Contribution
It names and operationalizes a form of regulatory capture that the transactional literature misses: capture through the design of the regulatory framework itself, by experts motivated by ideology rather than by firms through lobbying. It makes the “public interest” an observable counterfactual, measured by surveying which values drive public judgements, and it shows a large expert–public cleavage on the antitrust consumer welfare standard (CWS) that partisanship does not explain.
Key insight
Capture need not divert a decision. It can set the criterion by which every later decision is made. “Ideological capture, in other words, is not simply about what language is put into any given standard; it is also about the public goals that are left out of the standard.” Once the criterion excludes a value, no lobbying is needed: “If business interests get what they want (merger approval) most of the time under the new guidelines, we would not expect to find pervasive evidence of intense lobbying or business influence.” Whether a criterion is captured is then an empirical question about whose values it omits, answerable from measured public values rather than readings of legislative intent.
Assumptions
- Ideology in Lukes’s sense: “systems of ideas and values that shape the way people interpret social life, but which also establish and maintain power relations between social groups”, not left–right position. Capture works through the third face of power, following Li (2023): interest groups “constrain and distort” how regulators define the public interest.
- The public interest is “the counterfactual state of regulation that would have obtained in the absence of expert influence” (after Carpenter and Moss), which is “in a regulatory process that broadly reflects public values”. Note 9: “we do not equate the public interest with public preferences”.
- On technocratic issues the public “routinely extrapolates from core political values”, and these can be measured.
- Samples. Public: a nonprobability online sample of 2,092 US adults (Bovitz/Forthright, January 2024), quota-matched and post-stratified on education, partisanship and income. Lawyers: a convenience sample of 103 antitrust lawyers, 48 recruited by Facebook advertising and 55 by email to merger practice groups (November 2023 to March 2024), 51.4% Democrats against 28.2% of the public sample.
- Pre-registration of hypotheses H1–H10; deviations listed in an appendix not read here (two are flagged in the main text, notes 33, 39 and 40).
Key results
- Definition. Ideological capture occurs “when (1) regulators create standards that (2) are based on an ideologically motivated (expert) view of the ‘public interest’ and (3) consistently and repeatedly marginalize important public values and goals”. Disagreement alone is not capture: “so long as that framework does not marginalize important public values, it does not meet our definition of capture.”
- Strength, not existence, is normative. Capture runs “from weak to strong” on two conditions: “(1) whether the experts promoting a given framework have any special claim to scientific legitimacy (a question of truth); and (2) whether the regulators elicited and considered public views before making a decision (a question of process).” The lead-pipe hypothetical: adopted against public opposition after comment, on evidence of harm, it “would not be normatively troubling”.
- History (secondary). The CWS arrived through the Supreme Court’s 1979 adoption of Bork’s revisionist reading (Reiter v. Sonotone) and the 1982 and 1984 merger guidelines, which were exempt from notice and comment (note 12), not through legislation. Expert support was bipartisan, and party platforms did not mention antitrust (none from 1992 to 2012, note 15). Kwoka’s meta-analysis is cited: across 42 mergers and 119 products, prices fell in 38.6% of cases and rose in 61.3%, by about 9% on average (note 25).
- Marginalized values (conjoint, H8, Fig. 1). Effects on support for challenging a merger, in percentage points: a 5% price rise +11.1; price cuts of 5–20% −2.6 to −7.7, significant only at 10% and 20%; worse quality +26.6 (+26.7 later in the text); better quality −7.2; more lobbying +8.1; workforce cuts +14.8 to +26.3; more bailout risk +4.7 (+4.8 later). “Just about any of these risks alone (except for increased bailout risk) more than offsets the public demand for a price reduction as large as 20%.”
- Trade-offs by group (H9). Republicans and Democrats do not differ significantly in the weights they give any attribute. Lawyers are about 30 points less likely to support a challenge overall, and a further 14 points less likely when the merger cuts 20% of the workforce.
- Expert–public cleavage (H4–H6, Figs. 2–3). Support for replacing the CWS with a standard giving “equal weight to the impact on consumers, workers, small businesses, and local communities”: 80.6% of Democrats, 81.4% of independents, 74.8% of Republicans. Partisan gap 5.9 points; public–lawyer gap 42.6 points; lawyers’ support 36.7%, more than 30 points below strong Republicans (69.5%).
- Values behind the cleavage. Lawyers do not differ from the public on whether business regulation is needed to protect the public interest (an unregistered placebo, note 33), but are about 18 points less likely to say large corporations have too much power or that government is run by a few big interests. Each belief predicts public support for replacing the CWS.
- Beliefs about effects (H1–H3, Figs. 4–6). 21.8% of the public agree that mergers lower prices and 46.8% disagree; 72.3% agree that mergers give companies too much political influence. Lawyers are more sanguine on benefits and much more likely to disagree that mergers create social risks (by 0.28 to 0.33).
