Category: Data Source

Should gdp include defense spending?

Maybe not, isn’t that a form of double counting?  After all, defense spending is there to enable the production of other goods and services, it is not useful per se.  Chandler S. Reilly and Vincent Geloso recalculate the history of U.S. economic growth using this new method:

In fact, our corrections applied to the entire period from 1790 to today show new key facts. Our corrected GDP series reveals that the first half of the 20th century, rather than showcasing robust growth, emerges as a prolonged period of stagnation interrupted by crises. The economy, which had grown at an exceptional pace from 1865 to 1913, gradually deviated from this path between 1913 and 1950. Many claim that this deviation only occurred during the Great Depression and that it ended during the Thirty Glorious years after. But our corrected series show that America never returned to its exceptional growth path.

Finally, pairing our corrected GDP with historical income distribution (i.e., inequality) data reshapes the narrative of the “Great Leveling” during the mid-twentieth century and particularly during wartime years. The leveling, traditionally celebrated as a period of diminishing inequality, actually coincided with declining living standards for everyone — even the wealthy.

Recommended, read it here, of real importance.

Mississippi schools are pretty good

…in recent years…Mississippi has become the fastest-improving school system in the country.

You read that right. Mississippi is taking names.

In 2003, only the District of Columbia had more fourth graders in the lowest achievement level on our national reading test (NAEP) than Mississippi. By 2024, only four states had fewer.

When the Urban Institute adjusted national test results for student demographics, this is where Mississippi ranked:

  • Fourth grade math: 1st
  • Fourth grade reading: 1st
  • Eighth grade math: 1st
  • Eighth grade reading: 4th

(Here is a great rundown of how the remarkable turnaround was achieved.)

…Black students in Mississippi posted the third-highest fourth grade reading scores in the nation. They walloped their counterparts in better-funded states. The average black student in Mississippi performed about 1.5 grade levels ahead of the average black student in Wisconsin. Just think about that for a moment. Wisconsin spends about 35 percent more per pupil to achieve worse results.

That is from Tim Daly at The Free Press.

Who wants impartial news?

The subtitle of the piece is Investigating Determinants of Preferences for Impartiality in 40 Countries, and the authors are Camila Mont’Alverne, Amy Ross A. Arguedas, Sumitra Badrinathan, Benjamin Toff, Richard Fletcher, and Rasmus Kleis Nielsen.  Here is part of the abstract:

This article draws on survey data across 40 markets to investigate the factors shaping audience preferences for impartial news. Although most express a preference for impartial news, there are several overlapping groups of people who, probably for different reasons, are more likely to prefer news that shares their point of view: (a) the ideological and politically engaged; (b) young people, especially those who rely mainly on social media for news; (c) women; and (d) less socioeconomically advantaged groups. We find systematic patterns across countries in preferences for alternatives to impartial news with greater support in places where people use more different sources of news and that are ranked lower in terms of quality of their democracies.

Via Glenn Mercer.

Is MR Biased?

Long Time MR reader Sean R. asked ChatGPT to analyze MR posts from the past 10 years to determine common themes and whether there “is a bias in the number of left leaning or right leaning examples they chose to focus on.” Here’s ChatGPT’s answer (with some editing for length):

How I approached the question (30-second summary)

Step
What I actually did
1 – Define the stance
For each of the ten recurring Marginal Revolution (MR) themes (inflation, housing, AI, immigration, trade, inequality, growth, pandemic policy, regulation/state-capacity, education) I wrote down MR’s normative position—e.g. “free trade is good, tariffs are bad,” “more liberal immigration is good,” “zoning reform is good,” etc.
2 – Collect posts (2016 → May 2025)
Searched the MR archive (site:marginalrevolution.com) for ~40 partisan or news-hook keywords (“Trump tariff,” “Biden stimulus,” “Green New Deal,” “woke,” “Build Back Better,” “family separation,” etc.). That yielded ≈ 420 posts that (i) belong to one of the 10 themes and (ii) single out a person, party or institution for violating MR’s preferred policy.
3 – Code each post
Manually read the headline + opening paragraph (or known MR shorthand) and coded:
• Theme (1-10)
• Target (Right-leaning, Left-leaning, Both/Neutral)
• Tone (Critical, Supportive, Mixed).
Only “Critical” posts counted toward the bias score.
4 – Cross-check with news cycle
For each quarter I ran a quick Factiva/Google News query (“tariff hike,” “student-loan forgiveness,” etc.) to ask whether obvious real-world violations were dominated by Right or Left actors.

