Out of sample SWR (non-US)

Research on Safe Withdrawal Rates

Moderator: hocus2004

adrian2
* Rookie
Posts: 7
Joined: Sat Oct 23, 2004 9:40 pm

Out of sample SWR (non-US)

Post by adrian2 »

Hello everyone,

Is any of you aware of SWR studies (similar to Trinity) for non-US investors, using their local stock and bond markets?

Just from eyeballing the Japanese data, it seems very likely that a 100% Nikkei investor, starting at the top of the market at the end of 1989, with a 3 to 4% initial withdrawal, would be in quite an uncomfortable position - in the past few years, with the index at about 10k, the initial withdrawal would be now magnified 4 times, i.e. 12 to 16% (I'm ignoring inflation, which would make matters worse, and dividends, which would make it better - both of these are quite low in Japan).

I'm bringing up the latest bear in Japan just as an example, there could be other developed markets in the past century with even lower maximum SWR.

Even if most of you view the world from a U.S. perspective, I'm sure you are aware of potential benefits in comparing historical data from other stockmarkets.

Adrian
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

I don't think markets such as Japan have stable track records as long as the US one. Investors before WWII pretty much lost everything when they lost the war, which makes their SWR very low (perhaps zero) in the period since 1871 (the US SWR study period). Other war ravaged (and hyper inflation ravaged) countries have had their markets similarly traumatized during the period.

To get a reasonable SWR for the period requires economic stability for the period. If our country had our factories decimated by enemy bombers, the corporations owning those factories would not have done quite as well. The complete destruction of our currency, as in Germany and several South American countries, would have eliminated the bond portion of the portfolio, and temporarily dropped the stock market to near zero. The historical SWR would not have been 4% under these circumstances. A 60/40 stock bond portfolio SWR may very well have been nearly zero. Retirees in many other countries lost everything, including their pensions, and had to start all over again.
adrian2
* Rookie
Posts: 7
Joined: Sat Oct 23, 2004 9:40 pm

Post by adrian2 »

Mike wrote:I don't think markets such as Japan have stable track records as long as the US one. Investors before WWII pretty much lost everything when they lost the war
IIRC, from Triumph of the Optimists, it was the bond market that was wiped out (definitely in the case of Germany).

Even then, excluding periods of war, US investors using the SWR of 4% should be concerned, IMO, if that 4% did not work in a major market like Japan in the post-war era. Do you really think that Japan in 1989 was a once in humanity's history kind of event? And it could well be that it wasn't only Japan among the developed markets where 4% did not work.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"Are any of you aware of SWR studies (similar to Trinity) for non-US investors, using their local stock and bond markets?"

Welcome to the community, Adrian.

I recall a link that was posted to several of the boards that was a study of SWRs in non-U.S. markets. I believe that I copied the link to my files. But I just spent an hour trying to track it down and was unsuccessful. Perhaps someone else has a better recolllection of where the links appeared and can re-link that study here for us.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

adrian2 wrote:Even if most of you view the world from a U.S. perspective, I'm sure you are aware of potential benefits in comparing historical data from other stockmarkets.
Yes. Economists do this all of the time. It allows them to separate what is typical of markets in general as opposed to what is specific to a particular market such as the NYSE in the US.

You may be aware that we use Professor Robert Shiller's data in our historical sequence calculators. Here is his website.
http://www.econ.yale.edu/~shiller/
This is the paper that Dr. Robert Shiller of Yale and Dr. John Campbell of Harvard put together for their presentation to the Federal Reserve Board of Governors. They comment directly about foreign markets and the availability of data. It seems that, prior to 1970, data are limited.
http://www.econ.yale.edu/~shiller/online/jpmalt.pdf

I have looked into data sources. I do not know of any free source of detailed data (beyond the US S&P500 data that Professor Shiller provides). Here are a couple of links to let you know what you can buy:

Visit peteyperson's thread about Current dividends in different stocks markets/indices? dated Sat Sep 06, 2003.
http://nofeeboards.com/boards/viewtopic.php?t=1359
See this post.
http://nofeeboards.com/boards/viewtopic ... 113#p16113
And this post:
http://nofeeboards.com/boards/viewtopic ... 670#p19670

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

adrian2 wrote:Just from eyeballing the Japanese data, it seems very likely that a 100% Nikkei investor, starting at the top of the market at the end of 1989, with a 3 to 4% initial withdrawal, would be in quite an uncomfortable position - in the past few years, with the index at about 10k, the initial withdrawal would be now magnified 4 times, i.e. 12 to 16% (I'm ignoring inflation, which would make matters worse, and dividends, which would make it better - both of these are quite low in Japan).
This kind of analysis can be helpful. You are looking at a worst case historical sequence to see what happens under stressful conditions. This can tell you quite a bit.

