The risk everyone holds at the same time

Each of these owners had taken out cover thinking of their own misfortune, a fire, a flood, an isolated administrative closure. Business-interruption insurance existed for that, to put back on its feet a business struck by a singular loss, while the others kept running and paying their premiums. The pandemic overturned this silent assumption. It closed the restaurant, the gym and the cinema on the same day, in the same city, in the same country, and across the whole planet almost at once. Insurance built for separate misfortunes found itself facing a common one, and it understood, too late, that it had never been built for that.

Here is the thesis. The pandemic revealed that insurance has a blind spot, not for rare risks, nor even for severe ones, but for simultaneous ones. Insurance survives severity by spreading it, over different people, at different times, in different places. A pandemic collapses both axes at once, time and space, by striking everyone everywhere at the same instant. What makes it uninsurable is its synchrony. In the spring of 2020, the same lockdown order hit, within days, thousands of businesses that business interruption portfolios had treated as independent. And synchrony reveals a terrifying thing, that risks the insurer thought independent were secretly the same risk under different names.

Diversification and its collapse

To grasp the reach of this observation, one must return to what insurance is at bottom, a machine for diversifying across time and space. It rests on the law of large numbers, which operates only on one condition, the independence of losses. If the misfortunes of some bear no relation to those of others, then across a large book the swings cancel out, and the insurer can foresee with enough certainty what it will pay in the year. All the solidity of the edifice hangs on this assumed independence, it is the invisible floor beneath the actuary's feet.

A pandemic is the common cause in its pure state. It does not strike a thousand distinct risks, it flips the same switch a thousand times, and the law of large numbers, deprived of independence, does not degrade gently, it collapses all at once. The book that seemed diversified because it held a thousand different names is revealed as a single bet, the instant the cause becomes universal. The right image is a financial portfolio thought prudent because it held a thousand securities, until a common shock showed they all rose and fell together. The pandemic did not add a risk to the book, it revealed a hidden correlation that had always been there, never priced because it had never fired.

The loss with no outside

It will be objected that insurance has always faced enormous catastrophes, hurricanes, earthquakes, without collapsing. True, but those catastrophes have an outside. A hurricane ravages one region and spares a thousand others, it passes in a few days and leaves the following years intact. Reinsurance lives precisely on that outside, it diversifies across space, by covering regions that are not struck together, and across time, by counting on the calm years to rebuild what the hard years have cost. A pandemic has neither a geographic outside nor a temporal one. It leaves no region unharmed to draw upon, no neighboring year to recover in, it occupies all of space and stretches over a long time. It strips insurance of the elsewhere and the later on which its whole absorptive capacity rested.

Worse still, the pandemic strikes the insurer on both sides of its balance sheet at once. An ordinary catastrophe touches only the liabilities, it raises the claims to be paid while the investments backing the reserves stay stable. A systemic shock, by contrast, makes claims explode at the very moment markets collapse, so that assets shrink as liabilities swell. The insurer must pay more just when what it holds is worth less, and this correlation between its commitments and its holdings is the nightmare of solvency, because it attacks at the same time the promise and the means of keeping it. In March 2020, claims notifications were climbing while equity markets lost about a third of their value in a month, and the same shock hit the insureds and the assets meant to pay them.

The correlation no one had seen

Here the lesson turns truly disturbing, and reaches far beyond the health case alone. Before 2020, business-interruption cover spread over thousands of firms with no apparent link seemed beautifully diversified. A restaurant, a factory, a cinema, a dental practice, each looked a risk apart, governed by its own chances. The pandemic showed that an invisible thread bound them all, a common cause no one had seen or priced, because it had never fired. The insurer's diversification was not a fact, it was an assumption, and an assumption never put to the test is only a well-held belief.

Here is the epistemic point that should chill. An insurer can be blind to correlations that exist but have never shown themselves. Its book can be concentrated without its knowing, its apparent prudence resting on independences never verified, until the day a novel common cause reveals, retroactively, that the supposedly diversified book was concentrated from the start. The danger is therefore not only the pandemic, it is the whole class of latent correlations, those links that sleep in a book awaiting their trigger, invisible precisely because they have not yet struck. One prices only what has already happened, and synchrony, by definition, hides until its first passage.

Faced with this peril, insurance has certainly tried to equip itself, accumulation control, stress scenarios, tail-correlation modeling, a whole apparatus meant to spot concentrations before they strike. But this apparatus runs into a prior impossibility, one cannot measure a correlation that has never shown itself. The stress test presupposes guessing the shape of the shock in advance, and a truly new shock escapes, by nature, the imagination that preceded it. One can model yesterday's correlation, not the one awaiting its first day, so the most sophisticated instruments stay blind exactly where the danger is greatest, in the unprecedented. Quantitative prudence reassures on the known synchronies and leaves the sleeping ones intact, which is another way of saying that the true risk is always the one no model has yet seen.

The pandemic was only a template

Follow this thread, and the pandemic ceases to be an isolated accident and becomes the template of a whole family of modern risks, all defined by a hidden common cause. Cyber is the clearest example, a single piece of software, a single flaw, a single compromised provider can link thousands of insureds that nothing seemed to gather, turning one attack into a portfolio loss, exactly as a virus turns a thousand businesses into a single claim. Climate is another, a single driver, warming, correlates perils once thought distant, drought here and flood there, until it converges losses believed independent. Artificial intelligence prepares a third, the day a single failing model, adopted everywhere, produces the same error everywhere at the same moment.

