The pen that makes the law

What the withdrawal has already done

This mechanism is in no way hypothetical. Leading insurers and reinsurers have in recent years adopted exclusion policies targeting coal, then certain unconventional fossil fuels, refusing to insure new projects and sometimes to renew old ones. For a while they banded together in climate alliances to coordinate these withdrawals and give them weight. It worked: projects lost their cover, then their financing, then their viability. Then came the instructive reversal. These alliances were then accused of collusion, because coordinating a sector boycott looks a lot like an anticompetitive practice. Several members left under legal pressure. The main one, the Net-Zero Insurance Alliance, was launched in 2021 under the aegis of the United Nations. Targeted in 2023 by state attorneys general in the United States, it disbanded in the spring of 2024. The episode is illuminating on two counts. It confirms first that the power exists, since its concerted exercise became a cause for concern. Second, it shows that the law cannot classify it: climate virtue or cartel? It cannot tell whether these insurers were saving the planet or locking down a market, because the move is exactly the same under both readings.

A power no one delegated

This power might have been legitimate had it been entrusted. A regulator gets its authority from a mandate: a statute, an election, a delegation. That mandate lets it compel, and makes it answerable to those it compels. The insurer that exercises this de facto power has none of that. It answers to its shareholders, its reinsurers, its own appetite for risk and its reputational exposure, not to a political body. Where it draws the line reflects what is good for its balance sheet and its image. That may match the public interest or not, and nothing guarantees it will.

Insurers often have good reasons. They still decide, without a mandate, what once belonged to public debate. Banning coal, restricting certain forms of energy, excluding activities deemed incompatible with a climate pathway: these are major societal choices. A democracy normally leaves them to its legislatures, with their slowness, their compromises and their accountability. When a few private groups settle them in underwriting committees, on criteria the public cannot see or contest, a piece of economic sovereignty gets outsourced to an industry built to price risk, not to govern lives. Whatever the direction taken, no one has to account for having taken it.

On top of the missing mandate, the criteria don't line up. Each group sets its own thresholds, its own exclusions, its own idea of what is unacceptable. So what stays insurable depends less on coherent public policy than on the private morals of a few underwriting departments. An activity rejected by some finds a taker among others, until reputational contagion gradually closes the doors, and the result is neither a democratic debate nor a free market, but an opaque in-between where de facto rules impose themselves without ever having been written. We saw, with the cybersecurity directive, the law explicitly redistribute responsibility along value chains, in the open and under judicial review. Here it is the reverse, a redistribution of the right to exist takes place quietly, without text, without judge, without appeal, by the mere addition of private refusals.

The pen has always governed

It will be objected that this power is new, it is not, only its visibility is. Insurance has always shaped the world by deciding what it covered and at what price. It was insurers who, to limit their losses, imposed fire-safety standards, building codes, loss-prevention measures that then became the common rule, thus governing the city and the factory through the condition of cover alone. Insurers were also the ones who drew the neighborhoods they deemed too risky on maps, refused to insure them and sealed their decline. That was American redlining, and its social harm is still being measured today. The underwriter has always been a discreet legislator, dictating conduct by price and access rather than by law, rewarding what it wished to see and smothering what it feared. ESG does not create this power, it simply aims it, for the first time, at openly moral and political ends, and it is this new frankness that suddenly makes it visible and debatable.

The same pen excludes coal and the poor

There is something more troubling still, and it is here that the argument must become uncomfortable. The gesture applauded when it targets coal is exactly the gesture denounced when it targets something else. The insurer that turns down a coal plant for the sake of the climate does the same thing, with the same pen, as the one that turns down a low-income family's home in a flood zone, for the sake of the same climate. In both cases it exercises its power to exclude, it withdraws from an activity or a household the access to protection without which it can no longer hold. Society celebrates the first withdrawal as an act of conscience and laments the second as an abandonment, though the gesture is the same. The reason may differ: the flood-zone house is usually refused because its expected loss exceeds what the household can pay, the coal plant most often on a criterion that is not the year's loss. And in France, the second refusal has a remedy: a household refused because of the size of its natural catastrophe risk can apply to the Bureau central de tarification, which compels the insurer to provide the requested cover (Article L. 125-6 of the French Insurance Code).

