Matching with Learning: A Bayesian Assignment Game Approach

Matching with Learning: A Bayesian Assignment Game Approach — Saeed Najafi-Zangeneh and Olivier Gossner, Mathematical Social Sciences (2026).
Authors

Saeed Najafi-Zangeneh

Olivier Gossner

← All research

Mathematical Social Sciences, 102583 (2026), available online

Available online 8 September 2026. Article 102583; journal pre-proof.

Summary

The paper examines matching between workers and firms when the surplus from a match is learned through public signals. Posterior beliefs define an assignment game at each date, allowing the analysis of stable allocations as information accumulates. Under a condition ensuring a unique efficient matching, the surplus-maximising matching eventually becomes efficient almost surely, with an exponential bound on the probability of an inefficient match. The analysis also considers how signal precision and industry-specific comparative advantage affect outcomes.

Citation

BibTeX citation:
@article{najafi-zangeneh2026,
  author = {Najafi-Zangeneh, Saeed and Gossner, Olivier},
  title = {Matching with {Learning:} {A} {Bayesian} {Assignment} {Game}
    {Approach}},
  journal = {Mathematical Social Sciences},
  pages = {102583},
  date = {2026},
  url = {https://gossner.me/papers/matching-with-learning-a-bayesian-assignment-game-approach.html},
  doi = {10.1016/j.mathsocsci.2026.102583},
  langid = {en}
}
For attribution, please cite this work as:
Najafi-Zangeneh, Saeed, and Olivier Gossner. 2026. “Matching with Learning: A Bayesian Assignment Game Approach.” Mathematical Social Sciences, 102583. https://doi.org/10.1016/j.mathsocsci.2026.102583.