I am so honored and grateful to receive this
@NSF
CAREER Award that will help push my research forward.
And feel blessed for having such amazing co-authors in the projects that will be supported by the award (
@Nathan_Zorzi
,
@david_yang
, Noam Yuchtman, Andrew Kao, Paco Buera).
Congratulations to
@MartinBeraja
, who has been awarded an
@NSF
CAREER Award to support his research on the role of government policy in stabilizing business cycles and responding to challenges posed by new digital and automation technologies.
Taxi to JFK. Driver asks: where are you from? Argentina, I say. Driver turns around. Opens his jacket. He is wearing an Argentinean football jersey with the number 10 in it. I’m from Bangladesh, he says. We are big fans! You don’t believe me? Let me show you. Hilarity ensues…
This story now ends. He gave me a hug when I got off the car. We looked at each other, united in that brief moment by one thing.
It cuts across class, age, race, etc., as well as, I learned, nationality. This love for ⚽️ and the World Cup.
At this point, I am baffled. Bangladesh’s population is about 150mill (thanks again Wikipedia). By his calculation there are more Argentina fans in Bangladesh than Argentineans. There will also prolly be more Bangladeshi fans in Qatar (0.8 x 400k) than Argentineans there.
First he shows me pictures of his house. Big Argentina flag is hanging from his balcony. Then pulls up Youtube. Shows me this video of hundreds of Bangladeshi on the street with the 🇦🇷 jersey.
He tells me that 80% of the country supports Argentina. And that Qatar is an important migrant destination. I check Wikipedia. 400k Bangladeshis are there.
He calls his brother next (i think). Who do you support in the World Cup? Some short back and forth in Bengali happens (had to look this up, I must confess. Didn’t know the language’s name). The brother starts chanting “Argentina! Argentina!”
Nunca me sentí tan extranjero como ayer. Mi vecino me deslizó una carta, muy educada, por "algunos ruidos a la noche." Fueron los goles de
@RiverPlate
que grité. Lo comprendí, pero cómo explicárselo? La distancia cultural era insalvable.
@KenoBilli
Comercial de
@Quilmes_Cerveza
. Se suceden imagenes de personas en distintas partes del mundo. Tomando mate. Con camiseta argentina. Comiendo dulce de leche. Bailando cuarteto. Etc. Termina con alguien tomando una Quilmes y una voz en off: El que es argentino nace donde quiere.
🚨New paper🚨 with
@Nathan_Zorzi
Automation raises productivity but displaces workers and lowers their earnings. When these workers are financially vulnerable, the degree of automation can be excessive.
How should the government respond?
A thread 🧵…1/n
@MITEcon
#EconTwitter
Automation can be excessive when it displaces workers who are financially vulnerable. This may justify slowing down automation, even when accounting for generous social insurance, from
@MartinBeraja
and
@Nathan_Zorzi
I used to be an F-1 student. Many of my friends and colleagues too. To those of you who are F-1 students now, I am so sorry. To those of us who teach, imagine our classes without them. This cannot be.
🚨New paper🚨 with
@ChristianKWolf
Consider two recessions. In one, people stop buying cars and furniture. In the other, they stop going to restaurants and the dentist. Would you expect the economic recoveries to differ?
A thread 🧵... 1/n
@MITEcon
#EconTwitter
Recoveries from recessions concentrated in services and non-durables will tend to be weaker than from recessions biased towards durables, as the recovery is hampered by weaker pent up demand, from
@MartinBeraja
and
@ChristianKWolf
Do you listen to music while working?
I do. Different styles for different tasks. Indie for writing, funk/soul for proving stuff, electronic for data or coding.
I am often looking for new music. If you are too, stay tuned. I’ll start sharing what I listen to.
#ResearchSounds
I respectfully disagree. While the “ultimate” macro questions can’t be tackled this way, the tools for estimating cause/effect have helped us provide evidence for mechanisms and discipline macro models. Lots of recent papers are in this vein, in particular by young macro folks…
The credibility revolution has had less impact in macro, and that's because it's difficult to think of experiments that also account for the spillover effects from one person or one market to the other. Those "general equilibrium" effects are absolutely central to macro.
This made me think about what
@JohnHCochrane
and Tom Sargent used to tell us in their class
@UChicago
.
John: economics is about conditional predictions, not unconditional ones.
Tom: economics is about the “ifs” and the “thens.”
Understanding = Conditional if/then prediction
Predictability and understanding are not the same thing. Something can be well understood but at the same time completely unpredictable. This is obvious with earthquakes. But somehow many people have a hard time with this notion for economic phenomena. 3/3.
Come see
@DrDaronAcemoglu
,
@JeanTirole
, and Noam Yuchtman present research on Public Policy and Political Economy in the Age of AI.
Our
#ASSA2022
Session is tomorrow Friday, Jan. 7, 2022, 3:45 PM - 5:45 PM (ET).
#EconTwitter
Details:
If you are interested in AI and Governments/Policy, here is the Lecture I gave at the
@AEAjournals
Continuing Education program.
AI is a data-intensive technology that can be used for automation and surveillance. What role for governments in each case?
The more credible approaches (in my view) match conditional moments, like responses to a particular shock, not unconditional ones, like this blog.
There is a nice exercise illustrating why in my paper on regional business cycles:
1/n
This is a really interesting blog post about DSGE models -- almost makes me want to get into the identification / estimation issues around macro structural models! (almost)
From the paper:
🤖🛑 Hitting the breaks on AI?
(1) Alignment worries: perhaps legit. But don't have much to say.
(2) Job automation concerns: LLMs ≠ robots. Economic arguments for slowing down automation are much weaker for LLMs.
Why? Let's dive in!
#EconTwitter
#FutureOfWork
#AI
#ChatGPT
Forthcoming! Much work to do on the normative side.
Firms mostly use privately collected data. But gov'ts have troves of data too --> Gov't data provision can be desirable IF civil liberties violations are unlikely. We need to measure such risk + models to quantify trade-offs.
🚨New paper🚨
AI innovation may both entrench modern autocracies and be promoted by them in a sustained manner.
Read on for evidence of such mutually reinforcing relationship in the context of facial recognition AI in China.
A thread 🧵 1/13
@MITEcon
#EconTwitter
Study finds a mutually reinforcing relationship between frontier innovation and autocracy in the context of facial recognition AI in China, from
@MartinBeraja
, Andrew Kao,
@david_yang
, and Noam Yuchtman
Classics:
- Chris Sims's "Macroeconomics and Reality"
- Cowles Commission (Koopmans, 1950; Marschak, 1974; Hurwicz, 1962); Heckman and Vytlacil (2007)
In DSGE models:
- Del Negro-Schorfheide (2009)
- My JMP on semi-structural counterfactuals:
#EconTwitter
Hi
#EconTwitter
!
On top of the paper written by Lucas himself, is there any important piece of literature about the Lucas critique one should read?
Thanks! RT appreciated 🙂
A propósito del DNU de
@JMilei
, me puse a buscar qué sabemos sobre los efectos macro de la desregulación, tanto en el corto como largo plazo.
Me gustó este paper de
@ojblanchard1
y Giavazzi. Tiene algunos lindos resultados; en particular sobre la diferencia entre sacar
🚨New paper🚨
States play an important role in the age of data-intensive innovation. Evidence from AI firms in China + Macro implications of government data provision to firms.
See it Thursday July 16th, 11ET, @ NBER Income Distribution and Macro:
(1/18)
This is Twitter at its best. An expert writes a piece on an interesting/important topic that I am not an expert on.
@ethanbdm
writes a (very) well argued rebuttal. I am now better informed, without the need to have spent any time thinking about the topic myself.
The mathematician Richard Hamming wrote about the tension between keeping your door open vs. keeping it closed.
He said: On any given day, you'll get more done if you work with the door closed. But over the long arc of time, you'll achieve more if you work with the door open.
Hace 27 años me caía de la cama.
#AMIA
Vivía por Av. Cordoba, entre Once y Barrio Norte, y la ventana de mi cuarto daba al pulmón de manzana. Como iba al colegio a la tarde, a las 9:53 estaba durmiendo. Recuerdo un estruendo y levantarme en el piso. 1/
Come for the evidence, stay for the q-theory...
In our new WP with Rodrigo and Nitya we explain why some technological transitions are particularly unequal and slow to play out.
Ungated:
Cognitive-biased transitions are slow and unequal because they are mostly driven by skill changes across generations rather than worker reallocation within a generation, from Rodrigo Adão,
@MartinBeraja
, and Nitya Pandalai-Nayar
I teach lots of recent papers (in particular JMPs) with the following template:
-interesting question
-credible identification strategy that nails down a mechanism (thanks Nobels)
-macro model to derive implications of said mechanism
Very much in line with my experience teaching 2nd year Macro. A PSet often involves taking a dataset (eg PSID) and finding evidence for/against a hypothesis related to a topic we have covered. We then discuss results and choices made in class. 1/2
When testing the same hypothesis with the same data, different teams find (sometimes drastically) different results.
Researcher degrees of freedom in analyses create another layer of uncertainty in published results.
Source:
How do inequality, skills, and economic activity adjust over time to new technologies?
Theory of tech. transitions driven by: (i) worker reallocation within a generation and (ii) changes in skills across generations
Evidence: adjustment to cognitive-biased innovations
1/2
VMACS Junior 21/07 @ 12pm ET:
@MartinBeraja
(MIT) presents
"Technological Transitions with Skill Heterogeneity Across Generations",
joint work with Rodrigo Adao (Booth) and
@nityanayar
(UT Austin)
Sign up @
Very cool to see
@Lagarde
@ecb
cite our work w/
@ChristianKWolf
at 10'14".
"recovery from a services-led recession tends to be slower than from a durable goods-led recession, as services create less pent-up demand than consumer goods."
New draft soon!
Help from the
#EconTwitter
crowd, please.
I am looking for examples of applied theory papers (macro or micro) with a short section at the end that discusses/interprets some evidence through the lens of theory.
No quantification or fancy empirics. Just figures/tables with the
After the initial shock of Diego's passing, one of my first thoughts was: the world will mourn him for who he was as a football player, but how could anyone else even begin to understand what he meant to many Argentineans? I was wrong.
@EmmanuelMacron
got it, in beautiful prose.
Leí la nota de
@pgerchunoff
que generó revuelo. Para entenderla hay que haber leído a Pablo antes (o sido su alumno). Sigue su lógica habitual, que puede sonar a excusa, pero no lo es. Me permito una crítica sobre el final: la locura quizás sea tal. Sigo...
Don't miss the
#ASSA2022
Econometric Society's Job Market Lightning Round, 12:15pm ET January 7th.
Ten job market candidates. Fascinating research across fields: from macro to applied metrics, to political econ, to innovation.
Details:
@p_ganong
I like to tell my students to think about macro as a sea that feeds from many rivers. Historically, it has fed from math and engineering for studying dynamic models. It has then fed from in time-series metrics. It has recently fed from applied micro because of breakthroughs there
🇧🇷Hay que agradecer a Brasil por muchas cosas: las mujeres, la samba y el fútbol.
Este país nos ha regalado leyendas, sobretodo goleadores. Pero hubo un defensa que marcó historia con sus golazos🚀
Hoy, en su cumpleaños, hilo del mejor lateral izquierdo de la historia: RC3 🧵
I love you
@Spotify
. You're in my top 5 of products. But your interface gets more incomprehensible with every update.
Please, simplify it around 3 things:
1) What I have - library/just listened
2) What I'd like - recommendations/your mixes
3) What is new - albums/songs/podcasts
💯. If you want to see how to use state-level evidence + models to tackle questions about aggregate business cycles and policy, check out these two papers of mine.
So be careful in using state-level evidence an arguments when trying to infer what would happen with a national gas tax holiday. Supply is much more inelastic at the national level--and much more of the benefits go to producers.
A Passover table question that caught me off-guard: if they were short on time, why didn’t they cook a Mexican tortilla which is much better than Matzah? 😂🤷🏻
From what we know so far, LLMs are likely to automate cognitive-intensive jobs. These tend to be middle-to high-wage jobs (think lawyers or software devs). So LLM-led automation would likely decrease inequality. The redistributive rationale for slowing down automation is gone.
