We have a precise word for a politician taking an envelope. We have no everyday word for a decade of decisions that all happen to land on the same side of the same argument. That gap in the vocabulary is not neutral — it is doing work.
The bribe is the exception
The picture of corruption most of us carry around is a transaction. Someone pays, someone decides, and if things go well there is eventually a date, a name, and a trial. It is a satisfying picture because it has a villain in it.
Almost none of the influence that actually shapes public policy looks like that.
The common pattern has no moment you can point at. A regulator spent fifteen years in the industry they now supervise, and will probably return to it. A public consultation closes with nine of its ten substantive submissions written by the same sector. An expert committee's members do research funded by the firms the committee advises. A draft amendment reaches a ministry already written, by people who understand the file better than anyone on the public payroll — because the public payroll cannot match their salaries.
Every one of those is legal. Several are, taken alone, desirable: you want technically competent people advising on technical rules. But run them together for ten years and the rulebook acquires a direction, and the direction is not one anybody voted for.
That pattern has a name, and the name is not corruption. It is policy capture.
What the OECD actually calls it
The Organisation for Economic Co-operation and Development (OECD) is not an activist group. It is a club of rich-country governments whose house prose runs to phrases like "regulatory management practices." In 2017 it published Preventing Policy Capture: Integrity in Public Decision Making, and defined the thing in one sentence:
Policy capture is the process of consistently or repeatedly directing public policy decisions away from the public interest towards the interests of a specific interest group or person.
Two words in that sentence carry the entire argument: consistently and repeatedly. Not illegally. Not secretly. The definition is about a pattern over time, and it is silent on whether any individual step in the pattern broke a rule.
The report is explicit that most of it doesn't. Capture, it says, "can be achieved through a wide variety of illegal instruments, such as bribery, but also through legal channels, such as lobbying and financial support to political parties and election campaigns." And its foreword contains a line that is genuinely striking coming from an intergovernmental secretariat:
Capture escapes clear-cut legal definitions, and happens in the grey areas between ethics and laws.
Read that again as an admission rather than a description. The most institutionally cautious body that could possibly have written this sentence is saying that the failure mode it considers most corrosive to democratic government is one the law is not built to see.
A fifty-year-old idea that keeps getting wider
None of this is new. What has changed is how much of government the concept is understood to cover.
The economist George Stigler set it out in The Theory of Economic Regulation in 1971, with a claim that was shocking at the time and is now close to conventional wisdom: as a rule, regulation is acquired by the industry and operated primarily for its benefit. Regulators do not get corrupted as an accident of weak character; they get captured as a predictable result of who shows up, who pays attention, and who has a concentrated stake in the outcome.
Forty years later, Daniel Carpenter and David Moss's Preventing Regulatory Capture complicated the picture in two useful ways. They gave a name to cultural capture — the regulator who, through years of shared conferences, shared vocabulary and shared assumptions, simply comes to think like the industry and can no longer easily imagine another way of framing the problem. And they identified corrosive capture, where the win is not a favourable rule but the absence of one. Capture doesn't have to produce bad regulation. It can just produce silence.
The OECD's move in 2017 was to take the idea out of the regulator's office and apply it to the whole apparatus: agenda-setting, drafting, adoption, implementation, evaluation. Anywhere a public decision is made, it can be pulled.
Corruption is a verdict; capture is a direction
Here is where I think the language actually fails us, and it isn't a matter of one word being stronger than the other. The two words describe different kinds of object.
Bribery is a claim about an act. It has a location in time. You establish it by proving that one specific thing happened, and the proof either works or it doesn't.
Capture is a claim about an aggregate. It has no location in time. You establish it by showing that a long series of things kept happening the same way — which is not a fact any single one of those things contains.
The OECD puts the distinction plainly: unlike bribery, capture "is not related to a specific transaction, but is usually characterised as a more stable (undue) relationship achieved over time."
A legal system built to prosecute acts is therefore structurally blind to it, and no amount of enforcement effort fixes that. It is not that prosecutors are lazy or that the laws are too weak. It is that there is nothing in the right shape to prosecute. Capture is a property of a distribution, and you cannot indict a distribution.
This also explains why "is it lobbying or is it capture?" has no answer at the level of a single meeting. The same act sits at very different points on this line depending entirely on what surrounds it:
Nothing visible changes as you move left to right. The meeting looks the same. The minutes look the same. What changes is who else was in the room over the previous five years, and how the decisions came out.
Where capture gets in
Because capture is cumulative, it doesn't need a single point of entry. It needs several small ones, used repeatedly.
Set those channels side by side and the interesting thing is how little the legality of each tells you about its power.
| Channel | What it is | Legal? | Publicly documented | Risk of sustained capture |
|---|---|---|---|---|
| Bribery | Payment for a specific decision | No | ||
| Party and campaign finance | Money to the people who will later decide | Yes, within limits | ||
| Lobbying | Direct argument to decision-makers | Yes | ||
| Revolving doors | Staff moving between the regulator and the regulated | Yes, often unrestricted | ||
| Funded evidence | Studies and expert panels paid for by the interested party | Yes | ||
| Social ties | Shared schools, careers, boards, friendships | Yes |
I have deliberately left the legality column uncoloured, because colouring it would imply that "legal" is the reassuring answer. It isn't. Bribery — the only illegal row — carries no more sustained risk than ordinary lobbying, not because it is harmless but because it is prosecutable, and a prosecutable channel can be closed. Every channel in that table with a high risk of sustained capture is a legal one.
The revolving-door row deserves singling out, because it is the clearest case of a risk that is fully documented and simply tolerated. The OECD's own survey of economic regulators found that over half place no restrictions at all on the pre- or post-employment of professional staff[1]. These are not secret arrangements. Appointments are announced. The information is public, sitting in plain view, and mostly nothing follows from it.
