The 2010 Martin Feldstein Lecture – National Bureau of Economic Research

Remarks by Roger W. Ferguson, Jr.

Cambridge, Massachusetts, July 28, 2010

(As prepared for delivery.)

Good afternoon. I am honored to deliver the 2010 Martin Feldstein lecture, honoring Marty’s outstanding tenure as president of the National Bureau of Economic Research.

It is especially appropriate that I will be discussing the issue of retirement security, because throughout the course of his distinguished career, Marty Feldstein has improved our understanding of the U.S. Social Security system and retirement policy more broadly.

Marty has pushed policymakers and researchers to think creatively about how best to fund a universal, guaranteed system of retirement security for all Americans.

In the best traditions of academia and economics, Marty has never let his own views keep him from soliciting the ideas and opinions of others and giving those ideas a careful – and fair – reading.

I currently have the privilege of serving with Marty on President Obama’s Economic Recovery Advisory Board. I think it’s safe to speak on behalf of my fellow members and say that we benefit greatly from Marty’s insight, his experience, and his probing questions.

Marty, thank you for your extraordinary contributions to economics and public policy.

And thank you to the many distinguished economists and other specialists in the room. I’m pleased to know many of you personally. I’m also pleased to rely on your work.

As the chief executive of America’s largest private retirement system, I spend a great deal of time thinking about how to improve the way we help people prepare for retirement and live well in retirement.

Many of you have done research, some of which I’ll cite today, that is changing the way we think about retirement, from the perspectives of employers, employees, retirees, and public policy makers.

We clearly are at a pivotal moment in the national discussion on retirement security. Over the past 30 years, the responsibility for funding retirement and the associated risks has shifted from employers to individuals.

Social Security provides a retirement income floor for retirees, but the system will soon annually pay out more than it collects. It is unclear when policymakers will develop an appetite for making difficult choices to return the system to more stable footing.

The largest generation in American history has begun retiring. Some 76 million Baby Boomers will stop saving for retirement and begin to draw down their accumulated nest eggs.

And as these changes are unfolding, Americans are recovering from a deep plunge in financial markets and a recession that left many if not most people less confident about their ability to achieve financial security.

Today I’d like to approach the question of how we can advance retirement security in America by examining how we arrived at the current situation.

Then I will outline a few core features that should be built into a retirement security plan or system – while keeping in mind the prohibition on making overt policy recommendations.

After outlining these areas, I will consider the question that is increasingly asked by researchers: How can we design retirement plans and systems that ensure an adequate and secure lifetime income?

I will conclude by highlighting a few areas in which we need more research to drive the design of systems and products that accommodate how people actually behave.

Throughout my remarks, I will draw upon research on the economic and policy foundations of the nation’s retirement system and the work of behavioral economists who are contributing fresh perspectives on the habits and motivations of workers and retirees. These perspectives are having a tangible impact on retirement plan design.

The Changing Retirement Landscape

Let me begin with a brief overview of how the contours of the U.S. retirement system have changed over the past few decades.

Most significantly, defined benefit pension systems that previous generations relied on for secure retirement income have become increasingly rare.

According to the Employee Benefits Research Institute, only 33 percent of private-sector employees had access to a defined benefit pension plan in 2008 – down from 84 percent 30 years ago.ⁱ

In their place, a patchwork of individual accounts has placed greater responsibility and risk for retirement income security on individual workers.

401(k) plans, which were initially envisioned as a way for Americans to supplement the pensions made available by their employers, have instead become most workers’ primary means of saving for retirement. In doing so, 401(k)-type products have also fostered a focus on asset accumulation rather than income in retirement.

Many workers now devote the bulk of their attention to their quarterly account balances, with little thought as to how their accumulated savings will translate into a steady stream of income when they retire.

Throughout the 1980s, the 1990s, and much of the past decade, an increasing reliance on market gains to fund retirement accounts was largely unquestioned. But the collapse of financial markets in 2008 and the ensuing global recession have caused many Americans – especially those nearing retirement – to question whether they will be financially secure after they stop working. Indeed, can they stop working and enjoy anything approaching the standard of living to which they are accustomed?

Evidence is mixed, but for many people the answer seems to be no.

Last year, research from McKinsey and Company found that the average American couple will face a savings gap of $250,000 at the time of retirement.ⁱⁱ

Average 401(k) balances have recovered some of their lost ground in the market recovery of late 2009 into 2010. But by any measure, most American workers have not saved enough.

