Reality Matters

Research on Safe Withdrawal Rates

Moderator: hocus2004

Post Reply
hocus2004
Moderator
Posts: 752
Joined: Thu Jun 10, 2004 7:33 am

Reality Matters

Post by hocus2004 »

There are a good number of community members at the various FIRE/Passion Saving/Retire Early boards who, when faced with the reality of what the historical data says re SWRs, have in essence responded: "I don't like this particular reality and I don't intend to permit it to influence my thinking on SWRs."￾ It is when this happens that you see arguments made to the effect that there is no right or wrong in the determination of SWRs, that just because the SWR is a mathematical construct does not mean that SWR researchers need be bound by the laws of mathematics in putting forward their "findings"￾ as to what the historical data says.

I reject that line of thinking. The reason why I founded this board was so that community members interested in pursuing the realities of what the historical data says would have a safe place to go to for purposes of holding such discussions. That's why I don't permit word game posts at this board. The purpose of word game posts is to confuse people about the realities. The purpose of the discussions held at this board is to bring about a stronger understanding of the realities. Word games do not serve our purpose; they undermine it.

There is a thread going on today at the Early Retirement Forum

http://early-retirement.org/cgi-bin/yab ... 1087764331

that illustrates well the tangles that communities get themselves into when they make decisions to ignore the reality principle.

The question raised in the thread-starter is--Is it proper to adjust one's withdrawal rate after the beginning of a retirement in which SWR anlaysis was used to determine the initial take-out number?

The answer for those following the conventional methodology studies is generally "no."￾ There are some "puzzling"Â￾ exceptions to the general rule; the exceptions were examined in depth at the earlier "SWR of 6.21 percent"Â￾ thread, also at the Early Retirement Forum. The general rule, however, is that one should continue withdrawing the same 4 percent from the retirement starting-date portfolio amount. The conventional studies are rooted in the assumption that the SWR is an unchanging number. If the DOW were to rise to 30,000, the SWR would still be 4. If the Dow were to drop to 2,000, the SWR would still be 4. Recognition is generally not given to the effect of changes in valuation levels.

The conventional methodology approach is an unrealistic way to think about the question being posed in SWR analysis--What is safe? The reality is that the SWR changes each time valuation levels change. When valuation levels go up, the SWR goes down. When valuation levels go down, the SWR goes up.

Those following the data-dased SWR approach do not face a "puzzle"￾ as to whether to change their personal withdrawal rate (PWR) when stock prices go up or down. They of course may do so; it is a matter of personal choice and if their life circumstances have changed (if, for example, they now require a higher or lower level of safety than they required at their retirement starting-date), changes are of course appropriate and it is of course appropriate to take the SWR that applies at the new valuation level into account in deciding on what changes to make. But changes in valuation levels do not change a withdrawal amount that was 90 percent safe at the beginning of a retirement into a withdrawal amount that now provides something greater or lesser than 90 percent safety. A withdrawal rate that was truly 90 percent safe at the beginning of a retirement will remain 90 percent safe regardless of upward or downward changes in valuation levels during the retirement.

Say that you began taking a 5 percent withdrawal from a high-stock-percentage stock portfolio at a time when the historical data showed that you enjoyed 90 percent safety with that withdrawal level. Then prices go down. You now have a smaller portfolio. Do you need to change your take-out amount to continue to enjoy 90 percent safety? You do not. The drop in prices raises the SWR so that afterwards you enjoy 90 percent safety at a higher withdrawal rate, perhaps 6 percent. Since your portfolio is now smaller, you are taking more than 5 percent out each year to support your planned annual spending target, again perhaps 6 percent. Your withdrawal is still 5 percent of the retirement starting-date portfolio amount. But it is 6 percent of the reduced portfolio amount. And the SWR for high-stock-percentage portfolios has increased to 6 percent. It all works out nicely.

