Flat Curves and Stock Prices

The flattening of the yield curve in the bond market has become a hot topic lately.  There are a number of excellent analyses out there that explain why … in a nutshell a flattening yield curve has tended to foreshadow declining stock returns, in particular a yield curve that inverts.  This is not only an empirically observed phenomenon but one that has a logical justification; it means that bond investors anticipate lower growth and inflation in the future than they expect in the near term.  Rather than rehash the numerous arguments that support this relationship, let’s take a look at the rebuttals.  One that appeared on Bloomberg yesterday even holds that a flat-to-modestly-upward sloping yield curve is one of the best environments in which to buy stocks.

In Flat Yield Curves Are No Reason to Sell Stocks, Aaron Brown cites monthly return statistics showing that on average,

“… You get the best average monthly return, and a lower-than-average volatility. How can it be that with recession looming, stocks have high average return with low variance? We can get more insight by looking at a scatter plot of S&P 500 one-month returns versus yield curve slopes at the beginning of the month.

Months with losses exceeding 15 percent happen when the curve is steep. There have been few big disasters once it flattens to the point where the spread between two- and 10-year yields is less than about 0.8 percentage point. And when the curve inverts, really good months of 10 percent plus returns are scarce…”

Let’s unpack this.  First you take a series of yield curve observations, and list the returns for months following each of them as a function of the slope of the yield curve at the beginning of the month.  In other words, we take a continuous timeline of stock market returns, chop it up into individual months, throw them in the bin and shake them up, and then line up all the pieces with their corresponding yield curve stats.  This is as crazy as it sounds.  This statistical technique discards any semblance of continuity beyond a one month time frame.  Why does it seem to mean something?

A look at market history reveals why.  Late in a bull market cycle, short term interest rates begin to rise as the Federal Reserve responds to increasing growth and inflation.  Once these short term rates reach their peak relative to longer term rates (the yield curve is flat to inverted), it takes some time before the effects are manifest in the economic data.  If the effects were obvious immediately, there would be no need for short rates to rise that far in the first place.  Later, once the economy begins to cool and stock prices to head down, short term interest rates begin to decline and the yield curve steepens again.  But having become too elevated during the bull market, stock prices have yet further to go down before they’re in line with the slower economy.  The bear market continues for a while, and the deepest selloffs typically occur at the end of a bear market.  So of course you get an association between the steepest of yield curves and negative stock returns.  Once the economy is in recovery and stocks are solidly in a bull market again, there is no longer any need for very low short term interest rates and the yield curve begins to adopt a more normal positive slope.

Yield curve inversion is an early warning indicator.  It can take anywhere from 6 to 24 months from inversion for the stock market to respond.  Why would you just look at the following month?

Slicing and dicing a long period of stock returns – a cycle that extends over several years – into little monthly snippets and throwing out all the longer term sequence information is a highly artificial exercise that can’t possibly illustrate the major relationships involved.  Wall Street must want to sell its stocks badly to contrive such clever arguments.  My advice for anyone looking for clarity on the question is to consult any of the numerous sources that show you the big picture, the whole picture, not just a peephole squint at pieces of it.  Nothing chopped up and glued back together in a scatter plot.  Nothing withheld from the eye.  Here’s one example which plots the difference between the ten year UST yield and the one year UST yield versus the annualized rate of total returns on the global stock market.  The only manipulation to the data is a high frequency rolloff to filter out the short term noise.