Was inflation A Series of Unfortunate Events or Original Sin?
My discussion of the Bernanke-@ojblanchard1 paper at Hutchins @BrookingsEcon.
Short version: their model/results are agnostic. My view is *core* inflation is mostly original sin.
A 🧵
The authors are clear about what their model/results do and do not find and Bernanke reiterated it in his response to my comments. But a lot of commenters have misunderstood it.
I hope this slide (added post-conference) clarifies what their model/estimates do & don't say.
You can find the Bernanke-@ojblanchard1 (henceforth BB) paper and the other papers/slides/discussion from "The Fed: Lessons learned from the past three years" conference here. All of it is excellent so if you have the time try to watch or read.
BB develop a simple & elegant four equation model:
i. price setting based on wages and productivity
ii. wage setting based on expected prices, desired real wages & slack (measured by V/U)
iii & iv. short- and long-run expectations, a function of actual and expected inflation.
They estimate their model (plus lags) to generate this decomposition of headline CPI inflation.
The striking result is that labor market tightness (measured by v/u) played ~zero role in 2021 and a small one today.
Instead was energy, food, "shortages" & an unexplained residual.
(My comments take their results as given & discuss interpretation. Is also worth kicking the tires on how robust the results are to alternative specifications etc. I have some quibbles/questions but are for another time.)
It turns out the food and energy shocks they estimate work out to be almost exactly the same as the food and energy contributions to the overall CPI. Which is to say, their empirical breakdown is consistent with essentially no passthrough to core.
(And BTW, the paper includes a principal components analysis of commodity prices, finds most are consistent w/ demand increases, so BB largely interpret these shocks more as demand/policy than "unfortunate events." I'm closer to the "unfortunate events" interpretation.)
BB don't do a breakdown for core inflation but here is my *rough* attempt to map their results into core inflation, showing the excess above 2.3% inflation.
The biggest overage in 2021 was "shortages" and it was large in 2022 & meaningful in 2023-Q1 as well.
So let's talk about "shortages," which are based on Google search frequency.
Is, of course, endogenous. Could go up because demand rises due to macro stimulus. Or demand rises because of an exogenous shock like COVID. Or supply chains worsen.
Thus the model/results agnostic.
The model/results themselves do not answer the question. You have to use a collage of evidence from outside the model/results to decide whether the shortages are more the result of something exogenous and unpredictable (COVID "taste shock" or COVID-induced supply problem) or not.
The authors lean into but do not completely endorse an interpretation that have some have called the Peloton Economy thesis: (1) COVID caused spending to shift to goods and (2) even general equilibrium reduced services spending that supply elastic so inflation net up.
On the first, I don't see COVID as exogenously causing a shift from services to goods in 2021 (the year inflation emerged). In fact, the big increases in goods spending followed the stimulus checks. & they happened *while* the economy was rapidly reopening & services were rising.
Moreover other countries took longer to vaccinate, longer to reopen, and had longer-lasting lockdowns in 2021. If goods spending was exogenously caused by COVID they should have had bigger increases in goods spending. Instead they were mostly flat or down.
So broadly speaking, the evidence is mostly consistent with durable goods spending going up in 2021 because people had more money to spend--and possibly even spent that money disproportionately on durables as the evidence suggests they did in 2008.
Also not sure about the 2nd part of the argument. If goods more expensive people buy less services. If goods inelastic and services elastic (left picture) then inflation up. But a lot of services were inelastic (right picture) so would just affect relative prices not price level.
So where does that leave us?
Their models/result finds no support for the pessimistic argument that stimulus would overtighten labor markets and unleash inflation...
...But it is consistent with the pessimistic argument many were making that stimulus would create too much nominal demand, that real demand could only increase so much, and so inflation was inevitable.
Note this is the precisely the argument/analysis one would use for Turkish inflation or a stimulus plan that gave everyone $1m checks. You wouldn't expect a linear labor market model to pick up all the inflation that would ensue.
Identities don't prove anything. But add a little structure and they can help. Real GDP growth was as fast as could be reasonably expected post-pandemic. But the economy was awash in support for nominal spending--which manifested in higher prices.
Sorry, this was very long. But you can watch the video, read my slides, or other parts of the excellent conference here.


i. 賃金と生産性に基づく価格設定
ii. 予想物価、望まれる実質賃金、およびスラック(V/Uで測定)に基づく賃金設定
iii & iv. 実際と予想インフレの関数である短期と長期の予想




著者たちは、ペロトン経済命題とある人が呼んだ解釈に傾いてはいるが、完全に是認してはいない。その命題とは、(1) COVIDにより支出は財にシフトした、および、(2) 一般均衡においてさえ供給弾力的なサービス支出が減るため、インフレはネットで上昇した。








*1:cf. About the Center

*2:ここで挿入されている図の元ネタの映画はそれぞれ「レモニー・スニケットの世にも不幸せな物語 - Wikipedia」と「映画 キング・マフィア/偽りの報酬<未> (1989)について 映画データベース - allcinema」。ファーマンはそれぞれの映画の原題に寄せて、ワクチンの効果があった/なかったこと、半導体不足、港湾の混雑、ロシアのウクライナ侵攻を一連の不幸な出来事、金融財政政策を原罪に準えている。

*3:スライドでは、100万ドルを各家計に配ったら、GDPの514%の財政刺激策になるので、線形モデルでは、乗数を0.8として実質GDPは412%上昇し、失業率は0%に低下し、0.15の傾きのフィリップス曲線を使えばインフレ率は2.6%に上昇する、という極端な試算例を示している。その上で、実際には、実質GDPは少し上昇し、大インフレが生じる明確な「品不足」ショックが生じることになる、と指摘している。またスライドの結論部では、「It may be more fruitful to ignore the labor market in assessing large non-linear shocks.」とも書いている。