A review of international experience

weighted average pension wealth, multiple of economywide average earnings
SOURCE: OECD 2011.

32 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Coverage of Private Pensions
In 14 of the 34 OECD countries, private pensions are mandatory or quasi-mandatory 
(figure 2.5). Occupational plans are compulsory in Finland, Iceland, Norway, and Swit-
zerland; they cover 70–80 percent of the working-age population. Quasi-mandatory occu-
pational plans have very high coverage rates in Denmark, the Netherlands, and Sweden, 
as a result of collective bargaining agreements at the sectoral or national level. Mandatory 
personal pensions are in place in eight OECD countries 
Voluntary schemes can be optional in two senses. For occupational plans, employ-
ers are free to choose to establish a plan or not. Once a plan is established, it is possible 
for employers to make the plan a compulsory part of the employment contract in some 
cases. In some countries, employees either must be offered a choice as to whether to join 
(the United Kingdom and the United States, for example) or this is common practice. For 
personal plans, participation is entirely up to individuals.
Measuring coverage from administrative data can be difficult, because individuals 
can be members of both occupational and personal voluntary pension plans. Therefore 
total voluntary pension plan coverage cannot be obtained by summing occupational and 
personal coverage data.
3
The coverage of voluntary pension plans (both occupational and personal) exceeds 
50 percent in the Czech Republic, Germany, New Zealand, Norway, and the United 
Kingdom. It is very low in a range of countries at the bottom of the chart, such as Greece, 
Luxembourg, Mexico, Portugal, and Turkey.
Figure 2.6 examines some of the patterns in private pension coverage. Coverage 
increases with earnings (figure 2.6a) and is hump-shaped with respect to age (figure 2.6b). 
FIGURE 2.4  Impact of pension reforms on lifetime retirement income benefits in selected OECD 
countries 
−70 −60 −50 −40 −30 −20 −10  0 
Mexico 
Korea, Rep. 
Portugal 
Turkey 
Italy 
Slovak Republic 
Finland 
France 
Sweden 
Germany 
New Zealand 
Japan 
Austria 
Czech Republic 
United Kingdom 
Poland 
% change
SOURCE: OECD 2009; see Whitehouse and others 2009 for more details. 

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
33
FIGURE 2.5  Private pension coverage in selected OECD countries, 2009 
0  10 20 30 40 50 60 70 80 90 
Greece 
Portugal 
France 
Mexico 
Korea, Rep. 
Austria 
Ireland 
Chile 
New Zealand 
Poland 
Czech Republic 
Germany 
Australia 
Denmark 
Finland 
Iceland 
voluntary 
mandatory 
% coverage
Sweden 
Switzerland 
Netherlands 
Estonia 
Norway 
United Kingdom 
Canada 
United States 
Belgium 
Slovak Republic 
Spain 
Hungary 
Italy 
Turkey 
Luzembourg 
SOURCE: OECD 2011. 
NOTE: Data for Australia, Belgium, Canada, and Switzerland are for 2008. Information for Hungary was adjusted to refl ect 
the closure of nearly all mandatory individual accounts.
FIGURE 2.6  Relationship between private pension coverage and age and earnings in selected 
countries 
b. Earnings 
1 2 3 4 5 6 7 8 9 10 
decile of earnings distribution 
0 
10 
20 
30 
40 
50 
60 
70 
80 
20–24 25–34 35–44 45–54 55–64 
% of total employment 
age
0 
10 
20 
30 
40 
50 
60 
70 
80 
90 
100 
% of total employment 
Australia 
Canada 
Finland 
Germany 
Ireland 
Norway 
United
Kingdom 
United
States
a. Age
SOURCE: OECD 2011. 

34 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
In most countries, it peaks at prime working ages (35–44 or 45–54, depending on the coun-
try). Exceptions are the United States and the two countries with low overall coverage of 
voluntary plans (Finland and Norway), where membership continues to increase with age. 
The relationship between private pension coverage and both age and earnings is 
similar in Canada, Ireland, the United Kingdom, and the United States. Germany is a sig-
nificant outlier in both respects, achieving much higher rates of coverage of younger and 
lower-income workers than other countries do. One explanation for this phenomenon 
might be that English-speaking countries have public pension systems that redistribute 
from high earners to low earners, through basic pension schemes (Canada and Ireland), 
a redistributive formula in an earnings-related scheme, or both (the United Kingdom). 
Means-tested benefits are a significant component of retirement incomes, received by a 
quarter to a third of pensioners in all countries except the United States. As a result, there 
is a much smaller “pension gap” for low earners than for high earners in these countries: 
replacement rates for low earners are high relative to richer workers. This is not the case in 
Germany (see Antolín and Whitehouse 2009 for a more detailed analysis of the pension 
gap and voluntary retirement savings).
Financial Incentives 
A standard policy for encouraging private voluntary retirement saving is the granting of 
preferential tax treatment to pension plans. The idea is that a higher net rate of return on 
savings will encourage people to save more. These tax incentives tend to come with condi-
tions, usually governing the duration of saving and restrictions on the way benefits can be 
withdrawn. It is such conditions that qualify these as “retirement savings.”
Matching contributions are another type of financial incentive. The difference 
between a tax incentive and a matching contribution can be simply a matter of nomencla-
ture: whether government support is labeled a “tax” or a “match” can be irrelevant to the 
financial flows involved. There can, however, be significant structural differences between 
the two that affect the financial flows. Tax relief may be provided at individuals’ marginal 
rates. A sizable group of people of working age does not pay income tax and does not 
generally benefit from tax relief unless it is provided in the form of “nonwasteable” what 
are sometimes call “refundable” tax credits (that is, relief granted to both nontaxpayers and 
taxpayers). Similarly, fiscal incentives will tend to benefit people who pay tax at the high-
est marginal rates more than those paying lower or standard rates.
IMPACT OF FISCAL INCENTIVES
Economic theory is ambiguous regarding the effects of tax incentives for savings on 
individual behavior, even in a simple world in which the only motivation for saving is to 
provide for retirement.
4
 The various effects include the following:
 
• Because pensions are typically illiquid, they increase overall savings for house-
holds that face binding borrowing constraints (Hubbard and Skinner 1996).
 
• Tax incentives increase the net rate of return on pensions relative to other kinds 
of saving. The income effect of the higher return reduces saving; the substitution 
effect increases saving.

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
35
 
• Pensions are an annuitized form of saving. By providing longevity insurance, they 
should both increase welfare and reduce saving to protect against an uncertain life 
span (Hubbard and Judd 1987).
 
