Ronald Reagan once remarked that the nine most terrifying words in the English language are: “I’m from the government and I’m here to help.” Similarly, the financial economist Campbell Harvey recently wrote that the four most dangerous words, feared by central bankers and economists alike, are “This time is different.”
Despite much postcrisis writing on our inability to spot what seem like similar events, every technological or financial innovation allows practitioners to invoke those four magic words. And sometimes they’re right to: the Amazons and Netflixs and Visas of every era started off small and genuinely revolutionized their industries. For them, times were different.
Are the Yield Curve Inversions Different?
Beginning earlier this year, yield curve inversions have been in the news a lot. Going back to the 1960s, the inversion of the yield curve — that is, when the yield of long-dated government bonds falls below that of short-dated bonds — has successfully predicted recessions. As some yield curves inverted in March and again in late August this year, what everyone wants to know is: is there a recession coming, or is this time different?
For bond prices, it’s very easy to make credible “this time is different” stories. Former Fed Chair Janet Yellen recently said that yield curve inversions are no longer reliable indicators for future recessions. The reason, she argued, is that QE and other unconventional monetary policies have blurred the yield-inversion signal from the bond market. Heavy objections notwithstanding, Yellen has a point; when the Fed has bought up to one-fifth of all outstanding government debt with the explicit purpose of pushing up its prices (thus lowering its yields), it seems believable that the old pattern of upward-sloping yield curves no longer means what it used to.
To the yield curve–blurring effects of QE, we can add the heightened demand for short-term government securities because of Basel III liquidity regulations. Banks and insurance companies no longer hold bonds, as they did in the past, for investment purposes but almost exclusively for regulation-induced liquidity reasons — severing the link to the real economy.
It would be astonishing if the outcome of unprecedented monetary and financial experiments did not change the informational content of the world’s bond markets.
The events that yield curve reversions are intended to forecast — recessions and accompanying asset price falls — are moreover predicted only vaguely and with very large variances. While, as Richard Salsman recently explained in a popular AIER piece, the yield curve has reliably forecasted previous recessions, its predictive power has been highly variable. For the handful of postwar yield curve inversions, the recession they preceded arrived any time between a few months and two years later.
That’s not a very precise indicator. And it could happen by chance alone. Former Caltech professor Leonard Mlodinow writes in his superb The Drunkard’s Walk about our tendency to find patterns where none exist:
We should keep in mind that extraordinary events can happen without extraordinary causes. Random events often look like nonrandom events, and in interpreting human affairs we must take care not to confuse the two.
Which Yield Curve?
There are many different versions of “the” U.S. yield curve: The oft-cited yield spread between 10-year bills and 3-month bills (10y-3m) fell below zero in March 2019 and has been reliably negative for almost five months straight. Another version, the yield spread between 10-year bills and 2-year bills (10y-2y), has trended downward for several years — dipping below zero in late August — but has stayed positive since before the financial crisis. One can create almost endless varieties of yield curves, limited only by the current maturities of outstanding government debt.
Low–interest rate regimes have prompted both governments and companies to issue longer-dated bonds, a return of a market that had been dormant for decades. Those new maturities will allow ever more versions of yield curves to be calculated, the inversions of which can worry us even more.
For reasons of spurious statistics, that’s a warning sign right there. With enough variables and detailed-enough statistics, we’re bound to stumble onto correlations that seemingly predict anything.
Recently, Harvey, whose original research established the predictive power of yield curve inversions 30 years ago, has received a lot of media attention. His Ph.D. dissertation studied the 10y-3m together with the 5y-3m, which has been inverted for even longer than the usually quoted 10y-3m and 10y-2y yield curves. Before a signal can be trusted, Harvey argues, the yields have to remain negative for a full (calendar) quarter. Why that cutoff point, precisely, is rather unclear.
Indeed, I can construct a new one, 30y-3m, that didn’t invert in March when many of the others did; it inverted in late August and is now back on a positive slope (admittedly, only by around 30 bps). This specification did signal recession ahead of 2001 and 2007–9, but, by Harvey’s own quarter restriction, missed 1990–91 and was too late for the 1981–82 recession. Selecting one’s yield curves carefully can make them invert when appropriate.
Since there’s nothing special about one yield curve spread relative to over another, I am inclined to believe that the predictive power of various inversions are accidents of the data — noise — rather than genuine and universal reflections of underlying financial realities. That’s ironic, considering that Harvey’s own research on factor investing has successfully drained the factor zoo from that literature’s many spurious correlations.
What Should Investors Do?
In the middle of the last few months’ inversion hype, Eugene Fama and Kenneth French tried to figure out if yield curve inversions carry valuable and actionable information for investors. Using monthly returns for U.S., World, and World ex-U.S. between 1975 and 2018 and checking six different spreads (5y-1m, 5y-1y, 5y-2y, 10y-1m, 10y-1y, 10y-2y), they found “no evidence that inverted yield curves predict stocks will underperform bills.” Like Harvey, they only count yield curves that went negative for at least three months. For the vast majority of portfolio strategies (67 out of 72), switching from equities to bonds after a yield curve inversion reduced payoffs — and that’s before accounting for fees and transaction costs:
The results should disappoint investors hoping to use inverted yield curves to improve their expected portfolio return…[I]nverted yield curves provide no information about future excess stock returns.
I’m sure somebody, somewhere, can find a portfolio switch that overperformed following some carefully selected version of a yield curve that inverted. Our times might not be different, but banks’ demand for liquid assets sure looks pretty new.
Last week, Luke Delorme provided a lot of useful guides for investors and savers. In light of the yield curve signal’s inability to overperform, it seems that for most investors the takeaway is that yield curve inversions are mostly noise. Ignore them and stick to your strategy.
Made available by the American Institute for Economic Research.