Why is Immaculate Disinflation seen as win for #TeamTransitory?

Here need to look conditional (not just unconditional) predictions.

Big disinflation with no unemp hike is signature of supply healing. And supply healing was the key point of TT.

h/t @mtkonczal Yes, rates rose a lot. Rate hikes could have led to:
1) reduction in inflation through inc in unemp lowering demand (main channel proposed)
2) increased unemp and little change in inflation (if supply stayed snarled).

Thankfully 2 didn't happen. Importantly, neither did 1.

Measuring wage trends during downturns is hard as low-wage workers are more likely to get laid off. Esp true during Covid.

This is why comparing wages from 2021 to 2023 is misleading. Has caused confusion lately.

Here's the evidence you need to see this clearly.

🧵

1/ Background: we see clear rise in real (inflation adjusted wages) for most American workers that today have not only exceeded inflation, but are about where they'd be based on pre-pandemic trends.

2/

The simplest point Dems can make:

Yes inflation has been painful. Both here and throughout the world.

But here's what we did: we kept companies hiring, jobs growing, raising pay. This is why U.S. workers have seen wages rise more than prices, more so than many other countries. Voters may not link the tight labor market to rising wages, but it's a fact.

We show this in our paper, where tightness strongly predicts wage growth especially for bottom half of pay scale.

Even after accounting for any impact on inflation.

https://twitter.com/arindube/status/1724147820788355124

This great 1997 article by Bob Shiller has survey evidence on to why people dislike inflation: they don't believe wages would be very different if inflation were lower.

They think inflation makes them worse off from higher prices, as wages would be the same w/o inflation.
🧵 Key point:

In a better world, the American left would be defending the strong pro-full-employment policies pursued by Biden administration, and argue back against the specious claim much of the rise in inflation were caused by those.

Esp since it helped raise wages and lower inequality. Instead, we have TikTok takes on the Silent Depression.

The political reality is depressing, indeed.

But it's, unfortunately, not silent.

An update to work with @davidautor & @AnnieMcGrew1.

The Unexpected Compression in wages has persisted, even as labor market tightness subsided.
Tl;dr ~40% of the rise in wage gap between 10th & 90th percentile in 1980-2019 was reversed in past 3 years. 🧵
nber.org/system/files/w…
Wage inequality fell over the past 3 years. There was a major compression in 2021-22.

2/

Mr. Dube and two co-authors found that...in just two years, the economy undid about a quarter of the increase in inequality since 1980. Much of that progress, they found, came from workers’ increased ability — and willingness — to change jobs.
nytimes.com/2023/07/06/bus… Pay is no longer rising faster for low-wage workers than for other groups. But importantly in Mr. Dube’s view, low-wage workers have not lost ground over the past two years, making wage gains that more or less keep up with inflation and higher earners.

🚨 Difference-in-differences working paper alert 🚨

Our Local-Projections DiD offers a unified approach that encompasses many popular alternatives as specific instances; allows for extensions; and does it all using an OLS regression.

nber.org/papers/w31184

🧵 Why yet another DiD estimator?
* Simple, fast, easy to code (single regression command with an "if" statement)
* Flexible (eg, recurring treatments)
* Easily allows matching on pre-treatment outcomes and covariates
2/

Some great grad macro teaching resources and slides here from @JonSteinsson and Emi Nakamura

https://twitter.com/JonSteinsson/status/1613580005329764353

These slides on "Identification in Macro" from Emi and Jon are fantastic. eml.berkeley.edu/~jsteinsson/pa…

I think some people are getting really confused about what it means when there is inflation and real wage declines. Textbook models of wage and price Phillips curve tells us that if price growth>nominal wage growth, it's unlikely from labor market tightness. This comes right off the 1999 analysis by @lkatz42 and Krueger in their BPEA. Take the difference between equations 4 and 5: that's the real wage growth. Tightness will raise nominal wage and price similarly. If price growth > nominal wage growth, gap likely from supply shocks.

I'm pretty sure that if an overly tight labor market were the primary cause of the current (too high) an inflation rate, it would require a simpler diagram.

The reality is that real wages are rising only at the bottom, workers who contribute a very small amount to labor costs Link: crfb.org/papers/fiscal-…

The Russian regime might be the most costly institutional failure afflicting the world today. Though there are (sadly) other strong competitors. Pro-democratic societies will have to (and are being forced to) make tough and costly choices.

But the costs of wavering far outweigh the costs of weak complacency, IMHO.

Just saw that synthetic DID R-code (by Arkhangelsky, @Susan_Athey, et al.) is available on Github now.

Really simple but powerful idea: doing weighted DID w/ synth weights (+regularization) can be doubly robust and reduce bias of each approach.

github.com/synth-inferenc… Paper: arxiv.org/abs/1812.09970

Binned scatterplots are great.

"binscatter" in Stata is very helpful to assess how the partial linear fit in OLS is being estimated.

However, Stata binscatter does not provide a binned version of the partial linear regression. For that, use binsreg. Namely, if you want partial effect of X on Y, controlling for X', you should use binsreg. The realignment of binscatter (based on partialing under FWL) does not correctly align the impact on *unconditional* X (after linear controls) which is what a partially linear model does.

From today's class, based on Sorkin 2018.

Yes, mining pays a lot more to otherwise similar workers (AKM firm effects, Y axis)

But based on how many workers choose to move to mining, mining isn't a "good job" (X axis). Rather, it's compensating differential for job risk, etc Here's Sorkin's 2018 QJE paper: …a-a-62cb3a1a-s-sites.googlegroups.com/site/isaacsork…