The example used in this paper fails to show the end game (i didn't run it either, but can make some assumptions). Yes, Joe only has 590K vs Janes 880K due to sequence of returns during accumulation. However, Joe retires in 1994 and is about to see the biggest market boom... pretty much ever. While Jane in 2007 is about to see her's get cut by more than 30 percent even with a 50/50 allocation. I would MUCH rather retire with Joe's scenario. Knowing the future, of course. In general, one could argue that even if accumulation sequences cut you down with poor returns later in that phase (ie with more accumulated), the same market cycles and mean reversion will help correct this problem later in retirement. "Dynamic" portfolios is a fancy word for market timing. The only time I see changing a well researched AA as helpful, perhaps, is to decrease stock allocation slightly (to a more stable asset class) during the first few years of a prolonged retirement. This would help avoid a sequence of returns disaster once the buying stops. Even then, I would prefer flexibility to changing my AA.
Since no one can predict the future, It's best to use what we do know and that is 4%WR with flexibility is safe provided you keep half your assets in stocks. 3.5% for nervous nellies who will absolutely never work again, want to give lot's upon their death, and believe SSI will dissolve. Anything more and you have worked too much.
As always Tyler's expertise on AA is appreciated! Every time you post I play with your tool and get think'en!