Tuesday, March 5, 2013

Retirement Income Newsbeat #6

Safe withdrawal rates have been back in the news as of late. At the Wall Street Journal, Kelly Greene tells us to "Say Goodbye to the 4% Rule." She discusses three alternatives, which include aiming for the efficient frontier of retirement income, using the required minimum distribution tables from the IRS, and Michael Kitces' take on the relationship between market valuations and safe withdrawal rates (which was developed in May 2008 and is decidedly more optimistic that my own take on the matter).

About the efficient frontier, my new column at Advisor Perspectives first summarizes the research article, and then digs into some of the assumptions to provide new material about how the results change when the assumptions change.

In Research Magazine, Moshe Milevsky wrote a column about his research showing that for those who own guaranteed income writers on variable annuities, the best strategy is often to start taking income as swiftly and quickly as possible. That was a story I had the pleasure of breaking with his permission after attending his conference last November. Now you can read his take on the matter.

On a personal note, I am now within the last couple of weeks before moving from Japan to the United States. Once I get adjusted, I will get back to a more regular blogging schedule with more new content. I've got lots of ideas for blog posts, but am struggling to find the time to implement them these days. Please stay tuned.


  1. Some thoughts about your first and third paragraphs (I’m still digesting your column in the second paragraph … good stuff!). From your 3rd paragraph link (Nov 23, 2012 post): “As long as the assets in the VA account are bigger than zero, then withdrawals do not cost the insurance company anything. You are taking back your own funds. The GLWB becomes costly to the insurer (which is good for you) in the case that the contract value of your assets reaches zero. Then you still receive an income for life and this is the expense which the insurance company needs to be hedging against with the income they receive from the guarantee rider they charge you.”

    A point here: The rider expenses are typically charged from the assets within the annuity. So when those assets are consumed by the income, and eventually reach zero (and now the income is all coming from insurer) there are no assets for the rider that is supposed to hedge against this either. This problem is part of the too big to fail problem blog too (http://wpfau.blogspot.com/2013/02/this-is-special-guest-post-of-email.html). The VA promised income needs to come from the insurer who is also on the hook for SPIA promises from their general fund.

    Everyone invests in the same global markets and if there is a sustainability problem under one method of income sustainability (and you reference a couple in this blog) then that is an indication that other income methods are likely under stress as well (note the issues are popping up in all forms of retiree income). Yes, the rates of return are less today than calculated in earlier times. And this goes to my general theme of dynamic planning and simulation updates; sustainable income is a function of the asset base and the two (income and assets) need to be connected to each other at all times in order to properly manage them over a person's lifetime.

    First paragraph thoughts: The academic search for a single lifetime withdrawal rate, a.k.a. SWR, will always find that SWR value is sensitive to the cyclical nature of returns over time. The current thought trend is resulting in large changes to the SWR, from 4% to under 3% … a 25% change. Clinically, none of my retired clients required such a large spending cut during the last big market decline. Basing their spending reduction on Probability of Failure (JFP Nov 2011) resulted in some of them reducing spending by about only 5%.

    This is because the single universal value (SWR) approach is trying to adjust a retiree's entire future income immediately today, rather than manage that income year by year as they age. Yes, dynamically over time if the historical market data used in simulations trends lower, then that will be slowly and slightly through each increment, reflected in withdrawal rates going down as well at that time. And at that time it would reflect in what is prudent during that year’s review when all the data including simulation data, the portfolio value, retiree current age, their updated expected longevity age (because they aged a year).

    So the profession has been seeking a single safe rate for withdrawals. There is a series of prudent rates that change based on the age of the retiree. Withdrawal rates can actually go higher as the retiree ages because their remaining distribution time gets slightly shorter each year. So the first question to ask, before one can consider a prudent withdrawal rate, is how old is the retiree (how old are EACH of them if they’re a couple) because the prudent withdrawal rate depends on time remaining (distribution period) - which should only be determined by a period life table (those statistics are better than guessing how long a retiree/couple may live). Note that WR is not the same as withdrawn income because the latter depends on portfolio value.

    Apologies for the long comment Wade. The topic has many idiosyncrasies due to its dynamic nature. The great thing about research – learning from each other. Best wishes on the move and settling in at the American College!


    1. Larry, thank you.

      Your point here is important. My understanding is that many state guarantee associations do not guarantee the guarantee riders, and so a situation in which the company is on the hook to support a guarantee means they will probably have to support the guarantee for a large customer basis, and this could be a problem if they were not doing a good job of hedging the risk.

