Tuesday, May 19, 2009

Why Long-Term Timing Works Even Though Short-Term Timing Doesn’t

This is a guest post from Rob of A Rich Life. If you would like to be a guest writer on Money and Such, shoot me an e-mail at shadox1 at the domain name gmail.com.

I wrote an earlier guest blog entry entitled “Passive Investing Is a Strategy for Extremists” that argued that investors should change their stock allocations in response to big shifts in valuations, going with higher-than-normal stock allocations when prices are low and going with lower-than-normal stock allocations when prices are high. Shadox expressed some skepticism in a critique of that blog entry, asking: “Why make the assumption that investors will be better at predicting the long-term peaks and troughs in the market than they are able to predict short-term ones?” This guest blog entry is my response to that question (Shadox and I agree that short-term market timing does not work).

At the top of the huge bull market that ended in January 2000, stocks were priced at three times fair value. That means that for every $300 you put into stocks, you obtained $100 worth of shares in U.S. business enterprises and $200 worth of cotton-candy nothingness. When you buy something that is overvalued, you overpay for the thing purchased. You pay more than that thing is worth. There is no reason to believe that you will obtain an investment return on the amount of the overpayment any more than there is a reason to believe that you obtain a better car than those who pay $20,000 when you pay $60,000 for a car with a fair market value of $20,000.

The average long-term stock return is a bit above 6 percent real. It is realistic to expect to see a return somewhere in that neighborhood. But not on the entire $300 you invested in stocks. You obtain the 6 percent real return only on the $100 that you invested in stocks, not on the $200 invested in nothingness. So your likely long-term return on the entire $300 payment is not something in the neighborhood of 6 percent, but something in the neighborhood of 2 percent. A 6 percent return on $100 is $6. A return of $6 on a payment of $300 is a percentage return of 2 percent.

That is why I believe that investors should be changing their stock allocations in response to big shifts in valuation levels. Common sense tells us that the long-term value proposition of investing in stocks must be better at times of low and moderate prices than it is at times of insanely dangerous prices (I will explain in a follow-up guest blog entry what valuation metric I use to determine when stock prices are “insanely dangerous”). We all consider risk and return when setting our stock allocations. Since our realistic assessments of risk and return must change with big changes in valuations, our stock allocations should change with big changes in valuations as well.

That’s the case for long-term market timing. Nothing fancy. It’s plain old common sense.

The reality, however, is that this common-sense argument is a highly controversial argument today. Millions of smart people, ordinary investors and experts both, strongly believe that market timing is impossible. Can it really be that millions of smart people have become convinced of something that defies common sense?

Yes, that is precisely what I believe to be the case. To understand how this strange state of affairs came to be, you need to consider how our knowledge of how stock investing works has developed over the years.

The common goal of stock investors in the days before the popularity of Passive Investing was to buy low and sell high. That’s market timing. The reason why we have long divided the community of investors into bulls and bears is that for a long time the name of the game was to anticipate in which direction stock prices were headed.

The Passive Investing Revolution brought an end to that for millions of investors and for most investing experts. The Passive Investing advocates told us that it was a mistake to act on the intuitive belief that stocks must offer a better deal at low or moderate prices than they do at sky-high prices. They didn’t put forward this claim as a matter of personal opinion. They backed it up with the hard stuff -- academic research SHOWING (not just claiming) that market timing doesn’t work. 

There are indeed hundreds of well-executed studies showing that timing doesn’t work. It is not hard to understand why many became excited about these breakthrough findings. It is not hard to understand why many became convinced that the best way to invest is not to guess which way prices are headed but to determine the proper stock allocation and then stick with it for the long run.

It turns out that those studies were misinterpreted. I mentioned that there are hundreds of studies showing that timing doesn’t work. Do you know how many of those studies examine whether long-term timing works or not? The answer is -- not one of them. All of the studies showing that timing doesn’t work examine short-term timing; they look at whether changing one’s stock allocation in response to price changes pays off in six months or a year or perhaps two years. These studies are silent on the question of whether long-term timing works (long-term timing is changing your stock allocation in response to big price changes with the understanding that you may not see benefits for doing so for five or perhaps even ten years).

Given that the studies are silent, I believe that we should default to our common-sense take that timing MUST work (for the reasons explained at the top of the blog entry). But we don’t need to base our belief in long-term timing in common sense. There are studies that look into the question of whether long-term timing works (Robert Shiller, author of the book “Irrational Exuberance” is the lead researcher in this area). Do you know what these studies say? They say that long-term timing works. It has always worked. There are no exceptions in the historical record.

