US and UK Deep History: Safe Withdrawal Rates

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sat Aug 04, 2018 2:53 pm

Largely thanks to Simplegift's research (see his recent posts), I assembled multiple 1871+ data series about historical stock returns and bonds returns, in real terms (inflation-adjusted), taking the perspective of either a US investor or a UK investor (hence adjusting for local currency and exchange rates). I then used a customized version of the Simba backtesting spreadsheet to analyze those data series.

One of the goals was to take a good look at safe withdrawal rates (SWRs) over such deep history. The following charts look at 30-years long retirement periods, and show the maximum fixed (inflation-adjusted) withdrawal rate that would have avoided portfolio depletion before the end of the period. The horizontal axis shows the various starting years for such retirement periods, while the vertical axis shows the resulting SWR.

Here is the classic SWR chart, for a US-only portfolio, from the perspective of a US investor, comparing a 70% stocks / 30% bonds portfolio to a 30/70 portfolio. Bonds are Intermediate-Term. No expense ratio has been taken in account. Unsurprisingly, this shows the usual finding that roughly 4% would have worked for even the worst starting years, and that for rosier years, a higher stock exposure would have made quite a difference.

Image

Now let's look at a similar chart, for a UK-only portfolio, from the perspective of a UK investor, and comparing again 70/30 and 30/70 portfolios. Note that the historical data we have only includes long-term bonds and short-term bills, so bonds are allocated fifty/fifty each, to coarsely approximate a IT bonds portfolio. Here are the findings, displaying quite a different outcome than in the US. 4% doesn't exactly seem safe anymore. And a heavy reliance on bonds would have made the situation even more dire.

Image

User avatar
knpstr
Posts: 2093
Joined: Thu Nov 20, 2014 8:57 pm
Location: Michigan

Re: US and UK Deep History: Safe Withdrawal Rates

Post by knpstr » Sat Aug 04, 2018 3:02 pm

Great easy to see analysis showing that bond heavy portfolios are riskier than equity heavy portfolios!
Very little is needed to make a happy life; it is all within yourself, in your way of thinking. -Marcus Aurelius

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sat Aug 04, 2018 3:04 pm

Then, we might ponder if some level of International diversification would have helped (e.g. the US retiree investing in some UK securities, while getting returns expressed in USD; or the UK retiree investing in some US securities, while getting returns expressed in GBP). Let's compare two 70/30 portfolios, one purely domestic, the other one splitting the 70% equities in 35% domestic, 35% international.

Somewhat surprisingly, this would have helped the US investor quite a lot in the 70s during the oil crisis (while being somewhat useful for most of the other starting years). This appears to be primarily due to exchange rate dynamics.

Image

What about the UK investor? Well, this would have helped most of the time, except for the oil crisis where it would have made things a tad worse. Still, we're getting closer to the 4% rule, which is comforting.

Image

I wouldn't draw too much of a conclusion from this quick domestic vs. international analysis though. This experiment was limited to two countries, and the vagaries of exchange rates can be really erratic and unpredictable. A much more extensive analysis (restricted to 1970+ though) can be found here.

Feedback welcome... I'll probably write a more extensive blog article on those US/UK 'deep history' data series at some point.

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sat Aug 04, 2018 4:28 pm

Forgot to mention something. I used the CPI numbers for inflation in the UK, and not the Retail Price Index (RPI) numbers. The RPI is often used in the UK, although apparently on its way out. There is quite a lively debate out there when comparing both measures of inflation. From my limited understanding, the UK CPI appears to be more consistent with what is done in other countries, including the US, so this seemed more appropriate for this analysis.

So all nominal returns were adjusted with US CPI for the US investor perspective, and adjusted with the UK CPI numbers for the UK investor perspective. If I had used RPI, the SWR numbers for the UK investors would have been worse... :shock:

david1082b
Posts: 369
Joined: Fri Jun 09, 2017 12:35 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by david1082b » Sat Aug 04, 2018 5:02 pm

knpstr wrote:
Sat Aug 04, 2018 3:02 pm
Great easy to see analysis showing that bond heavy portfolios are riskier than equity heavy portfolios!
Since this is past data, we could say that they *were* often more risky, not *are* (since future performance is not known I believe). A few times the bond-heavy portfolio seemed to provide a safer withdrawal rate, especially around 1929-1930. A bond-heavy portfolio for Japanese in the last 30 years would have worked out better, as well as including ex-Japan stocks. Diversification into ex-US stocks is often disaparaged by Jack Bogle and some posters on Bogleheads, but the historical charts show that it was sometimes very useful for US investors. The fact that it hasn't worked in the last 25 to 30 years seems to get cherry-picked by Jack Bogle to say that ex-US stocks are "not needed", as if they could never be useful again in future just because of the recent past. Naturally no one knows the future and the debates will continue.

User avatar
knpstr
Posts: 2093
Joined: Thu Nov 20, 2014 8:57 pm
Location: Michigan

Re: US and UK Deep History: Safe Withdrawal Rates

Post by knpstr » Sat Aug 04, 2018 7:12 pm

david1082b wrote:
Sat Aug 04, 2018 5:02 pm
knpstr wrote:
Sat Aug 04, 2018 3:02 pm
Great easy to see analysis showing that bond heavy portfolios are riskier than equity heavy portfolios!
Since this is past data, we could say that they *were* often more risky, not *are* (since future performance is not known I believe). A few times the bond-heavy portfolio seemed to provide a safer withdrawal rate, especially around 1929-1930. A bond-heavy portfolio for Japanese in the last 30 years would have worked out better, as well as including ex-Japan stocks. Diversification into ex-US stocks is often disaparaged by Jack Bogle and some posters on Bogleheads, but the historical charts show that it was sometimes very useful for US investors. The fact that it hasn't worked in the last 25 to 30 years seems to get cherry-picked by Jack Bogle to say that ex-US stocks are "not needed", as if they could never be useful again in future just because of the recent past. Naturally no one knows the future and the debates will continue.
Sure, "nobody knows nothing" when it comes to the future, I get that. All arguments are made in light of "past performance" since that is all we have to go on. I thought that much was understood by all, but thanks for clarifying just in case anyone thought any of us can predict the future.

