NewsMacroMarginal Revolution's Thursday Assorted Links: Publication Bias, UCLA Deficits, SF Dishwashers, Treasuries, and AI Alignment

Marginal Revolution's Thursday Assorted Links: Publication Bias, UCLA Deficits, SF Dishwashers, Treasuries, and AI Alignment

Author: Marginal Revolution·

Key Takeaways

  • •A new JEL paper reports that statistically significant results are 30 times more likely to be published in experimental economics, and fewer than 2% of empirical political science papers report null-only findings in abstracts.
  • •UCLA's athletic department lost $52 million last year and $83 million in 2024 when including the campus subsidy, pressured by rising salaries, facilities spending, athlete compensation, and new Big Ten travel costs.
  • •A report found most laid-off San Francisco tech workers are not qualified for a $50-per-hour dishwasher job at a local Vietnamese restaurant.
  • •Steve Williamson published an analysis of the US Treasuries market, which has drawn attention amid heavy issuance and debate over term premiums.
  • •A discussion explores applying mechanism design, the incentive-engineering field recognized by the 2007 Nobel memorial prize, to AI alignment.
Marginal Revolution's Thursday Assorted Links: Publication Bias, UCLA Deficits, SF Dishwashers, Treasuries, and AI Alignment

Marginal Revolution's latest Thursday assorted links roundup touches on econometrics, college athletics finance, Papua New Guinea, the San Francisco labor market, the US Treasury market, and AI alignment.

1. Publication bias in economics. Jon Fiva highlights striking statistics from a new paper in the Journal of Economic Literature (JEL): instrumental variable (IV) estimates are on average 3–10 times larger than OLS estimates according to a meta-analysis; statistically significant results are 30 times more likely to be published in experimental economics (per Andrews & Kasy); and fewer than 2% of empirical political science papers report null-only findings in their abstracts (per Briggs et al.). The numbers feed a long-running concern in empirical economics: journals' preference for significant findings may distort the published evidence base, which is why practices like pre-registration and replication have gained traction in the field.

Some crazy stats from this new JEL paper: – IV est. average 3–10x OLS (meta analysis) – Sign. results 30x more likely to be published in experimental econ (Andrews & Kasy) – <2% of empirical polisci report null-only findings in abstracts (Briggs et al.)

2. UCLA athletics finances. The UCLA athletic department lost $52 million last year, and $83 million in 2024 when including the campus subsidy. The shortfalls come as major-college athletics finances face mounting pressure from rising coaching salaries, facility spending, and athlete compensation following the NCAA's name-image-likeness changes; UCLA also took on new travel costs after joining the Big Ten in 2024.

The UCLA athletic dept lost $52 mil last year and $83 mil in 2024 when including campus subsidy.

3. Papua New Guinea fact of the day. A featured fact about Papua New Guinea.

Papua New Guinea fact of the day

4. San Francisco tech workers and a $50 dishwasher job. Most laid-off San Francisco tech workers are not qualified for a $50-per-hour dishwasher job, according to a report on a local Vietnamese restaurant's hiring struggles. The item is a small window into a bigger question in the current labor market: how readily displaced white-collar workers can shift into other sectors, even relatively high-paying service jobs that demand physical stamina and schedule availability.

Most laid-off SF tech workers aren't qualified for this $50 dishwasher job.

5. The US Treasuries market. Steve Williamson examines what is going on in the US Treasuries market. The Treasury market — the world's largest government bond market and a benchmark for global borrowing costs — has drawn heightened attention amid heavy issuance and debate over term premiums, making such analysis useful context for anyone tracking funding conditions.

What's going on in the US Treasury market?

6. Mechanism design for AI alignment. A discussion of how mechanism design — a field of economics concerned with designing rules and institutions that align incentives, recognized with the 2007 Nobel memorial prize to Leonid Hurwicz, Eric Maskin, and Roger Myerson — can be applied to AI alignment. The idea is that the same incentive-engineering toolkit used for auctions and matching markets may help structure how AI systems and their operators interact.

Mechanism design for AI alignment