Is the US Treasury yield curve really “Mr Reliable” at predicting recessions?

In the spring of last year, the US Treasury yield curve inverted for the first time since 2007, leading investors to speculate that a recession was coming. The yield curve has signalled (or at least preceded) every US recession over the last 50 years (chart below). The curve typically inverts towards the end of economic cycle, on average, 12 months ahead of a recession.

With the term spread – the difference between short dated and long dated yields - turning negative last May, the curve suggested that the US would head into recession towards the end of the first half of this year.

Why does the yield curve invert towards the end of the cycle?

After a period of expansion, with the economy growing robustly and inflation accelerating, the Federal Reserve (Fed) tends to “put the brakes on” by hiking interest rates. This typically lifts yields at the front-end of the curve, but dampens them at the long-end, which leads to the curve flattening and sometimes inverting. In other words, higher interest rates increase the cost of borrowing for companies and households now, but this is expected to have a negative impact on future demand and inflation in the economy.

Moving into 2020 and with the emergence of Covid-19, the US economy is set to take a significant hit to activity in Q2. Global growth is expected to experience the worst year since the 1930s (see here). While the yield curve did not predict coronavirus, it did highlight that the US was late in the longest period of expansion in two centuries. In addition, the expansion increasingly resembled a “wobbly bicycle”, one which could topple given the slightest bump in the road. The bump, in this case, was the coronavirus pandemic.

But is the yield curve truly reliable?

Despite the yield curve’s enviable recession-predicting credentials, some commentators have questioned the reliability of the measure. They have argued that the curve had lost its predictive power, as quantitative easing (QE) has depressed long-term bond yields and the term-premium artificially. The term premium can be viewed as the extra compensation that investors need to hold a long-term bond compared to a shorter-dated one. Without the impact from QE, the curve should have been steeper, so the recessionary signal may have been misleading.

The other case against the yield curve is the lack of obvious triggers to tip the US into recession as both corporate and household balance sheets are in a better place compared to previous cycles. In the run up to the global financial crisis (GFC), the US economy became vulnerable due to excess leverage that had built up in the financial system, particularly household mortgage debt.

Corporate debt had also ballooned to double-digit figures. In comparison, current credit growth is around 5% per year, which is rather modest by historical standards given the length of the US expansion. Overall, in this cycle, there are fewer imbalances in the economy and inflation is relatively low. 

Source: Refinitiv Datastream, Schroders Economics Group, 13 April 2020

Finally, recessions are unlikely to start during US presidential election year, as there tends to be pre-election stimulus by the incumbent administration. Since the 1950s, there have only been two recessions that began in election year – John F. Kennedy in 1960 and Ronald Regan in 1980. Instead, recessions are more likely to occur post elections and during the first year of the newly elected president when this pre-election spending is reversed.

Curve inversion occurs when the economy is late cycle

In May 2019, when the arguments above were made, they had merit in explaining why investors should not be concerned about a recession. However, the inversion of the yield curve highlighted that the economy was at the late stage of the cycle. In the past, the inversion of the yield curve (the US 10-year Treasury bond yield versus 3-month Treasury bill rate) typically occurred when the economic cycle was around 80% complete (table below).

The cycle length is measured from one recession to another, where US recessions are defined by the National Bureau of Economic Research (NBER). We also did the same analysis for the US 10-year versus 2-year Treasury bond curve and found similar results.  

With the US experiencing the longest period of expansion since the 1800s, the economy was not going to grow forever (chart below). Although this cycle has been one of the weakest, with growth sub-par compared to previous expansions, this is to some extent due to less borrowing by the household and corporates, as highlighted earlier. Meanwhile, the length of this expansion has been partly thanks to the Fed keeping the cycle journey going with loose monetary policy – it took nearly 10 years before interest rates went up after the GFC.

Since the end of 2015, the Fed has increased interest rates nine times, four of which occurred in 2018. While consumer spending has been the key support of growth over the past few years, higher interest rates have put a squeeze on households’ income. US consumers have become more bearish about the future compared to present and this measure has a close relationship with the yield curve (chart below).

Meanwhile, the other catalyst for an inversion of the curve is when the growth is vulnerable to a shift in market sentiment or a geopolitical event given the late nature of the cycle. Last year, the cycle was made more unsteady by US-China trade tensions, which led to greater uncertainty and weighed on consumer and corporate sentiment with the latter holding back on capital spending intentions. Growth stalled, although not enough to trigger a recession, unlike the emergence of Covid-19.

While there is an element of luck, and the coronavirus may have caused a recession at an earlier stage of the cycle, the inverted yield curve highlighted the vulnerability of the “wobbly bicycle” economy to a shock. In a year’s time, when civilian life hopefully returns to normal and the US has gone through a recession, the yield curve will have kept its recession predicting track record.