• Pensions may induce earlier retirement, which will increase pension saving (Feld-
stein 1974). 
Do tax incentives increase coverage of private pension plans? The OECD has mea-
sured incentives to save in pension schemes by comparing the effective tax rate on pen-
sions with the effective rate applied to benchmark savings (typically the bank deposit rate) 
(Yoo and de Serres 2004a, 2004b).
5
 The scale of tax incentives is calculated as a percentage 
of contributions, measured in present value terms over a given time horizon. Because the 
calculation considers revenues foregone from deductible contributions and tax-free invest-
ment returns from the nontaxation of accrued income and takes account of revenues col-
lected when benefits are withdrawn, it can be said to be on a net basis. The calculations 
average over nine age groups with different investment horizons.
The results are shown in figures 2.7 and 2.8. Figure 2.7 separates the tax treatment 
of pension savings at three stages: when contributions are made, as investment returns 
accumulate, and when benefits are withdrawn. Relative to benchmark saving, the size 
of the tax incentive for investing in private pensions varies significantly across countries, 
ranging from about zero in Mexico and New Zealand to nearly 40 percent of contribu-
tions in the Czech Republic. Most countries provide incentives of at least 10 percent of 
contributions; the average is more than 20 percent.
The incentive is naturally highest in countries that exempt pension contributions from 
tax and lower in countries, such as Sweden and Italy, that tax investment returns of pension 
plans. However, given that the net tax cost reflects the generosity of tax treatment of private 
pension savings relative to alternative nonpension savings vehicles, there is no systematic 
pattern (figure 2.8). In fact, several countries that exempt private pension contributions and 
investment returns (Greece, Iceland, Mexico, Poland, Korea, the Netherlands, and the Slo-
vak Republic) also provide generous tax breaks for alternative savings vehicles. Hence, there 
is only a small incentive to save in the form of pensions relative to benchmarks.
Figure 2.9 explores whether coverage of voluntary private plans is higher in coun-
tries with more generous tax incentives. The link between the two is very weak and a 
possible causality can go both ways. There is a cluster of countries at the left of figure 
2.9 with relatively low coverage of private pensions and another group at the right with 
much higher coverage. But the level of tax incentives is not very different between the two 
groups: 24 percent in the left-hand cluster and 28 percent in the right-hand cluster. A 
simple regression shows that coverage increases somewhat with the size of the tax incentive 
but that the effect is far from significant.
This simple cross-country analysis fails to capture many of the nuances of sin-
gle-country studies, which are able to address two policy issues: 
 
• Are contributions to a tax-privileged pension plan “new saving,” or are they sim-
ply diverted from other savings vehicles? 
 
• If additional household saving is motivated by the tax break, does it exceed the 
revenues foregone from the tax incentive? 

36 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
FIGURE 2.7  Tax treatment of investment returns, private pension contributions, and withdrawals 
in selected OECD countries 
−40
−20
0
20
40
60
relative effective tax rate (%) 
contributions
investment returns
withdrawals
total
Mexico 
New Zealand 
Finland 
Korea, Rep. 
United States 
Turkey 
Canada 
Japan 
Netherlands 
Great Britain 
Ireland 
Denmark 
Norway 
Portugal 
Greece 
Sweden 
Spain 
Italy 
Germany 
Czech Republic 
Poland 
Luxembourg 
Belgium 
Slovak Republic 
Hungary 
Australia 
France 
Austria
Switzerland
Iceland
SOURCES: Yoo and de Serres 2004a, 2004b. 
FIGURE 2.8  Tax treatment of benchmark savings and private pensions in selected OECD countries
0
10
20
30
40
benchmark saving
private pensions
effective tax rate (%) 
Mexico 
New Zealand 
Finland 
Korea, Rep. 
United States 
Turkey 
Canada 
Japan 
Netherlands 
Great Britain 
Ireland 
Denmark 
Norway 
Portugal 
Greece 
Sweden 
Spain 
Italy 
Germany 
Czech Republic 
Poland 
Luxembourg 
Belgium 
Slovak Republic 
Hungary 
France 
50
Australia 
Austria 
Switzerland 
Iceland 
SOURCES: Yoo and de Serres 2004a, 2004b. 

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
37
The impact of pension plans and tax incentives on savings is the subject of a long-
running debate, dating back to the 1960s. Much of the literature concerns the United 
States, particularly individual retirement accounts (IRAs) and 401(k)s, a type of employer-
based defined contribution plan (discussed in detail in chapter 3). Figure 2.10 summarizes 
the results of these studies.
Some researchers (for example, Poterba, Venti, and Wise on the United States and 
Venti and Wise 1995 on Canada) find a large, significant effect of tax-favored retirement 
saving schemes in terms of increasing net saving. Other researchers report little or no net 
effect, merely diversion of savings from less tax-favored instruments (see, for example, 
Milligan 2002 on Canada; Attanasio and Banks 1998 and Attanasio, Banks, and Wake-
field 2004 on the United Kingdom; and Engen, Gale, and Scholtz 1994, 1996 on the 
United States). 
In principle, it should be possible to gauge the effect of tax incentives on saving by 
comparing the total savings of individuals who contribute to such schemes with the sav-
ings of individuals who do not contribute. But this approach is valid only if eligibility is 
exogenous to the propensity to save. In practice, higher savings of people participating in 
tax-favored savings plans could reflect only their greater underlying preference for saving 
rather than a genuine net increase in overall savings. Different approaches to control-
ling for heterogeneity in saving preferences (and other unobservable characteristics) are an 
important factor behind the breadth of the range of empirical results. 
Venti and Wise (1990, 1991) compare assets of households with contributors to 
IRAs with assets of noncontributors, controlling for initial wealth. They conclude that 
most IRA contributions are new saving. However, two individuals with the same initial 
FIGURE 2.9  Correlation between coverage of voluntary private pensions and tax incentives for 
private pensions relative to benchmark savings
0
10
20
30
40
0
20
40
60
80
Australia
Austria
Belgium
Canada
Czech Republic
Finland
France
Germany
Greece
Hungary
Ireland
Italy
Korea, Rep.
Luxembourg
Mexico
New Zealand
Norway
Poland
Portugal
Slovak Republic
Spain
Turkey
United Kingdom 
United States 
coverage = 7.78 + 0.897 × tax incentive
(11.7)   (0.473)
R
2
 = 0.195
coverage
(% of working-age population, 2009) 
effective tax rate on private pensions
relative to benchmark savings (%)
SOURCES: Data on tax incentives are from Yoo and de Serres 2004a, 2004b; data on coverage are from OECD 2009.

