Market power in US manufacturing: it's about wage markdowns, not price markups!
Firms with market power in product markets restrict all variable inputs symmetrically. Firms with market power in labor markets restrict labor inputs *more* than other variable inputs like materials.
Fun as can be, to dunk on Gen Z...
It looks like the US result is an artifact of a huge outlier in 2022, after the sampling methodology changed.
Here's what the young men time series looks like with 2022 dropped.
Code available here:
NEW: an ideological divide is emerging between young men and women in many countries around the world.
I think this one of the most important social trends unfolding today, and provides the answer to several puzzles.
Alice Wu was the first to quantify EJMR's toxicity *in her undergrad thesis.*
I haven't seen it mentioned at all today. Which feels... Weird?
I can't seem to find the words as to why. But I appreciate it so I'm tweeting it.
What role does structural change play in the fall of the US labor share? ๐งต 1/n
Ben Bridgman (
@BenBridgmanEcon
) and Berthold Herrendorf offer an illuminating framework built on Input-Output links in their ๐จ new draft ๐จ!
๐จ Public-firm profit rates have fallen to *half* their 1980 values ๐จ, which explains why financial market rates fell despite rising/steady *aggregate* profit rates.
New draft with
@ASollaci
&
@CarterDavisFin
, & outstanding support from
@TonBobrov
...๐งต
Super cool work on markups and economic profits!
Micro + Aggregation = Macro
Aggregation research is only going to get more important as we develop richer large-scale microdata.
(But let's keep all our micro-macro jokes ๐).
More on aggregate markups ๐
โThe profit share has been constant at 18% of GDP because the increase in monopoly rents has been completely offset by rising fixed costs and changes in technology.โ
By Thomas Hasenzagl and Luis Pรฉrez
Bruno Pellegrino (
@BPellegrino_UMD
): Hybrid IO-macroeconomic view: need to look across many different sectors. Network theory of market power- an economy is a network made up of nodes (products). Distribution of centrality in the network indicates concentration
@Caffar3Cristina
I donโt care if itโs real analysis in particular. But one has to be deeply incurious not to take a proof-based math course before playing professional football.
Saying "supply and demand" can explain any price movement. A parrot can do that.
Unfortunately, the alternatives are worse!
Ex: some inflation stories are "prices went up because people raised prices."
Now that's a theory that can explain everything!
Where we work is not where we buy!
๐จ New working paper ๐จ from the BEA develops *consumption* zones.
The overlap with commuting zones is like 50/50!
I'm excited to share a new working paper Consumption Zones, joint with three of my BEA colleagues.
We use credit card data to calculate local areas of consumption for services that require physical presence.
@AllenDowney
@janzilinsky
Fwiw, I tried all the weight variables to no avail.
It's something really weird about that specific data point.
We might all just be puzzling over a tail draw!
Does concentration = market power?
Quantitatively, how far away have we moved with richer models?
I see dynamics as key.
But lots of folks assume repeated static oligopoly, no entry threats etc.
Is variation in concentration still de facto driving our collective inference?
Metrics musings on two-way fixed effects (TWFE) --
I've been sleeping. Is this just Simpson's paradox? Can we just use multilevel modeling?
i.e. if you think the model is
y_it = b_i*x_it + ...
Use each panel to estimate b_i, then average?
@pedrohcgs
@jfiksel1
We really need to stop defining random effects as uncorrelated by default!
Most econ students learn this definition and move on thinking fixed effects are nearly always better.
Meanwhile...
Post-pandemic, firms aren't just raising prices.
They're strategically testing new pricing and customer capital strategies!
"We may have prices changing more quickly than they have before."
@_amackay
from
@HarvardHBS
NK economists take note!
Look what my good friend
@AliHortacsu
just brought to my office! Every economist should take a look. I know that I will learn a lot from it. Can't wait to dig in.
On a personal note, Ali has long been one of my favorite colleagues and people. Always supportive and generous.
This paper is above all *practical*:
It solves a common problem without just saying โget more dataโ or โadd more structure.โ
It's also highly educationalโI strongly recommend giving it a look, and adding it to your IO course lists!
It is sobering to recall that the shape of the past decadeโs inflation curve almost perfectly shadows its path from 1966 to 1976 before it accelerated in the late 1970s.
This was Berthold Herrendorf's final paper before his untimely passing.
His contributions to economics were invaluable.
I'm in awe of Ben Bridgman for continuing with the paper.
It's a wonderful way to honor Berthold's memory and legacy.
().
10/10
Good thread, and consistent with my favorite slide deck in all of economics (from
@PhilHaile
). Even "reduced form" estimating equations are (derived from) some theory! Anyway:
openai.Completion.create( model="text-davinci-003", prompt="Write an academic economics tweet about non-compete agreements that will get a lot of retweets.", temperature=0, max_tokens=4000, top_p=1, frequency_penalty=0, presence_penalty=0)
My most popular work in econometrics is with Stรฉphane Bonhomme. Very accessible, no proofs.
Nothing is robust. Nothing is identified. Bonhomme ร la Bresson
#EconTwitter
#metricstotheface
@PikettyWIL
Excellent thread!
In the spirit of open inquiry, since they don't really have a platform, if they write a reply, would y'all be up for tweeting it?
@Jabaluck
I parsed it a little differently: Should we ask questions that are only highly informative conditional on findings?
These questions distort researcher incentives. We're agents, too!
@gabriel_zucman
OK, trying to follow seriously.
In my first look at the Auten-Splinter work, it seems the biggest change is from measuring the tax base consistently through time.
But this thread doesn't mention that at all? No โsample,โ โmarriage,โ โdependent,โ or โresident.โ What's going on?
@wwwojtekk
Yep! I'd be surprised if Walmart behaved differently. Contrarian view is Walmart might have more local monopolies:
I'm surprised someone hasn't combined the two insights yet. Just check if average local concentration predicts firm pricing rules.
We propose an alternative hypothesis:
Aggregate profit rates track the return on capital for all firms, but financial market rates track the cost of capital for *public firms only*.
(Private equity literature suggests private firm returns have *not* fallen).
Just noticed a ๐จ new draft ๐จ was posted, an excellent opportunity to start my New Year's resolution to tweet about papers I really like.
Here's my sketch simplification, imprecise, but I hope accurateโฆ
@BrianCAlbrecht
Compustat concentration measures are iffy. Well-known in finance but maybe not in economics. See the canonical or more recently
@DrJanKeil
's
Our results indicate estimates of economic profits and aggregate markups *using public-firm data* might be substantially biased.
e.g., Compustat, or mixing national accounts with financial markets dataโฆ
In Opinion
โAmericans have been living as subjects in a large-scale experiment in letting big companies do as they please, and the consequences are increasingly apparent in daily life,โ the editorial board writes.
@ChadSyverson
on the macroeconomics of antitrust enforcement at the
@StiglerCenter
competition conference this summer---
๐จ๐๐งต
2 remarks at the end of the thread!
Two remarks:
A big rise in markups would imply an implausibly big fall in the aggregate labor share, cf. discussion in Barkai and Benzell (
@SBenzell
):
2. Might be labor markdowns, not product markups ๐()!
8/n
@Andrew___Baker
@momin_rayhan
@achenfinance
Here's another fun one from the accounting literature:
Compustat itself makes a lot of iffy standardization choices. And they're big enough to drive key results behind highly cited papers.
But at least we're all using it so that's good I think?
@BrianCAlbrecht
@HasvanVlokhoven
For a slight wrinkle, theory doesn't tell you which weight to use. The EMX result suggests cost weights, but that's for a model with homogenous production technology, which isn't how we estimate it in the microdata!
@BrianCAlbrecht
@ArshiaHashemi
@BasilHalperin
@luiscanyamel
@jamesbrandecon
Rents: Payments Without Production
Market Power: Markups, Markdowns, etc.*
*You can add qualifications here. For example, does a firm "have low market power" if it sells at a low markup today to cause a high markup tomorrow? cf. Viner (1932) on marginal costs.
I know a few classic arguments for the characteristics approach, like dimensionality and heterogeneity.
But it also seems like we've got much better data and tricks now.
@BrianCAlbrecht
You need an input wedge to rationalize the data! Non-neutral productivity, adjustment costs, or (my own view) input market power. Excellent paper and post.
The gap is a โprofit puzzleโ:
Why don't they profit by borrowing at the lower rate and investing at the higher rate?
Markups? Risk? Intangible capital?
@florianederer
@NateHMiller
@conlon_chris
Yes, though I believe if DLEU markups are estimated with a national price index, we mechanically shouldn't expect any subnational (i.e. industry) correlation with prices.
Directly follows from Klette & Griliches (1996) or Bond et al. (2021), right
@ArshiaHashemi
?
@xftrebbi
Each first-order condition gives you a joint wedge -- output and input. (e.g. markup and markdown). If you assume input wedges for two different inputs are both zero, they should recover the same output wedge. Devesh's excellent work shows that assumption is bunk!