A review of international experience

Implementation Issues in Low- and 
Middle-Income Countries
Robert Palacios and Mike Orszag
Experience in a number of countries indicates that matching contribution schemes can 
potentially increase pension coverage and the level of retirement savings. However, this 
hopeful assessment is derived mainly from outcomes observed in a relatively small number 
of high-income settings with relatively small programs supplementing extensive and long-
established national pension systems. These schemes have been set up in environments char-
acterized by high levels of public and private institutional development and benefit from 
well-established and sophisticated financial and information management systems. Imple-
menting matching contribution pension schemes in other settings requires careful consider-
ation of the necessary enabling conditions and system parameters that will ensure viability. 
Enabling conditions include the legal, institutional, and information systems required to 
maintain and control individual accounts; the capacity to invest accumulated assets effi-
ciently; and the ability to exercise oversight and control to ensure the integrity of the match-
ing payments. Design parameters include the target benefit level, the total contribution 
required to achieve the benefit level, the extent of the match itself, the age of withdrawal, 
and the rules for accumulation and withdrawal. These parameters will need to be adjusted 
as incomes grow, the size of the formal sector expands, and the population ages.
I
n response to persistently low coverage rates, a growing number of low- and middle-
income countries have been exploring alternatives to traditional social insurance models 
that link retirement benefits to payroll tax payments. Targeted or universal cash transfers 
to the elderly are one response to the coverage gap, especially in countries where broader 
social assistance programs do not already exist. Another approach is to increase incentives 
to participate in contributory programs, especially for the self-employed and informal sec-
tor workers.
These programs can be seen as a subset of a broader category of voluntary contribu-
tion-based pensions. In richer countries, the incentive to sacrifice current consumption or 
liquidity in order to save for retirement typically takes the form of tax incentives that exclude 
or defer income taxes on earnings that are saved in designated retirement savings vehicles or 
through the provision of tax credits that effectively subsidize contributions. The tax incen-
tives result in foregone revenues and are therefore often characterized as “tax expenditure.” 
Tax-deductible contributions are common in Organisation for Economic Co-operation 
and Development (OECD) countries, although tax credits are limited to a few countries 
(for example, Turkey). Participation in voluntary private pensions is correlated with these 
incentives although not to the extent that is typically expected (see chapter 2).
The use of tax expenditures to encourage pension savings is premised on the public 
policy objective of generating adequate retirement income for some portion of the popu-
lation. This premise implicitly assumes that this type of incentive increases the net level 

312 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
of retirement savings. As discussed in earlier chapters, however, the empirical evidence 
is mixed. Some studies show a positive savings impact, while others find that only the 
composition of savings is affected. At the same time, progressive income tax schedules 
combined with deductions that take place at higher marginal rates for the rich often lead 
to a regressive subsidy, especially when there is a significant substitution of tax-preferred 
savings for savings that would otherwise have occurred.
In countries with very low or zero marginal tax rates for low-income individuals, 
this preferential treatment provides no meaningful incentive for pension-related saving. 
In poorer countries, tax incentives are less relevant, because only a small share of the 
population actually pays income taxes. The rationale for a tax-based subsidy that could be 
expected to accrue only to the highest quintile of the income distribution is therefore even 
weaker than in richer countries.
An alternative approach that equalizes the incentive regardless of income tax status 
is to provide the subsidy directly, in the form of matching pension contributions. This 
approach is being implemented in several low- and middle-income countries, including 
Colombia, Mexico, and Peru; China; and India (see chapters 10, 11, and 12, respectively). 
Others have passed laws to create matching programs which are in the very early stages of 
implementation. There is little evidence yet of the likelihood of success or the outcomes to 
be achieved by these initiatives. 
To some degree, these new programs are motivated by the positive results in expand-
ing voluntary pension coverage in several high-income countries, notably the United 
States, Germany, and New Zealand (see chapters 3, 4, and 5, respectively). These results 
have been achieved under very different circumstances. The schemes have all been con-
ceived as complements to long-established national pension systems that have already 
achieved very high rates of coverage and participation and that provide basic levels of guar-
anteed old-age income support for the vast majority of people. Most have been introduced 
as complements to occupational pension systems that benefit from a well-developed legal, 
financial, and institutional infrastructure that either predates the pension system or has 
been developed to support it. This infrastructure includes reasonable, functioning mar-
kets for financial services and products and a vibrant, competitive industry that provides 
recordkeeping and asset management for pensions and other savings and investment insti-
tutions. Further, these pension systems have arisen in countries where property rights and 
reliable third-party management of financial assets are relatively effective. They also have 
social security and tax administration institutions with information systems that can issue 
and control individual identification numbers and maintain data to track transactions to 
ensure the reliability and integrity of contributions and payments.
Although matching contributions are not restricted to defined contribution schemes, 
these are the most common variant found in developing countries, largely due to the dif-
ficulty of applying a defined benefit or quasi–defined benefit approach to informal sector 
workers where it is extremely difficult to track earnings that tend to be highly variable. For 
this reason, this chapter focuses on the case of matching defined contributions (MDCs). 
Specifically, it reviews the key challenges, minimum enabling conditions, and some of 
the key parameters to be determined when designing a matching scheme in low-income 
countries—keeping in mind that, like other pension policies, design choices should antic-
ipate an evolution of conditions over the long run and be designed accordingly.

