In this series, I am attempting to describe the fundamental relationship between yields and inflation.
In the previous articles in this series, I said I would attempt to show that the real price of primary commodities (e.g., grains, metals, oil), but especially industrial commodities, were highly correlated with equity yields, and that this was likely the source of Gibson's Paradox.
Reliable equity data for the U.S. goes back to at least the 1870s, and is easily accessible from Robert Shiller's website. And although one always wants more data, there is a fair amount of historical commodity data from such sites as the St Louis Fed, the World Bank, the Global Price and Income History Group (GPIH), the International Institute of Social History (IISH), MeasuringWorth, the BLS, Long-Term Returns and other sites.
Demonstrating the relationship between commodity prices and equity yields is not that difficult for the U.S. If it is not Gibson's Paradox, it sure is a lot like Gibson's Paradox.
It doesn't matter what we call this commodity/equity yield connection, but if one could show that this was not merely a quirk of the American data, that it had a longer pedigree from a noble European line, if you will, it would suggest that we are witnessing a remarkably durable law of economics.
The problem is that there is, outside of the U.S. experience, no readily available data on long-term earnings yields and only limited data on dividend yields. And, the data on dividend yields doesn't go back much farther than the U.S. data.
All hope is not lost, however. Setting aside the question of equity yields, if this Gibsonian effect that I have identified is the same thing as the Gibson Paradox, then commodity prices should be highly correlated with global interest rates from the eighteenth century up until the early twentieth.
The real questions (from my perspective) are not so much whether commodity prices are highly correlated with gold standard-era interest rates. That has already been more or less established at the level of wholesale goods. Rather, the two most difficult questions are a) what was the relationship between equity yields and interest rates during the gold standard, and b) what do we mean by "real" prices?
These two questions really stemmed from a third question, which was namely, why? Why should primary commodities be correlated with yields of any sort at all?
Looking back at the historical data by itself does not give an answer, but if one could identify whether or not it was a particular class of primary commodities, that might narrow things down a bit.
That is easier said than done, however.
I won't bore the reader any more than I have to with the problems of collecting historical market data. I will only say that it is considerably amplified when one is trying to find a coterminous set of data for interest rates, equity yields, CPI, and a broad, representative basket of primary commodities. That difficulty makes me all the more grateful for the data I have found, and which is generously provided by a number of researchers and government agencies, either free of charge on their websites or as appendices to books and articles.
The patchwork of data that I now have strongly suggests to me that my original thesis, that real commodity prices are highly correlated with equity yields, is the source of Gibson's Paradox. It is difficult to present such a patchwork in a convincing or economic way, though.
I think the best way to do that is to frame the fundamental problems in this article, and then to present the individual cases in the next couple of articles. Some of those principles may be more controversial than the claim they are intended to support.
First, I propose to get around the problem of a lack of equity yield data by assuming that during the gold standard period, global equity yields, interest rates, and prices were all highly correlated with one another.
For some readers, I suspect that they will have difficulty accepting the inclusion of equity yields into this mix. As we have all seen from the collapse of the Fed model and its debunking by means of historical data going back to the 1940s, equity and bond yields can go shooting off in very different directions.
The very limited evidence of the 40-some years of Shiller's data from the gold standard period suggests, however, that it is this breakdown that is the anomaly. And, I think, although some will find it a paradoxical argument, that the history of the last century also suggests that gold standard yields trended together.
Let me break this argument down into five parts.
1. From 1870-1913, yields were correlated with one another.
2. From 1870 up until today, the yield complex interacts in a systematic way. This can be roughly modeled as EY - DY + 1y - 10y = 0, where "EY" and "DY" refer to earnings and dividend yields and "10y" and "1y" refer to long- and short-term Treasuries, respectively. As I've written before, that particular equation over-simplifies the relationship, but I think it's dollar-standard permutation,
EY - DY + 1y - 10y = CPI%,
indicates the original equation's durability.
3. Since these four yields seem to exhibit systematic behavior with respect to one another and inflation, it is not a coincidence that the most significant breakdowns in the correlation between equity yields and bonds (having in mind chiefly the earnings and long-term yields) occurs when short-term yields approach or break below "1.0". That was true during the Depression, and that has been true for the last decade. In other words, the so-called "risk premium" has only spiked when short-term yields start going ZIRP.
(Source: Shiller data, except BAA yield from St Louis Fed).
4. There is no evidence anywhere of short-term yields behaving in this crisis manner during the gold standard, and if we can assume the same (or at least a similar such) relationship in other historical markets, there is more reason to assume that equity and bond yields tracked with one another than that they did not.
5. Fortunately, during the gold standard period, we have some idea of how equity yields performed when bond yields were at their lowest point in the nineteenth century, because that was during the 1890s. I can find no evidence of extreme equity yield movements in the U.S., UK, Germany, or France at that time.
In a word, then, I cannot prove that movements in equity yields or the "rate of profit" coincided with movements in interest rates, but in the absence of any such data, I hold that it is more likely than not that equity yields did track with bonds.
As I said, I think that most readers will find this assertion about equity yields the most troublesome aspect of my statement above.
That is not the part of the argument that keeps me up at night, however. The problem is identifying and defining "real" commodity prices. This was a problem I had really hoped to avoid, but as I dug deeper into the historical data and looked at the composition of price indexes over time, it became increasingly clear to me that not only could it not be avoided, but that exploring it points to (or begins to point to) a new understanding of inflation and perhaps the deeper causes of Gibson's Paradox.
Take one of the current controversies about the way inflation is calculated now. There is some bitterness and anguish over the question of whether or not fuel and food prices should be included in the measure. They are considered too "volatile" to be included in "core inflation."
But I am not so much interested in wading into that controversy as using it to illustrate my problem at the historical level.
Today, we tend to strip out CPI of as much volatility as possible. One could say that we seek to strip it of those elements that are, coincidentally or not, most closely tied to primary commodities. Bread does not grow in fields, and unleaded gasoline does not spring from the ground, but there is not that much of a gap between primary commodity and finished product in the food and fuel categories of CPI.
Historically, however, consumer price indexes were composed almost entirely of food and agriculture, especially grains. (I am still unsure of whether I should regard livestock as a primary commodity or not). So, to exaggerate things a bit, what does it mean to deflate the price of wheat by the price of flour in the nineteenth century? Or iron by the price of nails?
This is not an academic or remote question, I think. Why did we have steadily falling nominal commodity prices in the 1980s and 1990s, but steadily rising consumer prices? And then why explosive commodity prices over the last decade with comparatively mild inflation? One could grind one's teeth at statistical manipulation of CPI, but over the long-term, it would be impossible to hide. There's something else going on.
We could also ask ourselves, why should we regard consumer prices in any form as a good deflator? And, why draw a great dividing line between producer and consumer prices?
Frankly, I am somewhat doubtful about the whole producer-consumer paradigm. I think it is a useful fiction. For the businessman or woman, it is probably more useful than fictitious, but for the economist or political economist, how does one draw a line between production and consumption? Is a school lunch "productive"? Or is it mere "consumption"
At the moment, without a greater effort on the part of the establishment to explain what "consumer prices" really represent, I think it has to be regarded as a largely political conception. We want to know what "consumers" pay, what "people" pay. But, is there any clear economic reason why we should deflate by an index of consumer goods? And, what good does it do to have a "consumer" index if we are going to strip it of food and fuel anyway?
In other words, we have an index of prices that we are utilizing to identify "consumer" and "core" inflation, when these might be two real but wholly different things.
Let me put this another way, because it bears directly on the problem of Gibson's Paradox, I think.
While writing and rewriting drafts of this article, I continuously bumped into the problem of how to characterize the collection of commodities that I am arguing are tied to equity yields. So far in this article, I've been labeling them with the quite innocuous term "primary" commodities, and it's not a bad description.
Previously, I have written that this connection between primary commodities and yields is to be found chiefly among "industrial" commodities. But it seems to be evident in precious metals, too, and I don't think that their industrial uses can account for why they should behave like industrial commodities. If one were to find gold listed as an industrial commodity, you couldn't help but cock a skeptical eyebrow (or two if you are a bit weird).
But, to say that Gibson's Paradox is "primarily to be found in industrial commodities and precious metals"… has an odd ring to it. It suggests that you haven't put your finger on the underlying connection. So, then I read an old paper while researching the history of aluminum and nickel prices, and it occurred to me: "nonrenewable natural resources!" Of course! We can grow as much corn and coffee and as many bananas and tomatoes as we please (Malthus be damned!), but we have a finite amount of oil and copper and gold.
Looking back over the patchwork of historical data I have, though, I am somewhat less confident that we can simply exclude agriculture from consideration. Although agricultural commodities have not been as strongly correlated with equity yields as have nonrenewables in recent decades, it is unclear to me whether there is a shift in the Gibson effect from agriculture to industry that has coincided with the process of global industrialization, or whether there is "interference" from the "real economy." That is to say, if you take a single commodity like aluminum, for example, and look at its history, it has undergone a complete transformation from precious metal in the nineteenth century to roadside rubbish today. Technological breakthrough appears to have had a far greater impact on its historical price trajectory than did any Gibson effect.
One could imagine that either the real economy (such as in the case of aluminum) or even the transition from gold to the dollar has somehow impacted agricultural prices differently than it has impacted nonrenewables and therefore, distorted this Gibson effect in some fashion.
I tend to think that if there is a transition of the Gibson effect from agriculture to nonrenewables, this has something to do with the rise of industrialization, but I cannot be sure. And, in any case, agriculture hardly seems to be immune to this Gibsonian tendency, even today.
