Review of Russell Investments target date fund methodology

Transcription

1 Review of Russell Investments target date fund methodology Yuan-An Fan, Ph.D., Senior Research Analyst Dan Gardner, CFA, Defined Contribution Analyst John Greves, CFA, Portfolio Manager Steve Murray, Ph.D., CFA, Director of Asset Allocation Strategies Russell s original research on glide path construction established our retirement income-based approach to designing target date funds. This paper restates Russell s convictions regarding a sound target date investment process and updates our research on the major assumptions that drive our glide path modeling: 1. Russell s foundational approach to glide path construction remains intact today 2. New insights into participant and retiree behavior and a changing capital market environment drive an updated glide path 3. Our research and refinement of the target date funds is designed to put plan participants on a better path to retirement. Introduction The importance of a sound target date investment strategy has never been greater. Reductions in corporate benefits, the freezing of defined benefit retirement plans and uncertainty about future Social Security benefits all contribute to increased individual responsibility for retirement planning. In recent years, defined contribution (DC) plan participants use of target date funds has grown, and this trend is expected to continue. 1 In light of these converging developments, DC plan fiduciaries should want their target date funds to be explicitly designed to support participants progress toward meeting their retirement savings goals. Russell s original research on glide path construction (Fan and Gardner 2008 revision) established our retirement income based approach to designing target date funds. This continues to remain our foundation. Since then, we have gained additional insight into retiree Russell Investments // Review of Russell Investments target date fund methodology SEPTEMBER 2014 Frank Russell Company owns the Russell trademarks used in this material. See Important information for details.

2 and participant behavior due to our work with custom target date funds and ongoing revisiting of our target date assumptions. Further, the capital markets outlook has changed in today s low interest rate environment. With our advances in knowledge and understanding of a changing investment landscape, we believe it is time to restate Russell s convictions regarding a sound target date investment process and also to update our research on the major assumptions that drive glide path modeling. This paper includes a brief review of our foundational target date research, a refresh of key assumptions for a typical DC plan participant, a discussion of regulatory constraints and a performance comparison of projected outcomes from our original glide path with the updated version. Quick review of original research Russell s original research on glide path construction (Fan and Gardner 2008 revision) highlighted three important characteristics that influenced the design of our target date funds. 1. The relationship between human capital (the ability to work, and thus save for the future, based on wages) and financial capital. In essence, early in a participant s career, total wealth is dominated by the bond-like behavior of human capital. Over the course of a career, the value of human capital as a proportion of total wealth decreases as monetary contributions accrue and financial assets typically grow from ongoing savings and investment returns. 2. The desire to balance total risk throughout a participant s savings horizon. Over time, the gradually changing proportions of these two sources of capital necessitate a shift in portfolio allocation from more volatile, growth-oriented assets to capitalpreservation assets a shift that defines the glide path and helps assure that total risk is optimally balanced throughout a career. For the sake of thoroughness, we note that there may be other valid explanations for a glide path strategy, although time diversification the notion that stocks are less risky over a long time horizon is not one of them The need for an income replacement target to measure the potential for success or shortfall in a target date investment strategy. Identifying an accumulation target as defined by typical expense levels of retirees facilitates analysis of participants risk of falling significantly short of sufficient income to cover retirement expenses. To incorporate this risk in the assessment of possible investment solutions, Fan and Gardner employed a utility function that prefers glide paths designed to maximize the level of expected wealth at retirement while mitigating the risk of falling short of the targeted replacement income. These characteristics were highly influential in the design of Russell s original glide path and remain fundamental to Russell s target date philosophy and construction process today. The need for a typical participant in designing target date funds As a vehicle for managing retirement savings, target date funds offer strategy differentiation by age cohort but limited precision for individual circumstances. They serve as one-size-fitsall retirement savings vehicles for large and diverse participant groups within given age ranges. Yet the optimal investment solution for any particular person depends on the specific circumstances of that individual. For instance, research indicates that differences in individual circumstances can lead to a wide range of total income replacement rates appropriate for retirees, depending on factors such as marital status, number of children, general state of health, home ownership, for example. 3 A single asset allocation strategy, such as a target date fund, cannot be best at addressing every individual situation. Despite the imprecise nature of target date funds, target date fund managers need some basis for designing the funds, and plan fiduciaries need some basis for determining whether a set of target date funds is appropriate for their plan. As explained in Fan and Gardner (2008 revision), this gives rise to the concept of the typical participant. The characteristics of the typical participant provide a reasonable representation of a broad population of participants in a target date fund and help anchor assumed contribution levels to potential future expenses. 4 Appropriately defining the typical participant enables the target date fund manager to better understand the potential risks and rewards associated with differing investment strategies Russell Investments // Review of Russell Investments target date fund methodology 2