To better clarify my first post: People say consistently on this board that equities are "riskier" (without defining what they mean by risk) than bonds (even when it comes to solely US based assets) and they point to past performance when making that argument. They'll often tout bond heavy allocations due to the "safety" of such an allocation. I think that this chart shows that they should rethink that popular but wrong position.
Very little is needed to make a happy life; it is all within yourself, in your way of thinking. -Marcus Aurelius

B4Xt3r
Posts: 222
Joined: Thu Sep 29, 2016 5:56 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by B4Xt3r » Sat Aug 04, 2018 8:17 pm

If you'll entertain it, I have a quick question since you seemingly have the data.

What "safe withdrawal rate" exists at the 90% level (i.e. like 90% of the retirement periods the portfolio didn't run out) for the 70% equities equally split between 35% domestic and 35% international (and the remaining in 30% bonds)? Is that more like 5% perhaps?

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sat Aug 04, 2018 9:27 pm

knpstr wrote:
Sat Aug 04, 2018 7:12 pm
To better clarify my first post: People say consistently on this board that equities are "riskier" (without defining what they mean by risk) than bonds (even when it comes to solely US based assets) and they point to past performance when making that argument. They'll often tout bond heavy allocations due to the "safety" of such an allocation. I think that this chart shows that they should rethink that popular but wrong position.
Yup. I actually wrote a long blog article making the case that, for retirees, the SWR metric could be viewed as both a measure of risk and reward, when analyzing past returns. Don't forget that SWR is a metric computed in hindsight though. To take advantage of the rosier situations as they unfold, there are implications on which withdrawal methods one plans to use...
Last edited by siamond on Sat Aug 04, 2018 9:54 pm, edited 1 time in total.

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sat Aug 04, 2018 9:51 pm

B4Xt3r wrote:
Sat Aug 04, 2018 8:17 pm
What "safe withdrawal rate" exists at the 90% level (i.e. like 90% of the retirement periods the portfolio didn't run out) for the 70% equities equally split between 35% domestic and 35% international (and the remaining in 30% bonds)? Is that more like 5% perhaps?
Here you are. First line is the absolute worst case. Second line eliminates the 5% worst cases. Third line eliminates the 10% worst cases (i.e. what you call 'the 90% level'). Remember that no expense ratios were subtracted to the returns, so reality would have been lower than that.

Image

Be careful to not generalize too fast though. Again, it's just a single test with two specific countries.

golfCaddy
Posts: 648
Joined: Wed Jan 10, 2018 10:02 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by golfCaddy » Sat Aug 04, 2018 10:14 pm

Interesting post, what happens if you look at 40-year SWRs?

CurlyDave
Posts: 652
Joined: Thu Jul 28, 2016 11:37 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by CurlyDave » Sun Aug 05, 2018 2:44 am

What looks interesting to me is that there appears to be some periodicity in the SWRs.

For US investors, the 70/30 portfolio peaks in 1921, 1949 and 1982 . Best of all there are "magic periods" 1918-1927, 1943-1954 (with the exception of 1946), and 1978-19??, where WRs of 7% would have been safe.

And 2012, which just happens to be when I started portfolio withdrawals, may be close to the center of another one. Stay tuned for another 25 years to see if this is really the case.

Valuethinker
Posts: 35351
Joined: Fri May 11, 2007 11:07 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by Valuethinker » Sun Aug 05, 2018 7:01 am

siamond wrote:
Sat Aug 04, 2018 4:28 pm
Forgot to mention something. I used the CPI numbers for inflation in the UK, and not the Retail Price Index (RPI) numbers. The RPI is often used in the UK, although apparently on its way out. There is quite a lively debate out there when comparing both measures of inflation. From my limited understanding, the UK CPI appears to be more consistent with what is done in other countries, including the US, so this seemed more appropriate for this analysis.

So all nominal returns were adjusted with US CPI for the US investor perspective, and adjusted with the UK CPI numbers for the UK investor perspective. If I had used RPI, the SWR numbers for the UK investors would have been worse... :shock:
Your understanding of CPI and RPI is broadly correct.

RPI has been downgraded to a National Statistic from, I forget the term, but basically from an international one. By the Office of National Statistics, the official record keeper.

RPI has historically been higher, pa, than CPI, by up to 1.0% p.a. The Bank of England targets 2.0% CPI inflation as its policy objective, with a range of 1.-3.0%.

The other thing with RPI is that it is fixed. Once a final figure has been set and published, it cannot be historically revised. This gives comfort in one way, but also means it's not necessarily comparable over very long periods of time.

Valuethinker
Posts: 35351
Joined: Fri May 11, 2007 11:07 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by Valuethinker » Sun Aug 05, 2018 7:10 am

siamond wrote:
Sat Aug 04, 2018 2:53 pm
Largely thanks to Simplegift's research (see his recent posts), I assembled multiple 1871+ data series about historical stock returns and bonds returns, in real terms (inflation-adjusted), taking the perspective of either a US investor or a UK investor (hence adjusting for local currency and exchange rates). I then used a customized version of the Simba backtesting spreadsheet to analyze those data series.

