Contribution
It separates capture from corruption and from lobbying. Corruption is a transaction. Capture “is not related to a specific transaction, but is usually characterised as a more stable (undue) relationship achieved over time through both legal … and illegal instruments and channels.” Lobbying is legitimate advocacy that should be transparent and open to all. Capture is “not necessarily illegal” but “always illegitimate”. The report gives a working definition, three criteria for when influence counts as capture, a catalogue of vehicles and risk factors, and a four-part strategy against it.
Key insight
Capture is defined by its outcome relative to a stated goal, not by its means. “While policy capture entails clearly identifying and violating policy goals, no particular means of influence are presupposed.” So the same channel, a consultation, an expert group, a study supplied to an official, can be participation or capture, and “participation then shifts from promoting inclusion to become a tool for exclusion.” Capture “works through the system, rather than undermining it”: captured laws “are adopted through legal means and are therefore legally effective.”
Assumptions
- A public interest exists and policy goals can be stated against which diversion is measured. Criterion 1 requires that “An overarching policy goal(s) must be explicitly defined.”
- Influence is legitimate when transparent and inclusive, and capture is marked by opacity and exclusivity (Figure 1.4, after Boehm 2007).
- The empirical claims are drawn from cited studies (Box 1.1: Fisman 2001, Faccio 2006 and others) and from correlations in World Economic Forum survey indicators (Figures 1.2 and 1.6). The report does no analysis of its own beyond OECD surveys of lobbyists, legislators and regulators.
Key results
- Definition, executive summary. “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.” Chapter 1 states it as “the result or process by which public decisions over laws, regulations or policies are consistently or repeatedly directed away from the public interest and towards the interests of a narrow interest group or person, by the intent and action of this group or person (see also Carpenter and Moss, 2014).”
- Without the decision maker’s knowledge. “Undue influence can even be achieved without directly involving (and without the knowledge of) public decision-makers, by manipulating information or establishing close social or emotional ties with them.” Chapter 2: officials can be captured “unawares (by providing false information to skew a decision …)”.
- Three criteria (adapted from Carpenter and Moss, 2014): “1. An overarching policy goal(s) must be explicitly defined. 2. The captor (group) must intend to extract rents. 3. The policy-making process … must have shifted away from the policy goal(s) in the public interest towards the captor/group’s interest.” Without all three, actors “can still influence a particular policy field … but cannot be said to have captured the policy process”.
- Across the policy cycle (Table 1.1): agenda-setting, development, adoption, implementation and evaluation, each with its targets, including “invited experts” and consultants at evaluation.
- Vehicles (Table 2.1): reciprocity (payments, campaign funding, future jobs), direct influence (“Providing research and analysis”, threats), personal ties (family, networks, revolving doors), strategic communication (“Broad concerted action through media ownership”), and expertise (“Using think tanks to produce research”, “Providing manipulated information and expertise”). Repeated interaction alone can produce capture as public attention wanes (citing Martimort 1999).
- Risk map (Figure 2.1): opportunity (unchecked discretion, technical complexity, opacity) and capacity (resources, concentration and inequality, recurrent benefits, stable networks), and expectations of reciprocity. Technical complexity matters because “external analysis and opinions, even if recognisably biased by a special interest, can be considered useful guidance”.
- Self-reinforcement. “Benefits obtained through capture enable the interest group to reinvest in further influence-seeking”; capture “nurtures a vicious circle of inequality”. Rent seeking “is subject to increasing returns” (citing Murphy et al. 1993).
- Oversight bodies are targets. Audit institutions, competition authorities and regulators “are a natural target for capture”. Box 3.3, from a survey of 48 regulators in 26 countries: 88% of those funded by the executive get annual budgets, most heads are appointed by the executive, over half place no restriction on pre- or post-employment of staff. Independent regulators “can be subject to undue influence from government, regulated industries and powerful consumer groups.”
- Four strategies (Figure 3.1): levelling the playing field by engaging stakeholders with diverging interests (“capturing a process or decision equates excluding others from it”), transparency (necessary, “not sufficient”), external accountability through audit institutions, competition authorities and regulators, and organisational integrity. Consultation late in the process “may indicate use of the tools of stakeholder engagement as a mere formality”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Capture is a durable relationship, not a transaction, and can proceed by legal means | moderate | definitional; Johnston’s “influence market corruption” (Box 1.2) |
| C2 | Capture is identified by a shift away from an explicitly defined policy goal toward a captor that intends to extract rents; no particular means are presupposed | moderate | definitional, adapted from Carpenter and Moss (2014) |
| C3 | Capture can be achieved without the decision maker’s knowledge, by manipulating the information supplied | moderate | asserted; Helm (2010) cited; the energy case’s “biased scientific expertise” |
| C4 | Political connections carry measurable value to firms | strong in the cited studies | Box 1.1, secondhand (Fisman 2001; Faccio 2006; Khwaja and Mian 2005; others) |
| C5 | Perceived undue influence correlates negatively with transparency and trust, and positively with perceived wasteful spending, across OECD countries | weak | cross-country correlations of survey indicators (Figures 1.2, 1.6), R² 0.61–0.78, no controls |
| C6 | Capture is self-reinforcing through reinvested rents and increasing returns to rent seeking | weak | asserted, with Murphy et al. (1993) cited |
| C7 | Inclusive engagement of diverging interests, transparency, independent oversight and organisational integrity together mitigate capture | weak | policy recommendation; Box 3.1 shows transparency’s effects are context-dependent |
Concepts
- Policy capture: as quoted above. State capture: of central government at large scale. Regulatory capture: “capture of a regulator, industry regulatory agency or regulatory process”, for example when private actors “persuade regulators to create, avoid or alter rules, or to show leniency in enforcing them”.
- Influence market corruption (Johnston 2006, 2014): private wealth seeking influence within strong institutions, where “abuses are often a matter of pushing legal, even desirable, activities and connections to unacceptable extremes”.
- Due and undue influence: distinguished by transparency and inclusiveness (Figure 1.4).
Connections
It builds on Stigler (1971) for capture of regulation as a cheaper source of rents than market strategies, on Olson (1965) for the collective-action dilemma of diffuse interests, on Krueger and Bhagwati for rent seeking, on Carpenter and Moss (2014) for the criteria, on Johnston for influence markets, and on Martimort (1999) for capture that grows as public attention wanes. It opens chapter 1 with North (1994): formal rules “are created to serve the interests of those with the bargaining power to create new rules.” The record holds North (LIT-489, THEORY-429) and, unread, Stigler (LIT-1324).
Bearing on the record
- TERM-066. The TERM cites the OECD sense through A20’s paraphrase
and calls it “not held”. The report’s own wording differs from the
paraphrase in two ways that matter to the TERM:
- It requires the captor’s intent (“by the intent and action of this group or person”; criterion 2, “must intend to extract rents”). The TERM’s fidelity-relative sense has no intent condition: in CLAIM-330’s loop, capture can arise with nobody intending it. So the two senses differ by intent as well as by the standard. That supports the TERM’s choice to keep them apart, and it is a third difference the TERM’s “The sense not adopted” should state.
- It requires an explicitly defined policy goal (criterion 1), the TERM’s “standard from outside”, made explicit. The TERM’s two conditions (an outside standard, durability) match the report’s criterion 1 and its “consistently or repeatedly”.
- A20’s “a repeated diversion of public decision-making toward narrow interests, through potentially legal or illegal channels” is a fair paraphrase, minus the intent clause.
- CLAIM-404. The report states the informational route outright: undue influence “can even be achieved without directly involving (and without the knowledge of) public decision-makers, by manipulating information”, and among its vehicles “Providing research and analysis”, “Using think tanks to produce research” and “Providing manipulated information and expertise”. Its risk factor of technical complexity, where biased analysis “can be considered useful guidance”, is the condition under which the informational route works. It has no evaluative route: nothing on capture of the criteria by which evidence is weighed, except that capture is measured against stated goals. So it supports the claim’s first route and is silent on its second.
- CLAIM-404, evaluators. “SAIs, competition authorities and regulatory agencies are a natural target for capture”, and independent regulators can be unduly influenced “from government, regulated industries and powerful consumer groups”. That is the claim’s “formal independence of the evaluator’s procedure does not make its evaluation independent”, stated as a policy finding, with Box 3.3’s survey of where the pressure enters: budgets, appointments, revolving doors.
- Consultation. “This timing is often too late to influence the process, and may indicate use of the tools of stakeholder engagement as a mere formality” is A27 §6’s “satisfy a nominal requirement to consult affected agents while denying them effective participation”, which CLAIM-404 quotes.
- THEORY-146. The energy case’s “polluting energy companies fund biased research to underpin their business case” is the phenomenon THEORY-146 models, stated without a model.
- Self-reinforcement. The report’s “vicious circle”, rents reinvested in influence, is CLAIM-330’s loop in prose, and its “increasing returns” to rent seeking is LIT-1314’s mechanism carried to influence. It derives neither.
- CLAIM-359. The four strategies are a policy-level counterpart of the claim’s safeguards: audits “independent of the audited” appear as independent audit institutions, which “must also … insulate themselves from undue influence”.
No THEORY is filed. The report is definitional and advisory, and its empirical content is secondhand. No instruction for machine-learning practice; nothing for the anthology.
Limitations
- Not research. The definition and criteria are adopted from others. The evidence of Box 1.1 is cited, and the cross-country figures are correlations of perception indicators without controls.
- The public interest is taken as given. Criterion 1 needs an explicitly defined goal. Where goals are contested or implicit, the criteria cannot be applied, and the report does not say what then.
- Intent is hard to observe. Criterion 2 requires the captor’s intent to extract rents. The report gives no way to establish it, and its own example of capture “unawares” through repeated interaction sits awkwardly with it on the official’s side.
- Recommendations are not evaluated. Box 3.1 itself reports that the effects of transparency are context-dependent.
Open questions
- How should capture be identified where the policy goal is contested? The report’s criteria presuppose that it is not.
- Is intent necessary? The report’s risk factor of repeated interaction, with attention waning, describes a drift no one need intend.