      There is a related issue here as well, that VA/GLWBs may be lapse supported products, which means that they provide offers that are really too good to be true, but they rely on most customers surrendering the products before ever getting to the stage that the guarantee kicks in. If people hold on to the guarantees that insurance companies were not expecting, it could be a problem.

      The way you are describing your POF framework here makes it sound like dynamic programming. Are the adjustments accounting for all the possible future adjustments as well? That would be impressive to do in an excel spreadsheet. I'm looking forward to reading your articles one day soon.

      Best wishes, Wade

    2. Yes Wade, the excel program Annually Recalculates and Serially Connects each year of retirement income at the 95th, 75th, 50th, 25th a 5th simulation percentiles (David Blanchett is a maestro programmer!). What happens in a prior year affects what is possible the following year and they are chained together. Each year's length also adjusts based on Period Life Table longevity (also can be adjusted by the percentile within the Period Life Table JFP March 2012) to take into account that a person can never actually attain their current age life expectancy age - it's always out ahead of you (JFP Dec 2012). The SSRN papers appendices for all three papers have screen shots of all this. This also factors in the third dimension of allocation. Finally, these all may be calculated at various planes of POF (they layer like an onion in a 3D model) which is important to get a sense of sensitivity to sequence risk exposure (JFP Nov 2011).

      All the program does is gives you a sense of exposure over a given lifetime this year. Each year during a review, or mid year if POF goes up too much (poor market sequence), all the data is updated including retiree portfolio value and their spending may be evaluated, based on then current facts. Good years mean more dollars may be available ... maybe for spending, maybe for reserving for poor years (income smoothing if you will).

      The model is fluid and dynamic to reflect how life really is and reflects how a review and prudent decision making each year helps preserve portfolios for a lifetime.

      PS. It is this model that Dr John Mitchell and I are using in our current project for AFS which I've been cc:'ing you for you to join later this Summer once you settle in should time permit.

      Best wishes as well, Larry

  2. Wade
    Figure 2 notes an almost 99% lifetime spending need with a 40% equity/ 60% bond portfolio. If one were to prefer using stocks and bonds would it be reasonable to use a 3% fixed(variable dollar withdrawal) with a 40/60 portfolio, revisited yearly as a plan? Thanks

    1. Hi, that 99% spending is at the 10th percentile, which basically means there is less than a 10% chance for wealth depletion with that strategy. Not that it's full-proof.

      That being said, 3% is certainly getting low enough that with a few potential minor adjustments (related to what Larry is describing), it should work out fine. I think that is what you mean by revisiting yearly, right?

      The lower the withdrawal rate, the less of a need there is for annuitization.

  3. For those of us planning to live on our investment income, and other sources, I struggle with the question of buy and hold versus rotation (like the IVY portfolio suggests). If you rotate out of an investment, you also lose the dividends from that investment and would have to withdraw capital to replace the lost dividends.

    IVY seems to only make sense if you can stay out of significant drawdowns -- and buy back at a lower price. The more shares purchased, and hopefully slightly higher dividends, would overcome any withdrawn capital.

    What few studies I've seen usually don't deal with dividends and income flows -- only price and portfolio value. Have you done anything that deals with investment versus rotation strategies for retirees living on investment income?


    1. Hi,

      I'm not familiar with IVY, but I am generally quite leery of anything other than buy-and-hold for individual investors.

      I have dabbled in this area a bit in the past, and I do have an article in the Journal of Financial Planning about how, historically, withdrawal rates could be boost if one made asset allocation adjustments based on the level of the cyclically-adjusted price earnings level at the start of the year. You can see it here:


  4. Any chance you could post the figures from your Advisor Perspectives March 5 paper on the blog? The resolution on the figures is very poor. Thank You

    1. John,
      To make a long story short, it's not very convenient to make those figures again now.
      But I have a Word file with better resolution for the figures. If you send me an email, I can send you that file. wadepfau@gmail.com

      Thanks for the interest, Wade

  5. As retirement is a step that most of us employees will face in the future, financial planning must be set beforehand. But I would like to add that the term 'safe withdrawal rate' can be an ambiguous term. Take note that the initial methods used historical data to statically determine what would have been safe given the actual results that past portfolios would have generated with the given variables. The next logical step is to use that information in order to predict future SWRs. It would be clearer to use it in reference to past or projected SWRs.