a pension promise to oneself

November/December 2013 www.cfapubs.org 13

Financial Analysts JournalVolume 69 · Number 6

©2013 CFA Institute

P E R S P E C T I V E S

A Pension Promise to OneselfStephen C. Sexauer and Laurence B. Siegel

Saving for retirement is not hopeless. Well-run DB pension plans provided retirement income for genera-tions. When plans failed, it is because they broke the rules. The same applies to individuals. By understanding a set of rules on how much to save and how to invest and then sticking to those rules—that is, by making a pension promise to oneself—retirement income goals can be met. Fulfilling this promise requires more saving than most people are accustomed to.

“Don’t have a pension? Don’t worry. Most people don’t. They will get to retire, and so will you.

“What you and almost everybody else do have is the ability to make a pension promise to yourself that is the economic equivalent of the promise that an employer could make. With or without your employ-er’s assistance (it helps but isn’t really required), you also have the ability to deliver on that promise. You will make pension payments to yourself from the moment you retire until the end of your life or your spouse’s life, whichever comes later.

“You may not know that you have this ability, but you do have it. You are about to receive the basic components of a toolkit for making this happen, and over time, you will receive the rest of the tool-kit. The strategy for making this happen involves a lot of saving but almost no risk. When you do face risk, you will have been provided with the tools necessary for managing it. You’ll be just fine.”

Why aren’t these words spoken to every employee who begins to work at a company, gov-ernment agency, or nonprofit organization? Why aren’t they taught in school? Why aren’t they part of the advice lovingly given by parents to their chil-dren as adulthood looms?

The reason is that many of the people who should be delivering this message do not even know that it is true. Yet, every word of it is true. There is nothing magical about the pension promise that an employer with a traditional defined benefit (DB)

pension plan makes to an employee. The employer saves and invests money on the employee’s behalf, according to a set of rules designed to make sure that enough money is available to pay the promised benefit, and then pays it to the employee as a postre-tirement income stream. Failures do occur, but that is because people—employers and employees—do not stick to the rules.

Individuals can provide pensions for them-selves almost as easily as employers can. There is no magic in this either. It requires saving a lot of money—almost exactly the same amount, if oppor-tunities to pool longevity risk are taken advantage of, that is needed to fully fund and pay out a DB promise. This fact leads us to an important point: If the amount of money needed to fund a DB plan can be made available to defined contribution (DC) plan investors, that amount is also enough to make the DC plan work.1

We say “almost as easily” because there are some economies of scale involved in providing a pension for a large group of people. These econo-mies can mostly be replicated by individuals.2 It is not very hard, and those who make a pension promise to themselves and do what is needed over time to keep that promise are made vastly better off by doing so. They will be able to sleep well at night.

The Pension SettingMuch ink has already been spilled on the pension catastrophe we are facing, so we will not go into the details. It suffices to say that private sector DB plans are all but extinct, mostly because spon-sors hoped, despite evidence to the contrary, that stock market profits would substitute for adequate

Stephen C. Sexauer is the chief investment officer, US multi-asset, at Allianz Global Investors, New York City. Laurence B. Siegel is the Gary P. Brinson Director of Research at the Research Foundation of CFA Institute, Charlottesville, Virginia.

Financial Analysts Journal

14 www.cfapubs.org ©2013 CFA Institute

contributions. Until recently, public sector DB plans were propped up by the ability to tap a growing tax base, but when tax revenues turned sharply down-ward in 2008 and 2009, the plans fell on hard times. Some defaults have occurred and many more are looming. The inability or unwillingness to control benefit growth is a secondary reason for both pri-vate and public DB plan failures.

The DC plans that have almost universally replaced DB plans suffer from even worse out-comes: Waring, Siegel, and Kohn (2007) reported that the median DC plan participant was retiring with an investment balance of $44,000; the mean balance of $150,000, skewed by a few affluent retirees, is not much better. The Federal Reserve’s 2010 Survey of Consumer Finances (SCF) shows little improvement, with households that are approaching retirement having a median DC balance of $63,000; when IRAs are included, the median balance is $120,000. Assuming a 4% with-drawal rate, the SCF balances, as of 2010, support monthly spending of $210 and $400, respectively. These are hardly pensions at all; they are better characterized as “beer money” savings plans. We simply have to do better for our valued employees and for ourselves.3

We concede that the battle to save DB plans in their original form is almost entirely lost. This essay is about using DB plan thinking to achieve better DC plan outcomes. The thinking behind DB and DC plans is the same: The task at hand is to spread the earnings from one’s working life over one’s whole life, and one can only consume what one has. Thus, contributions to a hypotheti-cal fully funded DB plan are also enough to fully fund a DC plan in the sense of providing an aver-age benefit across employees that is as good as the DB plan benefit.

How DB and DC Plans Are the SameWe noted earlier that the amount of money needed to fully fund a DB plan promise is also sufficient to fund a DC plan that is equally beneficial to the par-ticipant. In this section, we explain this comment and indicate how DC plans can be made more like DB plans.

Dollar Equivalence of DB and DC. The func-tion of any pension or savings plan (DB or DC) is to shift consumption over time. There is only one way to do that: to participate in the investment markets.4 With both DB and DC plans, participants accumulate assets by forgoing consumption; later, they decumulate assets, enabling consumption. They can consume only what they have saved plus investment return.5

In a DB plan, there are two intermediar-ies: (1) the sponsor, who performs the savings and payout functions and may also guarantee the benefit with her balance sheet, and (2) the financial markets themselves, which provide the time shift in consumption.6 As investment gains or losses become apparent and as other changes in the environment occur (e.g., life expectancies lengthen), the sponsor adjusts the contributions to the fund accordingly. Finally, the sponsor pays out benefits as promised.

A DC plan does exactly the same thing except that you, the investor, are the “sponsor.” DC plans have only one intermediary—the financial markets (with employers playing a minimal supporting role). Participants first project the cash flows they are going to need in retirement. Then, they develop and execute a savings plan in an amount sufficient, when investment returns are taken into account, to fund the retirement income requirement. (Savings or contributions to DC plans, whether from par-ticipants or an employer match, are analogous to sponsors’ contributions to DB pension funds.) Next—and this is a key concept—DC plan partici-pants make adjustments along the way for invest-ment gains and losses and for other changes in the environment. Finally, participants use the money when they are retired.

Note that these two stories are very similar. Money is neither created nor destroyed in either one. What you get is what you put in plus or minus investment returns after fees. The only major dif-ferences are (1) who makes the contributions and (2) who manages the process. And any economist will tell you that for a given set of payouts and ignoring such frictions as fees, there is no differ-ence, in terms of overall cost, between employ-ers making the contributions directly (DB plans) and employers giving the money to employees to invest (DC plans).

Enabling DC Plans to Be Experienced More like DB Plans. Many researchers and practitioners have tried to improve DC plans so that the experi-ence they provide to participants is more like that of DB plans. This article is an element in that effort. A key ingredient is to achieve in DC plans the mortal-ity risk sharing that is inherent in the DB plan payout structure. This structure replicates the payout from a hypothetical life annuity that is profit free and fairly priced. Most researchers have recommended that this outcome be accomplished using commercial annuities, acknowledging that, at present, access to these annuities for individual investors can be dif-ficult. We hope that the annuity industry and regu-latory structures evolve to bring greater scale and transparency to this vitally important product.

A Pension Promise to Oneself


At the time when investors want to convert assets to income, DB plans offer the resources of a long-lived organization with a staff and a process for annuitized payout. In DC plans, investors are on their own in selecting an intermediary for convert-ing assets to income, and the providers competing for this business typically offer little transparency and charge high fees. To make DC plans more like DB plans, sponsors need to provide, or contract for, institutional-quality investment management with low fees, low administrative costs, and institution-ally scaled resources for providing access to longev-ity pooling.

A Three-Part Rule Set for Personal Pension Plans. Count Leo Tolstoy could be forgiven—and might be considered more up to date—if he had begun Anna Karenina with the lines, “All suc-cessful DB and DC plans are alike. Each failure is unique.” Plans that understand and play by the rules succeed. Those that do not—whether DB or DC and no matter what rule they think they can break—fail.

Because our point is that a well-run DC plan can and should use the same principles that large institutions have used to manage DB plans on an economically sound basis, we identify the three basic steps that DB or DC plans need to take:

1. Liabilities must be appraised and discounted back to a present value, which is what actu-arial firms do for DB plans: They estimate the combined pension promise a company has made to all its employees and then discount this liability to a present value. Individuals can do the same, and we will show how it is done and set forth a shortcut (the retirement multiple) for arriving at the amount of money that needs to be saved.

