All Things William P Bengen and his New 4.7% Rule

Sounds like Craig Israelsen (7Twelve Portfolio) and Mr. Bengen are converging. Real estate and commodities appear unmatched, as yet.

Hey @chrisb, thanks for your comment. Not sure what you mean by the unmatched asset classes. Anyway, this is what Mr. Bengen said about other asset classes and why he did not include them:

"In this book, I intend to focus on asset allocation only from the perspective of increasing SAFEMAX without increasing investment risk."

"Adding a few more asset classes would likely, in fact, increase Universal SAFEMAX ... from 4.7% to something higher." "...I increased the number of asset classes included in my (2001) analysis from three to seven, and SAFEMAX rose from about 4.5% to about 4.7%. That's pretty minimal!. I suspect we are now in the realm of diminishing returns..."

I suspect you are correct.

My "unmatched" comment was only meant to highlight that Bengen's choices do not [yet] match Israelsen's for real estate and commodities. The quote(s) you mention from Bengen's book seems to support his reasoning. I've looked at the 7Twelve approach off and on over the years and it always seemed liked too much work for too little benefit. I find myself leaning more and more toward a KISS philosophy. It's easier and adequate for my purposes.

So what are the historical returns for the proposed portfolio? Attached are some quick backtests of the proposed portfolio by Asset Class and some example ETFs using Portfolio Visualizer.

Asset Classes:

For International Stocks, I also compared 2 different Asset Class Allocations: Global ex-US Stock Market and International Developed ex-US Market

I also compared the addition of 5% Gold (at the expense of Intermediate Term Treasuries) to the Asset Class in "Version C".

The benchmark is the Vanguard Balanced Index

And for Exchange Traded Funds:

For ETFs, I selected some with low expense ratios (as possible) and a reasonable track record for comparison. A Developed ex-US ETF was chosen for International Stocks. The addition of Gold is also included for comparison.

@wallace471 Wow! That is great information. Mr. B does not talk about rate of return (ROR), he is only concerned about SAFEMAX. He did use a different AA that was 63% / 37% (stock/bonds). In 95% of the test cases using the new AA, the SAFEMAX was higher. But Universal SAFEMAX (worst case) dropped from 4.68% to 4.60%. Plus this AA it more volatile, so you have to consider your risk tolerance. He also noted that SAFEMAX is about the same over a wide range of stock allocations (still using the 11% allocation for the individual stock portion) from 46% to 73%. So there is a big range of options in your AA and still have a good SAFEMAX. What happens to your stocks and bonds in your first 10 years of retirement will determine how well your finances will do over 30 plus years. This is were inflation, timing of Bear Markets and Shiller CAPE 10 come into play. Nobody knows what those 3 parameters will do over the next 10 years. Everyone will need to determine their risk tolerance when considering their preferred 30 year ROR.

Another facet of his research was portfolio rebalancing interval. Using his standard configuration, he selected 20 retirees from the 269 "stalwarts" (those having a full 30 years of historical data) spaced evenly between 1 January 1926 and 1 January 1993. The result was that it didn't matter much if you rebalance every quarter or every three years, SAFEMAX varied only from 7.37% to 7.53%. However, if you looked at individual retirees, the results varied greatly. If you retired on July 1, 1941, you were better off never rebalancing over your entire 30 year retirement. His general conclusions:

• "Generally, retires who experience a larger bear market early in their retirement are better off deferring rebalancing (i.e. taking longer intervals between rebalancing).
• Retires who do not encounter a large bar market early in their retirement improve their SAFEMAX by more frequent rebalancing.
• Rebalancing every fourth quarter (i.e. at the end of every investment year) produced a decent SAFEMAX for virtually all historical retirees and represents a reasonable compromise.
• The choice of rebalancing interval is not permanent and may be altered during retirement to take advantage of changing conditions."

Posted by: @wallace471

↑

So what are the historical returns for the proposed portfolio? Attached are some quick backtests of the proposed portfolio by Asset Class and some example ETFs using Portfolio Visualizer.

Attachment : Bengen-2025-AssetAllocation_Backtest.pdf

If you use asset classes rather than specific funds in portfolio visualizer, you can get another seven years of results, back to 1986: (At least in the free version.)

https://www.portfoliovisualizer.com/backtest-asset-class-allocation?s=y&sl=6KYypgV5U0W7BsdZx0WWum

In the subprime mortgage crisis Bengen's new portfolio would have lost 28%. The CAGR during this period would be 8.6% (5.5% inflation adjusted) with a 7.5% safe withdrawal rate. With this particular starting point, the old 60/40 would have actually beaten the new mix.

If you do a bootstrap style monte carlo on this portfolio during this same period to generalize the findings, you'd end up with this:

https://www.portfoliovisualizer.com/monte-carlo-simulation?s=y&sl=3tir96KL2AfRv4sMWYLl3f

During this span, it certainly looks like you could easily get away with very high withdrawal rates. With a 10% failure during this period you could still withdraw 5.47% annually.

But, to say the obvious, we'd really want to go back to 1871 or 1928 if we could, so not sure how much meaning only fifty four years of data contains.

