“Repeat” Performance

Nevin AdamsBy Nevin Adams, EBRI

A few months back, I was intrigued to catch several episodes of “Cosmos,” an updated version of the classic 1980 Carl Sagan series.  Along with the significantly expanded and enhanced visuals and (to me, anyway, generally annoying) animations, the series recounted the work, travails and accomplishments of Edmond Halley, who, even today, is probably best remembered for the comet whose 75-year cycle he identified and which still, as Halley’s Comet, bears his name.

Halley wasn’t the first to see the comet, of course – in fact, it had been recorded by Chinese astronomers as far back as 240 BC, noted subsequently in Babylonian records, and perhaps most famously shortly before the 1066 invasion of England by William the Conqueror (who claimed the comet’s appearance foretold his success).  Halley noted appearances by the comet in 1531, 1607 and 1682, and based on those prior observations – and the application of the work and mathematical formulas of his friend Isaac Newton – predicted the return of the comet in 1758, which it did, albeit 16 years after his death in 1742.

Of course, the importance of repeated, measured observations isn’t restricted to celestial phenomena.  Consider that individual retirement accounts (IRAs) currently represent about a quarter of the nation’s retirement assets; and yet, despite an ongoing focus on the accumulations in defined benefit (pension) and 401(k) plans that have, via rollovers, fueled a significant amount of this growth, a detailed understanding as to how these funds are actually used during retirement has, to date, not been as well understood.

To address this knowledge shortfall, the Employee Benefit Research Institute has developed the EBRI IRA Database, which includes a wealth of data on IRAs including withdrawals or distributions, both by calendar year and longitudinally, which provides a unique ability to analyze a large cross-sectional segment of this vital retirement savings component, both at a point in time and as the individual ages and either changes jobs or retires.  Indeed, as a recent EBRI publication notes, the rate of withdrawals from these IRAs is important in determining the likelihood of having sufficient funds for the duration of an individual’s life, certainly where these balances are a primary source of post-retirement income.

Previous EBRI reports[i] have explored this activity for particular points in time, but a recent EBRI analysis[ii] looked for trends in the withdrawal patterns of a longitudinal three-year sample of individual post-retirement withdrawal activity, specifically those age 70 or older (in 2010), the point at which individuals are required by law to begin withdrawing money from their IRAS.

The EBRI analysis concluded that, when looking at the withdrawal rates for those ages 70 or older, the median of the average withdrawal rates over a three-year period indicated that most individuals are withdrawing at a rate that not only approximates what they are required by law to withdraw, but at a rate that is likely to be able to sustain some level of post-retirement income from IRAs as the individual continues to age.

Furthermore, the report notes that an examination of these trends over this period suggests that, based on the resulting distribution of average withdrawal rates over time as a function of the initial-year withdrawal rate, the initial withdrawal rate for those in this age group appeared to be one that these individuals are likely to continue to make the next year.

Of course, while the median withdrawal rates suggest many individuals would be able to maintain the IRA as an ongoing source of income throughout retirement, further study is needed to see if these individuals are maintaining those withdrawal rates over longer periods of time.  Moreover, the integration of IRA data with data from employment-based defined contribution retirement accounts currently underway as part of initiatives associated with EBRI’s Center for Research on Retirement Income (CRI) will allow for an even more comprehensive picture of what those who may have multiple types of retirement accounts do as they age through retirement.

And we won’t have to wait 75 years to see how it turns out.

  • Notes

[i] See “IRA Withdrawals, 2011” online here.   See also ““Take it or Leave it? The Disposition of DC Accounts: Who Rolls Over into an IRA? Who Leaves Money in the Plan and Who Withdraws Cash?” online here

[ii] See “IRA Withdrawals in 2012 and Longitudinal Results, 2010–2012” online here

The Status Quo

Nevin AdamsBy Nevin Adams, EBRI

While the prospects for “comprehensive tax reform” may seem remote in this highly charged election year, the current tax preferences accorded employee benefits continue to be a focus of much discussion among policymakers and academics.

