Who Governs the Certifier Behind Your Certified Parts

Short answer: In New York, "certified" on a repair estimate means certified by a private body, and in practice that means one body: CAPA. The state's insurance superintendent declared the Certified Automotive Parts Association acceptable in a single sentence on 15 July 1993, and no later Departmental document naming it is publicly indexed. Two instruments check that certifier, and there is a gap between them: New York's regulation audits the certifier on eight criteria, every one of which is about what it does and none of which is about who controls it, while the standards industry's own anti-capture rule, which does test composition, reaches the committee that writes CAPA's standards and stops before the board that governs CAPA.

By Anthony Braswell for Quorum Industries LLC, The Autobody Directory · Updated 2026-09-05 · How this was written, and what the machine may not do

How this page was produced: Researched, drafted and checked with AI assistance under human direction, and signed off by the named author. How this site is written

The word on your estimate is doing legal work

New York's Regulation 64 tells your insurer what it may put on the estimate. At 11 NYCRR 216.7(b)(5)(v):

"the insurer shall specify only certified crash parts, in regard to any part that has been duly certified by a qualified certifying entity acceptable to the superintendent;"

Read that carefully, because it is the hinge. The state does not certify parts. The state does not test parts. The state decides which private body is fit to do it, and then the word "certified" on your estimate carries the state's regulation behind it. So the whole weight of that word rests on one prior question: who is acceptable to the superintendent, and on what basis?

The eight things the superintendent looks at

The basis is the next paragraph, 11 NYCRR 216.7(b)(6), reproduced in full:

"(6) In determining whether a certifying entity is qualified and acceptable for purposes of paragraph (5) of this subdivision, the superintendent shall consider the extent to which the entity:

(i) has adopted written standards containing conditions to be fulfilled by a manufacturer of crash parts;

(ii) tests, or contracts with an independent testing organization that tests, crash parts, using suitable equipment and techniques;

(iii) administers its certification program in a nondiscriminatory manner regarding any manufacturer or supplier of non-OEM crash parts;

(iv) provides a system to determine that certified non-OEM crash parts continue to conform with standards prescribed in subparagraph (5)(iii) of this subdivision and, failing to so conform, to decertify and advise crash part users of withdrawals of certification for any such part;

(v) provides mechanisms for quickly receiving inquiries and promptly resolving disputes that arise under the program in regard to consumers, insurers or repair shops;

(vi) provides a means of identifying each certified non-OEM crash part and provides a system of security that guards against misuse of the identification;

(vii) provides updated lists of certified non-OEM crash parts on at least a quarterly basis; and

(viii) provides the superintendent with an annual report, and such other reports as the superintendent may require, highlighting any significant developments, problems or changes relating to certification procedures or requirements."

Eight criteria, and there is no ninth: asked directly, both hosts we read returned NOT PRESENT for a subparagraph (ix), in the same call in which each returned the text of (i) through (viii).

Every one of those eight is about conduct and output.

Not one of them asks who owns the certifier, who sits on its board, who elects that board, or who pays for it. The words "board", "ownership", "membership", "governance" and "funding" do not appear anywhere in the paragraph, on either host we read.

That absence is specific rather than general, and the text shows why. The drafter was plainly capable of writing an impartiality condition, because the drafter wrote one. Criterion (iii) requires the entity to administer its programme "in a nondiscriminatory manner regarding any manufacturer or supplier of non-OEM crash parts". So the regulation is not silent about even-handedness. It requires even-handedness as a standard of conduct toward manufacturers. What it does not do is ask whether manufacturers are represented in the body that owes them that duty. Conduct and composition are different instruments, and only conduct is in the regulation.

The money, and why the money is not the finding

CAPA's revenue comes overwhelmingly from the programme it runs. On IRS Form 990 extract data for the two most recent filed years: for FY2024, program service revenue of $22,108,483 against total revenue of $23,319,357, which is 94.8 per cent, with contributions of $694,600, about 3.0 per cent. For FY2023 the same split is 94.55 per cent and 3.04 per cent. The only class of fee payer CAPA publishes is manufacturers presenting parts for approval and the subcontractors who stamp, mould, assemble or paint them. Participation "is open to any manufacturer" that meets CAPA's standards and signs its licence, and "A fee must be paid to become either type of Participant." No fee amount or schedule is published anywhere, and the 990 extract does not break the revenue out, so that the fee payers are the certified manufacturers is an inference from the programme's published structure. It is a strong inference and it is not a fact we have in a document.