One of the goals was to take a good look at safe withdrawal rates (SWRs) over such deep history. The following charts look at 30-years long retirement periods, and show the maximum fixed (inflation-adjusted) withdrawal rate that would have avoided portfolio depletion before the end of the period. The horizontal axis shows the various starting years for such retirement periods, while the vertical axis shows the resulting SWR.

Here is the classic SWR chart, for a US-only portfolio, from the perspective of a US investor, comparing a 70% stocks / 30% bonds portfolio to a 30/70 portfolio. Bonds are Intermediate-Term. No expense ratio has been taken in account. Unsurprisingly, this shows the usual finding that roughly 4% would have worked for even the worst starting years, and that for rosier years, a higher stock exposure would have made quite a difference.

Image

Now let's look at a similar chart, for a UK-only portfolio, from the perspective of a UK investor, and comparing again 70/30 and 30/70 portfolios. Note that the historical data we have only includes long-term bonds and short-term bills, so bonds are allocated fifty/fifty each, to coarsely approximate a IT bonds portfolio. Here are the findings, displaying quite a different outcome than in the US. 4% doesn't exactly seem safe anymore. And a heavy reliance on bonds would have made the situation even more dire.

Image
Enormously helpful, and thank you.

I can see where my scepticism re 4% withdrawal rates comes from.

At current UK indexed linked gilt (TIPS) yields of -1.5% (yes, minus) or 1.4% nominal it's hard to see a 4% SWR working out. And given the high levels of inflation the UK saw during the postwar years (2nd only to Italy among the G7) depleting capital is a major issue.

One thing to understand is that the UK stock market dropped c. 80% in real terms, total return basis, in c. 1972-74. Indexed Linked Gilts did not exist yet. Inflation was over 20%. There was, effectively, *no* SWR.

The question an American investor has to ask is "are there reasons why the USA, now, might be more like the UK than it will be like the USA in the past?"

My own view is that recent high stock returns have caused recency bias. We tend to look for evidence that it will always be OK, and find that support in historic numbers. Rather than considering whether things might have changed. Confirmation Bias. Real interest rates, as measured by the US TIPS, are very low compared to historic out-turns. Thus either:

- real returns will be a lot lower going forward (a la Japan), and if returns for TIPS are lower, they will be so, in equilibrium, for other financial instruments (otherwise you borrow at the "cheap" rate to invest in the higher return asset, driving its price up, thus lowering expected returns; this is exactly what I think the stock market has done via leverage and stock buybacks, takeovers, Private Equity transactions etc.)

- real returns will return to historic levels, in which case markets have to correct pretty severely to make that likely

However I could equally be accused of Confirmation Bias the other way ;-).

Dandy
Posts: 5221
Joined: Sun Apr 25, 2010 7:42 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by Dandy » Sun Aug 05, 2018 8:00 am

So, I guess historically a 3% or slightly more withdrawal rate looks pretty safe. So that or a bit higher might be a reasonable starting point when trying to make that initial withdrawal plan. Other than as a basis for an initial withdrawal approach I don't think that results in a set it and forget it withdrawal plan. It should help you not panic if there are a few bad years of portfolio shrinkage but, for me, I would not be comfortable dogmatically following my initial withdrawal plan - at some point I would make an adjustment to it. I think most people would.

So, I think we have good research to establish an initial withdrawal approach and some good guides on what asset allocation and maybe even glide path we should use in retirement. How/when to adjust the withdrawal approach is vague -- stay the course doesn't seem to make it when your retirement nest egg is shrinking dramatically for several years and you are making withdrawals on top of that. Historical research, no matter how well done and how far back, won't seem very comforting for very long when facing my expenses, my diminished nest egg and my loss of human capital.

For investing, especially in the accumulation stage you can almost select an allocation, a rebalancing plan and go golfing. We often seem to want to do that with a 30 year potential retirement i.e. have a withdrawal plan and go golfing. I don't think most will adhere to their original plan, they will cut back on withdrawals, but not much that I have seen on when/how to adjust that initial withdrawal plan.

User avatar
knpstr
Posts: 2093
Joined: Thu Nov 20, 2014 8:57 pm
Location: Michigan

Re: US and UK Deep History: Safe Withdrawal Rates

Post by knpstr » Sun Aug 05, 2018 8:12 am

Dandy wrote:
Sun Aug 05, 2018 8:00 am
So, I guess historically a 3% or slightly more withdrawal rate looks pretty safe. So that or a bit higher might be a reasonable starting point when trying to make that initial withdrawal plan. Other than as a basis for an initial withdrawal approach I don't think that results in a set it and forget it withdrawal plan. It should help you not panic if there are a few bad years of portfolio shrinkage but, for me, I would not be comfortable dogmatically following my initial withdrawal plan - at some point I would make an adjustment to it. I think most people would.

So, I think we have good research to establish an initial withdrawal approach and some good guides on what asset allocation and maybe even glide path we should use in retirement. How/when to adjust the withdrawal approach is vague -- stay the course doesn't seem to make it when your retirement nest egg is shrinking dramatically for several years and you are making withdrawals on top of that. Historical research, no matter how well done and how far back, won't seem very comforting for very long when facing my expenses, my diminished nest egg and my loss of human capital.

For investing, especially in the accumulation stage you can almost select an allocation, a rebalancing plan and go golfing. We often seem to want to do that with a 30 year potential retirement i.e. have a withdrawal plan and go golfing. I don't think most will adhere to their original plan, they will cut back on withdrawals, but not much that I have seen on when/how to adjust that initial withdrawal plan.
3% has been extremely safe.. remember that "most" of the time something like 6% is safe. At 3% you'd likely end up with many times more dollars at the end of your retirement. This isn't a bad thing if you want to build a legacy to pass to your heirs.

