Wednesday, July 24, 2013

Decomposing SPIAs: Rising Equity Glidepaths vs. Mortality Credits

Michael Kitces and I have completed a research article together called, "The True Impact Of Immediate Annuities On Retirement Sustainability: A Total Wealth Perspective." That link will bring you to the complete research article at SSRN. Michael is also blogging about the article today. His post provides an excellent and detailed summary of the article. What I will try to do here is to provide a shorter and more intuitive explanation for what we are doing.

This article is about what happens when you partially annuitize your portfolio at retirement. We have known since the classic article called “Making Retirement Income Last a Lifetime," which was written by John Ameriks, Bob Veres, and Mark Warshawsky in 2001, that purchasing a SPIA with part of one's assets at retirement can actually reduce the failure rate for the portfolio, and possibly even allow for larger bequest after enough time passes for remaining assets to grow in value. This idea was also a big part of the article on the efficient frontier for retirement income which I published earlier this year, which showed that portfolios combining stocks and SPIAs could support the meeting of a spending goal more effectively while also leaving a bigger pool of liquid financial assets available for other purposes than portfolios of stocks and bonds held at differing fixed asset allocations.

What we do in this article is to take a more critical look at those sorts of conclusions by decomposing the impact of a SPIA into two different components. In building retirement income strategies, it is important to look at a household's entire balance sheet, which includes financial, social, and human capital. Not just financial assets. When someone annuitizes assets, the value of their financial portfolio drops, but for the purposes of understanding their retirement income strategy, we need to include the present value of their remaining SPIA payments on their balance sheet. Then we can look at their asset allocation from their entire balance sheet perspective. What percentage of their assets are in stocks, from the perspective of their total balance sheet and not just their financial assets?

When thinking about things in this way, what we can see is that one practical implication of a SPIA is that it ends up behaving like a bucket strategy in which the retiree disproportionately spends down their fixed income assets first, leaving their stocks alone to grow (more often than not), which results in an increasing in stock allocation over their retirement.

This brings us back to an issue I first discussed several years ago here at the blog, and it is something that Michael and I will explore more in subsequent research. As it turns out, having a rising stock allocation (a rising equity glidepath) during retirement can actually support lower failure rates and better retirement outcomes than any fixed stock allocations. Essentially, target date funds would have a declining equity glidepath over the accumulation phase (which they currently do) and then an increasing equity glidepath over the retirement phase (which they currently do not). The intuition for this is that you want to have the lowest stock allocation when you are most vulnerable, which tends to be at around your retirement date when you are the most exposed to sequence of returns risk. Here are four things that can happen to you in retirement:

-Good market returns throughout retirement: you are not in the worst-case scenario situation, your portfolio will continue to grow in spite of your withdrawals, and a rising equity glidepath would beat a fixed equity glidepath in terms of providing a greater legacy value for your portfolio

-Good market returns in the first half of retirement followed by bad market returns in the second half of retirement: as your portfolio continues to grow in the first part of retirement, the withdrawal rate needed to meet your lifestyle goal will continue to decline from its initial level in your retirement will be quite sustainable. These are also good outcomes, and the rising equity glidepath might not support as high a legacy in this case, this retirement will still be successful with either sort of glidepath

-Bad market returns in the first half of retirement followed by good market returns in the second half of retirement: this is the usual worst-case scenario for retirement. It's when the sequence of returns risk impacts your portfolio, causing a loss of portfolio value in the first part of retirement which is never able to recover as assets continue to be spent and the portfolio plummets toward zero. The 4% rule, the worst-case scenario in US history, comes from this type of situation. For a 30 year retirement, the halfway point for a 1966 retiree (worst-case in history) was 1981. Market conditions were quite awful in the first half of their retirement and quite good in the second half their retirement. In this situation, the rising equity glidepath helps. It makes you less vulnerable to sequence of returns risk in the first half for retirement and then helps to support higher withdrawals and greater legacy in the second half for retirement.

-Bad market returns for the entire retirement period: This is truly some terrible bad luck, and retirement outcomes are not going to look good no matter what happens. In this case, a fixed stock allocation would outperform the rising equity glidepath.