What if stock valuations go up in the years immediately following your retirement? May you begin taking out more each year? Not if you want to maintain the same level of safety that you started with. The number that provides you with a 90 percent level of safety for a retirement beginning at a specified valuation level does not change just because stock valuations change. The SWR associated with the higher valuation level is a lower number, of course. If you continue withdrawing the same percentage of your retirement starting-date portfolio amount, that amount will be a smaller percentage of the new portfolio amount. Again, it all works out nicely.

Things work out nicely with the Data-Based SWR Tool because the Data-Based SWR Tool is rooted in the realities of what the historical data says. The historical data says that the SWR changes with changes in valuation levels, and the tool incorporates those changes into its determinations of the SWR. The conventional methodology does not do this. That's why consideration of the conventional methodlogy findings lead to mind-numbing "puzzles"Â￾ as to whether it is reasonable to change one's withdrawal rate when valuation levels change.

There are no satisfying answers to these puzzles because the entire methodology is built on a fantasy-land assumption that changes in valuation levels have zero effect on SWRs. Buy into that one, and you are buying into belief in an alternate universe in which stocks perform in ways in which they have never performed in the one we actually live in. Ongoing puzzlement is a natual consequence of trying to make sound money decisions in this world while following methodologies designed for application only in others perhaps existing in galaxies far far away.

This community's understanding of SWRs will prevail.

Not because we are in the majority; we obviously are not. Not because we are better liked; we obviously are not. Not because we will employ dirty posting practices to obtain an edge; we obviously will not. This community's understanding will prevail because the SWR concept is a data-based construct, and this community's approach to determining SWRs is rooted in what the data says. When it comes to SWRs, data trumps opinion polls, data trumps personal loyalties, and data trumps dirty posting tactics. When it comes to SWRs, data trumps everything.

The SWR is what the historical data says it is. That's our board motto.

Reality matters. That's the guiding principle we aim to keep in mind when offering money advice to fellow community members.
modified eutectic
* Rookie
Posts: 1
Joined: Sun Jun 20, 2004 1:34 pm

Post by modified eutectic »

The reality is that the SWR changes each time valuation levels change. When valuation levels go up, the SWR goes down. When valuation levels go down, the SWR goes up.
This line of reasoning implicitly assumes that these changes in valuation will, at some point in the withdrawal period, be cancled out by an opposite change in valuation.

Since it is impossible to know when this return to "normal" valuations will occur in the future (i.e. the future is unknowable), what happens to the SWR if the valuation levels increase or decrease and remain at these new levels for the entire withdrawal period?
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

This line of reasoning implicitly assumes that these changes in valuation will, at some point in the withdrawal period, be canceled out by an opposite change in valuation.
No. That does not have to happen.

But the assumption that the future will be similar to the past implies that the relationship between Safe Withdrawal Rates and valuations is (reasonably) stable.
Since it is impossible to know when this return to "normal" valuations will occur in the future (i.e. the future is unknowable), what happens to the SWR if the valuation levels increase or decrease and remain at these new levels for the entire withdrawal period?
[I strongly object to the throw away phrase the future is unknowable. We can know quite a bit about what is likely to happen and what is unlikely to happen.]

It is not necessary for valuations to return to normal. It is only necessary for the relationship between Safe Withdrawal Rates and valuations to remain stable.

Underlying the relationship between valuations and SWRs is the effect of valuations and stock market total returns. [There are other factors as well.] What you describe is a permanent change of the price that investors are willing to pay for earnings (and dividends): a prolonged change that is considerably different from the historical range.

The historical range includes P/E10 values from 5 through 27.

Permanently higher valuations would mean permanently lower dividend yields, well below the 2.9% to 3.0% floor inside the historical range and more like todays 1.5% to 1.8%.

For the relationship between SWRs and valuations to shift greatly requires a tremendous, permanent shift in the fundamental (or intrinsic) return of stocks, substantially different from that of the past.

We conduct sensitivity studies to handle reasonable variations in the relationship between SWRs and valuations.

