It is no secret that the yield curve does a fairly good job of anticipating the business cycle. With a little tweaking, I think it also does a decent job of anticipating oil spikes and stock market crashes, usually months to quarters in advance. Where the spread on 10-year and 3-month Treasuries falls to something like 1.0, a spike in oil, a stock market crash and/or a recession with all the trimmings tends to follow. In this article, however, I would like to argue that this Swiss Army Knife of market indicators also has something to say about the secular disposition of stock and commodity markets.
My interpretation of the secular phenomenon seems to be different from common usage and understanding, and so there are two primary longer-term goals I have for this article and a secondary, shorter-term goal, as well, namely;
1. To do something to convince the reader that the yield curve has something to say about the secular disposition of the stock market, whether the reader understands the secular phenomenon in the same way that I do or not.
2. To do something to demonstrate that the secular phenomenon is not merely a trend, but a different mode of market behavior that appears to hold true across all asset classes.
3. Less important in the long run, but perhaps of more immediate interest, it indicates that we are in the midst of a powerful secular bull market in equities. This does not preclude the possibility of 1987-style, cyclical corrections, but there are no clear indications, by historical standards, that such a correction is on the horizon between now and Christmas of next year.
I will do my best to meet these three goals in economic fashion while keeping the confusion to a minimum. I am going to begin with the most difficult point, number two, and I hope the readers will bear with me. If that discussion feels too abstract or dense, feel free to jump to the section "The Relationship Between Yields and the Yield Curve" below.
Does Cyclical Behavior Contain Information About the Secular Trend?
First, I believe that this change in the secular behavior of interest rates is not as distinct as in the comparisons that I have done elsewhere with respect to the surprising behavior of stocks relative to changes in earnings and interest rates. So, I do not want to argue that this is a stand-alone indicator, only that it is one more analytical tool, albeit one that I hope can do something to demonstrate this secular dimension.
When comparing the cyclical behavior of stocks to the cyclical behavior of earnings and interest rates, we will find that over the last forty years, in secular bull markets, stocks react inversely to changes in earnings and interest rates, while in bear markets stocks correlate positively to changes in earnings and interest rates. In my last article, among other things, I argued that even if one predicts future earnings perfectly, a perfectly wrong trade could ensue if the secular dimension is not observed. Therefore, the slowdown in the growth of earnings this year and the decline in yields last year have coincided with a surprisingly strong run in stocks.
(Note on sources: all interest rate, inflation, and unemployment data in this article come from St Louis Fed; all stock and earnings data from Robert Shiller; commodity price data from Grilli-Yang Commodity Price Index updated by Stephan Pfaffenzeller; producer prices from Roy Jastram's The Golden Constant.)
This relationship has been true, I argue, over the last forty years. So, one might ask, what are the chances that it will hold over the next ten, twenty, thirty, forty years? That is difficult to answer. Again, in my attempt over the last year or so to trace the changes in the markets over the last century, I think that these relationships are most likely due to a profound and growing distortion in market behavior that began with the establishment of the Fed. I think nobody can seriously doubt that there have been a series of structural changes in markets, especially since the Second World War. Take, for example, the yield curve itself. Prior to the establishment of the Federal Reserve, the yield spread was almost always negative, except in times of crisis. Its ability to reflect the status of the business cycle in the way we are accustomed to only began roughly five decades ago.
My suspicion is that the utility of the yield curve itself, in the future, will diminish, because of the deep, structural imbalances growing in the bowels of our economic system. But, because I am unable to identify the precise mechanism of these imbalances, I have to doubt both my ability to predict future systemic changes and my judgment that they are, in fact, "imbalances" and therefore "unsustainable." My solution to that problem is, therefore, to simply be aware of the possibility that systemic, structural change could come without warning but that it could also be decades away.
Many people who share this sense of systemic imbalance often seem to claim that they have special insights into the timing of a coming implosion, but I think that they too often confuse cyclical and secular changes with systemic ones. At present, although it is probably still too early to say, the crisis of five years ago was primarily secular and cyclical rather than genuinely systemic, although its causes may have been largely systemic. After all, all systems generate crises.
In other words, except for those who have deeper insights into the structure of our economic system, we are stuck with the historical relationships that we have. Perhaps the yield curve is already meaningless, but at present, it suggests to me that nothing has changed over the last five years, except the phase transition from one secular mode to another.
A Word or a Thing? A Trend or a Mode? Identifying and Redefining the "Secular"
The most obvious manifestation of the secular mode of the market is the broad behavior of the earnings and dividend yields (the inverse of the P/E and P/D, respectively), commodity and producer prices, consumer prices, and interest rates. My sense is that these have always tended to move together. This is probably what accounts for the various discoveries of correlations of these factors since Tooke first noticed that the general price level moved together with interest rates back in the mid-19th Century, or when Gibson noticed the correlation between producer prices and interest rates, or when Kondratieff noticed supercycles across a range of economic measurements and political trends.
