Inflation And Yields: The Evolution Of Gibson's Paradox And The Revolution In Prices, Part I

by: John Overstreet

Note: In this "Inflation and Yields" series, I am outlining the fundamental relationships between and within yield and price complexes. This article focuses chiefly on the price complex, but it is built in no small part on a previous article on the yield complex.

Since 1871, the "real price" of primary commodities (that is, commodities deflated by CPI) has been strongly correlated with equity yields.

1871-2010 New Gibson ParadoxClick to enlarge(Source: Please see text for information on data.)

This bears such a striking resemblance to Gibson's Paradox--Keynes's name for the observation that, from 1730-1913, the nominal general price level was highly correlated with bond yields--that it is hard to believe that these are two discrete phenomena. That Gibson's Paradox was even stronger with respect to the wholesale price index (WPI), i.e., producer prices (PPI), lends credence to that belief.

Click to enlarge

The question is how to reconcile these correlations.

In my last article, I asserted that by observing the nature of the yield complex in America over the last 140 years, we can reasonably assume that under a metallic monetary standard, equity and bond yields were highly correlated with one another. Unfortunately, the lack of reliable data on stock markets and yields makes it difficult to judge that claim empirically during the 1730-1870 period, and it seems unlikely that we will ever be able to reconstruct long-term equity yields by direct appeal to data.

But, even granting this assumption about the behavior of pre-1870 equity yields only addresses half of the problem, the yield half.

It is somewhat trickier to demonstrate my solution to the price half of the problem. If these correlations are the same phenomenon, why is it one correlation involves nominal prices and the other deflated prices?

My answer to that question is that this is, in part, a statistical trick.

From 1871-1913, when the two correlations overlap, commodity prices moved more dramatically than consumer prices, allowing for this correlation to appear. As I will show, this happened again in 1914-1959.

The more important part of the solution to the price problem is that under a metallic monetary standard, nominal prices were real prices. Under a metallic standard like gold, there is no inflationary bias built into the economic system. All "inflation" and "deflation" are relatively ephemeral under a gold standard, whereas under a fiat standard like the global dollar standard, it is understood that there is a constant inflationary bias (which we will dissect later).

"Real" prices are the nominal prices of goods times the value of the currency. Under the gold standard, the currency's value (or, so I am contending) was effectively stable. Under the dollar standard, the currency's value is variable, at times strengthening but generally weakening (which is why we have had chronic inflation since the late 1940s). Unfortunately, there is no perfectly objective way of measuring that decline, so we generally use a basket of goods, such as CPI or measures of "core inflation."

In a fiat currency system, that makes perfect sense. One can quibble about which basket to use, but any broad basket tends to produce the same results over the long run: chronic inflation (that is, chronic devaluation of the dollar). But to measure real prices under a system built around an effectively stable currency such as we had under the gold standard by deflating a given price by a basket of goods is anachronistic. It would be like measuring the real price of consumer goods by deflating CPI by CPI. This might be a useful device in certain instances (as I will show), but it wouldn't make a lot of sense to say that this would give you "real" consumer prices.

In sum, when we deflate commodity or producer prices by consumer prices during the gold standard, we may get some useful information about economic conditions (such as "real wages"), but we would not get the "real" price of commodities from those relative prices. We use the term "real" price nowadays because we understand that the unit we use to measure prices - the dollar or the euro or the pound - is not stable, that it is almost always in a state of decay.

In the hope of reducing confusion or at least imposing a degree of consistency, from this point forward, I am generally going to refer to a) "deflated" prices for all nominal prices deflated by CPI and b) "real" prices as nominal prices under the gold standard and deflated prices under the dollar standard.

A better way of going about this argument, however, is to simply review the history of the relationship between prices and yields for the last three centuries, and that may help readers to get a grasp of what I mean, instead of talking about things in this abstract manner.