This awkward symmetry reveals that we have no theory of legitimate exclusion. We can say the insurer was right to exclude the polluter and wrong to exclude the vulnerable claimant. But we could not state the rule that separates the two. The insurance contract knows only one rule, expected loss, and it is not enough to justify the first refusal. The underwriter segments: it sorts good risks from bad. That is the job. Pushed to the extreme, as we saw, the same logic dissolves solidarity into behavioral pricing. ESG underwriting is only an application of that logic turned toward an end deemed noble, but the end does not change the nature of the gesture. The power to include and exclude from economic life is the same, whether it serves the climate or abandons the fragile, and we never decided to entrust it to anyone.

The territorial side of this exclusion is the cruelest. By withdrawing from zones exposed to floods, to fires, to rising waters, insurers do not condemn an activity, they condemn places and those who live in them. A territory that insurance deserts becomes unsellable, uninvestable, gradually unlivable, and this governance by abandonment strikes first the populations who can least afford to leave. It is the mirror image of governance by prohibition. On one side the pen shuts out the powerful polluter, on the other it shuts the future on the vulnerable claimant, and the same climate justifies both. People praise the first as an awakening and accept the second as fate, without seeing they are two sides of the same power.

The honest defense of responsible underwriting

One must now serve the opposing case in good faith, for it is strong. Insurers will answer first that they are only managing their own risk, and that is partly true. Climate change hits them on both sides of the balance sheet: they insure the exposed assets and they invest in the firms that produce them. Too much carbon exposure is a real threat to solvency. It is the same double hit to liabilities and assets they took in the pandemic. Reducing that exposure is plain prudence, and one cannot fault an insurer for not wishing to sink with the assets it insures.

They will add that underwriting is a faster and more effective lever than regulation. A climate law takes years to pass, and gets negotiated, watered down, worked around. An underwriting policy bites within one renewal season. Using the insurance market to speed up a transition that governments struggle to impose would then be legitimate, and it would not have to wait for political consensus. Some even see in it an inverted democratic virtue, the insurer as a goad to states too slow, a private pressure that compensates for public failure. The case deserves to be taken seriously, for it touches a true point, the climate emergency is real and the public channels are slow.

A power one cannot refuse

And yet the answer falls short. It mixes up two separate questions: is the goal good, and who has the right to pursue it this way? That prudence justifies reducing carbon exposure says nothing of the legitimacy of a power that decides, beyond all control, which lawful activities may survive. An effective lever in hands without a mandate remains a lever without a mandate, and the effectiveness of a power has never made its legitimacy, it has often even served as its excuse. That is the danger. By greening their books, insurers govern without answering to anyone, and they dress political decisions in the neutral language of risk.

The most vertiginous point is that they cannot abstain. Even underwriting that meant to be neutral governs, for by making some things cheap to do and others impossible, it draws the map of what is economically practicable. Not choosing is still a choice, that of leaving the market as it is to decide, and that market is no more elected than the committees that correct it. The underwriter's pen has always been a discreet form of government, ESG has only made it visible by aiming it deliberately at moral targets. Insurers already hold this power and cannot give it back. The real question is how to make its use transparent and open to challenge. That is a constitutional question, and insurance has no framework to settle it.

Making this power accountable would require things insurance has never known, public and reasoned exclusion criteria, an avenue of challenge open to those deprived of cover, an external control verifying that the refusal truly serves the end it invokes and not a mere comfort of the balance sheet. Nothing of the sort exists, and it is hard to see what authority would impose it, since the power in question arises precisely from the aggregation of private decisions no regulator coordinates. Hence the paradox: the power is as diffuse as it is real. There is no seat where you can reach it, no text that bounds it, and so nothing to grab for anyone who wants to rein it in. One does not easily reform a sovereignty that has no address.

An economy that lets its insurers decide what has the right to exist has delegated, without saying so, a fragment of its sovereignty to actors designed to weigh probabilities, not to arbitrate values. Perhaps this power is exercised today in a direction many approve, but a power one approves when it serves our ends is a power one will suffer when it serves another's. Beneath the climate, the ESG debate in insurance is about sovereignty, and the only thing more dangerous than an underwriter who refuses to govern is an underwriter who governs without anyone ever having authorized it to. The climate will give it the occasion, it will never give it the mandate.