[UTMD-116] The Upside of Unreliability: Screening via Sunk Costs in Platforms (by Hiroaki Odahara)

Hiroaki Odahara

Platforms often require users to incur sunk participation costs (fees, effort, or waiting) before learning whether service will be delivered. We study such environments with a model that separates actual non-delivery risk from perceived risk. When users are naïve about risk, unreliability creates scope for ex post manipulation: within a benchmark class of efficient two-bidder mechanisms, the war-of-attrition (second-price all-pay) format maximizes deviation revenue. When risk is common knowledge but budgets and valuations are misaligned, unreliability instead screens participation. Higher reliability can attract deep-pocket, lower-valuation entrants who crowd out budget-constrained high-valuation users, while intermediate unreliability can induce their exit and may increase consumer surplus discontinuously at the exit threshold. We do not advocate intentional unreliability; rather, we highlight a design tradeoff between trust-enhancing reliability and screening through endogenous entry in settings with unavoidable execution risk and limited commitment.