Back to asset allocations (AA). Mr. Bengen (Mr. B) looked at "rising glidepath" as described by Wade Pfau and Michael Kitces, which "describes a continuous, gradual increase in equity allocation during retirement." First Mr. B used his method to look at decreasing glidepath. This is similar to the common idea that you should subtract your age from 100 and the resulting number is the percentage of your stock allocation. Meaning your stock allocation decreases with age. Mr. B's results show that "a declining glidepath deleteriously affects retirement withdrawal rates." A rising glidepath, say 1% a year, "produces a higher SAFEMAX for every retiree, without exception." Using the 1% rising glidepath with his standard AA resulted in the Universal SAFEMAX (worst case) increasing from 4.68% to 4.84%. Oddly he makes the point he is not recommending rising his 4.7% rule, not sure why.

OK, now we have a Universal withdraw rate of 4.7% that just about everybody can count on, as much as you can count on anything in this life, based on Mr. B's asset allocation and annual rebalancing, bumping up to 4.84% if you use a 1% rising equity glidepath. But this withdraw rate only applies to one retire out of the 349 retires that Mr. B evaluated. The safe withdraw rate for the remaining 348 retirees is greater than 4.7%. So, that begs the question, what withdraw rate should I use for myself? Well, that depends on the financial conditions at the start of your retirement. The two most important conditions according to Mr. B is stock valuation and inflation rate. Using these two items is a big advance to the original 4% rule. He suggests that a beginning withdraw rate can be estimated by combining these two items into a series of tables. He breaks down inflation into three groups; Low Inflation (0% to 2.5%), Moderate Inflation (2.5% to 5.0%) and High Inflation (above 5.0%). Within each of the three inflation groups, you sort with the Shiller Cape 10. For example, assume your are in the Moderate Inflation group, say 3% inflation like we are experiencing today, and a Shiller Cape of 23.50. Your SAFEMAX would be 5.68%. Figure 2.7 attached. Thoughts??

Mr. B doesn't use Shiller Cape above about 24 because high Cape values have not been super high until after 1997, meaning the present high cape values have not gone through 30 years of retirement data. The record high of 44.2 occurred in December 1999. There is evidence that once you get above a Cape of about 30 it's effects start to flatten out. If you look at the 17 retires between July 1997 and July 2001 (even though they do not have 30 years of data), when the Cape was between 32 and 44, the SAFEMAX did not go below 5.4%.

https://www.bengenfs.com/charts-tables-for-you/

@pizzaman But this approach is still for folks who want to withdraw the same amount every year (with inflation additive), right? I haven't come across any clients who want to do that. They want to spend more in their healthier early go-go years of retirement, and less in their sedentary years (assuming the big expenses then like medical and LTC are already earmarked/set aside). Isn't this why we have more intelligent planning tools? It's interesting that the 4% is bumped up though, and folks do use it as a benchmark when scanning down their year-by-year withdrawal/spending rates (i.e. if it seems over the "4% now 4.7%" rule of thumb on average they get a bit concerned)

Hey Bill. I have only discussed about half of what is discussed in his book. Mr. B makes it clear that his analyses are not meant to be one size fits all and that you should adjust your personal plan as needed. His SAFEMAX numbers and AA are meant to be a base level that you can be mostly assured will be safe, and then build upon to fit your own situation. Also note that his analyses only applies to tax advantage accounts (IRAs) and does not include Social Security, taxable accounts or pensions. You have to look at all your retirement incomes and compare to your expect annual expense and see where you are at. More ideas to be discussed!

I guess this is an obvious point, but what data is actually available for International stocks historically? Most databases don't start until 1970 and just substitute US stocks prior to that. Only Europe was well enough developed prior to WWII to have much of a stock market and that was pretty well flattened by the war. And WWI had demolished things just a generation before that. I looked at a couple of tables from Bengen's website and it's odd that his data shows the same growth for international through 2022 as US Large Cap. I have international data post 1970, and during that time, the S&P500 smoked it, returning 2.5X as much. So we are to believe that in spite of WWII, other countries outperformed the US by a factor of 2.5 prior to 1970 in spite of getting bombed to rubble in WWII?

It was also a very different historical environment for small and micro cap stocks before computers - most of their outperformance came from the days when it was possible to be the only analyst that had looked at a small company so you might really discover value that others hadn't. Nowadays by the time a company goes public, it's been well studied, so the founders might make money, but it's harder to earn money buying their stock.

Even Treasury yields are synthetic prior to the 1950s or for some durations, the 1970s, as there were years without regular auctions until the US really dove into deficit spending.

I think it's a stretch to draw conclusions and hope for 2+ digit accuracy about the future from such different environments and sparse data.

@ricke Mr. Bengen told me he uses the Center for Research in Security Prices, LLC (CRSP) database. https://www.crsp.org/

CRSP's flagship databases include:

• Common stocks on the NYSE from 1926, AMEX from 1962, and NASDAQ from 1972
• CRSP Indexes
• NASDAQ and S&P 500 Composite Indices
• NASDAQ and AMEX Industry Indices
• US Treasury bonds
• Survivor bias-free mutual funds
• Market capitalization reports;
• Proxy graphs for 10-K SEC filings
• Other custom datasets

In partnership with Compustat, CRSP provides the CRSP/Compustat Merged Database, and in partnership with the Ziman School of Real Estate at UCLA's Anderson School of Management, the CRSP/Ziman REIT

https://en.m.wikipedia.org/wiki/Center_for_Research_in_Security_Prices

As far as International stocks, there are no good databases (or useful information) prior to about 1970.