Plan S: An Economist’s Perspective

Many scholarly journals charge high prices to research libraries and generate high profits. Open access regulation, in its various forms, can mitigate this problem. This essay examines a particular policy, “Plan S”, which aims broadly to require regulated authors to publish their research in open access journals, which among other drawbacks of the policy greatly limits their publishing options.


Introduction
In this essay I discuss the scholarly journal market, including forms of market failure the market might su¤er in the absence of regulation. The market failure most prominent in current debates involves the excessive prices that some publishers charge readers for access to their articles, especially within science disciplines. Various kinds of open access regulation have been proposed to mitigate this problem. One recent proposal is known enigmatically as "Plan S", the many and changing details of which will be explained later, and in this essay I discuss the pros and cons of this plan alongside other potential regulatory policies. I will use an economist's perspective, both in terms of using economic tools to discuss this particular market and in terms of a focus on how Plan S would be likely to impact the economics discipline.
Department of Economics and All Souls College, University of Oxford. This essay is based on an invited presentation with the same title at the 2019 Royal Economic Society's annual conference in Warwick, UK. I am grateful to two referees, Manuel Bagues, Imran Rasul and John Vickers for comments, and to the European Research Council for research funding under Advanced Grant 833849. 1 A useful perspective on the scholarly journal market is that it is a "two-sided" market, providing value to readers (who gain access to articles they wish to read) and to authors (who gain exposure for their work to readers). 1 Journals incur costs when providing their services, and these costs are usually met from charges levied on one or both sides of the market. For this essay, the two principal purposes of a journal are (i) to disseminate an author's work to potential readers and (ii) to certify various attributes of that work.
In more detail, purpose (i) involves the journal making available its content to those readers-and only those readers-permitted to consume its content. With the advent of online journal access, the costs associated with dissemination to subscribers have fallen drastically in recent years. Authors care (some more than others) about the size of their journal's readership, but plausibly care mostly that their scienti…c peers have access to their work (either in its …nal journal form or as a working paper if available) rather than the wider public. Nevertheless, there can be much value in the wider population gaining access to up-to-date scholarly material, at the time of writing perhaps most notably in the area of health.
Purpose (ii) is to certify that the article's content is novel, correct, concerns certain specialist topics, and it may also signal how important and/or interesting the article is.
Certi…cation is valuable for both sides of the market, although for di¤erent reasons. For readers, (ii) helps them to discover the content they will likely want to consult, from out of the vast pool of available content. 2 Given that reading a paper involves a sunk cost, a reader bene…ts from ex ante information about a paper's quality and topic before deciding whether or not to read it. For an author, on the other hand, (ii) provides a useful signal of the quality of her work, including to people who may not actually want or be able to read it. The quality of an author's publications collected in her CV, as judged by various committees, determine the hiring, promotion, and research funding decisions which are so crucial for her career. This certi…cation function is particularly important for more junior researchers, who have not yet built up a widely perceived reputation. Many junior researchers would be willing to pay much money out of their own pocket to place a paper in a prestigious journal. 3 The costs associated with certi…cation (e.g., paying the journal's editors and, sometimes, its referees) have not fallen nearly as much over recent years as those of dissemination. Moreover, as the internet makes various kinds of self-dissemination by authors so easy and discoverable, purpose (ii) is nowadays arguably the more important of these two roles.
