Follow-on CPI Data Flags a Bit More Caution
Today’s CPI report was obviously a bit of a mixed bag with some volatility that reads a bit dovish when extrapolated out but also had a number of somewhat more hawkish components. On net, I read it as a bit hawkish but not so obviously so as to definitively change any of my qualitative views on the economy or the near-term trajectory of Fed policy.
As happens later in the morning, the trimmed mean and median measures came out as did the Atlanta Fed’s sticky and flexible prices measures. While the trimmed mean and median measures deserve a lower weight than usual given their tendency to cluster towards CPI shelter inflation (given its large weight and fairly low m/m vol) they don’t’ deserve 0 weight because Fed officials still reference them with some regularity. The Atlanta Fed’s sticky prices ex shelter presents a similar concept as the more popular core services ex housing but does so in a more useful way in my opinion.[1]
When looking at these 3 measures, as well as core services ex housing and core ex shelter and used autos, one somewhat troubling pattern looks to be common across all of them: it looks like their higher frequency growth rates all found a bottom in summer 2023 and have since perked up, to varying extents. This lines up reasonably well with the bottom and trough seen in the ISM pricing measures. Some of this could be lagged-inflation dependent residual seasonality[2] but that doesn’t go all the way in explain it in my view and fails to account for the late-23 upticks seen in these CPI measures.
Inflation is likely still on a broad, if quite bouncy, downwards trend but any shifts to underlying inflation dynamics and the very long-lived effects of the covid- and Ukraine invasion-related shocks makes assessing its medium-term dynamics very challenging. This is why the Fed is giving us macro forecasts and the dot plot as a base case and guide to the reaction function, but telling us they are primarily data dependent. I don’t think the recent upturn in some of these measures of underlying inflation is enough to break the Fed’s momentum towards cuts but this data, along with the broader outlook for cyclical outperformance relative to the Fed’s December forecasts, puts the skew towards fewer cuts, relative to a 3 cut 2024 baseline, rather than more.
Currently, the Fed is more concerned about inflation than growth, but their default next move has been signaled as a cut. Hawkish (inflation) or less dovish than expected (growth and labor) data seems to be more about pushing cuts farther out than anything else. This pattern of falsifying dovish moves rather than pricing outright hawkish policy will likely to continue until the Fed cuts enough for the default next policy move to become more symmetrically balanced between hikes and cuts (obviously a recession or a clear and persistent reacceleration in the underlying inflation data would change this too).[3]
Further out in the forecast, these signs of somewhat stickier inflation, along with my views on the neutral rate, the current and expected financial conditions impulse, and probable growth outperformance, suggest that the Fed is apt to underdeliver on cuts. With my own view of nominal neutral centered around 3.5%, I expect, as an admittedly quite tentative and imprecise baseline, the Fed will ultimately deliver around 6 cuts before the next recession hits, with perhaps 8 cumulative cuts a reasonable non-recessionary floor.
I’m generally of the view that ex ante categorical exclusions make less sense than do volatility or cyclicality-based ones. Here other central banks, especially the BoC and RBA, are more in the lead than the Fed. ↑
The frequent references to residual seasonality seem to explain it away too flippantly; the issue isn’t just a January effect but rather a Jan-Feb pricing reset dependent upon the lagged level of inflation. This positive feedback loop is different than a pure residual seasonality effect and deserves partial but not full down-weighting, even if it was entirely responsible for all the inflation bump seen in Jan and Feb. ↑
This same dynamic is what keeps medium-term option-implied distribution negatively skewed below the modal outcomes. I doubt that STIR distributions can be symmetric for <3y forward rates until roughly 4 cuts have taken place, then reacceleration and hawkish risks might be able to balance out the downside recessionary tail. ↑
