Category: Data Source
USA employment facts of the day
According to the Federal Reserve Bank of New York, the college majors with the lowest unemployment rates for the calendar year 2023 were nutrition sciences, construction services, and animal/plant sciences. Each of these majors had unemployment rates of 1% or lower among college graduates ages 22 to 27. Art history had an unemployment rate of 3% and philosophy of 3.2%…
Meanwhile, college majors in computer science, chemistry, and physics had much higher unemployment rates of 6% or higher post-graduation. Computer science and computer engineering students had unemployment rates of 6.1% and 7.5%, respectively…
Here is the full story. Why is this? Are the art history majors so employable? Or are their options so limited they don’t engage in much search and just take a job right away?
Via Rich Dewey.
New data on the political slant of AI models
By Sean J. Westwood, Justin Grimmer, and Andrew B. Hall:
We develop a new approach that puts users in the role of evaluator, using ecologically valid prompts on 30 political topics and paired comparisons of outputs from 24 LLMs. With 180,126 assessments from 10,007 U.S. respondents, we find that nearly all models are perceived as significantly left-leaning—even by many Democrats—and that one widely used model leans left on 24 of 30 topics. Moreover, we show that when models are prompted to take a neutral stance, they offer more ambivalence, and users perceive the output as more neutral. In turn, Republican users report modestly increased interest in using the models in the future. Because the topics we study tend to focus on value-laden tradeoffs that cannot be resolved with facts, and because we find that members of both parties and independents see evidence of slant across many topics, we do not believe our results reflect a dynamic in which users perceive objective, factual information as having a political slant; nonetheless, we caution that measuring perceptions of political slant is only one among a variety of criteria policymakers and companies may wish to use to evaluate the political content of LLMs. To this end, our framework generalizes across users, topics, and model types, allowing future research to examine many other politically relevant outcomes.
Here is a relevant dashboard with results.
You Can See the End of the Great Stagnation Everywhere but in the Productivity Statistics
Eli Dourado continues to keep his eye on the most important number in the world, total factor productivity. It continues to be bad, -3.88% on an annual basis for the first quarter of 2025. It’s too early for Trump’s tariffs to have made an effect and too early for AI.
You can see the end of the great stagnation everywhere but in the productivity statistics.
New results on Facebook advertising
There has been so much misinformation about this topic, much of it still persists. Here is a new paper by many researchers, Hunt Allcott and Matt Gentzkow are the first two names. Here is the abstract:
We study the effects of social media political advertising by randomizing subsets of 36,906 Facebook users and 25,925 Instagram users to have political ads removed from their news feeds for six weeks before the 2020 US presidential election. We show that most presidential ads were targeted toward parties’ own supporters and that fundraising ads were most common. On both Facebook and Instagram, we found no detectable effects of removing political ads on political knowledge, polarization, perceived legitimacy of the election, political participation (including campaign contributions), candidate favorability, and turnout. This was true overall and for both Democrats and Republicans separately.
Here is the full link.
Digital tech sentences to ponder
The first generation who engaged with digital technologies has reached the age where risks of dementia emerge. Has technological exposure helped or harmed cognition in digital pioneers?
…Use of digital technologies was associated with reduced risk of cognitive impairment (OR = 0.42, 95% CI 0.35–0.52) and reduced time-dependent rates of cognitive decline (HR = 0.74, 95% CI 0.66–0.84). Effects remained significant when accounting for demographic, socioeconomic, health and cognitive reserve proxies.
So maybe digital tech is not so bad for us after all? You do not have to believe the postulated relatively large effects, as the more likely conclusion is simply that, as in so many cases, treatment effect in the social sciences are small. That is from a recent paper by Jared F. Benge and Michael K. Scullin. Via the excellent Kevin Lewis.
Sentences to ponder
In a landmark 2013 paper, David Autor, David Dorn and Gordon Hanson found that America lost an average of 90,000 jobs per year between 1990 and 2007 because of imports from China. But put that in perspective. According to Strain, five million Americans currently separate from their employers per month. Plus, in a 2019 paper, Robert C. Feenstra, Hong Ma and Yuan Xu found that the China shock job losses were largely offset by job gains, owing to higher exports.
Here is more from David Brooks (NYT).
Economics coauthorships in the aftermath of MeToo
We study changes in coauthorships in economics, after the MeToo movement, using NBER and CEPR working papers between January 2004 and December 2020. We identify three main shifts in collaboration patterns. First, compared to pre-MeToo levels, collaborations across genders in an author’s seniority group increased: we estimate a 12.3% increase of women coauthors per 100 men-authored papers. Second, coauthorship shares of senior with junior economics declined by 3.0%, indicating a shift towards sorting of collaborations by seniority. Third, shares of new coauthorships declined by 5.4%, driven by drops in senior economists’ shares of new junior and new junior women by 18.4% and 48.0%, respectively. The results are robust to different specifications.
That is from a new paper by Noriko Amano-Patiño, Elisa Faraglia, and Chryssi Giannitsarou. Via the excellent Kevin Lewis. And here is a related paper on who receives credit for cross-gender co-authorships.
No Evidence of Effects of Testosterone on Economic Preferences
There is conflicting evidence on whether testosterone affects economic preferences such as risk taking, fairness and altruism, with the evidence suggesting significant effects coming from correlational studies or small underpowered testosterone administration studies. To credibly test this hypothesis, we conducted a large pre-registered double-blind randomized controlled trial with N = 1,000 male participants; 10–20 times larger than most previous randomized controlled studies. Participants were randomly allocated to receive a single dose of either placebo or intranasal testosterone. They thereafter carried out a series of economic tasks capturing social preferences, competitiveness and risk preferences. We fail to find any evidence of a treatment effect for any of our nine primary outcome measures, thereby failing to conceptually replicate several previous studies reporting positive findings that used smaller sample sizes. In line with these results, we furthermore find no evidence of an association between basal testosterone and economic preferences, failing to also conceptually replicate previous correlational studies.
