Sunday, November 25, 2012


Retirement planning is a complex and highly individualized process, but many people find it easier to start by focusing on a single, specific target number.

For those interested in a single number for health care expenses in retirement, a recent EBRI report provides that. Among other things, the report noted that a 65-year-old man would need $70,000 in savings and a woman would need $93,000 in 2012 if each had a goal of having a 50 percent chance of having enough money saved to cover their projected health care expenses in retirement. A 65-year-old couple, both with median drug expenses, would need $163,000 in 2012 to have a 50 percent chance of having enough money to cover health care expenses.1

Determining how much money is needed to cover health care expenses in retirement is complicated. It depends on retirement age, the length of life after retirement, the availability and source of health insurance coverage after retirement to supplement Medicare, the rate at which health care costs increase, interest rates, market returns, and health status, among other things. That said, it is possible to project health care expenses with some accuracy, and EBRI’s recent analysis uses a Monte Carlo simulation model to estimate the amount of savings needed to cover health insurance premiums and out-of-pocket health care expenses in retirement.

However, those recent “single number” projections specifically excluded the financial impact of long-term care.

EBRI has long acknowledged the critical impact that health care expenses can have on retirement finances, and considering that EBRI has long incorporated both the costs of health care and long-term care in its Retirement Savings Projection Model® (RSPM), one might well wonder why this particular report specifically excluded those long-term care projections.

For all the complexity in those calculations, the reality is that everyone won’t have to deal with the expenses associated with long-term care. For those who will, the impact on retirement finances could be significant, even catastrophic.2 That’s why EBRI has modeled their impact in the RSPM since 2003.

As noted above, for those interested in a single number, the recent EBRI report provides that, along with variations that permit one to take into account different likelihoods of success and gender/marital combinations. We are able to do that because we treat longevity risk and investment risk stochastically,3 and the fact that those expenses (and the costs of insurance) are, at least relatively, predictable.

But while it is possible to come up with a single number that individuals can use to start setting retirement-savings goals, it is important to bear in mind that a single number based on averages will be wrong for the vast majority of the population—and that those who rely exclusively on that single number run the risk of running short.

Nevin E. Adams, JD

1 Unlike reports produced by a number of organizations, the EBRI report also provided estimates for those interested in a better-than-50-percent chance of success. See ”Savings Needed for Health Expenses for People Eligible for Medicare: Some Rare Good News,” online here.

2 The EBRI Notes article above illustrates the difference: If you ignore the impact of nursing home and home health care expenses, more than 90 percent of single male Gen Xers were projected to have no financial shortfall in retirement—but when that impact was included, just 68 percent of that group was projected to have no financial shortfall in retirement. The error of ignoring nursing home and home health care costs is even more profound if one focuses on the percentage of individuals with shortfalls in excess of $100,000.

3 For an expanded description of the difference stochastic modeling can make, see “Single Best Answer.”
See also: “Employment-Based Retiree Health Benefits: Trends in Access and Coverage, 1997-2010”, and “Effects of Nursing Home Stays on Household Portfolios.”

Sunday, November 18, 2012

“Grayed” Expectations?

Even the best retirement planning requires a fair number of assumptions: the age at which you hope to retire, for one thing; the amount of income that living in retirement will require; the length of time over which your retirement will last; the rate of return on your savings prior to, and following, retirement; the sources of retirement income that will be available to you, and in what amount(s).

Consider that in the 2012 Retirement Confidence Survey while worker confidence in having enough money to pay basic expenses in retirement wasn’t exactly high (only 26 percent were very confident), workers were noticeably less likely to feel very confident about their ability to pay for medical expenses after retirement (13 percent) and even less likely to feel very confident about paying for post-retirement long-term care expenses (9 percent) — levels that have remained statistically unchanged since 2010.

Indeed, the lack of employment-based retiree health insurance may result in unanticipated expenses in retirement. In the 2011 RCS, one-third of workers reported that they expected to receive this type of insurance from an employer (36 percent), though only 27 percent of retirees in that survey actually received it.

Earlier research found little impact of reductions in coverage on retirees, but the report notes that that may be because initial changes employers made to retiree health benefits affected future retirees, rather than those retired at the point of change. A recent EBRI Issue Brief highlights that, over time, more and more retirees have “aged into” those program changes, resulting in the greater impact found in more recent studies. The report also notes that most employers that continue to offer retiree health benefits have made changes in the benefit package they offer, changes that impact both the cost and availability of the benefit, including raising premiums that retirees are required to pay, eliminating employer subsidies, tightening eligibility, limiting or reducing benefits, or some combination of these.

