Sam Gyetvay

@GarettJones @Ahlurahul @mean_field_zane Arrow discovered CES, though msuweb.montclair.edu/~lebelp/acmsca…

@Ahlurahul @mean_field_zane Dixit-Stiglitz Rules Everything Around Me

I mean, it’s entirely silly to refer to one of the most foundational economic theorists of all time as “niche”. His work is the foundation for literally every single model in modern macroeconomics and also the study of adverse selection & moral hazard.

My paper "Work Hours Mismatch" with @MartaLachowska @raffaelesaggio and Steve Woodbury is now in print at Econometrica.

Interpreting event-studies from recent difference-in-differences methods link.springer.com/article/10.100…

📢New Professor at SMU We are thrilled to announce that Fabien Postel-Vinay @fpvinay is joining @SMU as a Full Professor of Economics in the Fall. 📈Fabien is a distinguished researcher with seminal contributions, particularly in labor macroeconomics. Welcome to Dallas!

@KRoyMyers @mattsclancy @paulnovosad @ChadJonesEcon nice metrics paper on this idea arxiv.org/abs/2412.07031

@mattsclancy @paulnovosad @ChadJonesEcon Thoughts on “vintage LLMs” as a (the?) solution to creating a lot of this sort of data that is implicitly supposed to be state dependent? (ie, generating data about some point only with information available in that same point)

Two recent examples of how AI lets you do "expert judgment at scales large enough to do quantitative analysis on": @ChadJonesEcon and Christopher Tonetti's new paper on automation and growth. It has a model where production is a huge number of tasks. They use structured prompts to ask LLMs to come up with 100 production tasks per sector and estimate how much of the task was performed by machine or humans at different points in time, and the share of production costs associated with the task. Then they can use the results to do historical analysis and forward looking simulations on the contribution of automation to productivity growth. web.stanford.edu/~chadj/JonesTo… @_brianpotter's latest blog post asks claude how much earlier 160 inventions could have been invented, given the important components in the inventions and the state of technology in earlier years. He can then do descriptive data and trends on the overall set. construction-physics.com/p/how-long-do-… I'm sure there will be a bunch of methodological work to hammer out how well this works and what the failure modes are, but it seems like an exciting new tool to me. In the economics of innovation, we often rely on these very noisy proxies for subjective judgement: how many citations received, as a measure of a paper's impact/importance; clever transformations of citation data to assess how disruptive a paper is; overlap of technological classifications in patents to assess how similar two technologies are, etc. AI judgment (if validated) could let you come up with more precise measures of what we care about in all these cases.

Firms hiring under demand uncertainty face a tradeoff between holding permanent workers as a buffer and hiring from a spot market that thickens with use, from @AMLabEcon nber.org/papers/w35148

Canada’s mandatory retirement bans reduced retirements at 65, raised post-65 employment and earnings, and showed no adverse effects on firms or younger workers, from @ToddMorrisEcon, @stefan_staubli, and Benoit Dostie nber.org/papers/w35111

Very interesting paper now published at QJE Effects of Mandatory Profit-Sharing on Workers and Firms: Evidence from France* academic.oup.com/qje/article-ab…

A job ladder model with incomplete markets links wealth to job search over the life cycle. It provides a new microfoundation for all three inputs in F(K, AL) and offers novel insights into how redistributive policies affect aggregate output and welfare. econometricsociety.org/publications/e…

Forthcoming in the AER: "The Effect of High-Tech Clusters on the Productivity of Top Inventors: Comment" by Michael Wiebe. aeaweb.org/articles?id=10…

🚨 New paper alert 🚨 Are you also torn between small sample performance and robustness to heteroskedasticity? You can have both! One can produce an exact test for subvectors that is asymptotically robust to heteroskedasticity. Forthcoming Econometrica

Pat Kline: "I don't like the phrase 'causal inference' either because it's usually 'causal estimand identification' are what those papers actually are; there's nothing to do with inference typically; there's no distribution theory in those papers." youtu.be/ifp_BUKl3ig?si…

What do workers at the margin of informality value at work? We surveyed 700 residents of Brazil's largest favela complex to find out, using open-ended questions and discrete choice experiments. Joint work with Beatriz Marcoje, @iedarm, and @rodmclara. Here's what we learned. A thread 👇:

🚨New in AER🚨: "Negative Control Falsification Tests for IV Designs" (@DanieliOren, @DanielNevo, Itai Walk, Bar Weinstein, @danzeltzer). doi.org/10.1257/aer.20… Common placebo test implementations may reject valid IVs. We're coming to save your IVs! 🧵

🧠New RFBerlin Research Insight! 💸Germany's tax system rewards you for getting married. Not for having children. New research by @hanna_wang_ shows this might be exactly why the birth rate is stuck. 🔗Read about it here: rfberlin.com/research-insig…

Erwin Diewert should win the Nobel in economics this year. His work on price indexes -- while some find it boring -- is basically the thing that is everywhere around us without us seeing it. It is like the air we breathe as economists. You go to the BLS website to get the price index? Diewert is there! You get a GDP series from the FRED database -- Diewert is there. You open your macro textbook to teach undergrads about inflation, Diewert is there. You open your micro textbook to teach undegrads about substitution effects, Diewert is there. Turn around, Diewert is there too! Yet, we never recognize that his work gave us powerful insights bridges to gap between theory and applications. Not giving it to him at this point is the same slight as not giving it to Harberger, Demsetz, Alchian, or McKenzie (or the Arrow-Debreu-McKenzie)

Today's cool young researcher #econtwitter #econsky is @JuanFelipeRiano @Georgetown who works (quite prolifically!) on topics related to development, political economy and economic history

An absolute banger of a paper. “A Gentle Introduction to Matrix Calculus” by econometrics legend Jan Magnus — one of the clearest explanations of matrix derivatives ever written. If you work in econometrics, machine learning, statistics, or optimisation, this paper is pure gold.

Very excited for this Friday's Chamberlain Online Seminar, where we'll muse about "The Future of Econometrics." Check out this panelist lineup! Friday 3/27 at Noon ET. Come see!

If the Middle East conflict triggers mass refugee movements, where would people go? New research from @CReAM_Research at RFBerlin finds the answer lies where refugee communities already exist, and in Europe, that means Germany. 📄 𝗥𝗲𝗮𝗱 𝗶𝘁 𝗵𝗲𝗿𝗲: lnkd.in/deZevBFq