2. Assets must be accumulated according to an economically sound plan, and wishful think-ing about markets must not be allowed to substitute for rational savings rates. Although not all DB sponsors are this virtuous, the suc-cessful ones play by the rules and have close to fully funded pension plans. DC plan investors must do the same. Hope is neither an invest-ment strategy nor a retirement plan; this prin-ciple applies to both DB and DC structures.7

3. Assets must be decumulated (spent or paid out) in a sensible manner that preserves the tremen-dous value of longevity pooling. Decumulation is not the main point of this essay, but Waring et al. (2007) and Sexauer, Peskin, and Cassidy (2012) have described ways that a stable income lasting for the rest of one’s life can be generated from an asset pool.

One advantage DB plans have is the ability to make adjustments over time when their estimates are off because of economic events that materially change the value of the promises (the liability) or the matching assets. They can do this by increas-ing pension contributions and lowering profits, decreasing future compensation and benefits, or both. Individuals also have the ability to make adjustments, and they do this all the time. We call this concept a personal fiscal adjustment (PFA) and regard it as the “dark matter” that, over time, enables resources needed to be made equal to resources available.

Personal fiscal adjustments—decisions to increase or decrease consumption or production or to shift the time period in which consumption or production occurs—will be made as surprises occur. For example, in an extreme case, a PFA can include moving in with family (typically, with children, the millennia-old retirement plan). It can also include working almost full-time in “retire-ment.” The key point is that people can and do make the required adjustments to match needs and resources. DB plan sponsors also make adjustments to changing circumstances by contributing more to their plans, taking contribution holidays, or rene-gotiating benefits.

The retirement decision-making framework that we set forth is usable by the vast majority of the US working population. Those in, say, the top 7% or so when ranked by income (or assets) can afford professional financial planning and often have access to institutional-quality investment manage-ment. Although our approach applies to the top 7%, many of these investors are already doing fine and need little additional help.8 Our goal is to bring the needed technologies and resources to everyone, especially the other 93%.9

Why Is This So Hard? Why Are There Failures?Because we have been conditioned by employer-provided pensions to think narrowly about the retirement question, most thinking about pensions and retirement savings is anchored in a static and oversimplified world. In this world, people work from age 20 to age 65 and then magically—almost mystically—transition to a state called “retire-ment,” wherein they live for another 20–40 years with a minimal, or at worst a modest, drop in consumption.

Let’s see how many holes we can find in this story without filling an encyclopedia. When was the last time you met a 20-year-old who was working at what would become his or her life occupation?



Who works for 45 years without quitting, getting fired, or becoming ill or disabled? Don’t quite a few people these days live to 100 or 105, potentially making a mockery of the idea that saving for 20 or 30 years of retirement is adequate? Don’t some people run up medical bills in the hundreds of thousands of dollars? Aren’t we all afraid that we will end up in a nursing home?

OK, enough doom and gloom. Let’s look on the bright side. Don’t quite a few people experi-ence comfortable retirements, thanks to a series of promotions or career changes, successful sav-ing and investing, or both? When setbacks arrive, don’t most people adjust and go forward, some even being prudently prepared with low-cost but valuable disability insurance? Aren’t many employers careful and responsible, offering a prudent combination of current wages and future income and fully funding the pension promises; or if they offer a DC plan, do they not work with employees to help them accumulate a DC plan balance that is at least adequate? What might we learn from the success stories that can be used to build the toolkit for everyone?

The retirement problem, which we propose to help solve through a pension promise to oneself, needs to be understood as follows. We are trying to distribute the income from one’s work years over one’s whole life. Life is long and getting longer, and work years are short (owing to increasing educa-tional requirements at the beginning and the desire for a successful retirement at the end)—although we have learned that some level of involvement with work after age 65 brings the benefits of human contact and purpose, as well as income to pay for life-enhancing technologies and consumption goods. We face the unprecedented challenge of dealing with 80 or more years of uncertainty, begin-ning as one reaches adulthood and potentially end-ing as late as age 105. Yet, we are charged with cre-ating some sense of certainty or security in old age, which we define as the part of life when the option to materially change one’s financial well-being through work is mostly gone and which could last for 40 more years.

No wonder retirement planning seems hard! We are now going to make it very simple. Like diet-ing, however, it is simple to understand but not easy to do; it’s not easy because it involves saving a very large fraction of income.

The Personal Pension PlanIn this section, we advocate for simplicity in pen-sion plans and set forth a three-step process for managing them.

Simplicity Is Paramount. The most important principle in building a framework, a personal pen-sion plan, is to keep it simple. It is hard to gener-alize about human beings, but it is a pretty safe bet that nurses, airplane mechanics, lawyers, and restaurant workers are not crying out to become experts in portfolio theory. In our experience, many investors do not fully grasp the difference between a stock and a bond, much less the difference between an expected return and a return one can count on. Even experienced investment profession-als suffer from an illusion of precision regarding return expectations, overweighting the likelihood of attaining the expected return and materially underestimating the variance of real-life outcomes. We write this not to be demeaning but to be realis-tic: A complex strategy involving fancy math and beautiful statistical simulations will simply not be widely adopted or will be adopted incorrectly in a way that hurts investors.

There is another reason to keep the pension promise simple: In the middle and late retirement years, our cognitive skills leave us—and do so at an increasing rate. So, investors do not want to couple this almost assured fall in clear thinking with a retirement plan full of complexity and risk taking. It should be simple, and as we get older, it should become more automatic.

Elements of the Personal Pension Plan. Thus, the personal pension plan that we set forth here is strikingly, even childishly, elementary:

1. We determine the income stream that the inves-tor will need in retirement—that is, the liability. It can be as simple as an educated guess—a rule of thumb, such as 70% of what I am earn-ing today. We then subtract expected Social Security payments to determine the yearly amount the investor will have to generate from personal savings.

2. We determine the multiple of that annual retire-ment income number that the saver needs to accumulate (this number is called the “retire-ment multiple”).

3. We determine a savings rate that will produce that amount of money when invested in low-risk assets.That is all. We are done. It comes down to two

numbers: the amount of income desired in retire-ment and the retirement multiple. Investing for retirement really is that simple.

The Personal Fiscal AdjustmentOK, we are not quite done. We noted earlier that all successful DB plans adjust to changing cir-cumstances. We also said that DB and DC plans



involve the same thinking and essentially the same operational steps. We now discuss what individual people, including DC plan participants, do to adapt to changing circumstances. When such an adapta-tion relates to one’s financial affairs, we call it a per-sonal fiscal adjustment in homage to the currently popular term “fiscal adjustment,” which is used to describe similar behavior by governments. Thus, when either the asset side or the liability side of an investor’s balance sheet changes in an unexpected way, he or she undertakes a PFA to bring it back into balance.

Continuous adaptation to changing circum-stances is a good start at fixing what is wrong with any financial arrangement. If DB pension plans had reacted to poor market returns by increasing con-tributions and limiting the growth of benefits, they would still be the dominant retirement program. The PFA is the most natural behavior imaginable, and we see it in every aspect of life. When you have less, you spend less. When you have more, you can spend more or save more, with the trade-off between current and future consumption deter-mined by the usual factors—the extent to which one’s liability is unfunded, the attractiveness of various investment and consumption options, and so forth. Let us apply the PFA to managing DC plans. Without such adaptive behavior, the pension promise to oneself is almost doomed; with it, the promise can be kept—in most cases, easily.

Why is the PFA so crucial to success? Human life is far too long, too uncertain, and too lep-tokurtic for the standard finance models to apply. (“Leptokurtic” means fat tailed, with the left, or downside, tail the troublesome one; you might get a bonus equal to 10% or 20% of salary, but when you lose your job, you lose the whole job, not 10% or 20% of it.) One of us heard Paul Samuelson, the originator of much of the standard model, say at a conference, “Assume that a schoolteacher saves $10,000 a year for 50 years.” Where we live, schoolteachers have to retire at age 66, so this imaginary teacher must have started working at age 16, which means she must be a genius because schoolteachers are also required to have a college degree. This particular rendition of the standard model does not account for marriage, children, strikes, layoffs, illness, disability, the desire not to do the same thing for 50 years, or, for that matter, good fortune (e.g., her husband succeeds in busi-ness and she no longer wants to work). Investment planning must conform to reality, not pretend that reality can be forced to conform to the model. The real, flesh-and-blood version of Samuelson’s idealized schoolteacher-saver makes PFAs all the time, in all aspects of her life.

PFAs differ in the short run and the long run. In the short run, the main adjustment is to consump-tion. In the long run, however, one can work harder, increase the number of workers in the family, work smarter (by pursuing strategies to enhance one’s human capital), or plan to work longer. Large changes in one’s consumption plans can be accom-plished by relocating, by moving in with relatives, and through other means.10 People are far more dynamic and adaptive than any planner thinks!

