Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.
For more than a decade, researchers have been sounding the alarm that those approaching retirement face unprecedented portfolio depletion risk. Specifically, the traditional metrics applied by financial advisors to client portfolios to assess sustainability may be overly optimistic relative to the economic reality and that popular retirement spending paradigms, such as the vaunted 4% rule, put consumers at serious risk of running out of money before they run out of time.
My frequent research partner, Seattle Pacific University Professor Jack De Jong, Ph.D., CFA, MBA, and I examined this threat for ourselves in, “A Case Study in Sequence Risk: A 20-year Retrospective on the Impact of the 2000-2002 and 2007-2009 Bear Markets on Nest Egg Sustainability.” The paper was posted to SSRN in May 2021 and was published in the Journal of Wealth Management in January 2022. I summarized our findings in a June 2021 Advisor Perspectives article titled, “Be afraid, very afraid of retiring in the 2020s.” As the AP article title suggested, our conclusions largely aligned with those of prior researchers. Using monthly return data from 2000 through 2019, we illustrated that a 4% inflation-adjusted withdrawal rate applied to a classic 60:40 portfolio was in serious danger of failing well short of the standard 30-year time horizon for consumers retiring at or around normal retirement age (65).
Our findings also led us to caution that even the 2000-2001 and 2007-2008 bear markets in stocks did not represent a worst-case scenario for investors, since the dismal returns from stocks during that first decade of the 21st century were partially offset by a roaring bull market in bonds as interest rates steadily declined. Were a similar bear market in stocks to begin now and be paired with today’s near-historic low interest rates or, worse, with a prolonged period of rising interest rates (i.e., a bear market in bonds too), folks who retire in the early 2020s will be sailing into an unprecedented perfect storm of sequence risk.
With conditions ripe for an existential threat to individually managed retirement portfolios, some researchers have suggested that consumers’ desire for financial security in retirement will be better served with annuities. By pooling their risk and transferring their retirement savings to an insurance company, they will get a guaranteed lifetime stream of payments. Over the past decade, academic researchers have approached the annuitization question from many perspectives and have reached different conclusions. Drawing on a rather extensive research survey, Jack and I were inspired to explore the issue for ourselves. Our latest paper, “Is the Annuity Puzzle Really So Puzzling? An Analysis of the Consumer Lifetime Annuitization Decision in a Low Interest Rate World,” was posted to SSRN in May (2022) pending journal submission. In it, we sought to quantify the tradeoff that 65-year-old consumers make when annuitizing their retirement savings by purchasing a single-premium immediate annuity (SPIA) from an insurance company relative to managing the spending portfolio themselves. Our findings have important implications for how financial advisors should approach the annuitization decision. The purpose of this article is to share them with you.
The conclusions researchers (including us) make are heavily dependent upon model design and explicit and implicit assumptions. For instance, a common criticism of some early annuitization research is that Monte Carlo simulations were based on historical mean return assumptions for stocks, bonds, and cash that are higher than what consumers could expect to earn today. Similarly, assumptions regarding portfolio management expenses and distribution strategies (e.g., spending equity first, maintaining a constant allocation with rebalancing, and spending cash and bonds first) all influence the results. Conversely, research models that apply a theoretical annuitization structure may produce different outcomes than papers that consider real-world annuity contracts available in the retail market.
For example, some researchers have sought to advance the case for SPIAs by noting that academic researchers long ago established the superiority of annuitization of pensions and that consumers have been naïve in failing to adopt the same approach with their own retirement savings. However, pension annuities tend to have higher distribution rates than retail SPIAs due to the latter’s sales and distribution expenses and the insurance companies’ profit motive.
