A review of international experience

% of compensation
100%, up to 5% 
75% up to 6% 
50% up to 6% 
75% up to 2%, 50% for 3%–5% 
0% 
SOURCE: VanDerhei and Copeland 2001.

68 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
HOW IMPORTANT IS THE SHIFT FROM VOLUNTARY ENROLLMENT TO 
AUTOMATIC ENROLLMENT IN 401(K) PLANS?
The previous sections summarized the importance of employer matching contributions 
on participation and contribution decisions in 401(k) plans. But the real issue, certainly 
in terms of public policy research, is whether these decisions result in adequate retirement 
wealth accumulation. It is difficult to accurately simulate these results, given the complex 
interactions between plan design parameters and the participation, contribution, invest-
ment, and cash-out decisions of employees. Fortunately, the EBRI/ICI 401(k) database 
includes a detailed administrative record of tens of millions of employees from tens of 
thousands of plans, with longitudinal linkages for both plans and participants going back 
as far as 1996 in some cases.
In the first simulation publication using these data, Holden and VanDerhei 
(2002) provide an analysis of workers participating in 401(k) plans with voluntary 
enrollment. They show that workers participating in these voluntary enrollment 401(k) 
plans may be able to generate account balances (either retained in the original plan or 
rolled over to a successor employer plan or an IRA) capable of replacing a significant 
percentage of their pre-retirement income if they participate in a 401(k) plan their 
entire working life.
However, if workers are not always eligible to participate in a 401(k) plan, the ade-
quacy of the retirement income generated by these plans drops substantially (Holden and 
VanDerhei 2002). Several studies simulating the likely retirement income generated under 
these plans for all (or a significant portion of ) the workforce also find evidence that sig-
nificant numbers of future retirees could reach retirement age with account balances too 
small (even when combined with expected social security income) to achieve replacement 
rates within conventionally defined targets (70–85 percent of pre-retirement income); this 
literature is reviewed in Appendix II of GAO 2007.
Another simulation study (Holden and VanDerhei 2005) demonstrates how valu-
able automatic enrollment designs would be to savings accumulations of low-income 
401(k) contributors. A year after this study was conducted, the Pension Protection Act of 
2006 included several advantageous provisions for 401(k) sponsors with a specific type of 
automatic enrollment design.
VanDerhei (2010) simulates the likely impact of this change on future 401(k) accu-
mulations for a significant portion of workers (not just current 401(k) participants or 
people eligible to participate) if the switch to automatic enrollment is made. It updates 
previous EBRI research (VanDerhei and Copeland 2008) by using actual plan design 
modifications of large plan sponsors to estimate the employer response. The analysis indi-
cates that the adoption of automatic enrollment is likely to have a very significant impact 
on retirement savings for many workers, especially low-income workers.
For example, under the baseline set of assumptions used, the median 401(k) accu-
mulations for the lowest-income quartile of workers currently age 25–29 (assuming all 
401(k) plans were voluntary enrollment plans as typified by the large plan sponsors used 
in the analysis) would be only 0.08 times final earnings at age 65 (this is largely due to the 
fact that a sizable percentage of workers—as opposed to participants—were assumed to 
have zero balances at age 65).