Everyone makes PFAs at some point. Some do this throughout their lives, and some, only when a crisis begins to be perceived around age 50 or 55. Our objective is to encourage individuals to make PFAs sooner in life—not wait for a crisis—and to make PFAs better thought out and better executed.

The concept of PFAs enables us to understand how anyone gets by at all in a world where DB pension plans are failing and DC plan participants are saving way too little and investing poorly. The answer is that people survive with available resources and sometimes make more resources available. They muddle through. They do what has distinguished the human species since it began its time on earth: They adapt.

Much effort is going into looking for bul-letproof systems and turnkey solutions that will make retirement “work” without the ability to make ongoing adjustments. There aren’t any. The retirement establishment acknowledges this by calculating the probability of failure. Most income solutions are framed exactly this way, as though failure were an acceptable outcome. We know that an adult lifetime can encompass 80 years, which is a very long time; over such time frames, forecasts are almost completely useless. However, what we always do have is the ability to make adjustments, adapt, and go forward. This is an invariant compo-nent of human behavior and is the hidden option on the personal balance sheet that keeps it in bal-ance, thus making the personal pension plan work when other methods do not.

The Retirement MultipleThe retirement multiple (RM) figures prominently in the personal pension plan. It is the number of years of income you need to save in order to retire while investing risklessly (or as close to risklessly as markets allow), where “income” is not your cur-rent pay but the cash flow you need to generate, over and above Social Security benefits, in retire-ment. An example using current market rates is that if you need to generate $50,000 per year in retirement, you need to have assets amounting to 21.47 times that amount, or $1,073,500. So in this case, the RM is 21.47.



This simple heuristic reduces the retirement calculation to a multiplication problem that a fifth grader can solve. For most investors, the RM itself will have to be provided externally—that is, by a data provider that calculates the relevant RM on the basis of interest rates and other market data.

Riskless Investing for Lifetime IncomeTo develop a plan for making and keeping a pen-sion promise to oneself, we start by assuming that the investor wants to guarantee the desired income level, not merely have a high probability of achiev-ing it. Thus, in building a base case, we assume riskless investing (or, more precisely—because no investment is riskless—investing with as little risk as possible), both before and after retirement. Before retirement, in the accumulation phase, we treat Treasury Inflation-Protected Securities (TIPS) as the riskless asset.11

After retirement, in the decumulation phase, minimum-risk investing means acquiring a lad-dered portfolio of TIPS to provide for the first 20 years of retirement—that is, from age 65 to age 85—and a nominal deferred life annuity to provide for age 85 and beyond. The laddered TIPS portfo-lio is structured to provide income that rises with inflation over the first 20 years of retirement; the TIPS portfolio is exhausted (has a zero balance) at the end of the 20th year. The deferred annuity is nominal because there are no inflation-indexed, or real, deferred life annuities available; if one were available, we would buy it. The deferred annuity payouts are scaled so that the first annual payout (and, because the annuity is nominal, all other pay-outs as well) is equal to the last annual payout from the TIPS ladder.

The postretirement decumulation strategy is described in greater detail in Sexauer et al. (2012) and is called the “DCDB12 strategy,” for “defined contribution decumulation benchmark”—although the initials are also supposed to suggest a “DC to DB” transmutation. The DCDB yield is the blended rate, stated in nominal terms, on the port-folio consisting of laddered TIPS and the deferred annuity. As of December 2012, the DCDB yield was 4.657%, much of which is return of capital, rather than return on capital. Updates of the DCDB yield, along with other information, are available at www.dcdbbenchmark.com.

We need to know the decumulation strategy at this level of detail so that we can know how much money to save in the accumulation period. Specifically, the RM is the reciprocal of the DCDB yield: 1/4.657% = 21.47.

This simplified, almost-riskless rendition of the personal pension plan assumes that investors make no attempt to increase investment returns through risk taking and that they use longevity pooling to minimize the savings requirement. Although almost no one invests this way, the riskless strategy can be understood as a benchmark, or measuring stick, by which investors can gauge their progress in a riskier strategy. The rare investor who wants to shun all avoidable risks can allocate 100% to TIPS in the accumulation phase and “index” to the DCDB benchmark portfolio in the decumulation phase, demonstrating a key point: The benchmark strategy for a retirement plan should be, and is, executable by a typical investor using standard investments.

Conventional Investing (Risk Taking)Suppose you do want to take investment risk. We will confine this discussion to risk taking before retire-ment and investing risklessly to generate income during retirement. (Because—by definition—you cannot work longer or save more if you are truly retired, the only available PFA if postretirement risk does not work out is to cut consumption, which is an uncomfortable decision.) With a risky preretirement portfolio, you still need the same amount of money upon retiring as with a riskless one (the RM is the same), but accumulating it may require less saving. Specifically, the expected value of the amount you should plan to save is lower in each period, but instead of being able to map out the entire savings schedule in advance, you have to adjust it in each period for realizations (the difference between the return you achieved and the return you expected or planned for). In other words, after a disappointing market return, you have to “true up” by either increasing saving in one or more later periods or decreasing your expected income in retirement. The former is usually the more attractive option.

As we mentioned previously, DB and DC plans face the same underlying dynamic. You can get out only what you put in plus market returns. Most DB plan failures have occurred because the sponsor took risk (with the beneficiaries’ deferred wages) and was unprepared for shortfall risk at a scale that overwhelmed the sponsor’s ability or willingness to pay up the difference.13 DC failures happen the same way.

Therefore, we use a riskless rate as the invest-ment assumption in the accumulation period, as well as in postretirement decumulation, because we need a clean reference point. We realize that, in practice, investors will choose other portfolios.

Next, using an example, we examine our two variables: the desired retirement income and the retirement multiple.



A Teacher’s Pension Promise to Himself: A Numerical ExampleWe consider a public high school teacher in a large US city (Columbus, Ohio, one of the largest cities that offers a DC-only structure as an elective alter-native to a DB plan) to be the prototype for a middle-income worker.14 Suppose our teacher is early in his career and earns $37,328 per year.15 We need our teacher to do three things. First, he must take a guess at what he will want for retirement income. One easy way is to look at what senior colleagues earn—say, a teacher who is about to retire with 40 years’ experience and a master’s degree, which for the city of Columbus is $79,904—and multiply by a number representing the percentage of preretire-ment income that he will need in retirement—say, 70%. For this teacher, it would be $55,933.16 Of this amount, Social Security will provide $24,912, so he will have to generate the $31,021 difference from savings and investments.17

The second step is to apply the market-based retirement multiple to the annual retirement income number. The retirement multiple, as mentioned, is 21.47. The required savings amount, or savings target, is the expected or target retirement income multiplied by the retirement multiple.18 Therefore, the savings goal is 21.4729 × $31,021 = $666,111.19

The third step is to spread the required savings amount (net of balances already accumulated) over the remaining years of work. We assume that our teacher is 25 years old and will work for 40 years, through age 65.20

That is it. From here, our young teacher will behave as an experienced and practiced Bayesian. As time goes on and more information unfolds—updated needs or desires, wage levels, asset levels, and so forth—our teacher will simply adjust the retirement income number, something that can be done without computers, models, or simulations. The specific amounts will be updated over time as conditions change. The power of the approach is that it is a simple rule that (1) works and (2) allows people to compress 40–80 years of dynamic com-plexity into something that they can manage and understand. Table 1 summarizes the numbers used in our calculations.

Our teacher’s goal for accumulated savings, $666,111, is a fairly large amount of cash. Are you shocked by these numbers? Do you think they are unrealistic or unattainable? Well, they are reality. They are what a DB plan would need to save, on the employee’s behalf, if it did not accept shortfall risk by investing in equities and other risky assets. And because the calculation is transparent and easy, both the goal and the reality are clear: One either saves the

amount shown or makes a personal fiscal adjustment and decreases expectations for retirement income.

The retirement multiple heuristic is powerful. It informs and shapes behavior. Basic economic theory teaches us that knowledge and incentives are two of the most powerful forces in human behavior. Behavioral finance teaches us the impor-tance and power of framing. A pension promise to oneself is a toolkit with the knowledge, incentives, and framing for making a lifelong series of high-utility decisions about one’s retirement.

We also believe that the economics “works,” even with such a high required savings rate. Remember that if the goal could have been attained in a DB plan by the employer withholding the money from the employee and investing it on his behalf, then it is attainable by the individual mak-ing a pension promise to himself.

Let’s look at an example of both the overall sav-ings level and the pattern of change in the savings level over time that achieves the goal of a pension promise to oneself. It is important to note three things. First, the growth in human capital and the associated higher wages are explicit: If wages grow from $37,328 to $68,897 over the first 20 years of the teacher’s career, the growth rate is 3.11% per year. Second, to meet the savings goal, we apply a key successful concept from behavioral finance: Our teacher saves a large portion—52%—of the real (inflation-adjusted) portion of his raises. Third, the expected return is what can be earned in risk-free inflation-indexed bonds, which today is effectively zero.21 Given a starting savings rate of 10%, Table 2 shows what the accumulation plan looks like.