However, as that Issue Brief also notes, very few private-sector employers currently offer retiree health benefits, and the number offering them has been declining, even in the public sector: Between 1997 and 2010, the percentage of non-working retirees over age 65 with retiree health benefits fell from 20 percent to 16 percent. Still, expectations seem to outpace reality; in 2010, 32 percent of workers expected retiree health benefits, while only 25 percent of early retirees and 16 percent of Medicare-eligible retirees actually had them.

Circumstances change, expectations matter, and retirement planning that relies on flawed or outdated expectations can, unfortunately, leave us short of where we need to be.

Nevin E. Adams, JD

See Employment-Based Retiree Health Benefits: Trends in Access and Coverage, 1997‒2010

Sunday, November 11, 2012


Last week the Center for Retirement Research at Boston College provided an update on its National Retirement Risk Index (NRRI).¹ The impetus for the update was the triennial release of the Federal Reserve’s Survey of Consumer Finance (SCF), published in June, reflecting information as of December 2010.

Now, many things have changed since 2007, and in the most recent iteration of the NRRI, the authors note five main changes: the replacement of households from the 2007 SCF with those from the 2010 SCF; the incorporation of 2010 data to predict financial and housing wealth at age 65; a change in the age groups (because a significant number of Baby Boomers have retired, according to the report authors); the impact of lower interest rates on the amounts provided by annuities; and changes in the Home Equity Conversion Mortgage (HECM) rules that lowered the percentage of house value that borrowers could receive in the form of a reverse mortgage at any given interest rate.

And, when all those changes are taken into account, the CRR analysis concludes that, as of December 2010, anyway, the percentage of households (albeit those from a partially different cohort) at risk of being unable to maintain their pre-retirement standard of living in retirement increased by 9 percentage points² between 2007 and 2010 (from 44 percent at risk to 53 percent).

When the baseline for your analysis is updated only every three years, it’s certainly challenging to provide a current assessment of retirement readiness. In previous posts, we’ve covered the limitations of relying solely on the SCF data³and, to some extent, the apparent shortcomings of the NRRI (see “’Last’ Chances”), and retirement projection models, generally (see “’Generation’ Gaps”).

On the other hand, the impact of the decline in housing prices and the stock market were modeled by EBRI in February 2011 (see “A Post-Crisis Assessment of Retirement Income Adequacy for Baby Boomers and Gen Xers”), while the impact of the rising age for full Social Security benefits has been incorporated in EBRI’s Retirement Savings Projection Model (RSPM) since 2003. Moreover, EBRI has also included the potential impact of reverse mortgages in our model for nearly a decade now.

Meanwhile, as a recent EBRI report noted (see “Is Working to Age 70 Really the Answer for Retirement Income Adequacy?”), the NRRI not only assumes that everyone annuitizes at retirement, and continues to ignore the impact of long-term care and nursing home costs (or assumes that they are insured against by everyone), but it also seems to rely on an outdated perspective of 401(k)-plan designs and savings trends, essentially ignoring the impact of automatic enrollment, auto-escalation of contributions, and the diversification impact of qualified default investment alternatives.

It’s one thing to draw conclusions based on an extrapolation of information that, while dated, may be the most reliable available. It’s another altogether to rely on that result in one’s retirement planning, or the formulation of policies designed to facilitate good planning.

Nevin E. Adams, JD

¹ The report, “The National Retirement Risk Index: An Update” is available online here.

² The report notes that, between 2007 and 2010, the NRRI jumped by 9 percentage points due to: the bursting of the housing bubble (4.5 percentage points); falling interest rates (2.2 percentage points); the ongoing rise in Social Security’s Full Retirement Age (1.6 percentage points); and continued low stock prices (0.8 percentage points).

³ As valuable as the SCF information is, it’s important to remember that it contains self-reported information from approximately 6,500 households in 2010, which is to say the results are what individuals told the surveying organizations on a range of household finance issues (typically over a 90-minute interviewing period); of those households, only about 2,100 had defined contribution (401(k)-type) retirement accounts. Also, the SCF does not necessarily include the same households from one survey period to the next. See “Facts and ‘Figures.’”

Sunday, November 04, 2012

Out of Cite?

Our industry pays a lot of attention to the investment choices that retirement plan participants make; we fret about the type and number of choices on their investment menu, the efficacy of target-date funds, the utilization of active versus passive investment strategies, and the prudence of the asset allocation choices that individuals make—with or without the benefit of tools and/or professional guidance. Unfortunately, once they leave that part of our private retirement system, not so much.

