Decision Analysis · Stable Matching

Proposer
Advantage
in Job Markets

Comparative analysis of Applicant-optimal vs Employer-optimal stable matchings using the Gale-Shapley algorithm on real LinkedIn job postings and resume skill profiles.

Vakulich Anastasia · БАСБ251 · Roshchina Nadezhda · БАСБ252

Simulation Parameters

Theoretical Foundation

01

Gale-Shapley Algorithm

Iterative deferred-acceptance algorithm guaranteeing a stable matching in O(n²) time. Each round proposers make offers; receivers tentatively hold the best and reject the rest.

02

Proposer Advantage

The proposing side receives their best possible stable matching. The receiving side receives their worst possible stable matching — this is the core structural asymmetry we measure.

03

Cosine Similarity

Preference lists are generated via cosine similarity of skill vectors: sim(a,e) = (a·e)/(|a||e|). Higher similarity = higher rank on the preference list.

04

Blocking Pairs

A pair (a,e) blocks a matching if both prefer each other over their current partners. A stable matching has no blocking pairs — verified after each run.