However, do not look for guaranteed safety, it won't be a fun journey. I have found from most retirees posting here (I'm still 30 years away) that withdrawing isn't as scary as it seems (which I admit seems intimidating).
Very little is needed to make a happy life; it is all within yourself, in your way of thinking. -Marcus Aurelius

Dandy
Posts: 5221
Joined: Sun Apr 25, 2010 7:42 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by Dandy » Sun Aug 05, 2018 9:14 am

I have found from most retirees posting here (I'm still 30 years away) that withdrawing isn't as scary as it seems (which I admit seems intimidating).
I was forced to retire in 2008 at age 60- really bad timing. :oops: I had lots of risk mitigation e.g. a modest pension and health insurance from prior employer, a nice termination payment, house paid for, unemployment for a month or 2 (no jobs available), and a nice size portfolio which went from likely more than enough to not enough during the financial crisis. Yeah it was intimidating facing a possible 30 year retirement under those conditions. Oh and while it was great to have the house paid for - the prospects of selling for assets or to move to a less cost of living area were almost non existent.

We had a fairly rapid, significant and prolonged recovery which kind of smooths over the fear/concern experienced. I don't believe that such a quick and dramatic recovery is always the case. I am in great shape now somewhat justifying that things will always work out but recall I had lots of advantages others didn't/won't have e.g. pension, settlement, health insurance, etc.

So my main point was don't feel once you determine a "safe" withdrawal rate -- while a good starting point it is not an approach to blindly follow all of retirement. There are no do overs -- it is up to each retiree to manage their assets so you may need to have some guidance as to when/how to adjust that "safe" withdrawal plan.

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sun Aug 05, 2018 10:25 am

golfCaddy wrote:
Sat Aug 04, 2018 10:14 pm
Interesting post, what happens if you look at 40-year SWRs?
Unless you have a specific request, maybe I'll skip posting the whole set of charts, but here is the synthesis table for 30 and 40 years retirement cycles. See more explanations here.

Image

Image

golfCaddy
Posts: 648
Joined: Wed Jan 10, 2018 10:02 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by golfCaddy » Sun Aug 05, 2018 10:39 am

Valuethinker wrote:
Sun Aug 05, 2018 7:10 am
siamond wrote:
Sat Aug 04, 2018 2:53 pm
Largely thanks to Simplegift's research (see his recent posts), I assembled multiple 1871+ data series about historical stock returns and bonds returns, in real terms (inflation-adjusted), taking the perspective of either a US investor or a UK investor (hence adjusting for local currency and exchange rates). I then used a customized version of the Simba backtesting spreadsheet to analyze those data series.

One of the goals was to take a good look at safe withdrawal rates (SWRs) over such deep history. The following charts look at 30-years long retirement periods, and show the maximum fixed (inflation-adjusted) withdrawal rate that would have avoided portfolio depletion before the end of the period. The horizontal axis shows the various starting years for such retirement periods, while the vertical axis shows the resulting SWR.

Here is the classic SWR chart, for a US-only portfolio, from the perspective of a US investor, comparing a 70% stocks / 30% bonds portfolio to a 30/70 portfolio. Bonds are Intermediate-Term. No expense ratio has been taken in account. Unsurprisingly, this shows the usual finding that roughly 4% would have worked for even the worst starting years, and that for rosier years, a higher stock exposure would have made quite a difference.

Image

Now let's look at a similar chart, for a UK-only portfolio, from the perspective of a UK investor, and comparing again 70/30 and 30/70 portfolios. Note that the historical data we have only includes long-term bonds and short-term bills, so bonds are allocated fifty/fifty each, to coarsely approximate a IT bonds portfolio. Here are the findings, displaying quite a different outcome than in the US. 4% doesn't exactly seem safe anymore. And a heavy reliance on bonds would have made the situation even more dire.

Image
Enormously helpful, and thank you.

I can see where my scepticism re 4% withdrawal rates comes from.

At current UK indexed linked gilt (TIPS) yields of -1.5% (yes, minus) or 1.4% nominal it's hard to see a 4% SWR working out. And given the high levels of inflation the UK saw during the postwar years (2nd only to Italy among the G7) depleting capital is a major issue.

One thing to understand is that the UK stock market dropped c. 80% in real terms, total return basis, in c. 1972-74. Indexed Linked Gilts did not exist yet. Inflation was over 20%. There was, effectively, *no* SWR.
No SWR? From the UK Chart, it looks like a UK retiree in 1970 would have done fine with a 4% withdrawal rate.

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sun Aug 05, 2018 10:40 am

Valuethinker wrote:
Sun Aug 05, 2018 7:10 am
One thing to understand is that the UK stock market dropped c. 80% in real terms, total return basis, in c. 1972-74. Indexed Linked Gilts did not exist yet. Inflation was over 20%. There was, effectively, *no* SWR.
Dandy wrote:
Sun Aug 05, 2018 8:00 am
So, I guess historically a 3% or slightly more withdrawal rate looks pretty safe. So that or a bit higher might be a reasonable starting point when trying to make that initial withdrawal plan.
Well, from a *perception* standpoint, without the advantage of hindsight, some past crises (great depression in the US, oil crisis in the UK) really looked like the sky was falling, and I'm sure that most people would have had the view that any SWR past results were going down the toilet while living through such major disruption.

Furthermore, this is just an analysis focusing on two specific countries. Wade Pfau ran a 'deep history' test for more developed countries, and found some SWR outcomes that were even more frightening (e.g. 2%, even 1% if I recall). In more recent history (1970+), not only Japan, but also Spain and Italy went through an incredibly rough patch with similar numbers.