3 from the participant s perspective. For plan fiduciaries, assumptions for the typical participant provide context for evaluating how their plan might align with an overall target date strategy. Review of typical participant assumptions Building an investment solution for a typical participant requires a thorough understanding of the key variables that define the investment problem. For participants saving for retirement, those key variables include the target retirement date, pattern of intended contributions, projected ending salary, income needed for a comfortable retirement as a percentage of ending salary, flexibility associated with the retirement spending goal, and attitude toward risk and return. The following sections describe our latest research on the key input assumptions for a typical participant and how they are used to model a sound target date investment strategy. Income needed to support a comfortable retirement A fundamental consideration in determining an investment strategy for retirement savings is the level of income a typical participant will need to support a comfortable retirement. Since Russell s original paper was published, research on reasonable baseline income replacement rates has been updated. AON Hewitt published the results of its Real Deal study in 2012, which included updated statistics suggesting a baseline rate of 78% of final earnings for a typical defined contribution plan participant. 5 The study s calculation of income replacement rates relies on a well-established methodology, described in Aon (2008), which accounts for typical expenses before and after retirement, taxes and retirement savings contributions. 6 While we believe this represents an appropriate starting point, we also believe deeper analysis is warranted. The 78% starting point does not take into account our detailed understanding of contribution rates, retirees spending patterns or the possibility of health care expense shocks during retirement. We have evaluated research on each. Savings rate adjustment to baseline The 78% baseline income replacement rate reflects a specific assumption for contribution rates in the Real Deal study. We are more confident in our estimate of contribution rates, which is based on several sources, than in the contribution rate baked into the Real Deal study. Our estimated contribution rates are slightly lower, which increases the 78% baseline target to an adjusted 79% target. The intuition supporting this adjustment is that participants who save less than the percentage of income assumed by the study would be accustomed to spending more than assumed by the study, and thus would need to target a higher income replacement rate to maintain their pre-retirement levels of spending. Unexpected health care expenses Even individuals with Medicare and Medigap coverage may find that some health care needs (notably, for prescription drugs and long-term care) can involve painful out-of-pocket costs. For instance, the median cost of a private nursing home room in the United States is about $80,000 per year. 7 While most people who require long-term care will receive it from unpaid family members or friends, 8 this solution merely shifts costs from the individuals to their loved ones. Research from the Employee Benefit Research Institute (EBRI) sheds light on the potential impact of unexpected health care expenses on retirement savings goals. 9 Allowing for 50thpercentile unexpected health care expenses increases the amount of savings necessary to fund a retirement starting at age 65 by a fairly modest $35, If a plan participant desires more certainty i.e., savings sufficient to avert running out of money in 90 percent of retirement scenarios allowing for unexpected health care expenses increases the required savings amount by about $260,000. Russell believes that prudent retirement planning should acknowledge, in addition to the average expenses included in the Real Deal study, the possibility of unexpected, not-fullyinsured health care expenses. We believe setting aside a provision that lies between the two aforementioned estimates 11 takes into account a rainy-day scenario, without devoting an extreme amount of resources to protecting against it. This provision amounts to about $75,000 in present-value terms at age 65 and is depicted as the middle column of Figure 1. It increases our adjusted 79% targeted spending to 85%. Russell Investments // Review of Russell Investments target date fund methodology 3

4 Figure 1: Additional savings needed by 65-year-old to meet common unexpected health care expenses in retirement, at varying confidence levels $300,000 $250,000 $200,000 $150,000 $100,000 $50,000 $0 50% 75% 90% Confidence level Source: Park 2011, Russell analysis. Retiree spending patterns We found that retiree expenses tend to decrease in real terms until around age 80. Intuitively, the aging process typically slows activity, and retirees may downsize as they get older. Referencing a variety of sources, we found that inflation-adjusted expenses tend to decline by 1% to 2% per annum in the first 15 years of retirement. 12 Nonetheless, absolute expenses continue to grow in nominal terms, simply at a slower pace than inflation. Our 85% target with below-inflation spending increases is equivalent (in terms of present value) to a 79% target with spending increases in line with inflation, based on our assumptions for interest rates and inflation. In addition to these adjustments, we revisited our Social Security assumptions. The Congressional Budget Office currently anticipates a deficit for the Social Security program in the future. 13 Yet, how Congress eventually addresses the issue remains unclear. 14 As a result, we continue to model current benefits and taxes associated with the program. The size of current Social Security benefits depends on certain assumptions. We obtained a copy of the Social Security Detailed Calculator 15 so as to incorporate our new data on participants directly into the benefits estimate. Finally, we had originally assumed that Social Security income estimates should be based on a couple, including a spousal benefit. Recent research from the Government Accountability Office indicates that the use of the spousal benefit is becoming the exception, rather than the norm. The number of women age 62 and older receiving Social Security checks based on the spousal benefit has decreased from 56%, in 1960, to 25%, in In light of this evidence, we have eliminated the spousal benefit from our model. TRI = Target Replacement Income. It is a specific percentage of one s final, pre-retirement salary. Income replacement rate conclusion The cumulative impact of this research is an increase in the target replacement income (TRI) from our original white paper (Fan and Gardner revised 2008). Using the updated baseline rate from the AON/Hewitt study along with the adjustments mentioned above, we have modified the TRI from its original value of 42% to a new value of 49%, as shown in Figure 2. Further, since total expenses are slowly rising over time, we believe it is prudent to incorporate a cost-of-living adjustment of about 2.2% per annum, in line with anticipated inflation. Russell Investments // Review of Russell Investments target date fund methodology 4