And so back to our story, in this article we compare three scenarios. First, we look at the outcomes from using a fixed 50/50 stock/bond allocation throughout retirement. Second, we look at partial annuitization, in which half of assets are used to purchase a SPIA at retirement, and the other half of assets are left in stocks. At retirement, the stock allocation as a percentage of total assets is 50%, but since half of the assets go into the SPIA, the stock allocation for the remaining financial assets is 100%. Since there is no ability to rebalance from SPIAs to stocks over retirement, the allocation to stocks as a percentage of total assets will usually increase over retirement. This figure shows the median allocation to stocks with this strategy as a percentage of total assets for both fixed SPIAs and inflation-adjusted SPIAs:

In the third scenario, we only use stocks and bonds, but in each time period for each simulation we rebalance the portfolio so that the stock allocation in this stock/bond portfolio matches the corresponding stock allocation as a percentage of total assets in the previous scenario when we had annuitized. The reason for doing this is very simple: it allows us to decompose the impact of partial annuitization into two components. We can now see the impact of annuitization that comes from the mortality credits (the idea that the annuity provider can make payments based on one's life expectancy, because those annuitants who die early subsidize payments to those annuitants who live long) as distinct from the impact of annuitization that comes from the implied rising equity glidepath. This is important, because the impact of annuitization that comes from the implied rising equity glidepath can be replicated without annuitizing, since one simply increases their stock allocation in their stock/bond portfolio over the retirement.

Here are the key takeaways for this analysis:

-SPIAs do indeed provide benefits to their owners, such as unparalleled longevity protection. However, it must be clear that one needs to live well past life expectancy before the unique benefits of SPIAs can be enjoyed. Earlier research gives too much credit to SPIAs by mixing in the impact of the rising equity glidepath through comparing it to a strategy with a fixed equity glidepath. This compares apples to oranges.

-Inflation-adjusted SPIAs do look a lot better than fixed SPIAs with this approach. They generally provide lower failure rates, and more of their benefits come from the mortality credits as they imply a less steep rising equity glidepath than with fixed SPIAs.

-Rising equity glidepath really do help to reduce failure rates at many different time horizons. In this regard, it is really a less risky strategy than a fixed stock allocation.

-Many retirees may be uncomfortable with rising equity glidepaths, and so we could say that a benefit of SPIAs is that they imply this rising equity glidepath without the person even realizing what is happening.


  1. Very interesting research.

    Do you plan on evaluating the effect of serial partial annuitization (e.g. annuitizing half of remaining liquid assets at 10 year intervals) as well as evaluating the annuitization of a smaller portion of liquid assets (e.g. 25% or whatever it takes to satisfy current income needs for the individual) and then rebalancing the remaining liquid assets back to a 50/50 or utilizing the implied glide path of increasing equity allocation?


    1. Scott, about your first suggestion, any time you want to simulate annuity purchases in the future, it will work best if you have Monte Carlo simulations that are also able to keep track of evolving future bond yields and how they relate to bond returns. I'm still working on that.

      As for your second suggestion, we can certainly consider more scenarios, but the same general patterns will keep emerging, so I probably won't pursue that too much further. We are working on another article just about rising equity glidepaths which shifts away from the SPIA question.

    2. Wade, thank you for the follow-up.

      First, related to Anon's comment below, I didn't find a clear definition of what "failure" meant in the paper. I assume it is the inability to generate income consistent with the initial 4.5% withdrawal rate (annually increased for inflation?). If this is true, then failure in the scenario without the SPIA would literally mean that the money had run out without hope of any more in the future. For the SPIA scenario it would mean a continued income at some level less than was previously possible but definitely above zero. The state of 'failure' of these two scenarios represent very different situations. In other words, both situations may have hit the floor, but the floor is at a different height in each situation.

      Let me know if I completely misinterpreted the simulations.

      To clarify my original second question: the point I was wondering about was whether you were going to address what introducing the ability to rebalance when less than the entirety of the bond portion is annuitized does to the success of the portfolio. From your response, it sounds this will be addressed in your increasing equity glidepath paper.

  2. I like the paper, but it seems to me it mostly slides over an issue that you have raised elsewhere. "Failure" means something quite different in the different scenarios. In the 50/50 allocation and glidepath scenarios, it means no more portfolio income. In the scenarios with SPIAs, it means only annuity income remains, which is much less damaging, especially for the inflation-adjusted SPIA (because the failure usually occurs well downstream, so the nominal SPIA has lost a good bit of ground to inflation). That difference may not matter much for very small failure probabilities, but is a major consideration for larger ones. Or have I missed something?

    1. Hi, you are right. When we revise the article, we will make more clear that a limitation is that we only look at probability of failure.

      I think the best way to deal with this will be to include rising equity glidepaths in the efficient frontier article I wrote before and see how they look. I will try to do this sometime soon.

  3. Thanks for a very thought provoking article.

    The biggest weakness I see is the singular focus on failure percentages without considering the magnitude of failure.