38 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
wealth do not necessarily have the same underlying preferences toward saving. Gale and 
Scholz (1994) allow saving behavior to vary according to whether the individual is an 
IRA contributor or not, assuming different marginal propensities to save in IRAs and 
other savings vehicles. They then identify the impact of IRAs on saving by looking at 
the effect of a change in the IRA contribution limit, distinguishing between contributors 
who reached and contributors who did not reach the established ceilings. They conclude 
that a negligible fraction of contributions to IRAs represents new saving. However, their 
approach does not eliminate the possibility of inferring incorrectly that IRA saving dis-
places other forms of saving (Bernheim 1999). Moreover, their results are highly sensitive 
to small changes in the sample chosen for the analysis (by income level) (see Poterba, 
Venti, and Wise 1996a, 1996b). 
Attanasio and DeLeire (2002) exploit the idea that correlations between IRA sav-
ing and non-IRA saving can be particularly informative in the case of new contributors. 
They compare consumption growth in households that recently opened an IRA with 
growth in households that had already contributed to an IRA. They find that households 
finance IRA contributions not from a reduction in consumption but rather from existing 
or planned saving. They estimate that 9–20 percent of total IRA contributions were new 
saving.
Poterba, Venti, and Wise (1995, 1996a, 1996b) compare the financial assets of 
households eligible for 401(k)s with the assets of households without access to a 401(k) 
plan. They find little substitution between saving in 401(k)s and other financial assets. 
They also find that 401(k) eligibility correlates significantly with financial wealth. They 
therefore conclude that virtually all contributions to 401(k)s represent new saving.
FIGURE 2.10  Percentage of IRA and 401(k) saving that is new saving
0
25
50
75
100
New saving,
% of total contributions
100% 
(Venti
 and Wise 
1990, 1991)
0 
(Gale and Scholz 1994)
10–20% 
(Attanasio and DeLeire 2002)
35–40% 
(Attanasio, Banks, and Wakefield 2004)
75–100%
(Poterba, 
Venti, and 
Wise 1995, 
1996)
0–10% 
(Engen, Gale, and Scholz 1994, 1996)
−10% 
(Engelhardt 2002)
20% 
(Benjamin 2003)
IRAs
401(k)s

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
39
Engen, Gale, and Scholz (1994, 1996) and others challenge the assumption of exo-
geneity of eligibility for a 401(k) that underlies these studies. They argue that employ-
ees with greater preference for saving tend to gravitate toward jobs with good retirement 
provision (see Allen, Clark, and McDermed 1993 and Even and Macpherson 2000 for 
evidence). Furthermore, employers may use 401(k) programs to attract employees with 
such tastes (because preference for 401(k) plans may correlate with other desirable charac-
teristics, such as loyalty and long-term planning). Alternatively, they may establish 401(k) 
plans to meet the preferences of existing employees. Engen, Gale, and Scholz conclude 
that only a negligible part of 401(k) contributions represents new saving.
A second difference between these studies is the range of assets considered. Poterba, 
Venti, and Wise find an upward shift in the relative financial assets of people eligible for 
a 401(k), concluding that all contributions are new savings. Engen, Gale, and Scholz 
include home equity (property value less mortgage) in their measure of assets, arguing 
that some 401(k) contributions might be financed by equity withdrawal, leaving total 
net wealth unchanged. By allowing for substitution between real (property) and financial 
assets, Engen, Gale, and Scholz find a much smaller effect of 401(k) eligibility on total 
wealth than Poterba, Venti, and Wise find for financial assets alone.
Caveats apply to all of these studies. First, all studies compare the wealth of different 
groups of workers. But outside factors, such as the roller coaster path of equity markets 
since the mid-1980s, have a huge impact on wealth at any point in time that is likely to 
dwarf the behavioral effects of the availability of tax-favored retirement savings plans. 
Similarly, Engen, Gale, and Scholz’s results are affected by changes in housing prices over 
time. Second, most studies are based on a time series of cross-section data. However, the 
composition of both eligible workers and plan participants is likely to change over time. 
For example, coverage of 401(k)s and IRAs expanded over time. If newly covered individ-
uals were less motivated savers than early adopters, there would be a spurious downward 
shift in participants’ wealth that would offset some of the behavioral effect of tax-favored 
retirement savings plans. Third, employers have played a major role in wealth and savings 
composition with the substitution of 401(k)s for other kinds of retirement income provi-
sion (Andrews 1992; Papke 1995, 1999).
The impact of 401(k) on saving probably lies somewhere between the two extreme 
conclusions of “no new saving” and “all new saving” (Hubbard and Skinner 1996), 
although a wide range of estimates is plausible. Börsch-Supan (2004) finds that the evi-
dence of new saving in European countries is even weaker than it is in the United States. 
More recent studies report different results for different kinds of people. Engen and 
Gale (2000) find that tax incentives for 401(k)s raise savings for low earners and low savers 
but have little effect on high earners and high savers. Benjamin (2003) finds that 401(k)s 
are more effective in increasing new saving by people in rented housing and households 
without IRAs than they are for homeowners and people with IRAs.
FISCAL COST OF TAX INCENTIVES
Tax incentives for pensions necessarily involve a fiscal cost in terms of revenues foregone. 
These costs have been quantified as “tax expenditures.” This expenditure concept of tax 
preferences, originally put forward by Surrey (1973), has been used in many OECD 
reports (1984, 1995, 2010) that look at tax breaks for a range of purposes. These reports 

40 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
review national reporting of tax expenditures. They include national calculations of the 
cost of tax incentives for private pensions.
6
The literature on tax expenditures provides a number of warnings. Tax expenditures 
should not be added up (because they interact but the calculations of their value are inde-
pendent), they should not be compared across countries (because they are calculated rela-
tive to different benchmarks), and they should not be compared with direct expenditures. 
The analysis that follows breaks two of these three rules and so comes with caveats.
Figure 2.11 presents information from the OECD’s Social Expenditure Database 
(SOCX) on tax expenditures for private pensions, which come from national authori-
ties. The bars show the proportion of national income in revenues foregone, a figure that 
ranges from close to zero in France and Italy to more than 2.5 percent in Australia. The 
numbers next to the bars show these tax expenditures as a proportion of direct expendi-
ture on public pensions and other benefits for older people. Tax incentives cost an average 
of 14 percent of direct pension spending; the figure is much higher in Australia (80 per-
cent) and in Canada and Iceland (about 50 percent). The share is also high in Ireland and 
the United Kingdom, where tax expenditures represent more than 1 percent of national 
income. 
To address their public finance difficulties, many OECD countries are embarking on 
fiscal consolidation. Some countries have already made changes to private pension taxa-
tion in order to reduce the revenues foregone through fiscal incentives. In other countries, 
FIGURE 2.11  Revenues foregone from tax incentives for private pensions in selected OECD 
countries as a percentage of GDP and a percentage of public expenditure on pensions, 2007
foregone revenues as % of GDP, 2007
0% 
0% 
1% 
1% 
1% 
1% 
2% 
3% 
14% 
2% 
3% 
8% 
13% 
14% 
8% 
14% 
8% 
54% 
23% 
34% 
48% 
80% 
0.0 0.5 1.0 1.5 2.0 2.5 3.0 
France 
Italy 
Portugal 
Finland 
Austria 
Czech Republic 
Belgium 
Slovak Republic 
Mexico 
Poland 
Spain 
Luxembourg 
Norway 
OECD
Japan 
United States 
Germany 
Iceland 
United Kingdom 
Ireland 
Canada 
Australia 
SOURCE: OECD 2011. 
NOTE: Percentages noted alongside the bars are the cost of tax incentives in relation to direct pension spending.

