informal 
sector workers. An effort to expand the system was initiated in 2010, by which an annual 
matching contribution of Re 1,000 (about $20) was offered for all workers who enroll and 
pay contributions of Re 1,000–12,000 (about $20 to $225) a year, with no means test 
applied. To enhance the decision architecture, the scheme uses “aggregators” at the village 
level to collect contributions and has a simplified account structure with lower fees. The 
scheme’s very recent implementation makes it impossible to draw conclusions about its 
success. An early look at a small set of data indicates that participation in the scheme may 
be associated with income and education levels and might be negatively associated with 
access to alternative sources of retirement savings.
Thailand, discussed in chapter 14, initiated a national MDC scheme for informal 
sector workers in 2012. Individual deposits can be made at any time, with a minimum 
deposit of B 50 (about $2). The government match and ceiling are graduated by age, with 
a 50 percent match for people age 15–30 up to a maximum of B 3,000 (about $100), an 
80 percent match for people age 31–50 with a maximum of B 4,800 (about $155), and a 
100 percent match for people age 51–60 with a limit of B 6,000 (about $200). 
These efforts in India and Thailand trigger many questions that will require further 
evidence and rigorous evaluation to answer, including the following:

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
17
 
• What is the appropriate matching structure for informal sector workers in low- 
and middle-income countries? In particular, how important are matching rates 
compared with contribution ceilings?
 
• What is the role of matching compared with other participation determinants, 
such as the decision architecture and potential members’ perceptions of the ser-
vice providers?
 
• How effective are aggregators and efforts to increase access through “points of 
presence” (the establishment of a means to make contributions in a village such as 
a bank or post office) in increasing participation into rural areas?
 
• What mechanisms encourage continued pension saving efforts beyond increased 
participation?
EXPANDING COVERAGE OUTSIDE THE MANDATED SOCIAL INSURANCE SCHEME 
(SECTOR- AND GROUP-SPECIFIC APPROACHES)
In countries with a large informal sector and the desire to increase formal sector participa-
tion, a universal voluntary system for informal sector workers may not be the best way 
to achieve higher levels of pension coverage. A focus on specific groups may be justified, 
however. Self-employed workers lend themselves to such an approach, as they are difficult 
to integrate into a formal mandated scheme.
A number of countries have started to move in this direction. China, discussed in 
chapter 11, started a voluntary MDC pilot for the rural sector in 2009; the program was 
expanded to the urban sector in 2011, and full national coverage is envisaged by 2013. 
These voluntary schemes will coexist with the mandated urban pension scheme, which 
covers about half the urban workforce. 
China’s introduction of the National Rural Pension Scheme and the Urban Resi-
dent Pension Scheme has been one of the most ambitious voluntary pension saving and 
minimum elderly assistance schemes in a low- or middle-income country. Both schemes 
have innovative features. They provide a basic pension benefit from age 60 on if a vesting 
period of 15 years is fulfilled. Individuals select a contribution level of between Y 100–500 
($16–$80) per year. A partial match is then provided by local governments, with a mini-
mum required match of Y 30 ($5), although this may be at a higher level that is locally 
determined. The rapid expansion of these voluntary schemes that now are reported to 
cover more than 350 million participants may be linked to the minimum pension benefit, 
which is offered immediately if conditions are fulfilled. But it also reflects the influence 
the government has on inducing participation in public social insurance systems as well as 
a solid advocacy campaign.
Another important challenge in developing countries is reaching independent work-
ers, who offer their labor in often irregular patterns—such as fishers in offering their 
labor and skills to the owners of boats. Because this employment model does not lend 
itself to long-term relationships, these kinds of workers cannot make steady contribution 
payments into a social insurance fund. Their situation calls for innovative new payment 
and financing solutions adapted to the particular circumstances of the targeted workers 
that include nudging elements as well as matching-type contributions by contractors. An 

