Looking deeply into the SWR Equation

Financial Independence/Retire Early -- Learn How!
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

It could be that some especially bad result is more likely to occur once, sometime during that decade.

My sense is that this sentence is one that we need to examine in some depth. I'll offer a few thoughts on where I am coming from on this aspect of the question.

Bernstein says that in the short-term price changes are not predictable. The author of the book "Stock Cycles" also makes this claim. I presume that they are basing this claim on analyses of the historical data. So I accept that there is considerable support in the historical data for the idea that valuation affects prices in the long term but generally not in the short term.

Let me give an example of where using valuation levels to predict short-term price changes would have failed. I believe that stock prices were above normal levels in the beginning of 1995. Five years later, they were a whole lot higher. The fact that they were high at the start did not at all give you a sound basis for thinking that they would be heading downward soon.

I believe that the tricky thing in the short term is that the stock market has a tendency to go to extremes. Intuitively, you would think that, once prices go above the norm, there should be a downward tendency. There may be, but it may also be that there is an upward pull that is great enough to counter the downward tendency. Upward moves in price logically should cause downward moves in response, but because the market is not entirely rational upward moves often cause greater and greater upward moves instead.

Until they don't.

I believe that there is data supporting what Bernstein says on this point. But I also believe that there is data supporting what you are saying, JWR1945. The data that I am thinking about is data that you yourself presented in a post from the early days of The Great Debate. You were looking at the Dory36 section of the intercst study, and I recall you saying that all(or at least most) failed retirements had high-valuation start years.

If my recollection is correct, that suggests strongly that there is at least in some circumstances a connection between high valuations and really bad return years (the sort that cause busted retirements if they pop up in the early years of a retirement).

I am beginning to think that in general returns are not predictiable in the short-term. But that at extreme levels of valuation there is some predictability, enough to justify a special concern re the survivability of retirements beginning in years of extreme high valuation.

I am beginning to think that it is the way you describe it above. We need to be looking for one particular kind of bad retuirn year, the sort of bad return year that causes a busted retirement. Those types of return years seem to be correlated with times of high valuation. If a closer examination of the data bears this out, I believe that we might be onto something significant.

We would have identified the cause of a good percentage of all busted retirements, and we would have shown by looking at data that the likelihood of this bad thing happening is far greater in times of high valuation. Does that not suggest strongly that valuation matters when trying to determine what is safe?

This point, it should be noted, is independent of the valuation issue that causes Bernstein to put the SWR at 2 percent in the year 2000. If I understand things properly, the factor you are pointing to means that retirements beginning in high valuation years are less safe even if returns in the future are not less than average. This is an additional reason to believe that high valuation start years hold greater risk.
JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

hocus
I believe that the tricky thing in the short term is that the stock market has a tendency to go to extremes. Intuitively, you would think that, once prices go above the norm, there should be a downward tendency. There may be, but it may also be that there is an upward pull that is great enough to counter the downward tendency. Upward moves in price logically should cause downward moves in response, but because the market is not entirely rational upward moves often cause greater and greater upward moves instead.

Until they don't.

That is right.

raddr's post Mean reversion found on this FIRE board dated Monday, December 02, 2002 at 1:55 pm Central Standard Time had the graphs that showed the details. Those graphs and their references are no longer visible because they were stored on the MSN boards. The discussion is still worth reading.

http://nofeeboards.com/boards/viewtopic.php?t=83

raddr had generated some autocorrelation plots. (I think that they were actually auto-regression plots. An autocorrelation plot is normalized so that it cannot exceed plus or minus one.) They consistently went from positive to negative after 3 to 6 years. Excluding the S&P 500 they went strongly negative...strong enough to pass a 99% statistical confidence threshold. The S&P 500 index had similar behavior, but it was not strong enough to pass the threshold.

What that means is this: when the autocorrelation is high and positive, the next year is similar to the current year. When the autocorrelation is close to zero, the two years behave in an independent, random manner. When the autocorrelation is negative, the two years behave in the opposite manner. The fact that the autocorrelations became strongly negative in just 3 to 6 years shows strong evidence of mean reversion. If valuations started high, they corrected strongly downwards within 3 to 6 years and vice versa.

Relying on my memory, which is not perfect on this matter, year two was usually close to zero. That is, there was a year, possibly two, that maintained momentum (either high or low valuation), followed by a random year, followed by increasingly strong corrections.

This indicates that high valuations might last for two or three years, but not longer. The bubble was an exception, what statisticians call an outlier.

I do remember that I found, by looking at Professor Shiller's S&P 500 index values (manually), that big swings in prices tend to come in pairs. That is, I set a threshold at 15% or 20% and found those years. They pretty much offset each other. But you are right. There were a limited number of exceptions. The three eras of retirement disasters were associated with unanswered downward price swings. (Didn't I see 16% threshold recently, either by Ed Easterling or John Mauldin?) The 1930s had a couple of pairs plus a couple of unanswered price swings.

