
Professor @Harvard, Director @HarvardBizGov, Senior Fellow @PIIE, teaches Ec10, words @nytopinion, proprietor https://teapaw.org/t.co/4eOTjrd0x4, was Chair of Pres Obama's CEA.
Cambridge, MA
Joined April 2011
- Tweets22.4K
- Following2K
- Followers342K
- Likes7.4K
A gripping opening slide from Amy Finkelstein's excellent talk to the Chae macro policy seminar.
An intriguing set of findings about how manufacturing job loss cost lives but service job loss actually saves them. Answered some questions while raising lots of new ones.
Slides here, are based on several papers she's done with @ProfNoto and others. Will be interested in how this discussion progresses.
The relationship between health and macro is surprisingly complicated, at least at business cycle frequency.
economics.harvard.edu/sites/…
I just did this on Opus 3 (originally introduced 32 months ago, in March 2024).
What will these models be able to do 32 months from now, in August 2029?
P.S. I am glad that Levent Alpöge is a serious mathematician who was thinking about Riemann for years and did substantial iteration with Claude.
But it is hard not to believe that problems like this could be one shotted by anyone in the not too distant future.
P.P.S. on the P.S. teapaw.org/boazbaraktcs/status/21…
The Riemann hypothesis has resisted all human mathematician effort but has now been solved by a lab employee working on it as a “hobby” on his spare time with access to internal models and extremely high compute (latter of which we can tell even based on sheer size of this proof). I’m sure his mathematical skills were useful. I’m also sure that at best they were a relatively moderate compute saving that got it a month or two ahead of schedule.
I asked Claude (no internet, just training data which is through June 2026) for the hardest outstanding math problems and a prediction of when they would be solved by AI.
The first Millennium problem to fall with AI as a major contributor: most likely Navier–Stokes. Maybe 25–35% by 2030...
Riemann Hypothesis: probably among the last, median in the 2040s
And here is what Terence Tao thought about this way back in March. teapaw.org/dwarkesh_sp/status/212…
When I interviewed Terry Tao in March of this year, I asked him about AI solving the Riemann hypothesis. Here's what he said:
"Part of the reason we prize problems like the Riemann hypothesis is that we’re pretty sure a new type of mathematics has to be created.
We don’t even know what the shape of the solution is, but it doesn’t feel like a problem that will be solved just by exhaustively checking cases."
20
6,607
Jason Furman retweeted
As we brace for what might be another ugly day on Wall Street, @jasonfurman notes the stock market has called 17 (!) of the last 9 recessions (based on a drop of 20%).
And, markets "missed" 3 recessions that arrived without a drop of 20%+
This is a manufacturing boom in Taiwan & Korea.
Which is one of the beauties of globalization.
We have a great idea here in America, we borrow money from the rest of the world to finance it (financial account surplus) & buy the physical stuff from countries that make it better.
Matt is the most reliable, careful and honest analyst there is on this topic, among other topics. I hope some of the open-minded people who raised skeptical questions about this issue (you know who you are!) take the time to read and work through his evidence and arguments.
I opted for overkill on the new Econvitals Recession Explorer.
Pick your favorite recession and see how it compares. Rank them. Or track where we are now. One example in each post.
🆕 New on Econvitals: the Recession Explorer, every U.S. recession since 1948 side by side.
Measured against trend, consumer spending was the largest drag on GDP in 8 of the 11 recessions since 1953. In 2007–09, real GDP fell 8.0% below trend.
econvitals.org/tools/recessi…
How employment-population ratios fared in each recession (this is benchmarked against a baseline, the default in the system--which you can change--is the trend from the previous ten years).
And where we are right now compared to previous recessions--short version, not close. But data gets revised and can change quickly. (No bigger risk now than usual, that's a perennial caveat.)
econvitals.org/tools/recessi…
Still early to have much confidence on Q4. But the signs we have are reasonably strong.

















