Martha Gimbel
@marthagimbel.bsky.social
Sometimes Econ, sometimes rants, sometimes both. Current Budget Lab at Yale, former CEA, JEC, DOL, SF, Indeed.
If you know someone who might fit the bill - please send them our way! Happy to answer any questions.
October 6, 2025 at 2:09 PM
If you know someone who might fit the bill - please send them our way! Happy to answer any questions.
My take is: Did we lose exactly 30k jobs in September? Almost certainly not. Did we lose some jobs in September? Plausibly! Did we add enough jobs in September to keep up with population growth? Seems quite unlikely.
October 1, 2025 at 3:56 PM
My take is: Did we lose exactly 30k jobs in September? Almost certainly not. Did we lose some jobs in September? Plausibly! Did we add enough jobs in September to keep up with population growth? Seems quite unlikely.
And thanks to Claire Jones and @financialtimes.com for the coverage: www.ft.com/content/c9f9...
AI is not killing jobs, US study finds
Research shows little evidence that cutting-edge tech such as chatbots is putting people out of work
www.ft.com
October 1, 2025 at 3:54 PM
And thanks to Claire Jones and @financialtimes.com for the coverage: www.ft.com/content/c9f9...
The full piece is here, please give it and all of its many many graphs a read budgetlab.yale.edu/research/eva...
Evaluating the Impact of AI on the Labor Market: Current State of Affairs
budgetlab.yale.edu
October 1, 2025 at 3:54 PM
The full piece is here, please give it and all of its many many graphs a read budgetlab.yale.edu/research/eva...
Thanks to @mollykinder.bsky.social @brookings.edu for being such wonderful partners on this and to Josh and Maddie for objectively doing all the work
October 1, 2025 at 3:54 PM
Thanks to @mollykinder.bsky.social @brookings.edu for being such wonderful partners on this and to Josh and Maddie for objectively doing all the work
We plan to update this every month (assuming we get the CPS microdata!) and also to adjust our methodology in response to feedback/new data on usage/exposure. /end
October 1, 2025 at 3:54 PM
We plan to update this every month (assuming we get the CPS microdata!) and also to adjust our methodology in response to feedback/new data on usage/exposure. /end
The lack of widespread impacts at this early stage is not unlike the pace of change with previous periods of technological change. Preregistering areas where we would expect to see the impact and continuing to monitor monthly impacts will help us distinguish rumor from fact. 14/
October 1, 2025 at 3:54 PM
The lack of widespread impacts at this early stage is not unlike the pace of change with previous periods of technological change. Preregistering areas where we would expect to see the impact and continuing to monitor monthly impacts will help us distinguish rumor from fact. 14/
While generative AI looks likely to join the ranks of transformative, general purpose technologies, it is too soon to tell how disruptive the technology will be to jobs. 13/
October 1, 2025 at 3:54 PM
While generative AI looks likely to join the ranks of transformative, general purpose technologies, it is too soon to tell how disruptive the technology will be to jobs. 13/
While anxiety over the effects of AI on today’s labor market is widespread, our data suggests it remains largely speculative. The picture of AI’s impact on the labor market that emerges from our data is one that largely reflects stability, not yet major disruption. 12/
October 1, 2025 at 3:54 PM
While anxiety over the effects of AI on today’s labor market is widespread, our data suggests it remains largely speculative. The picture of AI’s impact on the labor market that emerges from our data is one that largely reflects stability, not yet major disruption. 12/
We should note that data in this space is still imperfect, and usage/exposure is still evolving. Here is a comparison of OpenAI and Anthropic's measures of exposure and usage for instance 11/
October 1, 2025 at 3:54 PM
We should note that data in this space is still imperfect, and usage/exposure is still evolving. Here is a comparison of OpenAI and Anthropic's measures of exposure and usage for instance 11/
And also duration of unemployment 10/
October 1, 2025 at 3:54 PM
And also duration of unemployment 10/
We see a similar story using Anthropic's usage data for both automation and augmentation in employment... 9/
October 1, 2025 at 3:54 PM
We see a similar story using Anthropic's usage data for both automation and augmentation in employment... 9/
And more exposed workers aren't accounting for a greater share of longer-term unemployed workers 8/
October 1, 2025 at 3:54 PM
And more exposed workers aren't accounting for a greater share of longer-term unemployed workers 8/
So far there's no sign that relatively more exposure to AI is at all impacting the share of employment in those occupations 7/
October 1, 2025 at 3:54 PM
So far there's no sign that relatively more exposure to AI is at all impacting the share of employment in those occupations 7/
We then look at employment shares and unemployment by duration among workers who are more or less exposed to AI (according to OpenAI) and whose occupations have more or less usage (automated or augmented) according to Anthropic 6/
October 1, 2025 at 3:54 PM
We then look at employment shares and unemployment by duration among workers who are more or less exposed to AI (according to OpenAI) and whose occupations have more or less usage (automated or augmented) according to Anthropic 6/
If we look at how similar the occupational mix of recent college graduates is to slightly older college graduates - there may be some sign that the difference is growing? But that could be due to 1) noise 2) non-AI factors (such as a slowing labor market), not just AI 5/
October 1, 2025 at 3:54 PM
If we look at how similar the occupational mix of recent college graduates is to slightly older college graduates - there may be some sign that the difference is growing? But that could be due to 1) noise 2) non-AI factors (such as a slowing labor market), not just AI 5/
In fact, taking a closer look at recent years, the data suggests that this recent trend is not necessarily attributable to AI (Figure 2). Shifts in the occupational mix were well on their way during 2021. 4/
October 1, 2025 at 3:54 PM
In fact, taking a closer look at recent years, the data suggests that this recent trend is not necessarily attributable to AI (Figure 2). Shifts in the occupational mix were well on their way during 2021. 4/
Although recent trends seemingly outpace historical shifts in the occupational mix, the potential effects of AI on the labor market so far are not out of the ordinary. 3/
October 1, 2025 at 3:54 PM
Although recent trends seemingly outpace historical shifts in the occupational mix, the potential effects of AI on the labor market so far are not out of the ordinary. 3/
First, we look at how quickly the overall occupational mix is changing. The job mix for AI appears to be changing faster than it has in the past, although not markedly so. BUT... 2/
October 1, 2025 at 3:54 PM
First, we look at how quickly the overall occupational mix is changing. The job mix for AI appears to be changing faster than it has in the past, although not markedly so. BUT... 2/
Reposted by Martha Gimbel
3. "Fine, we'll just rely on private data."
(This one is @marthagimbel.bsky.social bait)
There is lots of private data out there, and we'll be leaning on it heavily during a shutdown. But hardly anyone who has looked closely at these data thinks they can come close to replacing government sources.
(This one is @marthagimbel.bsky.social bait)
There is lots of private data out there, and we'll be leaning on it heavily during a shutdown. But hardly anyone who has looked closely at these data thinks they can come close to replacing government sources.
September 30, 2025 at 1:42 PM
3. "Fine, we'll just rely on private data."
(This one is @marthagimbel.bsky.social bait)
There is lots of private data out there, and we'll be leaning on it heavily during a shutdown. But hardly anyone who has looked closely at these data thinks they can come close to replacing government sources.
(This one is @marthagimbel.bsky.social bait)
There is lots of private data out there, and we'll be leaning on it heavily during a shutdown. But hardly anyone who has looked closely at these data thinks they can come close to replacing government sources.