16.  IMPLEMENTATION ISSUES IN LOW- AND MIDDLE-INCOME COUNTRIES 
313
Challenges
There is a vast literature on the determinants of informal economic activity (see, for exam-
ple, Loayza 1997). The very definition of this term poses a challenge (Godfrey 2011). 
The challenge of formalizing the labor force and economic activity is generally multidi-
mensional and requires a comprehensive approach. It is a subject well beyond the scope 
of social insurance policy. Nevertheless, changes to contribution- or payroll tax–based 
social programs can be important in influencing participation in the formal sector. Some 
analysis shows that the incentives that these programs create can matter, especially at the 
margin (see Auerbach, Genoni, and Pagés 2007).
For decades, governments and their social insurance agencies attempted to expand 
coverage of pensions through the traditional social insurance model. This model relied on 
the imposition of participation mandates and payroll tax collection in economies domi-
nated by agriculture and microbusinesses. In many cases, the legal mandate did not apply 
to large segments of the workforce—and, even where it did apply, evasion ensured that 
coverage levels remained low. Pension coverage in the social insurance sense was deter-
mined by the size of the formal sector, which itself was a function of many other factors 
outside the purview of the relevant agencies.
In low-income countries, roughly 1 in 10 workers contributes to a pension scheme; 
the figure is about 1 in 3 in middle-income countries, depending on the definition 
applied. Transition economies have higher coverage, thanks largely to the legacy of the old 
systems, in which the state was the main employer, although coverage is falling in many 
of these countries as the role of the state is reduced and the labor force becomes increasing 
informal.
1
TRANSACTION COSTS
By definition, informal sector workers are not captured by traditional social insurance or 
pension mandates. The new MDC initiatives in low-income countries recognize that par-
ticipation will necessarily be voluntary, which has several implications. Notably, the trans-
action cost of participating in a voluntary system becomes very important. Solutions that 
reduce these costs must be explored wherever possible; a design that allows for flexibility 
in contribution levels for workers with volatile income sources is preferable.
Recent experience in India demonstrates two possible ways of reducing transaction 
costs. The first is to harness the infrastructure set up for the formal sector as much as pos-
sible. This includes systems of recordkeeping and account maintenance as well as manage-
ment of funds. Regulation and supervision, where they overlap, can also be utilized. In 
the case of India, discussed in chapter 12, the defined contribution scheme established 
for civil servants provided much of the infrastructure for the system later opened up to 
informal sector workers. Use of this infrastructure reduced the marginal cost of expanding 
coverage (some elements specific to the informal sector scheme—in particular, a contribu-
tion collection system accessible by these workers—did have to be added).
The volume-oriented approach can also be used with government programs, partic-
ularly those that have targeted a population deemed to merit subsidy. In Mexico in 2006, 
a proposal by former president Vicente Fox to match the contributions of working-age 
members of households covered by the Oportunidades program was conceived but never 