3.  MATCHING CONTRIBUTIONS IN 401(K) PLANS IN THE UNITED STATES 
69
If, however, all 401(k) plans by large plan sponsors are assumed to have automatic 
enrollment provisions in place, the median 401(k) accumulations for the lowest income 
quartile jumps to 4.96 times final earnings, assuming that 401(k) participants revert back 
to the default contribution (3 percent of pay) when they change jobs and participate in 
a new 401(k) plan. Under the assumption that the participants retain their current con-
tribution level when they change jobs, the median 401(k) accumulations for the lowest 
income quartile jumps to 5.33 times final earnings.
15
 Even among the top 25 percent of 
these workers (when ranked by 401(k) accumulations as a multiple of final earnings), 
there are large increases: the multiple under a voluntary enrollment scenario is 2.41 times 
final earnings whereas automatic enrollment provides multiples of 9.15 or 9.81, depend-
ing on the assumptions for employee reversion to default contribution rates upon job 
change. 
IMPACT OF AUTOMATIC ENROLLMENT ON EMPLOYER CONTRIBUTION RATES
Soto and Butrica (2009) conclude that among a sample of large 401(k) plans, match 
rates are lower among firms with automatic enrollment than among firms without auto-
matic enrollment, after controlling for firm characteristics. There are, however, two major 
limitations of this analysis. First, the study was based on U.S. Department of Labor Form 
5500 data, which do not include specific information on 401(k) match rates. Rather, 
the authors constructed an estimate of the match rate as being the ratio of employer-
to-employee contributions for each 401(k) plan. Second, the authors merged the Form 
5500 data with information on automatic enrollment from the Pensions and Investments 
database of the top 1,000 pension funds, relying on a flag indicating whether plan admin-
istrators reported offering automatic enrollment in their plans. However, as this database 
does not report when the automatic enrollment provision was adopted, there is no way of 
knowing how long this design was in place.
The results of the regression analysis of this database suggest a negative relationship 
between automatic enrollment and match rates that is statistically significant at the firm 
level, according to the authors of the study. In particular, match rates are about 7 percent-
age points lower among firms with automatic enrollment than among firms without auto-
matic enrollment, after controlling for firm characteristics. The authors acknowledge that 
although the regressions suggest a relationship between automatic enrollment and match 
rates, they do not necessarily imply that automatic enrollment causes lower match rates—
a crucial qualification that has been ignored in many third-party accounts of their study. 
These conclusions conflict with previous EBRI research (VanDerhei 2007), which 
surveyed defined benefit plan sponsors administered by Mercer Human Resource Con-
sulting to gauge their recent activity and planned modifications to their defined ben-
efit (pension) and defined contribution (401(k)-type) plans. The survey also determined 
what, if any, increases in employer contributions to defined contribution plans were made 
in conjunction with reductions to the defined benefit plans.
Although the association between the adoption of automatic enrollment and 
employer contributions to 401(k) plans was not the focus of the study, it found that 
one-third of the defined benefit plan sponsors surveyed indicated that they had already 
increased or planned to increase their employer match to a defined contribution plan and 

70 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
that 20.9 percent indicated that they had already increased or planned to increase their 
nonmatching employer contributions to a defined contribution plan. There was some 
overlap between the two groups, but overall, 42.5 percent of the defined benefit plan 
sponsors surveyed indicated that they had already increased or planned to increase their 
employer match or nonmatching employer contribution to the defined contribution. This 
increase in contribution was particularly evident among defined benefit plan sponsors 
that had closed a defined benefit plan to new hires, frozen their defined benefit plan to all 
members in the last two years, or planned to do so in the next two years.
16
The 2007 EBRI study found an extremely high correlation between the adoption of 
automatic enrollment for a 401(k) plan and the freezing or closing of the defined benefit 
plan.
17
 Among defined benefit plan sponsors that had closed their defined benefit plans 
in the previous two years, 80.5 percent had either already adopted or were considering 
adopting automatic enrollment features for their 401(k) plans.
18
VanDerhei (2010) analyzes in detail plan-specific data on about 1,000 large defined 
contribution plans for salaried employees from Benefit SpecSelect (a trademark of Hewitt 
Associates LLC) in 2005 and 2009. A subsample of plan sponsors was created that had 
adopted automatic enrollment 401(k) plans by 2009 but did not have them in 2005 
(the last observation not influenced by the Pension Protection Act of 2006). For each 
plan, VanDerhei coded the default contribution rate for the automatic enrollment plan in 
2009, the match rate contribution formulas for both years,
19
 and all nonelective contribu-
tions paid to the defined contribution participants by the employer.
Whether plan sponsors were more or less generous after adopting automatic enroll-
ment was evaluated with three metrics:
 
• The average 2009 first-tier match rate was $0.8778 for each $1 contributed; the 
average 2005 first-tier match rate was $0.8126 for each $1 contributed. The dif-
ference of $0.0652 per $1 contributed suggests that, to the extent that this sam-
ple is representative of the universe of large 401(k) sponsors, sponsors adopting 
automatic enrollment were more generous to the 401(k) participants (when mea-
sured by this variable) after automatic enrollment was implemented than before.
 
• The average effective match rate was 4.32 percent of compensation in 2009 but 
only 4.00 percent in 2005. The increase of 0.32 percentage points suggests that 
large 401(k) sponsors adopting automatic enrollment were more generous to the 
401(k) participants (when measured by this variable) after the adoption of auto-
matic enrollment than before.
20
 
• The average total employer contribution rate (the sum of the effective match 
rate and the nonelective contribution rate) was 6.35 percent of compensation for 
2009 and 5.46 percent in 2005. The increase of 0.89 percentage points suggests 
that large 401(k) sponsors adopting automatic enrollment were more generous to 
the 401(k) participants when measured by this variable than before.
This information was then combined with the defined benefit information for 
the same sponsor in an attempt to analyze whether EBRI’s 2007 findings of the associa-
tion between defined benefit freezing/closing and enhanced 401(k) contributions were 
corroborated.

