SPARC Cost of Knowledge Study: 95% Public Investment in Articles

International Open Access Week 2026 runs October 19 through 25, and this year’s theme, “The Cost of Knowledge,” arrived with a number attached. More than 95 percent of the total investment behind a published research article is typically made before a publisher ever sees the manuscript, according to two independent studies from the University of Sydney Business School released October 14 through SPARC, the Scholarly Publishing and Academic Resources Coalition.

That percentage makes the headline. The deeper story is the method behind it. For the first time, researchers have a way to count the resources that go into an article across the full research process — from initial idea through grant applications, ethics approvals, data collection, analysis, writing, and revision — then set that number against independent estimates of what publishing itself costs. The comparison flips the usual conversation. Instead of asking what it costs to publish, it asks who already paid before publication started. The organizers of Open Access Week have framed the week around the financial, human, environmental, and infrastructural costs of scholarly publishing, and they have put one question under all of them: who captures the value?

Applied to one social science article at a US public doctoral university, the method lands on a conservative figure near US$70,000 in pre-submission investment. The public supplies the overwhelming majority of that funding, said Jane Andrew, Professor in the School of Accounting, Governance and Regulation at the University of Sydney. “Prior research has estimated various components of that investment but never brought them together across the full research process,” she said. “Our method does that, revealing the scale of the largely unaccounted-for public subsidy to private sector publishers.”

The accounting runs the entire pre-publication cycle, including failed and successful grant applications, ethics committee work, data collection, analysis, drafting, and the salaries and overheads those hours represent. That pre-submission money arrives through national research grants, university salaries, doctoral training programs, library staff support, and the indirect costs on every federal award. It then compares that figure with published independent benchmarks for scholarly publishing services, which put a publisher’s marginal contribution at a few hundred to roughly one thousand dollars per article depending on rejection rates. The gap between the public’s US$70,000 and the publisher’s invoice is where the “more than 95 percent” estimate comes from. A free online calculator lets any researcher or institution run their own numbers through the same model.

Where the return goes

The companion analysis tracks the four largest commercial academic publishers — RELX (Elsevier), Informa (Taylor & Francis), Springer Nature, and Wiley — from 2005 to 2025. Academic publishing was more profitable than the rest of the parent company in 68 of the 69 company-years examined. In 2025, margins in the academic publishing arms ran from 24 percent to 38 percent. Elsevier’s was 38.1 percent, against 31.7 percent for RELX as a whole.

The stability is the sharpest point. The publishers’ combined operating margin rose from 14.9 percent to 27.5 percent across two decades that included open access mandates, read-and-publish agreements, the 2008 financial crisis, and the COVID-19 pandemic. Open access has not squeezed the commercial side of the ledger; it has simply moved where the invoice lands.

Then the shareholder arithmetic. Between 2005 and 2025, the four parent companies returned US$38.8 billion to shareholders through dividends and buybacks, with academic publishing contributing disproportionately to those returns, according to the SPARC project page for the Cost of Knowledge report. RELX returned US$30.8 billion of that total. Its annual cash distributions climbed from US$349 million to US$3.63 billion — more than a tenfold increase — and reached 95.9 percent of operating cash flow in 2025 alone.

“This new analysis sharply highlights the way in which commercial companies inappropriately profit from the current academic publishing system and faculty labor,” said Nick Shockey, Associate Executive Director at SPARC. “These findings quantify just how efficient these companies have become in removing billions of dollars from the research ecosystem and returning them as cash to shareholders.” The full methodology and the companion profitability analysis are available through the University of Sydney’s announcement.

Photo by Markus Spiske on Unsplash

Here’s the catch

The confidence interval around that 95 percent figure is doing quiet work. Both studies are new this month — one still a preprint, the other a working paper from the same University of Sydney group — so they have not been through formal peer review. The accounting model is conservative by design: the US$70,000 example counts faculty time at base rates, standard overheads, and the minimum hours a serious project requires. But the headline percentage still depends on the publishing-cost benchmark a reader accepts. If a university believes a publisher adds US$4,000 of value per article rather than the benchmark range Alexander Grossmann and Björn Brembs put closer to US$1,000, the public share shrinks. It does not collapse, but it moves.