18 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
exploratory study of Cape Verde and Tunisia, presented in chapter 13, offers an outside-
of-the-box thought piece on this issue.
These country examples of innovative MDC schemes suggest that coverage can be 
expanded by moving beyond simple matching design to improving the decision architec-
ture. Linking ex post and ex ante transfers, as China has done, or offering new financing 
and payment structures, as proposed for Cape Verde and Tunisia, exemplify this approach.
Tentative Lessons from the Experience with Matching 
Contributions
The country experience and reported results offer a rich, although incomplete (and likely 
somewhat biased), body of evidence on the use of matching contribution arrangements. 
Most of the rigorous evaluation of the dynamics and outcomes of matching are focused 
on participation and contribution effects in 401(k) plans. These address particular groups 
(generally higher-income employers in a high-income country who are offered the chance 
to participate in an employer-sponsored plan) and therefore may not be relevant for other 
countries. These studies do not address some of the key questions for other settings such 
as impact on informality or overall fiscal effectiveness, as the matching is by employers 
and any fiscal effects are indirect. Most other country experiences have not (yet) been 
subject to rigorous evaluation; in many circumstances, there has been no evaluation at all. 
Consequently, the discussion presented below includes more hypotheses than lessons that 
have been inspired by empirical results and validated in other countries.
EXPANDING COVERAGE 
Participation
There is consistent empirical support across country income levels that matching is effec-
tive in increasing participation. The evidence from a few high-income countries (mostly 
the United States) indicates positive but modest effects of matching on participation, with 
overall effects increasing participation in the range of 5–10 percent of potential beneficia-
ries. The associated finding that a 25 percent match of individual contributions is associ-
ated with about a 5 percent increase in participation appears to be robust across a range of 
programs and analysis in the United States. This magnitude is also broadly consistent with 
results from Korea, where a 50 percent match for farmers and fishers increased the prob-
ability of making a pension contribution by 7.4 percent. The presence of a large initial 
match—a significant element of the KiwiSaver system in New Zealand—elicited enroll-
ment from many people with little or no earnings, including children, providing further 
evidence of the potential effectiveness of significant matches on at least initial enrollment.
Increase in Retirement Saving
The effect of the match on the saving rate is typically found to be small, and the sign is not 
always positive or statistically significant. This finding is consistent with the theoretical 
ambiguity arising from the conflict between substitution and income effects. What seems 
to emerge consistently is that the structure of the match—the matching rate, thresholds, 

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
19
and caps—does have significant consequences. Consistent across estimates (essentially 
from the United States and, to a lesser extent, the United Kingdom), the match thresh-
old seems to have a greater impact than the matching rate. Providing a lower match of 
25 percent on contributions for a higher level of 10 percent of pay will induce individu-
als to save more than a higher match of 50 percent for a lower level of 5 percent of pay, 
although both formulations result in similar costs to the organization providing the match 
(see chapters 3, 6, and 15). A possible explanation for this result is that matching acts as 
a signaling device or implicit advice on saving levels. Also notable is the “stickiness” of 
saving levels, as evidenced by the fact that most people remain at contribution levels that 
were established as defaults even when the defaults are subsequently reduced.
Increase in Overall Saving/Total Pension Wealth
Comprehensive evidence from the United States on the effect of scheme saving on other 
saving is mixed, but suggests up to 20–30 percent net increases in saving levels (see chap-
ters 3 and 15). In the United Kingdom (chapter 6), the saving rate increased but the 
measured net worth of individuals did not change. Evidence from Germany (chapter 4) 
suggests that matched saving did not squeeze out other saving.
Other Determinants for Participation and Contribution Efforts
The evidence from developed and developing countries strongly suggests that other fea-
tures of savings programs and related interventions may have a critical—perhaps even a 
dominant—effect on participation and contribution levels. Most of these features have 
not yet been subject to the rigorous testing across a range of settings that could begin to 
distinguish between inherent effects and those associated with a particular set of circum-
stances, cultural setting, or population group.
 
• Automatic enrollment and defaults. Evidence from the United States, the 
United Kingdom, and New Zealand suggests that making participation the 
default option has two to four times the impact of the match. (Of course, an 
automatic enrollment default option works only under formal employment con-
ditions.) The role of other default schemes on contribution efforts is mixed and 
at times negative, possibly because of inertia or the low default contribution rate. 
 
• Simplification of design and access. Empirical results for the United States and 
the United Kingdom and lessons from the German Riester pensions suggest that 
simplified design affects participation and, perhaps, contribution/saving efforts.
 