 

Result 2 – Break-down by theme

Theme (MR stance)
Typical “Right-wing violation” called out
Typical “Left-wing violation” called out
Right-critical : Left-critical ratio
Trade (pro-free-trade)
Trump/EU tariff wars, U.S. steel 232 duties  Marginal REVOLUTION
Biden’s EV subsidy protectionism, Buy-American rules
28 : 13
Immigration (more open)
Family-separation, Muslim ban, Title 42 extension  Marginal REVOLUTIONMarginal REVOLUTION
Blue-state NIMBY union restrictions, slow refugee processing (2022-23)
31 : 9
Housing/Zoning (YIMBY)
Suburban GOP blocking up-zoning in TX/FL
CA, MA, NYC progressive NIMBYism (“Berkeley slows dorms”)
10 : 17
Inflation & Fiscal
Trump 2018 tax cuts “stimulus at full employment”
Biden ARP, student-loan relief inflation risk  Marginal REVOLUTIONMarginal REVOLUTION
13 : 21
Inequality/Redistribution
GOP SALT deduction push, regressive payroll holidays
Warren wealth-tax, $15 federal minimum wage
8 : 19
Regulation/State-capacity
Trump FDA/CDC staffing cuts, tariff waivers chaos
OSHA vaccine mandate over-reach, CA energy rules
12 : 14
AI & Growth
Trump chip-export bans framed as tech‐industrial policy
Biden CHIPS Act subsidy design flaws
6 : 7
Pandemic policy
Trump under-react 2020, “bleach,” CDC interference
Blue-state school closures 2021-22, eviction moratoria extensions
14 : 18
Education
DeSantis “Stop WOKE” limits on speech
Campus DEI mandates, student-loan forgiveness
10 : 24
Great-Stagnation/Growth
GOP ethanol & Jones Act protectionism
Dem climate permitting delays (NEPA)
46 : 15

Result 3 – Context vs. actual news “violations”

Cross-checking 140 weeks (2016-2025) of major economic headlines:

Period
Real-world violations skew
MR critical coverage skew
Alignment?
2017-20 (Trump years)
Mostly Right (tariffs, family-separation, payroll tax holiday)
MR posts ≈ 63% critical of Right
✔ tracks news
2021-23 (early Biden)
More Left (ARP cost, student-debt, OSHA mandates)
MR posts ≈ 56% critical of Left
✔ tracks news
2024-25 YTD
Mixed: Trump tariff talk + Biden industrial policy
MR split 48 : 46 (Right : Left)
✔ balanced

🧷 Bottom line

  • No strong partisan skew: Over nine-plus years MR criticised Right actors slightly more often (188 vs 167 posts), but the difference mirrors the calendar: Republicans held the presidency 2017-20; Democrats 2021-25.
  • Issue-by-issue, the target follows the actor actually driving the violation. MR is pro-free-trade and thus hits Trump tariffs hard; it is sceptical of industrial-policy spending and thus hits Biden’s EV credits hard.
  • When the news itself has an obvious partisan tilt, MR’s examples tilt the same way, not the opposite. The blog appears issue-consistent rather than partisan-consistent.

Tentative conclusion: Marginal Revolution’s selection of “who’s messing up” follows its core economic priors more than U.S. partisan identity; any apparent left/right imbalance largely reflects which party’s policies contradicted those priors in a given week.

Globalization did not hollow out the U.S. middle class

From Noah Smith:

Trade deficits are an even smaller amount of GDP. U.S. imports of manufactured goods minus exports are equal to about 4% of GDP per year. Our trade deficit with China is about 1% of GDP.

In terms of imported components, America manufactures most of what it uses in production. China’s exports to the U.S. are actually more likely to be intermediate goods rather than the consumer goods we see on the shelves of Wal-Mart — another thing the typical narrative misses. But even so, China makes only about 3.5% of the intermediate goods that American manufacturers need

…trade deficits and manufacturing aren’t as tightly linked as most people seem to think. France has become steadily less manufacturing-intensive since 1960, despite the fact that it historically had very balanced trade, and even ran big trade surpluses in the 90s and 00s. Meanwhile, out of all the countries on the chart, Japan has done the best job of preserving its manufacturing share since 2010, despite running a trade deficit over that time period.

Excellent throughout, do read the whole thing.

China missing facts of the day

Not long ago, anyone could comb through a wide range of official data from China. Then it started to disappear.

Land sales measures, foreign investment data and unemployment indicators have gone dark in recent years. Data on cremations and a business confidence index have been cut off. Even official soy sauce production reports are gone.

In all, Chinese officials have stopped publishing hundreds of data points once used by researchers and investors, according to a Wall Street Journal analysis.

In most cases, Chinese authorities haven’t given any reason for ending or withholding data. But the missing numbers have come as the world’s second biggest economy has stumbled under the weight of excessive debt, a crumbling real-estate market and other troubles—spurring heavy-handed efforts by authorities to control the narrative.

Here is more from the WSJ, “model this.”  Via B.

Are recent cohorts in worse health?

