If you haven’t been worrying about the possibility of an inverted yield curve, now might be a good time to start.
If you’d purchased a 3-month U.S. Treasury bill at last week’s auction, you’d be earning an annual yield on your investment of 4.03%. A 6-month bill would be yielding 4.3%, about 25 basis points more. Does that mean that the 6-month security was a better investment for your money?
Not necessarily. Suppose you bought the 3-month bill and then, as many Fed-watchers are expecting, the Fed raises the fed funds rate another 25 basis points at each of its meetings in December and January. If that prediction turns out to be right, then when your 3-month bill matures in February, you could probably reinvest it at a new 3-month rate that’s 50 basis points higher than today’s 3-month rate and 25 basis points above today’s 6-month rate. If that indeed comes to pass, the 6-month return from rolling over two 3-month bills would be virtually the same as if you’d purchased a single 6-month bill right now.
A very simple model known as the expectations hypothesis of the term structure of interest rates posits that investors don’t particularly care which maturity they invest in, and as a result would always bid prices for different maturities so that the expected yield from rolling over securities of different maturities is identical. The blue line in graph at the right displays the current yield curve, which plots the annual yield of Treasury securities of different maturities on the vertical axis and the length of time to maturity on the horizontal axis. According to the expectations hypothesis, the current steep slope of the yield curve at the shortest maturities signals that investors are indeed anticipating a very rapid rise of 50 basis points or so in the short-term interest rate over the next three months, but expect subsequent hikes in interest rates to be more gradual and drawn out. The red line shows that the same curve was substantially steeper this time last year.
Although the expectations hypothesis has some descriptive power, a large number of academic studies have demonstrated that it is not a completely accurate quantitative description of how yields on different maturities are related. For one thing, the yield curve almost always slopes up, even in times when investors believe interest rates are likely to fall. A longer term security involves more risk of a capital loss and is less liquid than a shorter term security, so most investors would prefer the short rate even if they expect to earn a lower return. For this reason, it in fact is a fairly unusual development for the yield curve to become inverted, or actually slope down. A modified version of the expectations hypothesis holds that the long yield is an average of the expected returns from rolling over short rates plus a constant term premium. This modification, too, is not consistent with all the evidence, which shows that the term premia can change over time in significant ways.
The lower graph at the left records one quick summary of the slope of the yield curve at any given date, as provided by the spread between the yields on a 10-year Treasury bond and a 3-month Treasury bill that you would have seen in any given month. The spread became negative (signaling an inverted or downward-sloping yield curve) on only six occasions during the last half century. Five of these were associated with economic recessions (indicated as shaded regions on the graph), and the sixth (1966) with a significant economic slowdown.
Arturo Estrella, an economist at the Federal Reserve Bank of New York who has done a great deal of research on the term structure, has a a nice summary of the academic research on the relation between an inverted yield curve and economic recessions (hat tip: Economist’s View). One reason that the yield curve inverts prior to an economic slowdown is that economic recessions tend to mean lower interest rates, both because of lower demand for borrowed funds as well as the fact that the Fed is likely to lower rates aggressively in response to an economic downturn. According to the expectations hypothesis, to the extent that investors see this coming, the yield curve would start to slope down before the drop in economic activity.
In "http://dss.ucsd.edu/~jhamilto/kim.pdf">research with my former student Professor Dong Heon Kim, I found that the expectations hypothesis alone does not explain the ability of the yield curve to predict an economic slowdown, but that there is movement over the business cycle in the premium that investors require for long-term securities that also contributes to the ability of the yield spread to predict economic downturns. Research by another former student, Dr. Hibiki Ichiue of the Bank of Japan, suggested that this may be related to news about inflation, which seems to raise the market price of output risk and thereby lower the term premium prior to an economic downturn.
If the ten-year rate had stayed at the values below 4% that we’d seen at the start of the summer, we would be at another point of inversion right now. However, longer term rates have drifted upward over the last five months to keep the yield curve from acquiring a downward slope, though not fast enough to keep it from becoming more flat. With the 10-year over 3-month spread currently at 50 basis points, unless we see further hikes in long rates, we could face an inverted yield curve by the end of January.
The statistical evidence suggests that any time the yield spread gets below 100 basis points, it raises the probability of an economic slowdown even if not a full recession. And if two more FOMC hikes to 4.50 by January aren’t enough to do the job, four hikes that put us to 5.0 by May surely would. For all these reasons, I take comfort that some Fed-watchers are concluding that the Fed is likely to pause before then.
Let’s hope they’re right.