The inversion of the yield curve took another unusual step on Wednesday. It moved into double inversion.

The traditional yield curve inversion occurs when the yield on 2-year Treasuries falls below the yield on 10-year Treasuries, overturning the typical rule by which longer-dated government bonds have higher yields than shorted-dated bonds. This is thought to indicate that investors see the central bank cutting rates sometime over the next 12 months, which typically means a recession is looming.

The 10-year to 2-year curve has been inverted since early July. When the yield curve first inverted, those who thought a recession could be avoided pointed out that inversions only predict recessions when they stick around for several months. A momentary inversion was not predictive. Yet the yield curve not only stayed inverted for months and was the first serious inversion in 15 years, it also kept getting deeper.

Another line of criticism of those worried about the 10-year to 2-year inversion is that the more reliable indicator is really the 10-year to 3-month curve. Back in July, when the 10 to 2-year curve inverted, this part of the curve was 92 basis points into positive territory. There’s actually good academic work suggesting that this is the curve to watch when trying to figure out if bond yields are indicating a recession is in the offing. Economist Arturo Estrella has shown that since the late 1960s, this part of the curve has inverted an average of roughly a year before the start of a recession, although the start can range from six to 15 months.

Unfortunately, the 10-year to 3-month curve inverted on October 25. This too was not a momentary phenomenon. The curve inverted and stayed inverted. And, like the 10 to 2-year curve, it kept on getting more inverted.

Now something even more dramatic has happened: the 10-year to 3-month curve is more inverted than the 10 to 2-year. On the cart below, you can see that typically when the curve is uninverted, the distance between the 10-year and 3-month yields is greater than the difference between the 2-year and 10-year yield, indicated in the chart as the blue line being higher than the red line. Just before recessions, however, this relationship tends to invert: the 3-month yield curve goes more negative than the 2-year.

That is what happened in December. The inversion of the 3-month to 10-year yields is now greater than the inversion from 2-years. As of Wednesday, 10-year Treasuries were yielding 3.458 percent, and the 3-month Treasuries were yielding 4.325 percent, an inversion of 86.7 basis points. The 2-year yield was at 4.283 percent, an inversion of 0.825 percent.

Recall that an inversion more or less implies that the market believes that the Fed will pivot to lowering rates again, which would likely mean a recession has arrived. This double-inversion likely indicates that this is coming sooner rather than later. As we argued yesterday, this may be a mistake. But it appears to be what the yield curves are implying.