Personally, this reinforces me in the belief that only a proper withdrawal plan with a good chunk of adaptiveness (variability) can do the job of helping you navigate the really tough situations while also taking advantage of rosier outcomes. We just can't gate ourselves by the (past) worst case we know of... In addition, hedging one's bets with solid international exposure seems like the prudent thing to do, forget any personal bias, just trust the power of full diversification.

User avatar
ram
Posts: 1049
Joined: Tue Jan 01, 2008 10:47 pm
Location: Midwest

Re: US and UK Deep History: Safe Withdrawal Rates

Post by ram » Sun Aug 05, 2018 11:07 am

Thanks Siamond. Excellent analysis.
Ram

User avatar
grayfox
Posts: 4929
Joined: Sat Sep 15, 2007 4:30 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by grayfox » Sun Aug 05, 2018 11:36 am

siamond wrote:
Sun Aug 05, 2018 10:25 am
golfCaddy wrote:
Sat Aug 04, 2018 10:14 pm
Interesting post, what happens if you look at 40-year SWRs?
Unless you have a specific request, maybe I'll skip posting the whole set of charts, but here is the synthesis table for 30 and 40 years retirement cycles. See more explanations here.

Image

Image
Thanks for posting this info. I, for one, would like to see the whole graph for 40 years. I'm curious about how the SWR varied over time, and am interested in the average and max, not just the minimum.

jsprag
Posts: 18
Joined: Sun Nov 26, 2017 11:25 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by jsprag » Sun Aug 05, 2018 1:01 pm

knpstr wrote:
Sat Aug 04, 2018 7:12 pm
To better clarify my first post: People say consistently on this board that equities are "riskier" (without defining what they mean by risk) than bonds (even when it comes to solely US based assets) and they point to past performance when making that argument. They'll often tout bond heavy allocations due to the "safety" of such an allocation. I think that this chart shows that they should rethink that popular but wrong position.
I tend to see the opposite - relative to most other places on the internet, people here frequently point out that bond-heavy portfolios have a high risk of losing out to equity-heavy portfolios in the long term.

All depends, as you note, on the definitions. I don't think either statement that "X is riskier than Y" is wrong, just incomplete.

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sun Aug 05, 2018 1:17 pm

grayfox wrote:
Sun Aug 05, 2018 11:36 am
Thanks for posting this info. I, for one, would like to see the whole graph for 40 years. I'm curious about how the SWR varied over time, and am interested in the average and max, not just the minimum.
Here are the two main graphs. I really have to insist though on how crucial it is for early retirees to plan for variable withdrawals, and not let the worst cases of '40yrs SWR' overly shape their thinking...

Image

Image

Note: as previously mentioned, the 1975 starting year was really weird in the UK. 1973-74 saw a 80% drop (in real terms). It was followed by a 100% gain in 1975. Hard to imagine the emotions this triggered... :shock:

Bob
Posts: 183
Joined: Mon Feb 26, 2007 5:15 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by Bob » Sun Aug 05, 2018 1:24 pm

Very interesting analysis. Many things one can take-away from this work. Thank you for sharing your work.

As noted by CurlyDave, one of the take-aways may be that SWR seems to go up and down with some periodicity. So it made me wonder if the ups and downs correlate to anything? And if they do, would that correlation be useful in predicting current year SWR?

I am no statistician but using data from the Simba Backtesting tool it appears that T-Bills may be one factor. Since 1900, the T-bill rate in a year somewhat correlates to the SWR for that year (correlation = 0.69). And charting the two side by side shows a similar path. (Sorry I can't seem to post my chart).

Thoughts?

User avatar
grayfox
Posts: 4929
Joined: Sat Sep 15, 2007 4:30 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by grayfox » Sun Aug 05, 2018 1:35 pm

Thanks for the quick response!

Image

Image

US varied from about 3 to 10
UK varied from about 2 to 11

The variation for starting withdrawal rate is so wide, you can't just set it at year 0 and follow the withdrawal formula used in the standard analysis. In practice, you have re-evaluate every year and make adjustments.

Also, I have found that spending does not simply increase with CPI-U every year, so that is totally unrealistic as well.

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Sun Aug 05, 2018 1:47 pm

Bob wrote:
Sun Aug 05, 2018 1:24 pm
As noted by CurlyDave, one of the take-aways may be that SWR seems to go up and down with some periodicity. So it made me wonder if the ups and downs correlate to anything? And if they do, would that correlation be useful in predicting current year SWR?
Some people on this forum tend to go ballistic when the word 'valuation' is mentioned, but fact is there is an area where a strong correlation was displayed in the past. High valuations => low SWR. Low valuations => high SWR. It did NOT work all the time though, e.g. the oil crisis in the 70s was NOT due to a speculative bubble... You can find a more extended analysis of this topic in this blog article.

B4Xt3r
Posts: 222
Joined: Thu Sep 29, 2016 5:56 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by B4Xt3r » Sun Aug 05, 2018 6:46 pm

siamond wrote:
Sat Aug 04, 2018 9:51 pm
B4Xt3r wrote:
Sat Aug 04, 2018 8:17 pm
What "safe withdrawal rate" exists at the 90% level (i.e. like 90% of the retirement periods the portfolio didn't run out) for the 70% equities equally split between 35% domestic and 35% international (and the remaining in 30% bonds)? Is that more like 5% perhaps?
Here you are. First line is the absolute worst case. Second line eliminates the 5% worst cases. Third line eliminates the 10% worst cases (i.e. what you call 'the 90% level'). Remember that no expense ratios were subtracted to the returns, so reality would have been lower than that.