The main thing to keep in mind is that we are talking about an additional failure mechanism unique to retirement portfolios. It is not in effect unless there are withdrawals. This mechanism depends specifically on the sequence of market gains and losses during the distribution phase.

One troubling thought comes to mind about historical sequences. My investigations are usually based upon dory36's FIRECalc calculator. Yet, raddr has produced credible evidence that the actual historical sequences have been lucky for retirees. I have not included those specific findings in my research. In particular, when I scale safe withdrawal rates for valuations, I depend upon the historical relationship between P/E10 and safety. I do not make any adjustment because of lucky sequences.

In terms of whether a bad event will occur, having it happen at least once in a decade is much, much more likely than having it happen in a particular year. It is similar to tossing coins (but not fair coins since you do not expect this event to happen 50% of the time). Even if the odds are high that a coin will come up heads...say 80% of the time...the odds are not that good that it will come up heads ten times in a row. The odds that only a good event will occur may be high...say 80% of the time...but the odds that only good events will happen ten years in a row are not that good.

Have fun.

John R.
hocus
Moderator
Posts: 435
Joined: Mon Dec 02, 2002 12:56 am

Post by hocus »

If valuations started high, they corrected strongly downwards within 3 to 6 years and vice versa.

I had missed that. It' surprises me to hear that it happens that quickly.

From my e-mail conversations with Ed Easterling, I know that he has a particular interest in patterns of historical returns. We need to be sure to bring this up with him.
peteyperson
**** Heavy Hitter
Posts: 525
Joined: Tue Nov 26, 2002 6:46 am

Post by peteyperson »

Excellent post, John.

I was only thinking this today in fact. I felt at current market valuation both here and in the US, there was far more potential downside than upside and so the risk vs. reward and the exposed risk in general are higher than in the past. It made a market average return after inflation and costs i.e. UK mortgage rate, even more attractive as a guaranteed return rather than a high probability of getting a lower than average return on stocks starting from this point.

Petey


JWR1945 wrote: The proper way to look at this is to evaluate the upside potential and compare it to the downside risk. We are all familiar with this concept. It you start at a high valuation, your upside potential is much worse than if you start at a much lower price. Starting from a high valuation, your downside is much worse than if you start at bargain prices. Implicit in this discussion is an idea that there is a reasonable range of prices, both high and low, that exists even in the short term.
Last edited by peteyperson on Fri Jul 18, 2003 5:30 pm, edited 1 time in total.
peteyperson
**** Heavy Hitter
Posts: 525
Joined: Tue Nov 26, 2002 6:46 am

Post by peteyperson »

John,

This is very true. Most slides/corrections are 18 months to 3 years in length. I would anticipate needing to plan for at least one instance where we get as bad as history has seen, 15 years down and back to the same balance (inflation adjusted). To have two of those over a 50 year period would be unlikely but certainly possible.

As I said to you in an earlier post, 1929-1932 was a minor blip for those people who just held during that time. The end of 80s/early 90s crash was much the same, the market was back up within a year. These are the most dangerous times when psychology plays a big part, people sell out in fear, at low prices and wait until the prices go back up (and their fear subsides) and buy back into the market. It can catch out even the most sophisticated of investors.

Petey

JWR1945 wrote: This indicates that high valuations might last for two or three years, but not longer. The bubble was an exception, what statisticians call an outlier.
peteyperson
**** Heavy Hitter
Posts: 525
Joined: Tue Nov 26, 2002 6:46 am

Post by peteyperson »

John1945 wrote: If valuations started high, they corrected strongly downwards within 3 to 6 years and vice versa.

Why is this and why does it cause busted FIRE portfolios?

Is it because people took the literal value of their portfolio unadjusted for P/E ratio valuation and then a revision to the mean halved their assets? They sit there scratching their heads wondering where it all went wrong..

See, if you've looked realistically at the intrinsic value of your assets before you FIRE, a drop in 3-6 years as you suggest, would be no problem. You might get sub-par returns as the P/E corrects but that should even out over the course of your investments. The standard deviations range over 1 year is +67% to -40%, whereas over 10 years it varies from +11.2% to +2.4% (Siegel data, quoted by Bogle).

So I'm not sure the busted portfolios aren't due to something as basic as understanding valuation and intrinsic value.

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

Post by JWR1945 »

hocus
I believe that there is data supporting what Bernstein says on this point. But I also believe that there is data supporting what you are saying, JWR1945. The data that I am thinking about is data that you yourself presented in a post from the early days of The Great Debate. You were looking at the Dory36 section of the intercst study, and I recall you saying that all (or at least most) failed retirements had high-valuation start years.