As mentioned, a journal usually covers its costs by charging one or both sides of the market. 4 An open access journal makes all of its content free to readers at the time of publication. Such journals usually cover their costs by charging authors or their institutions to publish their work. Thus, anyone can read the 2020 article "Open access publishers: the new players" in PLOS ONE for free, but its authors have paid around $1749 (at current prices) to publish there. Within my own economics discipline, only a few well-known jour- or worse, the ability to publish articles in top 5 journals is an important route to career success for academic economists. 5 Some journals do not supply credible certi…cation at all, and publish almost anything in return for a fee from the author. These journals often position themselves as "open access", although they have minimal readership and could not generate much subscription revenue if they tried. Part of the author demand for these "predatory" journals is to fool those evaluation committees that are ignorant about which journals are discriminating about the articles they publish and which are not. 6 At the other extreme, a subscription-only (or "closed") journal sells all its content to readers (usually libraries in universities and other research institutions). Subscription journals often allow authors to publish their work without charge, so that authors have 3   Press, by contrast, has an embargo period of 24 months for most of its economics journals before an accepted version of a published article could be uploaded to RePEc.) In practice, though, it is hard to prevent authors from uploading the accepted version of their paper to a repository: an economics paper, for instance, will typically have been through several rounds of revision before being accepted, and it would take much e¤ort from the publisher to determine which precise versions have been posted online.
An important issue, and one central to the open access debate, is whether or not an unregulated journal market will implement an appropriate balance of charges across the two sides of the market. In particular, a frequent claim is that many publishers, left free to do so, charge libraries too much to subscribe to their journals. In the next section I explore why it is indeed likely that journals will levy high subscription charges rather than high author charges. But taking this pricing pattern as given for now, it is clear that high subscription charges have an adverse impact on both distributional and e¢ciency aims.
Even if libraries are just willing to pay them, high subscription charges have a welfare cost if a dollar of library budget is worth more in welfare terms than a dollar of publisher pro…t, as is plausibly the case when research libraries are ultimately …nanced from costly public funds. It is ine¢cient to exclude interested readers and libraries when it costs next to nothing to serve them. It may also be costly in political terms to exclude readers, if those who pay for public research through their taxes are denied timely access to its …nal product.
Related to this is the observation that scienti…c publishing can be highly pro…table.
As the UK's House of Commons (2004, page 5) put it, "There is mounting concern that the …nancial bene…ts from the Government's substantial investment in research are being diverted to an excessive degree into the pockets of publishers' shareholders." In the year 2020, for instance, Elsevier made pro…t of £1021m on revenues of £2692m, a pro…t margin of about 38%. 11 It is not just the most commercially minded publishers which bene…t from  open access.) In the economics discipline, the culture is such that almost all published articles are freely available online in working paper form, although in other disciplines it is rarer, or even forbidden, to post working papers online prior to publication. Third, many journals, usually non-pro…t journals, charge "low" but not zero subscription fees. The top 5 economics journals arguably all fall into this category. 12 In the economics discipline, this combination of a working paper culture along with relatively inexpensive journals (at least at the top end) has meant that discussions and controversies about open access have been muted relative to those within other subject areas.
Finally, there are various kinds of illegal open access, where published work is disseminated in violation of the publisher's copyright requirements. As mentioned, authors might post the accepted version of the published article online before a required embargo period, or even post the published version online. It is hard for published to monitor and enforce all such violations. There are also "pirate" websites, most notably Sci-Hub, which allow readers to download published articles that the website has somehow managed to obtain.
Open access regulation requires designated authors-for instance, those whose research was funded from a speci…c source-to make their research open access in a speci…ed manner, either in a "gold" form or in some "near open access" form. If an article is required to have gold open access, then any publisher who wishes to publish this article is constrained to set a subscription charge of zero for this article, which can be viewed as an extreme kind of price cap. Note that there is always "open access" for the certi…cation role of journals: one need not subscribe to the American Economic Review to discover that a particular article and author has been published there.

Some Economics of Open Access Regulation
To understand the impact of open access regulation, it is conceptually useful to distinguish between the dissemination and certi…cation functions of journals. 13 For the initial discussion I focus on the dissemination purpose of a journal, and subsequently add the extra complexities which arise from their certi…cation purpose.
Without regulatory intervention, the typical pattern of pricing in the commercial journal market is that subscribers pay high fees to access a journal's content, while authors pay little or nothing to publish, i.e., to access the journal's subscribers. 14 What is the asymmetry between the two sides of the market which induces this skewed pricing, whereby authors are o¤ered a "bargain" funded from "ripo¤" charges levied on subscribers? The crucial issue is that the peer-review process ensures an article is only published in a single journal and is di¤erentiated from other published articles, and so readers need to subscribe to several journals if they wish to see a wide range of content. (In the jargon of two-sided markets, the asymmetric is that authors "single-home" while readers "multi-home".) Each published article thus constitutes a mini-monopoly, and a journal enjoys market power in providing access for readers to its articles. For this reason, a journal in the absence of regulation is able to set high subscription charges which need bear little relation to the cost of running the journal. Even a tiny publisher with a single journal could set a high subscription charge for access to its content. 15 Because each article generates its own quantum of monopoly pro…t, a journal has strong incentives to attract articles to publish and so will wish to o¤er a generous deal to a suitably quali…ed author. An important and curious feature of the academic journal market, dating from its inception, is that publishers do not usually pay authors for their work. (Of course, an author may obtain …nancial reward for publications in prestigious journals, via job promotion and so on, but those rewards are …nanced from elsewhere. For some reason, academic books are di¤erent from journals and book authors are usually paid advances and royalties.) As a result, the most generous deal a publisher can o¤er is that an author can publish for free. Since authors are not paid, the often large revenues from selling subscriptions to institutions are not easily dissipated and publishers can enjoy supernormal pro…ts. From this perspective, high subscription fees and excess publisher pro…ts are due mostly to the monopoly nature of each individual article, alongside a constraint that monopoly subscription pro…ts are not fully passed back to authors. 16 This pattern of skewed pricing, with its "bargains and ripo¤s", would be reversed in an alternative world in which readers each subscribed to a single journal, and authors had to place their work in multiple journals in order to reach a large readership. (This is akin to the old market for print newspapers, in which most people read a single newspaper and advertisers had to place their advert in multiple outlets to reach a desired number of eyeballs.) In such a world, it would be readers who would be courted by journals, and authors would pay high fees for access to a journal's captive subscribers.
The journal market is an instance of what I have elsewhere termed a "competitive bottleneck". 17 In such a market, there are competing platforms which intermediate the interaction between two sides. For perhaps exogenous or institutional reasons, one side joins one platform exclusively (the single-homing side). The chosen platform can exploit the monopoly position it holds on this single-homing side by setting high prices to the other, multi-homing, side. Sometimes monopoly prices on the latter side are fully passed onto the single-homing side, in which case platforms are less likely to sustain excess pro…ts.
However, sometimes there are constraints which prevent full pass-through, for instance that the single-homing side cannot be paid, in which case the platforms may retain a portion of the monopoly pro…t, even when there is sti¤ competition between similar platforms. Other markets with competitive bottleneck features similar to scholarly publishing include credit cards (which consumers can often use for free if they pay the balance in full each month, while merchants pay high fees on each transaction), shopping malls (which consumers can enter for free, while retailers pay high rents for access to these consumers), online search engines (where people can search for free, but advertisers pay high prices to appear prominently in their search results), and call termination on telephone networks (where before the advent of regulation subscribers had a subsidized phone and their callers had to pay high charges to call them).
This competitive bottleneck logic can be illustrated in the following simple example.
First, imagine (as John Lennon might say) a world without journals. Author A obtains a bene…t of $5 when reader R sees her article, while R gains bene…t $15 from reading A's article. It costs A some small amount to deliver the article to R (say, the e¤ort of posting the article online, which is much lower than $5), and so she is willing to do this. The joint surplus from this exchange, which involves no monetary exchange, is therefore about $20. Now introduce a group of homogenous journals, which for simplicity incur negligible costs for disseminating the article to R. Since by assumption only one journal is able to publish A's article, the journal which obtains the article will be able to sell it to R for $15. If a payment to A is feasible, journals compete to o¤er the highest such payment, and the monopoly pro…t of $15 from R is passed back to A. Thus, in this stylized model, the introduction of the journal market induces no signi…cant e¢ciency gains, but operates to transfer surplus from readers to authors. However, if as is more realistic there is a 17 See Armstrong (2002, section 3.1, and2006, section 5). constraint that A cannot be paid, journals will o¤er A free publication, A will choose a random journal, which will then sell the article to R for $15. In this case, the combined surplus of A and R falls from about $20 to $5 (all of which is enjoyed by A), the di¤erence being siphoned o¤ by journals. 18 Provided there is su¢cient competition between publishers in a given subject area and quality tier, a "gold" open access regime, in which a regulated author must publish her article in a journal which make the article freely available at the time of publication, entirely overcomes this problem of monopoly pricing by journals. 19 Journals would then usually have to cover their running costs by charging regulated authors a fee to publish their paper. Like more familiar "one-sided" markets, journals would then compete for custom from authors in terms of publication fee, turnaround time, value-added (or not) from the refereeing process, and so on, and there is a greater chance that only normal pro…ts would be observed. As Brown et al. (2003, page 2) put it: "Open access would eliminate monopolies over essential published results, diminishing pro…t margins and creating a more e¢cient market for scienti…c publishing".
A gold regime has other attractions relative to alternative policies involving "near" open access. It is surely of some bene…t to the reader to see the journal article itself (rather than the author's own pre-print as in a "green" open access regime): the format may be somewhat more attractive, pagination for detailed citation is stable, she automatically knows the name of the journal which published the paper, and she knows she has the …nal, peer-reviewed version. A subscription price which is precisely zero (rather than merely 18 This highly stylized example has historical parallels in the story of how Robert Maxwell exploited this "competitive bottleneck" aspect of publishing when he started numerous new journals for his Pergamon Press (later sold and incorporated into Elsevier). He believed that "we don't compete on sales, we compete on authors". He o¤ered to publish the journals of scienti…c societies in return for a small fee to them. The editor of the Journal of Neurochemistry said Maxwell took over the journal after he was wooed with a lavish dinner and a cheque for a few thousand pounds. See Buranyi (2017) for these and many more details. Buranyi writes: "If a serious new journal appeared, scientists would simply request that their university library subscribe to that one as well. If Maxwell was creating three times as many journals as his competition, he would make three times more money. [...] And since there was no way to swap one journal for another, cheaper one, the result was, Maxwell continued, 'a perpetual …nancing machine'. Librarians were locked into a series of thousands of tiny monopolies. There were now more than a million scienti…c articles published a year, and they had to buy all of them at whatever price the publishers wanted." cheap) will reduce some journal costs, such as the selling costs associated with negotiating contracts with libraries. Relatedly, it is easier for regulators to ensure that authors are complying with gold open access obligations, rather than having to decide what counts as a "cheap enough" journal subscription or having to check that the author has indeed posted her accepted version in an appropriate online location.
Nevertheless, the gold policy essentially reverses the extreme skewness of pricing, moving from a situation where authors have free access to subscribers to one where readers have free access to articles. Thus the "paywall" is shifted from readers to authors. Is this opposite extreme pattern of bargains and ripo¤s likely to be the most e¢cient way for journals to cover their costs? First, there are sound public …nance reasons why readers should contribute something to the cost of publishing. Taking a parochial perspective, many readers of journal articles written by authors in a relatively small country will be overseas, and it is not obvious that national taxes should be used to fund free access for these readers. 20 Many users of scienti…c research are in the industrial and corporate sector, and it is unclear why such users should free ride on a subsidized author-pays regime. 21 Having to pay to publish will deter some authors at the margin from publishing at all.
Of course, not publishing an article at all is even more harmful to potential readers than having to pay a high price for access. It may be that regulated authors have publications fees built into their grants, in which case this is not an issue. However, if open access regulation succeeds in making publication fees widespread in the market, then "unregulated" authors will be impacted too. Even if many authors have access to funds which can be used to cover a publication fee, there will often be an opportunity cost when paying to publish a paper. Many scholars in the humanities and/or in poorer countries do not have research funds which cover publication fees. Indeed, a major drawback to a widespread move to an author-pays journal regime may be that scholars in poorer countries will not be able to adequately publicize their work to their peers in richer countries. 20 One option here might be to o¤er open access only to those readers in the geographical area which funded the research, as suggested in 2017 by Elsevier in www.elsevier.com/connect/working-towards-atransition-to-open-access. 21 House of Commons (2004, paragraph 175) reports that Elsevier obtains 20% of its journal revenue from this sector, and quotes the Biochemical Society as saying "in the open-access world it would appear that the only real winners are going to be corporate pharmaceutical companies who would no longer have to pay to access information." A "near" open access policy might require authors to publish in journals with a short embargo period, in journals with low but not zero subscription charges, or require authors to post the accepted version of their published work online (and to publish in a journal which permits this). Such policies mitigate the problems of high subscription charges and excluded readers. When a somewhat inferior substitute is freely available (say, when content is released after a short embargo period or if authors post accepted versions online), then libraries have a valuable outside option, and publishers are forced to charge less if they wish to continue selling their subscriptions. Moreover, the wider public has free access to the inferior variant of the published article, while before they may have been excluded altogether. Nevertheless, if the inferior variant is not too close a substitute, publishers may still be able to extract su¢cient revenue from libraries willing to pay for the premium published version to cover their costs, albeit with less to spare. Therefore, this "freemium" policy can be consistent with authors continuing to publish without charge.
Note that even if a publisher permits an author to immediately post the accepted version of her published paper online, the evidence is that-beyond a few subjects, including economics, computer science and physics-many authors do not go to the trouble to do this, which is not surprising if their article is anyway being distributed to their desired audience by a journal. Thus the "green" route to open access requires a degree of compulsion on regulated authors to disseminate the accepted version of their articles. A bene…cial byproduct of "green" open access regulations may be to stimulate a working paper culture within a wider set of subject areas than is the case at present. Such a culture often enables speedier access to research compared to a system where readers must wait for the published article to arrive, and may also enable comments from readers (rather than just from editors and referees) to be fed into the article's …nal draft.
This discussion so far has focussed on the dissemination role of journals. Journals also add value to the raw content, for instance in terms of certifying quality, attractive formatting, providing feedback from referees to authors, and generally in "polishing" papers for publication. Some of this added value bene…ts readers. If a journal is not permitted to charge readers (or more generally faces an in ‡exible price cap on its subscription charges), as in the gold regime, it cannot appropriate the extra bene…t it provides readers in the form of higher prices, and so will have less incentive to spend resources on such activities.

13
Thus, we expect that an open access journal will provide expert feedback from referees and editors and go through multiple rounds of revision only to the extent this is valued by its authors (including the indirect impact on boosting readership insofar as this is valued by authors), and the direct bene…t of such activity on readers is ignored in the journal's calculus.
If we focus more speci…cally on the certi…cation service, there is a downside to authors bearing the costs of processing papers, which is that some authors may be unwilling or unable to pay for certi…cation. More selective journals are likely to charge higher author fees than less selective journals in a gold regime, since they follow a more rigorous and costly peer-review process. (By contrast, in the traditional subscription model, the extra costs of peer-review for a selective journal are covered by readers.) As such, at least some authors with good papers may be less willing or able to publish in selective journals. The result is that the quality signal in a journal's name becomes less precise, which harms readers and (good) authors.
This issue can be illustrated as follows. 22 An author has a paper which might be good or bad. Readers in aggregate are willing to pay $10,000 to read an article known to be good, and willing to pay nothing for an article known to be bad. An author knows the quality of her paper, while readers cannot directly observe quality without investing in the costly e¤ort of reading. It costs a journal $2,000 to determine via peer review whether a paper is good or bad, but all other journal costs are zero. Journals come in two forms: a "discriminating" journal peer reviews all submitted papers (at the cost of $2,000 per paper) and publishes only good papers, while a "non-discriminating" journal will publish anything and incurs no costs. Because of its reputation, say, readers know whether a journal is discriminating or not. An author enjoys some intrinsic bene…t from being seen to publish a good paper (if she has one), and if she has a bad paper she will not submit to a discriminating journal since she knows her paper will be rejected.
In an unregulated subscription-funded market, the outcome is that discriminating journals compete for good papers, a journal which attracts a good paper charges readers $10,000 to read the paper certi…ed to be good, and since the certi…cation cost of $2,000 is covered by the subscription revenue, it allows an author to publish for free when she has a good paper. 23 Readers infer that a paper which appears in a non-discriminating journal is bad, and won't read it (even if it is free). By contrast, if the author must publish her article with open access in a journal, a discriminating journal will charge her $2,000 to publish if she has a good paper. If her intrinsic bene…t from being seen to publish a good paper is greater than her opportunity cost of funding the $2,000 publication fee, the outcome is as before and good papers are all published in discriminating journals. However, if this bene…t is below her opportunity cost (for instance, if she has limited research or personal funds), the author prefers to submit a good paper to a non-discriminating journal which charges her nothing. 24 In this case, a reader must consult a non-discriminating journal if she wants to see the article. If an uncerti…ed article is relatively likely to be good (and this depends on the underlying fraction of good papers and the distribution of author bene…ts of being seen to have a good paper), it is worthwhile for readers to take the gamble of reading a paper in a non-discriminating journal, although they su¤er the disutility of having to read some bad papers alongside the good ones. If an uncerti…ed article is unlikely to be good, though, readers may have no incentive to read undiscriminating journals at all, and so these good papers will go unnoticed. Likewise, authors of good articles, such as those in poorer countries, may not gain the reputation they deserve if they are …nancially constrained to publish their work in lower-ranked outlets.
In sum, the demand for journal certi…cation by authors may fall when authors must pay to publish. In a subscription-funded market, authors can usually submit and publish for free, and so an author has an incentive to place her article in the most discriminating journal willing to accept it. The result is that potential readers, as well as members of tenure and promotions committees, obtain a relatively precise signal of article quality from the journal in which a paper appears. In an author-pays regime, however, it becomes more expensive to publish a paper in a discriminating journal since the peer-review costs are 23 To implement this scheme, the journal could charge a submission fee of $2,000 which is refunded if the paper is judged to be good. 24 For instance, within the PLOS group of open access journals, the highly selective PLOS Biology currently charges authors $4000 to publish, while the less selective PLOS ONE charges $1749. Some good biology papers will be published in the latter outlet when authors are …nancially constrained (or don't care enough about the incremental impact on their reputation).
higher. As a result, some authors with good papers may be constrained or choose to submit to a less prestigious journal. For instance, authors with an already high reputation may reach their audience without the need for expensive journal certi…cation. 25 Readers and committees then have a less precise signal of quality than before, and good papers may be lost amongst the mediocre.
As before, a near open access regime can overcome these dangers. If the regime is not "too near" to open access, a discriminating journal will be able to cover its costs (including its costs of peer review) out of its subscription income, and authors of good papers do not need to pay to have their article certi…ed. As such, the policy may be compatible with authors continuing to publish in the most selective journal that will accept their paper, and the certi…cation role of journals is maintained. policy such that for a journal publication to be submitted to the current review, the article must either have appeared without embargo from the publisher itself (the gold route), or to use the green route the author's accepted version must be publicly available from a suitable repository no longer than 12 months after publication (or 24 months for humanities and social sciences). In essence, this makes most of the journal articles originating in the UK open access in some form, albeit with a very lengthy delay in many subjects. Although it is hard to get data at this point, I feel con…dent in predicting that a large majority of economics articles submitted to the current REF will follow the green rather than the gold route to comply with open access requirements.
Plan S is a di¤erent kind of regulation, which focusses less on the article than on the journal in which it is published. An explicit aim of the policy, at least initially, was to terminate subscription payments for journal content, and to move to a world in which only open access journals exist. The origins of the plan seem murky, as do several aspects of its implementation. The architect of Plan S, Robert-Jan Smits, the European Union's special envoy for open access, gave a conference presentation in July 2018 during which he announced a new radical plan to accelerate open access, the details of which would be clari…ed later in the year. 26 In this presentation he said that the plan's "S" stands for "speed, solution, science", and that he aimed to use the …nancial in ‡uence of research funders to put pressure on journals to change their funding model from a reader-pays to an author-pays regime.
These promised further details were published as a manifesto in September 2018 in Schiltz (2018), and Plan S was launched with the support of eleven European national funding bodies and the European Research Council (ERC) who would all follow its policies.
Although Plan S seemed to many in the university world to come "out of the blue", Schiltz The manifesto proposes one key principle and ten subsidiary principles. The key principle states that "After 1 January 2020 scienti…c publications on the results from research 26 The presentation can be viewed on YouTube by searching for "ESOF 2018". funded by public grants provided by national and European research councils and funding bodies, must be published in compliant Open Access Journals or on compliant Open Access Platforms". 27 Thus, within sixteen months of this announcement, the plan envisaged that all European grant-funded research would have to appear in open access journals.
Among the ten subsidiary principles, one suggests that for disciplines in which high quality open access journals do not yet exist "the Funders will in a coordinated way provide incentives to establish these and support them when appropriate", another suggests that "when Open Access publication fees are applied, their funding is standardized and capped (across Europe)", another states that funders will monitor and sanction non-compliance by authors, while another explicitly states that "the 'hybrid' model of publishing is not compliant". (It was later clari…ed that "mirror" journals would be considered to be hybrid journals, and so would not be compliant either.) Another principle suggests that open access will also be required for scholarly books and monographs, although the timeline needed to achieve this may be longer. It was not explained why it is harder to achieve open access for books, although one factor may be that authors would be unwilling to give up their royalties if their books had to be given away.
Clearly, unless very many subscription and hybrid journals " ‡ip" to become open access, these policies would greatly restrict where researchers can publish their work. The manifesto (Schiltz, 2018, p.3) acknowledges that "researchers need to be given a maximum of freedom to choose the proper venue for publishing their results", but immediately goes on to warn that "researchers must realise that they are doing a gross disservice to the institution of science if they continue to report their outcomes in publications that will be locked behind paywalls". 28 The manifesto also gives a strong steer away from the certi…cation role of journals, saying that researchers who commit this "gross disservice" might prefer to publish in non-compliant journals due to a "misdirected reward system which puts emphasis on the wrong indicators (e.g., journal impact factors). We therefore commit to fundamentally revise the incentive and reward system of science." 27 The precise meaning of "open access platforms" is unclear, although www.coalition-s.org (which is the umbrella group of funders who support Plan S) says that "open access platforms are publishing platforms for the original publication of research outputs (such as Wellcome Open Research or Gates Open Research)." It certainly does not include subject repositories such as RePEc.
28 In Enserink (2018), Schiltz is quoted as saying "The greater good of a well-functioning science system is more important than the right of individual researchers to decide where to submit their papers." There are profound problems with Plan S as it was originally proposed in this manifesto, both in terms of its aims and in terms of its feasibility. As argued in the previous section, it is by no means clear that a monoculture of open access journals is the most e¢cient way for authors to disseminate and certify their work. For example, an environment with widely available working papers and inexpensive high-quality journals, as is arguably already the case in economics and other disciplines, may work better than one where all journals are funded exclusively by author fees. Author fees might discourage some authors at the margin from publishing their work, which is even worse than science being "locked behind paywalls". Permitting green open access, whereby authors can post the accepted version of their paper online, might stimulate a working paper culture in a broader range of subject disciplines, which carries its own advantages. 29 In addition, the certi…cation role of journals may well be diminished when subscription fees are prohibited. Indeed, as mentioned it is one of the subsidiary aims of Plan S to lessen this certi…cation role, with its disdain for impact factors and the like, despite the e¢ciencies caused by certi…cation (especially for more junior scholars of high calibre).
What the stated aim "to fundamentally revise the incentive and reward system of science" might actually mean is clari…ed to some extent in the current set of Plan S principles, where principle 10 states that "The Funders commit that when assessing research outputs during funding decisions they will value the intrinsic merit of the work and not consider the publication channel, its impact factor (or other journal metrics), or the publisher". 30 How in practice a committee member asked to evaluate a grant application outside her own areas of expertise should grasp the "intrinsic merits" of the applicant's work is not explained. It is also hard to understand how this aim to diminish journal certi…cation sits with another of the Plan S principles, which is that "the Funders will ensure jointly the establishment of robust criteria and requirements for the services that compliant high quality Open Access journals [...] must provide"; what does this notion of "high quality" mean? 29 Enserink (2018) quotes the open access scholar Peter Suber as saying that Plan S's downplaying of green open access was an "elementary mistake", and that by allowing researchers to make their work freely available while publishing in a "conventional, venerable" journal, green OA helps young scientists who need the cachet of publishing in top journals. 30 See www.coalition-s.org/plan_s_principles for a description of the current ten principles.

19
The aim to standardize and cap publication fees is hard to interpret. Would they be standardized across disciplines, or only within the same subject area? (The latter seems the only reasonable policy, given the extent to which peer review costs di¤er across …elds.) Would a more discriminating and selective journal be permitted to charge higher publication fees than a less discriminating journal, given that its peer review costs will be higher? If not, and if the cap is set below that of a highly discriminating journal, this may force such journals to become less discriminating and hence weaken further the precision of journal certi…cation.
The share of world research funded by the initial supporters of Plan S is tiny. 31 This would not give major international journals a su¢cient incentive to " ‡ip" to an open access model, with the result that researchers with grants from Plan S funders would have a very restricted set of journals to consider for their work. In economics, for instance, none of the top 5 journals had any prospect of becoming open access, and so economists with grants from these funders would be unable to publish in the discipline's prime outlets. This would disadvantage the careers of good researchers who had grants from these sources, and also deter many of the best researchers from applying for grants from these sources.
(Funders compete for good applicants, just as journals compete for good papers, and Plan S promises an excellent opportunity for non-Plan S funders to attract good researchers by o¤ering them the freedom to publish where they wish.) Relatedly, it would be hard for a Plan S funded researcher to collaborate with other researchers who may not wish to be so constrained in where they could publish. The economics discipline is arguably already "dominated" by American scholars, and the inability of many European economists to publish in the best journals or to collaborate with American-based researchers would only exacerbate this problem.
While the manifesto suggests that funders will somehow establish new high quality open access journals in disciplines where these do not yet exist, it hard to see how that could happen in reasonable time, if at all. Journal reputations are notoriously "sticky", and for reasons of coordination it is a di¢cult task to launch a new quality journal, and journal "commits" to a path to become open access by means of a so-called transformative arrangement. 36 For instance, many Elsevier journals have become "transformative journals", as has Nature (albeit with a publication fee of 9,500 euros to gain open access for an article). What the sanctions would be if a journal did not satisfy all of the transformative requirements over time are unclear. So far as I understand, none of the top 5 economics journals is currently a transformative journal. Finally, the "green" route to open access is permitted, but with very tight restrictions: the accepted version of the article must be deposited in a repository without an embargo period and be published with an liberal re-use license (such as "CC-BY"). Few publishers would agree to such terms, at least without payment of an author fee (which would not be covered by the funder). 37 In sum, although the Plan S requirements have become weaker than those in the original manifesto, they nevertheless impose signi…cant restrictions on where a funded author can place her work, especially for those authors who cannot fund on their own the likely expensive "green" route to open access which is permitted under the policy.
In regulations for all disciplines, while in many subject areas beyond the sciences the journal market arguably works satisfactorily without the need for further intervention, with all its attendant costs. At the current time the bold policy appears to have lost its initial momentum, and to have become a European-centric policy followed by a number of the region's national research funders, which will probably harm the career prospects of many researchers in those regions, and which has not transformed the publishing market into one populated mostly with open access journals.