The allocation of US AID funds
According to Marco Rubio only 12 cents of every dollar spent from USAID went to recipients, the other 88 cents went to NGOs who pocketed the money.
I tried to fact check that with o3:
However you draw the line, before 2017 well over half—and usually more like 75-90 percent—of USAID money was channelled through third-party NGOs, contractors, and multilateral agencies rather than handed straight to the governments or other local actors in the partner country.
I do support PEPFAR and the earlier vaccine programs, but perhaps those estimates have been underreported as of late? I do understand that not all third party allocations are wasteful, nonetheless something seems badly off here. Nor were many US AID defenders keen to deal with such estimates when the major debate was going on.
Changes in the College Mobility Pipeline Since 1900
By Zachary Bleemer and Sarah Quincy:
Going to college has consistently conferred a large wage premium. We show that the relative premium received by lower-income Americans has halved since 1960. We decompose this steady rise in ‘collegiate regressivity’ using dozens of survey and administrative datasets documenting 1900–2020 wage premiums and the composition and value-added of collegiate institutions and majors. Three factors explain 80 percent of collegiate regressivity’s growth. First, the teaching-oriented public universities where lower-income students are concentrated have relatively declined in funding, retention, and economic value since 1960. Second, lower-income students have been disproportionately diverted into community and for-profit colleges since 1980 and 1990, respectively. Third, higher-income students’ falling humanities enrollment and rising computer science enrollment since 2000 have increased their degrees’ value. Selection into college-going and across four-year universities are second-order. College-going provided equitable returns before 1960, but collegiate regressivity now curtails higher education’s potential to reduce inequality and mediates 25 percent of intergenerational income transmission.
An additional hypothesis is that these days the American population is “more sorted.” We no longer have the same number of geniuses going to New York city colleges, for instance. Here is the full NBER paper.
Spain fact of the day
By 2039, nearly 4 in 10 Spanish residents will be either immigrants themselves or the children of immigrants.
When combined, these figures imply that, by 2039, approximately 43% of Spain’s workforce — over one in four working-age individuals — will be either first or second-generation immigrants.
Here is the full story, via Mario.
The economics of sleep
Full-time, prime-age male workers in the top income quartile sleep around half an hour less per day than those in the lowest quartile.
At the macro level, average sleep duration decreases as a country’s GDP increases.
Higher-income individuals allocate more time to other leisure activities, such as social outings and internet usage, substituting sleep.
Here is the paper by Cristián Jara, Francisca Pérez, and Rodrigo Wagner. Via the excellent Kevin Lewis.
Politically correct LLMs
Despite identical professional qualifications across genders, all LLMs consistently favored female-named candidates when selecting the most qualified candidate for the job. Female candidates were selected in 56.9% of cases, compared to 43.1% for male candidates (two-proportion z-test = 33.99, p < 10⁻252 ). The observed effect size was small to medium (Cohen’s h = 0.28; odds=1.32, 95% CI [1.29, 1.35]). In the figures below, asterisks (*) indicate statistically significant results (p < 0.05) from two-proportion z-tests conducted on each individual model, with significance levels adjusted for multiple comparisons using the Benjamin-Hochberg False Discovery Rate correction…
In a further experiment, it was noted that the inclusion of gender concordant preferred pronouns (e.g., he/him, she/her) next to candidates’ names increased the likelihood of the models selecting that candidate, both for males and females, although females were still preferred overall. Candidates with listed pronouns were chosen 53.0% of the time, compared to 47.0% for those without (proportion z-test = 14.75, p < 10⁻48; Cohen’s h = 0.12; odds=1.13, 95% CI [1.10, 1.15]). Out of 22 LLMs, 17 reached individually statistically significant preferences (FDR corrected) for selecting the candidates with preferred pronouns appended to their names.
Here is more by David Rozado. So there is still some alignment work to do here? Or does this reflect the alignment work already?
Partisan Corporate Speech
We construct a novel measure of partisan corporate speech using natural language processing techniques and use it to establish three stylized facts. First, the volume of partisan corporate speech has risen sharply between 2012 and 2022. Second, this increase has been disproportionately driven by companies adopting more Democratic-leaning language, a trend that is widespread across industries, geographies, and CEO political affiliations. Third, partisan corporate statements are followed by negative abnormal stock returns, with significant heterogeneity by shareholders’ degree of alignment with the statement. Finally, we propose a theoretical framework and provide suggestive empirical evidence that these trends are at least in part driven by a shift in investors’ nonpecuniary preferences with respect to partisan corporate speech.
That is from a recent paper by William Cassidy and Elisabeth Kempf. Via the excellent Kevin Lewis.
Claims about police shootings
There is a new book by Tom S. Clark, Adam N. Glynn, and Michael Leo Owens, called Deadly Force: Police Shootings in Urban America. Here are a few of the conclusions:
…we were more likely to obtain records [on police shootings] from cities with women mayors and more women on municipal legislatures.
And, most interesting to me:
…we also found that fewer police shootings occurred in cities with more police, all else equal.
And:
…Black and Hispanic officers are disproportionately the ones involved in police shootings. That is particularly true when a Black civilian is the subject of the shooting.
I do not follow this area closely, but the book seems of interest.