Fragile by Design: How Climate Risk Is Repricing Aviation

Aviation is usually discussed as a cause of climate change. This analysis treats it as an industry whose operating environment is changing faster than its economics can absorb. Three terms need to be kept apart: hazard is the physical event, such as a storm or flood; exposure is what lies in its path; and vulnerability is how badly the exposed system is damaged. Climate change raises the hazard, and aviation’s structure determines its vulnerability, meaning a more severe event produces a proportionately larger loss only if it strikes a system with little capacity to absorb it.

Aviation’s vulnerability stems from three structural features: networks with almost no slack, margins too thin to absorb recurring shocks, and dependencies concentrated in a few nodes. These features create an economic coordination problem: the benefits of resilience are distributed across airlines, airports, passengers, and the wider network, while the cost of maintaining spare capacity is borne directly by individual operators. What is privately efficient may therefore be less resilient than what is socially optimal. Northern Europe illustrates the vulnerability of network flow, while Thailand illustrates the vulnerability of node concentration.

Clear-air turbulence is the best-documented signal. Severe turbulence over the North Atlantic rose from 17.7 to 27.4 hours a year for an average point between 1979 and 2020, an increase of 55%. This measures turbulence in the atmosphere, not encounters, since pilots avoid much of it. Convective weather is more directly disruptive to schedules, and its swings are large: weather delay per flight in summer 2025 was 37% lower than in summer 2024, and Eurocontrol credits part of 2025’s improvement to less convective activity.

These figures show that weather is a major and volatile cost driver, not that climate change caused today’s delays. IATA finds that 46% to 65% of European Air Traffic Flow Management (ATFM) delay each year was within air navigation providers’ control, mostly capacity and staffing. The defensible claim is therefore that climate change is altering hazards that a stretched network already handles poorly. Why does the network handle them poorly? Because delay in a queueing system does not grow in proportion to demand. In the simplest queueing model, average waiting time scales with utilisation divided by the share of capacity left unused: 9 at 90% utilisation (0.90 divided by 0.10) and 49 at 98% (0.98 divided by 0.02), so eight extra percentage points multiply waiting time by about 5.4. The calculation is illustrative, but Eurocontrol finds the same shape: delay grows very fast once demand exceeds capacity, and at several high-density airports demand is planned so close to capacity that little resilience remains. Heathrow is reported to run its two runways at about 98% of capacity.

This exposes a trade-off between efficiency and resilience. Spare capacity looks wasteful in normal operations but works as insurance during disruption, and a network optimised for maximum utilisation removes the buffer that disruption requires. The economic implication is that some apparently unused capacity has an insurance value: maximising utilisation lowers costs in normal conditions but increases the expected cost of disruption when the system is exposed to shocks. The consequence is propagation: in 2025, each minute of primary delay generated 0.79 minutes of reactionary delay on later flights.

The delay data are consistent with this mechanism, though they mix weather with other causes. IATA’s wider measure shows ATFM delay rising from 14.1 million minutes in 2015 to 25.8 million in 2025, an increase of 82.7%, while flights grew 11.5%.

Eurocontrol’s narrower en-route measure peaked at 22.4 million minutes in 2024, valued at €2.8 billion, and fell to 18.4 million minutes and about €2.4 billion in 2025, even as traffic rose 4.3%. Delays nonetheless remain above 2019 levels and are forecast at around four times the target through 2030. The 2025 improvement reflected better staffing, coordination and weather, so performance tracks conditions rather than a stable margin of safety. And IATA’s estimate that ATFM delays cost airlines and passengers €17.5 billion over 2015–2025 excludes weather by design, and over 70% of it is linked to capacity and staffing, it prices the structural weakness that weather exploits, not weather itself.

That weakness is clearest in the tail. The claim that simultaneous disruption at several hubs could fully paralyse regional traffic is a tail risk, not an observed event. The nearest test was the March 2025 Heathrow closure, caused by a substation fire rather than weather. More than 1,300 flights were cancelled or diverted on the Friday, though planes were landing again within 18 hours, and IAG put the hit to its operating profit at about €50 million. The network absorbed it while neighbouring hubs were operating normally. The scenario that matters is a correlated event that removes diversion options when they are needed most.