5 Figure 2: Components determining target replacement income (TRI) as percentage of typical participant s final earnings 100% Initial spending need from DC plan + Adjustments increasing spending need from DC plan + Adjustments decreasing spending need from DC plan = Final spending need from DC plan 75% 1% 6% 6% 30% 50% 78% 25% 49% 0% Baseline Savings rate adjustment to baseline Unexpected health care Retiree spending patterns Social Security TRI from DC plan Inputs: Aon (2012); Banerjee (2012); Blanchett (2013); Fisher et al (2005); Hurd and Rohwedder (2011); MacDonald and Moore (2011); Park (2011); Pfau (2012); SSA (2014); Tacchino and Saltzman (1999). Patterns and sizes of defined contribution plan participants contribution amounts Understanding contribution behavior is also critically important for participants saving for retirement. As Fan and Gardner (revised 2008) noted, in a hypothetical scenario where participants could make all of their contributions in a lump sum at the start of their career (and where market forecasts did not change through time), it would not be optimal to gradually decrease the weight of growth-oriented assets over time: the glide path would be flat. However, participants almost always 17 need to make contributions over time. This makes it important to understand the pattern and size of the contributions. We utilized our clients participant data, external research and information from the Employee Benefit Research Institute 18 to clearly understand total contributions today. Since total contributions are calculated using savings rates, matching and non-matching employer contribution amounts, salary, salary growth rate and retirement age, we investigated each of these components. Savings rates We found data on individual savings rates to be relatively consistent across sources. For participants who are currently saving, we found that the starting rate averaged around 5% (as a percentage of earnings) at the start of the career and escalated to 10% near the end. We excluded 0% contribution levels from the analysis, since designing the funds to accommodate non-savers would penalize those who do save. Matching rates Plan sponsors typically offer matching contributions. The most common structure is for employers to match half of participant contributions on the first 6% of salary contributed. We applied the same contribution policy to the individual savings rates noted above. Non-matching contribution rates Roughly 40% of plan sponsors studied also offer a non-matching contribution in addition to the matching contribution, and the median discretionary contribution rate is 4%. 19 We believe a sensible assumption for the level of non-matching contributions is to multiply the percentage of plan sponsors offering a discretionary contribution by the median contribution amount. The result is a rounded non-matching contribution rate assumption of 1.5%. The combined impact of participant savings rates, matching policy and non-matching contribution rates is shown in Figure 3, alongside our prior total contribution estimate in Fan and Gardner (2008 revision), which increases from 6% total contributions up to about 11.7%. Russell Investments // Review of Russell Investments target date fund methodology 5

6 % of initial wage Contribution rate (% earnings) Figure 3: Total contribution assumptions and component amounts 15% 10% 5% Sources: Aon Hewitt (2012); EBRI (2013 queries); Russell (2013 research); Vanguard (2014). Salary 0% Based on our new review, we have concluded that our ending salary assumption should modestly increase from $80,000 to $90,000, a 13% increase. 20 Ending salary influences tax rates and the level of Social Security benefit. Salary growth rate Age Updated - participant Updated - employer non-matching Updated - employer matching Original As part of our update of this note, we researched typical wage growth patterns for individuals. While we prefer to see longitudinal data on the growth of salaries over time, such data for DC plan participants were difficult to find. As a result, we used cross-sectional salary data (e.g., current wages of individuals by age) to generate a wage curve. The data indicated that inflation-adjusted wages tend to increase at a faster rate early in a career, to plateau midcareer, and to decline slightly later in a career. The net impact over an entire career was average wage growth of roughly 1.3% per annum. 21 Previously, we assumed a constant rate of salary growth of 1.5%. As shown in Figure 4, we have updated our assumptions to reflect the pattern of wage growth and the modestly lower yearly average. Figure 4: Pattern of salary growth assumptions over participant career 200% 170% 140% 110% 80% Age Original Updated Source: Sources: Aon Hewitt (2012); EBRI (2002); EBRI (2013 queries); Russell (2013 research). Retirement age We believe maintaining our current retirement age assumption of 65 is still appropriate. Our research finds that current retirees have retired on average before age 65, and that nonretirees still plan to retire around age While the retirement age for full benefits from Social Security is later than age 65, eligibility for Medicare enrollment still starts at age 65. The net result of each of these assumptions leads us to the patterns and sizes of contributions shown in Figure 5. Russell Investments // Review of Russell Investments target date fund methodology 6

7 Total annual contribution (current $) Figure 5: Savings pattern assumptions throughout career $15,000 $12,000 $9,000 $6,000 $3,000 $ Age Original Updated Inputs: Sources: Aon Hewitt (2012); EBRI (2013 queries); Holden and VanDerhei (2002); Russell (2013 research); Vanguard (2014). Regulatory constraints The Department of Labor s Qualified Default Investment Alternative regulations require target date funds to hold an unspecified level of fixed income securities. When the regulations were finalized in 2007, we updated our glide path allocation to include a minimum 10% allocation to fixed income securities in each target date fund. 23 With the passage of time and further interpretation of the regulation across the industry, we decided to modestly relax our modeling constraint to include a minimum of 7% fixed income securities or maximum of 93% growth assets. From the perspective of our modeled typical participant, the highest possible exposure to growth assets early in the career provides for a better chance of achieving retirement objectives than does including a minimum allocation to fixed income securities. Final framework for the investment problem We incorporate each item discussed earlier into a framework for modeling optimal asset allocation decisions for retirement savings. Plan participants need to accumulate enough wealth, based on 40 years of contributions, to replace 49% of their final year s salary with a cost of living adjustment. To calculate the wealth needed to meet the 49% retirement income goal for the defined contribution assets, we price a hypothetical immediate, single life annuity for a 65 year-old male 24 that delivers the desired cash flows with the 2.2% average cost of living adjustment described earlier. While we acknowledge that annuity uptake among retirees remains low, an annuity represents the market price for retirement income and requires us to make fewer assumptions than those needed in alternative approaches to determining a wealth accumulation target. Russell Investments // Review of Russell Investments target date fund methodology 7

