Over recent months, we have been flagging the rising risks of a US recession, even if we still think a soft landing is most likely.

Below we outline why the recent changes in the shape of the yield curve haven’t shifted this view, but do highlight the need to be vigilant as we enter a dangerous phase for the business cycle

The yield curve has predicted every recession in the past 70 years

The yield curve for government bonds sets out the benchmark interest rate across different maturities in the economy. These interest rates are determined by the expected path for short-term real interest rates, inflation and term premia, which is the compensation received for the uncertainty of holding a longer maturity asset, for example, a 10-year government bond.

In normal times, the yield curve slopes upward. The shape of the curve contains information around the expected outlook for growth, inflation and central bank policy. And this signal comes under extra scrutiny when the curve inverts.

Historically, yield-curve inversions have tended to precede a downturn, albeit over different time horizons. An inversion often implies that the central bank will have to cut interest rates in the near future, likely on account of a recession. This means that long-term average interest rates will be lower than those in the shorter term.

In the post-World War II era, the one-10 year yield curve, for which we have the longest data available, has inverted 10 times. In nine of these, a recession occurred in the near future. The only exception was the mid-1960s. This suggests that yield-curve inversions have a strong track record of predicting recessions.

It’s worth highlighting that the most recent yield-curve inversion in late 2018 was followed by the Covid downturn in early 2020. Of course, the bond market wasn’t predicting a pandemic, but we’ll never know with certainty if the economy would have entered recession that year. Few were forecasting a downturn, though.

But in light of this historical pattern, the inversion of the US yield curve during the last week of March has put the question of impending recession under greater scrutiny.

But has the quality of this alarm deteriorated?

Judging signals from the yield curve has become more complicated in recent years. Central bank asset purchases and strong global demand for government bonds distort some of the messages that could be derived from the interest-rate structure.

The US Federal Reserve (Fed) holds about a quarter of outstanding US Treasuries. Other central banks hold even larger shares of local government debt. Meanwhile, concerns over low growth in recent years have supported private-sector demand for relatively safe assets, such as government bonds.

These forces suppress term premia. According to the Federal Reserve Bank of New York, term premia turned negative in the late 2010s and currently sit around -0.40%. This implies that investors were paying a premium to hold longer-dated government bonds, above the anticipated real return, given expected short-term interest rates and inflation.

These distortions mean that it’s harder to interpret what the yield curve is telling us. So we may not be able to conclude definitively that this most recent yield-curve inversion portends recession.

Consider that in 2006, when the yield curve inverted ahead of the 2007 recession that ushered in the Global Financial Crisis (GFC), term premia averaged around 0.50%. The 0.90% difference in term premia between then and now has a significant effect on the shape of the yield curve and, if not accounted for, could give misleading signals around the outlook for rates.

And the real yield curve is telling us a different story

When inflation expectations are broadly consistent across various maturities, inflation over the next few years is expected to be about the same as the average inflation rate over much longer periods. In this scenario, the shape of nominal and real rates curves will be similar. This means that there’s typically little to learn from looking at the real curve over and above the nominal one.

But, in the current environment, short-term inflation expectations are high, but they’re likely to moderate toward consistent levels over a longer period. So, now, the distinction between nominal and real yield curves matters much more.

Indeed, the real curve is still steeply upward sloping. This implies that the market expects the true stance of policy – the real rate – to have to increase in the future.

An inverted curve is typically interpreted as a recession signal because it suggests that the market thinks monetary policy will need to ease in the future, presumably to fight some expected downturn. But this is precisely the opposite signal we get from the very steep real curve. This again suggests we should put somewhat less weight than usual from the signal from an inverted nominal curve.

Regardless, the interest-rate structure is warning that recession risks are rising

All of this doesn’t mean that we should disregard the yield curve. Rather, we should treat the assumption that a nominal yield curve inversion guarantees recession in the near future with caution.

We can use our recession probability models to assess what the traditional yield curve, a yield curve adjusted for the decline in term premia and the real yield curve are telling us about recession risks. The different degrees of flattening across these yield curves show quite different recession risks (Chart 1).

Chart 1: Recession probability within 12 months

Source: Haver, FRED, abrdn, as of March 2022. Note that the real yield curve uses the 5s-10s spread instead of 1s-10s because of the volatility in short-term inflation expectations.

According to this chart, the nominal yield curve suggests the highest probability of recession in the next 12 months — 26%. The real yield curve also signals recession risks in the coming year rising, albeit to a lesser extent — about 17%.

Meanwhile, the term-premia adjusted yield curve, which strips out the effect of lower-term premia, continues to signal a relatively low risk of recession within the next year, at less than 10%.

In terms of timescales, our modeling suggests that the most accurate signal from the yield curve comes 12-14 months ahead of the beginning of a recession (Chart 2). In this modeling, we looked at the accuracy of recession models historically at different horizons. We considered, too, the trade-off between specificity and sensitivity of the models — this balances predicting recessions when they do occur versus being too sensitive and predicting recessions when there isn’t one.

Source: Haver, abrdn, as of March 2022. Projections are offered as opinion and aren’t reflective of potential performance. Projections are not guaranteed and actual events or results may differ materially.

Therefore, when considering recession risks based on the shape of the yield curve, both the measure you use and the time horizon you’re interested in matter. This analysis would suggest that while alarm bells aren’t ringing yet, we should be mindful of the scope for recession risk.

But US recession still isn’t our base case

The yield curve is a common harbinger of recession, but it’s just a single metric. Our preference is always to consider a wealth of data when considering end-cycle events because these can be driven by several factors. Without looking across the economy, you might miss something.

On this basis, our broader recession risk framework signals that overall recession risk is around 25% to end 2023. Therefore, we don’t yet see these risks as sufficiently high enough for us to make a US recession our base case. But we’ll need to monitor this closely, especially against the backdrop of a rapid adjustment in Fed policy and a sharp real income squeeze from higher energy prices.

Chart 3: Implied US recession risk by end 2023

Source: Haver, Bloomberg, abrdn, as of March 2022

US-060422-168955-1