Bless you! I had forgotten all about that post. It shows the kind of damage caused by the Great Debate with its distractions and disruptions.

Back at that time, I had not identified any reliable measure for valuation. But look at those tremendous swings. It isn't 15% or 20% that we need to look out for. It is 30% at a minimum.

I mentioned in my post that I thought that another 25% or 30% drop would be unpleasant. It was.

I have place the post on its own thread.

Have fun.

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

Post by JWR1945 »

peteyperson
Why is this and why does it cause busted FIRE portfolios?

Is it because people took the literal value of their portfolio unadjusted for P/E ratio valuation and then a revision to the mean halved their assets? They sit there scratching their heads wondering where it all went wrong..

That's right. Exactly right. There have also been some confessions on these boards about people being convinced that the dot.com bubble really was a new era.

Just before the Great Depression, there were similar statements. Stocks had reached a new plateau. Prices would remain permanently higher. In fact, everybody could become a millionaire (back in 1920's dollars, much more valuable than today's). You will recognize this key element. This is the secret of success: All it takes is leverage.

As for the details of mean reversion and the correlation periods, I must defer to raddr. He was surprised that it happened so fast. Perhaps, he can find a way to put up his graphs again. That includes his ground breaking sensitivity studies as well. They are treasures.

Have fun.

John R.
peteyperson
**** Heavy Hitter
Posts: 525
Joined: Tue Nov 26, 2002 6:46 am

Post by peteyperson »

Well raddr can put up the graphs, but I won't understand them without explanation that a non-mathematician could digest!

Good to see that mis-valuation was the cause for busted FIRE portfolios at high valuations. MUPPETS! That takes care of the overvaluation and cash buffer takes care of an extended undervaluation. A cash buffer limits both the exposure to risk and how much it can damage you. Did enjoy what Bogle said on investments in his book today: Whilst increased returns from an asset class usually expose the investor to increase risks, lower costs carry no such penality (not a word for word quote).

Petey -- The Great Escape - The Cash King :wink:
JWR1945 wrote: As for the details of mean reversion and the correlation periods, I must defer to raddr. He was surprised that it happened so fast. Perhaps, he can find a way to put up his graphs again. That includes his ground breaking sensitivity studies as well. They are treasures.
User avatar
BenSolar
*** Veteran
Posts: 242
Joined: Mon Nov 25, 2002 5:46 am
Location: Western NC

Post by BenSolar »

peteyperson wrote: John,
As I said to you in an earlier post, 1929-1932 was a minor blip for those people who just held during that time. The end of 80s/early 90s crash was much the same, the market was back up within a year. These are the most dangerous times when psychology plays a big part, people sell out in fear, at low prices and wait until the prices go back up (and their fear subsides) and buy back into the market. It can catch out even the most sophisticated of investors.


1929-32 was one hell of a 'minor blip'! :shock:An 80% drop that didn't recover to the level of the 1929 peak (excluding dividends and inflation) for more than 25 years! Umm, the drop in in 1987 doesn't even compare.

For an even scarier example of a historic bear, the Japanese market continued dropping for more than 11 years. Who knows when it will see it's prior peak of 1990 again? 2020? 2030?

Regards,
"Do not spoil what you have by desiring what you have not; remember that what you now have was once among the things only hoped for." - Epicurus
wanderer
*** Veteran
Posts: 363
Joined: Tue Nov 26, 2002 9:33 am
Location: anytown, usa

Post by wanderer »

For an even scarier example of a historic bear, the Japanese market continued dropping for more than 11 years. Who knows when it will see it's prior peak of 1990 again? 2020? 2030?

perish the thought! bpp has assured me he will not let this happen. :wink:
regards,

wanderer

The field has eyes / the wood has ears / I will see / be silent and hear
bpp
** Regular
Posts: 98
Joined: Tue Nov 26, 2002 6:46 am
Location: Japan

Post by bpp »

For an even scarier example of a historic bear, the Japanese market continued dropping for more than 11 years. Who knows when it will see it's prior peak of 1990 again? 2020? 2030?


Let's see, to get the Nikkei back to its old peak would require a quadrupling from current levels. This could be done by 2020 with an 8.5% total annual return. By 2030 with 5.3%. Hmm, seems like that time range may not be all that unimaginable, I'm afraid.
perish the thought! bpp has assured me he will not let this happen. :wink:


I don't know, I'm pretty winded from the 30% push in the last couple of months, and the footing is getting slippery. You may need to root for an even weaker dollar if you hope to boost your VPACX returns much quicker than that. :oops:

Cheers,
Bpp
Post Reply