8 Table 1: Summary of assumption changes FACTOR PRIOR ASSUMPTION UPDATED ASSUMPTION OBSERVATIONS Contribution rate 6% increasing to 12% Retirement age Career length Final salary (impacts Social Security and tax rates) 9% increasing to 14% Makes success easier: All else equal, increases lifetime savings by 26%. $80,000 $90,000 Complex and offsetting effects: Annuity target and portion of savings changes caused by increased final salary offset both increase by 13%; Social Security becomes smaller portion of retirement income target; lower real wage growth nearly offsets higher final salary level by shifting wages (and savings) toward younger age. Real wage growth 1.5% and constant 1.3% and decreasing with age Other retirement income sources Social Security only, 36% (for couple) Initial wealth $0 $0 Target replacement income (TRI) Maximum amount of growth assets Source: Russell analysis Social Security only, 30% (for individual) Makes success harder: Social Security change mostly due to final salary assumption. 42% nominal 49% real Makes success harder: Most of the change can be attributed to the incorporation of a COLA. The other largest contributors to change are the Social Security assumption and the set-aside for unexpected health care expenses. 90% 93% Modestly improves success rate. Impact of new participant and capital market assumptions on glide path Overall, the assumption changes shown in Table 1, above, provide for a higher chance of success. A significant driver of the improvement is the combination of timing and amount in the pattern of contributions. Total dollars saved increase by 47%, and this occurs slightly sooner, on balance, under the new assumptions. In addition to the assumptions shown in Table 1, capital market assumptions have changed. Our prior strategic planning for the glide path assumed notably higher return levels than are expected today, and interest rates (which influence the discount rate used to value retirement income) were higher as well. Under the updated assumptions, expected portfolio growth is lower, and the wealth target corresponding to 49% TRI is relatively larger. Table 2: Comparison of original with updated capital market return, inflation and interest rate (Treasury yield) assumptions ORIGINAL UPDATED U.S. Equity 8.4% 8.0% Non-U.S. Equity 8.4% 8.2% Fixed Income 6.0% 4.6% Inflation 2.5% 2.4% 10-Year Treasury yield at time of analysis 5.03% 1.85% Source: Russell. All are expected values except for the Treasury yield, which is a known value. Under the updated assumptions, a 60/40 portfolio 25 would grow to only about 75% of the amount it would have under the original assumptions. At the same time, the decline in interest rates, the introduction of the COLA and the change in the TRI from 42% to 49% significantly increase the hypothetical cost of achieving the retirement income goal. This increase in the cost of retirement, combined with lower market growth, more than offsets the 47% increase in savings. Consequently, the glide path necessary to support a successful retirement is more aggressive at most ages and ends up with a modestly higher exposure to growth assets at retirement. Russell Investments // Review of Russell Investments target date fund methodology 8

9 Percent allocation to growth assets The updated assumptions have led us to an updated glide path, shown in Figure 6. The allocation to growth assets is somewhat more aggressive than before at most ages along the glide path. For a few years in the middle, it becomes somewhat less aggressive, since the contributions are more front-loaded than before (recall that if they had been entirely frontloaded, and if market forecasts did not change through time, the glide path would be flat). While the standard deviation of returns would increase somewhat, commensurate with the modest increase in the growth asset allocation, we do not necessarily see the updated glide path as riskier from the perspective of our typical participant. In the next section, we discuss how, by our measure of risk, the updated glide path is actually somewhat less risky than before. Figure 6: Growth allocation of updated (2014) and original (2007) glide paths 100% 80% 60% 40% 20% 0% Years to retirement Original glide path Updated glide path Source: Russell Before and after glide path change: Projected outcomes under new assumptions The new glide path allocations represent an improvement over the previous glide path in terms of anticipated increased income replacement rates, account balances at retirement and decreased shortfall risk. As discussed in the original Fan and Gardner analysis, the target date fund is designed with reference to a hurdle level of wealth consistent with the cost of a hypothetical, immediate single life annuity 26 that is designed to provide the targeted retirement income. The risk associated with a candidate glide path allocation strategy is the possibility that the final account balance will fall below this hurdle level, with shortfall measured as the amount of any deficit relative to that hurdle. The allocation model s utility function finds candidate glide paths attractive if they support high replacement rates with limited risk of underperforming the hurdle, and unattractive if they generate low replacement rates and leave open a significant possibility for large shortfalls. Mathematical details of the utility function are documented in the Appendix. Table 3: Projected outcome comparison of original glide path with updated glide path under updated assumptions VALUES AT RETIREMENT 2014 GLIDE PATH 2007 GLIDE PATH Expected account balance ($1000s inflation-adjusted) $927 $904 Median account balance ($1000s inflation-adjusted) $820 $805 10th percentile account balance ($1000s inflation-adjusted) $492 $494 Expected replacement rate 88% 86% Median replacement rate 78% 76% 10th percentile replacement rate 46% 46% Probability replacement rate > 49% target 72% 71% Expected shortfall penalty ($1000s) Russell Investments // Review of Russell Investments target date fund methodology 9

