Occupancy-Aware High-Speed Rail Pricing with Aggregate Data: A Case Study in China

Aggregate origin–destination data and summary load factors do not identify train-specific flows; imposing each load factor as an exact margin can therefore overstate precision. We develop an auditable high-speed rail pricing framework combining pooled passenger-kilometer entropy completion, an occupancy-aware pivot logit, link-load equilibrium, and exact-grid Pareto enumeration. Historical train shares guide a sensitivity-tested prior, and target elasticities index fare response. For a reconstructed Beijing–Wuhan case, 97,336 discount vectors are evaluated in each of three core scenarios and candidates are screened across 729 crossed baseline and behavioral scenarios. Under the reference scenario (unit elasticity, crowding-equivalent ratio 0.10, threshold 0.80), the closest-to-ideal admissible vector (0.94,0.99,0.97) increases modeled-service revenue by 0.106% and yields a full-corridor-fare-equivalent surplus gain of 0.803%, with maximum link occupancy of 77.27%. It is admissible in only 368 scenarios and reaches 104.45% occupancy in the most adverse case; only the status quo is admissible in all scenarios. The framework thus identifies scenario-specific opportunities while separating exact numerical evaluation from empirical identification and operational robustness under aggregate-data limitations.