Our teacher thus begins his career saving only 10% of income, but by the 20th year, he is saving 28% of income, and by the 40th year, a robust 32%. Although these savings rates and levels seem high or even unattainable, we have evidence that large increases in savings can be achieved. One of the big-gest and most encouraging revolutions in DC plans has been the success of the Save More Tomorrow

Table 1. Retirement Planning Basics for a Columbus High School Teacher, Based on Minimum-Risk Investing

Initial salary $37,32820th year salary $68,897Final (40th year) salary $79,904Retirement income replacement ratio 70%Retirement income goal $55,933Expected Social Security $24,912Personal pension plan income $31,021Retirement multiple 21.47Lifetime savings accumulation goal $666,111



Tab

le 2

. A

sset

Acc

um

ula

tio

n P

lan

fo

r a

Co

lum

bu

s H

igh

Sch

oo

l Tea

cher

, Bas

ed o

n M

inim

um

-Ris

k In

vest

ing

Num

ber

of w

ork

year

s40

Rea

l inv

estm

ent r

etur

ns0.

00%

Hum

an c

apit

al g

row

th (Y

ears

0–2

0)3.

11%

Hum

an c

apit

al g

row

th (Y

ears

21–

40)

0.74

%H

uman

cap

ital

gro

wth

(who

le p

erio

d)

1.92

%In

itia

l sav

ings

rat

e10

.00%

Perc

enta

ge o

f rai

se s

aved

52.0

0%Sa

ving

s go

al$6

66,1

11To

tal s

aved

und

er a

ccum

ulat

ion

plan

$666

,419

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

1$3

7,32

810

.0%

$3,7

33$3

,733

21$6

8,89

729

.1%

$20,

038

$233

,085

238

,490

11.3

4,33

38,

066

2269

,409

29.3

20,3

0325

3,38

83

39,6

8712

.54,

951

13,0

1723

69,9

2629

.420

,570

273,

958

440

,922

13.7

5,58

918

,606

2470

,446

29.6

20,8

3829

4,79

65

42,1

9614

.86,

247

24,8

5325

70,9

7029

.721

,109

315,

905

643

,509

15.9

6,92

531

,778

2671

,498

29.9

21,3

8133

7,28

67

44,8

6317

.07,

624

39,4

0327

72,0

3030

.121

,656

358,

942

846

,259

18.0

8,34

547

,748

2872

,565

30.2

21,9

3338

0,87

59

47,6

9819

.19,

089

56,8

3729

73,1

0530

.422

,212

403,

087

1049

,182

20.0

9,85

666

,693

3073

,649

30.5

22,4

9342

5,57

911

50,7

1321

.010

,646

77,3

3931

74,1

9730

.722

,775

448,

355

1252

,291

21.9

11,4

6188

,800

3274

,749

30.9

23,0

6147

1,41

513

53,9

1822

.812

,302

101,

101

3375

,305

31.0

23,3

4849

4,76

314

55,5

9623

.713

,168

114,

269

3475

,865

31.2

23,6

3751

8,40

015

57,3

2624

.514

,062

128,

331

3576

,429

31.3

23,9

2854

2,32

816

59,1

1025

.314

,983

143,

314

3676

,997

31.5

24,2

2256

6,55

117

60,9

4926

.115

,933

159,

247

3777

,570

31.6

24,5

1859

1,06

818

62,8

4626

.916

,913

176,

160

3878

,147

31.8

24,8

1661

5,88

419

64,8

0127

.717

,923

194,

083

3978

,728

31.9

25,1

1664

1,00

120

66,8

1828

.418

,964

213,

047

4079

,314

32.0

25,4

1966

6,41

9



program introduced by Thaler and Benartzi (2004). Figure 1 shows the great success of this program at a midwestern manufacturing company. When a DC plan has the right structures and incentives, people can and do save more.

Before Save More Tomorrow was introduced, the savings rate was low and both the employees and the company strongly believed that at the employees’ (generally low) income levels, they did not have the ability to save more. Clearly, this was not the case: People can and do save more when the information, structures, and incentives to do so exist. Figure 1 shows that adoption of Save More Tomorrow can boost savings rates into a realistic range for the first part of employees’ pension prom-ise to themselves.

More broadly, a study that used Aon Hewitt data covering 55% of all US plan participants showed that from 2005 to 2011, the auto-enrollment rate went from 19% to 56% and the use of auto-escalation went from 9% to 51% (Benartzi and Thaler 2013). Whether a Save More Tomorrow–type approach can achieve the savings rates of greater than 30% that are required in the late-career years remains to be seen, but the early-career results suggest that it can.

We have been conservative and strict with our assumptions, including the use of today’s real risk-free rate of zero instead of the long-term average rate, which is positive. These assumptions establish a base case to build on. We believe the risk-free rate is likely to be positive in the future. We also know that most investors will wish to take shortfall risk (say, by holding equities) to try to reduce the sav-ings burden; if the risk does not pay off, the inves-tor has increased rather than reduced the burden, but the expectation is to reduce it. The calculations can be adjusted to reflect these variations on our base-case theme of essentially riskless investing.

Investing in Risky Assets during the Accumulation Phase: An ExampleIn this example, our teacher is willing to accept shortfall risk by investing during his working years to earn an expected real return of 2%. We start with the same retirement income goal and expected Social Security benefit as before and thus the same income goal from the personal pension plan, $31,189. We still use the DCDB rate—a version of the real riskless rate—to discount the postretire-ment liability because we assume that our teacher will want to invest risklessly to generate income after retirement. Thus, we arrive at the same sav-ings goal, $666,111.

While saving during his working years, how-ever, the teacher takes risk and expects to earn a real rate of 2%, which changes the savings sched-ule. If the 2% annual real returns were spread evenly over time, only 27% of the real part of raises would need to be saved and the savings rate (sav-ings as a percentage of income) would climb less steeply, to 20% by the 20th year and 22% by the 40th year. Thus, the 2% real investment return assumption makes the accumulation plan much easier and provides a big boost to consumption during the teacher’s working years. The accumu-lation plan at a 2% real return in the accumulation period is shown in Table 3.

But what is the downside? Because the 2% real rate is a risky rate, the return might not be realized. Any investment disappointments would have to be met with a PFA—either an increase in the savings rate or a decrease in postretirement consumption.22 The PFA will be large if market returns are much less than expected.

It is important to emphasize a key point that we made regarding the retirement multiple:

Figure 1. Increases in Savings Rates from Adoption of the Save More Tomorrow Program at a Midwestern Manufacturing Company, 1998–2002

20022001200019991998

9.4%

6.5%

3.5%

11.6%13.6%

Savings Rate (%)

15

10

5

0

Source: Thaler and Benartzi (2004, Table 2, p. S174).



Tab

le 3

. A

sset

Acc

um

ula

tio

n P

lan

fo

r a

Co

lum

bu

s H

igh

Sch

oo

l Tea

cher

wit

h 2

% R

eal I

nve

stm

ent

Ret

urn

in A

ccu

mu

lati

on

Per

iod

(S

om

e Ye

ars

Om

itte

d)

Num

ber

of w

ork

year

s40

Rea

l inv

estm

ent r

etur

ns2.

00%

Hum

an c

apit

al g

row

th (Y

ears

0–2

0)3.

11%

Hum

an c

apit

al g

row

th (Y

ears

21–

40)

0.74

%

Hum

an c

apit

al g

row

th (w

hole

per

iod

)1.

92%

Init

ial s

avin

gs r

ate

10.0

0%

Perc

enta

ge o

f rai

se s

aved

33.0

0%

Savi

ngs

goal

$666

,111

Tota

l sav

ed u

nder

acc

umul

atio

n pl

an$6

66,7

83

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

1$3

7,32

810

.0%

$3,7

33$3

,733

25$7

0,97

021

.0%

$14,

902

$286

,173

238

,490

10.7

4,11

87,

926

3073

,649

21.4

15,7

9139

6,23

53

39,6

8711

.44,

516

12,6

0035

76,4

2921

.916

,714

522,

484

440

,922

12.0

4,92

617

,779

3676

,997

22.0

16,9

0354

9,83

75

42,1

9612

.75,

349

23,4

8337

77,5

7022

.017

,093

577,

927

1049

,182

15.6

7,66

860

,761

3878

,147

22.1

17,2

8560

6,77

015

57,3

2618

.110

,372

115,

148

3978

,728

22.2

17,4

7863

6,38

320

66,8

1820

.213

,523

190,

617

4079

,314

22.3

17,6

7266

6,78

3



Investing in risky assets during one’s working years, and consequently hoping for higher invest-ment returns, does not change the savings goal! It simply changes the amount that needs to be saved each year in order to be on target with the goal, with the understanding that one might not achieve the goal unless one adjusts the savings rate for investment results. The RM is based on riskless investing in retirement, when personal fiscal adjustments are more difficult to implement (because one big lever, the opportunity for addi-tional savings, has been removed). Also note that even with a 2% real return assumption, achieving the savings goal requires a much higher savings rate than most investors ever achieve. This shows one of our major points: To spread the income from 40 years of work over up to 80 years of con-sumption takes a lot of savings, indeed.

