One common notion among market timers is that a negatively sloping yield curve suggests a future recession and poor stock market performance. In this study, we will use the available evidence to examine this notion, and then convert our findings into two simple decision rules. We will then evaluate the performance of these two rules relative to using a "buy and hold" strategy.
First, what is the yield curve? The yield curve is simply a graphical relationship of the yields available at different maturities for the same class of debt security. The most common type of security to use when constructing yield curves is government Treasury securities.
Let's look at yield curve data as of 10/15/2012 (Source - U.S. Treasury Department):
Yield curve data in graphical form:
Just an interesting note before we continue: The lower line is the government's own estimate of "real" interest rates after inflation. Notice that real interest rates are negative up to about twenty years, and real 30 year rates are less than .5%. If you are a skeptic like me and consider the government's method of calculating inflation to be defective, the situation for conservative savers is even worse than it first appears.
What we see is that interest rates are relatively low for short-term maturities, and relatively high for longer-term maturities. This is considered a normal, upward sloping yield curve.
There are many viewpoints on why the yield curve is shaped the way it is, and academics have written countless papers trying to explain what causes it to change. The NAS Trading "quick and dirty" viewpoint is that the yield curve is determined by two things:
The supply and demand for loanable funds at the various maturities
Direct government intervention
Most government manipulation has historically been on the short end of the curve. However, efforts like quantitative easing (where the Federal Reserve "prints" money to buy bonds) directly manipulate the long end of the curve as well.
Few investors realize that prior to the shift in the role of the Federal Government that occurred during the Great Depression, the yield curve in the United States frequently ranged from negative sloping to flat.
As Jean Mathieson Gray states in, The term structure of interest rates, 1884 - 1915:
Short-term rates may have been higher than long-term rates almost without exception over the thirty years, 1884-1930. The basic yields also show that negatively sloped yield curves were more common than those with positive slopes for the years between 1900 and 1930. From 1930 until the late 1950's, short-term rates were always lower than long-term rates. In more recent years, yield curves with negatively sloped segments have again been observed, but never for periods exceeding a few months' duration. Thus, 1930 appears to mark a major shift in the history of the relationships between long- and short-term interest rates in the United States.
In effect, the Federal Reserve's primary "job" throughout its history has been to manipulate short-term interest rates - usually down.
Regardless, today a negative yield curve is perceived as being bad for both the economy and the stock market, as it suggests monetary policy is tightening. This tightening culls asset price speculation and cuts into the profits at financial firms.
Let's get to the meat of our analysis:
For this study we will be using S&P 500 data from 1962 on, and the yield curve spread between 3 month Treasury bills and 10 year Treasury bonds. We are using S&P 500 index data rather than SPDR S&P 500 ETF Trust (SPY) data because it gives us a much longer time horizon over which to conduct our study.
The average monthly return for all months was .62%. When the yield curve had a negative slope at the start of the month, the return for that month was on average -.33%. I have represented this difference in average returns in the graphic below:
An investor who had the misfortune of only investing during months when the yield curve started the period negative would have lost -22% of their capital during a period where the market rose about 2230%.
This data clearly suggests that for at least a one month time horizon, a negative sloping yield curve has not been good for the stock market.
Does this observation hold true over a longer period of time? The shape of the yield curve is usually considered to be a long-term indicator. Let's evaluate how a negatively sloped yield curve has impacted returns over the following year:
I have included data using each month as a starting point because I believe it creates a more complete picture. Going from left to right, the first column is the month ("1" = January, "2" = February, etc.), the second column is the return over the next 12 months, and the third column is the number of those months since 1962, etc.
The first thing that stands out is that for every monthly starting point, the total return is lower when the yield curve starts the period with a negative slope. Seven of the months actually have a negative return over the next year. While this data is not statistically perfect, it adds solid evidence to the notion that a negatively sloping yield curve is bad for the stock market.
Let's convert the yield curve ideas expressed above into two timing systems. The first system will exit the market any month that starts with a negatively sloping yield curve, and not re-enter until the curve is positive sloping again. Here are the results.
This system more than doubles the total return of buy and hold during the entire period, however it failed to sidestep the 2000's bear markets and displays substantial volatility.
Let's examine the results using a different rule set:
If yield curve is negative, exit the market for one year
Do not re-enter until no month within the past 12 months has had a negative-sloping yield curve
This model outperforms both buy and hold and the first yield curve timing model ($5,750,000 vs. $4,930,000) but still displays substantial volatility.
Our conclusion is that even given the limitations of the available data, there is solid evidence that a negative yield curve is bad for stock prices, and as such it deserves a place in the market timing tool kit. However, the signal has not been reliable enough to use as a stand-alone timing approach.