• Social marketing and advocacy. Retirement saving remains an objective that 
most will embrace but find difficult to implement. In the United States, employ-
ers have found that information sessions and advocacy are a useful adjunct to 
the incentives of matching contributions. In Germany, take-up of the match 
increased with greater awareness of the scheme associated with information cam-
paigns. New Zealand has coupled introduction of its system with information and 
advocacy campaigns which are perceived to have had the expected effect. There 
is some very preliminary evidence of a positive impact through the introduction 
of account aggregators in India. Yet rigorous empirical evidence is missing on 

20 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
the effectiveness of information campaigns for short-term participation and long-
term contribution efforts.
REDUCING INFORMALITY
Evidence on the effect of MDC schemes on informality is very limited and mixed. There 
is no empirical evidence that the modest matching schemes in Colombia or Mexico have 
reduced informality or increased coverage, nor any clear expectation that the soon-to-
be-implemented matching scheme in Peru will lead to these outcomes. This is possibly 
because of the small size of the programs and important distortions in the labor market. 
Data from national household surveys in these three South American countries show an 
enormous potential for saving in the informal sectors. The estimate for Korea suggests 
that the match for farmers and fishers had a modest positive impact on participation in 
the national pension scheme. The matching programs in Chile to incentivize the partici-
pation of young workers in the formal labor market—and hence in the pension scheme—
increased participation, but the programs have not yet been subject to rigorous evaluation. 
The rural pilots in China that started in 2009 reportedly reached 358 million rural work-
ers as of the end of 2011, and full coverage (of some 500 million people) is envisaged in 
2013. The pilots increased coverage but, strictly speaking, had no effect on informality. 
This is likely to be because the target group had very little potential to become formal sec-
tor workers to begin with.
A number of countries (including China, India, and Thailand) have established or 
are planning voluntary matching programs in parallel to formal matched or unmatched 
schemes. Individuals joining these voluntary programs have no obligations to join the 
mandated scheme (as they do in Germany and the United States). Not enough evidence is 
available to determine whether these schemes create disincentives to formalization.
INCREASING FISCAL EFFICIENCY
Assessing the effect of matching schemes on coverage and informality is relatively straight-
forward, and effectiveness comparisons can be done within or across similar schemes. 
To measure fiscal efficiency requires the pricing of the intervention and comparison 
with alternatives or a counterfactual. This analysis is hardly ever done, however, as these 
schemes are designed and implemented—making consideration of potential fiscal effects 
more art than science. 
In considering the possible fiscal efficiency of matching schemes, two comparisons 
are proposed: (1) a comparison of the fiscal costs with the additional savings volume cre-
ated by the match and (2) a comparison of fiscal costs for the match with the costs of alter-
native interventions such as ex post subsidies. The most useful comparison may depend 
on the purpose of the intervention. If the objective is to promote supplementary coverage, 
evaluating the marginal increase in savings seems more relevant. If promoting basic cover-
age is the objective, the more relevant comparison is with alternative interventions.
Comparison to Additional Savings Created
Comparison of total fiscal costs with MDC savings created can be done on a flow and 
stock basis. Each requires some heroic assumptions.

1.  EARLY LESSONS FROM COUNTRY EXPERIENCE WITH MATCHING CONTRIBUTION SCHEMES 
21
 
• Comparing the annual contributions by participants to annual fiscal costs. 
Assuming that all contributions are new saving, fiscal effectiveness requires that 
the ratio of the annual flow of new savings to the annual cost of matches (a fis-
cal efficiency ratio) be larger than 1 so that the public expenditure and potential 
public dis-saving are at least compensated by additional contribution revenues of 
equal magnitude. Taking account of distortions (for example, through changes in 
general revenue collections) would increase the opportunity costs and therefore 
increase the required fiscal efficiency ratio. Using empirical results of new savings 
created of, say, only a third, fiscal effectiveness would require an efficiency ratio 
greater than 3. For the German Riester pensions, for example, the share of annual 
contributions to direct fiscal costs is slightly above 2 and falling (chapter 4).
 
• Comparing the additional national capital stock created with the accumu-