Common sense tells us that timing should work. And the research on long-term timing backs up what common sense tells us. The puzzle is not why long-term timing works; common sense explains that. The puzzle is -- why DOESN'T short-term timing work? How can it be that so many well-executed studies show that what common sense tells us should be so is in fact not so?

The puzzle is resolved by reaching an understanding of the difference in the influences on stock prices in the long term and in the short term. In the long term, stock prices are determined by the economic realities. The U.S. economy is sufficiently productive to support a long-term return of a little more than 6 percent real. So long as our economy remains roughly as productive as it has always been before, that number must continue to apply. Long-term stock returns are largely predictable. That’s why long-term timing works. When returns are predictable, timing is an effective strategy.

Timing doesn’t work in the short term because short-term prices are not predictable. Why? Because stock prices are set by humans and humans are emotional creatures. For a time, we can make stock prices whatever we want them to be. Stock prices were insanely high in January 1995. But those who shorted the market got killed. The rest of us reacted with insane emotion to those high prices, pushing prices yet higher and higher and higher for another five years. We have the power!

But not in the long term. In the long term, stock prices must reflect the economic realities or the entire market will collapse. By January 2000, prices had gone so high that all the legitimate economic gains for many years to come were already priced in to the current market price. That ensured that stock investors were going to be disappointed for many years running, eventually becoming disgruntled enough to sell their shares and send prices back to fair value (where they are today).

Short-term timing does not work. Long-term timing does. The reason why is that prices are set in the short term by investor emotion, which is unpredictable, but in the long term by the economic realities, which can to a large extent be known in advance. Our common sense from the pre-Passive Investing era did not mislead us -- price really does affect long-term returns, just as we long believed it must.

The Passive Investing finding that short-term timing does not work was a breakthrough insight. It changes the history of investing. But for investors to make constructive use of it, we must fix the great mistake that unfortunately was delivered to us in the same package as that insight.  It’s only short-term timing that doesn’t work. Long-term timing always works. Long-term timing is REQUIRED for the investor seeking a realistic chance of achieving long-term investing success.

Rob Bennett writes the “A Rich Life” blog. He has recorded over 100 podcasts explaining what investors need to understand to make the transition from the Passive Investing model of understanding how stock investing works to the new Rational Investing model. 

Editor's note: I will provide a brief critique to Rob's article later this week, but let me steal my own thunder, I think the arguments Rob makes are largely sound, as far as they go.

Enjoyed this post? Please consider subscribing to Money and Such by free RSS Feed or by email. You can also follow me on Twitter.


I need money said...

ya its true people need to think about the longterm, but realistically how many people will do that. nobody has any patience.

Rob Bennett said...

The lack of patience is indeed a problem, I Need Money. I believe that one of the big reasons why many "experts" are reluctant to advocate Valuation-Informed Indexing is that they believe that many or most investors do not possess the long-term mentality needed to make it work.

I see two solutions:

1) Show people how much sooner they will be able to retire if they elect to follow a valuation-informed approach. I have several calculators at the site that do this. When you look at the numbers and you begin envisioning a retirement five years sooner just because you were willing to follow a true long-term approach, you don't want to give it up. The short-term-focus stuff fades in significance when you see how much you can enhance your enjoyment of life by investing rationally.

2) Promote long-term rational investing widely. We all are born with a Get Rich Quick impulse. it's just part of being human. But success comes by following a true long-term approach. I believe that the role of experts should be to encourage us to focus on the long term. This is where Passive Investing has been a miserable failure. When you tell people that "timing never works" you are encouraging them to ignore price and focus on the short-term fun of participating in an out-of-control bull. If all of the experts were instead saying on a daily basis "short-term timing never works, long-term timing always works," I believe that many of us could overcome the influence of the Get Rich Quick impulse.

In the dietary area, we all like to eat too much chocolate cake. If the experts were telling us "trying to eat balanced meals never works," our weight problem would be much worse than it already is. At least when we are given good advice, we can try to do the right thing (knowing that we will not always succeed). I don't say that there will be no one who will go with too high a stock allocation at times of high prices if we allow accurate reporting of the realities. I say that the number who do so will be smaller if we let people know how big the risks are of investing passively. Most people I talk to want to do the right thing with their money even if doing the right thing does at times mean resisting temptations to ignore price.


Frank Curmudgeon said...

I don't think this post really makes the case in its title. Yes, over the short term irrationality dominates the economic truth, but the long term is made up of many shorter periods.

The logical conclusion from believing that one year is not predictable, but 20 is, is that an even more accurate prediction would be a 19 year one based on 1 year-old data. Surely we can all agree that this does not make sense.

