Valuation-Informed Indexing #256
by Rob Bennett
My friend Wade Pfau has written an important article on Valuation-Informed Indexing that I encourage you to read in full. The article is titled: Is High CAPE Cause for Alarm? Part Two: Valuation-Based Asset Allocation.
The article refers to me directly. It states: “We have historical evidence to suggest that mean reversion can be expected at some point, but it most certainly cannot be predicted. Since the precise timing of mean reversion is entirely random and unpredictable, expecting the result to occur in the short-term is unwise. One internet blogger has been using PE10 to predict a 65% drop in the S&P 500 by the end of 2015. This is not how the concept should be applied.”
I am the internet blogger being referred to here. There are points made in that paragraph re which I am in agreement with Wade. But I am very much NOT in agreement with him re the question of whether we should be expecting a price crash of roughly 65 percent by the end of 2016 (I once predicted that we would see the crash by the end of 2015, but that was a number of years back — I have been predicting it by the end of 2016 for a long time now). The question of whether there comes a time when relatively short-term predictions are feasible is an important one and one that needs to be much more widely discussed.
I agree 100 percent with Wade that mean reversion can be expected whenever prices get too far out of hand. I disagree with his statement that “it certainly cannot be predicted.” It is certainly true that there are limits to our ability to predict when mean reversion will take place. But it is not true that no effective predictions are possible. In fact, it is dangerous and irresponsible (in my view!) to pretend that this is the case.
Even the statement that mean reversion always takes place is a prediction. It is not at all a precise prediction. But it is a prediction all the same. The question is not: Are predictions possible? The question is: What sorts of predictions are possible?
I predict that we will see a price crash by the end of 2016. That is only 18 months out and so it is a short-term prediction. The general rule is indeed that short-term predictions do not work. The general rule is that only predictions that go ten years out work. So, looking at things on a surface level, what Wade is saying is in tune with that the research shows.
But there’s a complication.
The P/E10 level peaked in January 2000. The research showed at that time that we would see a crash within 10 years. We of course did indeed see a crash in that time-period — prices fell dramatically in late 2008. So that prediction worked.
But the bull/bear cycle that began in 1982 did not resolve with the 2008 crash. There has never been a bull/bear cycle that ended before the P/E10 level dropped to 8 or lower. This is why Robert Shiller was telling people in early 2009 to stay out of stocks until the P/E10 level dropped below 10. We never got there. The Fed pumped money into the market to block it from falling below 13 and they have continued trying to hold off the deeper price crash that we need to experience before a new cycle can begin. We still have to go to 8 or below, which is a price drop of a bit more than 65 percent from where we stand today.
It’s important to know when we will see that price crash. The answer affects all sorts of things. It obviously affects the efforts of investors to finance their retirements. It also affects the economic recovery we all want to see. Each time prices crash investors lose trillions of dollars of pretend bull-market wealth. That means they cut back on spending. That means that hundreds of thousands of businesses fail. That means that millions of workers lose their jobs. This question of whether it is ever possible to make relatively short-term predictions of where stock prices are headed is a matter of great public policy import.
I do not feel comfortable making short-term predictions. I agree wholeheartedly with Wade that as a general rule “this is not how the concept should be applied.” But I have a big problem with holding back from making the 65-percent-crash-within-two-years prediction in the circumstances that apply today.
The first crash took place in September 2008. As a society we permitted P/E10 levels to reach such insanely high levels in the late 1990s that a single crash was not enough to bring prices to the levels they need to reach for the old bull/bear cycle (which dates back to 1982) to resolve itself and for a new one to begin. So there has to be another crash. Using the ten-year rule (which is the rule supported by the 145 years of historical data available to us today), we should be expecting that second crash by the end of 2018.
So I acknowledge that the prediction that I often put forward — that we will see the next crash by the end of 2016 — may not come through. It is hard to be precise about these things. Shiller famously was “wrong” about the prediction he made to the Federal Reserve that we would see the crash we saw in 2008 by the end of 2006. I put scare quotes around the word “wrong” because it is my view that Shiller was more right than wrong in his 1996 prediction. No, he didn’t get it precisely right. But he got it a lot more right than the many “experts” who were saying that there might not be a huge price to be paid for the insane price levels we saw in the late 1990s. I would rather be roughly right than wildly and irresponsibly wrong (as was just about every expert in this field not named “Shiller” back in 1996).