Now the discipline, stated plainly, because a reader could easily be sold something here. Funding by the certified is the normal structure of product certification. UL is paid by the manufacturers whose products it lists. The ASTM and ASME test regimes work the same way. NSF's surviving programmes work the same way. Fee for service certification is the ordinary commercial form of that entire industry, and nobody treats it as remarkable. If the point of this page were "CAPA is funded by the people it certifies", the page would be a description of product certification generally and would be worth nothing to you. The money is context. It is not the finding.

The seats, and the rule that stops one room short

The finding is about governance, and it needs two rooms.

Room one, the board that governs CAPA. From CAPA's own published roster, enumerated person by person rather than taken from any printed total, there are eighteen seats.

Nine are insurers, exactly half, and half is not a majority, because a majority of eighteen is ten. Manufacturers hold none of the eighteen. The payer of the claim, the seller of the part and the buyer of the part hold fifteen of the eighteen between them.

Room two, the committee that writes the standards. Nineteen members in five categories.

The largest single category is four of nineteen, 21.1 per cent.

Now the rule. The industry's own anti-capture instrument is the ANSI Essential Requirements, edition January 2025, which governs the development of American National Standards, three of which carry the ANSI/CAPA designation. The requirement itself is at clause 1.3, "Balance":

"The standards development process should have a balance of interests."

"Participants from diverse interest categories shall be sought with the objective of achieving balance."

And the benchmark ANSI publishes for that requirement, at clause 2.3:

"Historically the criteria for balance are that a) no single interest category constitutes more than one-third of the membership of a consensus body dealing with safety-related standards or b) no single interest category constitutes a majority of the membership of a consensus body dealing with other than safety-related standards."

ANSI's separate rule against dominance is at clause 1.2, together with the definition it turns on:

"The standards development process shall not be dominated by any single interest category, individual or organization."

"Dominance means a position or exercise of dominant authority, leadership, or influence by reason of superior leverage, strength, or representation to the exclusion of fair and equitable consideration of other viewpoints."

The requirement attaches to the consensus body, the room where a standard is voted. It does not attach to the standards developer's board of directors. Apply it and both halves come out. The technical committee at 21.1 per cent for its largest category sits below one-third and far below a majority, so on its published composition it passes comfortably, with manufacturers holding four of its seats. The board is not subject to the requirement at all, and one interest category holds exactly half of it.

Both limbs are historical benchmarks rather than a pass mark, an alternative formulation can be approved by ANSI's Executive Standards Council, and the stated consequence of an unbalanced consensus body is outreach. Which limb applies turns on whether the standards are safety-related, which is not established here; the committee's 21.1 per cent sits under either.

So the mirror is this. The manufacturers who supply nearly all the money sit on the committee that writes the standards and hold no seat on the board that governs the certifier. The single consumer representative sits on the board and holds no seat on the committee. Neither body contains a vehicle manufacturer or a government body. The committee's three general-interest seats are held by consultancies; the roster gives employers, not professions, so it does not establish whether any holder is an engineer or how independent any of them is. That gap, between the instrument that tests composition but not the board, and the instrument that governs the board but never looks at composition, is the whole of this page.

Since 2019 there has been one candidate

The regulation is written for a market with several certifiers. It speaks of "a qualified certifying entity" and asks the superintendent to weigh "the extent to which" an entity meets eight criteria. That is comparative language. There is currently nothing to compare. On 20 June 2019 NSF announced:

"NSF today announced it will discontinue its Automotive Parts Certification program and affiliated automotive certification and registration programs, effective September 30, 2019."

"Parts manufactured after September 30, 2019 are not certified nor eligible to be certified."

"Any use of a sticker bearing the NSF mark on parts manufactured after September 30, 2019, is prohibited and such marked parts are not certified."

That matters at the counter as well as in the abstract: a part made after 30 September 2019 bearing NSF automotive marking is not a certified part. Parts made on or before that date are not affected, which matters because aftermarket stock sits in warehouses for years. The same release says manufacturers "may continue selling existing inventories of NSF certified parts after September 30, 2019 as long as the parts meet all requirements and were manufactured on or before September 30, 2019", and it treats parts manufactured on or before that date as "NSF certified" for the lifetime of the part.

New York's own position did not change in 2019. The single Departmental document in evidence names CAPA and no one else, so there is no record of a second entity ever having been acceptable to this superintendent. What 2019 removed was a competitor in the national market, not an alternative on a New York estimate.

We looked for a second certifier of aftermarket collision parts and did not find one, and that was a search rather than a survey.

Who checks CAPA, in documents you can open

One document. Insurance Circular Letter No. 7 (1993), dated 15 July 1993, signed by Superintendent of Insurance Salvatore R. Curiale:

"The Insurance Department has determined, pursuant to the criteria set forth in the Regulation, that the Certified Automotive Parts Association (CAPA) is qualified and acceptable for the purpose of certifying non-OEM parts."

