Sunday, July 28, 2013

Foregone “Conclusions”

Behavioral finance drives much of the discussion around retirement plan design innovations these days, for the very simple reason that it seems to help explain what might otherwise be viewed as irrational behaviors. For example, human beings are prone to something that behaviorists call “confirmation bias,” a tendency to favor information that confirms what we already believe. While doing so generally contributes to quicker assessments of information, there are some obvious shortcomings to that approach in terms of critically evaluating new information, particularly information that contradicts what we have already chosen (rightly or wrongly) to accept as reality.

Last month EBRI published an analysis of a direct comparison of the likely benefits under specific types of 401(k) plans and defined benefit (DB) pension plans.(1) As anyone who has worked with employment-based retirement plans knows, individual participant outcomes can vary widely based on a complex combination of decisions by both those that sponsor the plans and the individual workers who are eligible to participate in them, as well as a host of external factors (notably the investment markets) that lie outside their control.

The EBRI report took pains to not only select but to explain the key assumptions in our analysis. As it turns out, the analysis highlighted a number of circumstances in which traditional pensions (where offered) could produce a higher benefit at retirement—and some in which 401(k)s were superior in that regard.

However, some of the reporting and commentary that followed its publication suggest that some either did not read with care the entire analysis, or chose to ignore the totality of the report. Here, in a Q&A format, we deal with most of the mischaracterizations(2) we’ve seen thus far:
  • Did the report conclude that 401(k) benefits were better for almost every age and income cohort?
♦ No. While an analysis of just the median results for just the baseline assumptions might suggest this conclusion, the EBRI Issue Brief followed with seven different sensitivity analyses with different assumptions that produce different results. Moreover, the entire analysis was restricted to employees currently ages 25–29, because those individuals would have the opportunity for a full working career with those 401(k)s, as well as the modeled defined benefit results.
  • Did the study make assumptions that are biased toward 401(k)s?
♦ Quite the contrary. The baseline used historical averages and ran many sensitivity analyses that were biased AWAY from 401(k)s to test how robust the results were.
  • Why did you exclude automatic enrollment designs in 401(k) plans from the analysis?
♦ While a great many employers have adopted automatic enrollment with automatic escalation provisions in recent years, sufficient time has not yet passed since the establishment of most of these escalation features to know how long, and to what levels, participants will allow the escalation provisions to continue before they opt out.
  • How does excluding those automatic enrollment plan designs from the analysis impact the results?
♦ Some have argued that automatic enrollment is actually bad for retirement savings because, in some cases, the AVERAGE rate of deferral—when computed only for those who make a contribution—initially decreases.(3) However—and as previous EBRI research has documented—the decline in average deferrals masks the reality that while some individuals initially save at lower rates (certainly in the absence of provisions to increase those initial rates automatically), many lower-income workers become savers under an automatic plan design who would not contribute anything had they been eligible for a voluntary-enrollment 401(k) plan instead.
 
The June 2013 Issue Brief(4) specifically treats these individuals as making zero contributions and does NOT simply exclude them. This is the major reason why low-income employees do not do as well as high-income employees under THIS TYPE of 401(k) plan. EBRI has done previous analysis(5) showing the difference between automatic-enrollment and voluntary-enrollment types of 401(k) plans and specifically documents how much better automatic enrollment was for the retirement savings of lower-income cohorts.
  • Aren’t workers today changing jobs more frequently than they did during the heyday of defined benefit pensions?
♦ Actually, no. While it was not an explicit part of this report, a recent EBRI Notes article points out that the data on employee tenure (the amount of time an individual has been with his or her current employer) show that so-called “career jobs” NEVER existed for most workers. Indeed, over the past nearly 30 years, the median tenure of all wage and salary workers age 20 or older has held steady, at approximately five years. While the new analysis focused on the outcomes for younger workers, including the implications of their tenure trends on DB accumulations, the historical turnover trends suggest that this would have been problematic for full DB accumulations among previous generations of workers as well.
  •  Were the rates-of-return scenarios “realistic” for 401(k) participants?
♦ The EBRI analysis presented baseline results under historical return assumptions that were adjusted for expenses by reducing the gross return by 78 basis points. Additionally, sensitivity analysis was also conducted in this study by reducing the returns by 200 basis points to determine the robustness of the findings.
  • What about the fact that private-sector pensions are largely funded exclusively by employers, while much of that burden falls on individual workers in 401(k) plans?
♦ The report acknowledges this difference, but as the title of the Issue Brief specifically states, this is a comparative analysis of future benefits from private-sector voluntary enrollment plans vs. two stylized types of defined benefit plans—not the financial impact on the individual participants during the accumulation period, or how he/she might deploy assets not committed to retirement savings. However, as both the academic papers cited in the Issue Brief state, a comparison of ALL aspects of this difference would likely need to incorporate the assumption that more costly employer contributions to a plan (whether defined benefit or 401(k)) will be at least partially offset by lower wages over the long run, everything else equal.
  • Is it reasonable to compare defined benefit and 401(k) plan benefits?
♦ Retirement income adequacy studies have been undertaken in several reports by other organizations with the implicit assumption that 401(k) plans are unable to generate the same amount of retirement income (regardless of the source of financing). One of the primary objectives of the June EBRI Issue Brief was to actually test whether these implicit assumptions were correct.

So, which is “better”—a defined benefit plan or a 401(k)?

Well, as was noted in the Issue Brief, there is no single answer because a multitude of factors affect the ultimate outcome: interest rates and investment returns; the level and length of time a worker participates in a retirement plan; an individual’s age, job tenure, and remaining length of time in the work force; and the purchase price of an annuity, among other things.

The best answer depends on an incorporation of all the relevant factors, the tools to provide a thorough analysis, and an open mind with which to consider the results.

Nevin E. Adams, JD

(1) Jack VanDerhei, “Reality Checks: A Comparative Analysis of Future Benefits from Private-Sector, Voluntary-Enrollment 401(k) Plans vs. Stylized, Final-Average-Pay Defined Benefit and Cash Balance Plans,” online here.

(2) For example, see “Claim that 401(k)s Beat Defined Benefit Plans Stirs Controversy,” online here.

(3) This is the same type of mistake made in a 2011 Wall Street Journal article. See the EBRI blog “What Do You Call a Glass That is 60−85% Full?” as well as  “More or Less?”

(4) “Reality Checks: A Comparative Analysis of Future Benefits from Private-Sector, Voluntary-Enrollment 401(k) Plans vs. Stylized, Final-Average-Pay Defined Benefit and Cash Balance Plans,” online here.

(5) See Jack VanDerhei and Craig Copeland, “The Impact of PPA on Retirement Savings for 401(k) Participants.” Washington, DC: EBRI Issue Brief, no. 318, June 2008, online here.

No comments: