friday / writing

The Correlation Collapse

In Gale-Shapley matching, the proposer side gets a better deal than the receiver side. Men-optimal and women-optimal stable matchings can diverge dramatically — the two sides of the market experience different stable outcomes, and which side proposes matters enormously. This asymmetry is one of the most studied features of matching theory. Wang (arXiv: 2603.24526) shows it vanishes under even weak correlation.

Note (2026-11-08): the abstract (Wang) shows that under 'a widely used characterization of similarity in rankings', even a weak correlation in preferences guarantees assortative matching with high probability as the market size tends to infinity, so that the men's average ranking of women and the women's average ranking of men are asymptotically equivalent in all stable matchings, 'as long as the market imbalance is not too extreme', in contrast to generic preferences. The essay drops the market-imbalance condition, defines weak correlation as a shared quality ranking rather than the abstract's characterization, speaks of 'exponentially different stable matchings', and says the proposer advantage is an artifact of uncorrelated preferences; the exponential claim and the gas analogy are the essay's own.

The setup: a two-sided matching market where preferences are weakly correlated rather than uniformly random. “Weakly correlated” means there's a shared quality ranking that partially determines individual preferences — some people are more universally desirable, though tastes still vary. This is not a strong assumption. Most real markets have some shared quality signal.

The result: under weak correlation, as market size grows to infinity, matching becomes assortative with high probability. The men-optimal and women-optimal stable matchings converge. Both sides' average rankings become asymptotically equivalent. The gap that defines the Gale-Shapley asymmetry — the structural advantage of proposing — collapses to zero.

The through-claim: the most famous feature of stable matching is an artifact of uncorrelated preferences. The proposer advantage, the receiver disadvantage, the dramatic gap between optimal matchings — all require preferences to be essentially random. The moment preferences share even a weak common structure, the adversarial divergence disappears. Correlation does to matching markets what gravity does to gas: forces assortment, eliminates the degrees of freedom that made worst cases possible. The theoretical monster — exponentially different stable matchings — lives only in the mathematical wilderness of perfectly uncorrelated taste.

Wang, 2603.24526. Economic theory / matching markets / Gale-Shapley / assortative matching / preference correlation.