3.  MATCHING CONTRIBUTIONS IN 401(K) PLANS IN THE UNITED STATES 
71
The average improvements for all three metrics were much greater for sponsors 
that had frozen or closed their defined benefit plans than the overall average (table 3.4). 
For example, the change in the total employer contribution rate for all frozen plans was 
1.64 percent of compensation versus 0.89 percent for the overall average. Employers 
that had closed their defined benefit plans to new employees had an even larger average 
improvement (2.82 percent of compensation).
The defined benefit plan sponsors that had frozen or closed their plans were then 
separated into firms that had done so before adopting automatic enrollment and firms 
that had changed their defined benefit plans between 2005 and 2009. If the hypothesis 
that the 401(k) improvements were at least partially a result of a simultaneous quid pro 
quo for the decreased accruals in the defined benefit plan, one would expect that the ear-
lier modifications would be less generous than the modifications that took place at about 
the time of the conversion to automatic enrollment—exactly what is found for all six 
comparisons in table 3.4. For example, the average total employer contribution improve-
ment for firms that had frozen their plans before 2005 was 0.69 percent of compensation, 
compared with 2.45 percent for firms that froze their plans between 2005 and 2009. 
Similar evidence is found for firms that closed their pension plans to new employees: the 
average improvement in total employer 401(k) contribution was only 0.56 percent of 
compensation for firms that closed their plans before 2005, but it was 3.34 percent for 
firms that closed their plans between 2005 and 2009.
TABLE 3.4  Changes in employer contribution rates to 401(k) plans that adopted automatic 
enrollment between 2005 and 2009, by type of modification 
PERCENT
Modification to defined 
benefit plan
Change in first-tier 
match rate
Change in effective 
match rate
Change in total employer 
contribution rate
Frozen
14.68
0.73
1.64
Before 2005
10.29
0.24
0.69
2005–09
18.40
1.15
2.45
Closed to new employees 
15.06
0.58
2.82
Before 2005
5.56
0.22
0.56
2005–09
17.26
0.66
3.34
Total
6.52
0.32
0.89
SOURCE: EBRI analysis of plan-specifi c data from Benefi t SpecSelect (a trademark of Hewitt Associates LLC).
Persistence of Contributions
An important issue in understanding the behavior of workers contributing to 401(k) 
plans and interpreting the long-term implications is the persistence of behavior when 
the matching formula is changed. In the wake of the 2008 financial crisis, a number of 

72 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
employers reduced, suspended, or terminated their matching contributions, providing a 
natural experiment that sheds some light on this issue.
21
A Towers Watson analysis of 260 companies that made changes to employer match 
contributions in response to the economic crisis finds that 231 suspended their matches 
and 29 reduced them (Apte and McFarland 2011). According to the study, most of the 
companies chose to reinstate their match (75 percent). Of the firms that did so, 74 per-
cent reintroduced the original match amount. Among these plan sponsors, the most fre-
quent match formula before and after the crisis was 50 percent up to 6 percent of salary. 
For companies on which specific change dates were available, the median duration for 
match suspensions was 12 months, and most companies reinstated their match after 9–12 
months.
Separately, by the middle of 2009, almost 10 percent of defined contribution plans 
administered by Fidelity suspended or reduced their contribution dollars. By December 
2010, 55 percent of those plan sponsors indicated that they planned to reinstate their 
match within the next 12 months. Fidelity also reported that among larger companies 
(firms with more than 5,000 employees), 71 percent had reinstated or planned to reinstate 
their match. More than 60 percent of employers with a plan size of 500–999 employees 
had already reinstated or indicated they planned to reinstate their match, up from 38 per-
cent just 10 months earlier. Among employers with fewer than 1,000 employees in their 
plan, the percentage was 46 percent.
In February 2012, an interim report from the IRS of responses from its 401(k) 
compliance check questionnaire—which polled 1,200 randomly selected plan sponsors 
via a secure website— revealed that the share of firms that had suspended or discontin-
ued matching contributions in their plans increased from 1 percent in 2006 to 4 percent 
in 2008, the share that had suspended or discontinued the nonelective contribution in 
their plans increased from 2 percent in 2006 to 5 percent in 2008, and the share that 
had reduced nonelective contributions in their plans increased from 1 percent in 2006 to 
5 percent in 2008.
Given the impact of employer matching contributions on employee participation 
and contribution decisions (at least in voluntary enrollment plans) documented above, 
a common concern with respect to plan sponsors suspending their contributions is the 
potential impact on employee savings. For example, if an employee were contributing 
6 percent of compensation to receive the maximum match from a plan with a 50 percent 
match on the first 6 percent of compensation, would a suspended match end up decreas-
ing the total contribution for the employee from 9 percent to 6 percent? Or would the 
reduced incentive drive savings below 6 percent (perhaps to zero)?
In an attempt to provide preliminary evidence with respect to the impact of sus-
pending employer contributions on employee behavior, VanDerhei (2009) analyzed all 
401(k) plans in the EBRI/ICI 401(k) database with more than $100,000 in employer 
contributions in 2007 and none in 2008 (all plans were still active as of year end 2008).
22
 