So, this brings me back to the point above. If historical CPI measures are composed of things like wheat and barley and products derived from them, what good does it do to deflate commodity prices by CPI? Is there some other way of identifying a "core inflation?" Or, was there no core inflation under the gold standard?
It would be simplest for my argument, I suppose, to simply take the following tack: "Under the gold standard, nominal commodity prices were highly correlated with yields, but under the dollar standard, it is real commodity prices that are correlated with equity yields." And that would be nicely symmetrical with the point I tried to make with respect to the change in the dynamics of the yield complex in the equations above.
But, having looked over the data patchwork of the last two hundred years (and beyond), I just don't think that the change in monetary regimes has had any impact on the underlying relationship between primary commodities and equity yields. I think that it was always "real" prices and equity yields, but that under the gold standard, core inflation or baseline inflation was virtually null.
Moreover, I am increasingly under the impression that there are two poles of inflation, or two independent generators of inflation. On one end are primary commodities - "elemental" goods - and on the other end are composite goods, particularly composite finished goods. In other words, I intuit the decisive difference between primary commodities and core inflation to be the interference of human labor, in its broadest sense. The higher the amount of intellectual and physical manipulation of a good, the more immune it is to the cyclicality of commodities, and the closer it hues to core inflation, whatever "core inflation" refers to.
So, under the gold standard, one can imagine this process in the following idealized manner:
And, under the dollar standard, a system of chronic core inflation, one can imagine it thusly:
Again, the last 30 years have shown us that something is clearly up. Twenty years of falling commodity prices, a decade of rising commodity prices (e.g., a 1500% increase in oil prices, give or take), and yet CPI just kind of ticks along without a care. The great and sudden thud of commodity prices in the wake of the financial crisis gave us an ever so brief dip in inflation, but otherwise, hardly left a ripple.
If there is this polarity in the price complex between elemental, primary commodities and composite finished goods, there should be some evidence of it somewhere apart from the contrast we have seen in the behavior of commodity and consumer inflation since 1980.
Perhaps there is a better proxy of finished composite goods, but the first thing that popped into my mind was durable consumer goods, so I conducted an "experiment" comparing the ratio of the producer price index of raw commodities to the PPI of finished capital goods on the one hand with the ratio of consumer nondurable to durable goods on the other.
(click to enlarge)
(The blue line is the durable/nondurable ratio. The green line is the finished capital goods/raw materials ratio).
As you can see, over the last 50 years, durable goods have generally taken a shellacking at the hands of nondurables. It is difficult to argue that there is a long-term relationship between these two lines. Relative commodity prices have gone up and down. But I think that there is clearly some kind of relationship here. At one- to 10-year intervals, there seems to be a connection, but also the decline of the relative strength of durable goods prices accelerates during commodity booms.
The really interesting thing, I think, is that in the 1970s, like everything else, nominal durable goods prices rose quite sharply, but in the 2000s, when commodity prices exploded, durable goods prices steadily fell. If these are not the two poles, they seem to be very close.
So, how can that be? If I were to redraw those idealized graphs of the inflation complex, how would I fit this phenomenon on there?
If I drew it in the following manner, with "real" commodities being an inverse of "real" composite goods prices, that would suggest that the rise of commodities prices in the 2000s somehow pushed down durable goods' prices. I can't rule that out, but I am hard-pressed to conceive of how such a dynamic would function.
Or, we could redraw it so that composite finished goods constitute baseline inflation with the general price level and CPI falling somewhere in between. But, it is, to slip into scientific jargon for a moment, really weird that commodity prices should have moved so strongly upwards at the same time that durables collapsed, so I suspect that there is something else going on, but I am not sure what.
I have to say that at first blush, it is also somewhat paradoxical that durable goods would (unless I am much mistaken) more likely tend to be composed of material derived from nonrenewable commodities than would nondurables, and yet perhaps their composite nature indicates to us that they are actually composed more of human labor and ingenuity than anything else. Metals are virtually worthless without the ability to manipulate them into an "unnatural" purpose. Grain, I think it is safe to say, was "meant" to be eaten. It might be slightly more difficult to say that the aluminum surrounding my Coke was "meant" to serve that purpose by nature.
Instead of graphing this polarity in the dynamic manner above, however, it might be more useful to graph it statically for the time being, as two separate sources of inflation, primary commodities and finished composite goods, a polarity that resembles and overlaps the producer-consumer price polarity.
In my next article, I had intended to write about how we can utilize Gibson's Paradox as traders and investors in a dollar-denominated world, but because I have had to use this article to set the stage for the larger demonstration of the historical data, I will have to postpone that for the moment.
For readers who simply cannot stand the anticipation, I intend to incorporate observations I have been making over the last year with respect to the relationship between the oil/gold ratio, the cyclicality of interest rates, and the stock market. Many of the elements of the argument with respect to application can be found in my previous articles on Seeking Alpha.