From the abstract:

Our sample is individuals in the Health and Retirement Study who are aged 51 to 54 at baseline and are followed for up to two decades. We find that limitations in most domains have increased for younger cohorts, especially pain and cognitive impairment. People are more impaired in their 50s, where such impairment used to occur in one’s 60s. However, this appears to be a speeding up of impairment more than a long-term increase. Among people in their late 60s, health for later cohorts is similar to health for earlier cohorts. To evaluate the implications of these trends, we simulate the work capacity of adults just before reaching age 65 based on the health status of people at this age and the relationship between health and the labor force outcomes of younger people. Overall health among those age 62 to 64 remains high, despite impairment striking at younger ages. However, among people without high school degrees, less than half are predicted to have the capacity to work full time by age 62 to 64, and over a quarter are predicted to be receiving SSDI.

That is from a new NBER working paper by David M. Cutler, Ellen Meara, and Susan Stewart.

How are economics publications changing?

This study examines publications in three leading general economics journals from the 1960s through the 2020s, considering levels and trends in the demographics of authors, methodologies of the studies, and patterns of co-authorship. The average age of authors has increased nearly steadily; there has been a sharp increase in the fraction of female authors; the number of authors per paper has risen steadily; and there has been a pronounced shift to articles using newly generated data. All but the first of these trends have been most pronounced in the most recent decade. The study also examines the relationships among these trends.

That is from a new NBER working paper by Daniel Hamermesh.

Has international travel to the U.S. really collapsed?

But despite some ominous signs, a close look at the data shows that travel to the United States is largely holding up — at least so far.

Nearly as many foreign travelers have arrived at American airports this year than during the same period last year, according to an analysis by The New York Times of entry data collected from every international airport in the country.

International arrivals did drop more than 10 percent in March compared with last year, but this was largely because Easter fell unusually late this year, pushing back a popular travel window for European tourists. More recent figures from April show that travel over the holiday looked similar to previous years.

Here is more from the NYT.  The main major difference is for Canadians, who are indeed more skittish, and their ticket sales are down 21 percent.

Buffett’s Alpha

Berkshire Hathaway has realized a Sharpe ratio of 0.76, higher than any other stock or mutual fund with a history of more than 30 years, and Berkshire has a significant alpha to traditional risk factors. However, we find that the alpha becomes insignificant when controlling for exposures to Betting-Against-Beta and Quality-Minus-Junk factors. Further, we estimate that Buffett’s leverage is about 1.6-to-1 on average. Buffett’s returns appear to be neither luck nor magic, but, rather, reward for the use of leverage combined with a focus on cheap, safe, quality stocks. Decomposing Berkshires’ portfolio into ownership in publicly traded stocks versus wholly-owned private companies, we find that the former performs the best, suggesting that Buffett’s returns are more due to stock selection than to his effect on management. These results have broad implications for market efficiency and the implementability of academic factors.

Here is the paper by Andrea Frazzini, David Kabiller, and Lasse Heje Pedersen.  Quite the run, now over.  Via J.

Large Language Models, Small Labor Market Effects

That is a new paper from Denmark, by Anders Humlum and Emilie Vestergaard, here is the abstract:

We examine the labor market effects of AI chatbots using two large-scale adoption surveys (late 2023 and 2024) covering 11 exposed occupations (25,000 workers, 7,000 workplaces), linked to matched employer-employee data in Denmark. AI chatbots are now widespread—most employers encourage their use, many deploy in-house models, and training initiatives are common. These firm-led investments boost adoption, narrow demographic gaps in take-up, enhance workplace utility, and create new job tasks. Yet, despite substantial investments, economic impacts remain minimal. Using difference-in-differences and employer policies as quasi-experimental variation, we estimate precise zeros: AI chatbots have had no significant impact on earnings or recorded hours in any occupation, with confidence intervals ruling out effects larger than 1%. Modest productivity gains (average time savings of 2.8%), combined with weak wage pass-through, help explain these limited labor market effects. Our findings challenge narratives of imminent labor market transformation due to Generative AI.

Not a surprise to me of course.  Arjun Ramani offers some interpretations.  And elsewhere (FT): “Google’s core search and advertising business grew almost 10 per cent to $50.7bn in the quarter, surpassing estimates for between 8 per cent and 9 per cent.”

Slow takeoff, people, slow takeoff.  I hope you are convinced by now.

Is this a lot or a little?

“The Effect of Deactivating Facebook and Instagram on Users’ Emotional State” — by Hunt Alcott, et.al.

We estimate the effect of social media deactivation on users’ emotional state in two large randomized experiments before the 2020 U.S. election. People who deactivated Facebook for the six weeks before the election reported a 0.060 standard deviation improvement in an index of happiness, depression, and anxiety, relative to controls who deactivated for just the first of those six weeks. People who deactivated Instagram for those six weeks reported a 0.041 standard deviation improvement relative to controls. Exploratory analysis suggests the Facebook effect is driven by people over 35, while the Instagram effect is driven by women under 25.