The percentage of 401(k) participants continuing to contribute in 2008 after a suspension 
in employer contributions was analyzed as a function of a match rate proxy.
23
For all plans with a match rate proxy of less than 50 percent, the percentage of 
401(k) participants continuing to contribute in 2008 was at least 86 percent. However, 
the percentage among participants with more generous match rate proxies decreased 

3.  MATCHING CONTRIBUTIONS IN 401(K) PLANS IN THE UNITED STATES 
73
substantially. For participants with a match rate proxy of 50–100 percent, only 80 percent 
of participants continued contributing after the suspension. For participants with match 
rate proxies of more than 100 percent, the percentage was only 73 percent. 
Going Forward
In order to avoid the complications associated with nondiscrimination testing, a growing 
number of employers have embraced the safe harbor plan design outlined above. Rather 
than relying on a match incentive, this design requires certain minimum amounts of 
employer contribution (table 3.5).
Over time, the expanded availability of automatic enrollment may result in reduced 
reliance on the match as an incentive to participate (table 3.6).
From inception, the additional nondiscrimination tests imposed on the 401(k) 
design (and their subsequent iterations) have ensured that the benefits of these programs 
did not skew disproportionately to highly compensated workers by requiring certain min-
imum participation and contribution levels by NHCEs. Compliance with these rules has 
required that sponsoring employers be attentive to the participation of NHCEs, in order 
to preserve the qualification for tax preference, as well as the benefit levels for highly 
compensated employees. The employer match has long, and consistently, been promoted 
as a means of offering “free money” to workers who might otherwise have foregone par-
ticipation or done so at lower contribution levels. More recently, and following their 
strengthened codification in the Pension Protection Act of 2006, automatic enrollment 
plan designs have provided an effective means for plan sponsors to encourage or expand 
participation without necessarily tying participation to a financial incentive.
Automatic enrollment 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.
TABLE 3.6  Use of automatic enrollment in 
defined contribution plans, by company size
Company size
% offering automatic enrollment
Micro
19.9
Small
38.4
Midsize
50.3
Large
50.7
Mega
55.7
Total
33.4
SEE table 3.1.
TABLE 3.5  Inclusion of safe harbor plans in 
defined contribution plans, by company size
Company size
% offering safe harbor plans
Micro
44.6
Small
38.6
Midsize
37.5
Large
41.4
Mega
41.7
Total
41.5
SEE table 3.1.