10 Source: Russell analysis. IMPORTANT: The projections or other information generated by this analysis regarding the likelihood of various investment outcomes are hypothetical in nature, do not reflect actual investment results and are not guarantees of future results. Results will vary with each use and over time. Table 3 displays projected outcomes for the original and updated glide paths under our updated assumptions. The updated glide path provides comparable downside protection (as indicated by the nearly equal 10th percentile replacement rates and slightly lower expected shortfall penalty) while improving the typical outcome, including increased expected account balance and replacement rate. Discussion Research findings suggest lower retirement readiness for typical participant The research findings we have documented above have a net effect of reducing projected retirement readiness of the typical participant, relative to our prior glide path analysis and participant assumptions in Fan and Gardner (revised 2008). That may seem to be a curious finding given we are projecting our glide path to now deliver higher income replacement than originally. Yet while contributions are now larger, the goal to achieve is also higher. While greater assumed contributions and the updated asset allocation should improve retirement readiness, these changes are more than offset by the increase in our TRI assumption and, especially, reduced expectations for capital markets due to low interest rates. While projected probability of success is an imperfect measure of the appropriateness of a retirement plan, we will use it here to convey a sense of the magnitude of the combined changes. Based on the original target date glide path, the projected probability of achieving the TRI objective is About 80% under our original assumptions for participants and capital markets (including the original 42% TRI objective) 28 About 90% under new assumptions for participants and original assumptions for capital markets. The increase in estimated contributions more than offsets the increase in the savings goal. About 71% under new assumptions for both participants and capital markets, as seen in Table 3. This is a somewhat sobering projection, but not surprising given the low interest rate environment. The updated glide path modestly increases the projected chance of a successful outcome to 72%. Next steps in retirement solutions While target date fund solutions have helped support the shift to DC plans as the primary source of retirement income for U.S. workers, there are limitations to the funds ability to be tailored to individual circumstances. In these paragraphs, we discuss some of these limitations and suggest modifications or new products that could enhance the ability of workers to plan for retirement. Key characteristics of many participants may be dissimilar to the profile of the typical participant for whom we developed the glide path. Examples include mid- and late-career workers who may not have participated in retirement plans over their entire careers and may not have accumulated as many financial assets as their cohort peers, who have spent their entire careers within a retirement plan, and workers who may have failed to save at a rate comparable to that of the typical participant. Participants who have incurred large medical bills or experienced divorce or bankruptcy may fit this profile. Furthermore, the retirement account balances of all workers could suffer or benefit from market returns that differ dramatically from anticipated levels. An intermediate step toward a more tailored solution would be to allow the glide path for each target date fund to differ, based on the circumstances typical of that particular age cohort. This modest step has been evaluated by Russell for some plan sponsors. We believe a better solution would be more closely tailored to the characteristics and preferences of individual plan participants. Russell Adaptive Investing TM offers customized, adaptive glide paths for each participant that incorporate specific savings and retirement preferences in the determination of an effective investment plan. 29 As markets evolve and Russell Investments // Review of Russell Investments target date fund methodology 10