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
41
there is a lively debate about reform, such as reducing the ceilings on deductible pension 
contributions in Australia, Ireland, and the United Kingdom. In addition, Ireland is levy-
ing an annual tax of 0.6 percent of assets on pension funds for four years. Since its intro-
duction in 2007, New Zealand’s KiwiSaver plan (discussed in greater detail in chapter 5) 
has seen frequent changes in tax treatment, compulsory employer-matched contributions, 
and government contributions. The most recent changes have been to curtail financial 
incentives, which are costly, as the government provides 41 percent of the money going 
into accounts, according to the Retirement Commission (2010). Germany extended the 
financial incentives for Riester pensions that were due to expire at the end of 2008, but 
not without debate about their cost. As chapter 4 shows, 37 percent of contributions to 
these plans have come from state coffers.
Bucking this trend, Chile and Poland have expanded tax relief in recent years. In 
both cases, voluntary plans had failed to have much of an impact. Tax incentives were 
strengthened in order to increase take-up.
Mandating Contributions and Using Soft Compulsion 
Mandating contributions is an easy way to achieve both high coverage of private pensions 
and more uniform coverage across ages and incomes (figures 2.5 and 2.6). Two policy 
approaches to compulsion can be identified.
In countries such as Australia, Iceland, Norway, and Switzerland, voluntary private 
pensions historically had broad coverage (50 percent or more of employees). Governments 
made it mandatory for employers to organize and contribute to private pensions on their 
employees’ behalf. The mandatory level of pension provision was below the customary 
level that prevailed when private pensions were provided voluntarily, however.
A second policy has been to mandate private pension contributions as a substi-
tute for part of the public pension. Chile, Estonia, Hungary, Mexico, Poland, the Slovak 
Republic, and Sweden have all taken this approach, and the Czech Republic will do so in 
2013. In contrast, Hungary recently nationalized its private pension funds, and Poland 
has partially reversed its reform (OECD 2012).
The main arguments for compulsion are that it protects people from regretting not 
having saved enough for their retirement when they were younger and protects societies 
from having to pay for safety net benefits for people who did not provide for their own 
old age. Implementing this paternalistic approach is simple: it involves choosing a tar-
get replacement rate (which may or may not vary with earnings) and then ensuring that 
people reach that target through either public retirement income provision or mandatory 
private pension plans.
An important, but unresolved, question is whether compulsion is necessary. Are 
people myopic? Left to their own devices, will they fail to save enough for retirement?
One way to investigate this question is to exploit historical differences in the degree 
of mandatory pension provision. Comparing the outcomes for retirement incomes of 
today’s pensioners might show evidence of myopia in countries in which voluntary pen-
sion provision has long played a major role.
Figure 2.12 shows how pensioner incomes compared with the incomes of the 
population as a whole. The data are net incomes, adjusted for household size. There is 

42 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
significant bunching of countries with older people’s incomes of 75–85 percent of the 
population average, covering a dozen countries from Hungary to Finland; the OECD 
average is 82 percent. But there is no link between relative incomes and the type of pen-
sion system. Voluntary private pensions play an important role in Canada and the United 
States, where older people’s incomes are relatively high, as well as in Ireland and the United 
Kingdom, where they are well below average. The OECD (2001) describes this phenom-
enon as “convergent outcomes, divergent means.” These data provide some evidence that 
the myopia hypothesis does not hold.
Several arguments can be made against compulsion:
 
• Even if individuals are myopic, it does not mean that greater mandatory pension 
provision is always a good thing. Mandating retirement saving means choosing a 
target replacement rate. It is difficult but important to get this rate right, as losses 
of individual welfare from forcing people to oversave can be as great as losses from 
myopia and undersaving. Resources diverted to retirement savings, for example, 
might come at the expense of raising and educating children.
FIGURE 2.12  Pensioners’ incomes as a percentage of population income in selected OECD 
countries, mid-2000s
% of population income
Hungary
Japan
Iceland
Germany
France
Luxembourg
Mexico
0
25
50
75
100
Ireland
New Zealand
Denmark
Finland
Slovak Republic
Czech Republic
Portugal
Switzerland
Sweden
Italy
United States
Netherlands
Canada
Turkey
Poland
Austria
Korea, Rep.
Australia
United Kingdom
Belgium
Norway
Spain
Greece
SOURCES: OECD 2008, 2011. 

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
43
 
• Formal pension plans are not the only way people can and do save for retirement. 
People might want to invest in property or their own business. This perfectly 
rational behavior is not possible with large, mandatory saving through formal 
pension schemes.
 
• Mandatory contributions to pensions are often perceived as a tax, which is likely 
to discourage people from working.
 
• Providers of voluntary pension arrangements—especially occupational pension 
schemes—often oppose compulsion because it would crowd out existing plans. 
There is also the risk that existing provision is leveled down to the amount of the 
mandate.
In countries such as Denmark, the Netherlands, and Sweden, more than 85 percent 
of employers offer private pension plans, even though the plans are not mandatory. Cover-
age of this extent is achieved through industrial relations agreements in different sectors. 
Employers covered by the agreement must offer a pension plan, and their employees must 
join it (for this reason, these plans are called “quasi-mandatory” in OECD analyses). Cov-
erage of voluntary pension arrangements in Belgium and Germany has also edged upward 
in recent years, as a result of the establishment of industrywide pension plans. However, 
this model is difficult to export to other countries, where labor market and industrial rela-
tions structures are less amenable to achieving near-universal coverage of private pensions. 
Figure 2.13 compares coverage of voluntary private pensions with the simulated 
pension entitlements of full-career workers. These entitlements are shown relative to 
economywide average earnings. Taking a weighted average across the earnings distribu-
tion captures the redistributive features of the pension system (which reduce the need for 
low earners to save for retirement) and the impact of ceilings (which increase the need for 
high earners to save). The modeling includes all mandatory components of the retirement 
income system, including compulsory private pensions where appropriate. 
The relationship between voluntary private pension coverage and mandatory retire-
ment income provision is negative and statistically significant. Coverage is high in Can-
ada, Germany, the United Kingdom, and the United States, where public pensions are (or 
will be for today’s workers) relatively low. In Greece and Luxembourg, by contrast, public 
benefits are high: few workers need private pensions as a supplement, and very few have 
them. Comparing figure 2.13 with the analysis of tax treatment in figure 2.9 shows that 
the “space” left by the mandatory retirement income system appears to have much more 
impact on the extent of coverage than do fiscal incentives. Figure 2.13 is consistent with 
the finding of “convergent outcomes, divergent means” in the discussion of figure 2.12.
Compulsion has disadvantages, especially the risk of forced oversaving for retire-
ment in formal pension plans. But purely voluntary pension provision runs the risk of 
undersaving. Automatic enrollment offers a third way between compulsion and volun-
tarism. It is often therefore called “soft compulsion.” 
The idea is that people have to opt out of saving for retirement rather than opt in. 
Surveys of financial literacy, such as the OECD’s (2005), routinely find that people agree 
that saving for retirement is important and that they feel they should be planning for old 
age. Unfortunately, these beliefs often fail to translate into action: inertia and procrastina-
tion predominate.
7
 An obvious reason for this inertia is that the process of signing up for 

