The yield curve is predicting much worse than a recession.

Deep Inversion

Here’s the yield curve in what we may call the forward-rates representation. Each of the five graphs in the chart are spreads along maturity — the number of years until the payment on a zero-coupon bond. The first is the yield on the 6-month (mo) bill less the yield on the 3-mo bill (anchored at the Fed’s current policy rate). The second is 1 year less 6-mo, and so on and so forth. All the information contained in the yield curve out to ten years is contained in these forward rates. Bit crowded, right? Let’s break it down by horizon to better understand what exactly the yield curve is saying.

Our horizons will be very short-term (6mo less 3mo), short-term (from 6mo out to 2 years), medium-term (5 less 2), and long-term (10 less 5). We expect rate expectations to be absolutely dominant relative to the duration risk premium at the very short-term. Conversely, we expect the duration risk premium component to dominate at the far end. Intuitively, duration risk increases with distance along the yield curve.

Astonishingly, the 6mo-3mo spread shows that the market expects the Fed to already start cutting rates in 2022Q4. On July 1, the spread had reached a peak of 79 basis points (bp). It has now fallen, as of market close on Friday, to 53bp — a full quarter-point hike has been priced out in the past two weeks.

This is astonishing. The disagreement with the Fed’s dot plot diplomacy could not be more obvious. The market has essentially rejected the fed’s model of the world. The Fed is wrong, it is saying, after driving the economy into recession, you’ll be cutting in a few months. Again, no one doubts that the duration risk premium or any other bond risk premia confound the rate expectations embedded in the 6mo-3mo spread. This is why the short end — not the level but the fwd rate — is so informative about the expected direction of the policy rate.

Let’s step forward along the yield curve into short-term expectations. Henceforth, we’ll have to think symmetrically about rate expectations and risk premia. If one wants to measure the former, the signal is expectations and the noise is risk premia. And vice-versa, if one wants to measure the duration risk premium. You can tell they are confounded by risk premia because they fall below the Fed’s zero lower-bound. In order for the short-term spread to invert, you cannot have expected policy rate very far from zero. So, the yield curve is now saying, rates will be close to the zero lower-bound in 2023.

The signal for expectations of the policy rate in medium and long-term forward rates are definitely confounded by risk premia. Our medium-term forward rate is the 2y5y term spread. Like the shorter spreads, it predicted the “covid recession.” It reached 63bp in December last year, crashed in Q1 this year, before falling to zero on the first trading day after the Fourth of July. On Friday, it was -9bp. The yield curve is pricing a prolonged period of poor expected returns on duration risk assets.

Our long-term forward rate is the 5y10y term spread. (Economists say “5y5y breakeven rates” instead of 5y10y, the “correct” name in the forward rates representation.) Noticeably, the 5y10y peaked much earlier, in June 2021, when risk assets rocketed skywards. Recall that AQR’s real expected return on US equities is 3.6 percent per annum over this horizon. Long-term return expectations on risk assets are now so poor that you can get negative rates on bonds (guaranteed poor expected returns) even on the 10 to 5 year horizon. But not only. Note that the 5y10y did not invert when the short end was pricing recession in 2019. Clearly, the market now reckons that long-term prospects have deteriorated in a way it did not in 2019. Note also that this gradual repricing began as the same time as risk premia began collapsing in the premier risk asset — blue-chip US equities.

The inversion of the yield curve predicts recessions. But the information contained in 5y10y is subtly distinct from that contained in the 2y5y. The 2y5y contains more information about recessions and economic-activity in general, in the near-term “cyclical” future. The 5y10y is less controlled by the cyclical future and is therefore freer to encode information on longer-term expectations. So, we’re seeing a “double inversion,” with both 5y10y and 2y5y inverted.

In sum, the yield curve is saying: a recession is now guaranteed; the Fed will be "cutting by ‘23,” exactly as Zoltan had wagered; and, just in case you thought there was a silver lining, longer-term prospects have dimmed to boot.

Talk about a vibe shift.