The most recent entry was a report by the Urban Institute which simulated the short- and long-term effect of three policy options for “flattening tax incentives and increasing retirement savings for low- and middle-income workers.”  The report concluded that “reducing 401(k) contribution limits increases taxes for high-income taxpayers; expanding the saver’s credit raises saving incentives and lowers taxes for low- and middle-income taxpayers; and replacing the exclusion for retirement saving contributions with a 25 percent refundable credit benefits primarily low- and middle-income taxpayers, and raises taxes and reduces retirement assets for high-income taxpayers.”

However, and to the authors’ credit, the report also noted that “the behavioral responses by both employers and employees will affect the final savings outcomes achieved under reform but are beyond the scope of our estimates[i].”

In previous posts, we’ve highlighted the dangers attendant with relying on simplistic retirement modeling assumptions, the application of dated plan design information to future accumulations, and the choice of adequacy thresholds that, while mathematically accurate, seem unlikely to provide a retirement lifestyle that would, in reality, feel “adequate.”

However, one of the more pervasive assumptions, particularly when it comes to modeling the impact of policy and/or tax reform changes, is that, regardless of the size and scope of the changes proposed, workers – and employers – will generally continue to do what they are currently doing, and at the current rate(s), for both contributions and/or plan offerings.  Consequently, there is talk of restricting participant access to their retirement savings until retirement, with little if any discussion as to how that might affect future contribution levels, by both workers and employers, and there are debate about modifying retirement plan tax preferences as though those changes would have no impact at all on the calculus of those making decisions to offer and support these programs with matching contributions.  Ultimately, these behavioral responses might not only impact the projected budget “savings” associated with the proposals, but the retirement savings accumulations themselves.

EBRI research has previously been able to leverage its extensive databases and survey data (including the long-running Retirement Confidence Survey) to both capture  potential responses to these types of proposals and, more significantly, to quantify their potential impact on retirement security today and over the extended time periods over which their influence extends.  In recent months, that research has provided insights on the full breadth of:

  • A retirement savings cap[ii],
  • The proportionality of savings account balances with incomes[iii], and
  • The impact of permanently modifying the exclusion of employer and employee contributions for retirement savings from taxable income, among other proposals[iv].

While we can’t be certain what the future brings, considering the likely responses to policy changes is a critical element in any comprehensive impact assessment – not only because the status quo is rarely a dependable outcome, but because, after all, those who assume the status quo are generally looking to change it.

  • Notes

[i] From Urban Institute and Brookings Institution:  Flattening Tax Incentives for Retirement Saving“Our findings should be interpreted with caution. Actual legislation for flattening tax incentives requires more than the simple adjustments discussed here. For instance, if a credit-based approach is used, then the laws would need to ensure some recapture of those benefits for those who made contributions one year and withdrew them soon thereafter.

Additionally, the behavioral responses by both employers and employees will affect the final savings outcomes achieved under reform but are beyond the scope of our estimates. For instance, employees may save more in response to improved incentives, in which case the benefits to low lifetime income households would be greater than we find. On the other hand, employers might reduce their contributions in response to some of the policy changes outlined. In this case, the tax and savings benefits we find would be overstated.  While our policy simulations are illustrative, addressing these behavioral responses would be a chief concern in tailoring specific policies to create the best incentives.”

[ii] See “The Impact of a Retirement Savings Account Cap

[iii] See “Upside” Potential

[iv] See “Tax Reform Options: Promoting Retirement Security”, and “Modifying the Federal Tax Treatment of 401(k) Plan Contributions: Projected Impact on Participant Account Balances

Map “Quests”

Nevin AdamsBy Nevin Adams, EBRI

I’ve been hooked on the convenience of GPS systems ever since the first time one was included in the price of a rental car on a family trip in unfamiliar territory. After all, it combines the opportunity to tinker with electronic gadgetry alongside the convenience of not having to do much in the way of pre-planning trip routes—not to mention avoiding the need to stop and ask for directions (that is frequently associated with not doing much in the way of pre-planning trip routes).

There are, of course, horror stories about drivers who have blindly followed GPS instructions without paying attention to the evidence of their eyes. My family still chuckles at the memory of a trip where we were running late to our plane, and the rental car GPS, based on what appeared to be an outdated address for the return office, kept directing us to an address that was not only miles from the real office, but a place from which I wondered if we might never return.

