The Oligopoly of Academic Publishers in the Digital Era

Imagine you're a librarian in the nineteen nineties. Every extra journal you buy has a real, physical cost. Paper, shipping, shelving. That friction keeps prices tethered to reality and gives you some leverage. Then the web arrives. The cost of delivering one more digital copy plunges toward zero, and suddenly publishers can bundle everything into giant packages—big deals—that are hard to refuse and even harder to leave. That shift didn't just change how we read; it changed who holds the power. So Larivière, Haustein, and Mongeon asked a blunt question: in the digital era, how much of the world's research did a handful of firms come to control? To answer it, they built a forty-year map of ownership using the Web of Science, or WoS, featuring about 44.48 million records from nineteen seventy-three to two thousand thirteen. They turned this into a database that links each paper to its journal and each journal to its publisher. It sounds simple until you hit mergers. A journal might still list Pergamon Press for years after Reed-Elsevier actually bought it. So the team followed the money, not just the name on the masthead. They reconstructed merger and acquisition histories from the Munroe report through two thousand six, and then from LexisNexis profiles and press releases. If one company acquired at least fifty-one percent of another, they reassigned the journals to the acquirer as of the deal date. That's how a Pergamon title published in nineteen ninety-five ends up counting toward Reed-Elsevier. It's forensic accounting for the scholarly record. By the end of that reconstruction, the headline is crisp. In the natural and medical sciences—think biology, chemistry, engineering—the top five publishers controlled just over half of all papers by two thousand thirteen. Fifty-three percent. Most of that share sat with three commercial houses: Elsevier at twenty-four point one percent of papers, Springer at eleven point nine percent, and Wiley-Blackwell at eleven point three percent. Scientific societies still mattered—American Chemical Society titles accounted for a smaller slice at three point four percent—but the center of gravity was commercial. In the social sciences and humanities, the turn was even sharper. Through the nineteen seventies and eighties, the top five publishers had less than a tenth of the output. Fast-forward to two thousand thirteen, and they're over the halfway mark. Just above fifty-one percent. Elsevier led there too, with sixteen point four percent of papers, followed by Taylor and Francis at twelve point four percent and Wiley-Blackwell at twelve point one percent, with Springer and Sage rounding out the top tier. The names are familiar, but the scale is new: a few firms, spanning hundreds of disciplines, handling most of the world's peer-reviewed work. Now, that's the panorama. The texture depends on where you look. In chemistry, having a dominant society publisher like the American Chemical Society pushes concentration higher. In physics, societies act as counterweights. The American Physical Society, the American Institute of Physics, and the Institute of Physics keep more of the field in community hands, and you can see it in the shares: Elsevier's portion of physics journals fell from twenty-eight percent in two thousand one to twenty-one percent in two thousand thirteen, while Springer grew from about three percent to eleven percent. Arts and humanities look almost like a different planet. The five big commercial houses together held only around a fifth of humanities output in two thousand thirteen and about a tenth in the arts—much more dispersed, more local. Biomedicine shows the market can shift under your feet. Around two thousand nine, the top five were closing in on half the papers—forty-nine percent. Four years later, that dropped to forty-two percent, thanks in part to new entrants like PLOS ONE and the rise of open access megajournals. Concentration is high, but not static. What actually moved the numbers? Journal migrations. The authors trace waves of acquisition that are hard to miss once you line them up by year. Before nineteen ninety-seven, journals rarely jumped from one camp to another, and when they did in the natural and medical sciences, it was more often from big publishers to smaller ones. Social sciences and humanities barely showed any big-to-small motion at all. Then the late nineteen nineties hit. Taylor and Francis absorbed portfolios from Gordon and Breach, Harwood Academic, Scandinavian University Press, Carfax, and Routledge in a burst across nineteen ninety-seven to nineteen ninety-eight. Reed-Elsevier folded in Butterworth-Heinemann, Ablex, JAI Press, Gauthier-Villars, and Expansion Scientifique Française. Another wave crested in two thousand one. And in two thousand four, a third peak landed when Springer took over Kluwer Academic. Each deal moved fleets of journals in a single stroke. Here's the intuitive question: if your journal moves from a small house to a big one, do your papers get cited more? Better marketing, stronger distribution, richer platforms—you'd expect some lift. Larivière and colleagues looked at that directly. They compared citation impact in the four years before and the four years after a change, focusing on two migration windows—nineteen ninety-five to nineteen ninety-eight and two thousand one to two thousand four—so they were comparing like with like. In the natural and medical sciences, journals that moved from small to big publishers saw a slight uptick in the earlier window, but in the early two thousands window, their citation impact actually dipped after the move. Journals moving the other way, from big to small, showed no consistent pattern. And in the social sciences and humanities, switching camps didn't budge citations in