Asked directly, the letter names no other certifying organisation and states no term, no review, no renewal and no expiry. A site-restricted search of dfs.ny.gov returned exactly one document naming CAPA, that one. As a control, a differently worded search of the same site did surface a 2004 General Counsel opinion on Regulation 64 crash parts, so the search does find Departmental documents on this subject when they exist. The honest boundary is that a search index is not the Department's file: this shows that no later document is publicly indexed, not that none exists. We are not claiming that no annual report under criterion (viii) has ever been made to the superintendent. None is public, and the position is unknown in both directions. The route to an answer is a FOIL request to the New York State Department of Financial Services, which publishes these files nowhere else.

One further thing the eight criteria give nobody a way to ask about. Since 2 March 2018, Intertek has held "exclusive rights to operate the CAPA certification program around the world, which includes managing daily operations, administering certification marks and labels, performing marketing activities and conducting testing and inspection services", while CAPA "will continue as a non-profit standard development organization". Intertek also appoints one CAPA director under that same licence, and in three project notices ANSI published, the contact given for the standards developer is an intertek.com address. Hold that against criterion (ii), which asks whether the entity "tests, or contracts with an independent testing organization that tests". This page makes no claim that criterion (ii) or criterion (iii) has been breached. The point is narrower and it is about the instrument: nothing in the eight criteria addresses the relationship between a certifier and the operator of its programme, so the question has no place to be asked.

The case that nothing here is wrong, put at its strongest

It deserves to be put properly, because it is a good case.

Every fact in the seat count above came off CAPA's own website, because CAPA chose to publish it. It publishes its directors by name and employer, its technical committee the same way, its supporters, what it certifies, its complaint route and its decertification rule. An organisation concealing capture does not publish the roster that would evidence it.

The enforcement record is real and it runs against the funders. CAPA maintains a public register of notices on parts, 59 of them between 12 December 2018 and 15 May 2025, and its complaint programme states: "CAPA reserves the right to decertify the lot number or the part number immediately, prior to investigation." A certifier controlled by the companies paying it does not keep a public list of their failed parts and does not reserve the right to pull a certificate before investigating.

Six of the eight criteria map onto published programme features that anyone can check. And the balance argument cuts CAPA's way exactly where it operationally bites: if the worry is that insurers write the tests to be easy, the room where the tests are written is the balanced one. In thirty-three years no enforcement action, adverse determination or withdrawal of acceptability has been published, which, as above, is a statement about what is indexed and not about what is in the Department's file.

What we could not verify

We are not saying that any insurer has influenced any certification decision. There is no board minute, no vote, no agenda and no recorded decision in evidence, in either direction. The record at this control point is empty: no evidence of capture, and no evidence of anything that would prevent it. A reader who says "so what" at this point is entitled to say it, and the honest answer is that the argument is about structure.

We do not claim that CAPA's board is unusual among certifiers. No comparator certifier's governance was examined. The comparison here is against a written rule, not against UL's board or NSF's.

The seat count is dated only by when a web page was read, on two days in September 2026. CAPA's board page carries no "as of" date, and the filed federal return that would date it against a document could not be obtained. Do not read "nine of eighteen" as a fact about any tax year. The revenue figures are ProPublica's extraction of IRS data, twice read, not the return itself. Whether CAPA's technical committee is formally its ANSI consensus body is not established, so what is compared above is a published committee against a published rule, and it is not an ANSI finding of compliance or non-compliance either way. Three CAPA standards carry ANSI approval and are free on request; the other eleven are not published as text anywhere, so most of the programme a state regulation makes operative runs on documents you cannot read.

On other states, the scope is exact and small. New York appoints a certifier. Hawaii requires certified or approved parts without naming any certifier or any appointing authority. Georgia and Rhode Island both regulate aftermarket crash parts in detail and neither mentions certification in any form. That is five jurisdictions somebody in this project has opened, and it supports no sentence beginning "most states".

The specific walls we hit, listed so you can see them:

• CAPA's filed IRS Form 990 itself. Only ProPublica's extract of the IRS data could be obtained, so no filed return dates the board roster.
• The fee schedule. CAPA publishes that a fee must be paid and no amount, and the 990 extract does not break the revenue out, so the identity of the fee payers is an inference from published structure and not a documented fact.
• Eleven of the fourteen CAPA standards. Only the three carrying ANSI approval are available, so the text most of the programme runs on could not be read.
• Any Departmental document about CAPA later than 1993. A site-restricted search of dfs.ny.gov returned one, and a control search of the same site found other Regulation 64 material, so this establishes what is publicly indexed and not what sits in the Department's file.
• The current wording of 11 NYCRR 216.7(b)(5)(vi), the warranty subparagraph. It would not confirm identically on the second host: the Society of Collision Repair Specialists copy is a printing current only through 30 April 2002 and words that subparagraph differently, so only subparagraph (vii) is quoted here as squarely confirmed on both printings.