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Capture can operate through the regulatory criterion, with no lobbying or improper exchange, when experts’ framework omits values the public holds | moderate as a definition and argument | the conceptual section; the CWS case |
| C2 | In this public sample, layoffs, product quality and lobbying weigh more than price reductions in judging whether to challenge a merger | strong for the sample | pre-registered conjoint, n = 2,092, Fig. 1 |
| C3 | In these samples, being an antitrust lawyer predicts opposition to replacing the CWS far more than partisanship does | strong for the samples, moderate beyond them | Fig. 3; robust to demographics per appendix F (not read) |
| C4 | The cleavage is rooted in values about corporate power that law-and-economics teaching implicates | moderate | the 18-point belief gap and its prediction of public support; causal direction not identified |
| C5 | The CWS evidences a strong form of ideological capture | moderate | C2–C4 plus the historical case, which rests on secondary sources |
| C6 | Ideological capture is likely pervasive in American government | weak | conjecture in the conclusion |
Concepts
- Ideological capture: the three-part definition above.
- Regulatory capture, in the definition it widens: “the process by which regulation, in law or application, is consistently or repeatedly directed away from the public interest and toward the interests of the regulated industry, by the intent and action of the industry itself” (Carpenter and Moss 2014).
- Marginalized values: considerations that weigh heavily in measured public judgements but fall outside, or are sidelined by, the standard.
- Expert–public cleavage: divergence in policy preferences between a professional class and the public, larger than partisan divergence and traceable to differing values.
- Faces of power (Lukes): lobbying, agenda-setting, and shaping how interests are conceived. Ideological capture is the third.
Connections
It extends the regulatory-capture literature (Carpenter and Moss 2014; Carpenter 2014’s demand for a public-interest counterfactual; Dal Bó 2006) with Lukes’s third face of power and Li (2023)‘s account of experts defining the public interest. Its history draws on Berman (2022), Teles (2008), Khan (2018) and Short (2022a, 2022b). Its measure of merger effects is Kwoka (2014). Note 4 distinguishes it from informational lobbying models: ideologues “seek to change broad regulatory frameworks … not by leveraging an informational asymmetry per se, but by developing ideological perspectives that redefine the ‘public interest’.”
Bearing on the record
- CLAIM-404: prior art for the evaluative route, and empirical. The claim’s criteria route (“With U_λ(a) = v₁(a) + λv₂(a) and λ = 0 … The limitation is not missing evidence. It is the adopted evaluative framework”) is this paper’s definition, and the CWS is an instance with λ ≈ 0 on workers, lobbying and bailout risk. The claim’s “What it does not say” lists this paper among unread works on which nothing rests. It is now read, and it supports the evaluative half as a possibility realized in at least one measured case. Note 4 is the claim’s split stated by the authors: capture “not by leveraging an informational asymmetry per se” but by redefining the public interest. The claim’s evaluator-in-the-loop part gets nothing from this paper, which has no dynamics.
- TERM-066. The term requires “a standard specified apart from the captured process”. This paper supplies one kind: measured public values, explicitly not public preferences and not legislative intent. It also differs from the term’s adopted sense. Its capture is the institutional sense (diversion from the public interest), and its mechanism needs no loop and no durability through the organization’s own processes, only a criterion. A note on the term could cite it as the measured form of the sense the term set aside.
- CLAIM-348 (adequacy fixed by the evaluative target) is the formal side of the same point: the CWS is a target that its own descriptions cannot justify.
- CLAIM-403. The two normative conditions (truth, and whether regulators “elicited and considered public views”) separate capture that is troubling from capture that is not by the process rung of the claim’s ladder: aired grievances move conflict “from the third to the first face of power”. The guidelines’ exemption from notice and comment is the closed rung.
- CLAIM-277 and CLAIM-406. The lawyers’ values are attributed to professional training, and the profession staffs the agencies. That is a route by which a framework reproduces its own expert class, but the paper does not model or measure it.
- THEORY-146 (influence through which producers of evidence prosper) is the informational analogue. This paper’s route does not pass through evidence.
- THEORY-833 (the measured cleavage, Active) and THEORY-859 (capture through the criterion, Proposed) are filed from it.
No instruction for machine-learning practice; nothing for the anthology.
Limitations
- Samples. The public sample is nonprobability. The lawyer sample is a convenience sample of 103, recruited two ways, and skews Democratic, which the authors note makes the cleavage more striking but which also means it may not represent the antitrust bar or agency staff.
- The counterfactual is not identified. That the public would prefer another standard, and that the CWS reflects experts’ values, is suggested by the cleavage. That the standard would differ absent expert influence is inferred, not shown.
- The quality finding cuts against C5, as the authors acknowledge: the CWS allows quality to be considered. Their reply, that agencies may not do so in practice, is not measured.
- Lawyers as experts. The “experts” who designed the framework were judges, economists and legal scholars of the 1970s and 1980s. The surveyed lawyers are today’s practitioners. The link is assumed.
- History from secondary sources. The process claims rest on cited works (Berman, Teles, Short), not on evidence presented here, except the platform search and the Roper iPoll review in an appendix not read.
- Internal inconsistency. The introduction says the lawyer effect is “four times larger” than the partisan effect; the analysis says “seven times larger” (42.6 against 5.9). Two effect sizes are given twice with different values (quality 26.6 and 26.7; bailout risk 4.7 and 4.8).
- Some tests deviate from registration (notes 33, 39, 40).
Open questions
- Does ideological capture occur where no expert–public cleavage is measurable, for instance where the public holds no values on the question? The definition needs marginalized public values, so it cannot see capture of criteria the public has no view on.
- How would the strength scale (truth, process) be measured? The paper gives the two conditions and no instrument.
- Does professional training cause the cleavage, or does selection into the profession? The data do not separate them.