The danger in using 20 year periods for analysis is that we have very few of them on which to draw conclusions. There are only 6 non-overlapping 20 year periods in 120 years of stock market data, and 6 is too small a sample size on which to base any sort of statistical conclusion. Using 100 overlapping 20 year periods is perfectly legitimate, but n does not equal 100 for statistical purposes, and is a lot closer to 6. (I forget the actual math. I think there is a known adjustment.)

This may seem like a minor point in the context of the larger question, but I don't think it is. Setting aside analysis based on shorter periods, e.g. one year, greatly hamstrings the sort of analysis that can be done to answer the larger question.

Rob Bennett said...

Part One of Response:

This may seem like a minor point in the context of the larger question, but I don't think it is. It's not at all a minor point, Frank. It's a very important point. It goes to whether the theory behind the model makes sense or not. if the theory does not make sense, the model falls. I struggled with this question myself for a long time. I am okay with my resolution of it today. But I think that some intelligent probing re this point among people who are more disinterested than I am able to be at this point would be a big plus.

It's of course true that the long term is made up of many short-term periods. If this model is to stand up, we need an explanation of what causes the behavior of stocks to be transformed from emotion-based in the short term to economic-realities-based in the long term. The data shows that this is so. But we cannot go just by the data (which is, as Frank points out, limited). We need an explanation of how the market would "know" that the short-term had passed and that the long-term had arrived and that it should begin operating according to different principles.

There is research showing that about two-thirds of the return that investors obtain from stocks comes from dividends and only one-third from non-dividend factors. So let's focus on how the payment of dividends shapes things over time.

Say that the broad index is paying at fair value a total return of 6 percent real, 4 percent from dividends and 2 percent from non-dividend factors.

Now say that valuations go from fair value to double fair value. This cuts the dividend rate in half. The cutting of the dividend (which results from an increase in valuations) reduces the return from stocks. That would explain why in the long term high valuations lead to low returns.

But there is a complication that applies only in the short term. In the short term, momentum has an effect of precisely the opposite of the effect of valuations. Higher prices cause people to like stocks more, which causes even higher prices.

So you have two competing factors at work. Momentum is pulling prices up. The lowering of the dividend payout percentage is pulling prices down. For a time, there's a struggle between these two factors and either one can win out. That's why prices are unpredictable in the short term. There's no way to know which of the two factors is going to dominate in the short term.

Rob Bennett said...

Part Two of Response:

However, as we move into the long term, the dividend factor gets stronger and stronger. If we go to three times fair value, the dividend is reduced to something a little above 1 percent. The downward pressure on total return applied by the falling dividend payout percentage grows stronger and stronger over time.

Eventually, that factor has to win out. It takes more and more of the momentum factor to cancel out the dividend factor as time goes on and eventually the momentum factor just cannot pull the weight any longer.

At that point, the momentum factor reverses polarity. Investors get fed up that the momentum factor is no longer keeping prices going up enough to cancel out the dividend factor and start selling. That causes momentum in the opposite direction. Combine downward momentum with the dividend factor and you've got a stock crash.

After the crash, something interesting happens. The dividend factor reverses! Now you've got the opposite of the earlier situation. You've got the dividend factor pulling prices up and the momentum factor pulling prices down. Sooner or later, the bear is broken in the opposite way from how the bull was broken.

Stock prices are naturally self-regulating. If people just understood how stock prices change over time, there would not be these extreme highs and extreme lows. The problem is when investors come to believe things about markets that are not so (specifically, that it is not necessary to change one's allocation in response to big price changes). When that idea catches on, the self-regulation process is broken for a time and prices go to crazy places.

But the self-regulation process is never entirely broken. It always reasserts itself in time. We just cannot know when. That's why the long-term is not really just a collection of short terms. The long term is always rational because it is always a result of the application of the self-regulation process. The short term COULD be that (it is that at times when investors are rational). But the short-term can also be a time of great irrationality.

How do several short periods of great irrationality end up creating a rational long-term period? Irrationality on the high side combined with irrationality on the low side can combine to create a rational long-term price. Pick any year in the historical data and go out 30 years and you will see a rational price. Go out only 5 years and you will often not. That's because 5 years is not enough time for the self-regulation process to kick in (it is not enough time for the dividend factor to overcome the momentum factor).

The dividend factor is the rational factor. The momentum factor is the emotional factor. Humans can set stock prices wherever they choose. But they cannot keep them there. Because setting them too high reduces dividends. And reduced dividends cause reduced returns. And reduced returns cause emotions to flip from positive to negative. Eventually.