74 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Conclusion
Matching contributions have played a central role in the emergence of 401(k) plans, the 
most common type of defined contribution pension savings account in the United States. 
Employers have used various kinds of matching arrangements to encourage broad-based 
participation as a means of ensuring that these plans remain in compliance with a complex 
set of nondiscrimination rules designed to ensure that the value of contributions into these 
arrangements does not unduly benefit higher-income workers at the sponsoring firms.
Several empirical studies analyze the effect of matching contributions on the prob-
ability of participating in 401(k) plans that use voluntary enrollment. The magnitude of 
the results varies considerably, but generally the employer match has been found to have a 
positive impact on overall employee participation in 401(k) plans, though not necessarily 
for all groups of employees.
In a study using the most comprehensive set of administrative data for this topic, 
Mitchell, Utkus and Yang (2005) conclude that nearly two-thirds of NHCEs at a typical 
firm would participate regardless of the presence of a match and that the impact of a match 
is limited to a relatively narrow range of outcomes. More than a quarter of the workers 
who did not fall within the group defined as highly compensated did not participate in 
plans that employed the most common matching formulas used in this sampling, and 
even with a more generous match, more than one in five still did not choose to participate.
The literature on the impact of matching contributions on employee contribution 
behavior for 401(k) plans with voluntary enrollment is much less developed than that deal-
ing with the participation decision—understandably, given the complexity of the model-
ing. VanDerhei and Copeland (2001) provide a framework for integrating the impact of 
the participant’s age and wage with plan-specific incentives that look at the incremental 
match rates and maximum amounts matched in voluntary enrollment designs.
A more recent design evolution is the application of automatic enrollment in 
these programs. It appears that a significant percentage of workers participating in these 
arrangements do so at the default contribution rates and adhere to the automatic escala-
tion design parameters established by the plans. A number of employers that have adopted 
automatic enrollment features for their 401(k) plans have adopted the simplified safe har-
bor matching provisions of the Pension Protection Act of 2006. Some plan sponsors may 
be concerned that the increase in participation rate that frequently accompanies automatic 
enrollment adoption could lead to a sudden increase in matching contribution costs. One 
way to mitigate a potential increase in employer costs is to decrease the generosity of the 
employer match provisions of the plan. It is too soon to determine whether employers 
have done so, especially given the considerable financial turmoil in recent years, which 
will make it difficult to disentangle the effects of the design from other factors.
A considerable knowledge gap remains with respect to the impact of an employer 
match on participation in 401(k) plans utilizing automatic enrollment. Based on a sam-
ple of nine firms, Beshears and others (2007) estimate that eliminating the match under 
an automatic enrollment plan would likely reduce plan participation by 5–11 percent-
age points. Public policy analysts will be keenly interested in seeing whether these results 
continue to hold as the number of plan sponsors adopting automatic enrollment provi-
sions continues to increase.
24
 Ultimately, whether matching contributions are an effective 

3.  MATCHING CONTRIBUTIONS IN 401(K) PLANS IN THE UNITED STATES 
75
means of closing the retirement income savings gap depends to a large extent on the type 
of plan design (automatic versus voluntary enrollment) chosen by the plan sponsor and 
the income level of participants.
25
Notes
1.  In the United States, the legal definition of a defined contribution plan is a plan that pro-
vides for an individual account for each participant and provides benefits based solely on the 
amount contributed to the account, plus or minus income, gains, and expenses and losses 
allocated to the account.
2.  The Retirement Confidence Survey is sponsored by the Employee Benefit Research Institute, 
the American Savings Education Council, and Mathew Greenwald & Associates.
3.  Additional information about these plan designs and their applications can be found in the 
Employment Benefit Research Institute’s Fundamentals of Employee Benefit Programs, avail-
able at http://www.ebri.org/publications/books/?fa=fundamentals. 
4.  The Tax Reform Act of 1986 also changed the law such that, effective for plan years begin-
ning on or after January 1, 1987, contributions under qualified profit-sharing plans no longer 
needed to be limited to current or accumulated profits of the employer. As a consequence, 
employers could maintain qualified CODAs as part of profit-sharing plans, regardless of 
whether they actually had profits.
5.  Currently, two nondiscrimination approaches are permitted for the average deferral percentage 
and average compensation percentage tests. The first is “current year” testing, where current 
year deferral and contribution percentages are used to compare the percentages of both highly 
compensated and non–highly compensated employees. The other approach is “prior year” 
testing, where the deferral and contribution percentages for NHCEs in the prior year are com-
pared with the deferral and contribution percentages of highly compensated employees in the 
current year. This approach means that the average deferral percentage and average compensa-
tion percentage limits for highly compensated employees are known in advance, reducing the 
chance of a failed test at year end or the need for taxable refunds or other corrective measures. 
The method used must, however, be specified in the plan document.
6.  A safe harbor is a statutory, regulatory, or contractual provision that provides protection, usu-
ally from a penalty or liability.
7.  Graded vesting schedules provide for a gradual increase in the ownership interest in employer 
contributions over time. Under a cliff vesting schedule, the worker’s ownership (vesting) of the 
employer contributions generally happens at a single point in time.
8.  The original matching contribution (25 percent on the first 4 percent of pay contributed) 
was replaced with an employer contribution equal to 4 percent of pay plus an annual profit-
sharing contribution.
9.  The impact of matching contributions on employee contribution behavior has been stud-
ied extensively in voluntary enrollment 401(k) plans. Relatively little research has been con-
ducted on automatic enrollment plans. Nessmith, Utkus, and Young (2007) provide evidence 
that new employees hired under automatic enrollment 401(k) plans have participation rates 
that are nearly twice those of new employees hired under voluntary enrollment 401(k) plans 
(86 percent versus 45 percent). However, they show that overall plan contribution rates under 
automatic enrollment fall, because many new participants who would have voluntarily chosen 
a higher contribution rate remain at the low default levels. For additional research conducted 