Technological transitions are *slower* when skill specificity is *stronger* b/c it is driven less by the fast reallocation of older incumbent workers and more by the gradual entry of *younger* generations.
One insight from Adao,
@MartinBeraja
, and
@nityanayar
(2022).
A mini-🧵
The aesthetic beauty, the childhood memories and nostalgia (you had me at Panini album), the pride, the sadness and anger of a stupid war, the politics...it's all here. 👏🏽👏🏽👏🏽
@guido_lorenzoni
Exactly right. We were just having this discussion with
@alpsimsek_econ
(coincidentally, just after bumping into each other in the bathroom).
Even further, why not increase sales tax on "social-interaction" goods, decrease it for online goods, durables, etc.?
Just incredible... These two individuals, their connection through circumstance; their letters brimming with humor right out of a Russian literature classic.
Take 5 minutes, and read them. Well worth your time.
From the punishment cell of the gulag where he was killed, Russian political prisoner Alexei Navalny wrote to the former Soviet dissident
@natan_sharansky
.
Today, we are honored to publish the letters in their original, handwritten Russian and in English:
I read an early draft yesterday. My New Keynesian side of the brain first reaction was literally: isn't it just \beta?
Then, my Menu-Cost side took over and was like: no dummy, of course you have more than one-for-one passthrough.
Read it, but flatten your priors first 😅
What is the causal effect of a change in expected inflation on actual inflation? Which expectations matter most: short- or long-run?
In a new paper I try to address these questions. A link to a very early draft (updates coming soon):
A thread!
@jasonfurman
you are spot on. We show this with
@ChristianKWolf
in "Demand Composition and the Strength of Recoveries."
When spending cuts are concentrated in services as opposed to durables - like in this recession - recoveries will tend to be weaker.
I am unsure about pent up demand because normally see it in durables and people bought *more* of them in 2020 than normal. In fact, some of those purchases might have pulled demand forward resulting in less spending in 2021 (e.g., you got the car in 2020 so no new car in 2021).
For students interested in this and other questions related to new technologies and the role of governments, I'm giving an NBER Tutorial in March in the Digitization group run by
@ce_tucker
and
@avicgoldfarb
.
Be on the lookout for the call to register!
#EconTwitter
Very good thread. I took a writing class at the Little Red Schoolhouse
@UChicago
(students: I would recommend it if your univ offers it). They called this “managing the flow of information.” The off/on-ramp metaphors in this thread are on point.
One risk here is to make the
Have you ever read a paper (or seen a talk) where you feel the author is reading your mind?
Like any concern is immediately addressed right after it crosses your mind?
I think that is a hallmark of great writing, and I have a theory about it.
A thread... 1/12
@FabrizioZilibo1
One of my all-time favorites. I read it as a freshman and remember thinking: Ah! We can tell stories with elegant math about things that I care about from my social science and history courses? Interesting...
Any interesting policy implications?
If the
@federalreserve
treats services recessions as typical durables-led ones, it would hike rates too fast (compared to optimal) and output would be depressed for longer, as this ignores that pent-up demand will boost the recovery less. END
"Creativity is an act of magic rising up from your subconscious. It feels wonderful every time it happens, and I’ve learned to live with the anxiety of it not happening over long periods of time.”
Tagging some Twitter people that I am sure have more to add.
@JonSteinsson
talks about using "well-identified moments" in one of his papers.
@EichMartin
of course has thought about these issues at length, as has
@pablo_guerron
.
@PhilHaile
must have some wisdom to share too.
The Sputnik V vaccine, developed in Russia, has been shown to be 91.6% effective against the coronavirus, according to a new peer-reviewed analysis in the medical journal The Lancet.
AI misalignment remains a valid concern, in principle.
But I’d like to see more formal economic arguments rather than armchair theorizing; to spell out assumptions and evaluate them. At this point, I don’t feel we have such arguments to so confidently call for an AI moratorium.