That's my stab at explanation, anyway.


Rob Bennett said...

There are only 6 non-overlapping 20 year periods in 120 years of stock market data, and 6 is too small a sample size on which to base any sort of statistical conclusion. John Walter Russell (the fellow who does the statistical work that drives the calculators) addresses this question in an article at his site:


Shadox said...

I am going to strongly dispute the assumption that dividends have anything to do with returns. Saying that dividends somehow impact valuation is like saying that your bank account is somehow worth more money if you withdraw a set amount from it every month than if leave your money alone.

The total value of the assets you own (whether they are in the bank account or in your wallet) is unchanged. Call it the law of conservation of money… Same with dividends. Whether the company keeps the cash or you have the cash in your wallet, the value of your combined assets is unchanged.

I am not going to belabor the point, but I have written a long post explaining why dividends are meaningless for valuation purposes. There is a sea of academic papers that clearly demonstrate this (a link to some of which are in the post I pointed to).

If this is the basis for the long term timing theory, I think this is where our paths diverge.

Rob Bennett said...

I have written a long post explaining why dividends are meaningless for valuation purposes. If I am reading your earlier post properly, I believe that it is making the point that there is not necessarily an advantage to owning a stock that pays a high dividend (that it is just as good to receive the payout in the form of price appreciation as in the form of a dividend payment). I don't see that as being the same issue that is at play here.

I am saying that the total return is comprised partly of the dividend portion and partly of the price appreciation experienced. There is no question that higher valuations cause the dividend payout to drop. If you own one share of stock for which you pay $100 and the dividend payout is $4, that's a 4 percent payout. If the value of the share goes to $200 and the dividend payout remains $4, the percentage payout drops to 2 percent.

I am not making a claim as to whether it is better to receive dividends or price appreciation. I am saying that price increases reduce the dividend percentage payout. All other things held equal, that is a bad thing for the investor. All price increases reduce the dividend payout percentage and that has to be a bad thing.

It certainly can be that for a few years the capital appreciation experienced will be great enough to compensate for the smaller dividend payout. But the reduced dividend payout considered by itself is certainly a negative. I am saying that this negative at some price level becomes powerful enough to change investor emotions re stocks from a positive to a negative and that that emotional change in time causes stock price booms to turn into stock price crashes.


Market Timing said...

Hey this is a good article thank you for the read. I stumbled across this good market timing site myself that I have been using the past year. It's great I love market timing.

Jeff said...

I'm late to the game here but this article deserves a few comments.

First, while your main point is most likely right that long term timing can beat passive investing, you don't state your case well. There are some great points you could marshal to support your argument, but all you really come with is "Paying too much for stock leads to lower returns." Sure, you don't want to pay too much for stock. I don't. Passive investors don't either. We all agree it would be nice not to overpay. But that's very different than saying you can reliably tell when stocks are overpriced.

It seems to me the difference between long term market timing and passive investing stems from the belief whether it's possible to reliably time the market. You say it's possible, passive investors say it's not possible. Passive investors don't want to overpay, they just think there's no way to tell so why bother. That's the argument you have to refute, not just whether overpaying is bad for returns.

I agree that your main idea is very likely correct, but let me play devil's advocate.

First, you mention there's not a single academic paper out there dealing with long term market timing. You can't possibly believe this. How hard have you looked?

I'll briefly mention one counterexample. Read Geoff Considine at seekingalpha. He has written countless papers that are academic in nature. Sure he's not a finance professor, but he approaches thing in a largely academic way. He probably never uses the term "long term market timing" but that's essentially what he preaches. If some guy on seekingalpha is studying long term market timing, there are certainly "true academics" out there doing the same.

I have to play devil's advocate with another point you made. In discussing that short term market timing doesn't work (and you're very likely right that it doesn't) you mention that stock prices were "insanely high in January 1995." That was the exact criteria you mention to know when to exit stocks. Does that mean you would have exited stocks in January 1995? If so, when would you have gotten back in? By what criteria? Looking at a graph of the S&P500 since January 1995, if you had waited for the price to fall below Janaury 1995 levels, you'd still be waiting. You'd still be out of the market and missed one of the greatest bull markets in history. Sure you would have missed the dot com bust and 2008-9 crisis as well, but even at its lowest points in 2002 and 2009 it was up nearly 100% from its January 1995 level.

I believe that long term timing very likely works. I read your article hoping it would "hammer it home" for me. I'm left unimpressed. I think there is some great evidence out there. That's what I was hoping to see in your article.