Valuation-Informed Indexing #90
by Rob Bennett
The premise of the Valuation-Informed Indexing model is that the P/E10 value that applies today is a good indicator of the long-term stock return that will be obtained on the purchase of an index fund made today. It’s the strength of that correlation that discredits the Efficient Market Hypothesis and the Buy-and-Hold model and that informs investors as to how much they need to lower their stock allocations in times of high valuations to keep their risk profiles steady.
Few dispute anymore that the correlation is significant. Even diehard Buy-and-Holders acknowledge that the 140 years of historical data available to us shows a statistically significant correlation. However, widespread doubts remain as to how much actionable information is conveyed by the data revealing the correlation.
There is no correlation in the short-term. So most investors have given up on the idea of timing the market. To be persuaded that they should be engaging in long-term timing (changing your stock allocation in response to big changes in valuations with the understanding that you may not see benefits for doing so for as long as ten years), Buy-and-Holders need to be persuaded that the correlation is too important and too strong to ignore.
I believe that the correlation that has been demonstrated fits the bill.
You can see the correlation between the P/E10 level that applies on the day of a purchase and the 10-year return by viewing this graphic. You can see the correlation between the P/E10 level that applies on the day of purchase and the 20-year return by viewing this graphic.
You can compare the yield predicted by P/E10 and the yield that we saw apply in real life here. The poster who created this graphic observed that: “I can’t help feeling that, if anyone looks at that chart and concludes that it contains insufficient information to affect their non-valuation-based strategy, and that there’s no reason, everything else being equal, why their equity allocation should be any different in 1929, 1968 and 2000 than it was in say 1920, 1933 or 1982, then no amount of data points will change their mind.”
You can see a graphic showing the payoff to the investor willing to adjust his allocation percentage in response to big changes in P/E10 levels here. Norbert Schlenker, the author of that graphic, commented: “The evidence is pretty incontrovertible. Valuation-Informed Indexing, i.e., market timing, is everywhere superior to Buy-and-Hold over ten-year periods.”
Associate Professor Wade Pfau shows that Valuation-Informed Indexing beat Buy-and-Hold in 102 of the 110 rolling 30-year time-periods now in the historical record with this graphic. Wade compares the nominal wealth accumulation for $1 invested in 1871 for those following a 100 percent stocks fixed-allocation strategy with the same for those following a long-term timing strategy here.
There’s a correlation.
There’s no getting around it.
You know what? Even knowing that much should alarm Buy-and-Holders.
Buy-and-Hold is a counterintuitive strategy. Price matters with everything we buy other than stocks. So the idea that it might not matter that much with stocks is an “out there” claim.
The only reason why Buy-and-Hold ever gained any traction whatsoever is that it was developed at a time when many academic believed in the Efficient Market Hypothesis. But the EMH argues that stock prices are set each day in response to economic and political developments. If the P/E10 value that applied 10 years ago is telling us today’s price, the fundamental premise of Buy-and-Hold is discredited. A showing that there is any correlation at all between valuations and long-term returns tells us that we need to rethink pretty much all of today’s conventional investing wisdom.
There’s another point that many trying to assess the significance of this correlation ignore.
The theory behind the VII model is that stock prices are random in the short-term. Today’s P/E10 value tells us nothing about what the stock price will be next year or two years from now or three years from now.
This reality greatly limits the extent of the correlation that we can realistically hope to see between P/E10 and the long-term return.
Say that the correlation were perfect. That would not support VII, it would discredit the theory behind it. Considered by itself, a strong correlation would of course be supportive evidence. But there is no way that the 10-year or 20-year correlation could be perfect in a world in which there is no short-term correlation (and a finding of a short-term correlation would be evidence against the theory advanced by Valuation-Informed Indexers).
The stock price that applies on any given day is a COMBINATION of short-term and long-term effects. If the theory behind the VII model is correct, the short-term effects should always be pulling the data in the direction of showing no correlation and the long-term effects should always be pulling the data in the direction of showing a perfect correlation. The combined effect of the two contrary forces should be data showing a strong but imperfect correlation and one that grows as the time-period lengthens.
That is precisely what we see when we look at the 140 years of data.
If the VII model properly explains how stock investing works, we should not be expecting to see a perfect correlation when we compare the P/E10 value that applies on the day of a purchase and the long-term return produced in the years following that purchase. We should be expecting to see a strong but less than perfect correlation. That’s the correlation that appears on our computer screens when we generate graphics showing to us in pictures the story that the historical stock-return data has been trying to tell us for 140 years now.