Teaching an Old Dog New Tricks: How lessons from
Retirement 20/20 can Improve DC Plans
By Emily Kessler, FSA & Andrew Peterson, FSA, The Society of Actuaries
The Society of Actuaries has undertaken a multi-year project to explore the possibilities for new retirement
plans that go beyond today’s defined benefit (DB) and defined contribution (DC) models. The project, titled
Retirement 20/20, aims to consider the needs we all have for a well functioning retirement system, by clearly
examining the needs and risks faced by the various stakeholders. Retirement 20/20 started with the premise
that the best retirement solutions for tomorrow might come from outside the DB/DC system we have today.
Its first task was to identify stakeholders and determine what needs and risks these stakeholders faced. These
were first outlined in our series of conference reports from 2006-2008. Subsequently we held a call for models
to ask actuaries and others to design new retirement plans that better meet the needs and risks for the
stakeholders that we identified. The prize winning papers of this call for models are available on the SOA’s
Retirement 20/20 website.
While the call for models was about future designs and in particular developing new ideas that move beyond
the current DB/DC paradigm, several common themes emerged from the collection of submitted models that
can make DC plans work better today.1 Many of the models work like DC plans, but offer significant
improvements from today’s 401(k). The focus of this paper is to identify those features which could be
implemented in today’s DC landscape. Our goal was to identify the changes that could have the biggest
positive outcomes for participants with relatively little regulatory (and in some cases administrative) changes
required. This paper will outline these ideas, and provide background on why we believe these features do a
better job meeting the needs of system stakeholders. We believe this is a necessary and important first step to
improving retirement security.
Who are the stakeholders?
Retirement 20/20 has identified four primary retirement stakeholders, with a goal of improving overall
retirement security and outcomes for each stakeholder.
• Society. Society represents taxpayers, both current and future generations. We all have an interest in
having a solid retirement system for all retirees (not just ourselves). If retirees aren’t able to meet
their needs in retirement, we might see political pressures to increase Social Security benefits, or to
lower tax rates on the elderly. Any time we spend more on the elderly, other social needs suffer (e.g.
While the focus of this paper is DC plans (as requested by the PostPartisan Foundation), the authors do not want
to diminish the important role that DB plans, when designed and managed effectively, can play in providing
retirement income. In fact some of the ideas presented suggest making DC plans more DB-like.
education, infrastructure). Therefore it’s important for society to create adequate, stable income for
the largest number of retirees possible. We know that we can’t do it all through Social Security –
Social Security does a good job at helping the poor, but it would be very expensive for it to support
the burden of the entire middle class.
• Individuals. Individuals are the people retiring. They have many needs in retirement, but our primary
focus in Retirement 20/20 are the needs surrounding a secure income stream. Individuals don’t
necessarily choose when they’ll retire (due to poor health of the individual, poor health of a family
member, economic downturns, etc. – see SOA study: Understanding and Managing the Risks of
Retirement, July 2010), and an individual who runs out of money at 80, 85 or 90 is going to have a
hard time returning to the workforce. Retirement 20/20 has focused on the needs of individuals for a
secure retirement income, and how to make plans manage that income stream to make it more
secure. In addition, we have focused primarily on middle income individuals – those between the 25th
and 75th percentile of income. These individuals need retirement income in addition to Social Security
but are not wealthy.
• Employers. Employers have been the cornerstone of the retirement system in the U.S. During the
20th century, we focused on building an employment-based retirement system, first through the DB
model, and now, more often, through DC plans. Traditionally employers have used retirement plans
to attract, retain, and provide an orderly exit for employees. But as time has passed, they have faced
increasingly complex administrative and fiduciary roles, which have discouraged some employers from
maintaining their plans and have even kept many employers (and their employees) out of the system.
• Markets. Markets are defined here as the capital markets, insurance companies, financial services
companies, and any other forum in which retirement savings are placed and income is withdrawn.