All these risks share the pandemic's structure, a latent synchrony awaiting its trigger. This is what unites subjects once thought separate, climate, cyber, hybrid war, AI. Each is a common cause asleep, each threatens to reveal that a diversified book was only a single bet in disguise. The great matter of insurance in the twenty-first century is therefore not the rise of a particular peril, it is the rise of correlation itself, the return, in a thousand new forms, of the risk everyone holds at the same time. The pandemic was simply the first to declare itself with enough brutality that it could no longer be ignored.

One then finds, gathered into a single point, the threads pulled one by one in the other pieces of this series. The insurers' retreat before climate, the reinsurance chain that carries a shock step by step, cyber defeating a solvency built on history, the denied war that blurs attribution, all were already telling, without naming it, the same phenomenon, correlation rising and independence receding. The pandemic does not add a chapter to this story, it is its key, the case where simultaneity is seen with the naked eye because it closed everything on the same day. What the other perils insinuated, it demonstrated, the coming century will be one of common causes, and insurance will have to learn to think no longer of separate risks it adds up, but of linked risks it can no longer separate.

Who can carry the simultaneous

The detail of what played out in 2020 lights up the mechanism. Most business-interruption cover required physical damage, a fire, a flood, in order to trigger, and pandemic closure caused none, hence the huge battle over so-called non-damage clauses. Where the contract mentioned an infectious disease or an administrative closure, the insurer might have to pay, elsewhere it refused, and the same catastrophe produced opposite verdicts depending on the letter of the policies. Conversely, the one line that had correctly priced the risk, event cancellation backed by explicit pandemic cover, paid without dispute, and Wimbledon, canceled in 2020, remains the most cited example: its organizers had been paying for explicit pandemic cover since the 2003 SARS scare, and it paid out. This is proof that a simultaneous risk can be insured provided it is named, reserved for and charged at its true price. The disaster did not come from insuring the pandemic, it came from the insurance that had pretended to ignore it.

There remains the practical question, what to do with a synchronous risk that pooling, by construction, cannot absorb. Three answers contend, and none is comfortable. The first is exclusion, the insurer simply removes pandemic risk from its contracts, as it removes war, leaving a gaping hole of protection that society hates but that the market, alone, cannot fill. The second is transfer to the capital markets, through pandemic bonds that promised to attract investors willing to bet against the scourge. The experience largely disappointed, triggers too complex, payouts too late and too small, a basis risk that widened the gap between what the instruments paid and the losses they were meant to cover, recovering all the flaws already seen in parametric cover and in the securitization of risk.

The third answer is the only one that truly holds, and it is also the most far-reaching in its implications, transfer to the state. For the state holds a power no private insurer has, it can diversify not across space, where everything is correlated during a pandemic, but across time. It can spread the shock over generations, borrowing from the future to fund the present, which no private balance sheet can do. The state is the reinsurer of last resort of the simultaneous, not because it is richer, but because it insures along an axis forbidden to the market, duration. To say this is to admit that pandemic insurance is decided on the state's balance sheet, and by its willingness to socialize the shock, far more than at the insurers.

This recourse to the state is not, however, without its downside, and it must be said. A permanent public net can dull vigilance, for if everyone knows the state will pay for the simultaneous, the market is tempted to underprice it, firms to prepare less for it, and moral hazard rises a notch, no longer with the individual insured but with a whole system that leans on the rescuer of last resort. The backstop that saves is also the one that distorts, it socializes the loss but risks socializing imprudence too. The difficulty is therefore not only to find who carries the synchronous risk, it is to carry it without extinguishing each party's incentive to reduce it, a delicate balance that neither all-market nor all-state resolves on its own.

The worst choice, the very one that produced the chaos of 2020, is the in-between, neither covering clearly, nor excluding clearly. Many pre-pandemic contracts had neither foreseen nor ruled out health risk with any sharpness, so the question of who paid was settled in the courts, case by case, in long and costly battles over the meaning of words. In the United Kingdom, it took a test case brought by the regulator, decided by the Supreme Court in January 2021, to settle whether these policies covered pandemic closure or not, revealing that the market had, at bottom, neither embraced nor refused the risk, but left it in a blur from which no one emerged a winner. Ambiguity, in insurance, protects no one, it merely shifts the loss toward litigation.

The dress rehearsal

The pandemic was not an anomaly, it was a dress rehearsal. It showed that the most dangerous risks of the century will not be the biggest, but the most synchronized, the ones that strike everyone at the same time and turn a diversified book into a single bet. Insurance knows how to price the rare and absorb the severe, it meets its limit at the simultaneous, and the simultaneous is precisely the shape of the risks now arriving, from cyber to climate, from AI to denied war.

The question left by the empty restaurants is therefore not whether we will be able to insure the next pandemic, but whether we have understood that the pandemic was only the first of a kind. Each time a single cause runs through a book thought scattered, the same astonishment will return, that of discovering a thousand contracts were only one. The lesson, if we hold it, changes the very way of thinking about risk, it is no longer enough to count the perils and measure their severity, one must hunt the hidden threads that bind them, because it is correlation, not size, that decides what remains insurable. The future of insurance will be played out there, in the tracking of sleeping synchronies, before they all wake at the same time.

There is, in this discovery, something that exceeds insurance technique. A society that multiplies its dependencies, one network, one cloud, one climate, one market, unknowingly manufactures common causes, and so makes its own risks ever more synchronous the more efficient it believes itself to be. Insurance merely mirrors this drift, it is the glass in which the technical solidarity of the world reveals its fragility. To insure the simultaneous is, at bottom, to ask how many of our protections are still truly independent, and the answer, each year, grows smaller.