Putting A Fixed-Income Rule of Thumb To The Test

We like to use rules of thumb, or heuristics, when facing choices.  We often default to rules of thumb because when finding the optimal choice is difficult, rules of thumb allow us to solve a problem quickly and move on in life.

While attending a lunch presentation by Charles Ellis at the recent Morningstar ETF Conference (which was a fantastic presentation and I am looking forward to reading his book “Falling Short”), I noticed that two investing/retirement rules of thumb were used in his presentation:

  1. Use Target Date Retirement funds
  2. The 4% withdrawal rule.

Being a geek, specifically regarding withdrawal rates in retirement, I immediately noticed an issue with these rules of thumb – they aren’t likely to be true if people follow both at the same time!  To understand why, we will need to 1) briefly review the history of withdrawal rates, then 2) move on to review some of the assumptions underlying the 4% withdrawal rule, and finally 3) do some Monte-Carlo simulations to see what the future might hold .


The field of withdrawal rate research is fairly young.  Prior to 1994, financial advisers would estimate a rate of return for a portfolio and, using an amortization table, solve for the annual withdrawal that would leave a balance of $0 at the end of 30 years.

Not pleased with the previous method, in 1994 Bill Bengen wrote an article in the Journal of Financial Planning that was the source of the 4% withdrawal rule of thumb.  Specifically, Bengen used historical Ibbotson data and simulated a person starting retirement in 1926 who lived to 1956.  The portfolio was composed of 50% large cap stocks and 50% intermediate term government bonds.  He solved for what starting percentage of the portfolio the person could withdrawal and then adjust that dollar amount up or down by inflation (subsequent years would have withdrawal rates very different than the initial percentage).  He then repeated this with a person starting retirement in 1927 through 1957 and continued repeating until he was out of historical data.  This produced a graph that looked like the following:

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged, do not reflect management or trading fees, and one cannot invest directly in an index. Additional information regarding the construction of these results is available upon request.

Bengen noted that a “safe” withdrawal rate would be the lowest of these, which is approximately 4%.  This 4% withdrawal rate has since been tested by others using similar historical data and the 4% initial withdrawal rate finding was validated.

After several more years, people began to use Monte Carlo simulation to test the 4% initial withdrawal rate.  The assumptions for Monte Carlo used historical data (i.e., they assumed that stock and bond returns, variation in returns and correlations would be the same as history).  The Monte Carlo simulations, using historical data as inputs, showed that a 4% initial withdrawal rate worked in 95% of the simulations.

The finding of the 4% initial withdrawal rate by Bengen and all subsequent studies are what is used as support for the 4% rule of thumb.  However, it is important to remember the assumption underlying the 4% rule of thumb:

  1. 50% stock portfolio & 50% intermediate term government bond portfolio re-balanced annually every year through retirement. Even at age 94 you would have the same 50% stock & 50% bond allocation that you started with at age 65.
  1. Withdrawals only have to last 30 years. Even if a 4% initial withdrawal rate would lead to bankruptcy after 31 years it wouldn’t matter because the money only had to last for 30 years.
  1. One must assume that stock and bond returns will be the same as history.

There are two issues that I see that affect the assumptions of the 4% withdrawal rule 1) Target Date funds and 2) current yields.