    Here is an extreme example. Imagine a person wants a retirement lifestyle of and inflation adjusted 5% of their portfolio. And lets assume that a rising equity portfolio has an 80% chance of success. We can compare this to using all the money to buy a inflation indexed immediate annuity which let's assume pays 4.5%. Clearly this second option has a 0% chance of success since it will never be able to support 5%.

    But what does failure look like for each of these scenarios? In the first one failure could mean a lifestyle drop of more than 50%. In the second one, while failure is guaranteed, the worst case lifestyle drop is only 10%. At this point it is clear that one is not necessarily the slam dunk choice the success percentages might lead one to believe.

    If you want to use a single metric, you probably want to use something akin to expected shortfall of utility - chance of shortfall x amount of lost utility (amount of loss plugged into utility function that takes a loss aversion parameter).

    Other thoughts that came to mind.

    * What happens if you keep allocation constant through buying additional SPIAs as person ages (e.g., when stocks become more than 60% of the allocation "rebalance" by buying additional SPIA - realistically this becomes problematic 85+). This would actually be a more traditional following of the bucket strategy.

    * How does using a immediate variable annuity for the stock portion change things?

  4. This is an interesting and provocative article. Thanks for the deep thinking, Wade!

    Some thoughts/reactions:

    1. In your paper, you assume a perhaps unrealistic overall real return of 1.54% for inflation-adjusted SPIA's, although the top of the yield curve for 30-year TIPS is currently 1.35% and the yield for 10-year TIPS is around 0.40%. Similarly, your assumption of a 4.49% nominal return for fixed SPIA's is very high in relation to today's available SPIAs, where even a 4% internal rate of return is hard to find.

    2. If you are going to value the annuity income stream from a SPIA portfolio in terms of the present value of the remaining lifetime income, are you also willing to incorporate the Social Security income stream into your analysis? If so, how might that alter your conclusions?

    3. You portray the early years of retirement as ones where there is an elevated sequence-of-returns risk and the latter years as ones where an increasing proportion of assets in stocks is appropriate. However, suffering a large loss from a stock-heavy portfolio in very old age could be one that cannot be recovered from -- just around the time that expensive health events become more likely. I question the wisdom of encouraging a stock-heavy portfolio in very old age unless leaving a bequest is the main objective.

    4. There are psychological and behavioral elements that could use attention. For instance, people often buy SPIAs for peace of mind-- so they are not so psychologically captive to volatile markets. How do you put a value on sleeping well at night, worrying less about returns and accumulations, and avoiding regret about investment decisions? Given that the pain of losing a certain $ amount tends to greatly exceed the joy of an equal $ gain, I submit that encouraging a larger proportion of volatile investments (i.e., stocks) as one ages may be a catalyst for unhealthy emotions, especially if a person is unlikely to live long enough to recover from a large market drop.

    5. Is there a place in your analysis for individuals who wish to hold no equities at all -- and perhaps only bonds? This would be useful for people who have more assets than they need to retire on and deliberately choose to dial back on volatility or risk with their investments.

  5. I am 75 years old, still working
    My wife is 70 years old retired
    No debt
    own home
    1 million savings (75% fixed (bonds, govmt, corporate) and 25% equity..Do you recommend partial annuity ?
    Like to sleep at night !
    Comments appreciated


    1. Thanks everyone for comments. Michael and I are working on some revisions to get magnitude of failure incorporated. I'll respond to your great detailed comments as we make progress in this regard.

      About your question... it's hard to come up with advice based on such very limited information. At least there is one essential detail that is missing: how much spending are you trying to support through portfolio withdrawals? But still, even with that knowledge, I would be comfortable providing any direct advise, except that I still think partial annuitization can play a useful role in protecting against longevity risk, and you are now getting into the age range when partial annuitization can be most effective.

    2. I see two typos, I meant:

      I would be uncomfortable providing any direct advice

  6. Wade,

    While I have immense respect for both you and Michael and the American College, I believe this research will have some very negative unintended consequences. We have finally convinced almost the entire world that you MUST at least cover your basic expenses with Guaranteed Lifetime Income. The Wirehouse community which fought this for years is now 100% on board. I never hear any questions of SHOULD a person annuitize (that is now a given), the questions I receive is HOW MUCH and WHEN?

    Now your research will put doubt into some people's minds. Advisors will second guess the OVERWHELMING research that says you should annuitize a portion of the portfolio.

    See, regular people are not rational - they still buy high and sell low, they still panic when the market crashes, they still underestimate inflation and overestimate stock market risk, they will still suffer from dementia as they age. None of your research will change any of that.