In a nutshell, the redistributive rationale is as follows. If automation displaces workers who are relatively poor, then this increases inequality. Absent other tools to directly redistribute (e.g., transfers to workers) a government should curb automation to reduce inequality.
@ShengwuLi
The issues with identification/inference in DSGEs are well known, when using unconditional observational data (as in that blog). But you wouldn't say that a demand-supply model is not useful because one cannot identify elasticities from data on prices and quantities alone.
In a class of about 15 students, I’m always surprised at how seemingly small differences in choices lead to very different results. They often come from the way the data was constructed or slight differences in specification that I wouldn’t have predicted would matter so much.
"People talk abt the story of immigration as one big happy tale. But in every immigrant story, there's sadness as well..sadness of a country, a culture, a family left behind. A mother who quietly weeps at night distant from the child she loves"
This hit a little too close to home
Great read. I’ll expand on one thought (for students):
I envy you. You are taking different courses at the same time. The ideas you learn keep “running in the background” of your mind. Makes it easier to arbitrage ideas across fields. See connections, analogies or a tool that
The essay I just published called "How to Do Great Work" () grew out of a single paragraph in another essay I was writing. It seemed such an important topic that I cut it out and made it into its own essay.
Ok. On to some music.
I will kick things off with two artists that I have been listening to: Black Pumas and Dope Lemon.
Both are soulful and have a bit of psychedelia. Mellow at times too. Strong voices and guitars.
Check them out and other
#ResearchSounds
in the future.
En cambio, lograr bajar la desigualdad quizás requiera de paciencia. Es un arduo proceso de décadas. De gobiernos consecutivos que con políticas correctas, pero no revolucionarias, y algo de suerte, porque siempre es necesaria, poco a poco le vayan haciendo mella al problema.
The redistributive argument is much weaker for LLMs than it was for robots.
Robots automate jobs intensive in routine tasks. These tend to be low-to-middle-wage jobs (think workers in a car manufacturing plant). So automation increases inequality.
Y seguí caminando por ahí muchos de los años que siguieron. Viendo distintas versiones de esta imagen. El triste recordatorio de la impunidad y de los intentos de
jueces y gobiernos de oscurecer, en vez de aclarar.
Transfers are the best tool for this redistributive goal. In practice, these are hard to implement or not generous enough.
In this case, "taxing robots" is second best. See the nice papers by Guerreiro-
@SergioRebelo6
-Teles; Costinot-
@IvanWerning
;
@akorinek
-
@JosephEStiglitz
. 4/n
Result 2. A government should tax automation (even if it does not value equity) when it lacks instruments to fully alleviate borrowing frictions. The optimal policy slows down automation but doesn't tax it in the long-run. 10/n
Both assumptions seem much less relevant for workers displaced by LLMs than by robots.
Again, LLMs will displace relatively richer workers in cognitive-intensive jobs. These workers are likely to have sufficient savings or able to borrow while relocating to a new job.
The efficiency rationale is in my paper with
@Nathan_Zorzi
. See this thread for the argument:
In a nutshell, firms automate excessively because
there is a conflict between how firms and displaced workers value the gains from automation.
🚨New paper🚨 with
@Nathan_Zorzi
Automation raises productivity but displaces workers and lowers their earnings. When these workers are financially vulnerable, the degree of automation can be excessive.
How should the government respond?
A thread 🧵…1/n
@MITEcon
#EconTwitter
A very nice write-up of our "AI-tocracy" paper. Here is a link to the
@QJEHarvard
version ()
Thank you Peter Dizikes from
@MIT
News for an excellent coverage.
China’s use of AI-driven facial-recognition tools appears to help repress dissent, a new paper reports: “AI innovation entrenches the regime, and the regime’s investment in AI for political control stimulates further frontier innovation.”
It has long been known that "pent-up demand" is stronger for more durable goods (like cars).
This basic consumer theory insight goes back to at least Hayashi (1985),
@gregmankiwblog
(1982) and Caballero (1993). 2/n
There is by now good evidence on the impact of automation on labor markets; for example from great work by
@DrDaronAcemoglu
-Restrepo or
@AndersHumlum
.
However, while workers may be worse-off, the evidence does not necessarily imply that governments should intervene. 2/n