Image

Be careful to not generalize too fast though. Again, it's just a single test with two specific countries.
Thanks.

Valuethinker
Posts: 35351
Joined: Fri May 11, 2007 11:07 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by Valuethinker » Mon Aug 06, 2018 7:35 am

siamond wrote:
Sun Aug 05, 2018 1:17 pm
grayfox wrote:
Sun Aug 05, 2018 11:36 am
Thanks for posting this info. I, for one, would like to see the whole graph for 40 years. I'm curious about how the SWR varied over time, and am interested in the average and max, not just the minimum.
Here are the two main graphs. I really have to insist though on how crucial it is for early retirees to plan for variable withdrawals, and not let the worst cases of '40yrs SWR' overly shape their thinking...

Image

Image

Note: as previously mentioned, the 1975 starting year was really weird in the UK. 1973-74 saw a 80% drop (in real terms). It was followed by a 100% gain in 1975. Hard to imagine the emotions this triggered... :shock:
People tended to focus on nominal returns (RPI inflation was c. 20%+ pa).

Thus c. 60% drop in nominal terms (dividend yield would have been something like 6%). Then, a rise - so from 100 to 60 then 120. Not sure when, exactly, you got your money back in nominal terms. In real terms, you didn't make your money back again until c. 1986.

Share ownership by individuals really was a rare thing in those days - only for the quite affluent. And no one had a Defined Contribution pension scheme-- it was all DB. I don't think DB schemes were legally inflation indexed at that time, but most raised payments to their (relatively small given the postwar generation was in mid career) number of pensioners. Self employed people, etc, typically saved via "with profits" insurance policies/ pensions (that paid periodic policy dividends), and again increases lagged inflation but I don't know by how much.

Commercial Union insurance did a fund raise (rights issue, to existing shareholders with first refusal) in that period where the indicative dividend yield on the shares (which they had to pay, having raised money from shareholders) was 12%. Good times but they didn't feel like it at the time!

Bob
Posts: 183
Joined: Mon Feb 26, 2007 5:15 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by Bob » Tue Aug 07, 2018 8:20 am

siamond wrote:
Sun Aug 05, 2018 1:47 pm
Bob wrote:
Sun Aug 05, 2018 1:24 pm
As noted by CurlyDave, one of the take-aways may be that SWR seems to go up and down with some periodicity. So it made me wonder if the ups and downs correlate to anything? And if they do, would that correlation be useful in predicting current year SWR?
Some people on this forum tend to go ballistic when the word 'valuation' is mentioned, but fact is there is an area where a strong correlation was displayed in the past. High valuations => low SWR. Low valuations => high SWR. It did NOT work all the time though, e.g. the oil crisis in the 70s was NOT due to a speculative bubble... You can find a more extended analysis of this topic in this blog article.
Thank you very much for supplying the link to your blog article "CAPE and Safe Withdrawal Rates" from earlier this year. Very useful and helpful analyses and implications. Thanks!

User avatar
grayfox
Posts: 4929
Joined: Sat Sep 15, 2007 4:30 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by grayfox » Tue Aug 07, 2018 9:46 am

siamond wrote:
Sun Aug 05, 2018 1:17 pm
grayfox wrote:
Sun Aug 05, 2018 11:36 am
Thanks for posting this info. I, for one, would like to see the whole graph for 40 years. I'm curious about how the SWR varied over time, and am interested in the average and max, not just the minimum.
Here are the two main graphs. I really have to insist though on how crucial it is for early retirees to plan for variable withdrawals, and not let the worst cases of '40yrs SWR' overly shape their thinking...

Image

Image

Note: as previously mentioned, the 1975 starting year was really weird in the UK. 1973-74 saw a 80% drop (in real terms). It was followed by a 100% gain in 1975. Hard to imagine the emotions this triggered... :shock:
Would it be too much trouble to ask for some additional statistics. e.g.

Yearly or Monthly Data?
Min US 3.2 UK 2.0 ?
Mean US about 6 UK about 5 ?
Median US about 5 UK about 4 ?
Max US 9.8 UK 11.1 ?
Standard deviation
Number of Samples - about 117 ?
Interquartile Range (25%,75%)
10%, 90% Range
5%, 95% Range

monthly data gives a better min/max (higher frequency)

Thanks.

GAAP
Posts: 549
Joined: Fri Apr 08, 2016 12:41 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by GAAP » Tue Aug 07, 2018 9:48 am

This does a great job of confirming several of my biases:
  • The value of international diversification;
  • The value of larger equity allocations in retirement;
  • The necessity to use a variable withdrawal method;
  • The necessity to base withdrawals on estimated/likely/expected future portfolio performance;
  • The absolute insanity of using a fixed inflation-indexed withdrawal rate;
  • The danger of making plans based upon past success rates.
Great job!
:happy

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Tue Aug 07, 2018 6:53 pm

grayfox wrote:
Tue Aug 07, 2018 9:46 am
Would it be too much trouble to ask for some additional statistics.
LOL, you are asking a LOT of questions... Let's see...

Yearly or Monthly Data?
=> this is based on annual data, this is all we have that far back in time.
=> also remember that for the UK, we had to emulate IT bonds with 50% bills and 50% LT gilts, so this is rather approximate.