    To think that an average person can invest just right, liquidate the right amount from the right place at the right time and do it for 30 or 40 years is totally unrealistic. Grandma Johnson will not be able to do this. And if she tries she will likely have a very bad outcome.

    The perfect portfolio in retirement is one that is SIMPLE and GUARANTEED:
    1. Cover basic expenses with Guaranteed Lifetime Income 2. Optimize the rest of the portfolio to protect against inflation 3. Have a Long Term Care Insurance policy 4. Have permanent Life Insurance for Legacy

    That is the message we should be drilling. There are 78 million baby boomers who need to hear THAT message. Not that they might be able to do as well, if they don't live a long life and invest just right. Yes, your research is fine for academics to banter about. But it is NOT for the average retiree. I hope you address that in your final release of the research. Otherwise, I believe your research will be misquoted, and misrepresented by financial advisors who have other agendas and it will be the American people that will suffer.

  7. This is so good and effective research. more useful for all traders in any segment......


  8. I have carefully read all of the posts associated with this study. This study is not representative of SPIAs (plural sense) rather it is representative in a (singular sense) "life only" SPIA. The inventory of SPIAs (plural sense) are very broad and the combinations that can be used to provide sustainability income solutions when properly designed in combination with traditional investment allocations, can create some of the most efficient retirement income plans, not only for sustainable income but also the opportunity for growth with greater "unfettered" and truly FREE liquidity than is originally perceived. These efficient solutions however require the architect of the retirement design to fully understand and comprehend the entire inventory of SPIA product design and allocation. The "life only" version of the SPIA is the least used option, unless utilized in combination with an arbitraged life insurance policy, and the one option type I see most often used for such research papers which often compare such "life only" SPIA feature to a traditional SWIP strategy (SWIP... aka assume and consume strategies). However, our retirement income research lab, which is devoted to the case design of current live client "real world" fears and concerns, when architecting such plans we must develop plans that address all aspects of sustainable lifetime income, inflation, liquidity and opportunity for growth of the portfolio. Each case of course depends on the client's health and family history, goals and fears. It is important to consider tax, fee-drag (expenses) and inflation when testing the "gross" income requirements over time. I can surely tell you that when all of these factors are fully considered and ALL available SPIA and DIA (deferred income annuity) income annuities are included in the inventory of product allocation used in concert with traditional bond and equity investments options, SPIAs and DIAs provide both un matched predictability and sustainability of of income while addressing the risk and possibility of dying too soon. When I say entire inventory of income annuities, I mean, period certain, reverse period certain (non-death benefit PC), installment and cash refund, cola adjusted options, living and death benefit commuted value options. Once the actual goals and fears of the client are known, only then can you properly design the most efficient retirement plan. Lastly I can tell you the "pure returns of today's SPIAs we are using in a properly designed way, are producing real and implied rates of 3.5% - 4.5%. When calculated based on tax-exclusion effect, no fee-drag of the dedicated assets and built-in inflation of 3% - 4%, the nominal real or implied returns are in the 5%+ range which is unfettered to market risk, investment defaults and uncertainly of traditional investments used without the incorporation of the secure income base provided by the income annuity sources.


    Curtis Cloke
    Thrive Income Distribution System

    Thrive provides an agnostic approach to the education and engineering design for retirement income solutions, providing expert consultative services to agents and fiduciary advisors for retirement in income designs that consider all investment asset classes and guaranteed income insurance product allocation options that focus on a Glidepath for a proper blend of 1) Secure Income, 2) Protected Assets and 3) Long-term Alpha Growth. Thrive was established in 1999 by fiduciary retirement income planner, Curtis Cloke. Software tools were later developed and designed in 2009 by partner and actuary, Garth Bernard, a former actuary of MetLife and the software development team of, Spencer Dillard, Chief Coder/Engineer and Leslie Prescott, Chief Market Officer, both formerly of Fund Quest.

  9. I realize that I'm oversimplifying with this comment but here goes, it seems that you are evaluating an approach producing a bell shaped asset allocation curve with the heaviest allocation to bonds occurring the year of retirement as opposed to the linear allocation one would see with an "age in bonds" approach. This is a very interesting concept. I think there is potential value to this approach if inflation risk during retirement is a major concern.

  10. Nice article.It discussed a lots of points about Equity Glidepaths and Mortality Credits.It gives a clear idea about the subject.
    financial advisor Maryland

  11. UK life insurance data that are generated exclusively for you. Looking at it from a sales perspective, the efficacy of our data is as good as your marketing approach and most importantly, timing.

  12. It is hard to make conclusion on retirement lifestyle. The scenario outcome is really provocative.
    Personal Asset Management