Yosh Halberstam
Causal inference and measurement for labor markets, AI and public policy
PhD economist with 15 years of research at the University of Toronto and the FDIC. I study how incentives, institutions and information shape behavior, using administrative, survey, social-network and audio data. I also evaluate the economic reasoning of generative AI models.
Causal inference
Labor economics
Political economy
Social networks
Survey methodology
AI evaluation
2.2M
Twitter users in network data
90M
Network connections
40K
Lawyers' voicemail greetings
15 yrs
Research in academia and government
Based in: Washington, DC
Email: yosh.halberstam@gmail.com
LinkedIn: yoshhalberstam
Focus: Labor markets, AI and public policy
Selected work
- Political information: majority-group users get more information, sooner, and like-minded information reaches like-minded users faster (Journal of Public Economics).
- Voice and behavior: about a third of women lawyers switch between two pitch ranges within a greeting, a pattern averages hide (working paper).
- Pay transparency: salary disclosure laws narrowed the gender pay gap among Canadian faculty by 20 to 40 percent (AEJ: Applied).
- Unions: faculty salaries rose about 10 percent at the bottom of the distribution; the top was unaffected (AER: Insights).
- AI evaluation: expert economics and finance tasks, reference solutions and rubrics for evaluating model reasoning.
What I’m looking for
- Research roles on information, AI and labor markets: AI labs, research and economics teams at tech and AI companies, and policy research institutes.
- Washington, DC or remote.