As far as global investing history, the go to research is by Dimson, Marsh and Stratton Global Investment Returns Database that goes back to 1900. Of course we mere mortals don't have access to it. UBS apparently keeps it going: https://www.ubs.com/global/en/investment-bank/insights-and-data/2025/global-investment-returns-yearbook-2025.html
Attached are a few overviews.
You can buy a book on Amazon that has data that goes from 1900 to 2020: https://www.amazon.com/Triumph-Optimists-Global-Investment-Returns/dp/0691091943/
Apparently I can only add one fine at at time see attached:
@pizzaman All of this is speculation, even if it's by smart people, so no you're not dense. But the idea is that, unlike storks and births, there is an intuitive causal explanation between stock valuations and subsequent returns (and therefore the safe withdrawal rate) in the years to come. We see this most clearly with CAPE, which is a measure of price as a relation to profit. (Bengen says this too, I think.) When stocks are priced higher in relation to profits, subsequent returns are less, and vice versa. (See diagram attached.)
Granted the market being "x% below its recent high" in a given month isn't nearly as directly connected intuitively (since it ignores profitability). But it is also widely acknowledged that stocks are not a "random walk" and tend to revert to mean. (Something captured by historical analysis like Pralana/Bengen/ERN, but missed by Monte Carlo.) BTW, he's talking about monthly data, not the exact day! To quote ERN from his blog,
"The stock market is a Random Walk and past returns have no bearing on future returns, right? Wrong! As is well-known in finance... stocks have the tendency to mean-revert. So, expected returns tend to be higher after a steep drop and lower after a long run-up in stock prices. That’s reflected in the failure probabilities of the 4% Rule below. The unconditional failure probability was just under 10%, but it was 18.55% when we were at the market peak. So, failure probabilities are indeed impacted by past returns."
I think you are correct to remain skeptical that this is close to proven; but it's not nearly as ludicrous as storks-births.
Having said that, I don't use that table of his as gospel. For that matter, I don't even use his CAPE : SWR chart as gospel either. But they're another data point which i find a little better than safe-max, which seems too conservative. He is saying: if CAPE is all time high or the market is all time high, stick with safe-max. (For him that is 3.98% SWR with an 11/11/11/40/5 portfolio not 4.5% though!) But if CAPE is lower, or the market has dropped from its high, you can push higher. Bengen says the same thing doesn't he? Doesn't he say for most of time you can go into 5-6% but the safe-max (worst case) is 4.5%?
So I guess it comes down to: (1) Do you trust ERN or Bengen's SWR data more (3.98 vs 4.7)? (2) Do you want to stay at the fail-safe regardless of valuations, or do you want to take valuation into account to try to push a bit based on the past 150 years when CAPE or drawdown is a bit lower than its all time high?
@jkandell Yea! I am not dense. I am not saying that stock valuations are unimportant, they clearly are, it's the time frame I have a problem with. Shiller CAPE is a 10 year average, when the S&P 500 hits a record high, it is one day, big difference. Averaging 30 subsequent SWRs is better, but still a very limited data set, irregardless how good it looks on a chart. Just my 2 cents.
I think there is a misconception about what the 4.7% rule signifies. It is definitely very conservative, that's the point. It is worst case, suggesting that if you are using a SWR less then that, you are unnecessarily leaving money on the table, heirs not withstanding. Mr. B's book makes it clear that something like 95% of retires can use a bigger SWR, you just have to play around with the different levers.
@hecht790 Interesting quote from Mr. B's book:
"...economics is a social science that deals with manifestations of human behavior. THe best we have been able to do so far is to study retiement behavior, describe what's happened in the past, and build mathematical constructs to replicate that behavior. We have no underlying theories and associated mathematics we can use to predict what will happen in the future. Ours is an empirical science relying on observation, categorization, and model-building. Any prediction we might make about future behavior is merely an extrapolation of past behavior and could be invalidated by a change in that behavior."
Well said!!!!
@pizzaman Yah. As you say, put another way, it's about how much you are comfortable leaving on the table.
I guess it comes down to: (1) Do you trust ERN or Bengen's SWR data with additional asset classes more (viz 3.98 safe-max monthly data since 1871 vs 4.7 safe-max annual data since 1926.)? (2) If you stick to sp500 data (no other stock classes), do you prfer ERN's monthly 1871 data more or Bengen's annual 1926 data? (Safe max 3.8% vs 4%=basically the same.)
And then what you call the "leave on table" question: (3) Do you want to stay at the fail-safe regardless of valuations, or do you want to take valuation into account to try to push your failsafe up a bit based on the past 150 years when CAPE or drawdown is a bit lower than its all time high? That is, do you want the lowest=safeest=safemax=failsafe of only those months with high valuations; or all months regardless?
BTW, it's SP500 x% from its recent high for a given month, not day, that he's tracking.
@jkandell Going back to 1871 doesn't do anything for me. The farther back in time you go the less pertinent the data is to today.
"SP500 x% from its recent high for a given month, not day, that he's tracking. " Not sure what that means. If the S&P 500 hits an all time high just once in a month that counts? And what do you do with that information, is it just to determine your SWR and/or when to get into/out of the stock market. Just trying to understand 🤔.
@jkandell I found the book Triumph of the Optimists: 101 Years of Global Investment Returns used for half price if you are interested:
@jkandell I found the book Triumph of the Optimists: 101 Years of Global Investment Returns used for half price if you are interested:
Thank you! (Btw the link you sent me was for Triumph of the Optimists at $61! I wouldn't have been able to retire at age 60 if I'd bought books like that. 🙂 )
I've been reading summaries of Bengen's book and listening to his interviews. It does look very good. From what I'm reading I don't think he is disagreeing with ERN theoretically much, maybe just in the specific swr 4.7% vs 3.8%. Bengen has a whole chapter on valuation (using CAPE), right? What is his conclusion about what to do with regard to 4.7% given the record high CAPE now, with only one data point from the past? From what I gather Bengen has a "two factor" model that takes into account the base swr but also inflation and CAPE. I'll bet Bengen has a nice chart incorporating all of that.
I've already posted ERN's charts for CAPE-related decrease to safemax, users can download his spreadsheet and play around if they choose. The failsafe/safemax from history for ERN with the 11/11/11/11/11/40/5 "bengen portfolio" is about 4% for all-time-periods. If we are willing to go up to a 10% failure rate, the SWR for ERN goes from 4.7% (!) for all-time-periods, but down to about 4.3% at CAPE >20. If one is skeptical about the added future returns of micro/small/intl (i.e. one feels they are crucial diversification yet offer "no free lunch" in returns), then ERN's data would say: safemax is 3.8% (almost 1% lower than Bengen). And with a 10% failure rate, about 4.5% for all time periods/4.1% cape>20.
So in short, for ERN safemax is about 4% (not 4.7%), and about 4% swr at current cape if willing to have a 10% historic failure rate. It comes down to more skepticism for ERN about small and micro giving you a free lunch.
I noticed both Jeske and Bengen were at the last Bogleheads conference panel together, so I think that's worth watching.
@jkandell Going back to 1871 doesn't do anything for me. The farther back in time you go the less pertinent the data is to today.
"SP500 x% from its recent high for a given month, not day, that he's tracking. " Not sure what that means. If the S&P 500 hits an all time high just once in a month that counts? And what do you do with that information, is it just to determine your SWR and/or when to get into/out of the stock market. Just trying to understand 🤔.
It's not a major part of ERN's spreadsheet. (The major part is CAPE related, like Pralana and like Bengen.) But FWIW, what Jeske's saying is that for all the 30 year long spans (starting 1/1/1926, then 2/1/1926, then 3/1/1926... ) and filtering only those that were at a recent "all time high" of sp500, the safemax for that group goes down. He looks at the resulting safemax once the market drops from that high (10%, 20%...50% drop etc), and noticed the SWR goes back up, and at crashes it's higher than the all time period SAFEMAX. He attributes this to the fact stocks have a return to mean and are not a "random walk" like monte carlo and many economic models assume.
There are several ways to use these parts of his spreadsheet. One could ignore it and just stick with his CAPE filtering (like I do) or ignore that as well. Or, each month, when one calculates the SWR for their portfolio, one could adjust for how far the market is off of the high. So for instance right now you might withdraw less this month (per his chart being "sp500 recent high"); but say next month the market goes down 25% due to the jobs report... Well for that next month, you'd withdraw the SWR listed under "25% from peak". Jeske himself used the charts to justify why he retired near a peak: he noticed that even if had to start off "low" it was typical that the market would "crash" and then he could take out more and be on track for 4%.
Like I said, it's not a major part of his analysis, just something he explored to incorporate the "return to mean" we see when the market is at a peak or at a trough.
I ignore it.
But I still prefer ERN's spreadsheet because I can adjust the length of the retirement from 30y; I can utilize his monthly data rather than yearly data (12 more runs per year!); and I like how he works in CAPE. It's a small thing, but I like that I can enter my cashflow (social security/pension and future expenses), and get the Failsafe Consumption Rate, not just the failsafe withdrawal rate. But I think he and Bengen are blood brothers, with about 70% overlap. And of course Bengen invented the whole idea!
@jkandell That is the correct book link, you were concerned earlier about finding info on international stock returns prior to 1970.
Bengen has a whole chapter on valuation (using CAPE), right? Not quite a whole chapter but he does talk about it a lot throughout the book. I explained some of it in my post dated Sept 1, 2025 @ 7:24PM. He uses CAPE in conjunction with inflation (see Figure 2.7 in my earlier post). Attached is an example Table 2.3 using the two together. He also makes the point that you can't just use Shiller CAPE alone. There are different time frames with similar CAPE values that resulted in very different SAFEMAX percentages.
Table 2.1.
SAFEMAX vs. Shiller CAPE for Selected Retirees
(Tax-advantaged accounts, 30 years longevity, 55%/40%/5% stocks/bonds/cash)
Retirement date Shiller CAPE SAFEMAX % Difference Low to High SAFEMAX
January 1, 1932 9.31 11.30 %
April 1, 1984 9.31 9.02 % +25 %
July 1, 1931 15.52 10.51 %
October 1, 1987 15.53 6.78 % +55 %
April 1, 1967 21.69 5.40 %
April 1, 1995 21.64 7.80 % +44%
Mr. B states that his research assumes a passive "buy-and-hold" strategy with your investments because it would be impractical to analyze retirement portfolios not using that assumption, way too many variables. However, he does not endorse that approach for individual investors. He strongly believes that retirees should use "risk management" techniques to protect their investments from the devastating effects of a major bear market. Risk management is not the same as "market timing". "Risk management means making gradual, incremental changes to your portfolio allocation (primary in equities, but the concept can be extended to fixed-income investments as well) in response to one's assessment of market risk. Risk in this context means the probability of a significant decline in prices in the near future."
@pizzaman I find it fascinating to compare Bengen's version of how CAPE affects things and ERN's. Bengen ingeniously finds a "best fit" curve that fits Failsafe/Safemax to CAPE. Love it. ERN cautions against getting that granular (because too few cases at the extremes) and has two broad "boxes" of CAPE (>20, <20). Bengen includes two factors (cape and inflation), ERN includes two other factors (cape and sp500 drawdown from high).
In my last post I forgot one important word in my sentence ...protect their investments from the devastating effects of a major market. Bear, as in "major bear market". I fixed it. Also attached is a better Table 2.1. SAFEMAX vs. Shiller CAPE for Selected Retirees.
