Regulatory grandfathering

Why do new safety rules so often exempt the very buildings that prompted them?

A fire kills people in an old tower. An inquiry reports. A new standard is written — and it applies to buildings not yet designed. The tower that burned, and the several thousand like it still standing, keep the rule they were built under.

Read plainly, that is perverse: the evidence came from the old stock and the remedy lands on the new. Yet the pattern is so consistent — fire codes, emissions law, vehicle standards — that calling it a failure of nerve explains nothing. A practice this regular is doing something. What?

Start with cost, because that part is honest. A second staircase drawn into a plan costs the price of the drawing plus a slice of floor area nobody has sold yet. The same staircase cut into a finished building costs demolition, structural work, displaced tenants, and a fight with a geometry never meant to accommodate it. The gap is not a small multiple. So per unit of risk removed, a regulator gets far more from the new stock than the old — and if the compliance budget is finite, spending it where it is cheap is arithmetic, not cowardice.

But ask what that arithmetic invites: how long does a new-build-only rule take to cover the stock? Suppose buildings are replaced at 2% a year — a figure chosen for the calculation, not a measured rate. After a year the rule covers 2%. After ten years, 0.98 raised to the tenth leaves 82% untouched, so under a fifth is covered. Half the stock is covered only after about thirty-five years, since 0.98 to the thirty-fifth is 0.49. A generation, from a rule written the month after a disaster.

Now the turn that makes this more than slow progress. The rule raises the cost of building new while leaving the cost of keeping old exactly where it was. What does that do to the 2%? It lowers it. So the exemption does not merely wait for the dangerous stock to disappear — it pays that stock to stay. This is the best-documented consequence of grandfathering, clearest in American air law, where existing plants were exempted in 1970 and new sources held to strict review, and old coal capacity ran well past the retirement anyone had assumed; Stavins's work on vintage-differentiated regulation is the standard treatment. The mechanism is not disputed even where its size is.

Two folk accounts are worth clearing away. It is capture — the owners lobbied is sometimes true but cannot be the general answer, because grandfathering appears where the affected owners are thousands of unorganised householders with no lobby, and in rules written while public attention is at its peak and any lobby at its weakest. The rules are simply too weak fails on direction: a grandfathered regime is often stricter on new build than a uniform regime would be, precisely because the cost falls on people not yet in the room. Exempting the present is what makes severity toward the future politically cheap.

That last point is a second, separable mechanism. Existing owners are identifiable, present at the hearing, able to name what the rule costs them this year. Future owners of buildings not yet drawn are nobody — they cannot object, and will pay in prices they never see itemised. A rule with a hole in it is the cheapest rule to pass, quite apart from whether the hole is efficient.

What would falsify the load-bearing claim, that grandfathering extends the life of the risky stock rather than merely postponing coverage of it? If grandfathered assets were retired at the same rate as comparable assets under a uniform rule — if exemption changed who complied but not how long anything lasted — the claim fails, and grandfathering is no worse than slow. That is testable wherever a rule commenced on a date and the retirement curve either bent or did not.

The account passes a second test too. If the exemption were really about retrofit cost, it ought to lapse the moment the building is opened up anyway, since the cost gap then collapses. And it does: change-of-use and major-alteration triggers, pulling a grandfathered building up to current standard when it is substantially altered, are close to universal. The exemption is written to expire on the event that removes its justification.

Two distinct things are being traded, not one. The cost argument is about efficiency — buy risk reduction where it is cheap. The representation argument is about incidence — put the cost on whoever is absent. They usually point the same way, which is why the practice looks natural, and they can be told apart by the expiry conditions: an efficiency-driven exemption ends when the cost gap closes, as the alteration triggers do, while an incidence-driven one runs until the asset dies.

Two neighbours here are worth reconciling, and one reaches the opposite conclusion. Technical standards argues that an installed base keeps a standard alive: the tools, skills and spares already bought make the incumbent expensive to leave, so the base sustains the rule. Grandfathering runs the other way — here the rule sustains the base, by exempting it from a cost the alternative bears. Both hold, because the subsidy points differently: there, staying with the old thing is cheap because everyone else has too; here, because the law has said so. The second neighbour, Persistent rules, explains why a rule outlives its reason. Grandfathering is a sharper case, and the fixed point of difference is timing: this is not a rule quietly failing to die, but one deliberately born with the exemption inside it, drafted by people who could see exactly whom it spared.