The reason a returns policy shapes who buys

Two shops sell the same jacket at the same price. One takes returns for a year, no questions; the other allows fourteen days with a receipt. They end up with different customers, different return rates, different margins and different reviews — from an identical product.

What everyone sees

Return terms are treated as a cost line, tuned to balance conversion against losses. That treatment misses the selection effect. The policy is read by prospective buyers as information about the product and about themselves, so it does not merely change behaviour within a fixed customer base — it changes which people are in the base.

What is actually happening

Rothschild and Stiglitz established the general result in insurance markets: contract terms sort applicants by unobservable type, so any change to terms changes the composition of who accepts, and terms cannot be evaluated independently of that sorting. Applied to retail, a generous policy attracts both buyers who value insurance against a genuinely uncertain fit — a large, profitable group — and buyers who intend to use the shop as a fitting room. Petersen and Kumar’s analysis of returns management found lenient policies raising both returns and subsequent purchasing, with net profitability depending on which type dominates. The policy is therefore a customer-selection instrument that happens to be filed under logistics.

Why it stays hidden

The selection hides because the metrics are recorded per transaction. Return rate rises and is investigated as an operational problem, order value rises and is credited to merchandising, and neither report contains the sentence that says both numbers moved because the buyer population changed. The instrument that did it is not owned by anyone who reads either report.