The Fair Inversion

Surge pricing has a simple logic: when demand rises, raise the price. The revenue-maximizing policy under standard assumptions is monotonically increasing in load. More customers, higher price. The math confirms what the intuition suggests.

Chen et al. show that adding a hard fairness constraint inverts this structure.

When heterogeneous user groups display varying price sensitivity and equitable access is enforced as a constraint rather than a soft penalty, the optimal pricing policy becomes non-monotonic in demand. Under heavy load, the revenue-maximizing price sometimes decreases. The optimal response to a demand spike can be a price drop.

The mechanism is specific. Price-insensitive users continue purchasing regardless of surge. Price-sensitive users are priced out first. A fairness constraint that prevents disproportionate exclusion of the sensitive group forces the platform to lower prices precisely when demand-driven pricing would raise them. The constraint does not merely cap the price — it redirects the optimization landscape, creating a local minimum where standard theory predicts a local maximum.

Their framework uses High Order Control Barrier Functions with robust optimization, handling unobservable price elasticity across groups. The approach maintains safety and fairness constraints across all user distributions consistent with available data. The result is not approximate — the non-monotonicity is a proven feature of the optimal policy, not a numerical artifact.

The implication cuts deeper than pricing. Any optimization problem acquires qualitatively different structure when equity constraints bind. The fair solution is not the unfair solution with a ceiling — it is a different function entirely. Constraints do not shrink the solution space; they reshape its topology.

The fair monopolist must sometimes charge less when demand is highest. Fairness is not a cost subtracted from the optimum. It is a mirror that reverses it.