While not a stakeholder in the same way as individuals, society, and employers, proper use of the
markets is important to achieve the best outcome for the other stakeholders, particularly individuals.
As such, Retirement 20/20 focused on using the markets in the best way possible to create a stronger
Making DC Plans Better
This section will cover the most common features found in the Retirement 20/20 call for models papers that
can be adapted into today’s DC plans. Of the ideas offered for change, we believe that some of them could be
accomplished within the existing regulatory framework, some would require minor changes to the regulatory
framework, while others would require rethinking the structure of the DC plan. We hope these ideas inspire
those with political influence to work toward a better DC system.
We will discuss six potential changes:
1. Focus on retirement income rather than accumulation
2. Set up strong defaults that emphasize retirement income
3. Provide better risk-hedging investment mixes
4. Build variability into retirement income
5. Encourage fewer and larger plans
6. Increase standardization among plans
We believe it is important today to challenge DC plans to do a better job at providing a dependable income
stream in retirement. DC plans were originally conceived as supplements to DB plans. As a supplement to
another plan that pays a stream of income, a plan focused on the accumulation of income to provide an
accumulated account balance was appropriate. Plans were designed with a wide range of investment options.
At retirement, most individuals were counseled to withdraw money as needed, to supplement other fixed
sources of income such as Social Security, DB pensions, or other annuities.
But today DC plans have evolved from their original role as supplemental plans to their current position as the
primary source of retirement income, in addition to Social Security, for most Americans. While many Baby
Boomers (particularly those nearing retirement) still have access to some form of DB income, many of these
plans are frozen and many individuals have had multiple jobs, limiting the benefits they earned in their DB
plan(s). In addition, younger workers are much less likely to have access to a DB plan during their careers.
This requires the DC plan to take “more of the load” in providing retirement income, and it means we need to
ask more of these plans. We’ll focus on what we can do to make DC plans work better at providing a
dependable stream of retirement income.
1. Focus on retirement income rather than accumulation
Most participants in a DC plan get regular statements which show how much money they’ve accumulated.
This follows the model of the typical investment portfolio; your focus is on whether that portfolio has gone up,
or down. The goal of the investment strategy is to accumulate assets. For a savings vehicle designed to
provide needed assets for emergencies and supplemental retirement income, showing accumulations can
provide individuals the necessary information for their retirement planning. But as that savings vehicle
becomes the primary source (other than Social Security) of retirement income, the focus on the pot of assets
does not help individuals understand how this money will have to work for them in retirement, namely to
provide a dependable source of retirement income.
One easy way to get individuals to think about income is simply to show them income projections based on the
assets accumulated. There are several benefits to this:
• It focuses people early in their careers on having enough income in retirement to meet needs.
• Most individuals don’t understand how to convert single sums into cash. However, individuals know
how much their monthly bills are, and can extrapolate, reasonably easily, into what they might need,
on a monthly basis, in retirement. They can easily see if they are or are not on track to meet those
• Focusing on income may encourage more people to convert savings into a stream of income in
retirement. In particular, if people understand the income advantages of purchasing annuities over
annual withdrawals at a 4% rate, this could help boost rates of annuitization from DC plans.
There are several ways to show an income stream. One way can be simply to show the amount of income
available, based on typical scenarios for creating income (e.g. withdrawing X% per year, buying a single
premium life annuity, taking single sum installments over a fixed period). A more elaborate communication
might establish a target income level (which could take expected benefits from Social Security into account)
based on retirement age and income goals. This could help participants determine each year whether they are
on target or not and provide direction on how to adjust retirement age goals and investment or contribution
strategies to reach that target.
There are some disadvantages to this strategy: the risks associated with different strategies are not shown or
discussed with the individuals (e.g. investment risk, insurance company failure, outliving assets) and favoring a
single form of income (e.g. withdrawals of X% per year) could be a nudge to individuals that that is the best
strategy for them. However, giving individuals a baseline of information, with better retirement counseling
later, could help become the basis of decisions.
