How You Frame the Discount Matters More Than You Think — But Only for Cheap Tickets

Swiss transit experiment finds that showing discounts as percentages instead of dollar amounts amplifies price sensitivity for low-fare trips, with a ~47% discount crossover effect.

You know how grocery stores put "50% OFF" on a $2 yogurt and "$5 OFF" on a $200 blender — and both feel like a deal even though one is a buck and the other is real money? That asymmetry is the entire engine of this paper. The question is whether the same framing trick works for public transit pricing, where the goal isn't selling more yogurt but pushing commuters off the 8:05 and onto the 8:35. The committed claim: price communication format — absolute fare, absolute deviation from current fare, or percentage deviation — does NOT change aggregate price sensitivity for transit riders. Full stop, no main effect. But dig into the interactions and two things emerge. First, for low-fare trips, percentage-based framing substantially amplifies price sensitivity compared to either absolute format. Second, percentage framing increases sensitivity for small discounts but decreases it for large ones, with a crossover at roughly 47% discount depth. The experimental design is clean for a stated-preference study. Wallimann, Sticher, and Weibel recruited 1,082 participants from Swiss public transport users, randomized them across three price-format arms, and anchored the choice scenarios in each participant's actual recent journey — not a hypothetical route. The estimation uses a multinomial logit model with price-format interaction terms. This is standard discrete choice machinery, well-suited to the question. On the ladder: there is no direct predecessor that tests price communication format for single transit fares, which is the paper's honest contribution. The broader behavioral pricing literature (Thaler's mental accounting, prospect theory, the percentage vs. absolute discount literature in retail) provides the theoretical frame, but nobody had run this experiment for transit departure-time choice before. The novelty is narrow but real — it's a first application, not a first mechanism. The integrity picture is decent but not bulletproof. This is stated preference, not revealed preference — people saying what they'd do, not doing it. The authors acknowledge this limitation. Randomization across three arms with over 360 per arm is adequate for the multinomial logit they're running. But the Swiss transit context is specific: high baseline service quality, relatively price-insensitive riders, strong cultural norms around public transport. The 47% crossover is estimated from the model, not observed as a discrete threshold in field data. The practical upshot is pointed: transit agencies already running off-peak discount programs could potentially amplify their demand-shifting effect simply by relabeling fares as percentage discounts rather than absolute amounts — but only for low-fare products and only for discount depths below ~47%. Above that threshold, percentage framing actually dampens sensitivity. This is a formatting decision, not a pricing decision, which makes it operationally cheap to test. The glaring gap is a field experiment. Stated preference gets you directional signal, but the next credible step is an A/B test on a real ticketing platform with revealed behavior. The authors almost certainly know this — a Swiss transit authority collaboration would be the natural follow-up, and it would require institutional access and timeline that a survey-based paper doesn't. This reads as (a): they'll seek the field trial next, not (c): they're hiding results.