44 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
a pension plan can be long and complex. Indeed, many people say that retirement plan-
ning is “more stressful than going to the dentist” (OECD 2005). Automatic enrollment is 
designed to capture such people and turn them into retirement savers.
Two countries have introduced automatic enrollment (with an opt-out clause) into 
private pension plans at the national level. The results have been mixed. New Zealand 
achieved a coverage rate of 43 percent with its KiwiSaver scheme. This figure understates 
the impact, as people are subject to automatic enrollment only when entering the labor 
force or when they change jobs (the government has discussed extending this provision 
to existing employees at some point). About a third of automatic enrollees have chosen 
to opt out. In Italy, severance pay is automatically paid into an occupational pension plan 
unless employees explicitly choose to retain severance pay. Although this provision has 
been in place for people working in companies with 50 employees or more since 2007, 
only 13 percent of the working-age population is covered by a voluntary pension plan in 
Italy. The United Kingdom’s new automatic enrollment scheme is being rolled out over six 
years beginning in October 2012.
Other countries have discussed adopting such an approach. The Retirement Secu-
rity Project in the United States has developed a bipartisan proposal for a national scheme 
with automatic enrollment (Iwry and John 2007). Both the current and previous govern-
ments in Ireland have backed this approach (Department of Social and Family Affairs 
2007). Germany has also considered introducing automatic enrollment for the salary con-
version (Entgeltumwandlung) (Leinert 2004, 2005). One reason for the prominence of 
this policy has been the rapid development of the discipline of behavioral economics, 
which has quickly achieved influence among policy makers.
FIGURE 2.13  Coverage of voluntary private pensions compared with tax incentives for private 
pensions relative to benchmark savings in selected OECD countries 
0
20
40
60
80
100
0
80
Australia
Austria
Belgium
Canada
Czech Republic
Finland
France
Germany
Greece
Hungary
Ireland
Italy
Korea, Rep.
Luxembourg
Mexico
New 
Zealand
Norway
Poland
Portugal
Slovak Republic
Spain
Turkey
United 
Kingdom
United States
coverage = 62.5 – 0.6703 × mandatory pension
(11.7)   (0.473)
R
2
 = 0.259
20
40
60
coverage
(% of working-age population, 2009) 
weighted average pension level as % of economy
SOURCES: Weighted average pension level data are from OECD 2011; for data on coverage, see fi gure 2.5 and its notes. 
NOTE: The weighted average pension level is the simulated pension relative to economywide average earnings for full-
career workers, with the weights refl ecting the national earnings distribution.

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
45
A number of employer-provided pension plans in the United Kingdom and the 
United States have long used automatic enrollment to increase coverage among their 
employees. The evidence from the United States is discussed in detail in chapters 3 and 
15.
This chapter therefore discusses some of the evidence from the United Kingdom. 
According to the Government Actuary (2006), 48 percent of occupational plans auto-
matically enrolled all new employees in 2005, and another 12 percent applied automatic 
enrollment to some employees. These figures represent a modest increase over 1995, when 
43 percent of employees were in plans that automatically enrolled everyone and 7 percent 
in plans with some automatic enrollment. However, there are some definitional questions 
about what constitutes automatic enrollment. The Department for Work and Pensions 
distinguishes four enrollment procedures. The department’s survey found that 44 per-
cent used a process of “streamlined joining,” which required the employee only to sign 
a completed form (McKay 2006). Only 19 percent of employees were covered by plans 
with full automatic enrollment—that is, plans that require an active opt-out. As in the 
United States, both of these enrollment procedures were more common among larger 
employers. Traditional opt-in accounted for 19 percent of plans, weighted by the number 
of members.
Most of the evidence on the effectiveness of automatic enrollment in the United 
Kingdom is based on case studies of a handful of schemes. Horack and Wood (2005) 
examine 11 company pension schemes in the United Kingdom that changed their enroll-
ment arrangements. Two firms that introduced automatic enrollment increased cover-
age (one from 25 percent to 58 percent, the other from 45 percent to 62 percent). The 
other two firms already had very high coverage rates (86 percent and 88 percent), most 
likely because the schemes did not require employee contributions. Automatic enrollment 
increased these shares to 92 percent at one firm and 100 percent at the other.
Hawksworth (2006) reports the most dramatic increase in coverage from automatic 
enrollment—from 15 percent to 100 percent—in the Building and Civil Engineering 
scheme. The Government Actuary’s survey of occupational plans in the United Kingdom 
finds coverage of 89 percent in plans that automatically enroll everyone, 73 percent in 
plans in which some employees are enrolled, and 59 percent in plans without automatic 
enrollment. The survey carried out for the Department of Work and Pensions finds cov-
erage of 41 percent in traditional opt-in plans and 60 percent in plans with automatic 
enrollment (McKay 2006). These figures relate to larger employers (firms with more than 
20 employees). Among smaller employers, coverage was virtually the same with traditional 
and automatic enrollment (at 67 percent).
The studies of the United Kingdom and United States suggest an important effect 
of automatic enrollment on coverage of private pensions. However, it can be difficult to 
disentangle the impact from other features of the pension plan, such as the scale of the 
required employee contribution and the amount the employer is willing to contribute.
In New Zealand, for example, coverage of private pensions was relatively low before 
the KiwiSaver reform: at about 20 percent, it was substantially lower than in other coun-
tries with similar pension systems (see figure 2.13). Two-thirds of KiwiSaver members 
actively chose to sign up, according to the Inland Revenue, with the other third automati-
cally enrolled. About half went directly to a provider to enroll; another 14 percent did so 

46 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
through their employer. These findings suggest that the main factor driving the high level 
of coverage of the new scheme was the financial incentive, comprising both government 
contributions and tax relief.
Conclusion
Private pensions provide about 20 percent of retirement income on average in OECD 
countries. In most countries, this share 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.
This volume is concerned principally with financial incentives for private pensions. 
However, these incentives should be seen in the broader context of other policy options, 
such as automatic enrollment. Both the World Bank and the OECD are exploring the 
issues of financial literacy and capabilities, along with the types of programs that can 
improve them. In principle, greater awareness of the need to save for retirement might 
expand coverage of and increase contribution rates to private pensions; better knowledge 
of how to save for retirement and facilitation of access to private pension plans, particu-
larly through the workplace, might make it easier to save for retirement. In practice, policy 
initiatives in this area—such as making it easier for smaller employers to set up occupa-
tional plans in the United States and the requirement for employers in the United King-
dom to offer stakeholder pensions—have not led to significant expansions of coverage.
OECD countries have traditionally provided tax relief on pension contributions at 
individuals’ marginal tax rates. This practice has been questioned on a number of grounds. 
First, estimates suggest that 80 percent of the value of tax relief in Ireland accrues to the 
richest quintile of the income distribution. Figures for the United Kingdom show that a 
quarter of the tax expenditure goes to the wealthiest 1.5 percent. Second, tax incentives 
have not proved effective at expanding coverage among low earners or younger workers. 
Riester pensions in Germany and KiwiSaver in New Zealand involve government match-
ing and flat-rate contributions. In Germany at least, these mechanisms have been effective 
in engaging hard-to-reach groups. Both KiwiSaver and the new automatic enrollment 
scheme in the United Kingdom include compulsory employer matching contributions as 
an additional financial incentive.
Regarding automatic enrollment, more evidence is needed to determine how effec-
tive it is in extending coverage of private pensions. Longer-term data are needed to assess 
the degree of persistence in pension coverage. For example, over time, workers may over-
come their inertia in the opposite direction and realize that opting out is a quick way of 
increasing current income. Moreover, schemes with automatic enrollment involve sizable 
subsidies to individual savings. All occupational plans in the United Kingdom and the 
United States include employer contributions of varying sizes. Care is therefore needed to 
isolate a “pure” automatic enrollment effect on coverage from the effect of the subsidies. 
The automatic enrollment approach to extending coverage of private pensions is 
likely to spread. Survey evidence suggests that automatic enrollment is much more popu-
lar with individuals than compulsion in the United Kingdom (Bunt and others 2006; 
Hall, Pettigrew, and Harvey 2006). And voters’ views are shared by many politicians, 