Disruption costs reach airlines through delay, fuel and thin margins, and fuel matters because unplanned holding and rerouting burn fuel that was not budgeted. IATA cut its 2026 profit forecast to $23 billion, a 2% margin against 4.2% in 2025, after the Middle East war pushed jet fuel to an expected $152 a barrel and fuel’s share of operating costs from 25.4% to 31.4%. That shock is unrelated to climate; it matters because it shows how little cushion exists. Profit is forecast at just $4.5 per passenger, and the industry has never posted a net margin above 5%. Climate-related disruption therefore does not need to become aviation’s largest cost shock to matter: recurring disruption consumes part of the financial buffer needed to absorb everything else.

Where costs land is conditional, not automatic: market structure determines whether they appear as higher fares, lower profitability or reduced capacity. Short-haul airlines will struggle to pass costs on, while long-haul and business routes can more easily. Eurocontrol’s €2.4 billion delay bill for 2025 equals roughly a quarter of Europe’s en-route navigation costs. That is material against thin margins but not, on its own, a fare shock. The better-supported conclusion is that disruption erodes a margin already too thin to absorb the next shock. Thailand shows what such a shock looks like when the fragility comes from concentration rather than congestion. It should be treated as a case study of the mechanism rather than evidence that this pattern is already general across aviation: one flood event cannot establish a broader climate trend.

Europe’s fragility is one of flow; Thailand’s is one of concentration, where a single gateway and the systems around it play the role that congestion plays in the sky. Bangkok has two international airports, which looks like redundancy, but the September 2026 flooding affected both. Substitutes that share a hazard offer less insurance than they appear to. The exposure was already documented: Newcastle University researchers ranked Suvarnabhumi first of more than 14,000 airports for coastal flood risk, a global-model ranking rather than an engineering audit, and it is defended by a 23.5-kilometre flood wall. The stakes are national: tourism earned 2.7 trillion baht in 2025, arrivals fell 7.23% that year and a further 3.08% over January–August 2026, and tourism’s share of GDP is estimated at roughly 12%.

The flood tested that exposure. Bangkok received more than 12 inches of rain in two to three days, nearly a full September’s worth. The physical defences held and both airports stayed open, yet services deteriorated. Flooding kept hundreds of workers from reaching the airports, and Thai Airways cancelled 59 flights between 26 September and 1 October. The baggage backlog reached nearly 10,000; 1,092 remained on 3 October, and the chief executive had been removed from duty.

This shows that physical exposure alone is an incomplete measure of risk: roads, workers and ground-handling systems became channels through which an urban flood reached airport operations. An airport can therefore remain physically protected while becoming operationally impaired through infrastructure and labour dependencies outside its perimeter. This distinction between asset resilience and system resilience is central to understanding aviation’s climate exposure. The attribution needs care, however. Airports of Thailand said the airport’s other two ground handlers kept working normally, and some equipment was described as defective, so part of the failure was specific to one operator. Tourism effects are also unsettled: hotels reported normal cancellations, but the tourism authority warned of a 15% to 20% cut in western China arrivals. Loss estimates should therefore be read as ranges. The University of the Thai Chamber of Commerce put losses for 25–29 September at 12.28 billion baht in its base case, 0.066% of GDP, within a range of 5.9 to 20.6 billion.

Three economic lessons follow, and they link back to Europe. First, a small aggregate loss can hide a concentrated, recurring one: Bangkok bore about 57% of the base-case loss, and the 2011 floods caused approximately 1.43 trillion baht ($46.5 billion) in damage and losses. Second, resilience is a property of the whole system. Embankments protect the airport, but staff access, handling equipment and road links determine whether it functions, and these have different owners. The benefits of resilience are shared across airlines, airports, workers and passengers while the cost falls on one party. This creates a positive externality and coordination problem: an individual airport or airline bears much of the cost of additional resilience but captures only part of its system-wide benefit. Private incentives can therefore produce less resilience than the socially optimal level. Europe has the same structure, since no single actor owns the network’s slack. Third, spending reveals priorities: Airport of Thailand’s expansion roadmap exceeds 365 billion baht, more than five times its FY2025 revenue of 68.58 billion so capacity expansion is a very large call on capital.

