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 my way on August 23.
Rob,
My punctuation emphasis regarding the idea of predicting long-term returns by looking at initial market valuation was deliberate. The reality is that a significant portion of the financial planning community also still ascribes to the passive investing approach – thus the “surprise” of the predictability of long-term returns. Not all of these planners necessarily believe fully in the Efficient Markets Hypothesis, but at the least they believe that the markets cannot be predicted in advance in any reliable manner (which means they still use a passive approach).
However, I don’t believe that this was necessarily an influence on much of the SWR research. The reason a single number was used for the SWR is simply because its purpose is to BE the “worst case scenario”. By definition, there’s only one worst-case scenario. In addition, bear in mind that Bengen set the framework for this research back in 1993, and at that time the historical valuation ranges he focused on WERE the most extreme points ever seen. The outsized valuations of the technology era didn’t exist when the research started; it was an underlying implicit assumption (granted, one that may be ultimately proven false) that the worst-case scenario markets being viewed in history (as of 1993) DID represent the worst case scenario, and that something materially worse wasn’t anticipated. Thus, the “one” SWR worst-case scenario was developed.
From my perspective, this is why it was important to put forth the research that I did. To believe that market returns CAN be predictable over the long run based on market valuation is still not part of the generally accepted viewpoint for most financial planners. Likewise, then, the safe withdrawal rate implications that accompany different market valuations are likewise not part of the outlook. Thus, my attempt to advance the body of knowledge in this area. As I’ve said earlier, the purpose wasn’t to suggest that valuation couldn’t also someday lead us to a different, even WORSE scenario than what the historical record suggests – I do believe there is a high probability that a 2000-2029 retirement time period will prove to be such a scenario – but I believe that exploring the hypothetical safe withdrawal rates that might come from valuations never before seen in history is a worthwhile but separate line of research in this area. Although at this point, I suspect the impact will be more retrospective for those who did retire earlier in the decade, and unfortunately not applicable again anytime soon – although markets can cycle to valuation extremes, I also do believe that the valuations we saw earlier in the decade are not likely to be revisited for many, many years in the future.
But yes, at the end of the day, many planners still focus on passive investing as well, and don’t believe future market returns can be reliably predicted. Thus, the financial planning body of knowledge has never explored the impact of valuation on SWRs. But in the context of the original research being designed in the way that it was – yes, I suppose it inherently assumed passive markets at some level, but it was also done at a time when the historical range of valuations from 1870 to 1990 really WAS the maximum historical range!
I hope that helps a little!
With warm regards,
Michael
John Walter Russell says
This causes me to think about the big hole in the academic research: the weird way in which they have addressed statistics.
They have strained at many details while overlooking the big picture. [I am substituting for gnat and camel.]
In the classic SWR study, they should have applied a Gaussian (bell curve) distribution as an approximation and established confidence limits as a minimum. Instead, they tried to avoid statistics altogether.
In the classic Efficient Market Hypothesis study, they should have applied confidence limits and calculated type 2 errors (probability of not seeing something that is there). The Efficient Market Hypothesis would have been laughed out of existence.
You are right. Investment research is very primitive.
Have fun.
John Walter Russell
John Walter Russell says
By definition, there’s only one worst-case scenario.
To be more precise: there is only one worst case OUTCOME.
[Confidence limits can be wider or narrower.]
Have fun.
John Walter Russell
Rob says
You are right. Investment research is very primitive.
Here’s a big question — Can this change?
Is investing inherently an intensely emotional project? If so, then we can never move ahead. Emotion will always trump logic. Research will never be done in legitimate ways.
Or is progress possible? Can we learn from our mistakes and move on to more realistic models for understanding how stock investing works?
If I weren’t an optimist, none of the investing materials presented at the site would exist. And I have seen lots of reasons for optimism in the hundreds of comments by community members expressing a desire that honest posting on SWRs and other valuation-related topics be permitted.
Still, I don’t feel 100 percent confident re this one. There’s a long historical record in which emotion has dominated. There are a lot of “experts” who don’t appear to possess much confidence in the ability of middle-class investors to come to appreciate the realities. Are they better informed on these questions than I am? I hope not.
Rob
John Walter Russell says
Here’s a big question — Can this change?
I believe that David Dreman’s research has the answer: YES.
There is also Professor Shiller’s research and that of other behaviorists. Psychology is beginning to assert itself. It shows up in the numbers.
Have fun.
John Walter Russell