Note: In this "Inflation and Yields" series, I am attempting to outline the profound relationship between and within the yield and price complexes.
Since the end of Bretton Woods in 1971, virtually all real commodity prices have been positively correlated with equity yields. This is especially true of industrial commodities and precious metals. The US natural gas market is one of the few exceptions to this rule, and this appears to be a rather recent development.
Is this general correlation accidental?
Since the early 1970s, when the dollar standard came into its full glory, we have only had one clearly defined peak in equity yields (early 1980s) and one bottom (late 1990s). In my opinion, circumstantial evidence indicates that the correlation is not accidental, but I have struggled with the problem of causality.
This correlation smacks both of Gibson's Paradox (the observation made famous by Keynes that interest rates used to be highly correlated with the price level) and Robert Barsky and Lawrence Summers' attempted resolution (see PDF) of that problem (namely, that real gold prices and a number of real metals prices are negatively correlated with real interest rates), summarized by Brad DeLong here.
The correlation found in Gibson's Paradox disintegrated a good century ago. I have shown that although there does appear to be a constant equilibrium within the yield domain, since sometime around 1960, a new dollar equilibrium has insinuated itself, one shaped around the rate of inflation (the rate of change in CPI or PPI) rather than the absolute level of CPI or PPI itself.
And, in my previous article, I showed that the Barsky-Summers thesis has held neither during the dollar standard nor the gold standard, although there is some suggestion within their paper that equity yields are the key to Gibson's Paradox.
The question that naturally asserts itself is whether this correlation holds only for the market equilibrium under the empire of the dollar or if it is a more general phenomenon indifferent to the monetary standard.
The answer appears to be that real commodity prices have been correlated to equity yields for centuries.
Real wholesale prices in the United States
First, I am going to use wholesale prices (WPI, or PPI) as a proxy for commodities and, by extension, WPI/CPI as a proxy for real commodity prices.
The correlation between real wholesale prices and the earnings yield for the period 1872-2007 is 0.53; for the dividend yield, it is 0.67.
Although I do not have the data, in a recent paper by Tyler Muir (PDF, Figure 3) in which he estimates dividend yields going back to 1834, the results seem to fit nicely with the real wholesale price levels charted above.
Real wholesale prices in Britain
In the United States, that is great; but the figures in Britain do not suggest this kind of relationship. Assuming that British equity yields broadly approximated American ones over the last century, we can see that it is unlikely that British WPI/CPI is correlated with equity yields on their side of the pond, especially in the last six decades.
(Source: Wholesale price from Jastram; consumer prices from Jastram and Bank of England)
As I mentioned in my last article, it is rather remarkable that British wholesale prices have fallen relative to consumer prices for the last three centuries with only intermittent interruptions. Even taking into account the errors that inevitably creep in over the span of a three-hundred year price index or set of indices, it seems unlikely that this general trend is simply the product of a miscalculation.
For the moment, though, we are going to set aside this problem and return to one of the original questions from Barsky and Summers, namely the functioning of real gold prices.
In considering the question of gold, I first turned to Roy Jastram's analysis of the history of gold prices over the last five centuries in The Golden Constant, updated in 2007 by Jill Leyland of the London Bullion Market Association (see summary in this PDF).
One of the most surprising things he found was that, contrary to conventional wisdom and apart from the 1970s (the book was originally written in the latter half of that decade), gold has historically been a terrible inflation hedge and an excellent deflation hedge.
He demonstrates that using charts similar to these:
(Source: Gold and WPI from Jastram; silver price spliced from World Bank Pink Data and Global Price and Income History Group (GPIH) at UC-Davis, attributed to Jastram, presumably his Silver: The Restless Metal; 'estimated WPI' splice of Jastram WPI and industrial price index)
I think he slightly overstates the case, but this is true under a gold standard, simply by definition. The exceptions come when a metallic standard is suspended or eliminated.
In England, this was the case somewhat during the Napoleonic Wars.
In the United States, this was true during the Civil War.
What appears to happen is that gold simply begins to behave like the broader commodity complex.
It tends to rise faster and crash harder. I have added silver to the mix in the chart below.
(Source: Gold and WPI from Jastram; silver spliced from GPIH and World Bank)
Silver was demonetized far earlier than gold (during the Civil War), and its behavior tends to lead to the same conclusion.*
(Source: Gold, WPI from Jastram; silver spliced from GPIH and World Bank; estimated copper price spliced from Fed data and World Bank Pink Data; all prices nominal)
Gold and Barsky-Summers
Returning to Barsky-Summers and the supposed equivalence of gold and nonferrous metals, a little imagination applied to the previous chart would suggest that the real price of gold has gone through a fundamental shift since the demonetization of gold while other metals such as copper, aluminum, and tin have always behaved simply as commodities.
Moreover, since real commodity prices tend to be highly correlated with equity yields, we should see a transition from gold being negatively correlated with nominal yields (especially equity yields) during the gold standard period to being positively correlated under the dollar standard.
And that's precisely what we find both in the United States and Britain:
(Source: Gold, WPI, CPI from Jastram; dividend yield from Shiller)
I don't have British equity yield data but using the long bond yield as a general proxy for the purpose of illustration, we can see that real gold prices (especially when deflated by WPI) have undergone a fundamental change with respect to yields, having once been negatively correlated and now positively so.
(Source: Consol yield and CPI from Bank of England; gold, WPI from Jastram)
The chart below does something to demonstrate this transformation in the behavior of gold prices. Whereas once, real gold prices were negatively correlated to real commodity prices (as represented by the WPI/CPI ratio), they have since become highly correlated. This is true whether we are speaking of gold deflated by WPI or CPI (as you can see from the chart above). But that also suggests that gold behaves like a super-commodity. Gold trounced WPI in the 1970s and again in the 2000s, when WPI was strongest.
And, that sounds nice, but this is also true of silver and of copper, as well, and a good many commodities.
So, is gold a "super commodity" or is WPI just an inadequate substitute for commodity prices?
My suspicion is that a) gold and silver have unique properties within the commodity complex and b) that wholesale and producer price indices represent such a broad spectrum of commodities at multiple stages of the production cycle that they can only serve as broad guides to commodity behavior at best.
That is to say, the correlation between real wholesale prices and equity yields in the United States was lucky, but not coincidental.
In my next article, I am going to bear down on primary commodities (i.e., raw industrial and agricultural commodities) to more precisely formulate the relationship between equity yields and real commodity prices. And, hopefully in the article following that, the way will be open to discussing how we can translate these observations into action.
* Note on silver prices during the American Civil War and historical data more generally:
Curiously, the data from the Federal Reserve suggests that silver did not budge during the Civil War. The series A04018GB00LONA286NNBR is listed as "Price of Bar Silver for London, Great Britain,...Dollars Per Fine Ounce, Annual, Not Seasonally Adjusted." This stability in dollar-denominated silver is confusing because although the silver price in London appears to have remained flat during the American Civil War, the exchange rate between the pound and the dollar went from $4/£1 in 1861 to nearly $10/£1 in 1864, according to the Bank of England. It is therefore difficult to imagine how the dollar-denominated price of silver in London could have remained flat during that period.
And yet, the GPIH data attributed to Jastram also lists both American and British silver prices as flat during that period. It also, however, lists an implied exchange rate of $4.53/£1 in 1864. It would be foolish for me to bite the hands of those who feed me data by claiming that the Fed and Jastram are wrong, but it is awfully strange that silver (of all commodities) should have kept its head when all about it were losing theirs!
Looking at the Fed's notes about the silver price, unless I am misreading the explanation, it seems to indicate that fluctuations in exchange rates were not taken into account in calculations of the dollar-denominated price until the 1930s.
Leaving aside this particular question, the lesson, I think, is that we should always take historical data with a grain of salt and try to keep our eye on the broad trends rather than incremental movements, understanding that that has its own dangers, too.
Additional disclosure: I am short gold and S&P 500 futures.