I am using Nikkei225 total return data in inflation-adjusted Yen. For bond allocation, I use US 10 year treasuries in inflation-adjusted dollars. I assume (hope) this doesn't cause major inaccuracy versus holding Japanese bonds (of course, a hypothetical Japanese investor could have held US bonds).
I use 25 year withdrawal period because that is the longest period that includes retiring during and immediately after the bubble years. I do not reorder the time series.
With these assumptions, the 4% rule has a 17% failure rate in Japan with a 60/40 stock/bond holding:
0% failure "safe" withdrawal rate is 2.5%:
Stepping back from this, the problem is partially with "blind FIRE" and partially with home country bias. If we just plot the total return over the whole time period:
stocks are only being purchased at really bad rates in six years around the bubble peak. We have risk from two sources:
1. over-enthusiastically retiring earlier than originally planned on the back of high PE asset appreciation
2. having a short accumulation phase (typical of FIRE/E-RE) that lies within the bubble period due to bad luck with birth year
The best simple answer for both Boglehead 40-year DCAers and wannabe early retirees in this scenario was to invest in countries other than Japan during years of very high PE (i.e. those six bubble years - it doesnt require hindsight). The Boglehead 40-year DCAer would have been fine anyway, because he will only have bought stock at bad prices for 15% of his career. The E-REr must be more careful.
2.5% WR may be bomb proof. Reducing WR provides nonlinear safety return. At some point, you're just spending the 1/PE each year or less, even at truly atrocious valuations, and that in turn means your low draw down rate alone extends the portfolio life into the "long run" where valuations mean revert.
Below 2.5% WR, you may be more likely to be ruined by confiscatory taxation, war, being hit by a car, etc. than by any failure in your financial planning.