There are two reasons why I predict that we will see a crash by the end of 2016 rather than by the end of 2018 (the far safer prediction).
One reason is that a lot of my critics insist that I offer claims that can be verified as proof that the Valuation-Informed Indexing concept works. I always explain that precise predictions generally cannot be made. However, I do have some sympathy for their position. To say “there will be a crash someday because prices must revert to the mean but we have no idea when that will happen” is to say just about nothing. It is not fair for critics of Valuation-Informed Indexing to demand more precision than is possible. But it is also at least a little bit unfair for Valuation-Informed Indexers to refuse to offer any precision whatsoever. I think we should always employ caveats when we make predictions. But I think we also need to try to be a bit more specific than to say “there will be another crash someday.”
Valuation levels have been headed downward for nearly 16 years. It’s been seven years since the first crash. Ten years is not the amount of time we should expect to see pass before we see a second crash; it is the longest amount of time we should expect to see pass before we see that second crash. At the end of 2016, we will have seen eight years pass. I think it is reasonable to expect to see the second crash by then. It hardly seems fair to say in 2008 that we should expect to see another crash within 10 years and then to continue to say in 2015 that we should expect to see another crash within 10 years. If the 2008 prediction was valid (it was), then it should be possible to say in 2015 that we will see a crash within three years. And, given that that is the longest it should take for the crash to take place, it does not seem too far out to me to say that there is a very strong chance that we will see the crash by the end of 2016.
I don’t offer guarantees. I certainly acknowledge that my prediction could be proven wrong. But I see strong support for that prediction in the 145 years of historical return data available to us today.
The second reason is that I think it would be irresponsible to offer no prediction at all. That’s the safe thing to do. My critics can never catch me in a wrong prediction if I never put forward a prediction.
But the stakes are high here. Millions of people are going to experience failed retirements as a result of the second crash in the event that it plays out anything at all as we should expect to see it play out based on the 145 years of peer-reviewed research available to us. Shouldn’t we at least be trying to warn people of what very much appears to be in the cards? If we cannot be 100 percent precise in our predictions (we cannot be), shouldn’t we at least try to be as precise as possible?
It seems to me that we should be as precise as possible. I do not say that I am certain that we will see a 65 percent price crash by the end of 2016. I say that that’s the best prediction that I am able to advance, given what I have been able to learn from studying the historical data for a long time and in great depth.
I’ll tell you what I would like to see. I would like to see lots of other people making the best predictions that they are able to make. Wade doesn’t think much of my prediction. That’s fair enough. But can he offer some prediction of his own? I think it would be a great learning experience for him and lots of others to try. It’s by trying to form good predictions that we learn more about “how this concept should be applied.”
Wade knows a lot. It may well be that he knows more than me. But I wish that he would put himself on the line a bit more. It’s hard for me to accept that he doesn’t think that we can know more than that prices will revert to the mean someday in the future, perhaps next year, perhaps 200 years from now. No, we cannot be entirely precise. But can we not be more precise than that? If we cannot be more precise than that, I don’t think that Valuation-Informed Indexing will ever catch on. If we cannot be more precise than that, I don’t see that it offers much value to anyone. People need (at least somewhat) actionable information.
I of course don’t mean to direct those words only to Wade Pfau. I’d like to see more predictions from Jack Bogle. I’d like to see more predictions from Robert Shiller. People in this field have a great worry about appearing foolish. But the most foolish thing of all is to let this crash come and not try hard to warn people about it. We cannot get it all right. That option is not available to us today. But we should be trying to learn as much as we can as quickly as we can. Trying to form intelligent predictions as to when that darn crash will arrive can be a great learning experience.
We don’t and can’t know it all. But we can and do know more than a lot of us are letting on. We all need to try harder to address these very important questions in public while of course always attaching the appropriate caveats to the predictions that we do advance.
Rob Bennett’s bio is here.