If it is surprising how high the required savings rate is, consider the following: You are trying to work for half the years in which you are likely to be consuming. With a real return of zero and no help from Social Security or from a lower standard-of-living expectation in retirement, you would have to save half your income. One more time, just for fun: Save half your income.

It is very difficult to save half one’s income, so we should obviously try to lower that number—through annuitization, positive investment returns, PFAs, and anything else that helps.

There are many variations of these plans or calculations. Our teacher could save less early on and then more during peak earning years, when household expenses can be falling. It is beyond our objective here to analyze individual savings plans. Rather, it has been our purpose to provide a framework, or toolkit, to make and keep a pension promise to oneself—a promise that one can keep with a low risk of failure, that allows for pension payments to oneself from the moment of retirement until the end of life, and that can be used to assess and measure the value (utility) of high-risk, high-return strategies.

There are many career and income paths one can take; our teacher represents a relatively suc-cessful middle-income earner. Appendix A and Appendix B show the retirement planning basics and accumulation plan for two other prototypical workers: (1) a San Diego sanitation worker who earns a lower-middle income and experiences limited human capital growth and (2) a software developer in Austin, Texas, whose human capital grows rapidly and who achieves an upper-middle income later in her career as she works her way up to a management position.

Are These High Savings Rates Achievable?What makes us think that such high savings rates are achievable? Savings rates rise as people get older, with a characteristic “hump” peaking around age 55. The low savings rates quoted in the news media are for the whole population, including retired people in the decumulation stage and low-income households where sav-ings rates are low or negative. Large-scale stud-ies show savings rates of over 30% for the 55–60 age group for the top three income quartiles in the United States (see Attanasio 1994; Bosworth, Burtless, and Sabelhaus 1991). These are the sav-ings rates and the age and income cohort that our examples capture.

But savings rates are quite dynamic and vary significantly by income, wealth, education, age, and market conditions, such as interest rates. That people respond to changing circumstances, includ-ing in their savings behavior, is the core idea of the personal fiscal adjustment.

We would also make the complementary point that effective savings rates under a well-managed DB structure are very high. We believe that the right incentives and structures can lead individuals to replicate those rates. If these sav-ings rates are not realistic, then retiring on 65%–70% of preretirement income at an age close to 65 is not realistic.

ConclusionWe have shown that individuals saving for retire-ment are not facing a hopeless task. Far from it: Defined benefit pension plans, equipped with the same underlying resources, have been able to pro-vide the same required level of retirement income for generations by understanding a basic set of rules for deciding how much to save and how to invest and then sticking to the rules. When DB plans have failed, they have done so because—in the hope of not having to pay the true cost of the retirement benefit—they have broken the rules. The same applies to individuals.

Fulfilling a pension promise to oneself requires a lot of saving, more than most people are accustomed to. When one thinks about spend-ing only, say, 70% of income and saving the rest, it is necessary to ask, Are there other people who make 70% of my income who have perfectly fine lives? What are their consumption patterns? If the answer is no, you should try to improve your human capital position (that is, your income). If the answer is yes, you can make and keep a pen-sion promise to yourself.



The state of Ohio, which sponsors the teacher’s savings plan that we used in our main example, implicitly recognizes the high savings rate needed for a pension promise to oneself by setting a starting savings rate of 20% for teachers entering the DC plan. Of this amount, 10 percentage points are contributed by the employer and 10 percentage points, by the employee. Thus, the true or economic salary is 10% larger than the headline salary, and the total savings rate is slightly less than 20% (it is 20/110, or 18.2%). We left this wrinkle out of the example because we did not think it was necessary to our point. Although 20% is not enough as a career average savings rate, it is a great start, whereas the tiny 6%–9% rates often seen in employee benefit programs are not even in the right ballpark and will lead to penury in the employees’ old age unless supplemented by massive (and tax-disfavored) additional personal savings.

We have also indicated that people can increase their savings rates greatly—by amounts large enough to matter in the present context—through such agree-ments as Save More Tomorrow. This is not just idle speculation. They have already done so in an impres-sive and growing number of work environments.

Presumably, our employees mean enough to us, in terms of the output they produce and the profits they help us generate, that we want them to be enthusiastic and satisfied workers who do not lose sleep over how they can survive in old age. The procedure that people need to follow in order to generate an adequate retirement income exists, and we already know what it is. We owe it to our valued employees to tell them about the procedure and to help them implement it.

This article qualifies for 0.5 CE credit.

Appendix A. San Diego Sanitation WorkerIn this appendix, we provide an example of a sani-tation worker in San Diego (see Tables A1, A2, and A3). We obtained initial and final wage estimates for a San Diego sanitation worker from www.sandiego.gov/empopp/pdf/saltable.pdf, accessed on 27 February 2013. We interpolated the annual wage rates using a constant (exponential) growth model. We assumed a 14% initial savings rate, instead of the 10% rate used in the teacher and soft-ware developer examples, because the sanitation worker’s employer or adviser is likely to be aware that the worker will have a very low human capital growth rate.

Table A1. Retirement Planning Basics for San Diego Sanitation Worker

Initial salary $37,04420th year salary $45,606Final (40th year) salary $56,148Retirement income replacement ratio 70%Retirement income goal $39,304Expected Social Security $20,964Personal pension plan income $18,340Retirement multiple 21.47Lifetime savings accumulation goal $393,760



Tab

le A

2.

Ass

et A

ccu

mu

lati

on

Pla

n f

or

San

Die

go

San

itat

ion

Wo

rker

at

Rea

l 0%

Inve

stm

ent

Rat

e o

f R

etu

rn a

nd

14%

Init

ial S

avin

gs

Rat

e (S

om

e Ye

ars

Om

itte

d)

No.

of w

ork

year

s40

Rea

l inv

estm

ent r

etur

ns0.

00%

Hum

an c

apit

al g

row

th (w

hole

per

iod

)1.

05%

Init

ial s

avin

gs r

ate

14.0

0%

Perc

enta

ge o

f rai

se s

aved

54.0

0%

Savi

ngs

goal

$393

,807

Tota

l sav

ed u

nder

acc

umul

atio

n pl

an$3

94,3

60

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

1$3

7,04

414

.0%

$5,1

86$5

,186

25$4

7,54

322

.8%

$10,

856

$197

,701

237

,431

14.4

5,39

510

,581

3050

,080

24.4

12,2

2625

6,06

13

37,8

2214

.85,

606

16,1

8835

52,7

5225

.913

,669

321,

489

438

,218

15.2

5,82

022

,008

3653

,304

26.2

13,9

6633

5,45

55

38,6

1715

.66,

036

28,0

4337

53,8

6126

.514

,267

349,

722

1040

,678

17.6

7,14

861

,537

3854

,424

26.8

14,5

7136

4,29

315

42,8

4819

.48,

320

100,

770

3954

,993

27.1

14,8

7837

9,17

220

45,1

3521

.29,

555

146,

051

4055

,567

27.3

15,1

8939

4,36

0



Tab

le A

3.

Ass

et A

ccu

mu

lati

on

Pla

n f

or

San

Die

go

San

itat

ion

Wo

rker

at

Rea

l 2%

Inve

stm

ent

Rat

e o

f R

etu

rn in

Acc

um

ula

tio

n

Peri

od

an

d 1

4% In

itia

l Sav

ing

s R

ate

(So

me

Year

s O

mit

ted

)

No.

of w

ork

year

s40

Rea

l inv

estm

ent r

etur

ns2.

00%

Hum

an c

apit

al g

row

th (w

hole

per

iod

)1.