The claim that mitigation will cost tens of billions of dollars is best tested layer by layer. Physical resilience (barriers, drainage, raised land) is partly supported: protecting the world’s coastal airports against sea-level rise to 2100 would cost $0.2 to 39 billion, or $23 to 38 billion to elevate them all, against $17.2 billion for Hong Kong’s new hub, and the lead author called this modest relative to global infrastructure spending. It is a one-off capital sum spread over decades, so permanently reshaping aviation economics overstates this layer. Kansai shows the logic: after Typhoon Jebi in 2018, with regional impact estimated at about $500 million; insurance recovered ¥6.2 billion of the airport’s ¥8.1 billion loss, and it committed ¥54.1 billion to protection. A tanker also struck its only access bridge, a landside dependency like Bangkok’s Road problem.

Operational resilience (coordination, rerouting, staff-access plans, spare capacity) behaves differently, and two kinds of slack need separating. Coordination can be cheap: pre-agreed rerouting scenarios saved at least 720,000 minutes of delay in summer 2025, though these are Eurocontrol’s own estimates against an assumed counterfactual. Buffer capacity is a different kind of cost. At the infrastructure level, unused runway and airspace capacity is a recurring opportunity cost in foregone slots and revenue, held by airports and air navigation providers under slot caps and regulation. At the airline level, standby aircraft, spare crews and longer turnarounds tie up assets that earn nothing in undisrupted periods. Unlike a one-off barrier, both are permanent charges, so the tens of billions claim holds, if at all, for physical defences, while slack is a continuing drag on the system.

This invites a fair objection to the pessimistic reading. Coordination already delivers gains, non-climate factors dominate European delay, sea-level risk builds slowly, and the Thai failure was partly operator-specific, the system might adapt without much new slack. The 2025 recovery also shows how much weather luck matters, which makes any single year hard to read as a climate trend, and Thailand is one event in one emerging economy, so it illustrates a mechanism rather than proving a trend. But the objection assumes operators will convert efficiency gains into slack, when airlines and airports are rewarded for utilisation. Better coordination can just as easily be used to run the network closer to its limit. The problem is therefore not simply whether aviation can become more efficient, but whether the incentives governing the sector allow efficiency gains to be retained as resilience rather than converted into additional throughput.

That is why the third layer, institutional resilience, matters most, because the benefits of resilience are shared and the costs concentrated. In economic terms, privately optimal utilisation may therefore exceed socially optimal utilisation: an individual operator captures the revenue from using an additional unit of capacity but does not bear the full network-wide cost when that lack of spare capacity amplifies disruption. The available levers include slot and scheduling rules that preserve buffers, regulated airport charges that allow recovery of resilience investment, and continuity standards for ground handling and landside access. Eurocontrol makes a similar point: insufficient capacity generates delay costs that generally far exceed the marginal cost of the missing capacity.

In conclusion, climate change raises the hazard, but economic damage is determined by the interaction of hazard, exposure and structural vulnerability. Europe shows how high utilisation removes the slack needed to absorb disruption; Thailand shows how concentrated, fragmented dependencies let disruption outside an airport’s perimeter impair its function. The evidence supports rising structural delay costs, long-term coastal capital risk, and adaptation that is affordable for physical defences but recurring, and harder to fund, for operational slack. It does not support an imminent fare shock or regional paralysis. Aviation’s economics will be reshaped less by one capital bill than by a gradual repricing of slack. The question is whether existing incentives will produce the right amount of resilience before the cost of operating without it grows.

Bloomberg. 2026. “Thai Airways Battles 10,000-Bag Backlog After Bangkok Floods.” October 1, 2026. https://www.bloomberg.com/news/articles/2026-10-01/thai-airways-battles-10-000-bag-backlog-after-bangkok-floods

Eurocontrol. 2025a. Performance Review Report 2024. March 2025. https://www.eurocontrol.int/sites/default/files/2025-03/eurocontrol-performance-review-report-2024.pdf

———. 2025b. Special Flash Briefing: Overview of the Performance of the European Aviation Network, Summer 2025. September 2025. https://www.eurocontrol.int/sites/default/files/2025-09/eurocontrol-special-flash-briefing-summer-2025.pdf

———. 2026. Performance Review Report 2025. March 2026. https://www.eurocontrol.int/sites/default/files/2026-03/eurocontrol-performance-review-report-2025.pdf