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
47
who worry that workers will view mandatory contributions to private pensions as an 
unwelcome tax on their earnings. If soft compulsion fails to deliver a sustained increase 
in private pension coverage, governments adopting this approach must keep the policy of 
compulsion in reserve.
Notes
1.  Data for Switzerland are not shown, because capital (mainly private pensions) and work 
incomes are aggregated in the database. Together they account for 52 percent of older people’s 
income on average, with the 48 percent residual coming from public transfers.
2.  Figures for the Czech Republic do not take account of the new mandatory defined contribu-
tion scheme, which will be introduced in 2013 at the earliest.
3.  Using Canada as an example, 34 percent of the working-age population is enrolled in occu-
pational plans, and 35 percent has personal pensions. Overall voluntary pension coverage in 
Canada is only 53 percent, however, because 48 percent of people with occupational pension 
plans also have personal plans.
4.  This discussion draws on Antolín and Lopez Ponton (2007). 
5.  An earlier but more comprehensive study of household saving is by the OECD (1994). White-
house (1999) presents results pertaining to pensions.
6.  Other OECD studies—Yoo and de Serres (2004a, 2004b) and Antolín, de Serres, and de la 
Maisonneuve (2004)—calculate their own estimates of the revenue effect of the tax treatment 
of private pensions.
7.  See chapter 15 of this volume and the references therein.
References
Allen, S. G., R. L. Clark, and A. A. McDermed. 1993. “Pensions, Bonding, and Lifetime Jobs.” 
Journal of Human Resources 28: 463–81.
Andrews, Emily. 1992. “The Growth and Distribution of 401(k) Plans.” In Trends in Pensions 
1992, ed. John Turner and Daniel Beller, 149–76. Washington, DC: U.S. Government Print-
ing Office.
Antolín, P., A. de Serres, and C. de la Maisonneuve. 2004. “Long-Term Budgetary Implications of 
Tax-Favoured Retirement Saving Plans.” Working Paper 393, Organisation for Economic Co-
operation and Development, Economics Department, Paris.
Antolín, P., and E. R. Whitehouse. 2009. “Filling the Pension Gap: Coverage and Value of Volun-
tary Retirement Savings. Social, Employment and Migration Working Paper 9, Organisation 
for Economic Co-operation and Development, Paris. 
Antolín, P., and E. Lopez Ponton. 2007. “The Impact of Tax Incentives on Retirement Savings: 
A Literature Review.” In OECD/IOPS, Global Private Pension Conference Proceedings. Paris: 
Organisation for Economic Co-operation and Development.
Attanasio, O. and J. Banks. 1998. “Trends in Household Saving Don’t Justify Tax Incentives to 
Boost Saving.” Economic Policy 13 (27): 547–83. 
Attanasio, O., J. Banks, and M. Wakefield. 2004. “Effectiveness of Tax Incentives to Boost (Retire-
ment) Saving: Theoretical Motivation and Empirical Evidence.” OECD Economic Studies 
39: 145–72.

48 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Attanasio, O., and T. DeLeire. 2002. “The Effect of Individual Retirement Accounts on House-
hold Consumption and National Saving.” Economic Journal 112: 594–38.
Benjamin, D. J. 2003. “Does 401(k) Eligibility Increase Saving? Evidence from Propensity Score 
Subclassification.” Journal of Public Economics 87 (5–6): 1259–90.
Bernheim, B. D. 1999. “Taxation and Saving.” NBER Working Paper 7061, National Bureau of 
Economic Research, Cambridge, MA.
Börsch-Supan, A. 2004. “Mind the Gap: The Effectiveness of Incentives to Boost Retirement Sav-
ing in Europe.” OECD Economic Studies 39: 111–44.
Bunt, K., L. Adams, Z. Koroglu, and E. O’Donnell. 2006. “Pensions and Pension Reform.” 
Research Report 357, Department for Work and Pensions, London. 
Department of Social and Family Affairs. 2007. “Green Paper on Pensions.” Stationery Office, 
Dublin. 
Disney, R. F., and E. R. Whitehouse. 2001. “Cross-Country Comparisons of Pensioners’ Incomes.” 
Research Report 142, Department of Work and Pensions, London.
—. 2003. “The Economic Well-Being of Older People in International Perspective: A Criti-
cal Review.” In Economic Outcomes in Later Life, ed. S. Crystal and D. Shea. Annual Review of 
Gerontology and Geriatrics 22. New York: Springer Publishing.
Engelhardt, G. V. 2002. “Have 401(k)s Raised Household Saving? Evidence from the Health and 
Retirement Study.” Working Paper 2002-023, Department of Economics, Syracuse Univer-
sity, Syracuse, NY.
Engen, E. M., and W. G. Gale. 2000. “The Effects of 401(k) Plans on Household Wealth: Differ-
ences Across Earnings Groups.” NBER Working Paper 8032, National Bureau of Economic 
Research, Cambridge, MA.
Engen, E. M., W. G. Gale, and J. K. Scholz. 1994. “Do Saving Incentives Work?” Brookings Papers 
on Economic Activity 1: 85–151.
—. 1996. “The Illusory Effect of Saving Incentives on Saving.” Journal of Economic Perspec-
tives 10 (4): 113–38.
Even, W. E., and D. A. Macpherson. 2000. “The Changing Distribution of Pension Coverage.” 
Industrial Relations 39 (2): 199–227.
Feldstein, M. S. 1974. “Social Security, Induced Retirement and Aggregate Capital Accumulation.” 
Journal of Political Economy 82 (5): 905–26. 
Förster, M., and M. Mira d’Ercole. 2005. “Income Distribution and Poverty in OECD Countries 
in the Second Half of the 1990s.” Social, Employment and Migration Working Paper 22, 
Organisation for Economic Co-operation and Development, Paris.
Gale, W. G., and J. K. Scholz. 1994. “IRAs and Household Saving.” American Economic Review 84 
(December): 1233–60.
Government Actuary’s Department. 2006. Occupational Pension Schemes 2005: The Thirteenth Sur-
vey by the Government Actuary. London: Government Actuary’s Department.
Hall, S., N. Pettigrew, and P. Harvey. 2006. “Public Attitudes to Personal Accounts: Report of a 
Qualitative Study.” Research Report 370, Department for Work and Pensions, London.
Hawksworth, J. 2006. “Review of Research Relevant to Assessing the Impact of the Proposed 
National Pension Savings Scheme on Household Savings.” Research Report 373, Department 
for Work and Pensions, London.