05%

Init

ial s

avin

gs r

ate

14.0

0%

Perc

enta

ge o

f rai

se s

aved

18.0

0%

Savi

ngs

goal

$393

,807

Tota

l sav

ed u

nder

acc

umul

atio

n pl

an$3

93,9

59

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

1$3

7,04

414

.0%

$5,1

86$5

,186

25$4

7,54

314

.9%

$7,0

76$1

92,5

972

37,4

3114

.05,

256

10,5

4630

50,0

8015

.07,

533

250,

863

337

,822

14.1

5,32

616

,083

3552

,752

15.2

8,01

431

7,64

54

38,2

1814

.15,

397

21,8

0236

53,3

0415

.28,

113

332,

111

538

,617

14.2

5,46

927

,707

3753

,861

15.2

8,21

334

6,96

610

40,6

7814

.45,

840

60,1

8838

54,4

2415

.38,

315

362,

220

1542

,848

14.5

6,23

198

,041

3954

,993

15.3

8,41

737

7,88

120

45,1

3514

.76,

642

141,

930

4055

,567

15.3

8,52

039

3,95

9



Appendix B. Austin Software Developer/ManagerIn this appendix, we provide an example of a soft-ware developer/manager in Austin, Texas (see Tables B1, B2, and B3). We assumed that a highly skilled software developer begins work at age 25 at $100,000 per year and has a 10% real growth rate for 10 years (Years 2–11). In Year 11 (at age 36), she transitions to a management position, where her real wage growth rate is 1%. In Year 28 (at age 53), she is terminated and sets up a consulting firm that earns $100,000 in Year 29, $200,000 in Year 30, and $300,000 a year in Years 31–35. In Years 37–40, she experiences wage declines of 13% compounded annually, and after Year 40, she retires. Because she is a high-income worker, her initial savings rate is assumed to be 15%.

Table B1. Retirement Planning Basics for Austin Software Developer/Manager

Initial salary $100,000Peak (28th year) salary $307,178Final (40th year) salary $171,869Retirement income replacement ratio 70%Retirement income goal $120,308Expected Social Security $28,836Personal pension plan income $91,472Retirement multiple 21.47Lifetime savings accumulation goal $1,964,194



Tab

le B

2.

Ass

et A

ccu

mu

lati

on

Pla

n f

or

Au

stin

So

ftw

are

Dev

elo

per

/Man

ager

at

Rea

l 0%

Inve

stm

ent

Rat

e o

f R

etu

rn

No.

of w

ork

year

s40

Rea

l inv

estm

ent r

etur

ns0.

00%

Hum

an c

apit

al g

row

th (Y

ears

2–1

1)10

.00%

Hum

an c

apit

al g

row

th (Y

ears

12–

28)

1.00

%

Hum

an c

apit

al g

row

th (Y

ears

37–

40)

–13.

00%

Init

ial s

avin

gs r

ate

15.0

0%

Perc

enta

ge o

f rai

se s

aved

25.0

0%

Savi

ngs

goal

$1,9

64,1

94

Tota

l sav

ed u

nder

acc

umul

atio

n pl

an$1

,981

,504

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

1$1

00,0

0015

.0%

$15,

000

$15,

000

21$2

86,5

1121

.2%

$60,

695

$926

,001

211

0,00

015

.917

,450

32,4

5022

289,

376

21.2

61,3

9798

7,39

83

121,

000

16.6

20,1

4552

,595

2329

2,26

921

.262

,106

1,04

9,50

44

133,

100

17.4

23,1

1075

,705

2429

5,19

221

.362

,822

1,11

2,32

65

146,

410

18.0

26,3

7010

2,07

525

298,

144

21.3

63,5

451,

175,

871

616

1,05

118

.629

,957

132,

032

2630

1,12

521

.364

,276

1,24

0,14

77

177,

156

19.1

33,9

0316

5,93

627

304,

137

21.4

65,0

131,

305,

161

819

4,87

219

.638

,244

204,

179

2830

7,17

821

.465

,759

1,37

0,91

99

214,

359

20.1

43,0

1824

7,19

729

100,

000

15.0

15,0

001,

385,

919

1023

5,79

520

.548

,270

295,

467

3020

0,00

019

.839

,500

1,42

5,41

911

259,

374

20.8

54,0

4734

9,51

431

300,

000

21.3

64,0

001,

489,

419

1226

1,96

820

.954

,682

404,

196

3230

0,00

021

.364

,000

1,55

3,41

913

264,

588

20.9

55,3

2445

9,52

033

300,

000

21.3

64,0

001,

617,

419

1426

7,23

420

.955

,972

515,

492

3430

0,00

021

.364

,000

1,68

1,41

915

269,

906

21.0

56,6

2757

2,11

935

300,

000

21.3

64,0

001,

745,

419

1627

2,60

521

.057

,288

629,

407

3630

0,00

021

.364

,000

1,80

9,41

917

275,

331

21.0

57,9

5668

7,36

337

261,

000

20.9

54,4

451,

863,

864

1827

8,08

421

.158

,631

745,

994

3822

7,07

020

.346

,132

1,90

9,99

619

280,

865

21.1

59,3

1280

5,30

639

197,

551

19.7

38,9

001,

948,

896

2028

3,67

421

.260

,000

865,

306

4017

1,86

919

.032

,608

1,98

1,50

4



Tab

le B

3.

Ass

et A

ccu

mu

lati

on

Pla

n f

or

Au

stin

So

ftw

are

Dev

elo

per

/Man

ager

at

Rea

l 2%

Inve

stm

ent

Rat

e o

f R

etu

rn in

Acc

um

ula

tio

n P

erio

d

No.

of w

ork

year

s40

Rea

l inv

estm

ent r

etur

ns2.

00%

Hum

an c

apit

al g

row

th (Y

ears

2–1

1)10

.00%

Hum

an c

apit

al g

row

th (Y

ears

12–

28)

1.00

%

Hum

an c

apit

al g

row

th (Y

ears

37–

40)

–13.

00%

Init

ial s

avin

gs r

ate

15.0

0%

Perc

enta

ge o

f rai

se s

aved

13.0

0%

Savi

ngs

goal

$1,9

64,1

94

Tota

l sav

ed u

nder

acc

umul

atio

n pl

an$1

,947

,400

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

Year

Inco

me

Savi

ngs

Rat

eA

nnua

l Sav

ings

Acc

umul

ated

Sav

ings

1$1

00,0

0015

.0%

$15,

000

$15,

000

21$2

86,5

1113

.7%

$39,

264

$759

,939

211

0,00

014

.816

,300

31,6

0022

289,

376

13.7

39,6

1981

4,75

73

121,

000

14.7

17,7

3049

,962

2329

2,26

913

.739

,995

871,

047

413

3,10

014

.519

,303

70,2

6424

295,

192

13.7

40,3

7592

8,84

35

146,

410

14.4

21,0

3392

,703

2529

8,14

413

.740

,759

988,

179

616

1,05

114

.222

,937

117,

494

2630

1,12

513

.741

,146

1,04

9,08

97

177,

156

14.1

25,0

3014

4,87

427

304,

137

13.7

41,5

381,

111,

608

819

4,87

214

.027

,333

175,

104

2830

7,17

813

.741

,933

1,17

5,77

39

214,

359

13.9

29,8

6720

8,47

329

100,

000

15.0

15,0

001,

214,

289

1023

5,79

513

.832

,653

245,

296

3020

0,00

014

.028

,000

1,26

6,57

511

259,

374

13.8

35,7

1928

5,92

131

300,

000

13.7

41,0

001,

332,

906

1226

1,96

813

.836

,056

327,

695

3230

0,00

013

.741

,000

1,40

0,56

413

264,

588

13.8

36,3

9637

0,64

533

300,

000

13.7

41,0

001,

469,

576

1426

7,23

413

.736

,740

414,

798

3430

0,00

013

.741

,000

1,53

9,96

715

269,

906

13.7

37,0

8846

0,18

235

300,

000

13.7

41,0

001,

611,

766

1627

2,60

513

.737

,439

506,

824

3630

0,00

013

.741

,000

1,68

5,00

217

275,

331

13.7

37,7

9355

4,75

437

261,

000

13.8

35,9

301,

754,

632

1827

8,08

413

.738

,151

604,

000

3822

7,07

013

.931

,519

1,82

1,24

419

280,

865

13.7

38,5

1265

4,59

239

197,

551

14.0

27,6

821,

885,

350

2028

3,67

413

.738

,878

706,

562

4017

1,86

914

.224

,343

1,94

7,40

0



Notes1. The same contributions would provide the same average ben-

efit across employees, rather than the same exact benefit for each employee, because many DB plans contain a redistribu-tive feature (with, for example, short-service employees sub-sidizing long-service employees). The same contributions would provide the same exact benefit for each employee if longevity pooling can be achieved at no cost, any redis-tributive aspect of DB plans is ignored, and individuals can invest at institutional (that is, low) fee schedules and with institutional-quality performance. See Waring, Siegel, and Kohn (2007), which is also worth reading because it exam-ines some of the issues in the current article from a behav-ioral finance perspective.

2. Because one big advantage that DB plans have is economies of scale for longevity pooling, what would help individuals is a federal charter, much like that used for national banks, that would allow insurance companies to offer and admin-ister longevity pooling contracts on a national basis, thus providing individuals economical access to longevity pool-ing. At present, the private life-annuity market offers some access to longevity pooling, but pricing in this market is not very competitive and the various offers from insurance com-panies are not easy to evaluate.

