Yosh Halberstam

Selected work

A compact, industry-friendly view of my research as applied measurement systems. For each project: what was measured, how it was measured, and why it matters for product, platforms, and AI deployment.

Looking for a full publication list and citations? Google Scholar.
1) Platform networks and information diffusion

Network structure, exposure, and diffusion in large social graphs

Relevance: recommender systems, virality, integrity, and growth dynamics
Homophily, Group Size, and the Diffusion of Political Information in Social Networks: Evidence from Twitter Journal of Public Economics (2016) · with Brian Knight
PDF
  • Builds large-scale network measurement to quantify how homophily and group size shape what people see and share.
  • Connects network structure to diffusion outcomes that matter for feed ranking, polarization risk, and information reach.
Methods: network measurement, reduced-form causal inference, robustness across specifications
2) Behavioral signal processing and measurement

Extracting behavioral features from unstructured signals

Relevance: ML pipelines, feature engineering, human behavior analytics
Gender Identity and Mode-Switching Behavior: Evidence from the Human Voice Working paper (last updated March 2023)
PDF
  • Builds an end-to-end pipeline to measure latent behavioral switching in voice frequency patterns.
  • Shows how identity, incentives, and context can be detected from high-dimensional behavioral signals.
Methods: signal processing, measurement design, large-scale observational analysis
3) Incentives, institutions, and distributional outcomes

Causal measurement in organizational and policy settings

Relevance: pricing & incentives, fairness, adoption, organizational design
Pay Transparency and the Gender Gap American Economic Journal: Applied Economics (2023) · with Michael Baker, Kory Kroft, Alexandre Mas, Derek Messacar
PDF
  • Evaluates how information policy (pay transparency) changes behavior and outcomes in labor markets.
  • Demonstrates how measurement can isolate causal effects in complex, real-world environments.
Methods: quasi-experimental causal inference, robustness and sensitivity checks
The Impact of Unions on the Wage Distribution: Evidence from Higher Education Forthcoming, AER: Insights · with Michael Baker, Kory Kroft, Alexandre Mas, Derek Messacar
PDF
  • Measures how institutional bargaining power shifts distributions, incentives, and within-firm outcomes.
  • Focus on distributional impacts and equilibrium responses — directly relevant to marketplace and policy design.
Methods: quasi-experimental inference, distributional analysis