The Disaster Response System Isn't Breaking Where You Think

The number that best explains this week in global disasters is 777. That is how many firefighters from 14 European countries the EU pre-positioned in high-risk zones across Cyprus, Greece, Italy, France, Spain, and Portugal this summer, a record for the bloc⁷, stationed before the fires started rather than scrambled after. Keep that number in mind as you scan the week's news, because it is the clearest clue to which parts of the world's emergency architecture are working and which are quietly rotting.

And what a week to scan. British Columbia declared a state of emergency Saturday¹ after the Bald Range fire forced more than 20,000 people from Summerland, Peachland, and surrounding communities, with more than 100 fires burning provincewide. Typhoon Dolphin, the strongest storm to hit China this year, made landfall in Zhejiang province² after more than a million people were moved, including over 900,000 relocated in Wenzhou alone. Puerto Rico began 48-hour rotating water shutoffs³ for more than 180,000 customer accounts amid record drought. The UK entered its fifth heatwave of the summer⁵, with a sixth building behind it. The natural question, and the one I set out to answer, is whether systems designed for one disaster at a time can survive everything happening at once.

The pessimistic case runs like this: mutual aid is a pooling arrangement, and pools only work when demands are staggered. Water bombers, incident commanders, hotel rooms, and reinsurance capital are all finite and shared. When six regions draw on the same pool in the same week, somebody gets triaged to the back of the line, and the whole edifice reveals itself as an accounting trick that works until correlation arrives. It is a coherent theory. This summer keeps failing to confirm it at the layer where it should show up first.

Start with the sharpest test on record for cross-border firefighting. In a single week last August, Greece, Spain, Bulgaria, Montenegro, and Albania all activated the EU Civil Protection Mechanism, the framework through which any member or neighboring state can request emergency help coordinated from Brussels. Every one of the five got aircraft, per the European Commission's own account⁶: Swedish planes to Bulgaria alongside five other countries' aircraft, French-stationed planes to Spain, Croatian, Bulgarian, Italian, Czech, and Slovak assets to Albania. This July, when France and Portugal activated simultaneously, rescEU planes from Sweden, Cyprus, Italy, and Spain moved within hours⁷, and Spanish ground crews arrived in Portugal the same day. Seven countries have activated the mechanism so far this season⁸. None has been left empty-handed.

But here is the honest complication, and it is where the ceiling theorists deserve real credit: the system is holding because governments concluded the old system would not. The voluntary swap-meet model, where states offer whatever happens to be spare, has been progressively replaced by standing capacity. The EU built rescEU as a strategic reserve precisely because voluntary offers run short in bad years, and it is now buying a permanent fleet of 12 planes and five helicopters⁹. Canada did the same thing this spring: Ottawa put $316.7 million over five years¹⁰ into a national surge fleet of 10 leased aircraft, the first direct federal investment in firefighting aircraft in Canadian history¹¹. Governments do not spend hundreds of millions patching a wall that was never going to crack. The simultaneity problem is real; it has simply been anticipated at the international layer, which is why the pessimists keep predicting a breakdown that keeps not arriving.

The financial layer tells the same story, only louder. Reinsurance, the insurance that insurers buy to cap their own catastrophe losses, is where correlated global risk should pinch first, because a bad fire year in California and a bad typhoon year in China land on the same balance sheets. Instead, global reinsurance capital hit a record $648 billion in 2025¹² by Gallagher Re's count, up 11 percent, with the sector posting a 19.3 percent return on equity. Aon puts total reinsurer capital at $785 billion, also a record¹³, and property-catastrophe prices fell 10 to 20 percent at the January renewals¹⁴ because supply is outrunning demand. A market approaching a correlation ceiling does not cut prices by double digits.

So where is the strain actually surfacing? Look closer at the two ugliest stories of the week, and notice that neither is a coordination failure. In BC, the province needed emergency powers to "secure accommodations for evacuees that might otherwise be unavailable," as Emergency Management Minister Kelly Greene put it¹. That is a last-mile problem: a sharp evacuation pulse hitting one valley's hotel stock in tourist season. And Puerto Rico's rationing is barely a climate story at all. The island's water utility loses roughly 59 percent of its treated water to leaks⁴, its drinking-water system carries a "D" grade from civil engineers, its reservoirs have lost up to 60 percent of storage capacity to sedimentation, and it rationed water in 1994, 2015, 2019, and 2020. San Juan neighborhoods logged 118 days of outages this year before the drought plan even began³. The drought found the failure; it did not create it. Meanwhile China's Dolphin response, whatever one thinks of the state behind it, demonstrated that moving a million people with over 1,000 shelters² is an executable logistics task, not a fantasy.

Two things narrow my confidence, and I want to be plain about both. First, the response system is flexing while the outcomes deteriorate: the EU's 2025 fire season burned more than a million hectares, the worst on record since 2006⁸, despite record deployments. Coordination can be adequate while suppression loses ground to the hazard itself, which climate research says will keep growing; the IPCC now treats concurrent and compound extremes¹⁷ as a distinct, worsening risk class. Second, the money is holding at the top and leaking at the bottom. Swiss Re estimates the global protection gap, the share of catastrophe losses nobody insures, widened to $424 billion in 2025¹⁵, with wildfire insured losses growing about 12 percent a year¹⁶ and a plausible peak-year scenario near $320 billion. Reinsurers can be flush while households in fire zones go bare. Both trends are real, and neither is a coordination ceiling. They are a hazard problem and a distribution problem wearing the ceiling's costume.

Which brings me back to 777. The visible, dramatic, telegenic parts of the simultaneity problem, the planes and the cross-border crews, are exactly the parts institutions have spent the past three years fixing, because aircraft on a tarmac photograph well and budget lines for them pass easily. The parts that failed this week are the ones no minister gets applauded for: pipes under San Juan losing six of every ten liters, hotel capacity in a fire-prone valley, premiums quietly becoming unaffordable in the places that burn. The systems built for concurrency held. The test they are failing is the slower one, and Puerto Rico has been failing it since 1994.