As a growing number of Americans near, and head into, retirement, policymakers, retirement plan sponsors, and individual workers alike increasingly wonder—will Americans have enough to live on when they retire? Unfortunately, as a recent EBRI publication[1] explains, the answers provided are as diverse, and sometimes disparate, as the projection models that produce those results.

While it is not always clear from their results, some of those models limit their analysis to households already retired, while others focus on households still working, but old enough that reasonably accurate projections regarding their future wages and prospects for accumulating retirement wealth are obtainable. Still others attempt to analyze the prospects for all working households, including those whose relative youth (and distance from retirement) makes accurate, long-term predictions somewhat problematic.

Moreover, there are varied definitions of retirement income adequacy. As the EBRI report explains, some either (1) model only the accumulation side of the equation and then rely on some type of preretirement income replacement rate measure as a threshold for success, or (2) make use of a so-called “life-cycle” model that attempts to smooth/spread some type of consumption-based utility over the decision-maker’s lifetime.

The problem with the former is that, since very few households annuitize all (or even most) of their individual accounts in retirement, a replacement-rate focus overlooks the potential risk of outliving their income (longevity risk). And while the annuity purchase price relied upon in a replacement-rate target does depend on an implicit assumption with respect to (at least some) future market returns, it does not typically account for the potential investment risk. Additionally, and as previous EBRI research has demonstrated, one of the biggest financial obstacles in terms of maintaining retirement income adequacy for households that might otherwise have sufficient financial resources at retirement age is the risk of long-term care costs for a prolonged period. In the real world, few retirees have long-term care insurance policies in place that would cover the potentially catastrophic financial impact of this exposure—and thus, simply adding the cost of long-term care insurance into a replacement-rate methodology will vastly underestimate the potential severity of this exposure.

As for the life-cycle smoothing model, the EBRI report notes that approach typically produces extraordinarily low levels of “optimal” savings for low-income individuals at retirement, and while some households may, in fact, have no choice but to subsist at those levels in retirement, from a public policy perspective EBRI chose instead to establish a threshold that would allow households to afford average expenditures (for retirees in the appropriate income category) throughout their retirement, while at the same time accounting for the potential impact of uninsured long-term care costs.

EBRI’s Retirement Security Projection Model®[2] takes a different—arguably unique and more realistic—perspective. Rather than relying on an individual’s projected ability to achieve an arbitrarily designated percentage of his or her preretirement income as a proxy for retirement income adequacy, RSPM grew out of a multiyear project to analyze the future economic well-being of the retired population at the state level, focused on identifying the point at which individuals would run short of money and perhaps become a financial obligation of the state.

As valuable a resource as a GPS can be, it can quickly become a nuisance—or worse—if the input destination point is incorrect, or the mapping system is out of date. Similarly, those who want a financially secure retirement may find that relying on a model based on flawed assumptions or outdated “destinations” may find themselves short of their goal and with little time to do anything about it.


[1] See ““’Short’ Falls: Who’s Most Likely to Come up Short in Retirement, and When?” online here.

[2] A brief description of EBRI’s Retirement Security Projection Model® can be found online here.

“Out” Takes

Nevin AdamsBy Nevin Adams, EBRI

My first car wasn’t anything special, other than it was my first car. It was an older model Ford, ran reasonably well, with one small problem— it went through oil almost as quickly as it did gasoline. At first I attributed that to being a function of the car’s age, but as the leakage grew, I eventually dealt with it by keeping a couple of quarts of oil in the trunk “just in case.” Eventually, I took the car to a dealership—but by the time they finished estimating the cost of a head gasket repair, let’s just say that, even on my limited budget, I could buy a LOT of oil by the quart, over a long period of time, and still be ahead financially.

“Leakage”—the withdrawal of retirement savings via loan or distribution prior to retirement— is a matter of ongoing discussion among employers, regulators, and policy makers alike. In fact, EBRI Research Director Jack VanDerhei was recently asked to present findings on “The Impact of Leakages on 401(k) Accumulations at Retirement Age” to the ERISA Advisory Council in Washington.[1]

EBRI’s analysis considered the impact on young employees with more than 30 years of 401(k) eligibility by age 65 if cashouts at job turnover, hardship withdrawals (and the accompanying six-month suspension of contributions) and plan loan defaults were substantially reduced or eliminated. The analysis assumed automatic enrollment and (as explicitly noted) no behavioral response on the part of participants or plan sponsors if that access to plan balances was eliminated.