314 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
implemented. The idea had the advantage of harnessing the existing recordkeeping and 
payments infrastructure. Using a targeted program can also help constrain the fiscal cost 
and, in principle, increase the progressivity of the subsidy, as discussed below.
Another cost reduction strategy used in India has been to harness the existing infra-
structure of various groups, ranging from dairy worker associations to microfinance agen-
cies. These groups tend to have registries and periodic interactions with their members 
to which an additional transaction can be added at relatively low cost. By plugging in 
the group records and contribution flows to the central system, costs can be drastically 
reduced.
PORTABILITY AND ACCESS TO FORMAL FINANCIAL SECTOR
Many informal sector workers have variable career and earnings paths. Farmers may 
migrate to urban areas and take up informal sector employment; people remaining in 
rural areas may have seasonal incomes, migrate during the course of the year, or both. The 
self-employed may become informal or formal sector employees. The transient nature of 
informal sector work makes ensuring that pension contributions are portable even more 
important than for formal sector pension schemes.
Informal sector workers also tend to have low exposure to the formal financial sec-
tor. Thus, using traditional venues such as banks or insurance companies to collect con-
tributions may not be an effective mechanism in this regard. In many settings, initiatives 
such as village-level banking correspondents or outreach to certain groups of workers have 
been put in place to improve financial inclusion. Technology that leapfrogs traditional 
modalities may also be useful in connecting people with little knowledge of or interaction 
with formal financial institutions. For example, the Kenyan Mbao pension scheme com-
bines technology and financial inclusion (using mobile phones for contribution collection 
and account balance checking) with efforts to harness existing groups (members of the 
informal workers association). The same scheme allows for extremely small contributions 
to be made cheaply (ISSA 2011).
OUTREACH
In addition to technology, which requires up-front investments from government or the 
private sector, outreach to remote areas and illiterate populations involves information 
and education campaigns and direct contact with communities that may not have access 
to mainstream forms of communication. Whether implemented by the public or private 
sector, this outreach costs money. In China, local governments tasked with implement-
ing the rural pension scheme bear this cost. In India, nongovernmental organizations and 
other entities licensed as “aggregators” implement the program (see chapter 12).
FINANCIAL NEEDS AND RESOURCES
Most low-income workers have low saving capacity and a high preference and need for 
liquidity and precautionary savings. In low-income countries where the informal sector is 
the largest, there will be significant variation in the potential for long-term saving across 
the income distribution. Nevertheless, at the bottom of the distribution, many house-
holds will be operating at, or just above, subsistence; it is not practical to expect them 

16.  IMPLEMENTATION ISSUES IN LOW- AND MIDDLE-INCOME COUNTRIES 
315
to forgo access to their savings in the face of income uncertainty. In contrast, in middle-
income countries, even programs targeting the poor may include households with some 
saving capacity; this was the assumption of the Mexican proposal cited above. 
Other characteristics that affect take-up include coverage of other insurance pro-
grams. Unlike formal sector workers, few informal sector workers have health or other 
kinds of insurance. The fact that they must self-insure against these risks reduces their 
ability to save for the long run. The potential link between attempts to expand pension 
coverage and health insurance coverage is not often exploited.
INCENTIVES
The last challenge implies the need for robust incentives to offset high discount rates. The 
key question for policy makers is the elasticity of participation for the target population. 
One study in Peru asked individuals about their willingness to contribute if contributions 
were subsidized. The results, shown in figure 16.1, are intuitive. Workers at all income lev-
els were more willing to contribute when offered larger subsidies, but the marginal impact 
of doubling the subsidy level was less than proportional. If behavior were to follow these 
stated intentions, the impact of the subsidies on participation would be huge, increasing 
coverage in the bottom 40 percent of the income distribution from about 10 percent to 
more than 60 percent with a 1:1 match. Unfortunately, answers to questions about will-
ingness are not necessarily good predictors of actual outcomes, and there does not appear 
to be empirical evidence on the actual take-up elasticity in developing countries.
2
FIGURE 16.1  Effect of subsidy on likelihood of participating in pension scheme in Peru, by income 
quintile
5.7% 
15.2% 
28.2% 
36.0% 
50.3% 
36.8% 
34.3% 
27.2% 
29.6% 
21.6% 
22.7% 
19.1% 
18.2% 
15.9% 
13.3% 
34.8% 
31.5% 
26.5% 
18.5%
14.9% 
0 
10 
20 
30 
40 
50 
60 
70 
80 
90 
100 
Q1  
Q2 
Q3 
Q4 
Q5 
does not contribute
if half subsidy 
if quarter subsidy 
contributing 
%
SOURCE: Pagés 2012.
NOTE: Q1 is the lowest income quintile, and Q5 is the highest.