a systematic way. With a few exceptions, the average impact before and after migration stayed below the world average. The takeaway is not that publishers don't add value. It's that the consolidation we see doesn't show up as a clean, universal citation bump for migrating journals. Zoom out from the bibliometrics for a second and think economics. An online journal issue is a non-rival good: my reading it doesn't diminish your ability to read it. Once the platform is built and the paper is uploaded, the cost of serving the next reader—or the next hundred readers—is effectively zero. Libraries fund most of this market—roughly sixty-eight percent to seventy-five percent of journal revenues come from library budgets—and they're price-insulated end users. A graduate student doesn't pay more each time they click. In a world of near-zero marginal cost on the supply side and inflexible, centrally budgeted demand on the buyer side, bundling everything into a single subscription that you can't easily drop becomes a very attractive business. And the business has been very good. In two thousand eleven, journal publishing pulled in about nine point four billion dollars in revenue. Elsevier's profits alone topped two billion dollars in two thousand twelve and two thousand thirteen, with its scientific, technical, and medical division posting margins that never dipped below thirty percent through the late two thousands and peaking at thirty-eight point nine percent in two thousand thirteen. Springer's margins were in that neighborhood too—about thirty-five percent in two thousand twelve—with Wiley and Taylor and Francis not far behind. Those are technology-platform margins, not the margins you see in most content businesses. You can see why it works if you look at costs. The expensive part is the first copy—editing, typesetting, platform overhead—often estimated in the range of twenty to forty dollars per page. After that, every additional download is essentially free. Hybrid open access models, where authors pay to make a single paper open while the journal remains subscription-based, can run around five thousand dollars per article. Meanwhile, the open access megajournal PLOS ONE showed what scale looks like in the other direction, publishing more than thirty thousand papers a year at its peak. Low marginal cost meets high throughput, and the business model follows. A lot of the durability of this structure, as Larivière, Haustein, and Mongeon emphasize, comes from incentives upstream of the checkout line. Careers are built on where you publish, not just what you publish, and evaluation systems lean hard on journal-level metrics. That prestige economy funnels manuscripts to a small set of brands that can then sell access back to the institutions where those authors work. In physics, the presence of arXiv and community funding models like the Sponsoring Consortium for Open Access Publishing in Particle Physics, or SCOAP3, shows there's another way to organize dissemination without ceding as much control to for-profit firms. But those are still exceptions, not the rule. Methodologically, this is one of those studies where the plumbing matters. The Web of Science doesn't index everything, especially outside the core English-language journals, so the results describe the universe that WoS captures. Publisher names lag reality during acquisitions. The team did the legwork to normalize identities and used a clear control rule—fifty-one percent ownership—to decide when to reassign titles, but disambiguating names and tracking partial deals is inherently messy. They also kept the migration analysis anchored in two fixed windows to avoid cross-period apples-to-oranges comparisons. And they were transparent about access: aggregate outputs were shared on Figshare, while the underlying bibliometric records remained under Thomson Reuters' license. All that is to say, the patterns they chart are sharp, but they're not omniscient. They reflect the best longitudinal ledger we have of the mainstream indexed literature. So where does that leave us? With a market that looks and behaves like an oligopoly, particularly after the mid-nineteen nineties. A small group of publishers controls a majority share of papers in the natural and medical sciences and a comparable share in the social sciences and humanities. In two thousand thirteen, they also held roughly mid-fifty shares of journals and citations. Discipline by discipline, strong societies can temper that dominance, and new models can shift the mix, as biomedicine's post-two thousand nine wobble shows. But the gravity of the bundle—low marginal cost, locked-in budgets, prestige-driven supply—keeps pulling in the same direction. The pressure isn't all one-way. Remember the Cost of Knowledge boycott that Timothy Gowers launched in two thousand twelve? That was a signal flare. So were high-profile standoffs where universities paused or canceled big deals, like the negotiations at Harvard and the cancellation in Konstanz. None of those change the numbers in Larivière and colleagues' dataset, but they hint at the countervailing forces any future map might have to include. If you're looking for a moral here, it's not that publishers are bad and open access is good. It's more structural than that. Digitization slashed marginal costs and made bundling irresistible. Evaluation systems concentrated prestige. Libraries underwrote the whole thing. The result was predictable: consolidation without a consistent, field-wide citation dividend for migrating journals. The way out, if there is one, probably looks like physics—community infrastructure, open dissemination, and incentives that reward venue-agnostic quality. The data in this study don't promise that future. They just make clear why the present looks the way it does, and where the few real levers might be.