Ironically, Gemini pro is getting very good at solving maths problems, but terrible at things that require search: i.e. repeatedly confuses academic papers for the wrong one even when supplied with the name and year of the paper, incorrect answers to things easily searchable on the web.
July 22, 2025 at 1:53 AM
Ironically, Gemini pro is getting very good at solving maths problems, but terrible at things that require search: i.e. repeatedly confuses academic papers for the wrong one even when supplied with the name and year of the paper, incorrect answers to things easily searchable on the web.
March 5, 2025 at 2:29 AM
I don't think we will eat out much more as we get richer necessarily. It's a labor intensive activity with few efficiencies to be gained via technology (post delivery apps). Costs will rise in line with general wages. We eat out less than poorer Asian economies where labor and food out is cheap.
January 5, 2025 at 1:14 AM
I don't think we will eat out much more as we get richer necessarily. It's a labor intensive activity with few efficiencies to be gained via technology (post delivery apps). Costs will rise in line with general wages. We eat out less than poorer Asian economies where labor and food out is cheap.
GDPNow will take a hammering today
January 2, 2025 at 3:52 PM
GDPNow will take a hammering today
No you have to scrape them afaik
December 10, 2024 at 11:17 PM
No you have to scrape them afaik
Lots of volatile components of unemployment but this continued rising trend is interesting
December 6, 2024 at 5:08 PM
Lots of volatile components of unemployment but this continued rising trend is interesting
Browsing the Skyline
December 2, 2024 at 3:42 AM
Browsing the Skyline
Biden did this in 2024 alongside tighter asylum eligibility. But only after it became clear it was a huge electoral liability and without any time for it to make a dent in perceptions.
December 1, 2024 at 10:35 PM
Biden did this in 2024 alongside tighter asylum eligibility. But only after it became clear it was a huge electoral liability and without any time for it to make a dent in perceptions.
It played almost no role in dem messaging for IRA/CHIPS in the runup to the election. Why no "bring manufacturing jobs home", "putting us on the tech frontier" messaging? Crazy.
December 1, 2024 at 12:42 AM
It played almost no role in dem messaging for IRA/CHIPS in the runup to the election. Why no "bring manufacturing jobs home", "putting us on the tech frontier" messaging? Crazy.
Gotta be the biggest export
November 29, 2024 at 7:24 PM
Gotta be the biggest export
On wage growth, the ECI is back to 2018-9 rates of growth, but there's something funky going on with wages and salaries in Q1 that is pushing up annual unit labor costs. If it gets revised away, cost pressures are going to look more benign.
November 27, 2024 at 6:35 PM
On wage growth, the ECI is back to 2018-9 rates of growth, but there's something funky going on with wages and salaries in Q1 that is pushing up annual unit labor costs. If it gets revised away, cost pressures are going to look more benign.
Good point. Market-based core PCE (excluding imputed prices like portfolio fees) coming in a lot closer to target recently.
November 27, 2024 at 4:16 PM
Good point. Market-based core PCE (excluding imputed prices like portfolio fees) coming in a lot closer to target recently.
A very different pattern than the quits data, which also proxies job-switching. No false signal after 2015 and still falling sharply.
November 27, 2024 at 3:35 AM
A very different pattern than the quits data, which also proxies job-switching. No false signal after 2015 and still falling sharply.
Don't we need a Trump TS reposter on BlueSky?
November 26, 2024 at 3:58 AM
Don't we need a Trump TS reposter on BlueSky?
Well they also measure inventory building as a positive contribution to GDP...
November 26, 2024 at 1:55 AM
Well they also measure inventory building as a positive contribution to GDP...
And all the more amazing that QCEW is pointing to downwards jobs revisions! It doesn't seem like the births/deaths model like in '09.
November 25, 2024 at 2:51 AM
And all the more amazing that QCEW is pointing to downwards jobs revisions! It doesn't seem like the births/deaths model like in '09.
A strong activity index, but continued weakness in hiring. The hiring/employment index has been at 50 or below for 6 months now, with a big downshift in payrolls over the same period.
