Sunday, June 3, 2012

Choosing a Retirement Income Strategy: A New Evaluation Framework

Today's blog post introduces another new research article, which follows from the article I introduced yesterday. This article provides my initial attempts to build an evaluation framework to compare and analyze any sort of retirement income strategy, including fixed and variable systematic withdrawals, fixed annuities, GLWBs, and building a floor with partial annuitization. This article is, "Choosing a Retirement Income Strategy: A New Evaluation Framework." After clicking on that link you do need to follow through and click two more links and then you will have the PDF of the article.

This article presents the initial stages of a new evaluation framework for choosing among retirement income strategies. The investigation includes eight retirement income strategies: constant inflation-adjusted withdrawal amounts, a constant withdrawal percentage of remaining assets, a withdrawal percentage based on remaining life expectancy, a more aggressive hybrid withdrawal percentage, inflation-adjusted and fixed single premium immediate annuities, a variable annuity with a guaranteed living withdrawal benefit rider, and a strategy which annuitizes the flooring level to meet basic needs and uses the hybrid withdrawal percentage for remaining assets. 
These eight strategies will be analyzed with six retirement outcome measures over a 30-year retirement period: the average amount whereby spending falls below the minimally acceptable level, the average spending amount, the remaining bequest at the end of the retirement period, the minimum spending amount for any year in the retirement period, a measure of whether spending increases or decreases over time defined as spending in the first year divided by spending in the 30th year, and the value of total spending after accounting for diminishing returns from increased spending for a client with somewhat inflexible spending needs. 
Do note that there is no mention of failure rates, as failure rates do not have much meaning here. What matters is how far income may fall as financial wealth is depleted.
The model is applied to three client scenarios representing a cross-section of Retirement Income Industry Association’s client segmentation matrix. These scenarios include a severely underfunded couple who would need a 15% withdrawal rate to meet their desired expenses, a constrained couple who would need a 6% withdrawal rate to meet their desired expenses, and an overfunded couple who would need a 3% withdrawal rate to meet their desired expenses.
It is built using Monte Carlo simulations which reflect current market conditions, so that systematic withdrawals and guaranteed products share compatible underlying assumptions.
It is hard to summarize the results, because it depends on the unique financial circumstances of retirees as well as their uniquely personal preferences about the tradeoffs between downside protection, upside spending power, and leaving a bequest.  Please have a look!


  1. Have you considered Variable Single Premium Annuities as a part of aa retirement income strategy?

    1. Hi,
      Not exactly. The variable annuity with the GLWB rider makes it act differently from a VA. I definitely should add a standard VA in subsequent revisions.

    2. I think that the inclusion of Variable Income Annuities would also be an interesting addition to the research. As it would provide guranteed income with the possibility of growth of the payment.

  2. I think that you are making an important contribution to the retirement income discussion.
    I have a question. Your prior posting on page 10 states "With a low enough withdrawal rate, annuitization is not necessary because it will lower the legacy value with only the slightest increase in the already high sustainability."
    Using your new framework and the AMOUNT withdrawal strategy, when do you hit this "low enough" threshold? 2.5%? 2%?

    1. Thank you, and that is a good question. I have to do some deeper digging before answering it. I think the metric to pay attention to would be the distribution of underfunding amounts and how that changes with withdrawal rates.

  3. Long before index linked inflation adjusted annuities came along, many companies offered annuities with payments that increased at a fixed, purchaser chosen rate; for example, you could choose to have payments increase at a fixed 3% a year. Among others, Mass Mutual still offers these.

    Index linked annuities mean that the insurance company takes on inflation risk. The cost of this risk is passed to the annuity purchaser. The true cost might be small, since an insurance company could easily buy and hold long term TIPS, but I wonder whether they aren't overcharging. It would be interesting to see how a fixed increase annuity compared with a real SPIA. It might be beneficial to choose a low increase rate, which would have the effect of increasing real spending early in retirement.

  4. ourbrooks, interesting question. I suppose one should compare the real SPIA to a SPIA with fixed increases equal to the breakeven inflation rate between TIPS and nominal Treasuries.

    But the real SPIA would probably still be a bit more expensive for a couple reasons: more risk about future inflation and any difficulties about fully hedging the inflation risk, and also less competition in the real SPIA market could allow slightly higher markups.

  5. Wade --

    This is great progress in addressing and illustrating the complexity and advancing it toward clarity.

    You have 8 strategies for placement-and-use of the $$, and comments above have suggested more, any of which could be "best" for someone. And six measures, which will rank differently for different folks, and I bet readers can propose more of those too.

    And I have another complication I think essential: I think investments' future-performance uncertainties are greater than reflected by our calculated SDs.

    Your graphs are a BIG advance in turning this complexity toward clarity, essential for both advisors and clients. Especially with dangers of declining mental capacities with age.

    I am enchanted by your introduction of your article as "the initial stages..." What can you tell us now of what you are working on and planning ahead, as further stages? ? ?

    Dick Purcell

    1. Dick,

      Thanks for the comment. I'm trying to get caught up on things now after a brief trip to Chicago. Sorry for the big delay.

      I think there is still a lot of work to do in this area. I need to add bond ladders, annuity ladders, delayed annuitization, optimizing the Social Security claiming decision, etc. And also combining the various strategies. And dig more into the outcome measures and how they can be presented.

  6. Re: the ourbrooks question, I've been looking at that from time to time over the past couple of years and it appears like, if the TIPS/Treasury spreads indicates 2.5%, the cost for the annuity is roughly equivalent to fixed increases of about 3.5%. My guess, although I can't prove it, is that insurers don't work very hard to "sharpen the pencil" on inflation-adjusted SPIA pricing. It's a very small market, despite being a favored product by many economists and retirement researchers.