2. Set up strong defaults that emphasize retirement income
Another step toward improving how individuals take money from their DC plan would be to set up
requirements or defaults for withdrawing funds, or a portion of funds, as income rather than as single sums.
In most cases, the Retirement 20/20 call for models papers preferred strong defaults. We know from the work
of behavioral finance that strong defaults can work nearly as well as strict requirements, while allowing those
individuals who need more options the ability to exercise them.
We discussed previously that individuals who receive information about an account as a single sum are likely to
continue to think about that account as a single sum, rather than as an income stream. Retirement income
provides important protections for both individuals and society. But, individuals might not be nudged toward
taking income without strong direction.
Defaults and requirements can work in many ways. The most draconian is to simply “not permit” any form of
retirement savings to be used for anything other than a life annuity; our Retirement 20/20 designs did not see
this as the preferred option. Many of them solved the problem by looking at either how to limit the
requirement to take income or to provide incentives to take income (or disincentives to take a single sum):
• Provide tax incentives (or tax disincentives): Annuity income could receive preferred tax treatment,
including tax treatment over other strategies which do not provide protection against outliving assets.
This could include strategies that allow someone to consider periodic distributions of some portion of
their 401(k) assets through an annuity to satisfy the minimum distribution incidental benefit rules and
then allowing them to retain the rest of their 401(k) balance as a single sum of assets (not subject to
distribution rules, or subject to a much less stringent set of rules).
• Provide requirements (or tax incentives) to take income but only up to a certain dollar level or
percentage of pay: One criticism of annuitization requirements is they are often set to be
annuitization of entire accounts. This approach is too simplistic; everyone needs some sort of savings,
high income individuals don’t need to annuitize their entire account, and ideally, you should allow for
other sources of income if they’re available.
First, individuals must be able to combine various accounts, including any other annuity streams
coming from DB plans. Second, while income is important to provide security, most individuals only
need a certain amount of income in addition to Social Security. The requirements could say a
minimum of X% of the total account balances, but no more than $Y in income (combined from DC and
DB sources). The income level could be calculated based on some combination with Social Security.
While this creates a complex calculation, there may be ways of letting individuals certify as of a certain
date that they’ve met this rule which allows them to retain their tax preferences.
3. Provide better risk-hedging investment mixes
This feature will be familiar to many individuals based on the widespread availability of target date funds.
However, this assumes a more sophisticated treatment of these funds and one again that is targeted to
achieve income goals rather than wealth accumulation.
Actuaries typically think about an asset portfolio as hedging a particular insurance need (pensions, life
insurance, annuities, etc.). Not only are assets chosen that best meet the risk profile of the insurance, but the
amount of assets is set based on the projected need (anticipated pension payments, anticipated life insurance
payouts, projected annuity payments). A portfolio to pay out annuities would be invested in more long-term
assets than a portfolio to pay term insurance within the current year.
In the Retirement 20/20 call for model papers, authors typically recommended putting more investment in
fixed income, particularly Treasury Inflation Protected Securities (TIPS) and much less investment in equity,
particularly at retirement. The advantages of a pre-selected investment mix are that individuals participate in a
large fund with lower administrative costs, the funds use professional investment advisors with the appropriate
training and skill, and the funds are designed to meet target risk goals at retirement (providing greater security
to individuals). While this may require that individuals contribute more to achieve the same retirement income,
it would also protect them against market crashes.
These pre-selected investment mixes could be standardized (much like Medicare Supplement plans) so that
participants could compare a “DC Retirement 20YY” investment mix offered by different companies.2 These
pre-selected investment mixes could be offered with target benefits (as discussed in the “focus on income”
section). This would allow a participant to combine the features of targeting an income with a mix that is
designed to be more stable (subject to less market volatility) as the participant approached retirement.