11 circumstances change, this solution provides for updated allocation advice based on the actual present situation of the participant, rather than the situation forecasted by past analysis. Aside from closer tailoring of the plan to the individual s circumstances, adaptive strategies benefits include the ability to respond more nimbly to plan and market changes on an ongoing basis. Conclusion This paper re-examines Russell s research on glide path modeling, which was originally published in 2006, 30 and provides fresh perspective to help participants improve the probability of successful retirement. We affirm our conviction in our approach to target date investing, document the refreshed inputs to the Russell target date funds and describe the impact of these refreshed inputs on asset allocation and projected outcomes. The research detailed in this paper drives the asset allocation of Russell s target date funds, enabling us to implement our latest thinking while still holding to our core commitment: managing the risk of retirement income shortfall. Although success cannot be guaranteed, we believe our research and refinement of the target date funds puts plan participants on a better path to retirement. REFERENCES Aon (2008) Aon/Georgia State Replacement Ratio study. Aon Hewitt (2012). The Real Deal: 2012 Retirement Income Adequacy at Large Companies. Banerjee, Sudipto (2012). Expenditure Patterns of Older Americans, EBRI Issue Brief No. 368, February. Blanchett, David (2013). Estimating the True Cost of Retirement. Working paper, Morningstar, November. Cohen, Josh, and Jeff Eng (2014 revision). Russell Investments ARA: Aiming for the bull s-eye. EBRI (2013 queries). From the EBRI/ICI 401(k) database. Fan, Yuan-An and Grant Gardner (2008 revision). Russell s approach to target date funds: building a simple and powerful solution to retirement saving. Fan, Yuan-An, Grant Gardner and Steve Murray (2007). Changes in the Russell Target Date Funds: The Effects of Department of Labor QDIA Regulations. Fisher, Jonathan, et al (2005). The Retirement Consumption Conundrum: Evidence From a Consumption Survey. Working paper. Center for Retirement Research at Boston College, December. GAO (2014). Retirement Security: Trends in Marriage and Work Patterns May Increase Economic Vulnerability for Some Retirees. Genworth (2013). Cost of Care Survey. Holden, Sarah, and Jack VanDerhei (2002). Appendix: EBRI/ICI 401(k) Accumulation Projection Model. Investment Company Institute Perspective, Vol. 8 / No. 3A, November. Hurd, Michael D., and Susann Rohwedder (2011). Economic Preparation for Retirement. Working paper, National Bureau of Economic Research, July. MacDonald, Bonnie-Jeanne, and Kevin D. Moore (2011). Moving Beyond the Limitations of Traditional Replacement Rates. Society of Actuaries, September. Park, Youngkyun (2011). Retirement Income Adequacy With Immediate and Longevity Annuities. EBRI Issue Brief No. 357, May. Pfau, Wade (2012). How Do Spending Needs Evolve During Retirement? Advisor Perspectives, March. Riffkin, Rebecca (2014). Gallup: Average U.S. Retirement Age Rises to 62. Referenced at April Russell (2013). Research: Survey of clients. Scholz, John Karl, and Ananth Seshadri (2009). What Replacement Rates Should Households Use? Working paper, Michigan Retirement Research Center, September. Skinner, Jonathan (2007). Are You Sure You re Saving Enough for Retirement? Journal of Economic Perspectives, Vol. 21, No. 3; pp SSA (2014). Social Security Detailed Calculator , released January 2014, as downloaded at Tacchino, Kenn, and Cynthia Saltzman (1999). Do Accumulation Models Overstate What s Needed to Retire? Journal of Financial Planning, February. VanDerhei, Jack, and Craig Copeland (2010), testimony submission to Senate Committee on Health, Education, Labor and Pensions Hearing: The Wobbly Stool: Retirement (In) security in America (October 7), pp Vanguard (2014). How America Saves 2014: A report on Vanguard 2013 defined contribution plan data. Russell Investments // Review of Russell Investments target date fund methodology 11

12 Appendix Explanation of unexpected health care expense calculations Overview of Park (2011) Model Park (2011) models retiree health care spending with a deterministic component and a stochastic component. The deterministic component includes Medicare premiums, private Medigap premiums and estimated out-of-pocket costs that are not fully reimbursed (or not covered) by Medicare and/or private Medigap. The stochastic component includes Simulated cost of prescription drugs, based on the Medical Expenditure Panel Survey. Assumed future inflation for this category is 7%. Simulated cost of nursing home stays and/or home care, based on National Nursing Home and National Home and Hospice Care Surveys. Assumed future inflation for this category is 5% for nursing home expenses and 3% for home health care expenses. While Park (2011) does not state how much each of the 2 stochastic components contributes, based on the October 2013 EBRI Notes Amount of Savings Needed for Health Expenses for People Eligible for Medicare: More Rare Good News it is clear both categories have a material impact. Incorporating Park (2011) simulation results into Russell s methodology Figure 1 is based on the output of simulations performed in Park (2011) and our understanding of how the stochastic component of the Park analysis compliments expected health care spending in Aon Hewitt (2012). Since Aon Hewitt (2012) does not reflect variation in health care spending, we use the stochastic component of the Park (2011) model to reflect that variation. The calculation steps can be described as follows: 1. Park (2011) presents the amount of wealth needed to cover retirement spending expenses, as a multiple of final earnings, for many sets of assumptions. The analysis we source from Park (2011) makes the following assumptions: a. The individual being modeled is a male in Income Category 4, which is most consistent with our own assumptions for gender and final earnings. b. The individual purchases an immediate single life annuity with 100% of his wealth upon retirement, consistent with how Russell models. Any annuity income that is not spent in a given year is placed in a portfolio of 50% stocks and 50% bonds. Details of the relevant assumptions can be found in Park (2011). 2. From there, we calculate the difference in the required wealth multiple with and without stochastic health care expenses. Park (2011) presents the required wealth multiple at three different confidence levels: 50%, 75% and 90%. Under the assumptions in 1., above, the difference in the required wealth multiple is 0.3, 0.7 and 2.5 for 50%, 75% and 90% confidence levels, respectively. 3. Multiplying each of those amounts by the Income Category 4 final earnings assumption of $103,584 yields the required wealth amounts shown in Figure 1: $31,000, $73,000 and $259,000 (rounded to the nearest $1,000). Russell Investments // Review of Russell Investments target date fund methodology 12

