Who you gonna believe, me or your lying eyes?
-- attributed to a Marx Brother
In part one, I argued that there is a tension in the way we think about the extent to which fundamental modes of analysis can be applied to financial asset markets and goods markets, and that there is growing evidence that these markets are not efficient or, more precisely, not efficient in the way that efficiency has been defined thus far. The rise of behavioral finance and the notion of "irrationality" as a driving force in asset markets is only the beginning.
I will now argue that, although neither asset markets nor goods markets are "efficient," they still cannot be regarded as irrational, because their behavior is too consistent and ordered for that to be likely. Moreover, I will argue that it is becoming increasingly clear that it is time to overhaul our archaic, 18th century assumptions about how markets should work. And, this has a direct bearing on where investors should be putting their money today.
Our Conflicted Understanding of Markets
This is a somewhat unlikely argument, so I will restate some main points from part one. Specifically, I argued that:
a) There is a tension in the way that students of the market such as Robert Shiller on one side and the players on the field (investors and analysts), on the other, think about the difference between financial asset markets and goods markets.
b) For economists, asset markets are "speculative" and "irrational," difficult to understand and hard to model, while goods markets are efficient and well understood. Goods markets tend towards equilibrium; asset markets don't.
c) For investors, however, traditional asset markets (stocks and bonds) have oodles of metrics they can use to make sense of prices while goods markets (e.g., coffee, aluminum, and wheat) are chaotic, "speculative," and driven by emotion (typically, fear).
If Robert Shiller had written the script for the movie Trading Places, perhaps he would have had the action on the floor of a stock exchange rather than in a commodity pit.
All Markets Are "Irrational"
This skepticism on the part of market participants about the rationality of the goods market suggests that the theoretical model for that market that we all instinctively subscribe to ("its simple supply and demand, stupid") is flawed. And, when you go back to check on the academic literature demonstrating that price is the equilibrium between supply and demand, there is not much to be found except intellectual relics preserved from the time of Adam Smith. This supposed fundamental relationship between supply, demand, and price in commodity markets -- even well before their much maligned "financialization" -- is exceptionally hard to find in the data. The supply-demand paradigm is more theological than scientific. It is asserted, not demonstrated.
Thus, the goods market can be said to be "irrational." That is, like asset markets, goods markets don't behave how the models say they should. Therefore, the markets must be fundamentally flawed.
Who are you calling crazy?
If that is true, however, this is a theoretical catastrophe. What good is economics if it cannot accurately model markets? In what other branch of knowledge is there both this failure to model the thing it purports to explain and the option to label whole swathes of the field of study "irrational" because of that failure?
In the 18th Century, Europe's sense of a benevolent and rational natural order was shaken by the carnage of the Lisbon earthquake, almost as if the continent had never been through the Black Death or the Inquisition or the martyr of the saints it once revered. A century ago, few imagined that the following three decades could have been as horrific as they turned out to be, because the optimistic march of science, progress, and democracy was too relentless for that to be possible.
That is to say, it may not only be markets that are irrational, but it may be assumptions and models, however logically derived, that are less than rational, too. Most other branches of knowledge have had to go through great paradigm shifts in recent centuries, whereas economics seems happy to roll around in the dogmas it came up with back when men wore wigs. It is curious that economics has never had to go through this kind of basic reformatting that physics and biology have undergone in the shape of relativity and evolution, for example. One would think that after the remarkable transformation of how we conceive of the texture of the physical universe, theoreticians in all domains would have become more apt to question their models when those models failed, rather than to assume that it must be the subject matter that is "crazy."
So, economics could have gone two ways with the realization that asset markets deviated from "reason." It could have chosen to critique the rationality of the markets, or it could have chosen to critique itself.
I think I can show that although the former path was chosen, the latter would have been preferable, both in the interest of maintaining economics as an autonomous realm of knowledge and as a means of understanding markets better. Below, we will reexamine Shiller's thesis on irrationality in equities, and in the third and final part of this article, we will see how this relates to the goods market.
Where Shiller Went Wrong
In Irrational Exuberance, Shiller suggests that the inverse correlation between P/E ratios and subsequent returns proves that markets are irrational. A more granular look reveals that there is much more to the story than that.
For starters, although the Cyclically Adjusted P/E ratio (a.k.a. CAPE, P/E10 or the Shiller P/E) may be highly useful for the analysis of individual securities, I am less sure that it is the best measure for the macro-market. Using annual or even monthly earnings to calculate the stock market's P/E, on a secular basis, produces largely similar patterns to CAPE. And, although CAPE might be useful for eliminating noise over the long-term, what if the cyclical component to earnings, the noise, provides valuable information?
(Source: All charts in this article are from calculations derived from Robert Shiller's data)
In fact, Shiller's elimination of this cyclical component does tend to skew our understanding of how markets work.
The Cyclical Perspective on Stocks and Earnings
If we look at the relationship between stocks and earnings on a cyclical basis, we will find that prior to 1914, stocks and earnings moved in tandem. From a cyclical perspective, stocks shadowed earnings in a way that they have not tended to do since. Prior to 1914, the cyclical relationship between stocks and earnings was crystal clear.
Since then, and especially since the early 1970s, when the US severed the last ties to the gold standard, equities and earnings (as well as commodities, bonds, and interest rates) have taken on a new relationship. I described this in greater detail a couple weeks ago, but in brief, when stocks are in a secular bull market (such as in the 1980s and 1990s, and so far, the 2010s), momentum in stocks tends to be inversely correlated with the trend in earnings (and interest rates and commodity prices) and positively correlated with bond prices. During secular bear markets (such as in the late 1970s and 2000s), stocks tend to be positively correlated with earnings, interest rates, and commodities.
