It's worth separating two issues in the *China Shock* debate. Autor-Dorn-Hanson was a positive development for academic research. It paved the way for a lot of interesting work on trade and inequality. Several follow-up studies estimate similar distributional effects using different data and methods (see Rafael's thoughtful thread on this). Outside academia, however, the results were widely misinterpreted. Many used them to push protectionist agendas and to argue that trade with China was a net negative for the US economy, which the results never supported. This kind of misinterpretation will keep happening until we adopt stricter guardrails on how to present cross-sectional estimates. As @ben_moll and @oliverhanney's Vox article argues, the practice of making aggregate claims from cross-sectional estimates is pervasive but generally invalid, because of the *missing intercept* problem.

🚨 New paper🚨 How will AI affect the economy via making software engineering cheaper and better? With great coauthors Alex Blumenfeld, @ChenLian92 and Andreas Schaab. (1/9)

New post with the excellent @FutureEconJacob examining the evidence (so far) on the impact of AI on the labor market. We reviewed 20+ papers and datasets and attempted to synthesize the evidence. Our take: 1) There is no strong signal that AI has had a negative or positive impact on employment in the aggregate. 2) When looking at specific sectors and demographics, the evidence is fairly mixed. While some have argued for negative impact on AI-exposed, early career hiring, others have identified moderators such as remote work or macroeconomic trends. There is also some evidence of positive impacts as well. So has AI hit the labor market yet? When looking at the employment statistics--not really. But that does not mean that jobs are not changing significantly, and that there aren't substantial changes in what skills are becoming less relevant versus in greater demand. That's a topic for another post! Link:

TLDR: whether wages go ⬆️ a lot as the K-share ⬆️ depends almost entirely on Pi:=cost of doing the newly automated tasks with L vs w AI. If you write Y=K^aL^{1-a} you're implicitly assuming Pi is huge, like 10x-35x 🤖 No surprise u'll find huge wages gains. Pls, report your Pi‼️

Hi Friends, Thanks for all the support for the Everywhere Millionaire. Based on bookscan sales data, it looks like we are right on the bubble of the best seller list in business for September, so if you could pick up a copy at a local bookstore *in the next few days*, suggest the book to friends, retweet this message, or post about it - we’d greatly appreciate it!

New BPEA paper by @SVNieuwerburgh argues that AI buildout will cost $10.3T over next 8 years. Furthermore, to get a 10% return on this, those investing will need annual revenues of $3.7T by 2032! (Approx 9% of GDP!)

We are again recruiting @bold_lab_ai - please share the love 🙏. We are looking for: 1) Postdocs ( -- deadline 9th of Oct at noon 2) research assistants ( -- deadline 9th of Oct at noon 3) Strategic Partnership Project Manager ( -- deadline 14th of Oct at noon

On Thursday, October 1st, 12:15 pm, we are thrilled to welcome Ben Moll, LSE, to our Macro-CRC TR 224 Seminar (@EPoS224, @EconUniMannheim ). He will present “Structural Reinforcement Learning for Heterogeneous Agent Macroeconomics.” More info – @ben_moll

Each Refine 5 review uses enough frontier tokens that we're flirting with losing money on it. To celebrate the launch, we decided to get crazier. Is your research outside social science? Reply "Try Refine 5 on [your field]" for free or dirt-cheap reviews. Rules below 👇

Refine 5, a strong new reviewing tool, is live today. It beats the amazing new frontier models at verifying technical work. Free with every Refine review through 10/15. We benchmarked on 108 papers including bio, engineering, physics, environmental science, econ. 🧵 1/

If AI raises the capital share of income, will workers ultimately benefit? Recent analyses using a workhorse macro model concluded: YES. I wrote a paper explaining why these conclusions are somewhat misleading. The future of wages might not be so rosy.

Interesting question: Who is taking the other side in a P(doom) bet? What Jakob is saying: if we ever see liquid prediction markets with implied P(doom) substantially >0 (after adjusting for fees etc), THEN that’s when we should be really worried because the counterparty is likely AI