Regulatory capture

Why does an agency created to regulate an industry so often end up defending it?

The instinctive explanation is corruption: someone was bought. But the pattern is far too regular for that. An agency is created after a scandal, staffed by conscientious people with a clear mandate, and a decade later it is issuing rules the industry drafted, defending the industry's interests in court, and treating its critics as amateurs. If no one took a bribe, what did the work?

Start by counting who cares, and how much. Suppose a rule transfers a modest sum from every household in a country to twenty firms. Split among households the loss is small enough to ignore. Split among the firms the gain is enormous each. Now ask what each party should rationally do about it.

For a firm, hiring specialists, filing detailed comments and attending every hearing is trivially worth it — the stake justifies almost any expenditure. For you, the cost of understanding the rule alone exceeds your entire share of the benefit from opposing it. Not apathy: arithmetic. Mancur Olson's argument about collective action is exactly this — small groups with concentrated stakes organise, large groups with diffuse stakes free-ride.

So the first mechanism needs no villain at all. It is simply who shows up. The proceeding is open to everyone; the room contains the industry.

The second is subtler. A regulator must know how the thing being regulated actually works — what a refinery costs to retrofit, what failure rate a device really has, which mitigation is feasible by when. Where does that knowledge live? Overwhelmingly inside the industry. The regulator can demand data, but the industry chooses how it is framed, which alternatives are presented as realistic, and which are quietly absent from the option set. Whoever supplies the facts substantially sets the menu, and doing so requires nothing dishonest.

The third is the revolving door. Staff arrive from the industry because that is where the expertise is, and leave to it because that is where the salaries are. The assumed effect is soft — do not antagonise a future employer. But the evidence is genuinely mixed, and there is a serious argument the other way: a regulator whose market value depends on being known as formidable has a reason to enforce harder, building a reputation to sell. Treat the door as a channel of influence whose sign is not settled.

Stigler's 1971 paper made the general claim sharpest: regulation is often acquired by the industry and operated for its benefit. The most telling cases are not weak enforcement but rules restricting entry — licensing requirements, route and rate controls, standards written around incumbent technology. A firm's largest regulatory prize is rarely a lighter fine; it is a rule that keeps competitors out, dressed as a safety measure and sometimes genuinely being one.

Which brings the honest difficulty. Capture is far easier to allege than to demonstrate. Any decision industry approves of can be labelled capture, and the accusation is close to unfalsifiable unless you can say what an uncaptured agency would have decided instead — without simply assuming the answer is whatever the critic prefers. Meanwhile some of what looks like capture is genuine dependence on expertise: sometimes the industry's account is correct, and a rule it likes is not thereby a bad rule. Cultural capture — absorbing the industry's sense of what counts as reasonable, with no transaction anywhere — is probably the most common form and the hardest of all to measure.

Nor does the tilt run only one way. Agencies are also pulled by organised advocacy groups, by the enthusiasms of their own staff, and by the incentive to avoid the visible failure while accepting the invisible one — an approval that harms someone is a scandal, a delay that costs the same lives is not.

The word "capture" invites a picture of an agency taken over, and that picture is usually wrong. What the mechanisms above produce is a gradient — a persistent slope in whose arguments get heard, whose data frames the question, and whose objections carry weight — not an event with a date. An agency can be tilted on entry rules and rigorous on safety, or captured for a decade and then not.

That also tells you what the counterweights have to do. They are not attempts to make officials more virtuous; they are attempts to change who shows up and who supplies the facts — standing for affected parties, publication of contacts and comments, funding independent of the regulated firms, technical capacity inside the agency so it need not borrow the industry's, judicial review that forces a written justification, cooling-off periods on the door, sunset clauses that make a rule re-argue for itself. Each addresses a specific asymmetry rather than capture in general.

Finally, hold the null hypothesis open. Before concluding capture, ask what evidence would distinguish it from an agency that examined the question and agreed. If nothing would, the claim is doing less work than it appears to.

• Why the strongest capture evidence tends to concern entry restrictions rather than enforcement intensity.
• How agencies with independent technical capacity, such as some safety investigators, behave differently from those without it.

Every term the collection defines is gathered in the glossary.