Set forth below is the text of an e-mail that I received recently from Adam Butler. Adam is Director and Wealth Management & Portfolio Manager at Butler, Philbrick & Associates.
I am intrigued by your ideas, many of which overlap with the approach that we espouse at Butler|Philbrick & Associates. I would be delighted to hear your thoughts on our techniques and assertions, many of which are described in detail on our blog: http://gestaltu.blogspot.com.
Some thoughts on your approach, as i understand it. I eagerly await clarification where I have misunderstood:
Regarding the regression of long-term stock returns that you use in your models, I wonder about the statistical significance of the regression using all of the the available months. When I regress Shiller PE data back to 1870 against forward 10 year returns my findings suggest that the linear relationship is not strong enough to form the basis of a prescriptive model. This is due to the fact that intermediate valuations provide little guidance to long-term returns However, when i isolate the top and bottom quartiles of valuations and regress against the forward returns from these valuation levels, my regression strengthens considerably. Please see our blog post entitle Why Now Is Not The Time For Buy and Hold (http://gestaltu.blogspot.com/2010/03/why-now-is-not-time-for-buy-and-hold.html ), and Expensive Markets Revisited for some of our (and others) thoughts on this.
I am extremely interested in how you have applied the findings from the Shiller PE regression analysis to find the safe withdrawal rate. Have you taken the expected return at a specific Shiller PE (based on your regression of all data) and combined it with the TIPS return, with specified allocations to each asset class, to arrive at an expected portfolio return, and then run a mean/variance analysis for various withdrawal rates to a 30-year horizon? What historical volatility numbers do you use? Do you include fees?
Have you read some of Moshe Milevsky’s work on safe withdrawal rates? If not, I encourage you to check out the following papers: (http://www.qwema.ca/pdf_research/2007APR_RUIN.pdf) and (http://www.qwema.ca/pdf_research/2007SEPT_SustSpending.pdf).
Our approach incorporates our findings on Shiller PE regression vs. long-term total real equity returns (1st and 4th quartile only) with Dr. Milevsky’s gamma distribution approach to sustainable withdrawal rates (See http://gestaltu.blogspot.com/2010/07/three-horsemen-of-retirement-apocalypse.html).
We have extended this approach by applying timing systems, such as Mebane Faber’s Multi-Asset Tactical Asset Allocation strategy (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1476225), and applied simple momentum overlays with extensive and robust back-testing, to offer strategies that substantially improve the safe withdrawal rates offered by buy and hold strategies from these lofty valuations. Note that we have adjusted the back-tested returns to our models by reducing them by the difference between long-termaverage stock returns (6.6% real total returns) and current regression prescribed long-term returns. This aligns our back-tests with the implications of lofty market valuations (See http://gestaltu.blogspot.com/2010/04/and-now-good-news.html).
My partner, Mike and I would be very interested in touching base over the telephone to discuss our mutual findings in an effort to swap best practices. You will find our contact info, and a little bit about us on our web site: http://www.butlerphilbrick.com.
I look forward to hearing from you soon.