2.  POLICIES TO ENCOURAGE PRIVATE PENSION SAVINGS: EVIDENCE FROM OECD COUNTRIES 
49
Horack, S., and A. Wood. 2005. “An Evaluation of Scheme Joining Techniques in Workplace Pen-
sion Schemes with an Employer Contribution.” Research Report 292, Department for Work 
and Pensions, London.
Hubbard, R. G., and K. L. Judd. 1987. “Social Security and Individual Welfare: Precautionary Sav-
ing, Borrowing Constraints and the Payroll Tax.” American Economic Review 77 (4): 630–46. 
Hubbard, R. G., and J. S. Skinner. 1996. “Assessing the Effectiveness of Saving Incentives.” Journal 
of Economic Perspectives 10 (4): 73–90. 
Iwry, J. M., and D. C. John. 2007. “Pursuing Universal Retirement Security through Automatic 
IRAs.” Policy Brief 2007-02, Retirement Security Project, Washington, DC. 
Leinert, J. 2004. “Automatische Entgeltumwandlung: Hohe Teilnahmequoten ohne Zwang.” Wirt-
schaftsdienst 2004-2.
—. 2005. Betriebliche Altersvorsorge: Automatik statt Zwang. Warum das opting-out Modell bes-
ser ist. Deutsches Institut für Altersvorsorge, Köln, Germany.
McKay, S. 2006. “Employers’ Pension Provision Survey 2005.” Research Report 329, Department 
for Work and Pensions, London.
Milligan, K. 2002. “Tax-Preferred Savings Accounts and Marginal Tax Rates: Evidence on RRSP 
Participation.” Canadian Journal of Economics/Revue canadienne d’économique 35 (3): 436.
OECD (Organisation for Economic Co-operation and Development). 1984. Tax Expenditures: A 
Review of Issues and Country Practices. Paris: OECD.
—. 1994. Taxation and Household Saving. Paris: OECD.
—. 1995. Tax Expenditures: Recent Experiences. Paris: OECD.
—. 2001. Ageing and Income: Financial Resources and Retirement in Nine OECD Countries. 
Paris: OECD.
—. 2005. Improving Financial Literacy: Analysis of Issues and Policies. Paris: OECD.
—. 2008. Growing Unequal? Income Distribution and Poverty in OECD Countries. Paris: 
OECD.
—. 2009. Pensions at a Glance: Retirement-Income Systems in OECD Countries. Paris: OECD. 
—. 2010. Tax Expenditures in OECD Countries. Paris: OECD. 
—. 2011. Pensions at a Glance: Retirement-Income Systems in OECD and G20 Countries. Paris: 
OECD. 
—. 2012. Pensions Outlook. Paris: OECD. 
Papke, L. E. 1995. “Does 401(k) Introduction Affect Defined Benefit Plans?” National Tax Associa-
tion Proceedings: 173–77.
—. 1999. “Are 401(k) Plans Replacing Other Employer-Provided Pensions? Evidence from 
Panel Data.” Journal of Human Resources 34: 346–68.
Poterba, J. M., S. F. Venti, and D. A. Wise. 1994a. “Targeted Retirement Saving and the Net 
Worth of Elderly Americans.” American Economic Review 84: 180–85.
—. 1994b. “401(k) Plans and Tax-Deferred Saving.” In Studies in the Economics of Aging, 
ed. D. A. Wise. Chicago: University of Chicago Press for the National Bureau of Economic 
Research.
—. 1995. “Do 401(k) Contributions Crowd Out Other Personal Saving?” Journal of Public 
Economics 58: 1–32.

50 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
—. 1996a. “How Retirement Saving Programs Increase Saving.” Journal of Economic Perspec-
tives 10 (4).
—. 1996b. “Personal Retirement Saving Programs and Asset Accumulation: Reconciling the 
Evidence.” NBER Working Paper 5599, National Bureau of Economic Research, Cambridge, 
MA.
—. 1998a. “Personal Retirement Saving Programs and Asset Accumulation: Reconciling the 
Evidence.” In Frontiers in the Economics of Aging, ed. D. A. Wise. Chicago: University of Chi-
cago Press for the National Bureau of Economic Research.
—. 1998b. “Implications of Rising Personal Retirement Saving.” In Frontiers in the Econom-
ics of Aging, ed. D. A. Wise. Chicago: University of Chicago Press for the National Bureau of 
Economic Research.
—. 1998c. “Lump-Sum Distributions from Retirement Saving Plans: Receipt and Utiliza-
tion.” In Inquiries in the Economics of Aging, ed. D. A. Wise. Chicago: University of Chicago 
Press for the National Bureau of Economic Research.
—. 2001. “The Transition to Personal Accounts and Increasing Retirement Wealth: Macro 
and Micro Evidence.” NBER Working Paper 8610, National Bureau of Economic Research, 
Cambridge, MA.
Retirement Commission. 2010. Review of Retirement-Income Policy. Wellington: Retirement 
Commission. 
Surrey, S. S. 1973. Pathways to Tax Reform: The Concept of Tax Expenditures. Cambridge, MA: 
Harvard University Press.
Venti, S. F., and D. A. Wise. 1990. “Have IRAs Increased US Saving? Evidence from Consumer 
Expenditure Surveys.” Quarterly Journal of Economics 105 (3): 661–98.
—. 1991. “The Saving Effect of Tax-Deferred Retirement Accounts: Evidence from SIPP.” In 
National Saving and Economic Performance, ed. B. D. Bernheim and J. B. Shoven. Chicago: 
University of Chicago Press.
—. 1995. “RRSPs and Saving in Canada.” National Bureau of Economic Research, Cam-
bridge, MA.
Whitehouse, E. R. 1999. “The Tax Treatment of Funded Pensions.” Pension Reform Primer Series, 
Social Protection Discussion Paper 9910, World Bank, Washington, DC.
Whitehouse, E. R., A. C. D’Addio, R. Chomik, and A. Reilly. 2009. “Two Decades of Pension 
Reform: What Has Been Achieved and What Remains to Be Done?” Geneva Papers on Risk 
and Insurance 34: 515–35.
Yoo, K. Y., and A. de Serres. 2004a. “Tax Treatment of Private Pension Savings in OECD Coun-
tries.” OECD Economic Studies 39 (2): 73–110. 
—. 2004b. “Tax Treatment of Private Pension Savings in OECD Countries and the Net Tax 
Cost per Unit of Contribution to Tax-Favoured Schemes.” Working Paper 406, Organisation 
for Economic Co-operation and Development, Economics Department, Paris. 