Ferrovial. 2025. “Heathrow Airport: Terminals, Passengers and Runways Information.” Accessed October 2026. https://www.ferrovial.com/en-gb/business/projects/heathrow-airport/

FlightGlobal. 2025. “London Heathrow Closure After Substation Fire Cost IAG Around €50m.” May 2025. https://www.flightglobal.com/international-airlines-group-iag/2025/05/london-heathrow-closure-after-substation-fire-cost-iag-around-e50m/

IATA (International Air Transport Association). 2026a. “Air Traffic Control Delays in Europe.” Economic Reports. Accessed October 2026. https://www.iata.org/en/iata-repository/publications/economic-reports/air-traffic-control-delays-in-europe/

———. 2026b. “Middle East Disruptions and High Fuel Prices Halve Airline Industry Profitability.” Press Release, June 7, 2026. https://www.iata.org/en/pressroom/2026-releases/06-07-middle-east-disruptions-high-fuel-prices-halve-airline-industry-profitability/

ITIJ (International Travel & Health Insurance Journal). 2026. “Update: Thai Airways Cancels 37 More Flights as Bangkok Flooding Disrupts Operations.” October 1, 2026. https://www.itij.com/latest/news/update-thai-airways-cancels-37-more-flights-bangkok-flooding-disrupts-operations

Nation Thailand. 2021. “Suvarnabhumi Tops Ranking of Global Airports Threatened by Rising Sea Level.” Accessed October 2026. https://www.nationthailand.com/in-focus/40001633

———. 2026a. “Thailand Steps Up Airport Climate Defences as Extreme Weather Threatens Aviation.” Accessed October 2026. https://www.nationthailand.com/news/general/40062095

———. 2026b. “UTCC Forecasts 12.3bn-Baht Flood Losses for September 25–29.” September 28, 2026. https://www.nationthailand.com/business/economy/40071602

———. 2026c. “Thai Airways Expects to Clear 1,092-Bag Backlog by Sunday.” October 4, 2026. https://www.nationthailand.com/business/corporate/40071851

Newcastle University. 2021. “Climate Change Puts Hundreds of Coastal Airports at Risk of Flooding.” Press Release, January 21, 2021. https://www.ncl.ac.uk/press/articles/archive/2021/01/coastalairports/

Prosser, Mark C., Paul D. Williams, Giles J. Marlton, and R. Giles Harrison. 2023. “Evidence for Large Increases in Clear-Air Turbulence Over the Past Four Decades.” Geophysical Research Letters 50 (11): e2023GL103814. https://doi.org/10.1029/2023GL103814

Quartr. 2026. “Airports of Thailand (AOT) Q4 2025 Earnings Summary.” Accessed October 2026. https://quartr.com/events/airports-of-thailand-public-company-limited-aot-q4-2025_FMwQpXAN

Reinsurance News. 2019. “Kansai Airport Recovers Over 75% of Typhoon Jebi Loss from Insurance.” Accessed October 2026. https://www.reinsurancene.ws/kansai-airport-recovers-over-75-of-typhoon-jebi-loss-from-insurance/

Royal HaskoningDHV. n.d. “You Can’t Stop Nature in Its Tracks: Climate Resilience in Japan.” Accessed October 2026. https://www.haskoning.com/en/projects/climate-resilience-in-japan

UK Parliament. 2025. “Disruption at Heathrow.” Hansard, House of Commons, vol. 764, March 24, 2025. https://hansard.parliament.uk/commons/2025-03-24/debates/41F8B035-AA6C-4F6C-9648-B481B3DB8F2D/DisruptionAtHeathrow

Visit Thailand Today. 2026. “Thailand Tourism Statistics 2026.” Accessed October 2026. https://www.visitthailandtoday.com/thailand-tourism-statistics

World Bank, and Government of Thailand. 2012. Thailand Floods 2011: Rapid Assessment for Resilience, Recovery and Reconstruction. Washington, DC: World Bank. https://documents1.worldbank.org/curated/en/677841468335414861/pdf/698220WP0v10P106011020120Box370022B.pdf

Yesudian, Aaron N., and Richard J. Dawson. 2021. “Global Analysis of Sea Level Rise Risk to Airports.” Climate Risk Management 31: 100266. https://doi.org/10.1016/j.crm.2020.100266