13 Considerations for this analysis 1. Our analytical approach, described above, would be problematic if there were redundancies or gaps in how we account for overall health care expenses. Russell believes the approach of adding Aon Hewitt (2012) s deterministic expected spending to Park (2011) s stochastic spending as a model for overall health care spending is reasonable given that the former is similar to Park (2011) s deterministic expected spending. Aon (2008), upon which the Aon Hewitt (2012) calculations are based, uses the Consumer Expenditure Survey for its health care expenditure assumption. Park (2011) utilizes a later version of the same survey in the out-of-pocket expense portion in its deterministic component (the Medicare premium assumption includes estimated Part B and Part D premiums at the time, while the private Medigap premium assumption was gathered from a website of AARP Medicare Supplement Insurance). 2. Since we assume the retiree purchases an immediate single life annuity upon retirement, the annuity price assumption would impact the amount of wealth required. Inconsistency in the annuity price assumption between this paper and Park (2011) could impact results. We do not believe any inconsistencies materially affect our analysis. Annuity prices in Park (2011) come from actual price quotes at the end of 2009, when interest rates were a bit higher than they are today. However, the analysis presented in this paper is based on our forecast of interest rates in the future, and we expect interest rates to rise in the coming years. 3. While Park (2011) cites national data for long-term care expenses, these vary by region, as seen in Genworth (2013). 4. Many recipients of paid long-term care rely upon Medicaid to cover expenses. Park (2011) s calculations exclude Medicaid. This modeling decision may lead to overestimation of the amount of assets required for covering retirement expenses, particularly for retirees with low income. We believe our decision to focus on retirees with higher earnings mitigates the impact of Medicaid exclusion, as these individuals are less likely to utilize Medicaid. Details of target date glide path utility model from Fan and Gardner (2008 revision) We derive our glide path by finding the mix of equity and bonds in each of the 40 years that maximizes: Expected Value of Wealth at Retirement (Risk Aversion Parameter x Expected Shortfall Penalty) The size of the risk aversion parameter reflects the investor s attitude about the trade-off between risk and reward. In principle, there is a different optimal glide path for each possible value of this parameter. Thus, in order to decide on a glide path of a typical investor, we must choose this parameter value for a typical investor. Unlike the savings rate and income replacement rate, this parameter cannot be directly observed or estimated from data. In the model we use to construct the glide path, risk is the danger of falling below hurdle wealth at retirement. Our measure of risk is the shortfall penalty function shown in Figure 7. The horizontal axis shows wealth at retirement. If wealth at retirement is greater than the hurdle, the penalty is zero. If the wealth falls below the hurdle, there is a shortfall. The greater the shortfall in wealth, the greater the penalty. Moreover, the curvature of the function means the penalty grows in increasing proportion to the shortfall. In other words, if the shortfall doubles, the shortfall penalty more than doubles. This aspect of the shortfall penalty function means that we heavily penalize large shortfalls. In using mathematical optimization to search for the best glide path, we reward candidate paths for producing a high level of wealth at retirement, and we penalize paths for falling below the wealth hurdle. Russell Investments // Review of Russell Investments target date fund methodology 13

14 Shortfall penalty Target wealth Figure 7: Shortfall Penalty Function Income well below target: big penalty Source: Russell Income near target: small penalty Wealth at retirement Income above target: no penalty Russell Investments // Review of Russell Investments target date fund methodology 14

15 1 According to Vanguard (2014), the proportion of participants in Vanguard-administered DC plans with a professionally managed asset allocation has increased from 7% in 2004 to 40% in 2013, and is expected to increase to 58% by Paul Samuelson made the classic case against time diversification in his paper Risk and uncertainty: A fallacy of large numbers. Lubos Pastor and Robert Stambaugh actually argue for the opposite of time diversification that stocks returns are more uncertain in the long run, reducing desirable stock allocation in target date funds in their 2011 paper Are stocks really less volatile in the long run? 3 For a good starting point, see Scholz and Seshadri (2009) and Skinner (2007). 4 With off-the-shelf target date funds, the typical participant is based on a national population of plan participants, reflecting the intended users of the funds. For plan sponsors with custom target date funds, the typical participant assumptions can be tailored to reflect the circumstances of plan participants at their organizations. 5 Aon Hewitt (2012), p. 24. Based on participants age 50 and up. 6 It accounts for specific major expenses, such as a mortgage or a child s education, only to the extent that these are reflected in typical expenses before and after retirement. A specific individual s changes in spending may be very different from what is typical. As discussed in the prior section, the full variety of individual circumstances cannot be adequately reflected in a strategy designed for a broad group of individuals. 7 Genworth (2013). 8 Source: 40 Must-Know Statistics About Long-Term Care, U.S. Dept. of Health and Human Services, as referenced at 9 Park (2011) and Russell analysis. 10 Ibid. Based on a 65 year-old retiree in the top quartile (in terms of income) who seeks to have enough savings to avoid running out of money in retirement, at a 50% confidence level. Please see the Appendix for a more thorough explanation. 11 I.e., that represents an allowance for unexpected health care expenses sufficient to avoid running out of money at a 75% confidence level. 12 Banerjee (2012); Blanchett (2013); Fisher et al (2005); Hurd and Rohwedder (2011); MacDonald and Moore (2011); Pfau (2012); Tacchino and Saltzman (1999). 13 SSA (2014). The 2014 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Federal Disability Insurance Trust Funds reports that reserves are projected to run dry in At the time of projected trust fund depletion, projected ongoing revenues would cover 77 percent of scheduled benefits. 14 While the type of adjustment is uncertain, the implications of certain types of changes are clear in the context of the Aon (2008) methodology. For instance, if Congress decides to reduce benefits to what could be supported by ongoing revenues, this would place greater demands on individual retirement saving (i.e., the Social Security bar in Figure 2 would become smaller, increasing the final spending need). 15 Social Security Detailed Calculator , released January 2014, as downloaded at 16 GAO (2014). 17 There is a nascent market developing around the idea of taking a stake in individuals future earnings, exemplified by the agreement reached between professional athlete Arian Foster and Fantex, Inc. 18 A special thank you to Jack VanDerhei and his colleagues at EBRI for assistance with database queries. 19 Vanguard (2014). 20 Consistent with realized average wage growth since the original design of the glide path. See, e.g., the National Average Wage Index page at 21 This assumption is higher than national average real wage growth over the past several decades as reflected in, the National Average Wage Index. Yet we believe our assumption to be appropriate, since we are developing a solution for the 401k marketplace, not the general population. Highly educated people tend to experience higher wage growth, per VanDerhei and Holden (2002), and are more likely to have access to a 401k plan, per VanDerhei (2010). 22 Riffkin (2014). 23 Fan, Gardner and Murray (2007). 24 The choice of a 65 year old male gives a hypothetical annuity price that is in the middle of the range of hypothetical annuity prices for individuals of both sexes between ages 60 and 70. In that sense it represents the typical retiree. Other assumptions include our mortality assumptions we use the Annuity 2000 Basic Table for Males and our interest rate assumption, which is based on our forecast for the 5-year Treasury rate at the time the participant would, hypothetically, purchase the annuity % U.S. equity/20% Non-U.S. equity/40% fixed income. 26 The same hypothetical annuity described earlier and in endnote The units of shortfall are not directly comparable to the wealth units of account balance. The value cited is the square root of expected penalized shortfall. The value is probably not informative, other than to note that the shortfall of the updated path is modestly smaller than for the previous glide path. 28 Fan and Gardner (2008 revision). 29 Cohen and Eng (2014 revision). 30 Fan and Gardner (2008 revision). Russell Investments // Review of Russell Investments target date fund methodology 15