The overarching practical finding from our research and analysis is that financial advisors would do well to apply critical thinking in assessing the validity of academic research models rather than blindly accepting the journal authors’ conclusions as gospel. With that disclaimer/disclosure out of the way, six important findings and conclusions, some of which challenge the current planning orthodoxy, are presented in our paper are as follows:
- SPIAs are more attractive to single than to married consumers, and fixed SPIAs are more attractive than SPIAs with a 3% COLA
Our findings were based on SPIA quotes for 65-year-old consumers provided by Immediateannuities.com on March 20, 2020. With respect to fixed versus inflation-adjusted SPIAs, the annuity illustrations showed that the payout from the inflation-adjusted (3% COLA) contract did not exceed that of the fixed SPIA until the annuitants reached age 79-80 and that the total return will not surpass the fixed SPIAs until the annuitants are in their early 90s. Assuming that retirees value earlier years of retirement more than later years, the lower initial distribution rate of the inflation-adjusted SPIA further tilts the scale in favor of the fixed SPIAs. For single 65-year-olds who purchase the fixed SPIA, it takes 17 years (to age 83) to get their principal back, while it takes 21 years for married 65-year-olds (to age 87) to get to nominal-breakeven.
- In individually managed portfolios, the choice of distribution strategy plays a significant role in determining portfolio sustainability, and the superiority of a strategy depends on investment conditions at the beginning of retirement.
In both our Case Study in Sequence Risk paper and our Annuity Puzzle paper, we compared sustainability results from simulation models that employed stocks-first withdrawals (declining equity glidepath), constant allocation with rebalancing (constant glidepath), and bonds/cash-first withdrawal (rising equity glidepath). In our Case Study in Sequence Risk paper, which used monthly return data for the model asset classes, we found that consumers who spent down the equity portion of their portfolios first (i.e., followed a declining equity glidepath) fared best through that 20-year period, while consumers who spent their cash and bonds first (rising equity glidepath) without rebalancing experienced shortfalls sooner. Surprisingly, when we used the same model design for considering portfolios beginning with the current low-interest rate environment, our findings ran exactly contrary. The rising equity glidepath (spending cash and bonds first) produced fewer shortfalls and greater total return (income + remaining balances) than constant allocation (traditional rebalancing) and declining equity glidepath spending strategies. Although these results seem counterintuitive, they make perfect sense. For the 2000-2019 time period, bonds outperformed stocks. Today, we do not know what the future returns for any asset class will be, but interest rates on fixed income investments cannot go much lower (i.e., the past 20+ year bull market in bonds is not repeatable). Regardless of whether returns stay low or turn negative if interest rates rise, bond returns represent a relatively low hurdle for stocks to clear moving forward.
Our conclusion that the choice of retirement spending strategy is dependent upon the investment conditions at the time of retirement is consistent with prior research by David Blanchett and Wade Pfau. While Blanchett’s research on this topic downplayed the significance of the choice of withdrawal strategy, we conclude that it is highly relevant, especially in the current low-interest rate environment. This presents a conundrum for financial advisors, since nearly all popular financial planning software implicitly assumes that constant allocation with rebalancing is optimal and does not allow advisors to test different spending strategies.
- Our simulation results suggest that, when starting from a low interest rate environment, the initial allocations that provides the greatest total return (lifetime income + remaining balances) and fewest shortfalls within the 30-year time horizon are those with initial equity allocations well north of the standard 60:40 model and closer to 80:20.
Although this finding may seem blasphemous, low or negative returns from bonds early in retirement enhances the likelihood of shortfalls in allocation models that are heavily weighted to them. Our findings are decidedly at odds with those of researchers who are advocating for the classic 60:40 model or even more bond-heavy allocations. In our model, the 20% allocation to cash/bonds provides sufficient protection against sequence risk in stocks to produce lower shortfall risk and greater remaining balances than the staid 60:40 model.
- While some researchers have cited mortality credits as a definitive reason for retirees to annuitize at retirement rather than take on sequence and longevity risk by managing their portfolios themselves, our results suggest that the value of mortality credits for retirees, particularly couples, in their 60s may be largely offset by the insurance companies’ sales and distribution expenses and profit motive.
We are not alone in reaching this conclusion. Other researchers cited in our paper have shown that value of mortality credits is more pronounced for retirees in their 70s and 80s than for those in their 60s. Our conclusion is highly relevant to retirement planning, as there has been a great push in recent years toward making annuitization a mandatory option within 401(k) plans. Our paper suggests that the election by retirees at normal retirement age to purchase a SPIA in a 401(k) plan or a QLAC in an IRA may be sub-optimal.