76 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
on a relatively small sample of 401(k) plans, see Choi, Laibson, and Madrian (2004); Choi 
and others (2004); and Madrian and Shea (2001).
10.  This result is from a regression on a sample of all participants (whether they contributed or 
not) for whom information on the match rate and level was provided or derived. The regres-
sion model included age, tenure, salary, plan loan provision (yes/no), the employer match rate, 
and employer match level variables to examine their effects on participant before-tax contribu-
tion rates.
11.  The EBRI/ICI 401(k) database includes detailed records (including demographic information 
and contribution behavior) on individual participants from more than 60,000 plans. Because 
of strict confidentiality standards, no information on the individual was included.
12.  This limitation affects elective deferrals to Section 401(k) plans and to the federal govern-
ment’s Thrift Savings Plan, among other plans. The limitation under Section 402(g)(1) on the 
exclusion for elective deferrals in 401(k) plans was $17,000 for 2012.
13.  Even for plans without nonelective contributions, several participants had employer contri-
butions that were not equal to the predicted amount based on the plan’s matching formu-
las, the employee’s before-tax or after-tax contributions, or both. This apparent discrepancy 
may reflect the use of different definitions of compensation applied by the individual 401(k) 
plan for these purposes versus the compensation data provided in the database. The authors 
attempted to control for this unknown effect by computing the difference between actual and 
predicted employer contributions (as a percentage of compensation) and excluding any par-
ticipant with more than a 0.2 percent of compensation differential.
14.  If the plan does not allow an employee to defer to the next level of contributions, that employee 
is not included in the next contribution interval. This feature allows for the examination of 
only those employees who can actually contribute the next percentage of compensation, based 
on either a prior choice or plan-specific constraints.
15.  Technically, the stochastic simulation model assumes that future 401(k) eligibility is a func-
tion of current eligibility.
16.  The proportion of defined benefit plan sponsors that indicated that they had already increased 
or planned to increase their employer match or nonmatching employer contribution to a 
defined contribution plan ranged from 62 percent (for sponsors that had frozen the defined 
benefit plan in the previous two years) to 81 percent (for sponsors that intended to close the 
plan for new members in the next two years).
17. As hypothesized in VanDerhei (2007), some employers that discontinued accruals in the 
defined benefit plans may want to continue to have a very large percentage of their eligible 
employees participating each year. As many industry studies have shown, participation rates 
among eligible young and low-income employees are significantly higher under 401(k) plans 
with an automatic enrollment feature.
18.  Similar levels applied to defined benefit plans that were to be closed or frozen in the next two 
years.
19. Many plans will use a multitier formula—another reason why using simple averages of 
employer-to-employee contributions, as Soto and Butrica (2009) do, is problematic.
20.  The effective match rate is a measure of the total amount of the employer’s contribution via 
the matching formulas if the employee contributes enough to receive the full match. This 
measure simultaneously controls for the match rate, the maximum amount matched, and the 
possibility of a multitiered formula. For example, an employer that matches 100 percent of the 
first 1 percent of compensation and 50 percent of the next 5 percent would have an effective 
match of 1 * 1 + (0.5 * 5) = 3.5 percent of compensation.

