The most remarkable thing about the last fifty years of market history is not the booms and busts that have dominated the period, but the surprising degree to which macroeconomic forces have become correlated with one another.
The majority of those correlations appear to be grouped around the behavior of the earnings yield. In previous articles, I have written about how real commodity prices, as well as government deficits and spending levels, have been strongly correlated with the S&P 500 earnings yield, especially since the implosion of Bretton Woods.
For example, in the chart below, I took all commodity price data provided by the World Bank and calculated each one's correlation with the non-adjusted earnings yield of the S&P. Unfortunately, I couldn't find a more economic way of presenting the data; the extreme left is energy (coal, natural gas, crude oil), the extreme right metals (gold, silver, platinum, copper, tin, iron ore, aluminum), and the center agriculture. The gaps are data series that are too short to permit easy calculation.
But the salient point is that virtually all commodities demonstrate this correlation. The two negative correlations are of US natural gas prices (the blue bar on the left) and potash.
In this article, I would like to offer some additional economic phenomena that appear to be connected with the behavior of earnings yields and then make a few tentative guesses as to how these things may be tied together.
The next chart should set the stage.
(click to enlarge) (source: Shiller data; St Louis Federal Reserve Economic Data (FRED))
This is a comparison of the S&P earnings yield (the P/E10 ratio inverted) with the misery index, which is simply the sum of the rate of inflation and the rate of unemployment. The correlation has been especially strong since the 1960s. Prior to that, the relationship was a lot less clear, but the Dow/gold ratio, which has had a close relationship with P/E ratios for the last century, seems to have had an inverse relationship with the misery index as well.
(source: FRED; Wren Investment Advisers)
In effect, the Dow/gold ratio appears to behave like an inverted moving average of the misery index, which is a testament, I think, to gold's role as a hedge against macroeconomic trauma generally rather than inflation alone.
In any case, if earnings yields are so tightly bound to the misery index, one naturally wonders how earnings yields should relate to the unemployment rate and inflation separately.
Earnings yields and unemployment
In the chart below, I compare the unemployment rate with earnings yields four months prior from 1948 to 2012.
(source: FRED; Shiller P/E data)
In other words, earnings yields seem to lead the unemployment rate by one or two quarters generally, especially since the late 1960s but even prior to that as well.
This may have something to do with why earnings yields and government spending and deficit ratios also are correlated. Unemployment rates and stocks effectively have an inverse relationship, so when times get bad, government stabilizers kick in (spending) but can no longer be paid for by higher tax revenue (deficits).
(source: usgovernmentspending.com; Shiller data)
Assuming that debt levels matter and assuming that P/E ratios are effectively an independent variable, it would seem as if governments should be more aggressive about reducing expenditures during periods of rising P/E ratios in order to have enough fiscal ammo to deal with periods such as we have experienced over the last decade.
Putting that aside, however, the relationship between earnings yields and unemployment are a little more complicated and interesting than they appear at first blush. While I was researching these topics, I came across this very interesting insight at the Political Calculations blog.
Gasoline prices tend to lead the unemployment rate by two years. For whatever reason, since the early 2000s, the two factors have not been as correlated as they once were, but the general relationship appears to remain.
On the one hand, this seems to confirm the relationship between earnings yields and real commodity prices I have talked about in previous articles and which, I believe, debunks at least to some degree the conventional wisdom on the behavior of commodity prices (specifically metals) first laid out by Barsky and Summers 25 years ago.
(source: FRED; Shiller data)
Prior to this, I would have guessed that if one were driving the other, that earnings yields were driving commodity prices, but this lead seems to call that guess into question.
In any case, this is a good place to segue into the question of inflation, the other unhappy member of the misery index.
Earnings yields and CPI
Here is the earnings yield compared to the rate of inflation for the last 140 years.
(source: Shiller data)
The long-term perspective suggests that earnings yields have largely 'underperformed' since the 1960s. Even so, the basic relationship appears to be intact.
The 2000s and the disinflation conundrum
The question is, why is it that the last decade of rising earnings yields have, as the last fifty or so years of history have suggested, coincided with rising unemployment, rising deficits and government spending, and sharply rising commodity prices, but disinflation? We could point our fingers at unemployment, but even though unemployment was rising off its dot.com boom low, it remained at broadly reasonable levels up until the collapse. Moreover, unemployment and inflation over the long-term actually tend to go hand-in-hand, with inflation leading by about half a dozen quarters, give or take.
In fact, it is all the stranger that this relationship has come so unhinged, except to the degree that they are still united by the smooth operations of the misery index.
Inflation generally leads unemployment, so it seems unlikely that unemployment would have much power to suppress the inflation trend.
The next culprit might be interest rates, but for the most part, central banks kept rates low, which we generally expect to have, all things being equal, an inflationary impact.
We had rising deficits and wars, each of which are regarded as inflationary as well.
In a sense, everything has pointed to inflation, perhaps with the sole exception of globalization. Richard Duncan has written of how the triangle of debt, dollar reserves, and disinflation have come to dominate the globalization process. And, this certainly seems a plausible argument, but then it is somewhat difficult to reconcile with the tremendous increases in commodity prices and it forces one to ask why the 1970s should have been so different from the 2000s, at least with respect to consumer inflation rates.
Gary Shilling recently pointed out that commodity inflation and CPI should be regarded as somewhat separate phenomena. He says that commodity prices are affected by producers stockpiling materials when they become afraid that they will not be able to keep up with demand. As for CPI, I understand him to say that this is largely determined by wage inflation and that the structure of the labor market nowadays has not permitted this inflationary feedback loop to take hold.
Whatever factor or factors is responsible for holding down inflation, it would seem that it is also responsible for keeping Treasury yields so low.
In my last article, I noted that the debt/GDP ratio has, for at least a century, had an inverse relationship with bond yields. During the three decades after World War II, when debt/GDP ratios fell, the average rate of inflation was nearly twice the level of the subsequent three decades (and counting) of rising debt/GDP.
It would seem, then, as if this mystery disinflationary force has managed to trump the inflationary bias of falling P/E ratios. I am not arguing that the debt/GDP is that disinflationary force, only that its strange relationship with bond yields suggests a possible link.
The most important factor for those of us interested in an attempt to formulate a long-term outlook on markets and the economy, I believe, is the behavior of the earnings yield and the market phenomena associated with it. Simply put, the question we have to ask is whether or not we believe that equity yields will rise, flatten, or fall over the next decade.
On an historical basis, even after a decade of rising equity yields, they still appear far too low to be sustainable.
(source: Shiller data)
They are roughly at or below previous bull market levels set in the 1920s and 1960s.
And, considering the degree to which equity yields have been correlated with commodity prices and the misery index, it seems hard but to assume that much of the next decade will be quite painful.
Does Fed policy matter?
Before commenting on how I believe the market will progress over that period, I want to make mention of the challenge before the Federal Reserve. It is somewhat of an oversimplification, but effectively its dual mandate equates to keeping the misery index low.
We tend to think of unemployment and inflation as forces that stand in opposition to one another. Macroeconomic underemployment contributes to disinflation while overemployment spurs price rises. But, the manner in which unemployment and inflation have been correlated with one another for at least the last half century suggests that they are effectively two sides of the same coin, with inflation playing the leading role by about one to two years.
That would make one assume that the sole focus of central bankers ought to be keeping inflation low, except that inflation (as the growth rate of CPI) was kept at relatively low levels for the decade prior to the explosion in unemployment following the credit crisis. So, there is certainly more to this story. I suspect that the role of absolute price levels (for both commodities and CPI) is underemphasized, but it is also possible that the Fed's policy instruments are simply not capable of managing the dynamic between inflation and unemployment.
Commodity shocks, either in terms of absolute levels or in rapid accelerations in price (such as formulated in Leeb's "Oil Indicator"), are effectively self-correcting. Sharply higher prices depress growth, which then depresses commodities. So, when the Fed finally gets around to raising rates according to the Taylor Rule or some such model, is it reducing the inflationary burden or piling on top of it? Once inflation (by whatever measure) picks up, it may already be too late.
But, the larger question is, does the Fed have a meaningful role to play at all? If equity yields dominate the economic cycle or at the very least represent the existence of a deeper macroeconomic cycle that lifts and depresses things like the misery index, does the Fed have any real power to modulate that cyclicality? Or, has it perhaps even unwittingly exacerbated it somehow?
How to play rising equity yields
Whatever role the Fed may play in all of this, it is our duty to try to come to grips with how these forces will affect future market action. If we expect equity yields to rise over the course of the next decade, then the surest source of profit is likely to come from commodities. Gold (GLD), silver (SLV), and oil (USO) have been some of the most responsive to rising equity yields, both in real and nominal terms. For long-term investments, trading futures is somewhat problematic because of the need to roll over contracts, so the alternatives are to buy physical gold and silver or to take stakes in commodity producers. The negative side of buying producers is that they are, at least to some degree, subject to the whims of the equities markets (DIA), which generally underperform in periods of rising yields.
The oil/gold ratio gives us a fascinating way of gauging how we might allocate our investments. The very volatility of the ratio is the source of its utility. Since the crisis, the oil/gold ratio has been at historically low levels.
That suggests that - setting aside the difficulty of gaining long-term exposure to the price of crude itself - oil offers less risk compared to gold. That is certainly a fair assumption. However, it is slightly more complicated than that, because when the oil/gold ratio is pushing the lower bound (which appears to be around the 0.05 level), both oil and gold tend to drop, as do equity yields.
In other words, although oil is more likely to rise than is gold over the long-term, the signal that a commodity bull has begun usually involves the oil/gold ratio pushing to the upper bound (somewhere around the 0.12 level), in which case, gold becomes the more reliable bet.
To further complicate matters, since the credit crisis, gold - somewhat similar to the 1973 crisis, which has had a number of similarities to our current problems (for example, rising equity yields and commodity prices, oil shocks, banking crises) - has continued to make higher highs, at least up until the summer of last year. I am of the opinion that gold is over-extended and that much better prices will be had over the next year or two.
Although the long-term trend appears to point to rising commodities and earnings yields, these intermediate-term indicators point to an interregnum period similar to the mid-1970s, which saw gold quickly give up a lot of its gains. It does not always pay to be too clever, however, and the most prudent solution may be to slowly accumulate precious metals, even with the expectation that gold is due for a very severe correction.
At the moment, I am continuing to short stocks (SPY) and gold, and will wait until the new year before I go long on equities again.
Additional disclosure: I am currently shorting December futures for the S&P 500 and gold.