Welcome back to TPI’s Research Roundup, our semi-regular compilation of recent outside research of interest to tech policy nerds. DISCLAIMER: Not all authors are affiliated with TPI. We do not necessarily agree with everything, or even anything, in these papers, but find them interesting.
In this Issue: Two new economics papers and an industry panel of economists from AI firms highlight critical questions about data integrity, digital access, and AI’s impacts on labor markets. The papers are from the NBER working paper series.
1. Data Integrity at the BLS
The Value of Reliable Statistics
Nicholas Bloom, Erica Groshen, Duncan Hobbs, and Michael Strain
Their question: What is the economic value of reliable government statistics?
Their answer: Using the August 2025 firing of the Bureau of Labor Statistics head Erika McEntarfer as a natural experiment, the authors estimate that reliable official statistics generate roughly $25 in economic benefits for every $1 of BLS budget (annual budget: $704 million). The disruption produced a sharp spike in economic policy uncertainty and an estimated $20 billion drag on growth and job creation in the near term. The mechanism is straightforward: when markets and policymakers can’t fully trust the numbers, they hedge their decisions, which reduces investment and hiring.
Why it matters: The $25-to-$1 return on the BLS budget is large enough that fiscal arguments for cutting statistical agencies don’t hold up on their own terms. Whatever one thinks about the underlying personnel decision, the paper suggests the institutional credibility of government statistics is itself an economic asset worth treating as one in budget discussions.
2. School Phone Bans and Student Outcomes
The Effects of School Phone Bans: National Evidence from Lockable Pouches
Hunt Allcott, E. Jason Baron, Thomas Dee, Angela Duckworth, Matthew Gentzkow, and Brian A. Jacob
Their question: What happens when schools restrict phone access during instructional time?
Their answer: New empirical evidence from a study of school phone bans shows modest positive effects from restricting phone access since 2023. The scholars find that students report modest improvements in psychological well-being and high school math test scores show positive effects. However, the study finds little evidence of improvement in broader test performance, school attendance, or self-reported in-class attention over the first three years. Online bullying measures show no effects, suggesting phone restrictions do not alter perceived online harassment.
Why it matters: To the extent that Universal Service Fund subsidies support broadband in schools through the E-rate program, this evidence directly informs spending priorities. Students access school Wi-Fi networks through their phones and computers during the school day. Whether digital connectivity meaningfully improves educational outcomes—and under what conditions—should shape how E-rate dollars are allocated. These findings suggest that simply providing access is necessary but not sufficient; how and when students can use digital tools matters as much as whether they can.
3. Labor Market Trends Tracked by AI Firms
AI and Economic Measurement Workshop at Stanford, Spring 2026
On May 7, the National Bureau of Economic Research and Stanford Institute for Economic Policy Research held their annual AI and Economic Measurement workshop, hosted by Erik Brynjolfsson and Karen Dynan. A panel of economists from OpenAI, Anthropic, LinkedIn, and Zillow shared how they study industry data to understand AI’s labor market effects. Watch the panel here: https://www.youtube.com/live/G29Qb6vRa8o?si=FGGmgrqfrMbeASj2&t=26770.
Their question: What do leading firms see in their data about AI’s effects on labor markets?
Their answer:
- LinkedIn finds hiring down across both AI-exposed and insulated occupations, making it hard to identify any single occupation that is disappearing, but AI-forward firms have different talent composition than traditional firms (fewer recruiters and marketers, more go-to-market specialists). Founder titles have grown 50 to 70 percent year over year in some countries, and demand for human skills is rising alongside demand for AI skills.
- OpenAI found that the productive use of AI is highly concentrated, with frontier firms (95th percentile) using roughly six times the tokens of the median firm overall and sixteen times for Codex.
- Anthropic’s Economic Index samples millions of Claude transcripts and maps them to O*NET tasks and occupations, offering a way to observe what kinds of work users are attempting with AI. Panelists emphasized that translating AI exposure into occupational employment effects remains difficult: even within heavily exposed occupations such as software engineering, outcomes can vary widely across firms and teams.
- Zillow framed the broader puzzle as “pilot purgatory”: firms are running AI pilots that don’t scale into measurable productivity gains because of organizational frictions and process redesign.
- Several panelists pointed to new business formation as a potentially better indicator of AI’s productivity effects than headline employment statistics.
Why it matters: The most striking finding is that extensive measurement across four firms showed no single occupation disappearing and no real increase in headline productivity growth. The evidence does not yet support the strongest public claims about rapid AI-driven displacement. What it does show is more subtle: changing job composition inside firms, a surge in new business formation, and shifted entry-level hiring. Better measurement is the prerequisite for answering questions about AI’s effects.
If you’ve read a paper you think might be interesting to include in the next Roundup, feel free to send it to us at [email protected].
Sarah Oh Lam is a Senior Fellow at the Technology Policy Institute. Oh completed her PhD in Economics from George Mason University, and holds a JD from GMU and a BS in Management Science and Engineering from Stanford University. She was previously the Operations and Research Director for the Information Economy Project at George Mason School of Law. She has also presented research at the 39th Telecommunications Policy Research Conference and has co-authored work published in the Northwestern Journal of Technology & Intellectual Property among other research projects. Her research interests include law and economics, regulatory analysis, and technology policy.
