CIO Blog: Curve Ball
All around us are investment opportunities, and all of them carry risk. Theories of what maximises the former, and limits the latter, abound.
In theory there is no difference between theory and practice. In practice there is.
All around us are investment opportunities, and all of them carry risk. Theories of what maximises the former, and limits the latter, abound. A leading one at present is centred on the US yield curve, which is flattening: short-term rates are rising going up as the Federal Reserve raises rates, but the long-end of the curve is staying relatively static, subdued by dim expectations for growth and inflation in the future. Theory suggests that if, or when, the short-end rises above the long-end – an “inversion” – a recession is imminent. Theory carries on to say recessions are bad for equity markets, and therefore an inverted yield curve should be a strong signal to cut risk positions. To add meat to the bones of supposition, all US recessions over the last seven decades have been preceded by inverted yield curve (see chart above).
But in practice, the picture is much more nuanced.
Ahead of the Curve
First, while an inverted yield curve does appear to occur before recessions, there has to be some time parameter for this to be a useful signal. Otherwise, it would be akin to saying a recession follows a solar eclipse; it will always follow the last solar eclipse, even if it happened 10 years ago! A 12-month period is a sensible starting point. Seeing the table, we can see an inverted yield curve only heralded a reces- sion over the next year three out of ten times1.
So, an inverted yield curve is not a confirmed signal on some impending recession. Even by widening that corridor to 18-months, a recession would have occurred six out of 10 times, a little better than half.
Figure 1: Market Returns Following Observations of “Inverted” Yield Curve
Negative Yield Curve
|Months to Recession||Forward Returns|
|1M Return||3M Return||6M Return|| |
|% of observations negative||40%||40%||30%||60%||40%|
Tan shaded areas shows a recession occurring within 12 months from the observation of a negative yield curve. Pink shaded areas shows a recession occurring within 18 months from the observation of a negative yield curve.
Second, sometimes, an inverted yield curve does not signal a recession at all. There was clearly no recession forthcoming for 50 months after November 1965, or for 34 months following June 1998, long enough – using common sense – to say the signal failed.
Third, while knowing a recession is coming is useful, in any given year, gross domestic product often falls while markets surge, and vice versa.2 Much more important for investors should be what a negative yield curve tells us about the future performance of equity markets. Here too the signal appears to have limited efficacy. Once a negative yield curve was observed, equity markets were down over the next year in six of the 10 observations. However, the average is still positive; and double-digit gains (in 1988, 1998 and 2006, respectively) occurred more often than double-digit losses (in 1965 and 1973, respectively).
Clearly, an inverted yield curve is far from a perfect signal for either recessions or equity sell-offs. Nonetheless, it does tell us a lot about market-wide perceptions of risk, and it is quantitative expression of investors’ expectations for monetary policy, economic growth and inflation. While these have little bearing on near-term market performance, equities and economies are highly correlated over the long run.
As such, the shape of the US yield curve is a factor in our Macro-Cycle Indicator. However, it is one of nine other signals that, in aggregate, have far more efficacy in heralding a favourable environment for taking equity risk than any single factor alone. Moreover, the economic climate in and of itself is also too narrow. The most effective method in making investment decisions remains to dispassionately assess the economic climate, in conjunction with maintaining an incessant focus on valuation, momentum and sentiment.
1 Conventionally, 10-year minus 2-year Treasuries yields are used to calculate the slope of the yield curve. We have used 10-year minus 3-year US Treasuries in the analysis given a longer available data set. The difference is marginal. We also apply some basic optical judgement to ascertain when the curve is actually negative, as opposed to just skirting the border.
2 In their highly regarded book, Triumph of the Optimists, analyse 16 markets over 100 years, and they find no relationship between economic growth and stock market returns.