-
The decision to annuitize at age 65 may be optimal in instances of extreme longevity (age 90-95+) or if there is a prolonged environment for stocks and bonds that is as bad or worse than any conditions experienced in the historical record. Conversely, within a normal 30-year retirement time horizon and in investment environments that are even modestly less apocalyptic, most retirees will be better served by maintaining control over their retirement nest eggs.
We reached this conclusion first by comparing simulation results (total distributions + remaining balances, if any) from standard static inflation-adjusted portfolio models applying a rising equity glidepath spending strategy to the total 30-year return produced by the annuities. This exercise enabled us to quantify the sacrifice consumers make by annuitizing under investment conditions that are far worse than the historical norm. Critics may still argue that even shortfalls below the bottom 20th percentile result may not be acceptable for certain risk-averse investors. A flaw in this argument is the assumption that the static models we and other researchers use for comparing annuities are realistic. While those models are useful for illustration purposes, they are not optimal in real-world planning. This view is supported by a growing body of research (beginning with Guyton’s 2004 and 2006 seminal papers) demonstrating that the application of dynamic distribution strategies may produce significantly fewer shortfalls and higher total lifetime returns.
Blanchett’s revelation of the “retirement spending smile” showed how retiree discretionary spending tends to be at its highest in the early years of retirement, declines when age-related infirmities begin to restrict activities, and increases in the later stages of retirement as non-discretionary spending on healthcare costs increases. This work, along with other research, suggests that retirees rationally value the early years of retirement more than the later years. To the extent shortfalls in the bottom simulation results occur in the later years of retirement, the benefit of zero-shortfall risk of annuitization is overstated. This is particularly so if the client has buffer assets such as home equity and/or long-term care insurance that could be tapped to make up for retirement shortfalls. The existence of social safety nets such as Medicaid and family support further diminish the need for 30 year zero-shortfall risk.
-
Our paper also considers the merits of substituting a SPIA for the bond portion of one’s retirement spending portfolio. We find that the inclusion of SPIAs does not add value relative to traditional 60:40 or 80:20 model portfolios.
Again, research conclusions on this issue vary and are heavily dependent on model design. Our model design and conclusions are most similar to those of Hanoi University of Science Professor Nugyen Dang Tue in his 2014 Retirement Management Journal paper, “Decision Rule Strategies in Retirement Planning.”
Conclusion
My goal is to share original research and critical thinking regarding the lifetime annuitization decision using retail SPIAs in hopes that it advances the planning profession and improves outcomes for consumers. I have not provided the details of our methodology. That information along with our extensive reference list may be obtained by reading the full paper on SSRN. Our paper considered only 65-year-old consumers and our results and conclusion might be different if we considered different age groups. Lastly, our paper considered only SPIA contracts purchased by retail consumers. Analysis of other forms of annuities, such as deferred annuities, index annuities, buffered annuities, and variable annuities may lead to different conclusions than the ones we have reached.
I close with a word of caution. I have recently read a number of articles extolling the theoretical value of these often-complex instruments as risk management tools. These articles are reminiscent of papers I read in the late 1990s and early 2000s by York University of Toronto professor of finance Moshe Milevsky (and others) detailing the theoretical merits of lifetime withdrawal benefits riders on variable annuity contracts. It was not long before variable annuity wholesalers were using reprints of the articles to promote the sale of their products, even though the product features and contract restrictions in the annuities they were selling were materially different from those used in the research papers. In reading some of the recent research, I have noticed authors imploring practitioners to take the time to understand the underlying math and theory. I hope the researchers will acknowledge and point out the manner in which the insurance companies that issue these contracts alter them for their own risk management and profit motives. The devil is in the details.
John H. Robinson is the owner/founder of Financial Planning Hawaii, Fee-Only Planning Hawaii, and Paraplanning Hawaii, and is a co-founder of software-maker Nest Egg Guru.
Read more articles by John H. Robinson