The shareholder numbers carry the same caveat in reverse. RELX and Informa run businesses far beyond journals — RELX in legal and risk analytics, Informa in events. The authors calculate that academic publishing contributed disproportionately to returns, but the US$38.8 billion payout is corporate-level. Critics will probe that attribution, and they should.

The harder catch is operational. SPARC’s calculator at costofknowledge.org lets a researcher plug in variables and find the public investment behind their own paper, and the tool is genuinely useful as a teaching device. But awareness does not renegotiate a read-and-publish deal. A figure only bites when someone with authority uses it at the negotiating table, and the people who fill in the calculator are seldom the people who sign the contract. That is the same gap that has slowed earlier transparency efforts such as the Plan S journal comparison service.

The name itself has history. In 2012, mathematician Timothy Gowers launched a boycott of Elsevier under the same phrase, “The Cost of Knowledge,” and more than 20,000 researchers signed it. The petition reshaped the argument but did not break Elsevier’s pricing power. Fourteen years later, four publishers are posting 24 to 38 percent margins during the same week universities are told to celebrate open access. That’s not irony. That’s the structure working as designed.

Grading the hype

What these studies establish is narrower and stronger than the commentary around them. They do not prove that publishers add no value. Typesetting, manuscript platforms, integrity screening, and long-term preservation are real services, and even the independent cost benchmarks assume a functioning publishing operation exists somewhere. What the studies do establish, with arithmetic rather than slogans, is that the public’s pre-submission investment dwarfs a publisher’s marginal cost, and that the largest commercial houses have converted that asymmetry into two decades of rising margins and cash returned to investors.

That reframing matters more than any single percentage. Funders and libraries can now produce an accounting artefact for any article, and the same method travels to the next one without losing its shape. Whether the artefact changes contracts, reshapes evaluation criteria, alters where authors choose to submit, or moves library budget lines is a separate question the studies deliberately leave open. On the hype scale, the diagnosis earns high marks. The remedy remains unspecified, which is honest and also a little convenient.

Open Access Week, with better numbers

This year’s official theme invites reckoning with the financial, human, and environmental costs of creating, sharing, and sustaining knowledge. The SPARC release gives libraries something more concrete than a theme. During the week, institutions from Berlin to Barcelona, Oregon to Alberta, are running panels, data workshops, and repository sessions. Some are moving beyond programming toward budgets.

The University of Oregon Libraries, for instance, let their article processing charge (APC) award sunset after the 2025-26 academic year. The fund had supported 130 Oregon authors over six years, clear evidence of demand; administrators decided that paying individual fees to commercial publishers was not a durable strategy, so the money is shifting toward a book publication subvention pilot run with the provost, the research office, and two campus colleges, according to the library’s Open Access Week note. That kind of move — from per-article APC funds toward institution-owned publishing — is the concrete policy echo of the Cost of Knowledge findings, and it arrives on top of months of APC cost and waiver activism.

In Latin America, the regional social science council CLACSO is running five panels, from privatization and commodification to sessions on artificial intelligence data extraction and epistemic sovereignty. The framing is sharper than in the global North. The region has maintained a largely non-commercial, diamond publishing system for decades, and commercial open access models threaten that system rather than disrupting a subscription status quo.

SPARC is pointing researchers to a low-tech sequence: use the calculator to establish the number, retain your rights rather than signing them away, deposit the accepted manuscript in an open repository, and back community-owned publishing where it exists. Each is low-tech. Rights retention, in particular, has been spreading across UK institutions, and the Sydney figure gives it a price tag: roughly US$70,000 of public money per social science article that can otherwise disappear behind a paywall.

Photo by Markus Winkler on Unsplash

Max Baker, Professor in the same School of Accounting at the University of Sydney, put the question plainly. “The bigger issue that we should continue to explore is why publishers should substantially profit from a research process to which they provide only a minimum input.”

That question, not the 95 percent figure, is the one Open Access Week should carry into library negotiations, into promotion and tenure files where researchers decide where to submit, and into the deposit policies funders write next year.