Looked at together, EBRI found that there was a decrease in the probability of reaching an 80 percent real income replacement rate (combining 401(k) accumulations and Social Security benefits) of 8.8 percentage points for the lowest-income quartile and 7.0 percentage points for those in the highest-income quartile. Said another way, 27.3 percent of those in the lowest-income quartile (and 15.2 percent of those in the highest-income quartile) who would have come up short of an 80 percent real replacement rate under current assumptions WOULD reach that level if no leakages are assumed.

The EBRI analysis also looked at the impact of the various types of “leakage” individually. Of loan defaults, hardships, and cashouts at job change, cashouts at job change were found to have a much more serious impact on 401(k) accumulation than either plan loan defaults or hardship withdrawals (even with the impact of a six-month suspension of contributions included). The leakages from cashouts resulted in a decrease in the probability of reaching an 80 percent real replacement rate of 5.9 percentage points for the lowest-income quartile and 4.5 percentage points for those in the highest-income quartile. That effect from cashouts—not loans or hardship withdrawals—turns out to be approximately two-thirds of the leakage impact.

However, and as the testimony makes clear, it’s one thing to quantify the impact of not allowing early access to these funds—and something else altogether to assume that participants and plan sponsors would not respond in any way to those changes, perhaps by reducing contributions,[2] potentially offsetting some or all of the prospective gains from restricting access to those funds.

Because ultimately, whether you’re dealing with an old car or your retirement savings account, what matters isn’t how much “leaks” out—it’s how much you put in, and how much you have to “run” on.


[1] EBRI’s testimony for the ERISA Advisory Council, U.S. Department of Labor Hearing on Lifetime Participation in plans is available online here.  

[2] An EBRI/ICI analysis published in the October 2001 EBRI Issue Brief found that, “[o]n average, a participant in a plan offering loans appeared to contribute 0.6 percentage point more of his or her salary to the plan than a participant in a plan with no loan provision.” Testimony provided to the ERISA Advisory Council testimony notes that it’s likely that a similar relationship exists with respect to the availability of hardship withdrawals. See “Contribution Behavior of 401(k) Plan Participants,” online here.

“Short” Changed?

Nevin AdamsBy Nevin Adams, EBRI

My wife is fond of recounting one of our early dates when we ran out of gas.  Now, we were in the heart of a Chicago suburb at the time, not the middle of nowhere, and while the hour was late, I continue to maintain that it was a simple case of my misreading the gas gauge in a relatively new car with which I hadn’t yet gained a full appreciation for just how far I could push such things.  My wife, of course, has always accused me of a more “nefarious” purpose.

It would be more difficult to explain such an outcome these days.  We’ve gone from vehicles that simply had a floating gauge and a range of red at the 1/8 tank line, to those that have a solid and then a blinking yellow light, to ones that beep and flash and tell you how many miles you have left before you run out.

As inconvenient as running out of gas late at night can be, it surely pales in comparison to the prospects of running short of money in retirement.  EBRI has, for more than a decade now, used highly sophisticated modeling techniques to gauge the retirement readiness of baby boomers and Gen Xers.  One of the primary outputs of EBRI’s Retirement Security Projection Model (RSPM)[i] is the production of Retirement Readiness Ratings (RRRs), which represent the percentage of simulated life-paths that do not run short of money in retirement.  The 2014 version of RSPM found that over half of baby boomer and Gen Xer households would not run short of money in retirement.  However, when the results were analyzed by preretirement wage quartile, we found that while 86.4 percent of the highest income quartile were projected to not run short, just 16.8 percent of the lowest income quartile would not.

While it is useful, certainly from a public policy perspective, to know not only how many but also what types of individuals are projected to run short of money in retirement, it begs the question: when will they run short?

A recent EBRI Notes article[ii] provides new results showing how many years into retirement baby boomer and Gen Xer households are simulated to run short of money, by preretirement income quartile and for a variety of assumptions, as well as taking into account the impact of the potentially catastrophic expenses of nursing home and home health care expenses.  Not surprisingly, it finds that those in the lowest income brackets are most likely to run short.