Have fun.

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

Post by JWR1945 »

Hocus has done an excellent job in describing the dominate first order effects of price changes.

There are subtle second order effects that also apply. There is a certain amount of memory that affects future returns and which modifies the Safe Withdrawal Rate as an actual sequence occurs. Much of this is captured by the phrase Reversion to the Mean, when properly defined.

Taken far enough out, however, by the tenth or eleventh year, a portfolio's outcome is fairly well established. Most of the time, it has either grown so much that it can handle just about anything or else it is in serious danger. As a result, the SWR at the beginning of a 30-year sequence is best determined by looking at 30-year sequences instead of looking at pairs of 10-year and 20-year sequences placed back to back.

Have fun.

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

Post by hocus2004 »

"the future is unknowable"

Say that you have a Wheel of Chance in which there are 100 landing spots. Ten landing spots are assigned to each of the numbers one through five. The remaining 50 landing spots are assigned to the number six.

You plan to spin the wheel 10 times. Can you say with a good bit of assurance that the number six will come up more frequently than any other number? You cannot. The future is to a large extent unknowable.

Now say that you plan to spin the wheel one million times. Can you now say with a good bit of assurance that the number six will come up more frequently than any other number? You can. The future has become to a large extent knowable.

Why is it that an unknowable future became knowable just because you elected to take more spins of the wheel? With more spins, the mathematical realities have a chance to assert themselves and dominate the determination of the outcome.

So it is with stocks. Stocks offer a better value proposition for your investment dollar when they are at reasonable valuations than they do when they are at extremely high valuations. But there are so many factors affecting stock prices in the short term that the short-term future is to a great extent unknowable. There are studies showing that it is difficult, if not impossible, to successfully time the market in the short term.

There are no studies showing that it is difficult if not impossible to time the market in the long term. That's because the historical data shows that it is entirely possibly to successfully time the market in the long term. In the long term, the statistical realities have time to assert themselves and dominate the outcome.

JWR1945's methodology provides you with three numbers. The Calculated Rate is the withdrawal rate that is most likely to turn out to be the highest surviving rate at the end of 30 years. The Safe Withdrawal Rate is the rate that is almost certain to survive, the one that survives in 95 percent of the possible return sequences. The High-Risk Withdrawal Rate is the rate that has only a 5 percent chance of surviving.

It is possible that through a quirk of fate a rate that is lower than the SWR will fail. And it is possible that through a quirk of fate a rate that is higher than the high-risk withdrawal rate will survive. But do you want to bet your retirement on it?

We are not telling you what is going to happen for sure. We don't know what is going to happen for sure. We are telling you what the odds are of various scenarios popping up in your retirement. We determine the odds by looking at what has happened in the past. We are assuming that stocks will perform in the future somewhat in the way that they have in the past.

It is possible that those with high-stock-percentage portfolios who take a 4 percent withdrawal for a retirement beginning at today's valuation levels will do OK. No one can see into the future. But the historical data says that counting on a 4 percent withdrawal in those sorts of circumstances is a high-risk bet. It's not safe. It's not even close.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

This line of reasoning implicitly assumes that these changes in valuation will, at some point in the withdrawal period, be cancled out by an opposite change in valuation.
Something else to consider:

William Bernstein came to a similar conclusion based upon the Gordon equation, which does not assume permanent long term reversion to the valuation mean. On page 234 of his 4 Pillars book, he uses the equation to calculate future real equity returns of 3.5%, and then applies the 2% reduction in withdrawal rates below equity return that was required in the past to keep the portfolio solvent. This leads to a SWR below 2% without any need for a permanent reduction in valuations.

IOW, stocks are only worth their future dividend income, discounted to the present. At low inital dividend rates, there will be less future dividends per unit of equity than in the past, resulting in lower total returns to support a long retirement. The studies he based his 2% reduction factor on were done on the S&P, and so may need adjusting if categories such as SCV are included.
Post Reply