3.  MATCHING CONTRIBUTIONS IN 401(K) PLANS IN THE UNITED STATES 
77
21.  Some plan sponsors may have turned to their 401(k) plans as a means of freeing up cash 
flow for their legally required minimum contributions to defined benefit plans. A review by 
Salisbury and Buser (2009) of 251 plan sponsors that had suspended matching contributions 
for their 4.4 million workers finds that firms employing half of the workers also maintained 
an open defined benefit plan. Sixteen percent of workers were with employers that were still 
obligated to fund a frozen defined benefit plan, whereas 8 percent were with an employer that 
had both an open and a frozen defined benefit plan that carried funding obligations.
22.  More refined analysis is currently under way to link the 2006 and 2007 contributions on a 
plan-specific basis and filter out midyear suspensions.
23.  The proxy was aggregate employer contributions divided by employee contributions for 2007. 
This measure is only a rough proxy; it will be inaccurate to the extent that nonelective contri-
butions exist for the plan, or employees contribute in excess of the maximum amount needed 
to obtain the full match. This analysis is currently being refined using year-end 2010 data.
24.  Given that such a small percentage of 401(k) sponsors adopted automatic enrollment before 
passage of the Pension Protection Act of 2006, it is possible that their workforces may not be 
representative of the broader distributor of all eligible workers.
25. VanDerhei (2010) simulates median accumulation multiples for employees currently age 
25–29, who are assumed to have 31–40 years of eligibility throughout their working careers. 
He projects that the highest income quartile will have sufficient accumulations to purchase a 
real annuity to replace at least one-third of their final earnings at age 65. The impact of lower 
participation rates under voluntary enrollment plans becomes more pronounced as income 
decreases.
References
Adams, Nevin E. 2011. “Points of Hue: PLANSPONSOR’s 2011 Annual Defined Contribution 
Survey.” PLANSPONSOR Magazine November.
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 Printing Office.
Apte, Vishal, and Brendan McFarland. 2011. “A Look at Defined Contribution Match Rein-
statements.” Towers Watson, U.S.—Insider October. www.towerswatson.com/united-states/
newsletters/insider/5641.
Bassett, William, Michael Fleming, and Anthony Rodrigues. 1998. “How Workers Use 401(k) 
Plans: The Participation, Contribution, and Withdrawal Decisions.” National Tax Journal 51 
(2): 263–88.
Beshears, John, James J. Choi, David Laibson, and Brigitte C. Madrian. 2007. “The Impact of 
Employer Matching on Savings Plan Participation under Automatic Enrollment.” NBER 
Working Paper 13352, National Bureau of Economic Research, Cambridge, MA.
Brady, Peter J. 2007. “Pension Nondiscrimination Rules and the Incentive to Cross Subsidize 
Employees.” Journal of Pension Economics and Finance 6: 127–45.
Choi, James J., David Laibson, and Brigitte Madrian. 2004. “Plan Design and 401(k) Savings Out-
comes.” NBER Working Paper W10486, National Bureau of Economic Research, Cambridge, 
MA. 
Choi, James, David Laibson, Brigitte Madrian, and Andrew Metrick. 2004. “Defined Contribu-
tions Pensions: Plan Rules, Participant Decisions, and the Path of Least Resistance.” In Tax 
Policy and the Economy, ed. James M. Poterb, 67–113. Cambridge, MA: MIT Press.