Our existing historical sequence calculators can be used with data other than that of the S&P500 index. It is a matter of making a copy of a calculator and then pasting new data into the right places. Getting the data can be difficult. Putting it into a calculator is not.

You would probably like to include prices (i.e., index values), dividends, inflation rates (in the form of index values) and an alternative investment (such as cash equivalents).

Mike has done this sort of thing already. He was interested in Small Cap Value stocks. Here are some links related to his investigations that you might find interesting.

See Mike's thread about 30% SCV 70% S&P dated Mon May 17, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2495
My thread about Small Cap Value and Commercial Paper HSWR dated Fri Jul 16, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2778
Mike's thread about SCV Switching dated Fri May 21, 2004.
http://nofeeboards.com/boards/viewtopic.php?t=2510

We have found that ten years of 30-year historical sequence data is sufficient to generate useful Historical Surviving Withdrawal Rate information as a function of Professor Shiller's percentage earnings yield (100E10/P or 100% / [P/E10] ). P/E10 uses the latest price and the average of the most recent ten years of earnings. (Because ten years of earnings are used, the values have to include adjustments for inflation.) More information is better, of course. But if you are limited to data starting in 1970 and if you are looking at 30-year sequences, not all of them will be complete. There is a fair amount of uncertainty because of limited data. But there is still quite a bit of useful information that you can extract.

Have fun.

John R.
adrian2
* Rookie
Posts: 7
Joined: Sat Oct 23, 2004 9:40 pm

Post by adrian2 »

JWR1945 wrote:Our existing historical sequence calculators can be used with data other than that of the S&P500 index. It is a matter of making a copy of a calculator and then pasting new data into the right places. Getting the data can be difficult. Putting it into a calculator is not.

You would probably like to include prices (i.e., index values), dividends, inflation rates (in the form of index values) and an alternative investment (such as cash equivalents).
Does anyone of you have Triumph of the Optimists (I don't)? It should have most, if not all, of the required data.

Frankly, I'm surprised the academics haven't done this already.

Thanks for the welcome, Hocus.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"Does anyone of you have Triumph of the Optimists (I don't)? It should have most, if not all, of the required data. "

Here is a link to the Amazon page for "Triumph of the Optimists."

http://www.amazon.com/exec/obidos/ASIN/ ... 85-0760157

I have had this one on my Wish List for some time. So what am I reading at the moment instead? Bob Dylan's "Chronicles." There are some puzzles even more unsolvable (and more important?) than the SWR puzzle, you know.

"Frankly, I'm surprised the academics haven't done this already. "

I still have not had success finding the link to the study that I recall seeing that dealt with the issue you raised above, Adrian. Perhaps it's something I only saw in a dream.

Here are two that I came up with in my search that are kinda, sorta related:

Effect of International Diversification on SWR:

http://www.fpanet.org/journal/articles/ ... -art10.cfm

Predicting Equity Returns for 37 Countries: Tweaking the Gordon Formula:

http://www.econ.duke.edu/Papers/Other/T ... eturns.pdf

Be sure to let me know if you come across any good studies on the true meaning of the song "Foot of Pride" from the outtakes of the Infidels album, Adrian.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

adrian2 wrote:Does anyone of you have Triumph of the Optimists (I don't)? It should have most, if not all, of the required data.
I went to hocus2004's Amazon link. I extracted this from one of the reviews:
Reviewer: Tim Josling (Melbourne, Australia) - See all my reviews

This is a very handsome book with lovely graphs etc. However I was after a useful summary of historical market performance.

This book was lacking in several respects:

1. The numbers behind the graphs are not provided and are not available so you cannot do any further analysis yourself. The graphs themselves are also drawn in such a way that it is hard to extract the numbers using a ruler.
Apparently, the book has the information, but not in a usable form.

Have fun.

John R.
adrian2
* Rookie
Posts: 7
Joined: Sat Oct 23, 2004 9:40 pm

Post by adrian2 »

From chapter 4 of Triumph of the Optimists, available here, and the help of a ruler on the graph of page 6 of the pdf (page 49 of the book) I found the following:

- from 1969 to 1973, cummulative real return on UK equities was essentially zero. The first two years were negative, the next two were positive. Please note that all figures include dividends.
- in the next two years, according to the graph on page 15 (page 58 of the book), the worst bear market in UK equities (1973-1974) had a cummulative real loss of 71%
- a UK investor, starting in 1969 and using a SWR of 4% (beginning of each year) would have spent 16% of her portfolio, and be left with about 84% in 1973.

- in the next two years, she would spent another 8%, plus the bear market of 71%
- at the end of the bear market, all that's left is 5% (100 - 6 x 4 - 71) of the initial portfolio, the SWR of 4% is 80% of the total left. Even though UK has the best returns of all countries studied (including US) since the bottom of their greatest bear, I don't think I have to convince anybody that a 80% withdrawal after 1974 could not have been sustained.

Another back of the envelope calculation with a 3% SWR gives 11% of the portfolio left. Post-bear, withdrawal becomes 27%. Don't think so, Jose.

With a 2% rate, 17% is left. Post bear, it's close to 12%. Maybe, just maybe, with the help of a great bull.

With a 1% rate, 23% is left. Post bear, it's about 4%. Trinity, now we're talking.

Food for thought? FYI, at the beginning of the 20th century, UK had the biggest equity and bond markets.
adrian2
* Rookie
Posts: 7
Joined: Sat Oct 23, 2004 9:40 pm

Post by adrian2 »

And since at no moment since 1969 onwards, for the next 14 years or so, the cumulative real return has been positive, my simplified calculations are erring on the optimistic side - the sequence of returns for the first 4 years is unfavourable (two negatives and two positives), then the two year big drop. The withdrawal amount of the remaining portfolio grows larger and larger.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

This is where rebalancing one's portfolio really helps.

Using the same, rough, back of the envelope calculations, but with annual rebalancing of stocks and a cash equivalent that matches inflation (something similar to commercial paper):

With 50% stocks:
1) At the beginning of 1973, 84% of the portfolio remains (100%-4*(4%))=84%. If starting from $100K, the balance would be $84K. Of this, $42K would be in stocks and $42K would be in the cash equivalent.
2) At the beginning of 1975, after losing 71% of their value, the stocks would be worth ($42K*0.29=) $12.2K. The cash equivalent would still be worth $42K. Withdrawals for two years would have added up to $8K. The net would be $12.2K+$42K-$8K=$46.2K.

With 80% stocks:
1) At the beginning of 1973, the stocks would be worth $84K*0.80=$67.2K. The cash equivalent would be worth $16.8K.
2) At the beginning of 1975, the stocks would be down to ($67.2*0.29=) $19.5K. The cash equivalent would still be worth $16.2K. Withdrawals would have totaled $8K. The net would be $19.5K+$16.8K-$8K=$28.3K.

The strong bull market after the 1973-1975 bear might have been sufficient to pull these portfolios through another 24 years. A couple of good years with 30%+ gains back to back would have helped a lot.

Rebalancing a portfolio limits its upside potential, especially when one of the components has a substantially lower long-term total return than the other (e.g., cash equivalents with 0% real returns versus stocks with 6% to 7% real returns). Rebalancing is justified when the downside risk of the growth component is substantial. This occurs during periods of high valuations. Rebalancing among assets with similar long-term total returns is usually advantageous.

Be careful, though. Most investigations about rebalancing ignore transaction costs.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

Thanks for the link, adrian2.

The UK certainly went through some bad times.

It is likely that the Historical Surviving Withdrawal Rates of the mid-1960s were even worse than those starting in 1969. A series of flat returns early on really takes its toll.

It is one thing to have a sharp drop follow a rapid rise. That is what happened in 1929. It is another thing to have years of flat returns draining a portfolio followed by a sharp drop. It is much worse.

Have fun.

John R.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"This is where rebalancing one's portfolio really helps. "

Wouldn't rebalancing mean purchasing more stocks after having suffered the sort of hit to one's portfolio described by Adrian2 in his post above? Is it even the tiniest bit realistic to assume that an investor who has experienced this sort of hit is going to respond by purchasing MORE stocks? It does not seem to be even the tinest bit realistic to me.

This is the assumption that FoolMeOnce was objecting to in the comments that I reported on in the "FMO Gets Hocomania!" thread. FoolMeOnce's point was that the conventional methodology numbers work only if this assumption proves to be true. And yet there is not a single report anywhere in the literature of a real live breathing investor following the dictates of this assumption. For all we know, the strategy advertised as "100 percent safe" has never yet once been proven successful in the real world.

