An earlier blog entry described the background of my recent correspondence with Michael Kitces on safe withdrawal rates (SWRs). Set forth below is the text of an e-mail that Michael sent to me on August 26.
Rob,
A few quick thoughts, as unfortunately my time is a little constrained this morning but I did want to get back to you!
— As a general framework, yes I agree completely that new research builds on old research. I most definitely do NOT view the existing body of research on SWRs as the final conclusion. It is a body of research that is still developing and evolving. So I welcome all new contributions and work that seeks to build upon the body of knowledge and advance it further (and hope that my own work on SWR valuation impact will be viewed as such a contribution).
— I do believe that Russell’s work has to be viewed as theorizing at some level. We do not KNOW how the subsequent market returns WILL operate coming off of P/E10 valuation levels upwards of 40 (i.e., circa 2000), because it’s never happened. We can apply reasonable theories to at least assume that the long term returns will be diminished – certainly, we have an ample framework within the economics body of knowledge to help us clearly understand the relationship that P/E contraction has to forward returns. But the ultimate safe withdrawal rates are not built solely upon the average returns over the time period. They are also built upon the return sequencing, and we do not know for certain how such overvalued markets ultimately will unwind (a point you make later in your email – impact of both starting valuation, and return
pattern/sequencing). Certainly, from an SWR perspective, there is a dramatic difference between correcting valuations through sideways markets with earnings growth (contraction via the denominator) versus price declines (contraction via the numerator). The timing can still be crucial from an SWR perspective, even though either can (and I expect will) validate the general framework of depressed long-term returns due to overall P/E contraction. So this is the context in which I suggest that Russell’s work still “theorizes”. I also believe it “theorizes” at some level, because to my knowledge NO ONE has good research on how the depth of today’s available diversification options will really impact the whole picture (e.g., easy ETF availability of gold, commodities, various forms of real estate, international, small caps, emerging markets, etc.). Granted, as the saying goes “diversification today isn’t what it used to be” – but there is still a lot of uncertainty here. Again, I don’t necessarily believe that Russell’s conclusions are wrong – my only point is that there’s a difference between what has actually BEEN observed, versus what we reasonably anticipate given our current framework to model the future (given the risks that our models, assumptions, logic, etc. may turn out to be wrong).
— Likewise, I don’t really view Russell’s conclusions as being “different” from the Old School. At “worst,” they are simply additive. The Old School has never put forth the safe withdrawal rate for a P/E10 of 40, because the Old School (via its methodology of historical observation) has never SEEN such a valuation period and the subsequent market returns and the safe withdrawal rate that would have occurred. I view the Old School as agnostic on this issue (at least, if their research results are stated properly), not in disagreement. Certainly, the Old School research can and has been overgeneralized by some (in a dangerous way sometimes, I agree), and incorporating valuation gives us a language to help explain WHY the year 2000 environment was different than anything the Old School studied. But I don’t view these bodies of research as contradictions (yet – ask me again in 2029 🙂 ), but as progressions.
— I agree with you on the absurdity of the paradox between the year 2000 and 2002 retirees. Thus, my research work in this area, and the reason why I framed my research specifically in the context of that paradoxical issue! 🙂
— It’s also worth noting that I agree with you that different valuations merit not only different SWRs, but also different investment strategies. Our firm applies a tactical asset allocation approach for managing our clients’ wealth, built on this exact framework. We do NOT believe in passive investing, primarily because it unnecessarily exposes clients to high risk during high valuation periods, notwithstanding the fact that we KNOW it’s risky because we KNOW the risk and long-term impact of high valuations. This philosophical approach on the investment side is what draws many of our clients to the firm. And I still intend to follow-up my prior research and valuations by exploring the impact of basic portfolio allocation changes in response to market valuation – I expect it will also yield very interesting results! I alluded to this somewhat directly when I briefly discussed the investment implications in my newsletter research article. But the bottom line is that I agree with you fully that there are significant risks inherent in the passive investing approach that are not understood by most (consumers and financial planners) when viewed from a withdrawal lens, and today’s markets are proving the point to the detriment of many.
– I do still disagree with you that the Old School methodology is so “rooted” in the passive approach. True, it does not make portfolio adjustments in the SWR allocations along the way, but as far as I can see neither does your calculator (it assumes that starting equity exposure is the same equity exposure throughout). These are all implicitly passive portfolios. And again, from my perspective that simply means that we still need to further explore the impact of being more “active” in investment management and its impact on SWRs – it doesn’t invalidate active management, but is simply agnostic on the issue. At the end of the day, the “worst” period studied in the Old School research was a P/E10 in the low-to-mid-20s, and it produced an SWR of about 4%. Your own methodology and calculator produces a fairly similar result. In other words, you’re both honing in on the same fundamental worst case scenario. The difference is that your research work is expanding the scope to say “what if valuations were WORSE than that” while the Old School simply never addresses that question. But I don’t view either as being particularly predicated on passive investing any more than the other. The difference is simply that the Old School assumes that the worst scenario in history by definition will incorporate the worst valuation in history (and it basically does), while you look forward at the possibility that valuations may be even worse (which the subsequently were). In essence, that means you were both right in your respective paths. Whether you should invest differently BASED on that outlook, though, is really a separate discussion from either body of work (albeit a natural extension from here – but that’s how the body of research builds on itself over time).
— The Old School never said that valuation doesn’t matter, per se. To the contrary, it said that you should ALWAYS take an SWR of about 4%, REGARDLESS of valuation, inherently forcing people to be conservative even when valuations are favorable. It is anchored ONLY to bad valuations (at least as measured by those we’ve seen in history with a subsequent 30-year track record to show the outcome). If valuations do matter, then the New School will tell you to target a 4% SWR with valuations in the low-to-mid-20s, and so will the Old School. The difference is that when valuations are MORE favorable, the Old School is unnecessarily conservative (one of the points of my research), and when valuations tread outside the range ever seen by history, the New School TRIES to theorize the likely outcome while the Old School waits to see what happens and makes prescriptions thereafter. The latter point is probably the most effective criticism of the Old School from a public policy perspective, because it doesn’t do anything for the “first” generation to actually live THROUGH a new-record high valuation period (as the retirees circa 1999-2001 did).
Well, that didn’t turn out to be as short as I’d planned. Time to wrap up a few other projects for the morning. 🙂
With warm regards,
Michael
John Walter Russell says
These discussions are excellent. Thank you.
Have fun.
John Walter Russell
Rob says
I agree, John.
The newcomer who reviews my exchanges with Michael pretty much obtains from doing so an overview of the major questions examined during the first six years of our investing discussions.
Rob