316 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Enabling Conditions
There are at least four key enabling conditions that are necessary to make MDC plans 
work:
 
• Capacity for informed choice within the target population. If individuals can-
not understand the implications of the matching provisions and make choices in 
an effective and consistent manner in response to the incentives, the flexibility of 
defined contribution is potentially a problem rather than a benefit.
 
• Administrative systems. Administrative systems consistent with the require-
ments for tracking individual contributions, maintaining records of account 
balances, and responding to individual choices are essential. If systems cannot 
handle the complexity of choice or can only implement additional choices and 
matching programs expensively, the benefits of defined contribution and match-
ing programs will be significantly diluted.
 
• Reliable governance. Reliable governance is needed to ensure appropriately 
designed contribution choices and investment options and to provide system 
oversight. If the choices are not appropriate, individuals are not likely to respond 
to the incentives in the manner intended. If management systems and processes 
are inadequate, confidence in the system will be impossible to sustain. Both of 
these conditions will require a well-functioning and reliable governance structure, 
including ongoing oversight and evaluation to adapt the system to changing con-
ditions. As with any pension system, the governance process will need to be suf-
ficiently insulated from political pressures to ensure that assets are protected and 
incentives are not manipulated to achieve short-term political objectives. 
 
• Sustainable commitment. A sustainable commitment to the system requires a 
reasonably stable political environment and design features that produce consis-
tent and predictable outcomes over time. This latter is particularly important, as 
pension arrangements involves long lead times between contributions and ben-
efits. In the absence of sustainability, a system will not be able to deliver secure 
retirement income. Any matching system that achieves meaningful coverage will 
inevitably entail a significant fiscal investment if public funds or tax incentives are 
included in the design. Consistency of governance and political support will be 
required to sustain public confidence.
Several structural conditions also must be present to enable an MDC system to 
function. These conditions are required for any defined contribution pension system, but 
are essential to the operation of an MDC-type system:
 
• Long-term asset classes consistent with the needs of pension investments. 
The membership of any successful matching system will be highly heteroge-
neous with regard to their investment requirements. The target groups will typi-
cally be diverse in age and have higher levels of volatility in earnings, liquidity 
preferences, and other risk management attributes. Addressing these needs will 
require underlying assets that range from short-term, highly liquid instruments 
to longer-term asset classes that can exploit time-related risk premiums. While 

16.  IMPLEMENTATION ISSUES IN LOW- AND MIDDLE-INCOME COUNTRIES 
317
matching provisions can induce short-term contributions, achieving the long-
term outcome of secure retirement savings requires the same range of longer-term 
financial products as in any pension system. Inducing contributions without the 
underlying investment products would likely result in an expensive system that 
produces little advantage in terms of retirement savings.
 
• Well-supervised financial markets and institutions. A perception of reliability 
and long-term confidence in the financial system will be essential to the success of 
any matching system. This perception can only be achieved with the appropriate 
products and effective regulation and supervision—especially in developing and 
transitional economies. Without long-term confidence, members are likely to 
make contributions in response to incentives and then withdraw them as quickly 
as possible, severely diminishing the capacity of the system to achieve long-term 
objectives and threatening the continuity of political support. The financial sys-
tem in the country needs to be sufficiently developed to handle the funds flowing 
into the defined contribution system, and needs to be thought of much more 
broadly than just in terms of asset management infrastructure. Intermediaries 
such as financial advisers can play a critical role in improving and facilitating 
appropriate choice. The presence of reliable third-party oversight to undertake 
audits and a system of information disclosure and individual account statements 
are essential for a system to succeed.
 
• Instruments and policy framework to convert accumulated savings into 
reliable income. Establishing a matching contribution arrangement as a pen-
sion system rather than simply as an inducement to increase savings in general 
will require a means of converting account balances to a stream of retirement 
income. Ideally, this would take the form of an efficient annuity market that 
would enable members to manage mortality risk while continuing to accrue 
some benefits from their accumulated savings. However, this ideal is extremely 
difficult to attain in many settings; alternatives include a policy framework and 
related instruments for programmed withdrawals or mandates for minimal or 
deferred annuity purchases where markets are less efficient. The essential condi-
tion is that there be an explicit framework to ensure that a significant portion 
of the savings induced by the matching arrangement be translated into old-age 
income of some sort.
Finally, certain elements are important for implementation of the MDC system. 
These may be listed as the “seven Cs”: