On rationing, queues and the open-access quota that runs out every autumn
On 2 October 2026 the university library at Leiden sent its researchers a short notice. The annual allowance of open-access articles under the Dutch national agreement with Springer Nature would, it said, probably be exhausted on 9 October, and any author who wanted an accepted article to be freely readable thereafter would be invoiced for it personally, with no central budget from which the charge could be met.[^1] Three days later the consortium that negotiates these agreements for the Dutch universities published its forecast for the other allowances of the year: Emerald in mid-October, Taylor & Francis and Lippincott Williams & Wilkins in the first week of November, and Oxford University Press’s hybrid journals already gone.[^2]
Where it is noticed at all, the episode is treated as library administration — a budget line that ran out a little early. I think this is the wrong reading. The question I want to ask is what kind of economic arrangement runs out every autumn, and my answer is that the Dutch quota is a rationing scheme in the strict sense: an allocation of a valuable thing by a rule other than price. Its existence is the clearest evidence yet that the central economic promise of the open-access transition has not been kept. That promise was that moving payment from readers to authors would expose publishing to price discipline. A quota is what a buyer adopts when it cannot discipline the price.
I first set out the arrangement as it stands, once and in full. I then examine it from four angles: why the promised competition never materialised; what the choice of a quota rather than a price reveals about the buyer; who bears the cost of the queue that results; and why the shortage renews itself each year.
Two thousand and sixty-seven articles
The number that governs the Springer Nature agreement is 2,067. That is the maximum number of articles by Dutch corresponding authors that the 2024–2026 deal will publish open access each year without further charge, across some two thousand hybrid journals — subscription journals that make individual articles free to read on payment.[^3] Articles are covered in the order in which they are accepted, until the allowance is used up. The author of the 2,068th article then has three options. She may pay an article-processing charge, which Radboud University put in 2025 at between 2,500 and 3,000 euros, from a grant or an institute budget if she has one.[^4] She may publish behind the paywall and deposit a copy in her university’s repository six months later, which Dutch law permits for publicly funded research.[^5] Or she may take the paper to a journal whose allowance has not yet run out.
The allowance has run out in every year of the agreement, and each year earlier. The University of Groningen reported the 2024 allowance reached by 6 November;[^6] Radboud reported the 2025 allowance exhausted by 27 October; and the 2026 allowance is forecast to go on 9 October. The table sets these dates out, together with a rough calculation of the full-year demand each implies if acceptances arrived at a constant rate through the year.
| Year | Cap exhausted (by) | Approx. day of year | Implied full-year demand at that pace |
|---|---|---|---|
| 2024 | 6 November | 311 | about 2,400 articles |
| 2025 | 27 October | 300 | about 2,500 articles |
| 2026 | 9 October (forecast) | 282 | about 2,700 articles |
Acceptances certainly do not arrive at a constant rate, so the last column is an order of magnitude rather than an estimate. What it shows is direction: the allowance is fixed, demand grows, and the date of exhaustion moves earlier. If the pace of 2026 held for the whole year, something like six hundred Dutch articles would fall outside the agreement — at the Radboud figures, an unfunded liability of between one and a half and two million euros, borne article by article by whoever happens to be late.
These are the facts. Everything that follows is an attempt to say what they mean.
The promise that prices would do the work
The case for the current arrangement deserves to be stated at its strongest, because it is a good case and I share much of it.
For most of the twentieth century academic publishing ran on subscriptions: a library paid for a bundle of journals and its researchers read whatever was in the bundle. The arrangement had two well-known defects. Readers outside subscribing institutions were shut out of research that was, in most cases, publicly funded; and the price of the bundle bore no visible relation to anything, because libraries could neither compare prices nor walk away from journals their researchers needed. Prices rose for decades well above general inflation, and librarians called the result, without much exaggeration, the serials crisis.
The open-access movement proposed to cure both defects at once. If the author’s institution or funder paid a fee on publication, the article could be free to every reader in the world. And if the fee attached to the individual article rather than being buried in a bundle, it would become a visible price that authors and funders could compare across journals and on which publishers would have to compete. This was the economic logic of Plan S, launched in September 2018 by a coalition of European research funders, which required funded research to be published open access and expected fees to be transparent and “fair and reasonable”.[^7] Transformative agreements such as the Dutch deal were the bridge: a national consortium pays a publisher a single sum covering both access to the subscription content and open-access publication of its own authors’ work, with the balance shifting over time from the former to the latter.
I have no quarrel with the diagnosis. The subscription system was opaque and extractive, and the Netherlands, which has pursued national agreements more doggedly than almost anyone, deserves credit for the share of Dutch research that is now free to read. The difficulty lies in one assumption buried in the remedy: that once there was a price per article, competition would discipline it.
A market of small monopolies
Competition disciplines a price only when the buyer can take its custom elsewhere. In academic publishing the buyer who matters — the author choosing where to submit — cannot, because the institutions that employ her have decided that where she publishes is a measure of what she is worth.
Consider how a young economist is assessed. Hiring committees, promotion panels and grant reviewers rarely have time to read her papers; they read the names of the journals in which the papers appeared. Business schools across Europe keep formal lists grading journals, and promotion files are scored against them.[^8] The global league tables that universities court so assiduously do the same for the institution. The Shanghai ranking allocates a fifth of its total weight to the number of articles a university’s staff publish in two journals, Nature and Science — the first of which is the flagship of the very publisher whose Dutch allowance runs out on 9 October.[^9] One could hardly design a more efficient device for converting a publisher’s brand into a university’s obligation.
The consequence is that journals are not substitutes for one another, and a market in which no product substitutes for any other is not a competitive market at all. It is a collection of small monopolies. A paper accepted by a leading journal in its field cannot be moved to a cheaper outlet without cost to its author’s career, so the author will accept whatever that journal charges, provided someone else pays. The ownership of these monopolies is, moreover, concentrated: Vincent Larivière and his co-authors showed in 2015 that the five largest commercial publishers had come to account for more than half of all papers published, against about a fifth in 1973.[^10] The evidence on how charges are set is what one would expect. Nina Schönfelder, studying payments by British institutions, found in 2020 that the charges of fully open-access journals rise steeply with their citation impact — the price follows prestige — while in hybrid journals, the kind covered by the Dutch deal, the link is much looser, as though the charge had been inherited from the subscription era rather than set by anything resembling a market.[^11]
Add to this that the person who chooses the journal is not the person who pays. The author decides where to submit; the consortium, the funder or the grant meets the bill. This is the structure of the market for prescription medicines, in which the doctor chooses, the insurer pays and the patient consumes, and it is notorious for exactly the reason one would predict: nobody at the point of choice has a reason to care about the price.
Taken together, these features make the remedy look not merely optimistic but inverted. To make the charge per article visible, in a market where the author is bound to particular journals by her employer’s evaluation rules and pays nothing herself, is not to expose publishers to price discipline. It is to invite them to set prices as monopolists do. The universities, having created the lock-in through their own habits of assessment, then meet the resulting bill in their collective role as buyer. They are, in a quite precise sense, paying for market power that they themselves confer.
The Netherlands has tried harder than most to break the habit. In 2019 the Dutch universities and research funders committed, in a position paper entitled Room for Everyone’s Talent, to recognise academics for more than their publication record, and Utrecht announced in 2021 that it would drop the journal impact factor from hiring and promotion altogether.[^12] But careers are international and league tables global, and a Dutch researcher who hopes one day to move to Zurich or Singapore still needs the journal names that Zurich and Singapore read. One country cannot dissolve a prestige market on its own. It can only decline to be its most enthusiastic customer.
Counting articles instead of pricing them
Anyone who must pay for something whose volume they cannot predict has, broadly, two ways to limit the bill. They can agree a price per unit and accept whatever total that produces, or agree a total and accept that, once it is reached, no more units will be bought on those terms. A government limiting pollution faces the same choice: tax each tonne and let the market decide how many tonnes, or issue a fixed number of permits and let the market decide what they are worth. Martin Weitzman showed in 1974 that the choice matters only under uncertainty.[^13] With perfect knowledge, a price and a quantity could be set to produce the same outcome. Without it, a fixed price lets the quantity wander and a fixed quantity lets the effective price wander, and one should fix whichever of the two is more costly to get wrong.
Seen this way, the Dutch consortium’s choice is revealing. It did not secure a price per article and let the number of articles find its level, which is what the theory of Plan S required. It secured a number of articles inside a negotiated total and left the price of every article beyond that number to the publisher. In Weitzman’s terms, the consortium judged that what it could least afford to get wrong was the bill, and so it fixed the bill. That is rational for a buyer with a fixed budget, but it is not the choice of a buyer who believes the unit price to be under competitive discipline. Nobody caps the number of loaves they will buy because they trust the baker’s price.
The shape of the resulting schedule makes matters worse. Seen from the author’s side, open access costs nothing for the first 2,067 articles of the year and the publisher’s full list charge for the 2,068th. Economists call this a non-linear tariff, and it is familiar from mobile telephone contracts: a bundle of minutes included, a steep charge per minute thereafter. Such schedules are characteristically a device of sellers rather than buyers, because they let a seller collect a bulk payment for the volume it knows will be demanded while charging the highest price it can for every unit above it.[^14] The cap that looks like the consortium’s protection is, at the margin, the publisher’s most profitable price.
A careful reader will have an objection ready. A national consortium negotiating for every university in a country is a single buyer — a monopsonist, in the jargon — and a single buyer’s most powerful weapon is the credible refusal to buy more. On this view the cap is a strength rather than a weakness: it is the threat that secured a lower total, and the queue at the end of the year is the price of that discipline. I concede readily that an open-ended commitment to pay for however many articles Dutch researchers produce would be worse. But the objection proves less than it appears to. A refusal to buy is credible only if the buyer bears the consequences of not buying, and the consortium does not. When the allowance is gone, Dutch research keeps flowing into Springer Nature journals, now paid at list price out of grants or published closed; the publisher still receives the marginal articles, and the consortium has merely moved the bill from its own budget to its members’. A buyer with genuine bargaining power would have insisted instead on a price for the marginal article, since it is the marginal price that determines whether a seller gains from more volume. The Dutch arrangement leaves that price at the publisher’s discretion and at the top of the market. Whatever the consortium won on the total, it conceded at the margin, and the margin is where markets are made.
This is why I call the quota a revealed failure. Had the price per article been disciplined by competition, the consortium would have negotiated over the price and let the quantity follow. It negotiated over the quantity instead, which is what one does when the price cannot be moved.
The queue nobody designed
When an allowance runs out before demand does, something other than price must decide who gets it. In the Dutch case the rule is the simplest available — first come, first served by date of acceptance — and it looks neutral. It is not, for reasons set out in the economics of queues half a century ago. When a good is offered below the price at which people would willingly buy all of it, the surplus does not vanish; it is competed for, in whatever form the allocation rule rewards. Where the rule is waiting, people pay in time, and Yoram Barzel showed in 1974 that in the limit the value of the time spent queuing can absorb most of the value of the good itself.[^15] Donald Nichols, Eugene Smolensky and Nicolaus Tideman had earlier made the complementary point about incidence: rationing by waiting shifts the good towards those whose time is cheap, which is a deliberate policy when the good is meant for the poor and an accident when it is not.[^16]
Academic authors stand in no physical queue, but they are in one all the same, and each author’s place is set by the date on which an editor accepts her paper. That date depends on things that have nothing to do with the merits of open access for the article: the field’s typical review time, the speed of a particular referee, whether revisions were requested in July or in December. A cardiologist whose paper is accepted in March is covered; an economist whose paper has spent two years in review and is accepted in November is not. Nothing about the economist’s research differs; only her position in the year.
This is the point I find most troubling. The cost of the consortium’s quantity limit is not spread across the system. It falls entirely on whoever is late, and lateness is correlated with discipline, with the pace of a journal’s editorial process, and with whether the author holds a grant from which an unexpected charge can be paid. A senior researcher on a large European grant can absorb the fee; a doctoral student in the humanities cannot. Leiden’s advice to use the grant or the institute budget is sensible, and presupposes a grant.
Barzel’s further insight was that a queue invites effort spent on moving up it rather than on producing anything. One should expect, in time, some bunching of submissions early in the year and some steering of late acceptances towards journals whose allowance is still open. Whether either is yet visible in the data I do not know.[^17]
A shortage that renews itself
The most striking feature of the table is not that the allowance runs out but that it runs out every year, and earlier each time. A shortage that recurs on schedule is not a disequilibrium waiting to be corrected. It is an equilibrium of its own kind, and the economist who described such equilibria best was János Kornai, writing about the planned economies of Eastern Europe. In Economics of Shortage (1980) Kornai argued that chronic shortage there was not the result of planners’ mistakes, which might be fixed, but of incentives built into the system: institutions faced what he later called a soft budget constraint, so that the consequences of over-demanding were absorbed by someone else and demand, never disciplined by cost, outran supply as a matter of routine.[^18] He also catalogued how buyers in such systems adapt. They queue; they search for substitutes; and they accept what he called forced substitution, taking something other than what they wanted because what they wanted was not to be had.
Dutch universities are not Hungarian state enterprises, and Springer Nature is not a planning ministry, but the structure is uncomfortably close. Below the cap the individual author faces a zero price and has no reason to economise; above it she is rationed; and the consortium renews the allowance every 1 January as though the previous October had not happened. The adaptations are Kornai’s almost to the letter. Researchers are told to queue by submitting early, to search by checking which journals still have allowance before submitting, and, failing both, to accept forced substitution by publishing closed and depositing a copy six months later. The last is the most telling. A system designed to make research open on publication responds to its own shortage by making research closed on publication, and open later.
What the next number will admit
I set out to ask what kind of arrangement runs out every autumn. The answer is a rationing regime: a valuable good offered below its market price, allocated by a rule nobody would choose if asked to defend it, and generating a shortage that renews itself because no one in the system bears its full cost. None of this was intended. The consortium wanted budget certainty and got a queue. Plan S wanted price competition and got a quota. Researchers were promised that their work would be free to read on publication, and some of them, depending on the month, are now advised to publish it closed.
The case matters beyond the Netherlands precisely because the Netherlands is the system’s best student. If the country that has pursued national agreements most consistently ends each year in a queue, the difficulty is structural rather than Dutch. A market in which authors are tied to particular journals by their employers’ evaluation rules, and in which a buyer pays on behalf of authors who face no price, cannot be disciplined by a price — and leaving that price to the publisher handed it to the one party in the market that faced no competition. The remedy, if there is one, lies less in the negotiating room than in the promotion committee.
I should be clear about the limits of what I have shown. I have not examined the financial terms of the agreement, and the total the consortium pays may well be a good bargain; my argument concerns the structure of the arrangement, not its headline cost. The implied demand figures rest on an assumption about timing that is certainly false in detail. The consortium, which holds the data, is better placed than I am to test both.
The current agreement expires at the end of 2026. The number in the next one will be its most informative figure — not for what it allows, but for what it admits.
[^1]: Leiden University Libraries, “Publisher Springer Nature: maximum open access articles almost reached”, announcement dated 2 October 2026. The notice applies to researchers at Leiden University and the Leiden University Medical Center, and also describes the six-month repository route discussed in the text.
[^2]: Openaccess.nl, “Updated forecast for open access publication caps (October 2026)”, 5 October 2026, using publication data updated on 1 October 2026. The forecast dates are indicative and the consortium notes that they move with publication patterns.
[^3]: Openaccess.nl, deal page for the UKB–Springer Nature agreement 2024–2026. The agreement covers Springer’s hybrid OpenChoice journals and excludes society titles, BioMed Central and SpringerOpen. UKB is the Dutch consortium of university libraries and the National Library.
[^4]: Radboud University, “Maximum number of SpringerNature open access articles 2025 reached”, staff news item dated 27 October 2025, which reports the 2025 allowance exhausted and gives the range of article-processing charges.
[^5]: Article 25fa of the Dutch Copyright Act, introduced in 2015 and known as the Taverne amendment, entitles authors of short scientific works funded wholly or partly from public funds to make them available free of charge after a reasonable period, which Dutch universities interpret as six months.
[^6]: University of Groningen Library, news item on the Springer Nature agreement, first posted in September 2024 as approaching the annual quota and updated on 6 November 2024 to report it reached. Leiden’s announcement archive records an equivalent notice in September 2023 under the predecessor agreement, which I have not been able to verify in detail and have therefore left out of the table.
[^7]: Plan S was announced on 4 September 2018 by cOAlition S, a group of national research funders led initially by Science Europe and the European Commission. Its principles have been revised since, and the language on fees has softened; the expectation of transparency and of fees commensurate with services provided is the part that matters for the argument here.
[^8]: The Chartered Association of Business Schools’ Academic Journal Guide in Britain is the best-known example; several continental systems maintain national equivalents.
[^9]: ShanghaiRanking, Academic Ranking of World Universities: Methodology 2026. The indicator counts research articles in Nature and Science over the preceding five years, with full credit to the corresponding author’s institution and diminishing credit to the others. Institutions specialised in the humanities and social sciences are exempted and the weight redistributed, a dispensation that does not extend to a social science faculty inside a general university.
[^10]: Vincent Larivière, Stefanie Haustein and Philippe Mongeon, “The Oligopoly of Academic Publishers in the Digital Era”, PLoS ONE 10(6) (2015), e0127502. Their series ends in 2013; consolidation since, including the 2015 merger that created Springer Nature, has not reversed the trend.
[^11]: Nina Schönfelder, “Article Processing Charges: Mirroring the Citation Impact or Legacy of the Subscription-Based Model?”, Quantitative Science Studies (2020), doi:10.1162/qss_a_00015. The data are payments by UK institutions recorded in the OpenAPC database, and citation impact is measured by the journal’s SNIP score.
[^12]: VSNU, NFU, KNAW, NWO and ZonMw, Room for Everyone’s Talent: Towards a New Balance in the Recognition and Rewards of Academics (2019); and Chris Woolston, “University drops impact factor”, Nature 595 (15 July 2021), p. 462. The Nature report itself noted the risk that researchers assessed under Utrecht’s new criteria might be at a disadvantage when applying to institutions that still rely on the old ones, which is the point made in the text.
[^13]: Martin L. Weitzman, “Prices vs. Quantities”, Review of Economic Studies 41(4) (1974), pp. 477–491. The result concerns the relative slopes of the marginal benefit and marginal cost schedules; I have stated it in its intuitive form.
[^14]: The literature on non-linear pricing is large; Robert Wilson, Nonlinear Pricing (Oxford University Press, 1993), remains the standard treatment. A cap followed by list price is a crude two-part schedule, and it should be added that the consortium may have had no alternative, since a publisher will not easily concede a negotiated price on articles it has not yet received.
[^15]: Yoram Barzel, “A Theory of Rationing by Waiting”, Journal of Law and Economics 17(1) (1974), pp. 73–95. Full dissipation is a limiting case that requires everyone in the queue to value the good identically; with differing valuations some surplus survives.
[^16]: Donald Nichols, Eugene Smolensky and T. Nicolaus Tideman, “Discrimination by Waiting Time in Merit Goods”, American Economic Review 61(3) (1971), pp. 312–323.
[^17]: The test would be straightforward for the consortium, which holds the acceptance dates: compare the monthly distribution of covered acceptances in the years since the cap began to bind with the distribution before, and compare the share of Dutch articles going to journals in agreements that run out early with the share going to those that do not.
[^18]: János Kornai, Economics of Shortage (North-Holland, 1980); and “The Soft Budget Constraint”, Kyklos 39(1) (1986), pp. 3–30. Kornai’s analysis concerned firms whose losses were covered by the state; the application here is by analogy, and the analogy is in the incentive structure rather than in any claim about ownership.