What you can actually do

Not much through the certifier, and it is better to say so than to invent a remedy. Ask the shop for the seal number: a CAPA seal carries a unique number and bar code and is built to self-destruct if it is peeled off, so it is checkable and there is a complaint route attached to it. Ask who warrants the part, because the certifier does not: CAPA puts the warranty between buyer and seller, and so does the regulation, which requires the non-OEM manufacturer to warrant the part. The current text at 216.7(b)(5)(vi) requires a written warranty running for at least the period of your ownership of the vehicle, though the older 2002 printing words that subparagraph differently, so read the current text before relying on it. What is squarely confirmed on both printings we read is 216.7(b)(5)(vii): "the insurer shall cause the damaged vehicle to be restored to its preloss condition consistent with the non-OEM warranty, at no additional cost to the insured and within a reasonable time, if the non-OEM fails to honor its warranty required in subparagraph (iv) or (vi) of this paragraph". Keep the estimate, keep the seal number, and keep the part number. And if you want to know whether anyone has looked at the certifier since 1993, the only route is a FOIL request to the Department of Financial Services, because the answer is not published.

Sources

• 11 NYCRR 216.7, New York Regulation 64. Two independent hosts, both read on 2026-09-04, returned paragraph (b)(6) and subparagraph (b)(5)(v) word for word identically: Cornell Legal Information Institute, https://www.law.cornell.edu/regulations/new-york/11-NYCRR-216.7 (source note: amended effective 2/1/2017 and 6/9/2021), and the Society of Collision Repair Specialists copy, https://scrs.com/wp-content/uploads/2017/09/NY-Unfair-Claims-Hndlg-REG64.pdf (printed currency note: "Text is current through April 30, 2002").
• New York Insurance Circular Letter No. 7 (1993), 15 July 1993, https://www.dfs.ny.gov/industry_guidance/circular_letters/cl1993_07 read on 2026-09-04.
• ANSI Essential Requirements: Due process requirements for American National Standards, edition January 2025, published by ANSI, https://share.ansi.org/Shared%20Documents/About%20ANSI/Current_Versions_Proc_Docs_for_Website/ER_Pro_current.pdf The clause numbering relied on above was confirmed on a second, independently hosted copy: IEEE's copy of the Essential Requirements, https://standards.ieee.org/wp-content/uploads/import/governance/audcom/ansi_essreq.pdf, which is the January 2008 edition and carries the same three clauses at the same numbers, 1.2 Lack of dominance, 1.3 Balance and 2.3 Balance. The quoted text above is from the January 2025 edition. Both copies read on 2026-09-04.
• CAPA Board of Directors, https://www.capacertified.org/About/BoardOfDirectors read on 2026-09-04 and read again on 2026-09-05.
• CAPA Technical Committee, https://www.capacertified.org/About/TechnicalCommittee read on 2026-09-04.
• CAPA Participant Program Overview, https://www.capacertified.org/Participants/ProgramOverview read on 2026-09-04.
• CAPA Complaint Program Overview, https://www.capacertified.org/Complaints/CAPAComplaintProgramOverview read on 2026-09-04.
• CAPA Public Notices and Product Safety Alerts, https://www.capacertified.org/PublicNoticesAndProductSafetyAlerts read on 2026-09-04.
• NSF, "NSF International discontinues automotive parts certification programs", 20 June 2019, https://www.nsf.org/news/nsf-international-discontinues-automotive-parts-certification-programs read on 2026-09-04.
• Intertek, agreement with CAPA, 2 March 2018, https://www.intertek.com/news/2018/03-02-agreement-with-capa-for-aftermarket-auto-parts-certification/ read on 2026-09-04.
• Certified Automotive Parts Association, EIN 52-1555372, IRS Form 990 extract data via ProPublica Nonprofits, https://projects.propublica.org/nonprofits/organizations/521555372 read on 2026-09-04.

Corrections

This page was published on 2026-09-05, and nothing in it has been corrected since. When a correction is made, it will be recorded here with the date, what the page used to say, and why it changed.

General consumer information: not legal, insurance, or financial advice. Requirements, coverage, and practices vary by state, policy, and manufacturer.

More guides