Nobody has to be a villain
This is the part of the concept people find hardest to accept, and it is the part that matters most.
Our entire moral vocabulary for public life is built on intent. We ask whether someone meant to do wrong, because that is what distinguishes a crime from a mistake. Capture does not respect that distinction. The OECD notes that undue influence "can also be exercised without the direct involvement or knowledge of public decision makers, by manipulating the information provided to them, or establishing close social or emotional ties with them."
Consider what cultural capture actually feels like from the inside. You are a senior official on a technical file. The people who understand it best work in the industry. They are, generally, decent and serious. You meet them constantly, because they are the ones who turn up. Their framing of the trade-offs becomes your framing of the trade-offs, because it is the only detailed framing anyone has put in front of you. You have taken nothing, promised nothing, and hidden nothing. You would pass a polygraph on your own integrity — and you would be telling the truth.
The output is still capture. Not because anyone was bought, but because the input was never balanced.
Which means the honest version of this argument has to give up something satisfying. If your model of the problem is bad people, then good people are the fix, and you can go on believing that the right election solves it. If the problem is which voices are structurally present in the room, then replacing the people changes very little, and the fix is boring and institutional.
So where is the line?
Lobbying is not capture. Advocacy is not capture. A democracy in which affected parties could not argue their case to government would be worse, not better, and the OECD's own Recommendation on Transparency and Integrity in Lobbying opens by affirming that lobbying is a legitimate act of political participation.
So the line has to sit somewhere. I don't think it sits where we usually look for it.
We look for it in the act — was that meeting improper, was that donation over the limit, was that job offer a reward. But every act in a captured process can pass that test individually. So the test has to move to the aggregate: across a long run of decisions in a policy area, whose interests keep prevailing, and who was never in the room at all?
That reframing has an uncomfortable consequence. It means capture is undetectable at the resolution at which we actually consume politics. News covers decisions one at a time. Scandal requires an act. A pattern spread across two hundred decisions and fifteen years has no news hook, no date, and no defendant — and so, in the only forum where public pressure gets generated, it does not exist.
It is not that the evidence is unavailable. Martin Gilens and Benjamin Page's much-argued-over 2014 study, Testing Theories of American Politics, is precisely an attempt to look at United States policy at the aggregate resolution rather than the anecdotal one — and reported that economic elites and business-oriented groups had substantial independent influence on outcomes while average citizens had little or none. Reasonable people dispute the model and the strength of that finding. What isn't disputable is the method's premise: this is a question you answer with a distribution, not with a scandal.
Push the aggregate far enough and it stops being capture of policies and becomes capture of the state itself — what Transparency International's Corruptionary calls state capture, where private interests shape not just decisions but the laws, institutions and rules of the game to their own advantage. Policy capture and state capture aren't different phenomena. They're the same phenomenon at different scales.
What actually works
If the problem is structural, so is the remedy. The OECD proposes four complementary strategies, and their most useful feature is that none of them depends on identifying a wrongdoer.
- Inclusive process. Deliberately engage the interests that have no professional representation, rather than waiting for balance to arrive on its own. The diffuse side of almost every policy question — consumers, future residents, small firms, the general public — has no lobbyist, and its absence is the default state, not an anomaly.
- Transparency. Lobbying registers, published meeting logs, disclosure of who funded which submission, real records of who was consulted. The European Union's Transparency Register is the best-known example, and its limits are instructive: registration works far better as a source of aggregate patterns than as a per-meeting verdict, which is exactly the resolution the problem lives at.
- External accountability. Supreme audit institutions, competition authorities and regulators that can look at a policy area from outside and are themselves insulated from the same pressures.
- Organisational integrity. Conflict-of-interest rules that bite, cooling-off periods between public and private roles, rotation of staff on long-running files.
None of it is thrilling. All of it is procedural. That is the point: procedural defences are the only kind that work against a problem with no individual perpetrator.
I would add a fifth, which is less a policy than a habit: judge a public body by its record rather than by its episodes. If you only ever ask whether the last decision was defensible, you will never detect a captured institution, because in a captured institution every individual decision is defensible. That is what capture is.
Why the word matters
Calling this corruption is not merely imprecise. It quietly hands the argument away.
"Corruption" sets the threshold at the criminal law. It invites the reply I have broken no law — and that reply is usually true, and entirely beside the point, and it works. A vocabulary that only recognises the bribe will treat everything short of a bribe as fine by default, which describes essentially the whole mechanism by which public decisions are actually bent.
"Policy capture" moves the question. It stops asking was this legal and starts asking who has this consistently been working for. That second question is harder, slower, and produces no arrests. It also happens to be the one with an answer.
Getting the word right doesn't fix anything by itself. But you cannot organise against a thing you have no name for, and for most of what is wrong with how public decisions get made, "corruption" is not that name.
References
- OECD, Preventing Policy Capture: Integrity in Public Decision Making (2017) — OECD Publishing (PDF)
- George J. Stigler, "The Theory of Economic Regulation" (1971) — The Bell Journal of Economics and Management Science, 2(1), 3–21
- Daniel Carpenter and David A. Moss (eds.), Preventing Regulatory Capture: Special Interest Influence and How to Limit It (2014) — Cambridge University Press / The Tobin Project
- Martin Gilens and Benjamin I. Page, "Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens" (2014) — Perspectives on Politics, 12(3), 564–581
- Transparency International, Corruptionary A–Z — definition of state capture
- OECD, Recommendation of the Council on Transparency and Integrity in Lobbying and Influence (2010, revised 2024) — OECD/LEGAL/0379
- European Union Transparency Register