Why is an inverted yield curve NOT a good predictor for recessions except in the US?

Question

I am trying to better understand the mechanism (if there is one) behind an inverted yield curve predicting - or causing, as some would argue - a recession. In order to do that, I'm trying to understand why an inverted yield curve has historically NOT been a good predictor in countries other than the US.

See e.g. this article by AQR, which discusses the track record of yield curve inversions in Australia, Japan and Germany. The authors qualitatively argue why that track record hasn't been great but don't provide much of a general framework to think about it. I suspect that they are on to something in the case of Japan: "It is possible to argue that the banking system in Japan is so broken that it doesn’t matter how steep the yield curve is, banks won’t lend anyway. But, more likely, the relationship has broken down because interest rates have been stuck at or near zero for so long. The yield curve can no longer tell you whether the Bank of Japan has raised rates too much or not, because it hasn’t raised rates at all." I agree that the banking system looks pretty zombified, which would explain why banks (and the economy in general) don't really react to cues from interest rates. In addition, the Bank of Japan has such an outsize weight in the market for government bonds these days that the true information content of interest rates is probably quite low. However, I don't think either of these characterizations would apply to Australia, so clearly a broader line of thinking is necessary.

Academic references and/or plain English welcome. Thanks.

Answer 1

I’m currently researching this topic, and I have not found any single reference that answers this (... besides my work). I will summarise what I have seen.

I would point out that from a research perspective, having the yield curve work is interesting, while a failure to predict recessions is not interesting. Very few journals publish articles on failed indicators, since the number of those is very large.

If we ignore term premia, an inverted curve normally indicates that the bond market is pricing a cut in interest rates. The way central banks react is that they normally cut rates when recessions hit. As such, we would typically expect inversions ahead of recessions. However, the logic can break down.

• Bond market participants are wrong about the forecast.
• The central bank cuts rates, but no recession. (U.S., 1998).
• Proximity to 0% makes inversion difficult (Japan, euro area).
• Central bank does not cut rates in response to recession (possible for euro periphery).
• “Technical factors” in bond market lead to flat/inverted curves (U.K., Australia.) For the U.K., pensions were forced to match the liability of duration, and there was not enough supply of long-dated bonds. For Australia, the government had surpluses, and I believe that duration supply was extremely low. (Would need access to paid bond index data to validate that claim.)

The issue is therefore not U.S. versus non-U.S., it’s just that the U.S. managed to avoid most of these factors (except in 1998 in the recent history).