People with a serious interest in coming to a realistic assessment of the risk inherent in a particular asset class should not do so by engaging in fantasy-land "What if?" games. The assumption that investors using the REHP study are going to buy more stocks when facing the sort of circumstances described in the Adrian2 post above is simply not a realistic one. In my view, this assumption is almost as absurd and almost as dangerous as the assumption that for the first time in the history of equity markets changes in valuation levels are going to have zero effect on long-term returns.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

"It is another thing to have years of flat returns draining a portfolio followed by a sharp drop. It is much worse. "

Those whose retirements began in the year 2000 need to make particular note of this observation.
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Post by hocus2004 »

Here's the full text of a post that I put to the Motley Fool board

http://boards.fool.com/Message.asp?mid= ... t=username

on April 3, 2003. The words at the beginning in quotes are the words of interest to which I was responding.

"Another interesting fact is that the Sept 16, 1929 retiree who started with a $1,000 portfolio had his portfolio value drop to $320 by July 1932. That made the 1932 annual withdrawal of $32 equal to 10% of the portfolio value at the low point. As long as the retiree continued to rebalance and didn't do something unwise like overweighting bonds), the portfolio survived to the end of the pay out period."

Thanks for sharing, intercst. I did indeed find that fact "interesting."
I'm going to take a brief screaming break now. I'll be back in a little bit.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

adrian2 wrote:
JWR1945 wrote:Our existing historical sequence calculators can be used with data other than that of the S&P500 index. It is a matter of making a copy of a calculator and then pasting new data into the right places. Getting the data can be difficult. Putting it into a calculator is not.

You would probably like to include prices (i.e., index values), dividends, inflation rates (in the form of index values) and an alternative investment (such as cash equivalents).
Does anyone of you have Triumph of the Optimists (I don't)? It should have most, if not all, of the required data.

Frankly, I'm surprised the academics haven't done this already.

Thanks for the welcome, Hocus.

It is likely that the academics (among others) have calculated means and variances and cross correlations for plugging into Monte Carlo models and for recommending allocations. I am sure that I have read several such reports that mention international markets. In fact, the objective has been to find out the benefits of including international securities in US portfolios.
I doubt that the academics have used such data with the historical sequence method. The reason is that they do not know how to do it right. Our research covers new ground. Our results are not widely known.

We are the first ones to discover the quantitative relationships among valuations, Historical Surviving Withdrawal Rates and Safe Withdrawal Rates. No one else knows how to extract useful information using historical sequences when data are available only for short time periods.

Have fun.

John R.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

hocus2004 wrote:"This is where rebalancing one's portfolio really helps. "

Wouldn't rebalancing mean purchasing more stocks after having suffered the sort of hit to one's portfolio described by Adrian2 in his post above? Is it even the tiniest bit realistic to assume that an investor who has experienced this sort of hit is going to respond by purchasing MORE stocks? It does not seem to be even the tiniest bit realistic to me.
If the stock allocation is small enough, a person might do that.

If the initial stock allocation is very small, many people would consider taking advantage of a rare opportunity. They might increase their stock allocations. Warren Buffett comes to mind.

Such circumstances would speak well of switching stock allocations gradually depending upon valuations.

If the initial allocation were large, one might question the sanity of the retiree who maintains such a large position. He would be trusting the reliability and accuracy of a study without question while suffering a serious financial setback. Is it sane to trust such a study blindly without even considering the possibility that the study has a hidden flaw? Especially since such a flaw could mean financial ruin.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

I read the links. It would seem likely that SWRs for several countries are likely to be lower than US ones, based upon their lower equity returns.
adrian2
* Rookie
Posts: 7
Joined: Sat Oct 23, 2004 9:40 pm

Post by adrian2 »

JWR1945 wrote:The UK certainly went through some bad times.

It is likely that the Historical Surviving Withdrawal Rates of the mid-1960s were even worse than those starting in 1969. A series of flat returns early on really takes its toll.

It is one thing to have a sharp drop follow a rapid rise. That is what happened in 1929. It is another thing to have years of flat returns draining a portfolio followed by a sharp drop. It is much worse.
1929 did not result in a lower SWR because there were sharp rises both before and after the bear market.

UK had a few flat years before theirs.

Japan has been having a few flat years after their bear (assuming it's over).

Next time, US might not be so lucky.
Post Reply