When an investment celebrity like Pimco’s Bill Gross calls for Dow 5,000, we can’t help but see it as a sign of extreme investor pessimism suggesting that we must be getting near a bottom. It reminds me of the kind of thing we heard at the top in 2000 — in reverse.
There are quantitative signs that also mirror 2000. Yesterday our “yield gap” measure of equity valuation registered the single most undervalued reading in its almost 19-year history. Yesterday’s reading is so extreme that it puts the S&P 500 precisely — to the very basis point — as undervalued now as it was overvalued at the top of the bull market in the first quarter of 2000 (measured at each month-end).
With an extreme reading like this, the equity market is probably in a bottoming period. But bottoms take time to form, and it’s extremely painful while they do. That’s why, in our Model Position long the S&P 500, we’ve been only gradually building up our allocation to equities over the last three months — even at today’s seeming extreme of undervaluation we are only allocated 55%, and have therefore borne only modest losses. And we’ve retained some of our Model Position short the NASDAQ 100 and long bonds throughout, racking up offsetting gains.
Tops take a long time to form, too — although in hindsight there always seems to be a false memory of a single moment recalled as “the top.” For example, contrasting with Gross’s call for Dow 5,000, investors we talk to all seem to remember James Glassman and Kevin Hasset’s book Dow 36,000 as a key sentiment marker of the top. But, in fact, this book came out well before the actual top — in September 1999, to be precise. It would be another six months till the actual top was reached in March 2000, and an entire year until that moment of shattered hope when the recovery rally failed in September 2000 and the bear market set in for real.
There were other even more wild-eyed predictions that came out before the top — Charles Kadlec’s Dow 100,000 came out in September 1999, and David Elias’s Dow 40,000 came out in June 1999. As downright silly as these predictions seem now, the fact is that all three books seemed correct for several months after they first came out — at least directionally.
The valuation evidence of the top also took a long time to develop. Surely there was no shortage of informal valuation evidence of topping — we can all take delight in saying “I told you so” about those IPO’s of billion-dollar companies that sold pet food on the Internet (whether we really did or not). But even that kind of evidence took several years to mature as valuations of that kind of company went from absurd to insane.
Rigorous evidence of broad-based extreme overvaluation of equities didn’t begin to show up until late 1999. At that point the S&P 500 became more overvalued than it had been just before the crash of October 1987. The most extreme overvalued reading from our yield gap model, viewed on a daily basis, was negative 2.61% — it was reached on January 18, 2000.
From where we stand now at the depths of a bear market that seems like a brilliant call — but in January 2000 it didn’t seem so brilliant. The extreme of overvaluation of January 18 did not perfectly coincide with the top of the equity market — it would be more than two months until the S&P 500 made its ultimate top on March 24, 2000. And after that it spent another six months in sideways action before today’s bear market really got underway. This highlights an important truth about any value model strategy — market valuation is not the same thing as market timing.
That said, the yield gap’s extreme undervalued reading on January 18 2000 did coincide — to the precise day — with a very important market milestone: that day was the peak in long-term Treasury yields at 6.74%. Since then, yields have never been higher, and the yield gap has never been lower.
This highlights an important truth about any value model that uses interest rates as an input: such models do not purely measure equity value — value models measure the relationship between the equity market and bond market. So an extreme valuation reading is, of necessity, a joint prediction about both stocks and bonds.
There is just no way around it –to concretize the abstract notion of “value,” the value of one must be understood in terms of the relationship of that thing to something else. In the case of the yield gap model, the relationship is between the expected return of stocks to the expected return of bonds.
The expected return of stocks is the “earnings yield” of the S&P 500 — the consensus earnings forecast divided by market cap. The expected return of bonds is simply the yield of long-term Treasuries. So when the yield gap registered its all-time most extreme reading on January 18 2000, we could have expected in the future any combination of lower stock prices, higher forecast earnings, and/or lower bond yields. As it turned out, as the charts above show, we got a mixture of all three in the year that followed the yield gap’s January 2000 extreme. [The fed funds rate is shown in the chart above purely as a matter of historical interest, because the rate hike to 6% on March 21, and the resultant inversion of the yield curve, coincided with the S&P 500's ultimate top three days later; it is not an input to the yield gap model.]