The strongest link, in my opinion, is between commodity prices and the earnings yield, and so I look primarily to those factors in order to draw out the secular disposition of the market. That is quite different from standard usage. Contrary to a simple extrapolation of the trend, the secular disposition should be identified (rather than defined) first of all by the trend in the earnings yield (or P/E, if you prefer) and commodity prices. Not by the trend in stock prices. The correlation that Professor Shiller assumes exists between stock prices and the P/E ratio only exists after the establishment of the Federal Reserve. Therefore, since the 1920s, and especially since World War II, we can take real stock prices as a fair proxy for the secular disposition of the market, but there is nothing inevitable or natural about this. As we saw at the height of the crisis a couple of years ago, even in the modern market, the P/E ratio can rise quickly on a sudden collapse in earnings rather than on a rise in stock prices.
This accounts for the central difference in the way I am talking about the secular mode of the market and the way that most people, as I understand it, refer to secular bulls and bears. The primary indicator is the rise and fall of commodity prices and the earnings yield (i.e., the inverse of the P/E multiple). In the modern market, P/Es are driven by the 'P' side of the ratio, so we can generally equate Ps with P/Es, but this is just a useful fiction. This is why I disagree with the way others, such as Doug Short, define secular markets, although on a literal level, they are right: a "secular bull" market in equities should be identified by the "secular" (strictly, "long-term") trend of the stock market. But, as with so much in the market, what is gained from being quite "correct" is lost in being simply "wrong," in confusing words for things.
So, it is true that the word "secular" (bear with me) refers to nothing more than a long-term trend. But, the phenomenon that we have been carelessly applying it to is something rather different. Unfortunately, there is no good word to substitute for this phenomenon that would not cause even more confusion, so I prefer to let the good triumph over the perfect. For the last half-century and even the last century, the general use of the term "secular" alongside my own partial misappropriation of it generally refers to the same thing. If one wanted to think in terms of "long waves" or "supercycles," that might be okay, too, for the post-World War II period.
But, what confused me when I first became aware of these relationships and what I have become increasingly convinced of is that a secular rise fall in P/E ratios (and a secular fall in commodities) coincides with a sudden, qualitative shift in market behavior rather than as a trend that can only be identified in hindsight. I am increasingly of the opinion that, although a secular market usually contains one or more complete business cycles (so that the secular bull of the 1980s and 1990s neatly contained two standard NBER business cycles, for example), a secular shift can occur over a short period of time and, however rare, in the middle of a business cycle.
I believe that in the mid-1970s, in the brief aftermath of the oil shock of 1973, the market temporarily shifted to "secular bull" mode (replete with a rise in P/E and rising stocks, as well as falling commodities, inflation, and interest rates), only for it to switch back to a "secular bear" by the end of the decade without the interruption of a cyclical shock.
Whether the reader will be convinced of that is not so much my concern at present. I am merely trying to make clear that I am arguing that there is a qualitative difference in these secular modes and that "business cycle thinking" will just confuse you if you try to apply it here. Although the business cycle is a legitimate level of analysis, it is a mine that has largely been exhausted. We are generally familiar with its patterns without being any wiser as to its causality.
One huge block to understanding may be ignorance, if I can use such a strong word, of the secular mode.
The Relationship Between Yields and the Yield Curve
The impetus for writing this article was my rereading of one of Shiller's old papers on the bond market. In there, he wrote that a high spread on any two Treasury yields is typically followed by a fall in the longer-term yield over the course of the duration of the shorter-term yield. That was a surprise for me, because in my articles touching on Treasury yields, I have found that because yields across the curve, but most especially on the lower end, tend to share a common cyclicality, a high yield curve tends to coincide with cyclical lows in interest rates. Prior to reading Shiller's paper, if someone had held a gun to my head and forced me to produce a gloss on the relationship between the yield curve and the future behavior of interest rates, I would have said that, typically, a high yield curve would, over the next couple years, be followed by a rise in yields across the curve. A reasonable gunman, I think, would let me off with no more than a flesh wound.
(click to enlarge)The most severe breakdowns in my generalization about the yield curve and the cyclical behavior in rates occurred in the early 1980s and the early 2010s, especially with respect to the 10-year. The rise in yields predicted by the yield curve in the late 2000s was followed by a fall in the long-term yield in 2011-2012. This was not what really caught my eye, however.
More interesting than this is the concurrent behavior of yields and the yield curve. What I have found is that, a secular rise in P/E ratios coincides with a tendency for yields, especially long-term yields, to positively correlate with the yield curve. During a secular fall in P/Es, yields revert to a default, inverse correlation with the yield curve.
So, if you think about our current ZIRP-regime, by definition, the yield curve must strongly correlate with movements in the long-term yield, and we would expect to find this indicative of a strong, secular rise in P/E, which in the modern era would coincide with a "secular bull" market in equities.