Moreover, while some in all income brackets—including the highest—may run short at some point during their retirement, the EBRI analysis also found that, when nursing home and home health care expenses are factored in, the number of households in the lowest income quartile that are projected to run short of money within 20 years of retirement is considerably larger than those in the other three income quartiles combined.

The EBRI analysis provides valuable insights for policymakers, providers and employers alike because, whether you’re concerned about running out of gas short of your destination – or short of money in retirement – it’s important that your gauges be accurate, and appropriate to the vehicle in which you’re riding there.

  • Notes

[i] A Brief Description of EBRI’s Retirement Security Projection Model® is available here.

[ii] The June EBRI Notes article, ““Short” Falls: Who’s Most Likely to Come up Short in Retirement, and When?” is available online here.

The Hassle Factor

Nevin AdamsBy Nevin Adams, EBRI

Much is made these days of the application of behavioral finance and the implications for plan design, as well as the role of choice architecture in helping workers make “better” (if not more informed) benefit decisions.  Valuable as these insights have been, I think much of human behavior (or lack thereof) in these matters can be more simply explained.

What’s at work is a concept a friend of mine described to me more than 20 years ago – something he called “the hassle factor.”  It was a philosophy he routinely applied in many aspects of his personal and professional life.  Simply stated, presented with a choice between doing something that is hard, time-consuming, complicated, or even inconvenient, and doing something else, my friend – and, in fairness, human beings generally seem to be – inclined to opt for the latter.

Of course, the “hassle factor” CAN be trumped by exterior needs or forces, as anyone who has endured the long lines at the DMV or sat through the background music on an interminably long customer service line can attest.  That said, things like an unduly complicated 401(k) enrollment form/process can certainly serve as a barrier to plan entry, and there’s every reason to expect that the same might apply when it comes time to exit the plan.

Job change is a point in time at which a lot of important decisions are made—some voluntary and some forced upon us—and the disposition of one’s retirement savings account certainly looms large among them.  A recent EBRI Notes article examined what workers age 50 and above did with their defined contribution account balances at the point of job change, looking at data from the Health and Retirement Study (HRS), a study of a nationally representative sample of U.S. households with individuals age 50 and over.  EBRI analyzed responses from 2008 and 2010 for this study.

In terms of demographic characteristics, no significant difference was found between men and women in terms of their DC account balances and what they chose to do with them at job change.  And while married or partnered individuals were less likely to withdraw their assets and more likely to roll them over into an IRA than singles, the differences were small.

The EBRI analysis did find that a decision to take a withdrawal in cash declined with higher account balances, higher incomes, existing ownership of an IRA, and higher financial wealth. Not surprisingly, the decision to cash out rose with individual debt levels.

However, among those who left their employer but remained in the workforce, the most common outcome was to leave their retirement account balance with their prior employer’s plan.  The EBRI report notes that, unlike the outcomes detailed above, there was no clear trend between the financial variables, and the decision to leave those DC balances in the prior employer plans.

As for what might explain that outcome, the report noted that it might simply be a decision to postpone taking the money until it was needed, or that there “may be behavioral factors, such as inertia, driving what might be seen as a ‘non-decision.’”

Or, as my friend might have been inclined to say, a non-decision based on the “hassle factor.”

  • Notes

“Take it or Leave it? The Disposition of DC Accounts: Who Rolls Over into an IRA? Who Leaves Money in the Plan and Who Withdraws Cash?” is published in the EBRI May Notes, available here.

Source “Spots”

Nevin AdamsBy Nevin Adams, EBRI

Individual retirement accounts (IRAs) have been around a long time – since the Employee Retirement Income Security Act of 1974 (ERISA), in fact[i].

Today IRAs represent nearly $6 trillion in assets, approximately a quarter of the $23.7 trillion in retirement plan assets in the nation. As an account type, they currently hold the largest single share of U.S. retirement plan assets with, as a recent EBRI publication notes, a substantial (and growing) portion of these IRA assets having originated in other tax-qualified retirement plans, such as defined benefit (pension) and 401(k) plans. Recognizing not only the significant growth but the increasing importance of these accounts to individual retirement security, the Department of Labor has proposed expanding ERISA’s fiduciary protections to these accounts.