Why has the 401(k) framework failed to adequately prepare workers for retirement? Shortcomings include:

• Lack of participation among many eligible workers;
• Insufficient employer and employee contributions;
• The failure or inability of many participants to implement an appropriate asset allocation strategy;
• The failure to preserve assets for retirement;
• And a lack of annuitization of accumulated assets in retirement to produce a lifelong income stream.

These shortcomings largely result from decisions made by workers. Fundamentally in the 401(k) context, retirement risk burdens – funding, investment, longevity, and mortality – fall disproportionately, often entirely, upon workers who are not equipped to manage such risks.

On the other hand, employers have benefited substantially over the past three decades by jettisoning defined benefit pensions. For instance, it was only by reshaping their retiree health and savings plans that the big three U.S. auto manufacturers could avoid extinction.

Across the country, state and municipal entities are struggling with large gaps between promised retirement benefits and current assets. While only 33 percent of private-sector employees are enrolled in a defined benefit plan, the figure is 79 percent for public-sector employees, according to the Employee Benefits Research Institute.ⁱⁱⁱ

A recent Standard & Poor’s report found that total unfunded public-pension liabilities in the U.S. increased to $457 billion in fiscal year 2008 from $368 billion in 2007.^iv

To cope with this yawning deficit, public employers are incentivizing new hires to select defined contribution retirement options rather than traditional defined benefit pensions.

While the decline of defined benefit pensions and the rise of defined contribution systems has removed an element of security from most Americans’ retirement equation, it is true that a more individualized retirement system is more aligned with the way Americans work today. With more frequent job changes, including spells of independent work, it makes less sense for Americans to have their retirement savings tied to a single employer.

So three facts emerge:

• First, defined benefit pension plans proved unworkable for the vast majority of American businesses and now the tide appears to be shifting in the public sector as well; 
• Second, defined contribution retirement plans, which shift responsibility to individuals, offer less security than defined benefit plans and put too much emphasis on asset accumulation rather than retirement income planning;
• And third, Americans’ work patterns have shifted, so that portability and individual control are attractive features that workers appreciate in the current system.

Given these facts, the challenge for policymakers, financial services companies, economists, employers, and individuals is how to design retirement systems that offer flexibility and individual choice, yet still provide genuine security to individual savers – so that when market declines occur, retirees and near-retirees have an income floor to meet their essential expenses.

In a sense, these retirement plans would be grounded in the realities of the present, while incorporating a measure of security associated with the past.

Core Elements of Retirement Security

At TIAA-CREF, we have focused a great deal of attention on determining which factors are most critical to enjoying a secure retirement. While there are many subsets of individual needs, we have identified three core elements that should be featured in a retirement system providing genuine financial security in retirement.

The first is sufficient funding, which requires greater contributions by plan participants and sponsors. Accumulated savings must account for the full spectrum of expenses that accompany retirement, especially health care expenses, which tend to rise with age.

The second core element is appropriate diversification and asset allocation, which for the average worker requires increased access to financial advice and education along with an appropriately designed investment menu.

The third core element is guaranteed lifetime income in the form of a low-cost, relatively transparent annuity.

Sufficient Funding

I’ll expand on all three areas, starting with funding.

Recent research has clearly demonstrated the overriding importance of retirement plan contribution levels relative to other factors, including asset allocation, for ensuring an adequate level of retirement income.^v

Workers who want to maintain a standard of living close to what they enjoy at the end of their working years should be aiming to replace at least 70 percent of their final salary in retirement. This means that individuals should save at least 10-15 percent of their gross annual income. This figure represents the combined contribution of both employers and employees.

Among 401(k) plans, sponsor contributions average about 3 percent of wages, but are typically in the form of matches dependent on the level of worker contributions (the most common match is 50 cents per dollar, up to the first 6 percent of pay). Among eligible workers who participate, contributions average less than 6 percent of pay for non-highly compensated workers and 7 percent for highly compensated.^vi

Currently half of American workers do not have access to an employer-sponsored plan. Among those who don’t have a workplace retirement plan, fewer than 10 percent have an individual account, such as a traditional or Roth IRA.

Asset retention is another concern. Savings can only grow if they remain in the plan. 401(k) plans, which allow for loans, hardship withdrawals, and lump-sum distributions when workers change jobs, are replete with opportunities for savings to leak out and be used for other purposes.

A major expense looming for retirees – one that requires advanced planning and saving – is health care. The Employee Benefits Research Institute reports that without an employer-sponsored health plan, a couple retiring at age 65 today is projected to need between $200,000 and $800,000 to supplement Medicare and cover out-of-pocket health care expenses during retirement.^vii

That is a staggering sum for most people – and it reinforces the need to accumulate adequate savings for retirement.

Appropriate Diversification and Asset Allocation

The second core element of a secure retirement plan is appropriate diversification and asset allocation.

