3 ms·

The Treasury yield curve has been inverted -- that is, long-term bond rates are lower than short-term -- for the last few months, which has traditionally (n=8)

The Treasury yield curve has been inverted -- that is, long-term bond rates are lower than short-term -- for the last few months, which has traditionally (n=8) been a reliable indicator of an upcoming recession. The market mechanism leading to this signal isn't well-understood, but is certainly related to inflation expectations. The good news here is that the market expects that long-run inflation will be brought under control (otherwise, the 10-year t-bill rate would be much higher to compensate for expected inflation losses), but monetary policy to do so will impact economic growth. The countervailing element here is that the economy is slowing down from a very fast clip: unemployment is historically low, job losses are pretty subsector-specific and tech employees are quickly finding new jobs elsewhere in the economy, the financial sector is healthy, and consumers aren't seeing the kind of destruction of wealth we saw back in '08. As a result, I'm still cautiously optimistic -- for whatever my opinion's worth -- that the Fed can manage a soft landing, though the ambiguous nature of current economic signals means that monetary over- or under-shooting is quite possible, leading to either a hard recessionary landing or continued inflation (noting that cost-push inflation due to factors such as COVID and the Russian war in Ukraine are not really manageable through monetary mechanisms, so hang-on inflation is likely to continue to eat away at economic performance).

dragonwriter 4y ago
> which has traditionally (n=8) been a reliable indicator of an upcoming recession. No, yield curve inversions have preceded each of the last 8 recessions (9 if you count 20-year to 3-month inversion), they've also occurred multiple times in that period without recessions before the inversion resolved, so they are not a reliable indicator of recessions (their absence is, by that history, a reliable indicator of not-recession, though).
- HillRat 4y ago
  Yeah, I think it depends on whether you consider it a prediction if the recession occurs during the inversion (Harvey's thesis, IIRC, considers it a prediction if a recession occurs within 1-4 quarters after the inversion, regardless of whether the inversion is maintained into the recession itself. Plus, I think his approach was less predictive prior to the 1976+ time period, though that might have been a data noise artifact.) If you accept Harvey's definition, then the major misses were the minor 10-2 inversion in the summer of 1998 (which prompted a quick round of Fed cuts that may have avoided a mild recession); the persistent inversion of 2000, which correctly predicted an economic slowdown but not a recession (in the US, anyway); the February 2006 inversion, which was in response to Fed attempts to cool down the housing market, which bit us a few years later; and the summer 2019 inversion, which I think was a true miss, though COVID pretty much upended any "normal" economic cycle, so it's possible that absent pandemic-era stimulus we might have had a mild recession. (Src: https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvs https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvs and https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvv https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvv) As you say, it's an imperfect metric by any light, even though I buy into the less-strict traditional prediction model, so take it a bit more seriously when the lights start flashing. Regardless, for what my opinion's worth (which ain't much), I think this is looking more like '98 at best and '01 at worst, at least as long as the Fed doesn't overshoot and slam us into the tarmac.
Rebelgecko 4y ago
Hasn't that happened multiple times over the last ~5 years? Or are there different definitions of long vs short term bonds that are more accurate (eg you can compare 30 yr vs 1 month, or you can compare 10 year vs 1 year)

dragonwriter 4y ago

> which has traditionally (n=8) been a reliable indicator of an upcoming recession. No, yield curve inversions have preceded each of the last 8 recessions (9 if you count 20-year to 3-month inversion), they've also occurred multiple times in that period without recessions before the inversion resolved, so they are not a reliable indicator of recessions (their absence is, by that history, a reliable indicator of not-recession, though).

HillRat 4y ago
Yeah, I think it depends on whether you consider it a prediction if the recession occurs during the inversion (Harvey's thesis, IIRC, considers it a prediction if a recession occurs within 1-4 quarters after the inversion, regardless of whether the inversion is maintained into the recession itself. Plus, I think his approach was less predictive prior to the 1976+ time period, though that might have been a data noise artifact.) If you accept Harvey's definition, then the major misses were the minor 10-2 inversion in the summer of 1998 (which prompted a quick round of Fed cuts that may have avoided a mild recession); the persistent inversion of 2000, which correctly predicted an economic slowdown but not a recession (in the US, anyway); the February 2006 inversion, which was in response to Fed attempts to cool down the housing market, which bit us a few years later; and the summer 2019 inversion, which I think was a true miss, though COVID pretty much upended any "normal" economic cycle, so it's possible that absent pandemic-era stimulus we might have had a mild recession. (Src: https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvs https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvs and https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvv https://fred.stlouisfed.org/graph/fredgraph.png?g=YAvv) As you say, it's an imperfect metric by any light, even though I buy into the less-strict traditional prediction model, so take it a bit more seriously when the lights start flashing. Regardless, for what my opinion's worth (which ain't much), I think this is looking more like '98 at best and '01 at worst, at least as long as the Fed doesn't overshoot and slam us into the tarmac.

Rebelgecko 4y ago

Hasn't that happened multiple times over the last ~5 years? Or are there different definitions of long vs short term bonds that are more accurate (eg you can compare 30 yr vs 1 month, or you can compare 10 year vs 1 year)