Who pays for a closed sky? Carrier-specific airspace closures, range thresholds, and airline network reconfiguration

Airspace closures in the 2020s are imposed by nationality: the same city pair becomes 1,000– 1,800 km longer for some carriers and not for others. In a logit frequency–gauge model with an outside option, a carrier-specific detour adds a business-stealing term to the market-size effect of a uniform shock, and the share of lost passengers that rivals capture falls to zero as their capacity becomes rigid. Payload–range limits add an ordered menu of responses: seat blocking, type switch, technical stop, exit. We compute carrier-month detours from closed flight information regions and study three closures of the US–India nonstop market (2019, 2022, 2025) and 178 markets during the 2022 Russian closure. Own detour lowers departures, seats and passengers by 18–39 log points per 1,000 km, and randomization inference over city pairs supports the departure and passenger estimates. Incidence flipped with rival fleet slack. In 2022 Air India captured 17–63% of the passengers US carriers lost; in 2025 US carriers captured none, and Air India’s losses rose with effective distance before the Air India 171 accident. Calibrated counterfactuals show that extending a closure to the rival recovers little traffic, shrinks the market, and adds passenger time costs and emissions.