Tuesday, December 13, 2011

Withdrawal Rates, Savings Rates, and Valuation-Based Asset Allocation

In discussing William Bengen's interview from the December 2011 JFP in a recent blog entry, I wrote:

In Question 9, William Bengen seems to suggest that the next step for withdrawal rate research could be something like using a valuation-based asset allocation, which I have done for the accumulation phase.  Michael Kitces has explored this for retirement in one of his Kitces reports, though I don't think any public link is available.  I've thought of looking at this, but based on what I could already see with accumulations, it will be a foregone conclusion that a valuation-based asset allocation will increase sustainable withdrawal rates.  The only question, really, is by how much? Perhaps I will write up a column about this for Advisor Perspectives some day.

Well, curiousity got the best of me, and I ended up writing a full research article about this. Not just a short column. The article is called, "Withdrawal Rates, Savings Rates, and Valuation-Based Asset Allocation." Interested readers can download the article from RePEc.


  1. Wade --

    Nice, your RePEc paper!

    I have two questions -- the second more constructive than the first I hope.

    1. As I remember, at Bogleheads cjking was building stock-market reversion-to-mean into his "Monte Carlo" and ran into objections, supported by John Norstad, saying it occurred in the last century in the USA but was absent elsewhere. Is my memory correct? If so, what does this mean for your conclusions?

    2. Can we combine hist-sim and Monte on this valuations topic? -- by (a) using hist to determine a slope of X years future return rate vs. PE10 and then (b) assessing with Monte? One reason would be overcome the few-and-overlapping weakness of the population of hist samples. My other reason is concern for the less-well-off masses who can't achieve safe -- I want them able to explore tradeoffs below safe.

    Dick Purcell

  2. Interesting article. Useful elaboration of of how to exploit the "reversion to the mean" idea. Of course, going back one step, we must ask ourselves if the conditions that created the mean still hold. What stage of the Roman (American) Empire have we now reached? Mr Bengen isn't too sanguine in the interview.

  3. Dick: Thanks. About John, I didn't see the studies he was referring to. I wonder how they tested mean reversion, such as just serial correlation over a short time period, or something like that. Because that statement doesn't seem to apply to this sort of long-term issue I'm looking at. Now I have a student applying safe savings rates to the set of development market countries, and it looks like there is a strong positive relationship between savings rates and withdrawal rates in most every country, suggesting that this long-term mean reversion is not just seen in the US.

    About Q2, we have since been discussing this at Bogleheads now


    CanadianInvestor: Thanks as always. Yes, your point is valid. If the US indeed enters a period of terminal decline then valuations could continue declining without any rebound. Meanwhile, you'd be stuck in a higher stock allocation. This is a potential downside. But I hope that the US still has some good days ahead!

  4. Wade --

    The source saying that mean reversion does not exist in other countries is a paper by Jorion, here:

    In a cjking Bogleheads thread titled "Best way to do Monte Carlo simulations?", it was first cited by Verde, here:

    As I remember it, John Norstad later in the thread referred to the same Jorion paper.

    In that thread, cjking built mean reversion into his Monte and got much flak for it. The Jorion paper reports tests of 30 other countries and says NO on mean reversion.

    Dick Purcell

  5. Rob: Thanks as always for sharing your thoughts.

    Dick: as you know, but in case anyone reading this blog doesn't know, we've been discussing your points at Bogleheads:


    There is an interesting new article up at Advisor Perspectives now about the valuations issue:

    Kay Conheady's "Does the Trend Matter?"


    And from her article I learned that Michael Kitces's 2009 article about applying valuations to asset allocation and safe withdrawal rates is available on the web as well:


  6. Good thing that you got really curious for we now have this really informative and insightful financial research paper!