PART II
High-Income
Country Experience

53
CHAPTER 3
Matching Contributions in 
401(k) Plans in the United States
Nevin Adams, Dallas Salisbury, and Jack VanDerhei
Matching contributions have long been used in the United States to encourage lower-
income workers to participate in defined contribution pension plans at levels necessary to 
ensure that these programs comply with certain nondiscrimination tests mandated by the 
Internal Revenue Service, the U.S. tax authority. A review of the literature suggests that for 
voluntary 401(k) plans, employer matches have a positive impact on plan participation 
and that automatic enrollment in a matched plan encourages a level of participation that 
is adequate to produce sufficient retirement income for people with a full savings career in 
that system. Whether more widespread adoption of automatic enrollment will affect the 
perceived need for, and level of, the employer match in providing a future with retirement 
income security remains to be seen.
D
efined contribution plans—and matching employer contributions—were fixtures on 
the U.S. retirement scene well before the advent of 401(k) plans in the early 1980s.
1 
Then, as now, matching contributions were used as incentives to encourage participa-
tion by less highly compensated workers, as a means of ensuring that these programs—
typically defined contribution stock bonus/thrift savings plans—were able to remain 
in compliance with nondiscrimination tests. These tests, designed to ensure that a dis-
proportionate share of the benefits do not accrue to the more highly paid members of 
the plan, must be fulfilled in order for the contributions and investment earnings of an 
employer-sponsored pension plan to receive favorable treatment under U.S. income tax 
laws (in which contributions and earnings on investments are excluded from the taxable 
income of the plan’s members until they are distributed). These employer contributions, 
which matched worker contributions in rates specified in plan documents, provided a 
financial incentive for saving by lower-income workers in the plan at participation and 
contribution levels that would enable the more highly paid group to make the level of 
contributions it desired, up to the limit permitted under the tax laws.
Industry studies and other surveys continue to find a high degree of correlation 
between the level of the employer match and voluntary contributions by employees. Of 
six plan features listed in the 2005 Retirement Confidence Survey, the feature most cited 
by respondents who did not participate in their employer’s 401(k)–type plan as likely 
to encourage them to join was a generous employer match of up to 5 percent of salary.
2
 
Three in 10 respondents reported that they would be much more likely to contribute to 
their employer’s plan if this option were available, and 4 in 10 reported that they would 
be somewhat more likely to do so. However, an employer match would not persuade 
all nonparticipants to contribute. The 2006 Retirement Confidence Survey found that 

54 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
87 percent of employees offered a plan with a match reported participating, compared 
with 70 percent without a match.
More recently, a 2010 survey by the Principal Financial Group found that the design 
of the employer match can be a powerful motivator in boosting the amount of money 
participants put into their 401(k) retirement accounts, even when the employer’s total con-
tribution does not change (Principal Financial Group 2010). A survey by Fidelity Invest-
ments finds that 92 percent of participants surveyed indicated that one of the main reasons 
they participate is to take advantage of company contributions (67 percent cited this factor 
as very important and 25 percent as somewhat important); about a quarter (23 percent) 
of surveyed workers who were not previously participating said that wanting to take full 
advantage of the company match was the reason for increasing their contributions when 
they did join. In contrast, the Principal Financial Well-Being Index for the second quarter 
of 2010 notes that half of employees participating in the survey report that the deciding 
factor in determining how much to contribute to the plan was how much they could afford.
The growing interest in, and adoption of, automatic enrollment plan designs—and 
the decision of a significant minority of employers to suspend their matching contribu-
tions in the wake of the 2008 financial crisis—continues to attract interest about the 
future role of the matching contribution in effective plan design.
This chapter explores the origins and emergence of the 401(k) plan as a dominant 
retirement savings vehicle in the United States, as well as the evolution and impact of 
certain design elements, notably the application of an employer matching contribution, in 
influencing individual saving behaviors. It is organized as follows: the first section explores 
the origins and development of the 401(k) as a dominant defined contribution plan struc-
ture in the United States, as well as the shifting legislative and regulatory environment that 
attended its rapid growth. The second section outlines the factors affecting the level and 
timing of matching contributions during the emergence and maturity of the 401(k). The 
third section examines the research on the link between matching practices and worker 
responses. The fourth section explores the response of participant-savers to changes in the 
rate of matching contributions. It also looks at future trends in plan design, notably auto-
matic enrollment, and their impact on the rate and prevalence of matching contributions. 
The last section summarizes the main findings.
Drivers and Outcomes
Defined contribution plans are retirement plans that specify the level of contributions by 
both employer and employee and in which the contributions are placed into individual 
employee accounts. Among the plan types included in this category in the United States 
are money purchase plans, in which employer contributions are mandatory and are usually 
stated as a percentage of employee salary; profit-sharing plans, in which total contributions 
to be distributed are often derived from a portion of company profits; stock bonus plans, 
which are similar to profit-sharing plans but usually make contributions and benefit pay-
ments in the form of company stock; savings and thrift plans, in which employees may 
contribute a predetermined portion of earnings, all or part of which the employer matches; 
and employee stock ownership plans (ESOPs), in which the employer contributes a desig-
nated amount into a fund that is generally invested primarily in company stock.

3.  MATCHING CONTRIBUTIONS IN 401(K) PLANS IN THE UNITED STATES 
55
The 401(k) is a type of defined contribution plan, generally a subset of what would 
technically be considered a saving and thrift plan. It did not emerge until several years 
after the passage of the Employee Retirement Income Security Act (ERISA), the federal 
legislation that in 1974 codified the rules and reporting structures for employer-sponsored 
private sector retirement plans in the United States.
3
EARLY DESIGNS
In pre–401(k) thrift/savings plans, employees could contribute up to 10 percent of their 
income on an after-tax basis. These employee contributions were treated as the deposit 
of compensation already received and in the control of the worker and thus subject to 
income taxation. However, as an incentive to encourage employee participation in these 
programs, employers were allowed to make a tax-deductible contribution to employee 
accounts. Whereas employee contributions were made on an after-tax basis, taxation to 
the individual worker on the value of the employer matching contributions credited to an 
individual participant’s account was (and is) deferred until distribution, as were any result-
ing investment gains in the individual account. The deferral of tax obligation on these 
amounts by the individual participant was predicated on deposit to a qualified pension 
fund held as a trust account, subject to certain restrictions.
A less common plan design at the time was a cash or deferred arrangement 
(CODA). In such an arrangement, the individual participant was essentially given 
the choice of receiving a form of cash compensation (generally some kind of year-end 
bonus payment) or deferring that payment until some future point in time. In legal 
terms, the employer could either provide a specified amount to the employee in the 
form of cash or some other taxable benefit, or contribute an amount to a trust or pro-
vide an accrual or other benefit to be received (and therefore subject to tax) at some 
point in the future. The underlying operating principle of taxation deferral is that the 
employee must decide to defer receipt before he or she has actual receipt, or control, of 
the compensation (in regulatory parlance, before the taxable benefit is “currently avail-
able” to the employee).
These programs allowed eligible employees to either take cash now (generally some 
kind of annual bonus or profit-sharing payment) or defer compensation by having it 
deposited into a retirement plan. However, questions arose over time regarding the elec-
tive nature of these contributions—namely, whether an employee who has the option to 
decide whether to take the cash is in “constructive receipt” of the funds and thus liable for 
current taxation of the contribution.
Through a series of court cases and tax rulings, by the early 1970s, the U.S. tax 
authority (the Internal Revenue Service [IRS]) had established that a CODA could be