* Best services/composite PMI since 2022
* U.S SERVICES PMI (NOV) ACTUAL: 57.0 VS 55.0 PREVIOUS; EST 55.0
* U.S S&P GLOBAL COMPOSITE PMI (NOV) ACTUAL: 55.3 VS 54.1 PREVIOUS; EST 54.3
* U.S MANUFACTURING PMI (NOV) ACTUAL: 48.8 VS 48.5 PREVIOUS; EST 48.9
* U.S SERVICES PMI (NOV) ACTUAL: 57.0 VS 55.0 PREVIOUS; EST 55.0
* U.S S&P GLOBAL COMPOSITE PMI (NOV) ACTUAL: 55.3 VS 54.1 PREVIOUS; EST 54.3
* U.S MANUFACTURING PMI (NOV) ACTUAL: 48.8 VS 48.5 PREVIOUS; EST 48.9
November 22, 2024 at 3:10 PM
A strong activity index, but continued weakness in hiring. The hiring/employment index has been at 50 or below for 6 months now, with a big downshift in payrolls over the same period.
I haven't seen anything substantive from the BLS on this issue, but I think that reporting bulk headcounts is very different than tax payments for individuals. But regardless, something has caused massive sampling difference, and it doesn't look like births/deaths can explain that much of it.
November 20, 2024 at 11:28 PM
I haven't seen anything substantive from the BLS on this issue, but I think that reporting bulk headcounts is very different than tax payments for individuals. But regardless, something has caused massive sampling difference, and it doesn't look like births/deaths can explain that much of it.
But intuitively, since the payroll survey is just a one page form you fill in headcount and pay on, you can imagine it being more revealing about undocumented workers. It's also hard to understand how real consumption is topping 3% if true empl. growth is as bad as QCEW says.
November 20, 2024 at 6:38 PM
But intuitively, since the payroll survey is just a one page form you fill in headcount and pay on, you can imagine it being more revealing about undocumented workers. It's also hard to understand how real consumption is topping 3% if true empl. growth is as bad as QCEW says.
I'm not sure I understand "they are on the books or not". Employers need to pay undocumented workers, and keep a record of their headcount and pay. They also need to not reveal them to authorities that provide source data to QCEW. There's no evidence either way on how prevalent the difference is.
November 20, 2024 at 6:36 PM
I'm not sure I understand "they are on the books or not". Employers need to pay undocumented workers, and keep a record of their headcount and pay. They also need to not reveal them to authorities that provide source data to QCEW. There's no evidence either way on how prevalent the difference is.
The elephant in the room is that QCEW probably not capturing lots of undocumented workers while payrolls are (to some extent at least). And we don't really know by how much.
November 20, 2024 at 4:39 PM
The elephant in the room is that QCEW probably not capturing lots of undocumented workers while payrolls are (to some extent at least). And we don't really know by how much.
It depends how the higher net tax revenues from dropping the rebate are spent doesn't it? Freed up fiscal space for stimulus would be a trade imbalances reducing measure.
November 19, 2024 at 4:49 AM
It depends how the higher net tax revenues from dropping the rebate are spent doesn't it? Freed up fiscal space for stimulus would be a trade imbalances reducing measure.
New vacancies still coming down but above pre-covid. Actual hiring (JOLTS) already below pre-covid rate and still falling. Hard to reconcile with booming consumption, but a big dilemma: is this whiplash from over-hiring in 2023 or something more sinister?
November 19, 2024 at 4:26 AM
New vacancies still coming down but above pre-covid. Actual hiring (JOLTS) already below pre-covid rate and still falling. Hard to reconcile with booming consumption, but a big dilemma: is this whiplash from over-hiring in 2023 or something more sinister?
Not great, certainly not terrible.
November 19, 2024 at 3:45 AM
Not great, certainly not terrible.
Reposted by Altered Bagehot
If you’d told me in 2012 that the next 12 years would look like this, I wouldn’t have believed you.
November 17, 2024 at 8:09 PM
If you’d told me in 2012 that the next 12 years would look like this, I wouldn’t have believed you.