3. Munnell (2012) calculated that a worker with annual prere-tirement earnings of $65,000 who contributed 6% with an employer match of 3% should reasonably have accumulated about $363,000 at retirement, well above the 2010 SCF bal-ances. We will show that to make a pension promise to one-self, the savings goal for this $65,000 per year worker needs to be higher—about $565,000 for a 70% income replacement ratio if he expects to receive $1,598 in monthly Social Security benefits, the amount shown on the Social Security Quick Calculator webpage (www.socialsecurity.gov/OACT/quickcalc/index.html) for a 66-year-old worker retiring today at that income level.

4. Alternatively, you could hire a financial intermediary, such as an insurance company, to participate in the investment markets for you.

5. One way in which DB and DC plans can differ is that the state-ment “You cannot consume more than you have saved plus investment return” is true for every individual in a DC plan but is true only in the aggregate for a DB plan. DB plans are often organized to provide transfers between participants—for example, favoring long-service employees at the expense of other employees.

6. If the sponsor sets aside less than the required amount, the plan is considered underfunded, which means that it is implicitly funded by a debt owed by the sponsor to the beneficiaries (which can also be seen as an involuntary loan by the beneficiaries to the sponsor). If this debt is not paid off in the interim through extra contributions to the fund, it is paid off when the sponsor pays out the benefits in full. If the beneficiaries receive anything less than a full payout, the sponsor is considered to have defaulted.

7. The general framework around which asset accumulation for retirement has developed—manifesting itself today in “target-date funds,” “life-cycle funds,” and other products that adjust the asset mix for the participant’s age or time to retirement—originates with Merton (1969), Samuelson (1969), and Bodie, Merton, and Samuelson (1992). The emphasis in the current article, however, is not on how to take risk effi-ciently but on how to avoid it—by saving enough to invest at the riskless rate and still meet one’s liability. This thread in the literature began with Ibbotson, Milevsky, Chen, and Zhu (2007) and Bodie and Taqqu (2011) but is very much a work in progress. If the decision is made to invest in risky assets, it should be based on the judgment that the partici-pant is knowingly taking on shortfall risk, even high levels

of shortfall risk, in an attempt to lower the expected sav-ings rate. The framing of “hope” in contrast to prudent and responsible investment solutions comes from Harris (2009).

8. We do not want to overstate the quality of investment deci-sions made by the affluent. Many such investors have con-centrated positions, large equity allocations at an advanced age, fat fee arrangements, and other portfolio quirks. One of us has a neighbor with assets greater than $1 million who says that she lost half her money in the crash of 2008—in money market funds. She is not alone in confusing money market funds with stock funds, and at any rate, she has no business being mostly in stock funds at her age (over 60 years). No investment professional who has heard this story is the least bit surprised by it.

9. Typically, those in the fifth (bottom) quintile of income find that their postretirement income needs are for the most part met by Social Security. But even these individuals will be helped by the framework we are setting forth. It provides a purpose for savings and a structure that works, and they may place great value on even a modest addition to Social Security payments if it will reliably always be there. The high adoption rate of the Save More Tomorrow program devel-oped by Thaler and Benartzi (2004), where future raises are saved, indicates that a pension promise to oneself can work for all income levels. (Save More Tomorrow is a registered trademark of Shlomo Benartzi and Richard Thaler.)

10. For example, families can pool, or share, their resources. The cohousing movement is a baby step in that direction. While thinking about these possibilities, it occurred to us that the vast amount of perfectly serviceable housing in every well-preserved working-class neighborhood in the United States is a resource crying out to be used by retirees who need a safe, pleasant environment but who place little value on school sys-tems and other amenities that drive housing prices upward.

11. The exact composition of the TIPS portfolio is beyond the scope of this article. In an ideal world where all maturities of TIPS were available and the investor knew when he or she was going to retire, the portfolio would be “bulleted” so that all bonds acquired through the accumulation process would mature on the date of retirement.

12. DCDB is a registered trademark of Allianz Global Investors Solutions LLC.

13. There is a long and well-documented history of pension fail-ures. As recently as 2012, a federal bankruptcy court ruled that the bankrupt city of Central Falls, Rhode Island, could decrease its pension payments to police and fire depart-ment retirees, some by as much as 50%. Central Falls both overpromised—allowing full retirement after 20 years of work—and underfunded its promises. This combination of overpromising and underfunding goes back a long time. For example, on 28 March 1980, after 75 years of making steel for International Harvester tractors and equipment, Wisconsin Steel closed its doors and 3,400 steelworkers lost their jobs. On 31 May 1980, the steelworkers’ pension plan ended in “termination.” On 31 December 1981, the Pension Benefit Guaranty Corporation assumed responsibility for the plan, with a maximum monthly pension payment of $1,261 for those age 65, well below the pension promises made by Wisconsin Steel. In contrast, the fully funded pension plan of Inland Steel (now part of ArcelorMittal), just a few miles away from Wisconsin Steel’s former headquarters, is still making full pension payments. Although employment at both companies shrank dramatically after the 1979–80 reces-sion and productivity-driven recovery, Inland Steel had fully funded its pension plan. In 1998, when Inland Steel was purchased by Ispat International N.V., the fair value of its pension assets was $1.991 billion and its projected benefits obligation, a measure of the liability, was $1.967 billion.



Equally important, Inland Steel made the decision to immu-nize its liabilities by buying US Treasury bonds that matched its expected future pension payments. Hence, Inland Steel made the funding of its pension promise to its workers inde-pendent of its future economic prospects and the prospects of the returns from higher-risk, higher-return assets.

14. Only Alaska and Washington, DC, have DC-only retirement plans for teachers, and these two jurisdictions are too unrepre-sentative to use in a general example. Ohio, Florida, and South Carolina allow teachers to select either a DB or DC plan. Ohio is close to the national average in a large number of demo-graphic and economic dimensions, and Columbus is Ohio’s largest city (i.e., incorporated city, not metropolitan area).

15. We proxy the salary of the early-career teacher using the 10th percentile salary for Columbus high school teachers, that of the mid-career teacher (with 20 years’ experience) with the 75th percentile salary, and that of the near-retirement teacher (with 40 years’ experience and a master’s degree) with the 90th percentile salary. Using a salary well above the median for the mid-career teacher is appropriate because the teacher population is skewed toward short service (that is, many young teachers drop out of the profession). The data are in 2011 dollars and are from www.teachersalaryinfo.com/ohio/teacher-salary-in-columbus-city, accessed on 8 January 2013.

16. The entire analysis is in current (in this case, 2011) dollars, meaning that the teacher can expect inflation-related increases beyond the salary increase from $37,328 to $79,904—and that retirement income requirements will increase correspond-ingly. A retirement income replacement ratio of 70% is con-sidered low by some commentators but is entirely consistent with the requirement to save more than 30% of late-career income; someone who can live on less than 70% of income before retiring can live on 70% of the same number when retired. The analysis in this article ignores taxes.

17. The Social Security Quick Calculator was accessed at www.socialsecurity.gov/OACT/quickcalc/index.html on 23 February 2013. Because of the quirks of the Quick Calculator, we use inputs that reflect a 65-year-old retiring now. The specific inputs are 31 December 1948 for the date of birth, December 2013 for the date of retirement, and “current” pay equal to our worker’s final pay. The Quick Calculator then backfills prior earnings using various assumptions, includ-ing a wage growth rate that is 2 percentage points greater than the national average; because the backfill starts with final pay and works back in time, a high rate of wage growth

means low pay in earlier years. We override this default input with a wage growth rate equal to the national average—in other words, an excess-over-the-average wage growth rate of 0%. Although the workers in our example have specific wage growth rates that differ from one another, the output of the Quick Calculator is unpredictable, and for this purpose, it is better to use a single wage growth rate for all worker catego-ries. The Quick Calculator is particularly inaccurate for high-income earners, whose year-by-year wages are backfilled as if final pay were equal to the Social Security withholding maximum, currently $113,700; thus, early-career wages are in fact higher than in our examples, and the Social Security benefit for high-income earners, likewise, will be higher than in our examples unless benefits are cut in the future.

18. The DCDB rate is a blended low-risk rate (as close to the riskless rate as practical) consisting of (1) the nominal inter-est rate on a portfolio of laddered TIPS held to match cash flow requirements for the first 20 years of retirement and (2) the implied interest rate on a nominal deferred annu-ity that begins its payout in the 21st year of retirement and continues until the retiree’s death (see Sexauer et al. 2012).

19. The exact dollar amount, $666,111, is calculated by multiply-ing the income requirement by the exact RM, 21.4729. For the sake of discussion, we use a rounded number for RM, 21.47.