On data sources

I tested a wide array of price data from the United States and Britain from 1730-2010. Because the data comes from different sources and tends to be sporadic, I have had to splice together a number of price series. I have tried to be careful only to use domestic prices for each market (except for silver in the U.S.). I cannot always be sure that the sources I have used have not interpolated foreign prices, but I believe that the price series I have analyzed are relatively pristine in that respect.

For the UK, I have relied chiefly on price series from Blattman, Hwang, and Williamson's paper "Winners and Losers in the Commodity Lottery: The Impact of Terms of Trade Growth and Volatility in the Periphery, 1870-1939" (pdf) for data from 1840-1913 (the data on Blattman's site actually runs from the 1840s up until the 1950s and includes a handful of American price series that I took advantage of, as well). I also used the rather ample data from Gregory Clark's paper "The Price History of English Agriculture, 1209-1914," (pdf) which also has a lot of non-agricultural prices and wage data going back to 1700. I got WPI data (both generally and by sector) from Jastram's The Golden Constant, and Jastram's silver data from the Global Price and Income History Group's website hosted by UC-Davis. The data from Clark's paper can also be found there. Bond yields and exchange rates (which I used to recalculate American silver prices) were obtained from the Bank of England's website.

My brief mention of British dividend yields is largely based on Acheson, Hickson, Turner, and Ye's 1825-1870 estimates, (pdf) DeLong and Grossman's paper "'Excess Volatility' on the London Stock Market, 1870-1990," Grossman's subsequent revision of British dividend yields in "New Indices of British Equity Prices, 1870-1913," (pdf) and Long-Term Return's estimates of British dividend yields since the 1990s.

For the United States, I used price data from Blattman, Hwang, and Williamson's paper, as well as the BLS and the St Louis Fed websites. I have also used Pfaffenzeller's updated Grilli-Yang Commodity Price Index (GYCPI) and its component prices, and the World Bank's Pink Data to round things out. I also got American CPI, WPI, and gold data (although I treated gold separately elsewhere) from Jastram's The Golden Constant, updated with data from the Fed and Robert Shiller's website. Stock market data, including yields, are based on data from Shiller. Much of the data for individual commodities and food goods from the nineteenth century, as well as for wholesale price indexes for different sectors, come from the U.S. Census (pdf). It was evident that the PDF published by the Census has some scanning errors, which I corrected by reference to similar price series wherever possible, including Blattman, Hwang, and Williamson.

For American bond yields, I relied on Homer and Sylla's History of Interest Rates from 1800-1870 and Shiller from 1871. And, for historical wage data, I relied on the Fed.

Finally, I calculated money velocity rates with GDP data from and money supply data from Richard Anderson's paper (pdf) and the St Louis Fed.

Because of the sheer size and complexity of the data, I hope the reader will forgive me if I do not cite the sources for each price series as I present them. I believe all of the data used in this article fairly reflect historical prices. Moreover, I did not knowingly exclude any historical price data that was negative to my thesis. So, with respect to long-range prices, I believe this examination is fairly comprehensive. For modern prices such as those provided by the BLS, the data are too voluminous to be dealt with comprehensively, but I believe the prices used herein accurately reflect the phenomena they are intended to represent.

The outline of this article is:

Part II. To examine British prices and long bond yields from 1730-1913 in the hope of finding a greater degree of correlation between bond yields and primary commodities than between yields and consumer prices.

Part III. To use the much more extensive American price and yield data to assess which yields and prices are most likely to reflect correlations and to determine when the transition from nominal to deflated prices occurred.

Parts IV and V. To use these results to characterize the complete revolution in the behavior of non-commodity prices that began with the establishment of the Fed and accelerated under the reign of the dollar from Bretton Woods, with an eye to the simultaneous transition in the yield complex that we outlined before.

Fortunately, the British experience, which we will look at now, is relatively straightforward and easy to characterize and will set the stage for the rather less prosaic twentieth century.

Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it. I have no business relationship with any company whose stock is mentioned in this article.