To help better understand the trends driving this critical retirement savings component, the EBRI IRA Database, an ongoing project that collects data from IRA plan administrators across the nation, was created. For year-end 2012, it contained information on 25.3 million accounts owned by 19.9 million unique individuals, with total assets of $2.09 trillion. The EBRI IRA Database is unique in its ability to track individual IRA owners with more than one account, thereby providing a more accurate measure of how much they have accumulated in IRAs.

For example, a recent EBRI analysis[ii] notes that the average IRA account balance in 2012 in the EBRI IRA database was $81,660, while the average IRA individual balance (all accounts from the same person combined) was $105,001. Overall, the cumulative IRA average balance was 29 percent larger than the unique account balance.

While almost 2.4 million accounts in the EBRI IRA database received contributions in 2012, compared with the 1.3 million accounts that received rollovers for that year, the amount added to IRAs through rollovers was 10 times the amount from contributions.

However, an annual-snapshot percentage of IRA contributions doesn’t show whether the same individuals were contributing over time, or if different people contributed in different years. Taking advantage of the ability to look at multiple years across multiple accounts of individual owners across the EBRI IRA database, the report notes that while approximately 10 percent of traditional IRA owners contributed at some point during the three-year period, only 6 percent contributed to their IRA each year. On the other hand, while approximately 25 percent contributed to their Roth IRA in any one year, 35 percent did so at some point over the three-year period.

Looking at the pace of contribution activity, the EBRI analysis found that among those who contributed to their IRA in each of the three years, the pattern seemed pretty consistent: 12.1 percent did so in 2010, 13.2 percent in 2011, and 13.1 percent in 2012. Looking at the specific sources of those contributions, among traditional IRAs, we find that the percentage that contributed to them rose from 5.2 percent in 2010 to 6.6 percent in 2012 – but among Roth IRA owners, 24.0 percent contributed in 2010, 26.0 percent in 2011, and 25.1 percent in 2012.

Consider too that, among traditional IRA owners, only 3.0 percent contributed all three years, compared with 15.0 percent of Roth IRA owners who did so. Moreover, Roth IRA owners ages 25–29 were the most likely to contribute in any year and all three years (56.1 percent and 24.3 percent, respectively). Indeed, more than 4 in 10 (43 percent) Roth owners ages 25–29 contributed to their Roth in 2012.

The EBRI analysis found significant differences in the distribution patterns among older IRA owners, specifically those ages 70 or older, due to the required minimum distribution (RMD) rules. Those rules require individuals to begin making withdrawals from traditional IRAs starting April 1 of the year following the calendar year in which they reach age 70½. However, the RMD rules do not apply to Roth IRAs, a factor that likely explains the continued increases in account balances for Roth owners in that age group.

In sum, while the gross accumulations of retirement savings in IRAs provide value in terms of quantifying an increasingly significant component of the nation’s retirement security, a focus that takes into account only aggregate movements, or isolated account holdings, one that ignores the original source(s) of the money in the account, and/or the accompanying restrictions, runs the risk of overlooking significant undercurrents.

Undercurrents that may provide a better understanding of the growth trends in this important savings vehicle and ultimately, of course, explain how – and when – these retirement savings are withdrawn.

  • Notes

[i] Originally designed as a means to provide workers who did not have employment-based pensions an opportunity to save for retirement on a tax-deferred basis, IRAs have undergone a number of changes over time. The Economic Recovery Tax Act of 1981 (ERTA) extended the availability of IRAs to all workers with earned income (including those with pension coverage), while the Tax Reform Act of 1986 (TRA ’86) brought with it some restrictions on the tax deductibility (and, in some cases, availability) of IRA contributions. A decade later the Taxpayer Relief Act of 1997 (TRA ’97) created a new type of nondeductible IRA—the Roth IRA—and allowed nonworking spouses to contribute to an IRA, subject to certain income restrictions.

[ii] The May 2014 EBRI Issue Brief, “Individual Retirement Account Balances, Contributions, and Rollovers, 2012; With Longitudinal Results 2010–2012: The EBRI IRA Database,” is available online here.