These may sound like today’s target date funds, but they have a difference. First, there would be some
standardization of mixes so that individuals could truly compare funds with a “20YY” target date and know they
have the same baseline investment mix. Second, that investment mix would be designed to secure retirement
income by the target date. The goal of the fund is to become less risky over time, so that as you approach the
date, you know you are “on target” to meet your goals. This will likely mean the funds earn less, particularly
as someone approaches the target date, than the today’s typical target date fund.
Companies could still offer other “20YY” funds – they would only be required to standardize the investment risk
for specific plans targeted to DC accounts.
These investment funds could also be used to provide the annuity payments at retirement. Insurers could set
up these funds but other financial services companies could be licensed to provide annuity payouts as well.
One advantage of investing in TIPS would be to allow these funds to provide inflation-linked annuities at a
lower cost post retirement.
4. Build variability into retirement income
One criticism of DB plans is that, by securing a benefit for participants, this places the sponsor (typically the
employer and its shareholders) at high risk, for they must secure and fund the benefit, no matter what the
fortune of the company or entity. Similarly, though, DC plans have been criticized for being too unpredictable
in their retirement provision.
One option is to move away from absolute guarantees and provide some certainty while allowing some
variability. The annuity portion within TIAA-CREF, while it is guaranteed for longevity purposes, does have a
participating element. Over their history, based on a conservative investment philosophy, they’ve been able to
give some increase each year, but they never guarantee what that increase will be. At a recent Retirement
20/20 conference, attendees discussed at length that guarantees are expensive, and we don’t need guarantees
for all portions of the income stream. Many individuals are used to having small variations in their income
during their working years; retirees also should be able to make spending adjustments for small variations in
What variability could be built into retirement income amounts? Most Retirement 20/20 authors and
conference attendees agree that a guaranteed baseline payment that continues for life is important, so an
individual doesn’t outlive his assets. However, there can be variation around that baseline amount (what is
received as a monthly check). One way to build in variability is to provide lower benefits that are inflation-
linked, but allow the inflation increases to vary. Another way to factor in variability is to have individuals work
toward a target benefit. The target is not guaranteed; it’s simply a target, although the plan is structured such
that if you make a certain level of contributions and invest in the specified investment mix (which reduces risks
closer to the retirement age), the individual could be no more than 5-10% below the target at retirement. To
make a target benefit such as this work, the individual may have to make additional contributions in some
years (or work later). As an example, a plan might be targeting income of 30% of an individual’s salary based
on an age 67 retirement age. However, at age 50 due to lower investment returns than assumed, the
individual is informed that to reach the target they need to increase their contribution percentage by X%, push
out their retirement age by Y years, or some combination of the two. This is another example of variation – in
this case variation in contribution levels or retirement age versus variation in income in retirement.
Another source of variability could be investment returns. As noted earlier, investment for the purposes of
providing retirement income should use assets that better hedge the income needs. Simply put, a high quality
bond, with a lower return, is a better bet for income than a stock, which, while it could return more, is more
likely to not provide the needed income. There would likely need to be caution in how much investment return
was built into funds, and as noted, some variability is acceptable, but large variations in investment return
makes planning for a stream of income difficult.
5. Encourage fewer and larger plans
We recognize that providing better hedges on retirement risk will tend to reduce potential returns from equity
investments. The Retirement 20/20 call for model papers saw these tradeoffs as important for improving the
predictability of income in retirement. Part of how these papers overcame the challenge of maximizing returns
was minimizing fees, particularly investment fees.
We know that one way to drive down the cost of the retirement system is to increase efficiencies in the
system. One easy way of creating efficiencies is by creating larger pools of assets. This decreases the per
person cost of administration and investment. It also can provide an opportunity for the plan to hire
sophisticated investment advisors who are able to put together investment options that better hedge
investment risks as individuals approach and enter retirement while minimizing fees. In addition, for the
portion of the fund seeking equity returns, larger pools will allow a wider range of investments. Therefore,
while having fewer plans might be seen to limit choice for individuals, the end result is providing better, more
efficient and more cost effective choices to individuals that allow for economies of scale, aligning stakeholders
skills with resources and sharing (or pooling) risk.