A significant percentage of that retirement plan money winds up in individual retirement accounts, or IRAs. In fact, today IRAs represent more than a quarter of all retirement assets in the U.S., according to a recent EBRI Issue Brief. But there remains a limited amount of knowledge about the investment behavior of individuals who own IRAs, alone or in combination with employment-based retirement plans.

In order to fill this gap, EBRI has undertaken an initiative to study in depth this connection between DC plans and IRAs, and created the EBRI IRA Database,(1) which links individuals (both within and across data providers) in this IRA database and with participants in DC plans. The asset allocation across ages within each IRA type(2) had some minor differences, but, in general, the percentages allocated to equities and balanced funds declined as the owner became older, and the percentage allocated to bonds and other assets increased. Additionally, as the account balances increased, the percentages of assets in equities and balanced funds combined decreased, while bond and “other” assets’ shares increased.

While gender differences abound in many walks of life, male and female IRA owners had virtually identical allocations in bonds, equities, and money in the EBRI IRA database, though males were slightly more likely to have assets in the “other”(3) category, and females had a higher percentage of assets in balanced funds.

Among IRA categories, Roth IRAs had the highest share of assets in equities (59.1 percent) and balanced funds (15.5 percent). Of course, Roth owners are younger, on average, than rollover owners, and Roth IRAs tend to be supplemental savings funded by individual contributions only, whereas rollovers tend to be the main or primary retirement savings for workers nearing retirement or retirees.

Indeed, the most significant difference among IRA types is that Roth owners were much more likely to have 90 percent or more of their accounts invested in equities than in other asset types, and were correspondingly likely to have less than 10 percent of their assets in bonds and money combined. Once again, the likelihood of these “extreme” allocations was very similar across genders.

When comparing the overall percentage of 401(k) assets held in equities (equities, equity share in balanced funds, and company stock) from the EBRI/ICI 401(k) database, the number is relatively close to that found in the IRA accounts (60.0 percent in 401(k) plans and 52.1 percent in IRAs), although the bond and money percentages are higher for IRAs than 401(k) plans (19.9 percent and 11.6 percent, respectively, for bonds and 8.9 percent and 4.4 percent, respectively, for money), while balanced funds constitute more of the assets in 401(k) plans than in IRAs. In sum, while it wasn’t the focus on the study, the average asset allocation found for IRAs was similar to that in 401(k) plans.

Of course, those IRA balances likely aren’t the totality of the retirement savings for these individuals, and thus, whether that particular allocation—looked at in isolation—is “good” news or not, remains to be seen.(4)

- Nevin E. Adams, JD

(1) The Employee Benefit Research Institute’s retirement databases (the EBRI/ICI Participant-Directed Retirement Plan Database, the EBRI IRA Database, and the EBRI Integrated Defined Contribution/IRA Database) have been the subject of multiple independent security audits and have been certified to be fully compliant with the ISO-27002 Information Security Audit standard. Moreover, EBRI has obtained a legal opinion that the methodology used meets the privacy standards of the Gramm-Leach-Bliley Act. At no time has any nonpublic, personal information that is personally identifiable, such as Social Security number, been transferred to or shared with EBRI. None of the three databases allows identification of any individuals or plan sponsors.

(2) The report considered four types of IRAs: traditional-contributions (traditional IRAs originating from contributions, in which distributions are taxable); Roth (in which contributions are nondeductible and distributions are tax free); SEP (Simplified Employee Pension)/SIMPLE (Savings Incentive Match Plan for Employees); and traditional-rollovers (traditional IRAs originating from assets rolled over from other tax-qualified plans, such as an employment-based pension or DC plans, in which distributions are taxable)

(3) “Other” assets includes assets that do not fit into the categories of equities, bonds, money/cash equivalents, or balanced funds. This could include stable-value funds, real estate (both from investment trusts and directly purchased), fixed and variable annuities, etc.

(4) As the EBRI IRA Database expands, more elaborate studies are being conducted. Linked with defined contribution account data, the tracking of movements of money between multiple retirement saving accounts (DC plans and IRAs) is being studied to see what, if any, asset allocation changes are being made when assets are moved between accounts. Furthermore, once individuals have reached retirement, the withdrawal or “spend-down” of those assets over time can be studied based on the longitudinal data that will be available, offering the potential of a far greater understanding of the retirement preparation and post-retirement behavior of Americans as these databases mature.