20. We assume that the worker retires at age 65, earlier than the Social Security full-benefit retirement age, for two reasons: (1) It makes the math easier because the DCDB rate of return (for postretirement investing at the lowest achievable level of risk) assumes retirement at age 65, and (2) retirement at age 65 or even earlier has become customary, even though benefits are somewhat reduced.

21. In this example, we assume that the real riskless rate of return is exactly zero. We recognize that our teacher’s life may not go as smoothly as we have assumed and that rates of income growth will change over time. But what matters here is the basic framework, a framework that works and that provides a usable frame of reference for making savings, investment, and consumption decisions as circumstances change.

22. If we knew that riskless rates would be higher than zero—say, 2%—in the retirement years, that foreknowledge would raise the DCDB yield, reduce the retirement multiple, and thus reduce the savings goal. (The historical real riskless rate is about 2%.) Being able to invest at higher-than-zero real riskless rates during the accumulation phase would also improve the picture.

ReferencesAttanasio, Orazio P. 1994. “Personal Savings in the United States.” In International Comparisons of Household Saving Volume. Edited by James M. Poterba. Chicago: University of Chicago Press.

Benartzi, Shlomo, and Richard Thaler. 2013. “Behavioral Economics and the Savings Crisis.” Science, vol. 339, no. 6124 (March):1152–1153.

Bodie, Zvi, and Rachelle Taqqu. 2011. Risk Less and Prosper. Hoboken, NJ: John Wiley & Sons.

Bodie, Zvi, Robert C. Merton, and William F. Samuelson. 1992. “Labor Supply Flexibility and Portfolio Choice in a Life-Cycle Model.” Journal of Economic Dynamics & Control, vol. 16, no. 3–4 (July–October):427–449.

Bosworth, Barry, Gary Burtless, and John Sabelhaus. 1991. “The Decline in Household Savings: Evidence from Household Surveys.” Brookings Papers on Economic Activity, vol. 1991, no. 1:183–256.

Harris, Brent M. 2009. “Hope Is Not an Investment Strategy.” Project M (http://projectm-online.com/#/global-agenda/hope-is-not-a-retirement-strategy).

Ibbotson, Roger G., Moshe A. Milevsky, Peng Chen, and Kevin X. Zhu. 2007. Lifetime Financial Advice: Human Capital, Asset Allocation, and Insurance. Charlottesville, VA: Research Foundation of CFA Institute.

Merton, Robert C. 1969. “Lifetime Portfolio Selection under Uncertainty: The Continuous-Time Case.” Review of Economics and Statistics, vol. 51, no. 3 (August):247–257.

Munnell, Alicia. 2012. “401(k) Plans in 2010: An Update from the SCF.” Issue Brief (Center for Retirement Research at Boston College), no. 12–13 (10 July).

Samuelson, Paul A. 1969. “Lifetime Portfolio Selection by Dynamic Stochastic Programming.” Review of Economics and Statistics, vol. 51, no. 3 (August):239–246.



Sexauer, Stephen C., Michael W. Peskin, and Daniel P. Cassidy. 2012. “Making Retirement Income Last a Lifetime.” Financial Analysts Journal, vol. 68, no. 1 (January/February):74–84.

Thaler, Richard H., and Shlomo Benartzi. 2004. “Save More Tomorrow: Using Behavioral Economics to Increase Employee Saving.” Journal of Political Economy, vol. 112, no. S1 (February):S164–S187.

Waring, M. Barton, Laurence B. Siegel, and Timothy Kohn. 2007. “Wake Up and Smell the Coffee: DC Plans Aren’t Working—Here’s How to Fix Them.” Journal of Investing, vol. 16, no. 4 (Winter):81–99.

Statement of Ownership, Management, and Circulation(All Periodicals Publications Except Requester Publications)

1. Publication Title 2. Publication Number 3. Filing Date

4. Issue Frequency 5. Number of Issues Published Annually 6. Annual Subscription Price

8. Complete Mailing Address of Headquarters or General Business Office of Publisher (Not printer)

9. Full Names and Complete Mailing Addresses of Publisher, Editor, and Managing Editor (Do not leave blank)Publisher (Name and complete mailing address)

Editor (Name and complete mailing address)

Managing Editor (Name and complete mailing address)

10. Owner (Do not leave blank. If the publication is owned by a corporation, give the name and address of the corporation immediately followed by the names and addresses of all stockholders owning or holding 1 percent or more of the total amount of stock. If not owned by a corporation, give the names and addresses of the individual owners. If owned by a partnership or other unincorporated firm, give its name and address as well as those of each individual owner. If the publication is published by a nonprofit organization, give its name and address.)

11. Known Bondholders, Mortgagees, and Other Security Holders Owning or Holding 1 Percent or More of Total Amount of Bonds, Mortgages, or Other Securities. If none, check box

12. Tax Status (For completion by nonprofit organizations authorized to mail at nonprofit rates) (Check one)

Has Not Changed During Preceding 12 Months

PS Form 3526, September 2006 (Page 1 of 3 (Instructions Page 3)) PSN 7530-01-000-9931 PRIVACY NOTICE: See our privacy policy on www.usps.com

Has Changed During Preceding 12 Months (Publisher must submit explanation of change with this statement)

None

7. Complete Mailing Address of Known Office of Publication (Not printer) (Street, city, county, state, and ZIP+4®)

_

Contact Person

Telephone (Include area code)

The purpose, function, and nonprofit status of this organization and the exempt status for federal income tax purposes:

Full Name Complete Mailing Address

Complete Mailing AddressFull Name

FINANCIAL ANALYSTS JOURNAL

Bimonthly

09/30/2013

$100.006 per year

Jennette Townsend

434-951-5496560 Ray C. Hunt Drive, Charlottesville, VA 22903

560 Ray C. Hunt Drive, Charlottesville, VA 22903

CFA Institute560 Ray C. Hunt DriveCharlottesville, VA 22903

Rodney N. Sullivan560 Ray C. Hunt DriveCharlottesville, VA 22903

Maryann Dupes560 Ray C. Hunt DriveCharlottesville, VA 22903

�

5 7 1 2 0 0

CFA Institute 560 Ray C. Hunt Drive, Charlottesville, VA 22903

PS Form 3526, September 2006 (Page 2 of 3)

Extent and Nature of CirculationAverage No. Copies Each IssueDuring Preceding 12 Months

No. Copies of Single IssuePublished Nearest to Filing Date

a. Total Number of Copies (Net press run)

c. Total Paid Distribution (Sum of 15b (1), (2),(3), and (4))

Mailed In-County Paid Subscriptions Stated on PSForm 3541 (Include paid distribution above nominalrate, advertiser's proof copies, and exchange copies)

d. Free or Nominal RateDistribution (By Mail and Outside the Mail)

Total Free or Nominal Rate Distribution (Sum of 15d (1), (2), (3) and (4)

Total (Sum of 15f and g)

17. Signature and Title of Editor, Publisher, Business Manager, or Owner

13. Publication Title

15.

Percent Paid (15c divided by 15f times 100)

If the publication is a general publication, publication of this statement is required. Will be printed

in the ________________________ issue of this publication.

Date

Free or Nominal Rate Distribution Outside the Mail(Carriers or other means)

Total Distribution (Sum of 15c and 15e)

14. Issue Date for Circulation Data

16. Publication of Statement of Ownership

b. Paid Circulation(By Mail and Outside the Mail)

Copies not Distributed (See Instructions to Publishers #4 (page #3))

Mailed Outside-County Paid Subscriptions Stated onPS Form 3541(Include paid distribution above nominalrate, advertiser's proof copies, and exchange copies)

(1)

(2)

(4) Paid Distribution by Other Classes of Mail Throughthe USPS (e.g. First-Class Mail®)

Paid Distribution Outside the Mails Including SalesThrough Dealers and Carriers, Street Vendors, CounterSales, and Other Paid Distribution Outside USPS®

(3)

(1)

(2)

(3)

Free or Nominal Rate Outside-CountyCopies Iincluded on PS Form 3541

Free or Nominal Rate Copies Mailed at OtherClasses Through the USPS (e.g. First-Class Mail)

e.

f.

g.

h.

i.

Publication not required.

I certify that all information furnished on this form is true and complete. I understand that anyone who furnishes false or misleading information on thisform or who omits material or information requested on the form may be subject to criminal sanctions (including fines and imprisonment) and/or civilsanctions (including civil penalties).

(4)

Free or Nominal Rate In-County Copies Includedon PS Form 3541

FINANCIAL ANALYSTS JOURNAL Sept/Oct 2013

CFA charterholders and other members of the investment profession.

118,141 123,093

230 200

230 200

118,371 1,232,933

2,731 3,057

121,102 126,350

99.8 99.8

121,102 126,350

118,141 123,093

Nov/Dec 2013X

9/27/2013Jennette Townsend, Publishing Solutions Specialist

a pension promise to oneself

Documents