Treating a value model such as the yield gap as a purely mechanistic instrument, one should be agnostic about about how extreme misvaluations will ultimately resolved — through stock price changes, earnings forecast changes, or bond yield changes. But that’s not how we think about it. The yield gap simply signals to us that capital markets are out of equilibrium — and we then apply our macroeconomic insights to forecast the most likely resolution.
For example, on December 10, 2001, when the yield gap model showed an extreme overvaluation in the Information Technology sector, we established our Model Position short the NASDAQ 100 and long bonds. We strongly disagreed with the prevailing consensus that there would be a robust “V”-shaped recovery, and that the Fed would soon begin hiking the funds rate. We forecasted that an anemic recovery would not support a technology company earnings rebound — so we shorted the NASDAQ 100. And we forecasted that the Fed would stand pat — so we bought long-term Treasuries. We were very, very right on both sides of the trade. Since then the NASDAQ 100 has fallen 49.4%, and bonds have returned 19.4%.
When we initiated our Model Position long the S&P 500 on June 11, it was not so much on the grounds that stocks were especially cheap, but rather than they were no longer rich. We stated at the outset that we fully expected stocks to get cheaper, and that we would conservatively build up the position as they did — and that is precisely what has happened. While the yield gap (like any value model based on interest rates) is telling us not so much that stocks are undervalued absolutely, but rather that they are undervalued relative to bonds, we have not so far added a short bonds component to this position. Since establishing a Model Position long Treasuries on March 26, we have believed that defeated expectations for Fed rate hikes would drive bonds higher.
My colleague David Gitlitz now thinks that bonds may have seen there best days for a while — but while the Fed shows no signs of hiking the funds rate we are still not ready to forecast a major back-up in long-term yields. Whatever’s in store for bonds, the yield gap is now at such an extreme that it hardly matters — it would take an increase in long-term Treasury yields of 250 basis points to restore the yield gap to fair value.
If we are not forecasting a rise in bond yields, then the only ways that the yield gap can retreat from its present extreme state are for the stock market to rally, or for consensus earnings forecasts to come down. At this point the latter seems almost inevitable — the consensus call for 16.8% S&P 500 earnings growth over the next twelve months seems most unlikely considering the tepid pace of recovery, especially after the cascade of downward analyst revisions for blue chip stocks that poured out last week. That said, the yield gap’s current state of undervaluation is so extreme that — believe it or not! — the forecast for 16.8% earnings growth could turn into a forecast for a 24.6% earnings decline, and that would bring the market only to fair value.
With the stock market now valued to discount utter calamity — some combination of a 250 basis point rise in bond yields or a reversal of 41.4% in earnings growth expectations — we don’t think it’s too heroic to think that equities could find a bottom here. With any unexpected sign of real recovery, equities could do a lot more. If bond yields and earnings forecasts are held constant, then the S&P 500 could rally over 50% before it hits fair valuation.
Many of our clients don’t agree with any of that. Many believe that value models like the yield gap are “broken,” or “don’t work when interest rates are so low.” We don’t see it that way. It is in the very nature of registering an all-time extreme of valuation that the instrument of measurement would appear “broken.” All the value models seemed “broken” in the first quarter of 2000, too. We are reminded of the scene from Tom Wolfe’s The Right Stuff when Chuck Yeager first breaks the sound barrier in the Bell X-1, and kids the skeptics on the ground by reporting, “There’s something wrong with this old machometer…it’s gone kind of screwy on me…”
What the seemingly “broken” models are telling us is that market psychology is “broken.” Remember, value models are, at the core, devices for measuring the price of bearing risk (the yield gap is, quite literally, a way of observing the equity risk premium). When riskier assets are extremely undervalued, and it cannot be fully explained by pessimistic input forecasts, then it must be the case that investors have temporarily become extraordinarily risk averse. This is entirely consistent with other observations of economy-wide risk aversion that we have been making all year.
That means, if nothing else, value models such as the yield gap are now telling us one thing for sure: investors are getting paid an historic premium for daring to bear the risk of owning equities. It’s precisely the opposite of what happened in the first quarter of 2000. Then investors were not only risk tolerant, they were risk seeking — willing to pay a premium for the privilege of taking risk. That state of affairs lasted longer than anyone would have thought possible, and this one may too — but that one eventually resolved itself, and so will this one.