In my last article, I wrote about the apparently strong correlation between commodities and the earnings yield, in something approximating Gibson's Paradox.
In that piece, I had simply multiplied the real prices of four benchmark commodities (gold, silver, copper, and oil) and compared them to the estimated earnings yield. Below is a chart which separates them out and compares them individually.
The weakest link here seems to be copper, but by and large, the commodity-yield connection appears to be strong. Interestingly, there are two occasions that jump out as radical reversals in the correlation, the early 1980s and the beginnings of the current crisis. In both instances, earnings yields jumped up sharply while commodity prices crashed.
But, I would like to focus on the relative performance of the commodities in question. I have spent a good deal of time writing about two commodity relationships: gold vs copper and gold vs crude oil. I feel fairly strongly that the gold/copper ratio, for whatever reasons, forecast (although I can't exclude causation) real interest rates sixteen months in advance. And, I have tried to argue that the gold/oil ratio has historically pointed to large-scale oil price movements (twelve months in advance) and interest rate movements (sixteen months in advance). The interaction with the yield curve seems to indicate primarily monetary factors.
For example, commodity prices and the earnings yield moved strongly upward in the early 1970s prior to the jump in oil in 1973. (Although not on the chart above, the yield curve was compressed below the key 1.0 level, as well). This "oil shock" is regularly attributed to the oil embargo, but Professors Barsky and Kilian make a strong case for regarding the magnitude of the move as monetary in origin. I don't intend to try to move at that level of theory, but there do seem to be some resonances between their interpretation of "oil shocks" and some of the themes we have addressed here before.
In any case, I would again like to draw the reader's attention to the relationship between gold and oil and its relevance to both the Gibson Paradox and the Fed Model.
Gold, oil, and silver all move in sympathy over the long term, but oil's movements are usually the most exaggerated. One obvious instance of this was at the turn of the century: by the time the earnings yield bottomed, oil had already started its great trek upwards.
(Source: commodity data, UNCTAD; CPI data, St Louis Fed)
An arithmetical view gives us a somewhat better look at some of the divergences.
Here is a comparison of the earnings yield with the oil/gold ratio:
(source: commodities, UNCTAD; earnings yield, Shiller)
There appears to be a vague-ish relationship between the oil/gold ratio and the earnings yield. It seems that where the oil/gold ratio is relatively low, yields tend to drop, and when the ratio is high, yields tend to rise. This says nothing (so far as I can tell) of cause and effect. There are times when each appears to lead the other (yields leading in the early 1970s episode mentioned above and oil/gold leading at the turn of the century), so one suspects a third element generating both, but to spare the reader any unnecessary suspense (or to inadvertently stimulate it), I haven't any notion of what the underlying cause might be in any precise fashion, other than to say that the earnings yields appear to run in a given direction until they reach once unimaginable extremes (say, irrational exuberance/despair+50%) and then reverse.
So far, I have been using commodity data from UNCTAD. For crude oil, they appear to be using an index of different grades. However, I would like to focus on the benchmark WTI, as well as end of month gold data. By raising the spectre of precision, I run the risk of undermining the ultimate claim I am trying to make: that the behavior of the oil/gold ratio relative to the Dow/gold ratio strongly suggests that commodity prices are tied to earnings yields and that it is virtually impossible for this correlation to be merely coincidental. Nevertheless, even if the precision I seek to employ should turn on my argument, I think that it still illustrates the strength of the connection between commodities and yields.
In any case, I mention the Dow/gold ratio, because it is strongly correlated with P/E ratios, and it has the benefit of being less dependant on human estimation.
(sources: Dow and gold, Wren Investment Advisers; P/E data, Shiller)
Or, at least it has been until now. There has been a fairly big divergence since the economic crisis kicked in. Effectively, since the "Dow" and the "P" of the S&P "P/E" are virtually the same thing (indexed prices of equities), this suggests at least a temporary breakdown in the relationship between gold and earnings.
(source: gold, Wren Investment Advisers; earnings, Shiller)
As far as I am aware, the correlation between P/E ratios and the equity/gold ratio goes at least as far back as the establishment of the Fed, but that is no reason to assume that it will continue in the future.
In the meantime, supposing that there is an underlying relationship between the Dow/gold ratio and P/E ratios, I am going to substitute the former for the latter. So, below is the oil/gold ratio (using WTI data from the Fed and end of month gold data) and the Dow/gold ratio.
(source: WTI, St Louis Fed; Dow and gold, Wren Investment Advisers)
If you look closely, you can see that the oil/gold levels of 0.05 and 0.12 appear to be meaningful indicators. A move above 0.12 appears to indicate a downward move in the Dow/gold ratio, while a move below 0.05 appears to indicate that Dow/gold is about to "turn and burn". Even if one wishes to disregard absolute levels in the oil/gold ratio, on a relative basis, these "signaling" moves are massive, moving from the top to the bottom of the range or vice versa. If one accepts that the Dow/gold ratio is linked to the P/E ratio, it becomes, I think, quite difficult to hold that commodities are not a function of earnings yields a la Gibson's Paradox.
For whatever reasons, the commodities markets (particularly gold, oil, and silver) are exceptionally--almost unbelievably--sensitive to future movements in interest rates (both real and nominal) and earnings yields. I do not believe that one need accept or even judge the validity of any absolute levels as key in order to appreciate this quality of the commodity markets.
I have to say, too, as an aside, that one never hears anything good said about commodity markets. They are the scenes, we are told, of all sorts of nefarious activities, but this link between commodities and these more generalized economic phenomena strongly suggest that conventional wisdom is badly misinformed about how commodities markets work and the vital service they provide--although I suppose someone could just as easily regard this as evidence of just how rigged the markets are.
I would like to push forward with consideration of these absolute levels, however. Before I do, I need to note a few objections. One is obvious. In the early 1970s, a plunge in the oil/gold ratio below the 0.05 line did not result in a reversal in the Dow/gold ratio for quite some time and the same could be said for the situation since 2009. In the 1973 instance, it took a full two years before the Dow/gold ratio moved up. In our current situation, it is possible that the August 2011 low in the Dow ratio was the bottom, but it is hard to judge without appeal to other tools, and I would like to reserve that attempt for my next article.
But, there is a second objection. In 1976 and 1980, the violation of these levels were cursory. Actually, it's somewhat worse than that. In 1976, the number on the chart above was 0.11659 rather than 0.12. This raises the data problem again. The Fed data is listed as first of month, but it appears to be a monthly average of spot prices, which is probably why it is only updated after the month is over. Oil markets were not yet deregulated until the early '80s, either, so it is rather difficult to track down data from periods prior to then.
Nevertheless, the chart below uses end of month data, and it appears to confirm the violation of these levels. It is a gold/oil ratio chart, so the levels to look out for are 20 (rather than 0.05) and the awkward 8.33 (instead of 0.12).
(courtesy Dan Denning at The Daily Reckoning)
Technically, this chart seems to put me on safe ground, but the brevity of the January 1980 violation is certainly a slender reed upon which to rest any absolute conviction that these levels constitute anything like economic laws.
But, it strikes me as increasingly unlikely that these commodity ratios are mere chartist anachronisms and trivialities. They are more likely expressions of the powerful ebb and flow of money in a world dominated by the vicissitudes of liquidity.
So, let's review for a moment what it is that the oil/gold ratio tells us (without forgetting the anomalies in the 1970s or in the current 'new normal'). In previous articles, I argued that the gold/oil ratio led interest rate movements sixteen months in advance, while here I am saying that the oil/gold ratio provides a more or less instant indication of long-range movements in equities yields.
A wholesale collapse in the oil/gold ratio, for example, would point to the beginning of a long-term trend of falling equities yields (equivalent to a rise in the equity/gold ratio) but an upward spike in treasury yields sixteen months later.
It is possible that the link between the gold/oil ratio and treasuries is more tenuous than I had previously believed. In other words, if on a short-term basis, treasury yields and earnings yields operate under a similar kind of cyclicality, the fact that the gold/oil ratio seems to point to movements in treasury yields may in fact be due to a larger synchronicity in treasury and equity yields rather than to any exclusive relationship between the gold/oil ratio and interest rates.
In a previous article, I put up the chart below, noting the breakdown in the correlation between the gold/oil ratio and interest rates.
Around the middle of the last decade, the correlation broke down dramatically and has remained negative ever since. Previous falls in the correlation almost always promised a flat or inverted yield curve, as it did in the last decade, but the correlation has generally tended to "correct" itself. Now that treasury and equity yields are no longer aligned, this change in correlation may mean that the oil/gold ratio has always been speaking to equity yields rather than to treasury yields. So long as treasury yields and equity yields are in a kind of cyclical sympathy, the gold/oil ratio can be looked to for important clues to interest rates, but it is important to keep these questions in mind.
The reason, to recap the argument, for this correlation between the oil/gold ratio and equity yields is due to the relative sensitivity of commodities to equities yields, a phenomenon that seems closely related to Gibson's Paradox. Or, more precisely, this correlation is due to the relative sensitivity of commodities to the conditions that also prompt sympathetic moves in equities yields.
If Gibson's Paradox is now to be found in earnings yields rather than in treasury yields, then this will require a reevaluation of many of the dynamics of economic and monetary forces. On the other hand, this apparent change in behavior that has occurred since the disintegration of Bretton Woods offers an opportunity to more clearly identify how and why the original manifestation of Gibson's Paradox operated the way it did.
Indeed, the implications for how we think about monetary policy, central banking, asset allocation, and economic theory generally seem endless.
In my next article, I would like to look more closely at the complicated and unusual conditions surrounding the mid-1970s divergence and our post-crisis one now. Although some steps can be taken to account for both divergences, our current situation, from the perspective adopted here, is quite unique and offers no easy solutions. In such situations, it is important to try to identify which relationships are economically enduring and which are ephemeral, especially when it comes to searching for returns.
It appears, on the basis of relationships outlined in our previous articles, that we are on the brink of a significant contraction that will torment virtually every asset class for the remainder of the year. Stocks may well be the least worst option of the major asset classes. As I have argued before, apart from a hedge on an oil spike (USO), the most profitable opportunities this year remain in volatility and shorting commodity currencies such as the Aussie dollar (FXA) in favor of traditional safe-haven currencies such as the Swiss franc (FXF) and the yen (FXY). The fact that the oil/gold ratio is pointing to a strong equity performance relative to gold gives one pause, but ultimately it seems that the totality of factors point to a stock market (DIA) caught in the pincers of high commodity prices (on an absolute basis per my previous piece) and a global banking crisis and the possible onset of rising real interest rates.
I almost feel obliged to say something about the correlations between whatever phenomenon I am trying to describe and the list of global banking crises in Reinhart and Rogoff's This Time is Different. Virtually all of these global banking crises occur during periods of secularly falling equity yields (implying secularly strong equities and weak commodities). The only exceptions would be the 2007 mother of all credit crises and the 1981-1982 Latin American crisis, both of which approximated tops in earnings yields and commodities. Were we to use the oil/gold ratio as a way of precisely timing secular shifts, then we would have to regard the two outliers as the 2007 crisis and the 1977 Spanish crisis, this time marking a return to rising equity yields that had been interrupted from 1973.
The question which would seem to be dogging us, then, is whether or not the 2007 crisis and its aftermath signaled the end of the decade-long commodities run or an interregnum akin to the mid-1970s. The vague sense that one gets is that some portion of the market is going to have to move first to break out of this grinding holding pattern we are in. If we are, as I have postulated before, on the brink of a sharp rise in interest rates coinciding with falls in virtually all major asset classes, this has the potential to break the neck of an already badly crippled gold market (GLD), which may set us up for at least a temporary market respite beginning sometime this winter.
Additional disclosure: I am long September WTI. I am short September S&P 500 futures, AUDCHF, and AUDJPY.