REF 2029 Output Eligibility: OA Exceptions

REF 2029 output eligibility does not require every journal article or conference paper to be immediately open access. The policy allows four exception categories — deposit, access, technical, and a catch-all “other” route covering staff circumstances, third-party rights and technical infeasibility — each keeping an otherwise non-compliant output eligible, provided the exception is identified, evidenced and stays within a unit’s five per cent non-compliance tolerance.

The REF 2029 open access policy is the UK funding bodies’ requirement that in-scope journal articles and conference proceedings, published between 1 January 2021 and 31 December 2028, be deposited, discoverable and freely accessible to be eligible for the Research Excellence Framework 2029, subject to defined exceptions.

Durham University’s Library Research Support service publishes a working exceptions log — updated as recently as 1 July 2026 — that maps each permitted exception to the evidence a research office should hold, offering a practical template for institutions building their own REF 2029 compliance file.

Contents

What counts as an REF 2029 output eligibility exception?

An REF 2029 output eligibility exception is a recognised circumstance under which an in-scope journal article or conference contribution is exempt from some or all open access criteria while remaining eligible for submission. The REF 2029 open access policy, last updated 13 June 2025, groups these into deposit, access and technical exceptions plus a further “other” exception for circumstances beyond an institution’s control.

Applying an exception correctly means the output is treated as no-detriment and does not count against a unit’s compliance tolerance. Misapplying one, or failing to evidence it, risks the output being reclassified as unclassified during audit.

Two separate exception frameworks apply depending on when the output was first published:

  • Outputs published 1 January 2021 to 31 December 2025 follow the carried-forward REF 2021 exception categories, with one additional exception added for REF 2029.
  • Outputs published from 1 January 2026 follow a revised, consolidated exceptions framework of four categories, alongside tighter embargo maxima (six months for Main Panels A and B, twelve months for Main Panels C and D, down from twelve and twenty-four months respectively).

Individual staff circumstances: who qualifies for an exception?

Staff-circumstance exceptions cover authors whose personal, employment or contractual situation made compliance genuinely impossible. Under the “other” exception, this includes extenuating personal circumstances such as extended leave, industrial action, institutional closure days, and software problems not already covered by the technical category.

A related technical exception applies where an output has a substantive connection to the submitting institution but was published after the author’s employment ended, and compliance could not be determined. A further exception, introduced for REF 2029, exempts outputs authored entirely by staff ineligible for the volume measure, since they had no expectation of needing compliance.

One risk is flagged explicitly: institutions relying heavily on the “other” category are more likely to attract a higher audit risk score, though deposit, access or technical exceptions do not themselves raise risk. A researcher on long-term sick leave whose manuscript could not be deposited within the three-month window would typically be recorded under “other”, with leave dates and repository correspondence kept as evidence — the record Durham’s log recommends holding before, not after, submission.

Third-party rights and technical infeasibility: the other routes

Third-party rights exceptions apply where an output reproduces content — images, data extracts, previously published material — for which open licences could not be obtained at reasonable cost or within the required timescale. Here, deposit and discovery requirements still apply; only access and/or licensing conditions are waived.

Technical infeasibility exceptions cover cases where the failure sits with infrastructure rather than the author: a repository’s short-term technical failure (excluding systemic issues), or an external service provider failure, such as a subject repository that did not lift an embargo on schedule or ceased operating altogether.

Exception category What it exempts Typical qualifying scenario
Deposit All open access criteria (deposit, discovery, access, licensing) Unlawful to deposit; publisher disallows repository deposit; delay securing final peer-reviewed text
Access Access, embargo and/or licensing only — deposit and discovery still required Embargo exceeds policy maxima; third-party content licensing unavailable
Technical All open access criteria Repository or external service provider failure; employment ended before publication
Other All open access criteria Personal circumstances, industrial action, closure days, unlisted software issues

Durham’s exceptions log as a working model

Durham University’s Library Research Support service maintains a dedicated REF 2029 Open Access Exceptions guide, structured as four tabs: deposit, access, and technical/other exceptions for outputs published to 31 December 2025, plus a consolidated four-category framework for outputs from 1 January 2026. Each tab pairs the qualifying scenarios with how the exception is expected to be evidenced during any future audit.

Two features make it a useful model: it separates outputs by publication date, since qualifying criteria and embargo maxima genuinely differ either side of 1 January 2026; and it is candid about what remains unresolved. As of its 1 July 2026 update, Durham states that “the internal process for applying exceptions is yet to be confirmed,” directing queries to its repository administration team meanwhile — a reminder that internal workflows keep evolving well after the REF policy text has stabilised.

Documenting exceptions for audit: what research offices must keep

REF 2029 audit will broadly mirror the risk-based approach used for REF 2021, per the funding bodies’ guidance, while final procedures remain under development. In REF 2021, ten institutions’ submissions were selected for second-stage audit on a risk basis, producing four data adjustments; three were separately selected for substantive sampling; one had to adjust its submission by a single output. For REF 2029, evidence may be required at individual-output level, including written justification for any use of the “other” category.

A minimum evidence file for each exception should include:

  1. The exception category claimed and the REF 2029 guidance paragraph it falls under.
  2. Dates of acceptance, publication and deposit, plus any embargo end date.
  3. Correspondence documenting the barrier — a publisher refusal, or confirmation of a service outage.
  4. A short narrative justification, particularly for “most appropriate publication” or “other” claims.
  5. Confirmation the output falls within the unit’s five per cent, or one-output, tolerance.

Institutions should also monitor exception mix over time: a submission weighted heavily toward “other”, rather than deposit, access or technical exceptions, is more likely to draw a higher audit risk score, even where every claim is legitimate.

Answer-first Q&A

What is the publication period for REF 2029?

The output types in scope of the REF 2029 open access policy are journal articles and conference contributions with an International Standard Serial Number, first published between 1 January 2021 and 31 December 2028. Outputs published from 1 January 2026 follow revised deposit, embargo and licensing requirements, while earlier outputs follow REF 2021-derived rules carried forward for REF 2029.

What publications are eligible for REF 2029?

Eligible outputs include journal articles and conference proceedings with an ISSN, plus long-form outputs such as monographs, book chapters and scholarly editions, which carry no open access requirement. Datasets, code, protocols and artistic creations are also welcomed, though they sit outside the open access policy’s scope entirely.

What does the REF output rating measure?

REF output quality is measured against three fundamental dimensions: originality, the development of new concepts or techniques; significance, the capacity to influence scholarly thought, policy or practice; and rigour, the intellectual coherence and robustness of the methodology. Open access exceptions affect eligibility, not the quality score itself, which panels assess independently.

What this means for research offices ahead of REF 2029

For research administration teams, exception management is now a compliance workstream in its own right, not an afterthought bolted onto repository deposit. With a five per cent (or one-output) tolerance per unit of assessment, a handful of poorly evidenced exceptions can tip a submission over the threshold and trigger removal of otherwise strong outputs. Institutions should build an exceptions log now, using Durham’s date-segmented structure as a template: separate outputs to 31 December 2025 from those from 1 January 2026, record the exception paragraph relied upon, and retain evidence when the exception first arises, not retrospectively. Given that REF 2021 audits produced adjustments even at confident institutions, a living exceptions file — not a retrospective justification exercise — separates a smooth submission from a last-minute scramble.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *