Looking for Worst Historical Real Return on 30/70 AA

[Cheego]

Given the information below, does anyone know how to calculate the worst real return?

- 30% Equities (S&P 500)
- 70% Fixed income (10 yr Treasuries)
- Inflation-adjusted withdrawals (linear, and with zero ending balance)
- Looking within any 40 year span since 1926

This sounds like a job for Nisiprius.

Thanks!

[Patzer]

Testing monthly periods starting with Jan 1926-Jan 1966 and ending with Aug 1983-Aug 2023.

For 30/70
The worst result was 3.03% and came from retiring on Feb 28th, 1946. The average result 4.37%

Worst Result by Decade:
1920s: 4.41% (Only 1926-1929)
1930s: 3.14%
1940s: 3.03%
1950s: 3.47%
1960s: 3.33%
1970s: 3.77%
1980s: 6.71% (Only 1980-1983)

You picked a pretty bad allocation!!

If you did 50/50.
The worst result was 3.50% and came from retiring on Nov 30th, 1965. The average result is 5.17%

Worst Result by Decade:
1920s 4.19% (Only 1926-1929)
1930s 3.76%
1940s 3.98%
1950s 3.99%
1960s 3.50%
1970s 3.97%
1980s 8.02% (Only 1980-1983)

How did I do the math?
The Early Retirement Now website has a spreadsheet as part of their Safe Withdrawal Series that lets you calculate data like this.
I have taken that spreadsheet and over the last few years highly customized it and vastly expanded the data, so I can calculate almost any scenario pretty much instantly.
Even the original spreadsheet there would give you pretty close to what you want, if you want to run the numbers yourself.

[Cheego]

Testing monthly periods starting with Jan 1926-Jan 1966 and ending with Aug 1983-Aug 2023.

For 30/70
The worst result was 3.03% and came from retiring on Feb 28th, 1946. The average result 4.37%

Thank you! I'll try to find that page on ERN's website.

Your 3.03%... is that inflation-adjusted?

I used FiCalc and found the worst 40 year span was from 1940 through 1979. It shows a geometric mean (CAGR) real return on equities at 5.72% and on bonds at 0.00%. With some caveman math and their output, I calculate a net real return of a 30/70 AA at 1.72%.

Here is the scenario...
https://ficalc.app?additionalIncome=%5B ... tantDollar

[Patzer]

Your 3.03%... is that inflation-adjusted?

I used FiCalc and found the worst 40 year span was from 1940 through 1979. It shows a geometric mean (CAGR) real return on equities at 5.72% and on bonds at 0.00%. With some caveman math and their output, I calculate a net real return of a 30/70 AA at 1.72%.

Sure, that's your CAGR, but that's not what you can spend, that doesn't take into effect how your withdrawals during highs and lows have a different impact on your return.
If you only spent your real return, you would never run out of money.

You stated you wanted to have zero ending balance.

3.03% is Inflation Adjusted Safe Withdrawal Rate.
That means that in the WORST situation, you could spend 3.03% of your initial retirement sum, inflation adjusted over 40 years and it would be 40 years before you ran out of money.

So, how does that play out...
If at retirement you had a million dollars, then you could spend $30,300 a year or $2,525 a month.
My calculation actually includes historic monthly inflation adjustments.
So, if inflation went up 1% in the first month of retirement, then in the second month of retirement you could spend $2,550 and so on.
Doing that, you would have been fine with an initial spend of 3.03% that was then adjusted up with inflation over time, without regard for how the portfolio was doing.

[Cheego]

Your 3.03%... is that inflation-adjusted?

I used FiCalc and found the worst 40 year span was from 1940 through 1979. It shows a geometric mean (CAGR) real return on equities at 5.72% and on bonds at 0.00%. With some caveman math and their output, I calculate a net real return of a 30/70 AA at 1.72%.

3.03% is Inflation Adjusted Safe Withdrawal Rate.
That means that in the WORST situation, you could spend 3.03% of your initial retirement sum, inflation adjusted over 40 years and it would be 40 years before you ran out of money.

Interesting. Thanks.
3.03% is the SWR, but what is the real return on the investments?

[Taylor Larimore]

Cheego and Patzer:

"Past performance does not forecast future performance." Total Bond Market in 2022 is an example--although it still helped a stock only portfolio.

viewtopic.php?f=10&t=156573&newpost=2349057

Best wishes.
Taylor

Jack Bogle's Words of Wisdom: "No analysis of the past, no matter how painstaking, assures future superiority."

[zie]

2022 was a tough year for bonds, one of the worst ever recorded.

AOK invests at a 30/70 AA, and in PortfolioVisualizer, you can see a Worst Year of -14.16% and a max drawdown of -17.55% However AOK doesn't strictly meet your definition, as it's total bond and not US treasuries, and the 30% equities is global equities, not the S&P 500.

Using strictly your definition, we can look up VFINX(S&P500) and VGIT(Intermediate Treasuries) and get Worst Year of -12.84% and Max Drawdown of -15.20%

I and many others used to tell people a 30/70 AA could expect a max drawdown of 15% or so. Now we know it's probably more like 20%.

[Cheego]

2022 was a tough year for bonds, one of the worst ever recorded.

AOK invests at a 30/70 AA, and in PortfolioVisualizer, you can see a Worst Year of -14.16% and a max drawdown of -17.55% However AOK doesn't strictly meet your definition, as it's total bond and not US treasuries, and the 30% equities is global equities, not the S&P 500.

Using strictly your definition, we can look up VFINX(S&P500) and VGIT(Intermediate Treasuries) and get Worst Year of -12.84% and Max Drawdown of -15.20%

I and many others used to tell people a 30/70 AA could expect a max drawdown of 15% or so. Now we know it's probably more like 20%.

Thank you, but I'm looking for the worst return over any historical 40 year span.

[zie]

2022 was a tough year for bonds, one of the worst ever recorded.

AOK invests at a 30/70 AA, and in PortfolioVisualizer, you can see a Worst Year of -14.16% and a max drawdown of -17.55% However AOK doesn't strictly meet your definition, as it's total bond and not US treasuries, and the 30% equities is global equities, not the S&P 500.

Using strictly your definition, we can look up VFINX(S&P500) and VGIT(Intermediate Treasuries) and get Worst Year of -12.84% and Max Drawdown of -15.20%

I and many others used to tell people a 30/70 AA could expect a max drawdown of 15% or so. Now we know it's probably more like 20%.

Thank you, but I'm looking for the worst return over any historical 40 year span.

Ah, I missed that. You can do it yourself with the Simba backtesting spreadsheet I imagine.

[martincmartin]

For a 30/70 stock/bond mix, using TSM and TBM from Simba's spreadsheet, the 40 year 100% SWR is 3.25%. If you retire Jan 1, 1937, you just run out of money after 40 years. The second worst year to retire is Jan 1, 1940, where after 40 years, you end up with $23k (in real, i.e. 1940, dollars) if you started with $1M.

Why do you want the S&P 500? It was created in 1957 as a kind of total stock market index when computers were more limited, and they could only handle 500 stocks in their index to be able to update it every hour. Computers are now more powerful than in 1957, so the S&P 500 should be considered out of date.

Why only 10 year Treasuries? Why not diversify with something like GOVT, which has all durations of Treasuries, or even better, a total bond market fund?

[martincmartin]

Using intermediate term treasuries from the Simba spreadsheet, in place of Total Bond Market, changes the 100% SWR from 3.25% (TBM) to 3.24% (ITT). Same worst & second worst starting years (1937 & 1940).

[seajay]

Testing monthly periods starting with Jan 1926-Jan 1966 and ending with Aug 1983-Aug 2023.

For 30/70
The worst result was 3.03% and came from retiring on Feb 28th, 1946. The average result 4.37%

Worst Result by Decade:
1920s: 4.41% (Only 1926-1929)
1930s: 3.14%
1940s: 3.03%
1950s: 3.47%
1960s: 3.33%
1970s: 3.77%
1980s: 6.71% (Only 1980-1983)

You picked a pretty bad allocation!!

If you did 50/50.
The worst result was 3.50% and came from retiring on Nov 30th, 1965. The average result is 5.17%

Worst Result by Decade:
1920s 4.19% (Only 1926-1929)
1930s 3.76%
1940s 3.98%
1950s 3.99%
1960s 3.50%
1970s 3.97%
1980s 8.02% (Only 1980-1983)

How did I do the math?
The Early Retirement Now website has a spreadsheet as part of their Safe Withdrawal Series that lets you calculate data like this.
I have taken that spreadsheet and over the last few years highly customized it and vastly expanded the data, so I can calculate almost any scenario pretty much instantly.
Even the original spreadsheet there would give you pretty close to what you want, if you want to run the numbers yourself.

Thanks for that https://earlyretirementnow.com/safe-wit ... te-series/ website and spreadsheet pointer. I was surprised to see it included gold monthly data as well. I wouldn't count gold as gold pre 1934 however, as money was gold/silver back then (gold standard) and many might have deposited that money (gold) in return for interest (more gold/silver). Adjusting to account for that (gold=cash interest pre 1934), and moving bond risk over to the stock side, 50/50 stock/gold



4% SWR was supported over nearly all 30 year periods since 1871 (monthly granularity).

Seen as a reserve for opportunistic purchases and gold had a historic tendency to move counter-cycle to stocks over multiple years. When stocks were down -25% so gold might have been up +33% and vice versa. If at such times you migrated gold into 'cheap' stocks at some point during the 30 years then you ended up with more shares being held than had you lumped into stocks alone at the start, and where around two-thirds of those shares were bought at a -25% discount. Both your risk of a early years bad sequence of returns risk was reduced, whilst the mid/longer term prospects were ... (very) good.

[Alpha4]

Using intermediate term treasuries from the Simba spreadsheet, in place of Total Bond Market, changes the 100% SWR from 3.25% (TBM) to 3.24% (ITT). Same worst & second worst starting years (1937 & 1940).

Isn't that because for any year before 1973 the Simba spreadsheet simply uses a mix of LT, IT, and ST Treasuries to (very crudely) simulate TBM? In other words, "TBM" in said spreadsheet before 1973 isn't really "total" bond market at all. Of course the SWR would be very close to an intermediate maturity Treasury; the "TBM" in the spreadsheet is made up of a broad range of Treasuries but no other types of bonds (unlike the real TBM).

[Alpha4]

Testing monthly periods starting with Jan 1926-Jan 1966 and ending with Aug 1983-Aug 2023.

For 30/70
The worst result was 3.03% and came from retiring on Feb 28th, 1946. The average result 4.37%

Worst Result by Decade:
1920s: 4.41% (Only 1926-1929)
1930s: 3.14%
1940s: 3.03%
1950s: 3.47%
1960s: 3.33%
1970s: 3.77%
1980s: 6.71% (Only 1980-1983)

You picked a pretty bad allocation!!

If you did 50/50.
The worst result was 3.50% and came from retiring on Nov 30th, 1965. The average result is 5.17%

Worst Result by Decade:
1920s 4.19% (Only 1926-1929)
1930s 3.76%
1940s 3.98%
1950s 3.99%
1960s 3.50%
1970s 3.97%
1980s 8.02% (Only 1980-1983)

How did I do the math?
The Early Retirement Now website has a spreadsheet as part of their Safe Withdrawal Series that lets you calculate data like this.
I have taken that spreadsheet and over the last few years highly customized it and vastly expanded the data, so I can calculate almost any scenario pretty much instantly.
Even the original spreadsheet there would give you pretty close to what you want, if you want to run the numbers yourself.

Thanks for that https://earlyretirementnow.com/safe-wit ... te-series/ website and spreadsheet pointer. I was surprised to see it included gold monthly data as well. I wouldn't count gold as gold pre 1934 however, as money was gold/silver back then (gold standard) and many might have deposited that money (gold) in return for interest (more gold/silver). Adjusting to account for that (gold=cash interest pre 1934), and moving bond risk over to the stock side, 50/50 stock/gold



4% SWR was supported over nearly all 30 year periods since 1871 (monthly granularity).

Seen as a reserve for opportunistic purchases and gold had a historic tendency to move counter-cycle to stocks over multiple years. When stocks were down -25% so gold might have been up +33% and vice versa. If at such times you migrated gold into 'cheap' stocks at some point during the 30 years then you ended up with more shares being held than had you lumped into stocks alone at the start, and where around two-thirds of those shares were bought at a -25% discount. Both your risk of a early years bad sequence of returns risk was reduced, whilst the mid/longer term prospects were ... (very) good.

One, what interest rate did you use for pre-1934? Commercial paper rates are too high before about 1922/1923 (or maybe even 1924) vs what would've been earned on an actual risk-free asset (can show proof of this if you want); before the early to mid-1920s commercial paper did carry at least some default risk/late payment risk and the rates on it reflected this to a degree. FWIW I have looked at the commercial paper rates used in the simba/Siamond bond returns simulator spreadsheet and they seem to overstate returns on what could've been achieved on a true risk free instrument (or at least the closest thing to it before Treasury Bills came along) by anywhere from maybe 55-60 basis points to 120 or 130 basis points (it varies by year).

Two, if I were you I would just simulate using silver back to 1900 (and until 1973 or 1974) in lieu of gold for that 50% of the portfolio rather than assuming half of the portfolio was in interest-bearing instruments....no doubt this will reduce the SWR vs assuming that 50% of the portfolio was invested in interest paying cash but--unless gold/silver start paying interest magically for some reason since today they obviously do not--this will provide a fairer equivalent to what gold and silver are today. How much does doing this reduce the SWR for each starting year from 1900 to 1974?

[Patzer]

Testing monthly periods starting with Jan 1926-Jan 1966 and ending with Aug 1983-Aug 2023.

For 30/70
The worst result was 3.03% and came from retiring on Feb 28th, 1946. The average result 4.37%

Worst Result by Decade:
1920s: 4.41% (Only 1926-1929)
1930s: 3.14%
1940s: 3.03%
1950s: 3.47%
1960s: 3.33%
1970s: 3.77%
1980s: 6.71% (Only 1980-1983)

You picked a pretty bad allocation!!

If you did 50/50.
The worst result was 3.50% and came from retiring on Nov 30th, 1965. The average result is 5.17%

Worst Result by Decade:
1920s 4.19% (Only 1926-1929)
1930s 3.76%
1940s 3.98%
1950s 3.99%
1960s 3.50%
1970s 3.97%
1980s 8.02% (Only 1980-1983)

How did I do the math?
The Early Retirement Now website has a spreadsheet as part of their Safe Withdrawal Series that lets you calculate data like this.
I have taken that spreadsheet and over the last few years highly customized it and vastly expanded the data, so I can calculate almost any scenario pretty much instantly.
Even the original spreadsheet there would give you pretty close to what you want, if you want to run the numbers yourself.

One, what interest rate did you use for pre-1934? Commercial paper rates are too high before about 1922/1923 (or maybe even 1924) vs what would've been earned on an actual risk-free asset (can show proof of this if you want); before the early to mid-1920s commercial paper did carry at least some default risk/late payment risk and the rates on it reflected this to a degree. FWIW I have looked at the commercial paper rates used in the simba/Siamond bond returns simulator spreadsheet and they seem to overstate returns on what could've been achieved on a true risk free instrument (or at least the closest thing to it before Treasury Bills came along) by anywhere from maybe 55-60 basis points to 120 or 130 basis points (it varies by year).

I built my spreadsheet a few years ago, and I remember that I didn't like ERN's data from that far back, so I went back and pulled old scans of New York newspapers, which if I am remembering right published an interest rate in them from the Federal Reserve Bank of New York, which I use as a proxy for the modern Fed Funds Rate or 3-Month T-Bills.

[seajay]

... snip

Thanks for that https://earlyretirementnow.com/safe-wit ... te-series/ website and spreadsheet pointer. I was surprised to see it included gold monthly data as well. I wouldn't count gold as gold pre 1934 however, as money was gold/silver back then (gold standard) and many might have deposited that money (gold) in return for interest (more gold/silver). Adjusting to account for that (gold=cash interest pre 1934), and moving bond risk over to the stock side, 50/50 stock/gold



4% SWR was supported over nearly all 30 year periods since 1871 (monthly granularity).

Seen as a reserve for opportunistic purchases and gold had a historic tendency to move counter-cycle to stocks over multiple years. When stocks were down -25% so gold might have been up +33% and vice versa. If at such times you migrated gold into 'cheap' stocks at some point during the 30 years then you ended up with more shares being held than had you lumped into stocks alone at the start, and where around two-thirds of those shares were bought at a -25% discount. Both your risk of a early years bad sequence of returns risk was reduced, whilst the mid/longer term prospects were ... (very) good.

One, what interest rate did you use for pre-1934? Commercial paper rates are too high before about 1922/1923 (or maybe even 1924) vs what would've been earned on an actual risk-free asset (can show proof of this if you want); before the early to mid-1920s commercial paper did carry at least some default risk/late payment risk and the rates on it reflected this to a degree. FWIW I have looked at the commercial paper rates used in the simba/Siamond bond returns simulator spreadsheet and they seem to overstate returns on what could've been achieved on a true risk free instrument (or at least the closest thing to it before Treasury Bills came along) by anywhere from maybe 55-60 basis points to 120 or 130 basis points (it varies by year).

Two, if I were you I would just simulate using silver back to 1900 (and until 1973 or 1974) in lieu of gold for that 50% of the portfolio rather than assuming half of the portfolio was in interest-bearing instruments....no doubt this will reduce the SWR vs assuming that 50% of the portfolio was invested in interest paying cash but--unless gold/silver start paying interest magically for some reason since today they obviously do not--this will provide a fairer equivalent to what gold and silver are today. How much does doing this reduce the SWR for each starting year from 1900 to 1974?

I used the earlyretirementnow data for 'cash' pre 1934

I had been using SImba spreadsheet data (yearly granularity) and on seeing the earlyretirementnow monthly granularity data ... tested using that. For yearly I do already have silver data, where I assumed stock/T-Bills pre 1934, silver from 1934 to 1975, gold from 1976, loaded into a modified Simba spreadsheet version, with the silver data sourced from the US Geological Survey, that I label as Precious Metal (PM) ...

[seajay]

As before, gold = cash pre 1934 (earlyretirementnow data), gold from 1934, thirds each stock, 'gold' and hard cash and the minimum of 30 year maximum withdrawal rates was 3.1%



In other words in the worst historic case you got back 93% of your inflation money via 30 yearly instalments (3.1% SWR x 30 years), whilst sleeping on two-thirds of your portfolio (gold and cash stuffed under a mattress, no counter-party risk).

I guess that could be considered as stock prices generally rising with inflation, along with dividends that compare to cash interest that generally offsets inflation, gold pacing inflation, hard cash losing to inflation but where that is supplemented by stock dividends dropping into the cash pile. But individually in a volatile manner, however where the combination of share price, dividends, gold smoothed down the volatility. With the added benefit that you could be 100% all in hand after a couple/few days (stocks T+2 trading time).

Simba spreadsheet, with SCV for the stock

[Cheego]

I appreciate all the replies, but what I'm looking for is not a SWR. I was hoping to get the worst historical real return for the 30/70 AA.

[seajay]

I appreciate all the replies, but what I'm looking for is not a SWR. I was hoping to get the worst historical real return for the 30/70 AA.

You stated inflation adjusted withdrawals and ending with zero balance remaining, which is the same as SWR. For each start year x% of the initial portfolio value is drawn in the first year, and in subsequent years that amount is uplifted by inflation ... where zero remained at the end of (in your case) 40 years. So a inflation adjusted withdrawal rate.

Given the information below, does anyone know how to calculate the worst real return?

- 30% Equities (S&P 500)
- 70% Fixed income (10 yr Treasuries)
- Inflation-adjusted withdrawals (linear, and with zero ending balance)
- Looking within any 40 year span since 1926

I use Simba spreadsheet as the calculator, which indicates for 40 year 30/70 TSM/ITT (total stock market/intermediate term treasury's, worse historic case of all of the historic cases within the 1926 onward period you specified)



indicates that the worst case had a 40 year 3.2% SWR for the 1937 start year (for 40 years).

Note that data has calendar year granularity, each start date was at the start of year. Within that there will tend to be even worst cases, for instance perhaps a June 1937 start date (I just randomly selected June here, I don't know whether that was actually a worst case start month).

[Billy C]

I appreciate all the replies, but what I'm looking for is not a SWR. I was hoping to get the worst historical real return for the 30/70 AA.

This is my attempt to answer your question. You may want to double check my math and methodology since I haven’t done a calculation like this before but thought it might be educational to try. And it’s for a 35 year period, not the 40 years you asked for, but it’s the best I can do.

Jeremy Siegel points out in Stocks for the Long Run that bonds had a negative real rate of return from 1946-1981. Let’s assume the real rate of return was 0% over that period.

According to the officialdata.org website, if you invested $100 in the S&P 500 at the beginning of 1946, you would have about $2,947.13 at the end of 1981, assuming you reinvested all dividends. This is a return on investment of 2,847.13%, or 9.85% per year.

This lump-sum investment beats inflation during this period for an inflation-adjusted return of about 532.22% cumulatively, or 5.26% per year.

So if your AA was 30% stocks and 70% bonds over this period your real rate of return was 1.578%.

[tibbitts]

I appreciate all the replies, but what I'm looking for is not a SWR. I was hoping to get the worst historical real return for the 30/70 AA.

Doesn't real return and a zero ending balance imply SWR? If not it seems like a fine distinction.

If the application for this isn't to determine a SWR, what's the application? While it might be marginally useful as a guideline, I'm sure nobody assumes there won't be worse returns going forward than have occurred in the past. And today you'd use some mix of global assets and not U.S.-centric measures.

[southernlucky]

Hi op. Play with the start and end dates with this online historical portfolio allocation returns calculator at www.wealthmeta.com to see nominal and real returns for various stock/bond AA. Doing a quick check at the start of each decade looks like 1940-1980 (~40 yrs) was 5.4% CAGR nominal and a low of 0.9% CAGR real for 30/70. There could be worse results found if doing it year-by-year.

https://www.wealthmeta.com/calculator/p ... enceLine=5

[GaryA505]

Hi op. Play with the start and end dates with this online historical portfolio allocation returns calculator at www.wealthmeta.com to see nominal and real returns for various stock/bond AA. Doing a quick check at the start of each decade looks like 1940-1980 (~40 yrs) was 5.4% CAGR nominal and a low of 0.9% CAGR real for 30/70. There could be worse results found if doing it year-by-year.

https://www.wealthmeta.com/calculator/p ... enceLine=5

Nice!

[Cheego]

Hi op. Play with the start and end dates with this online historical portfolio allocation returns calculator at www.wealthmeta.com to see nominal and real returns for various stock/bond AA. Doing a quick check at the start of each decade looks like 1940-1980 (~40 yrs) was 5.4% CAGR nominal and a low of 0.9% CAGR real for 30/70. There could be worse results found if doing it year-by-year.

https://www.wealthmeta.com/calculator/p ... enceLine=5

Thank you. Much appreciated.