15-20 fund options should give savers the ability to diversify their investments appropriately. More choices could be confusing and actually lead people to choose less-diversified investments.^viii

Workers often lack the knowledge to choose appropriate investments and diversify their savings.

In Pension Design and Structure: New Lessons in Behavioral Finance, Olivia Mitchell and Stephen Utkus write that participants tend to use “a naïve heuristic – avoid extremes, pick the middle option – rather than maintain a consistent set of well-ordered risk preferences to select from the investments offered.”^ix

Furthermore, according to Mitchell and Utkus, “[M]any plan participants seem to lack well-formed investment preferences.”

This confusion underscores the need for reliable, independent advice. Plans, however, have avoided providing individualized advice. Historically there has been a legislated firewall between plan administration and plan advice. Because of the cost inefficiencies of hiring multiple vendors – one for administration and one for advice – many plan sponsors opted only for administration. Plan sponsors also have been reluctant to assume fiduciary responsibility for the advice provided to their employees.

Recent legislative and regulatory changes have lowered the firewall between plan administration and advice. As a result, the defined contribution market has been moving to provide individualized investment advice in recent years. The percentage of 401(k) plans offering investment advisory services has increased from 37 percent in 2005 to about 50 percent in 2009.^x

Investment and financial education can have a positive impact on participant investment decisions.^xi Nonetheless, an individual cannot be made to learn, and the reality is that many workers do not have the time or interest to develop adequate financial literacy as it relates to retirement planning and saving.

Guaranteed Income

The third core element of a plan that would increase the likelihood of achieving retirement security is to make guaranteed lifetime income options more broadly available.^xii

Guaranteed income in the form of annuities – whose guarantee is subject to the claims-paying ability of the insurance company writing the contract – can make it possible to re-introduce the element of security that has been missing from most private sector 401k plans for the past three decades.

Annuities can be made available within a plan as an accumulation vehicle, as a distribution option upon retirement, or through both the accumulation and distribution phases.

At TIAA-CREF, we encourage people to annuitize at least enough of their savings, so that, combined with Social Security, they have an income stream to meet their basic expenses in retirement – housing, utilities, taxes, food, and health care to the extent those costs are knowable.

In addition, the value of these annuitized payments should ideally be protected – at least partially – against erosion by inflation.

***

These three core elements – sufficient funding, appropriate asset allocation, and guaranteed income – should be at the root of our efforts to help all Americans save for a secure retirement.

We see anecdotal evidence that plan sponsors and plan participants agree. I’ll cite one example from the market with which TIAA-CREF is most clearly identified – the higher education retirement market.

The defined contribution model commonly used in higher education – a model familiar to most of the people in this room – includes, to one degree or another, these three features.

This model is, on the whole, serving the academic community quite well.

Recent research from the TIAA-CREF Institute found that 80 percent of higher education employees – and I should note that this includes employees across higher education, not just TIAA-CREF participants – 80 percent described themselves as “somewhat confident” or “very confident” that they will have enough money to live comfortably in retirement.

Among all U.S. workers, just 54 percent report such confidence.^xiii

If we’re doing something right in academia, we ought to think about ways to emulate that success in other sectors.

Behavioral Aspects of the Retirement Challenge

This leads me to the next question we should consider: Even if we design retirement plans, or a comprehensive retirement system, that provides more options for people to save appropriately and turn those savings into a steady stream of guaranteed income, are individuals likely to take advantage of those options?

Behavioral economists have been doing extensive research to determine how system design can influence participant behavior. Much of the literature has focused on overcoming, or leveraging, apparently negative tendencies such as inertia and risk aversion with new plan features and approaches, including auto-enrolling workers in a plan and framing choices in a way that motivates optimal decision-making.

With 401(k) plans, participation has historically been voluntary on the part of eligible workers. One-quarter of those eligible choose not to participate in a plan.^xiv Some choose not to participate for financial reasons; others for behavioral reasons, such as simple inertia, which seems to be a powerful influencer in retirement planning.

In Aging Gracefully: Ideas to Improve Retirement Security in America, Peter Orszag states, “We should recognize the power of inertia in human behavior and use it to promote rather than hinder saving.”^xv

Economists, plan designers, and policymakers have looked for ways to use inertia to keep people on the right track, anticipating that they will not muster the motivation to deviate from it.

We’re all familiar with the classic line from the movie Field of Dreams: “If you build it they will come.”

These efforts might be characterized by a slightly more aggressive mantra: “If you build it and put them in it, they won’t leave.”

One idea that has been applied more broadly since the Pension Protection Act of 2006 is to make the process of saving as automatic as possible.

