suggestions, links, corrections (except hocus) are welcome
Historical swr studies and inflation adjusted safe withdrawal rates
The rate is expressed relative to initial portfolio value. Withdrawals in subsequent years are adjusted upward for inflation.
There are two general approaches to estimating "safe" withdrawal rates from retirement portfolios.
Historically based studies have looked at various portfolio compositions and based on the amount initially withdrawn have reported the probability of portfolio survival. (Example Trinity) or the maximum "100% Safe" withdrawal rate. (REHP) These studies have limitations. If planned withdrawals start at a period of overvaluation more extreme than occurring during the course of the study questions arise about the applicability of the historical swr. Interesting attempts to test the historical swr have included rearranging the yearly data, increasing the expense ratio to simulate possible lower future investment returns, and simply reducing "swr" in a linear fashion according to degree of overvaluation/ historical extremes.
Monte Carlo analysis allows computer simulation of safe withdrawal rates based on estimates of portfolio return and variability. One limitation of this type of study is that future portfolio returns and volatility are unknown.
Some people (ok, hocus) have said that historically based swr analysis is invalid.
Please explain how the Trinity study is invalid.
The Trinity study methodology does not take into account how the SWR changes with changes in valuation levels.
Note that this does not depend on the rationale of extreme valuation for invalidity. Not all of us agree:
About the Trinity Study:
It shows what would have happened had you invested this way or that, in the past.
It doesn't make any claims about the future.
How can it possibly be "invalid"? (Unless, of course, their data was invalid.)
Are these also "invalid" studies?
On the other hand, you can just define classes of studies as invalid:
The reason why I say that the 2.0 percent number is valid and the 2.3 percent number is valid, but the 4.0 percent number is invalid, is that the first two numbers both are the product of methodologies which consider the effect of changes in valuation, while the latter is not.
Some of us wasted effort trying to tell hocus that changes in valuation occurred during the historical period and were implicit in the study. A waste of effort since he has defined valid studies as those that make explicitly use of valuation.
I thought this was an interesting scattergram of maximal withdrawal rates vs. P/e ratio:
Apparently even devoted followers of the historical approach are aware that there is an association between valuation, investment results, and safe withdrawal rates.
(Thanks to Intercst and BenSolar)
Alternatives to this type of withdrawal include:
Annual withdrawal of a fixed proportion of the portfolio
Preservation of capital with spending dividends/ income
Multiple sub-portfolios/ dynamic asset allocation