16 ABOUT RUSSELL INVESTMENTS Russell Investments is a global asset manager and one of only a few firms that offers actively managed multi-asset portfolios and services, which include advice, investments and implementation. Russell Investments stands with institutional investors, financial advisors and individuals working with their advisors using our core capabilities that extend across capital market insights, manager research, asset allocation, portfolio implementation and factor exposures to help investors achieve their desired investment outcomes. FOR MORE INFORMATION: Call Russell Investments at or visit russellinvestments.com/institutional Important information Fund objectives, risks, charges and expenses should be carefully considered before investing. A prospectus containing this and other important information can be obtained by calling or visiting RussellInvestments.com. Please read the prospectus carefully before investing. Target date fund investing involves risk, principal loss is possible. The principal value of the fund is not guaranteed at any time, including the target date. The target date is the approximate date when investors plan to retire and would likely stop making new investments in the fund. Please note information shown is based on assumptions. Expected returns employ proprietary projections of the returns of each asset class. We estimate the performance of an asset class or strategy by analyzing current economic and market conditions and historical market trends. It is likely that actual returns will vary considerably from these assumptions, even for a number of years. References to future returns for either asset allocation strategies or asset classes are not promises or even estimates of actual returns a client portfolio may achieve. The assumptions do not take fees into consideration and all returns are assumed gross of fees. Asset classes are broad general categories which may or may not correspond well to specific products. Opinions and estimates offered constitute our judgment and are subject to change without notice, as are statements of financial market trends, which are based on current market conditions. Adaptive Retirement Accounts are a product of Russell Investments Capital LLC ("RICap"). The implementation of Adaptive Retirement Accounts in investors' portfolios and related investment advice are provided through investment advisers and other financial intermediaries that are independent of RICap and its affiliates. The advice provided by RICap in Adaptive Retirement Planner is based on asset-class level assumptions only. An Adaptive Retirement Account will review a participant's asset allocation on a quarterly basis and automatically shift the participant's portfolio based on the participant's current age, gender, account balance, savings rate and salary and Russell Investments capital markets assumptions. Based on the Adaptive Retirement Account methodology, the asset allocation may become more conservative by increasing its exposure to fixed income or more aggressive by increasing its exposure to asset classes that have historically had higher risks (e.g. stocks). Adaptive Retirement Accounts are based on a quantitative model (the "model"). There can be no assurance that the model will enable Adaptive Retirement Planner to achieve its objective. Models may be flawed or not work as anticipated. Investing in an Adaptive Retirement Account involves risk; principal loss is possible. The principal value of the account is not guaranteed at any time. Advice provided by the online Adaptive Retirement planner tool is complimentary, however, the participant will be charged a fee if they decide to invest in an Adaptive Retirement Account. Investments in an Adaptive Retirement Account are not guaranteed. Securities products and services offered through Russell Investments Financial Services, LLC, member FINRA, part of Russell Investments. Russell Investments ownership is comprised of a majority stake held by funds managed by TA Associates with minority stakes held by funds managed by Reverence Capital Partners and Russell Investments management. Frank Russell Company is the owner of the Russell trademarks contained in this material and all trademark rights related to the Russell trademarks, which the members of the Russell Investments group of companies are permitted to use under license from Frank Russell Company. The members of the Russell Investments group of companies are not affiliated in any manner with Frank Russell Company or any entity operating under the FTSE RUSSELL brand. Copyright Russell Investments Group, LLC. All rights reserved. This material is proprietary and may not be reproduced, transferred, or distributed in any form without prior written permission from Russell Investments. It is delivered on an "as is" basis without warranty. First used: September 2014 (Reviewed for continued use: October 2017) AI Russell Investments // Review of Russell Investments target date fund methodology 16