How valuable can large institutional funds be? A study done by Ambachtsheer3 calculated that the difference
between an institutional fund, smartly governed and invested, and a retail mutual fund where an individual
“buys high and sells low” can be as much as 6% real return. The table below takes their returns and
calculates how that can change the amount of annuity income an individual can pull from a retirement plan.
As we look for ways of improving 401(k) plans, driving down the administrative costs must be a key strategy
How real returns Institutional fund Mutual fund (retail)
achieved can affect
retirement savings Market less High retail
Smart governance institutional expenses, “buy
expenses high, sell low”
Real return 4.0% 2.6% 1.5% -2.0%
Final savings 485,200 348,900 273,300 137,200
Annuity payment 28,300 20,300 15,900 8,000
Income replacement 57% 41% 32% 16%
Author’s calculations based on $5,000 per annum savings and Ambachtsheer’s real returns as reported in Ambachtsheer, Keith.
2008. The ideal pension-delivery organization: theory and practice. In Frontiers in Pension Finance, 2008, Dirk Broeders, Sylvester
Eijffinger and Aerdt C.F.J. Houben, editors. Edward Elgar Publishing.
One consequence of improving efficiency may be that we have to look beyond the current model of employer
sponsorship. Currently, only the largest employers are able to sponsor 401(k) plans that can meet these goals
of investment and administrative efficiency. To be able to make these efficiencies available to all individuals,
we may have to offer small- and medium-size employers (and those large employers not wishing to sponsor a
plan) another way to access retirement savings for their employees. This gives individuals working at these
employers a way to achieve the same efficiencies in their retirement savings as those employees at large
employers (currently these individuals are only able to access IRAs generally at the retail level).
Another related issue would be to allow these funds to limit the range of investment options available to
individuals. Part of the reason large professionally managed plans are able to achieve superior returns is that
the investment managers understand how to use the available market instruments. Minimizing the number of
funds, encouraging employees (through defaults) into funds designed to maximize returns for a given level of
risk is one of the benefits to be obtained from large plan structure.
6. Increase standardization among plans
Closely related to the issue of having fewer larger plans is having more standardization within plans. This
allows individuals (and employers) to truly comparison shop plans’ administrative costs, investment fee
structures, real returns and other features. It also enables us to have a national retirement conversation,
where neighbors share a common retirement structure and can share experiences and understanding in the
same way that individuals can talk today about their mortgage terms or car financing. In addition, we can
Ambachtsheer, Keith. 2008. The ideal pension-delivery organization: theory and practice. In Frontiers
in Pension Finance, 2008, Dirk Broeders, Sylvester Eijffinger and Aerdt C.F.J. Houben, editors. Edward
transfer this understanding into the educational system, helping students understand their future
responsibilities as citizens to save for retirement while they learn other basic financial and civic responsibilities.
We note that the authors of the Retirement 20/20 papers generally did not envision a single country-wide
system; we recognize the value of innovation and creativity in finding new solutions and any standardization
must be balanced with opportunities for creativity, choice and difference. However, some standardization in
reporting, in benefit structures, and in options can help individuals understand choices, comparison shop, and
introduce more transparency to the market. It can also help individuals when managing several accounts
across several employers.
We are at a crossroads in the retirement system. We have evolved from a system originally focused on DB
plans, to a system balanced between DB and DC plans, to one evolving to a focus primarily on DC plans.
Within this context, we understand the political difficulties of a radical revision of the system. However, we
believe with the changes outlined in this paper – some simple and some more complex –we can evolve an
existing structure, the 401(k) plan, into something that can better meet the retirement needs of the middle
class. The Retirement 20/20 project has focused on what designs will produce optimal results. We have
recognized from those designs that there are certain key features which seem to be endemic to a well
functioning retirement system: a stream of income that is will last a lifetime, some ability for that income
stream to link to inflation, and better ways to design investment structures so there is more certainty about
retirement income as one gets closer to retirement. The Retirement 20/20 designs have also recognized that
providing guarantees is expensive; many designs have focused on the notion that a targeted, narrow range of
income can provide essentially as good an outcome as a guaranteed stream of income with less cost. We
believe that DC plans can be adapted to meet these goals. Some of these ways are simpler (although not
always non-controversial): using the principles of behavioral finance to focus individuals on income streams by
changing how plans are communicated, changing defaults (or requirements) to encourage individuals to take
at least part of their 401(k) plan as income and selecting investment mixes that do a better job of hedging risk,
particularly as individuals approach and enter retirement. Other ideas require more changes to our 401(k)
structure; such as creating income streams that both guarantee lifetime income but allow variability of
payment. Finally, changes that would increase standardization and provide for fewer, larger plans would start
to move out of a solely employer-based system into a system that considers the employer as the place to save
for retirement, but not necessarily the sponsor of the plan.
We believe that with these and other adaptations, DC plans can better meet the challenges of providing
consistent income in retirement for life, that provide some link to inflation, and that help individuals manage
the transition from work into retirement with a more stable stream of income. The challenge of Retirement
20/20 has always been to push the retirement system to do more and create better outcomes. However, we
can do so only if we recognize that the system today is being asked to do more than it was designed to do,
and if we acknowledge that we must make deliberate changes to the system to provide better outcomes for
To find out more about the results of the call for models, and read the prize-winning papers, go to
http://retirement2020.soa.org/new-designs.aspx. Selected papers will be printed in an upcoming monograph.
What is Retirement 20/20?
The Society of Actuaries’ Pension Section Council has been concerned about changes to the pension system
over the past few years. In looking at the issue several years ago, they concluded that they needed, with
assistance from others, to step back and look at the bigger picture. Retirement systems today are based on
20th century models, and there have been significant demographic and economic changes that may mean that
yesterday’s models no longer fit today. Retirement 20/20 is a process to bring together experts on retirement
issues to redesign a system from the ground up to meet 21st century retirement fundamentals. One goal the
Pension Section has in this process is to develop new retirement risk sharing models that better utilize risk
pooling and risk sharing.
Retirement 20/20 is a process of exploration and creation to envision new retirement systems to meet the
needs of the 21st century. The 2006 Retirement 20/20 conference, “Building the Foundation for New
Retirement Systems,” looked at the needs, risks and roles for the four major system stakeholders (individuals,
society, markets and employers). The 2007 conference, “Resolving Stakeholder Tensions: Aligning Roles with
Skills,” focused on determining and aligning the optimal roles for the various stakeholders. The 2008
conference, “Defining the Characteristics of the 21st Century Retirement System” discussed the optimal
characteristics for successful retirement systems, specifically the areas of changing the signals within
retirement systems, self-adjusting mechanisms for retirement systems and optimal strategies for retirement
In 2009 the Retirement 20/20 initiative launched a call for models. The call for models asked individuals to
submit their ideas for new “Tier II” retirement systems—i.e., what is typically thought of as employer-provided
retirement benefits that fit between social insurance and private savings. The call for models was the
culmination of the Retirement 20/20 work to date including three conferences which explored needs and risks
for stakeholders in the retirement system (individuals, society, employers and the markets). Submissions were
judged based on how well they met the criteria of the Retirement 20/20 Measurement Framework (which
considers needs and risks for the various stakeholders) and how well they handled issues of risk, governance,
administration, transparency and transition.
As a result of the call for models, the SOA received 18 paper submissions from Canadian and American
authors. The four best papers, and a few selected others, formed the basis for the “Retirement 20/20: New
Designs for a New Century” conference, held June 2 – 3, 2010 in Washington, D.C. A second conference, in
cooperation with the CD Howe Institute and the Canadian Institute of Actuaries, was held December 8, 2010 in
Toronto. A monograph is forthcoming.
To find out more go to www.retirement2020.soa.org.