Plan sponsors have begun automatically enrolling eligible workers in their 401(k) plan; if they do not wish to participate, workers must pro-actively opt-out. Evidence suggests that changing the participant’s decision from opt-in to opt-out has been successful in increasing participation rates.^xvi

According to the Government Accountability Office, auto-enrollment can increase participation in employer-sponsored plans to as high as 95 percent. But only 16 percent of employer-sponsored plans feature auto-enrollment, with higher rates among larger plan sponsors.^xvii

The Employee Benefits Research Institute reports that auto-enrollment has increased the number of near-retirees who are on track to have enough money to pay for basic expenses and health care costs – from about 41 percent in 2003 to a little over half today.^xviii

But auto-enrolling individuals in a plan is not necessarily a panacea. Issues still remain, again often driven by inertia. Harvard’s James Choi and David Laibson and colleagues have identified that while participation rates increase when employers auto-enroll employees, the overall level of savings may not increase because default contribution levels tend to be fairly low and employees remain “anchored” at these low contribution levels and in overly conservative funds.^xix

In an attempt to improve asset allocation, plan providers and sponsors have made available the lifecycle, or target date, fund – an investment vehicle that changes asset allocation over time, based on a target retirement year, without the saver ever having to make another decision.

In the wake of the financial crisis, however, questions have emerged about the design of lifecycle funds, as some funds may have been too aggressive for workers nearing retirement.

The idea of auto-escalation – automatically increasing an individual’s contribution rate over time – has gained broader appeal, though it is not yet as widely incorporated into retirement plans as auto-enrollment.

Chicago’s Richard Thaler and Shlomo Benartzi of UCLA developed the Save More Tomorrow program, which allows workers to schedule automatic increases in their savings rate at set future dates. By pushing the income effect into the future, Save More Tomorrow mitigates savers’ tendency to overvalue the present utility of their earnings and undervalue their future savings. It also harnesses the power of inertia, as savers have to opt out of decisions they make years in advance.

In Thaler and Benartzi’s first case study, participants increased their set-aside rate from 3.5 percent to more than 13 percent. Dr. Benartzi has recently written that more than half of large employers in the United States now offer the program.^xx

Just as important as encouraging appropriate decision making in the saving and accumulation stage is the need to encourage optimal choices upon retirement.

If we accept that annuitization – or the purchase of guaranteed lifetime income – is the preferred choice for someone entering retirement – particularly if that person hasn’t shown any interest in being an active manager of his or her savings to date – how can we influence their decision-making and encourage them to move in that direction?

Employers have shown reluctance to include annuities as a distribution option. All 401(k) plans offer a lump sum distribution option, but only 14 percent offer the ability to annuitize assets.^xxi

A commonly cited reason for plan sponsor’s reluctance to offer annuities is fiduciary uncertainty. Regulatory clarity could go some way toward encouraging more employers to make annuities available.

If annuities are more widely available, will employees purchase them?

William Gale and David John of the Retirement Security Project note that private annuities account for just two percent of retirees’ household income.^xxii

This low annuity purchase rate has been labeled the “annuity puzzle.” Researchers are trying to solve it.

Paul Yakoboski of the TIAA-CREF Institute has found that saving through an annuity during the accumulation phase substantially increases the likelihood of purchasing an annuity at retirement. Retirees who have annuitized their retirement savings are more than twice as likely, compared with retirees who have not annuitized, to have saved through an annuity in a DC plan while working.

Furthermore, it is also apparent that professional advice matters. Among retirees who have not annuitized, more than half have worked with a financial advisor and most of these tend to follow the advice received. But only 5 percent were advised to annuitize.

By contrast, 71 percent of retirees who annuitized report working with a financial advisor in deciding to purchase an annuity or implementing a decision to buy.

The good news is that there is no evidence of buyer’s remorse with regard to payout annuities; 89 percent of annuitants report being very or somewhat satisfied with the decision to purchase a payout annuity.^xxiii

Jeff Brown has explored the question of how the decision to annuitize is affected by the way the choice is framed.

The key distinction is whether the annuity is presented in the context of investment or consumption. Jeff and his colleagues have found that “when products are presented in a frame that emphasizes consumption consequences, 72 percent of survey respondents prefer a life annuity to a savings account. In contrast, when the same products are presented using investment terminology, only 21 percent of respondents prefer the annuity.”

He explains: “When consumers think in terms of consumption, they perceive the life annuity as offering valuable insurance against the risk of outliving one’s resources. However, when they think in investment terms, they view life annuities as increasing risk without increasing return, because of the potential for variation in the total value of payments based on how long they live.”^xxiv

The desire to avoid what is perceived as a loss has been identified as a powerful motivator for individuals.

