How Inflation Altered the Stock – Bond Relationship

The yield on the ten year Treasury note briefly passed 5% recently, the highest yield on the ten year note since before the financial crisis of 2007 – 2009.  This surge in yields has led some commentators to argue that bonds now appear attractive relative to stocks, given that a 5% yield on government backed bonds compares favorably to the S&P 500, which now has an earnings yield (the inverse of the P/E ratio) of ~5.5% if you use a forward earnings yield, or ~5% if you use a trailing earnings yield.

Put another way, does the surge in interest rates of the last couple years make bonds so attractive that it portends a period of poor performance for stocks?  This is what practitioners of the “Fed model” argue, who use comparable bond rates to determine whether stocks are over- or undervalued.  While it makes sense in theory, – all else equal, if bonds are offering anemic yields then investors would bid up stock prices as they are comparatively cheap relative to bonds, – the reality is actually quite different.

Stocks are a risky asset class, subject to periodic bouts of panic selling due to anything from recession-induced earnings fears or geopolitical uncertainty.  Government bonds, on the other hand, are comparatively safer, given that they are backed by the full faith and credit of the United States.  You would think that investors would demand a risk premium for owning stocks relative to government bonds given this, but the reality is that over the last several decades, the risk premium has been essentially zero (yellow line on the chart below).  In fact, during the tremendous equity bull market of the 1980s and 1990s, stocks would have been considered overvalued relative to bonds pretty much the entire period.

Taking the analysis a step further, we can see that while the earnings yield can be useful by itself as far as predictive power of subsequent returns is concerned, the earnings yield relative to bond yields is far less so:

It should be pointed out, however, that this was not always the case.  As can be seen in the time series below, the earnings yield relative to bond yields was fairly tightly connected with subsequent equity returns from 1926 – 1970:

In fact, while the earnings yield in isolation was also fairly highly predictive of subsequent equity returns over this period, the earnings yield less bond yield was even more so:

It is only in the post-1970 world that this relationship breaks down, with the earnings yield itself still demonstrating predictive power, but the earnings yield relative to bond yields being little better than random:

So what happened to alter fundamentally the relationship between stocks and bonds in this latter period?

The answer, I think, may be the unstable inflation introduced in the post Bretton Woods world which ended when President Nixon dissolved the dollar’s link to gold in 1971.  Prior to this, there were two major inflation shocks, – the end of price controls following the end of World War II, and then the Korean War in the early 1950s, – but in each case, the shoch was short-lived.  What made the inflation of the 1970s and 1980s different was that it was sustained at a higher level than previously and during a period of peace:

There is an interesting article from Basil Copeland in the June 1982 edition of the Financial Analysts Journal that is worth quoting in full on this topic:

Conventional wisdom has it that the market risk premium is positive and that the market requires a greater return on stocks than it does on bonds…

What is missing from this explanation is an account of the impact of inflation on the risk inherent in debt and equity securities.  In a world where inflation can be perfectly anticipated, a common inflation premium will be incorporated into investors’ nominal return requirements for all classes of securities.  Furthermore, in such a world the yield on U.S. government securities would constitute the nominal risk-free rate of return (as long as investors believed that U.S. government securities were free of the risk of default).  But in a world of uncertain inflation, all classes of fixed income securities are exposed to an element of risk that is not shared to the same degree by shareholders.  With uncertain inflation, the bond investor has to worry about the risk of substantial loss of capital if the rate of inflation turns out to be greater than anticipated.  The shareholder, however, receives some measure of protection against inflation because of the ability of the firm to raise its prices.  For stocks to be more attractive than bonds during periods of inflation, it is not necessary that stocks be perfect inflation hedges.  All that is required is that stocks be better inflation hedges than bonds. [my emphasis]

In other words, what investors are missing when they compare stock and bond yields is that the former is a real yield while the latter is nominal.  The difference may seem subtle, and in periods of low and stable inflation such as that which prevailed prior to the 1970s, it may not matter all that much, which likely explains a great deal of why the earnings yield less bond yield demonstrated utility as a forecasting tool.  But in periods where inflation is more volatile, it is bonds, not stocks, that become in many ways the riskier financial instrument over the long term, a sentiment with which the market appears to agree by assigning virtually no discernible difference over the long term in the valuations of stocks and bonds.

Whatever the reason for the breakdown of the stock and bond relationship since 1970, what does appear to be one takeaway from this analysis is that more useful to use equity valuations in isolation when it comes to forecasting equity returns than it is to introduce contemporary bond yields into the equation.