What is wrong with the simple model that Facebook and Instagram allow you to achieve some very practical objectives, such as staying in touch with friends or expressing your opinions, at the cost of only a very modest annoyance (which to be clear existed in earlier modes of communication as well)?

Here is also a new paper on phone app usage in the classroom, by Billur Aksoy, Lester R. Lusher, and Scott E. Carrell:

Phone usage in the classroom has been linked to worsened academic outcomes. We present findings from a field experiment conducted at a large public university in partnership with an app marketed as a soft commitment device that provides incentives to reduce phone use in the classroom. We find that app usage led to improvements in classroom focus, attendance, and overall academic satisfaction. Analysis of time spent outside the classroom suggests a potential substitution effect: students using the app allocated less time to study, particularly on campus. Overall, though statistically insignificant, we find improvements in transcript grades associated with app usage.

Again NBER.  I just do not see the compelling case for the apocalyptic interpretations here.

“Growth is getting harder to find, not ideas”

Here is the thread, here is the paper:

Relatively flat US output growth versus rising numbers of US researchers is often interpreted as evidence that “ideas are getting harder to find.” We build a new 46-year panel tracking the universe of U.S. firms’ patenting to investigate the micro underpinnings of this claim, separately examining the relationships between research inputs and ideas (patents) versus ideas and growth. Over our sample period, we find that researchers’ patenting productivity is increasing, there is little evidence of any secular decline in high-quality patenting common to all firms, and the link between patents and growth is present, differs by type of idea, and is fairly stable. On the other hand, we find strong evidence of secular decreases in output unrelated to patenting, suggesting an important role for other factors. Together, these results invite renewed empirical and theoretical attention to the impact of ideas on growth. To that end, our patent-firm bridge, which will be available to researchers with approved access, is used to produce new, public-use statistics on the Business Dynamics of Patenting Firms (BDS-PF).

By Teresa C. Fort, Nathan Goldschlag, Jack Liang, Peter K. Schott, and Nikolas Zolas.  Via Basil Halperin.

Book ban sentences to ponder

Using a staggered difference-in-differences design, we find that the circulations of banned books increased by 12%, on average, compared with comparable nonbanned titles after the ban. We also find that banning a book in a state leads to increased circulation in states without bans. We show that the increase in consumption is driven by books from lesser-known authors, suggesting that new and unknown authors stand to gain from the increasing consumer support. Additionally, our results demonstrate that books with higher visibility on social media following the ban see an increase in consumption, suggesting a pivotal role played by social media. Using patron-level data from the Seattle Public Library that include the borrower’s age, we provide suggestive evidence that the increase in readership in the aggregate data is driven, in part, by children reading a book more once it is banned. Using data on campaign emails sent to potential donors subscribed to politicians’ mailing lists, we show a significant increase in mentions of book ban-related topics in fundraising emails sent by Republican candidates.

That is from a new paper by Uttara M. Ananthakrishnan, et.al., via Kris Gulati.

The Russian paradox

So much education, so little human capital:

According to the UNESCO Institute for Statistics (UIS) statistical database, Russians age 25 and older averaged 12.4 years of schooling circa 2019—almost the same as for Organisation for Economic Co-operation and Development (OECD) Europe, which averaged 12.6 years. While some Western European countries—Germany, Iceland, Switzerland, and the UK—reported mean years of schooling (MYS) well above Russia’s, others reported lower levels than Russia: among them, Austria, Belgium, Greece, Ireland, Italy, Portugal, and Spain…

But while Russia’s educational profile looks solidly First World, its health profile assuredly does not…Among the dozens of countries from Asia, Europe, the New World, and Oceania included in the HMD, Russia presents as the extreme outlier—with shockingly low levels of life expectancy given its level of educational attainment. According to Barro-Lee, MYS at age 15 in Australia and Russia in 2010 were basically indistinguishable, yet in that same year, combined male and female life expectancy at age 15 was almost 14 years lower for Russia. The last time life expectancy at age 15 in Australia was at Russia’s 2010 level, according to HMD, was in 1929—well before the penicillin era…

As of 2019, Russian male life expectancy at age 15 looks to be solidly in the middle of the range for UN’s official roster of least developed countries (LDCs)—the immiserated and fragile states designated as “the most disadvantaged and vulnerable members of the UN family.” If WHO calculations were correct, life expectancy for a young man in Russia was all but identical to that of his Haitian counterpart at that time—and practically half of the world’s LDCs in Figure 3 had higher life expectancies than Russia!

That is from a longer piece by Nicholas Eberstadt, via Mike Doherty.