Min US 3.2 UK 2.0 ?
=> for a 70/30 (domestic) allocation, the min SWR was 3.8% for the US and 3.1% for the UK
=> for a 30/70 (domestic) allocation, the min SWR was 3.7% for the US and 2.5% for the UK

Mean US about 6 UK about 5 ?
=> for a 70/30 (domestic) allocation, the average SWR was 6.6% for the US and 5.7% for the UK
=> for a 30/70 (domestic) allocation, the average SWR was 5.5% for the US and 4.8% for the UK

Median US about 5 UK about 4 ?
=> for a 70/30 (domestic) allocation, the median SWR was 6.3% for the US and 5.3% for the UK
=> for a 30/70 (domestic) allocation, the median SWR was 5.1% for the US and 4.4% for the UK

Max US 9.8 UK 11.1 ?
=> for a 70/30 (domestic) allocation, the max SWR was 10.7% for the US and 11.8% for the UK
=> for a 30/70 (domestic) allocation, the max SWR was 9.5% for the US and 9.2% for the UK

Standard deviation
=> for a 70/30 (domestic) allocation, the stdev SWR was 1.7% for the US and 1.9% for the UK
=> for a 30/70 (domestic) allocation, the stdev SWR was 1.5% for the US and 1.6% for the UK

Number of Samples - about 117 ?
=> 118 cycles of 30 years

Interquartile Range (25%,75%)
=> for a 70/30 (domestic) allocation, the (25%, 75%) range was (5.0%, 7.9%) for the US and (4.4%, 6.7%) for the UK
=> for a 30/70 (domestic) allocation, the (25%, 75%) range was (4.3%, 6.5%) for the US and (3.4%, 5.8%) for the UK

10%, 90% Range
=> for a 70/30 (domestic) allocation, the (10%, 90%) range was (4.6%, 9.0%) for the US and (3.6%, 8.4%) for the UK
=> for a 30/70 (domestic) allocation, the (10%, 90%) range was (4.0%, 8.0%) for the US and (3.1%, 7.2%) for the UK

5%, 95% Range
=> for a 70/30 (domestic) allocation, the (5%, 95%) range was (4.3%, 9.4%) for the US and (3.4%, 9.5%) for the UK
=> for a 30/70 (domestic) allocation, the (5%, 95%) range was (3.9%, 8.2%) for the US and (3.0%, 7.9%) for the UK

If you spot an inconsistency, let me know, this was quite a lot of copy and paste... :wink:

User avatar
grayfox
Posts: 4929
Joined: Sat Sep 15, 2007 4:30 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by grayfox » Tue Aug 07, 2018 9:20 pm

siamond wrote:
Tue Aug 07, 2018 6:53 pm
grayfox wrote:
Tue Aug 07, 2018 9:46 am
Would it be too much trouble to ask for some additional statistics.
LOL, you are asking a LOT of questions... Let's see...

Yearly or Monthly Data?
=> this is based on annual data, this is all we have that far back in time.
=> also remember that for the UK, we had to emulate IT bonds with 50% bills and 50% LT gilts, so this is rather approximate.

Min US 3.2 UK 2.0 ?
=> for a 70/30 (domestic) allocation, the min SWR was 3.8% for the US and 3.1% for the UK
=> for a 30/70 (domestic) allocation, the min SWR was 3.7% for the US and 2.5% for the UK

Mean US about 6 UK about 5 ?
=> for a 70/30 (domestic) allocation, the average SWR was 6.6% for the US and 5.7% for the UK
=> for a 30/70 (domestic) allocation, the average SWR was 5.5% for the US and 4.8% for the UK

Median US about 5 UK about 4 ?
=> for a 70/30 (domestic) allocation, the median SWR was 6.3% for the US and 5.3% for the UK
=> for a 30/70 (domestic) allocation, the median SWR was 5.1% for the US and 4.4% for the UK

Max US 9.8 UK 11.1 ?
=> for a 70/30 (domestic) allocation, the max SWR was 10.7% for the US and 11.8% for the UK
=> for a 30/70 (domestic) allocation, the max SWR was 9.5% for the US and 9.2% for the UK

Standard deviation
=> for a 70/30 (domestic) allocation, the stdev SWR was 1.7% for the US and 1.9% for the UK
=> for a 30/70 (domestic) allocation, the stdev SWR was 1.5% for the US and 1.6% for the UK

Number of Samples - about 117 ?
=> 118 cycles of 30 years

Interquartile Range (25%,75%)
=> for a 70/30 (domestic) allocation, the (25%, 75%) range was (5.0%, 7.9%) for the US and (4.4%, 6.7%) for the UK
=> for a 30/70 (domestic) allocation, the (25%, 75%) range was (4.3%, 6.5%) for the US and (3.4%, 5.8%) for the UK

10%, 90% Range
=> for a 70/30 (domestic) allocation, the (10%, 90%) range was (4.6%, 9.0%) for the US and (3.6%, 8.4%) for the UK
=> for a 30/70 (domestic) allocation, the (10%, 90%) range was (4.0%, 8.0%) for the US and (3.1%, 7.2%) for the UK

5%, 95% Range
=> for a 70/30 (domestic) allocation, the (5%, 95%) range was (4.3%, 9.4%) for the US and (3.4%, 9.5%) for the UK
=> for a 30/70 (domestic) allocation, the (5%, 95%) range was (3.9%, 8.2%) for the US and (3.0%, 7.9%) for the UK

If you spot an inconsistency, let me know, this was quite a lot of copy and paste... :wink:
Wow! Thanks. That is some interesting stats. I like like to turn them into some interesting statements:

For 70/30 allocation, half of all SWR's for U.S. investors were greater than 6.3%. The average was 6.6% And that's for a 40 year withdrawal period!
3/4's of 40-year SWR for 70/30 U.S. investors were greater than 5.0%.