78 
MATCHING CONTRIBUTIONS FOR PENSIONS: A REVIEW OF INTERNATIONAL EXPERIENCE
Dworak-Fisher, Keenan. 2011. “Encouraging Participation in 401(k) Plans: Reconsidering the 
Employer Match, Industrial Relations.” Journal of Economy and Society 50 (4): 713–37.
Engelhardt, Gary, and Anil Kumar. 2007. “Employer Matching and 401(k) Saving: Evidence from 
the Health and Retirement Study.” Journal of Public Economics 91 (10): 1920‒43. 
Even, William E., and David A. Macpherson. 2005. “The Effects of Employer Matching in 401(k) 
Plans.” Industrial Relations: A Journal of Economy and Society 44 (3): 525–49.
Facts from EBRI. 2005. “A History of 401(k) Plans: An Update.” Employee Benefit Research Insti-
tute, Washington, DC. http://www.ebri.org/pdf/publications/facts/0205fact.a.pdf.
GAO (Government Accountability Office). 2007. “Private Pensions: Low Defined Contribution 
Plan Savings May Pose Challenges to Retirement Security, Especially for Many Low-Income 
Workers,” GAO, Washington, DC. 
Holden, Sarah, and Jack VanDerhei. 2001. “Contribution Behavior of 401(k) Plan Participants.” 
EBRI Issue Brief 238, Employee Benefit Research Institute, Washington, DC. 
—
. 2002. “Can 401(k) Accumulations Generate Significant Income for Future Retirees?” 
EBRI Issue Brief 251, Employee Benefit Research Institute, Washington, DC.
—. 2005. “The Influence of Automatic Enrollment, Catch-Up, and IRA Contributions on 
401(k) Accumulations at Retirement.” EBRI Issue Brief 283, Employee Benefit Research 
Institute, Washington, DC.
Ippolito, Richard A. 1997. Pension Plans and Employee Performance.” Chicago: University of Chi-
cago Press.
Kusko, Andrea, James Poterba, and David Wilcox. 1998. “Employee Decisions with Respect to 
401(k) Plans.” In Living with Defined Contribution Pensions: Remaking Responsibility for Retire-
ment, ed. Olivia S. Mitchell and Sylvester Schieber, 69‒96. Philadelphia: University of Penn-
sylvania Press.
Madrian, Brigitte, and Dennis Shea. 2001. “The Power of Suggestion: Inertia in 401(k) Participa-
tion and Savings Behavior.” Quarterly Journal of Economics 116 (4): 1149‒87.
Mitchell, Olivia S., Stephen P. Utkus, and Tongxuan Yang. 2005. “Turning Workers into Savers? 
Incentives, Liquidity, and Choice in 401(k) Plan Design.” NBER Working Paper W11726, 
National Bureau of Economic Research, Cambridge, MA.
Nessmith, William E., Stephen P. Utkus, and Jean A. Young. 2007. “Measuring the Effectiveness of 
Automatic Enrollment.” Vanguard Center for Retirement Research, Malvern, PA. 
Papke, Leslie. 1995. “Participation in and Contributions to 401(k) Pension Plans: Evidence from 
Plan Data.” Journal of Human Resources 30 (2): 311‒25.
Papke, Leslie, and James Poterba. 1995. “Survey Evidence on Employer Match Rates and Employee 
Saving Behavior in 401(k) Plans.” Economics Letters 49 (September): 313–17. 
Principal Financial Group. 2010. “New Data from the Principal Reveals Power of Employer 
Match.” News Room: News Release Archive, November 30. www.principal.com/about/
news/2010/ris-match-stats113010.htm.
Raish, David L. n.d. “Cash or Deferred Arrangements (Portfolio 358).” Bloomberg/BNA. 
www.bna.com/Cash-Deferred-Arrangements-p7555/.
Salisbury, Dallas, and Elizabeth Buser. 2009. “Many 401(k) Sponsors Suspending Matching Con-
tributions Also Funding Defined Benefit Pension Plans.” EBRI Note 30 (6), Employee Benefit 
Research Institute, Washington, DC.

3.  MATCHING CONTRIBUTIONS IN 401(K) PLANS IN THE UNITED STATES 
79
Soto, Mauricio, and Barbara A. Butrica. 2009. “Will Automatic Enrollment Reduce Employer 
Contributions to 401(k) Plans?” CRR Working Paper 2009–33, Center for Retirement 
Research at Boston College, Boston.
VanDerhei, Jack. 2007. “Retirement Income Adequacy after PPA and FAS 158: Part One: Plan 
Sponsors’ Reactions.” EBRI Issue Brief 307, Employee Benefit Research Institute, Washing-
ton, DC.
—. 2009. “Falling Stocks: What Will Happen to Retirees’ Incomes? The Worker Perspective.” 
Presentation at the Association for Public Policy and Management Fall Conference, the Eco-
nomic Crisis of 2008: What Will Happen to Retirees’ Incomes?
—. 2010. “The Impact of Automatic Enrollment in 401(k) Plans on Future Retirement 
Accumulations: A Simulation Study Based on Plan Design Modifications of Large Plan Spon-
sors.” EBRI Issue Brief 341, Employee Benefit Research Institute, Washington, DC.
VanDerhei, Jack, and Craig Copeland. 2001. “A Behavioral Model for Predicting Employee Con-
tributions to 401(k) Plans.” North American Actuarial Journal 5 (1): 80–94.
—. 2008. “The Impact of PPA on Retirement Income for 401(k) Participants.” EBRI Issue 
Brief 318, Employee Benefit Research Institute, Washington, DC.
Yakoboski, Paul. 1994. “Salary Reduction Plans and Individual Saving for Retirement.” EBRI Issue 
Brief 155, Employee Benefit Research Institute, Washington, DC.
Yakoboski, Paul, and Jack VanDerhei. 1996. “Contribution Rates and Plan Features: An Analy-
sis of Large 401(k) Plan Data.” EBRI Issue Brief 174, Employee Benefit Research Institute, 
Washington, DC.

81
CHAPTER 4