At present, there is lots of chatter about how the break in the link between stocks, earnings, and other markets proves that QE is distorting the market. That seems unlikely. The stock market (SPY, DIA), as well as the other markets, are exhibiting the exact same bullish behavior as they did throughout the 1980s and 1990s. If there is distortion, it is probably coming from institutional forces, not policy ones.
The point is, the relationship between stocks and earnings -- just from a cyclical perspective -- indicates that markets tended to behave rationally prior to 1914, but have since become irrational. But, even the current "irrationality" is curiously consistent since the closing of the gold window.
One might suspect that the establishment of a central bank whose duty it is to "guide" the economy through the manipulation of interest rates, as well as the switch from gold to the dollar over the 1914-1970 period, might have somehow impacted the mentality of the stock market. Shiller's description of this relationship between stocks and earnings under the gold standard does not mention this difference, however. Rather, Shiller implies that stocks under the original gold standard demonstrated the exact same "irrationality" as they have over the last 99 years.
The Supercyclical Perspective on Stocks and Earnings
Shiller's misinterpretation of that period stems from his insistence on using "cyclically adjusted" earnings. As we just saw, amplifying the cyclicality of stocks and earnings gives us a quite different picture. But, if we neither amplify nor diminish that cyclicality, and simply let the data "speak for itself," if you will, a different image emerges from the one in Irrational Exuberance.
For every post-1914 peak in CAPE, other measures of P/E ratios peak at roughly the same time. Even if you use the more volatile monthly earnings, the story is essentially the same: P/Es peak in the 1920s, 1960s, and 1990s, and the driver is always runaway stock markets.
That was not true under the gold standard. The primary driver of falling P/E ratios at the end of the century was not a runaway stock market, but one that slowly adjusted to falling earnings. In other words, earnings were far more volatile than prices under the gold standard, quite unlike the norm since 1914 (although, as mentioned in the previous section, that characteristic arose again during the bear markets of the 1970s and 2000s, when gold, somewhat coincidentally, was ascendant again).
So, how did Shiller come to the conclusion that gold standard valuations were just as irrational then as they are today? A statistical quirk in CAPE. If you use TTM or monthly earnings, P/E ratios peaked in the mid-1890s, not 1901. And, the peak in P/Es coincided with the bottom of the stock market, not the top. In 1900-1901, stocks did jump, along with earnings, but because Shiller is using a ten-year moving average of those earnings, it shifts the top of the P/E ratio to 1901. Using CAPE only makes sense if we assume that stock prices are more volatile than earnings, but that is not always the case. More to the point, when equities and profits are highly correlated, as they were under the gold standard, using a moving average of earnings will cause the resulting P/E to jump, even if current earnings are rising faster than the market.
Shiller's general solution to the apparent mispricing of equity markets never allows the data to breathe, and it distorts the relationship between P/Es and returns. He suggests that 1901 was a bad time to buy into the market. It wasn't. Stock prices remained steady up until the establishment of the Fed and World War I. Earnings and dividends rose. Bonds fell. By Shiller's own calculations, returns were positive up until the time of the Fed. Where else could investors have gone? Commodities rose, but this was the period before financialization. Certainly big players could have found ways to take advantage of the bull market in commodities, but the commodity market wasn't what it is today. Stocks were a good play to make, especially in the mid-1890s, but even in 1901, too.
We really have to strain to find irrationality in the stock market prior to 1914. We can only do it by using one measure of P/E ratios (Shiller's P/E10) and then by reading the clear divergence in modern prices back into the gold standard period.
To recap: under the gold standard, at a cyclical level, stocks and earnings were highly correlated, and on a secular basis, P/Es tended to be positively correlated with future returns. Once the dismantling of the gold standard began in 1914, those relationships became inconsistent and "irrational," insofar as P/Es became inversely correlated with future returns. And, to make things even stranger, the irrationality of the market today has a consistency that suggests a deeper underlying cause than emotion.
So, in the light of this history, are stock markets "efficient" or "irrational"? From a cyclical perspective, it seems we could argue that they were once efficient but they are no longer, or that they are efficient to the degree that they are unencumbered by exogenous structural forces, i.e. central banking.
Unfortunately, even if we can assign a greater degree of efficiency to the stock market under the gold standard, this only gets us so far, because under that standard, the dollar standard, and the transitional period in between, P/E ratios, by any metric, have never been stationary. Shiller's thesis about stock valuations being driven by emotion seems untenable now, but that doesn't explain why P/Es rise and fall.
In part three, the relationship between P/Es and the price of goods will be addressed, and it suggests that the models and assumptions we employ with respect to both asset and goods markets are badly misaligned with the realities of the market, as well as casting further doubt on the CAPE as a reliable macro-metric.
In the meantime, the cyclical and secular trends of the last two years remain intact: strong equities and weak commodities (GSC), especially precious metals (SLV,GLD) and emerging markets (GXC,EWZ), while bonds (UST) look lukewarm. One of the subtexts of this article, however, is that the existence of the central bank has impacted key structural factors. Historically high stock market capitalization ratios and profit margins may be both a sign of the power of the Fed in the economy and its unsustainability. Insofar as that is a structural issue, however, it is extremely difficult to judge if and when a systemic crisis will occur, although one would suspect that it would begin, as with the crisis in 2007, to show symptoms of a mundane business cycle crisis before becoming much more severe. At the moment, however, there are no signs visible of cyclical or secular clouds on the horizon. After part three and in coming weeks, I hope to discuss how investors can distinguish between cyclical, secular, and structural conditions.