Peter Lynch was not so wrong after all! (He guesstimated 7%. I don't know if he was even talking about inflation-adjusted either.) :wink:

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Tue Aug 07, 2018 10:34 pm

grayfox wrote:
Tue Aug 07, 2018 9:20 pm
For 70/30 allocation, half of all SWR's for U.S. investors were greater than 6.3%. The average was 6.6% And that's for a 40 year withdrawal period!
Hm, actually, I gave you the stats for 30 years periods...

User avatar
grayfox
Posts: 4929
Joined: Sat Sep 15, 2007 4:30 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by grayfox » Wed Aug 08, 2018 10:13 am

OK, for U.S. Investors with 70/30 asset allocation, for 30 year withdrawal period:

Average max withdrawal rate was 6.6% and half were greater than 6.3%.
75% were in the range 5.0% to 7.9%
90% were in the range 4.3% to 9.4%

They never showed these graphs and statistics in the Trinity Study and similar studies. They leave all that information out and only show the single worst number out of 118 years.

ryman554
Posts: 1062
Joined: Sun Jan 12, 2014 9:44 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by ryman554 » Thu Aug 09, 2018 10:47 am

grayfox wrote:
Wed Aug 08, 2018 10:13 am
OK, for U.S. Investors with 70/30 asset allocation, for 30 year withdrawal period:

Average max withdrawal rate was 6.6% and half were greater than 6.3%.
75% were in the range 5.0% to 7.9%
90% were in the range 4.3% to 9.4%

They never showed these graphs and statistics in the Trinity Study and similar studies. They leave all that information out and only show the single worst number out of 118 years.
But to what purpose will you use those numbers? You absolutely should not take away from these numbers that you should spend a lot more during retirement unless your crystal ball is better than mine.

To be "safe", you want the 100% (or 90% or whatever) point. The rest don't matter, so don't confound the discussion by reporting anything more than the headline figure. But now that you have all the numbers you now realize that, on average, you're going to create a lot of money for your heirs if you only spend 4%, but a tool like FIREcalc could have told you that.

I *believe* these numbers led in some way to the creation of the VPW <sic> strategy. You'll have to ask LongInvestor, but the idea that you could be spending more than the "4% minimum" when times are good is not a new one.

User avatar
grayfox
Posts: 4929
Joined: Sat Sep 15, 2007 4:30 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by grayfox » Thu Aug 09, 2018 11:00 am

ryman554 wrote:
Thu Aug 09, 2018 10:47 am
grayfox wrote:
Wed Aug 08, 2018 10:13 am
OK, for U.S. Investors with 70/30 asset allocation, for 30 year withdrawal period:

Average max withdrawal rate was 6.6% and half were greater than 6.3%.
75% were in the range 5.0% to 7.9% (mistake. actually 50% in range. 75% above 5.0%)
90% were in the range 4.3% to 9.4%

They never showed these graphs and statistics in the Trinity Study and similar studies. They leave all that information out and only show the single worst number out of 118 years.
But to what purpose will you use those numbers? You absolutely should not take away from these numbers that you should spend a lot more during retirement unless your crystal ball is better than mine.

To be "safe", you want the 100% (or 90% or whatever) point. The rest don't matter, so don't confound the discussion by reporting anything more than the headline figure. But now that you have all the numbers you now realize that, on average, you're going to create a lot of money for your heirs if you only spend 4%, but a tool like FIREcalc could have told you that.

I *believe* these numbers led in some way to the creation of the VPW <sic> strategy. You'll have to ask LongInvestor, but the idea that you could be spending more than the "4% minimum" when times are good is not a new one.
Exactly! The action I would take knowing that the average was 6.6% and 3/4 were above 5% would be to reject the Trinity-style withdrawal in favor of some other plan.

User avatar
siamond
Posts: 4042
Joined: Mon May 28, 2012 5:50 am

Re: US and UK Deep History: Safe Withdrawal Rates

Post by siamond » Thu Aug 09, 2018 4:34 pm

grayfox wrote:
Thu Aug 09, 2018 11:00 am
The action I would take knowing that the average was 6.6% and 3/4 were above 5% would be to reject the Trinity-style withdrawal in favor of some other plan.
Yup, I reached the same conclusion a few years ago, and will not look back. Actuarial methods and flexible decision rules have been around for a long time now. I am always baffled that so few retirees seem to use any of those variable withdrawal plans...

ryman554
Posts: 1062
Joined: Sun Jan 12, 2014 9:44 pm

Re: US and UK Deep History: Safe Withdrawal Rates

Post by ryman554 » Thu Aug 09, 2018 7:42 pm

siamond wrote:
Thu Aug 09, 2018 4:34 pm
grayfox wrote:
Thu Aug 09, 2018 11:00 am
The action I would take knowing that the average was 6.6% and 3/4 were above 5% would be to reject the Trinity-style withdrawal in favor of some other plan.
Yup, I reached the same conclusion a few years ago, and will not look back. Actuarial methods and flexible decision rules have been around for a long time now. I am always baffled that so few retirees seem to use any of those variable withdrawal plans...
I look at it this way: 4%(ish) tellls me when I'm done accumulating/saving money and can move into a different phase in my life. I'm financially independent forever at that point, at least if past is prologue.

How I choose to utilize the nest egg is a different question. I *want* to leave the kids a lot of stuff for when they are older. So 4% idea meets that goal, too. But I agree, its not the best strategy out there, but at least with it I'll sleep well at night.

VPW and the like really only work if you have slop in your budget. I'm not going to retire without slop, or at least my best guess as to what slop (2x needs) is. I suspect so to are you. The purists out there will tell me I'm doing a 2% WR at that point, it is hard to disagree, but i want expensive tastes for a change.

Post Reply