Thread
A very wonkish and probably too long thread on inflation. Not predictions, but to talk about my sense that we (myself included) have been working with the wrong model given the current situation 1/
Most policy-oriented macro ppl entered the pandemic using what we might call a core inflation Phillips curve — inflation depending on slack, but with volatile stuff carved out. And we're still trying to fit events into that framework 2/
But it feels more and more like Ptolemaic astronomy, relying on a growing number of epicycles to deal with the anomalies. Two examples of what I mean: 3/
Here's what happened to a wage Phillips curve using ECI wages and salaries. The line is the fitted slope 2000-2020Q1; the red dots 2021-2022 (skipping the height of the pandemic disruption) 4/ 
The deviations could reflect a surge in expected inflation — but there's no evidence of that happening. Or it could reflect a mysterious deterioration in the Beveridge curve. As I said, epicycles 5/
Another anomaly: huge but temporary inflation surges in sectors usually included in core inflation, notably housing rents 6/ 
So what if an aggregative Phillips curve is all wrong for the pandemic era? I was trying to think through a model, and realized that it had already been largely written 7/ bfi.uchicago.edu/wp-content/uploads/2021/09/BFI_WP_2021-111.pdf 
Guerrieri et al basically assume a big shift in the composition of demand, plus sluggish reallocation and downward wage and maybe price rigidity. They worried a lot about optimal policy, but just focus on description 8/
What a model along these lines would predict is a jump in overall prices, with very high but temporary inflation in some sectors. Also a lot of churn as labor reallocated to expanding sectors 9/
That's pretty much what has happened — after they wrote this, so it counts as a successful prediction. And it suggests that the worsened unemployment-inflation tradeoff won't persist 10/
Do we know that this is right? No. But rethinking along these lines looks to me like a better approach than trying to keep adjusting the core inflation Phillips curve to fit awkward data 11/