Short-term Yields and the Yield Curve
Let's look at the short-term yield first, using the 3-month as a proxy. The yield curve is dominated by the short-term yield. If the Fed funds rate is more or less a function of a Taylor Rule, then so is the yield curve. The spread in the rate of inflation and unemployment coincides with the yield curve spread.
(click to enlarge)But, if you look at a chart comparing the yield curve (flipped upside down in the chart image below) with changes in short-term rates, you will see that although there is always a pretty strong inverse correlation at all times, it has been weaker in the 1980s, 1990s, and 2010s. And, in the mid-late 1980s, late 1990s, and 2013, the correlation has been straightforwardly positive.
Long-Term Yields and the Yield Curve
Although the behavior of the 3-month yield can draw out instances of the most severe deviations from "standard" relationships, longer-term yields, such as the 10-year, seem to do somewhat better. In the chart below, I have flipped the yield curve back up in order to make it easier to see changes in the correlation. In the mid-1970s, mid-late 1980s, early-mid 1990s to 2000, and 2009-present, the 10-year and the yield curve appear to be positively correlated.
(click to enlarge)In the chart below is a rolling, five-year correlation in three-month changes in the 10-year yield and the yield curve. There is an uncanny resemblance to the P/E ratio and commodity prices, so that a rise in the correlation between the 10-year and the yield curve tends to be matched with a rise in the P/E10 and a fall in real commodity prices.
Looking at a similar rolling correlation between five-year changes in the 10-year yield and the yield spread produces similar, if less striking and more erratic results.
This suggests two things to me:
1. Because yields across the yield curve tend to move together, this is a difficult tool to use, especially in isolation, in real time. Rather, it is probably better to use it as a means of confirming readings picked up from commodities, stocks, and earnings, and trying to identify possible future deviations from historical patterns.
2. With that being said, very short-term changes in long-term yields seem to contain information about the broad, secular disposition of the market--more so than medium- term changes. Although in a somewhat different context, Shiller finds that short-term changes in yields do not reflect "efficient" behavior on the part of bond markets, I think that this takes for granted the possibility that we can confidently theorize what efficiency really looks like. If we account for the secular, whether as a mere trend or as a more complex mode, bond markets tend to look more ordered, if not more "rational."
The Bull Market in Equities
Of more immediate concern is what this says about the current state of markets. If my interpretation is correct, this suggests that the yield curve is telling us two things.
1. The yield spread is well into positive territory, which, on a cyclical basis, is bullish.
2. As long as ZIRP is in place, the yield curve will be dominated by movements in the long-term yield, making the yield spread and the long-term yield, obviously, correlate almost perfectly. On a historical basis, we are in a secular mode of rising P/Es (and presumably, rising stocks) and falling commodities. This would also tend to suggest that a precipitous collapse in bonds is not on the cards anytime soon.
In other words, in the context of research I've done over the last year, each of the three major asset classes, stocks, commodities, and bonds, are all indicating strong cyclical and secular bullish signals for equities (SPY, DIA, QQQ), bearish signals for commodities (GSC, GLD, SLV), and neutrality with respect to bonds (UST).
The Yield Curve And Combining Secular And Cyclical Analyses
In conclusion, this new way of looking at the yield curve, studying it for the purposes of extracting secular information, does nothing to alter the traditional use of it as a cyclical tool. A flat yield curve, even more so than an inverted curve, is still a strong indicator that a contraction is approaching.
What it does is indicate where we should be allocating assets and, somewhat paradoxically, which assets will be most directly affected by contractions that appear on the horizon. In other words, asset classes (thinking mostly of stocks and commodities) that are in strong secular bull markets tend to be most negatively impacted by cyclical changes. Cyclical contractions that have occurred during secular bull markets in commodities tend to hit commodities more than stocks. In fact, cyclical contractions in secular bull commodity markets seems to trigger a switch in the secular mode of the market. Of course, it is only my opinion, but in each of the three contractions that occurred during commodity bull markets (1973, 1980, and 2008), apart from the immediate system-wide volatility, equities turned out to be the chief beneficiaries. Although commodities were to recover by the late 1970s, what this indicates is that although we can analytically separate out the secular mode from the cyclical mode, combining them can provide even more information. So, similarly, when contraction hits during secular bull markets in equities, equities tend to suffer more than commodities on a cyclical basis, although secular bulls in commodities appear to be somewhat more resilient. The 1990 recession did not kill the secular bull, although the 2000 one seems to have.
That is, information we can gather from the secular level of analysis can do a lot to help us make sense of what is happening in the markets today and allocate assets accordingly. Therefore, I continue to believe that we are in a secular and cyclical bull market in equities and a secular bear market in commodities. Bonds look to be stable and in the middle of a relatively normal, if somewhat early, cyclical correction.