Daniel Kahneman and Amos Tversky’s groundbreaking research on loss aversion demonstrated that people are about twice as likely to shy away from potential losses as they are to seek potential gains.

Recently, Columbia’s Eric Johnson expanded on Kahneman and Tversky’s research by focusing specifically on retirees. He uncovered what he calls “hyper loss aversion” among retirees, who were up to five times more loss averse than the average person.^xxv

Interestingly, this hyper sensitivity to loss does not translate into a desire to purchase guaranteed income. Instead, Johnson has found, retirees who exhibit hyper loss aversion are less likely to annuitize because they see giving up immediate control of their savings as another type of loss.

The Need for Further Research

Despite actions taken to increase savings and highlight the benefits of guaranteed income, a distressingly large fraction of people pay almost no heed.

Annamaria Lusardi and Jason Beeler found in the year before the financial crisis that 30 percent of Baby Boomers – the people closest to retirement – had given no thought to retirement planning. As we would expect, “those that fail to plan have accumulated much less wealth than those who did some planning.”^xxvi

How can we reach the nonplanners?

And how can we reach more of the individuals who are planning, but who lack the knowledge to make informed decisions and may feel paralyzed by the process or turn to Olivia Mitchell’s “naïve heuristics” for making retirement-related decisions?

How can we convince people that their dollars could benefit them even more if saved today rather than spent?

New sources of information on these issues are emerging.

For instance, RAND, Dartmouth, and the Wharton School joined together last fall to launch the Financial Literacy Center. The Center, supported by the Social Security Administration, is led by Annamaria Lusardi, Olivia Mitchell, and Arie Kapteyn. They and their colleagues will drive research and development on key financial literacy topics related to retirement savings and planning.

Boston College has launched a similar center, with the goal of developing educational materials and programs to help people make better financial choices.

A novel approach we’re taking at TIAA-CREF is to tap the wisdom of the crowds in seeking answers to the savings challenge. We launched a contest on Facebook that seeks ideas to increase the national savings rate. For those of you who will be in the classroom this fall, I encourage you to direct your students to our Facebook page for more information.

We – and I’m sure many of our competitors – are also at work on products that meet the retirement savings and lifetime income needs of people with diverse goals, financial abilities, and levels of participation.

And scholars, of course, continue to offer innovative ideas. The Retirement Security Project, for instance, has touted so-called “trial income” and “phased or incremental acquisition” approaches to annuities. These more flexible annuity options would remove the “all or nothing” and “now or never” aspects to choosing an annuity versus maintaining personal control over savings.^xxvii

These approaches could alleviate the loss-aversion that may be keeping some retirees from purchasing annuities, while also introducing retirees to the benefit of regular monthly income and winning them over to the concept … or simply hoping their inertia prevents them from reversing their decision.

We need further research that we can apply to drive innovations in retirement plan design, to aid policymakers in strengthening the legal and regulatory framework that supports retirement planning, and to cultivate broader financial knowledge in America.

Among the questions needing further exploration:

• What is the appropriate mix of automatic plan features with education or advice?
• At what point in a career is it advisable for a participant to stop being an “auto-bot” and become a planner who saves and invests according to his or her plan?
• And how much income should people save? Consensus on this figure has been elusive, but if we can clarify the goal for most workers, we may have more success in helping people reach that benchmark.

Conclusion

Earlier this year, President Obama said plainly that all Americans should have access to “a secure retirement even if you’re not rich.”^xxviii

For too many people, financial security that lasts a lifetime is beyond their reach.

A comprehensive and sustainable system that provides genuine retirement security likely will incorporate elements of defined benefit plans and defined contribution plans. It would have to be affordable for employers and give employees a measure of security through guaranteed income that will last a lifetime.

At least as important as the design of the plan is the behavior of the people in it.

Behavioral economists are exploring not only how people should act given a certain set of facts if they are rational creatures pursuing their own best interest, but also how they do act, as individuals prone to biases, passions, and proclivities that are perhaps even more determinative of their actions than reason is.

In doing so, economists are shedding new light on how to design retirement systems to harness these tendencies.

Retirement planners need these insights. As Olivia Mitchell and Stephen Utkus have written, “[R]etirement saving decisions are at least an order of magnitude more complex than other economic decisions.” The central question is: “How can the various stakeholders strike the best balance between encouragement and compulsion?”^xxix

Through your research, and the research of colleagues throughout the fields of economics, psychology, sociology, and other disciplines, we are formulating answers.

Thank you very much.

And I eagerly anticipate your questions.