Economic thoughts about “gold” open access


There is increasing support in the scholarly communications community for “flipping” the standard journal publishing model from subscription-based to “gold” open access, which is to say a system supported by pre-publication fees (article processing charges or APCs), rather than post-publication fees (subscriptions), and in which there is free (unpaid) access to the published articles.[1]

The increasing support is especially strong in Europe, as illustrated by the recent signing of the “Expression of Interest in the Large-scale Implementation of Open Access to Scholarly Journals” which came out of the Berlin 12 Open Access Conference, sponsored by the Max Planck Society. As of this writing, 44 institutions have signed on to the EoI.

A number of my colleagues in US academic research libraries have expressed skepticism or downright opposition. And recently the Advocacy and Policy Committee of the Association of Research Libraries has circulated a one-sided briefing paper in advance of a discussion during the spring ARL business meeting on 27 January. (I say “one-sided” because support of gold OA was presented, tepidly, in just nine words — “the overall aim of this initiative is highly laudable” — followed by nearly a page of single spaced “concerns and criticisms”.)

I think critics of the Expression of Interest are jumping to unfounded and incorrect conclusions, and that organizing opposition to the EoI may greatly hinder a very promising effort to radically transform the scholarly publishing ecosystem much sooner and more effectively than other efforts (notably the “green” OA movement). I’ll offer a few observations here, written from my perspective as an economist. (See Caveat below.[2]) For this article I’m going to assume that open access is a good thing overall.

I first present a summary, followed by a considerably longer statement of my arguments.


Many decision makers and influencers — particularly in the research library community in the US — are expressing opposition to gold OA for reasons that I think are unsupported by either facts or simple economic principles.

  1. Will gold OA further strengthen the monopoly scholarly publishing firms? No. In fact, it is likely the most realistic path towards reducing or eliminating their market power.
  2. Will there be a change in the current market model? Yes. By engaging authors in the economic decision about where to publish, we will create article-level (submission) price competition between journals and publishers.
  3. Will research-production-intensive institutions be made worse off? No. The costs of scholarly communications (primarily subscriptions) are generally paid (mostly indirectly) by research funders today. Those total payments for scholarly communications will be less in a predominantly gold OA world. The research funders can and will redirect funds to where the costs are paid, raising reimbursements to those institutions whose costs go up (because they are paying for a disproportionate share of APCs) and reducing them to those whose costs go down (because they are saving more through the elimination of subscription payments than they are paying in APCs).
  4. Will gold OA hurt under-resourced institutions (such as those in the “global south”)? No. First, because they generally publish less per employee than better-resourced institutions, at first blush we should if anything expect them to benefit: they’ll save more in eliminated subscription costs than they will pay in APCs. At worst, since research funders routinely adjust research direct and indirect payments as costs change, under-resourced institutions might not get to keep the savings, but they won’t be made worse off. (And any current subsidies to reduce their subscription costs can simply be re-directed to be APC subsidies.)
  5. Will flipping to gold OA take too long and cost too much? Given our experience to date with green OA, and the fundamental problems with getting to effective, universal green OA sufficient to bring subscription prices to competitive levels, flipping to gold OA probably can happen much sooner. And though the transition may be somewhat costly, those costs will be moderated increasingly by negotiated offsets. And some transitional investment is justified by the great social benefits that will follow from open access and competitive rather than monopolistic prices for scholarly communications.

Will flipping to open access strengthen the monopolized scholarly publishing industry?

To address this let me say a few words about the market structure of the current industry. “Monopoly” generally means “a single seller”, yet there are many scholarly publishers (some for profit, some not-for profit; some with many titles, many with just one or a few titles). Nonetheless, it is reasonable, for short hand, to say that scholarly publishing is monopolized.

First, research output (e.g., articles) is what we economists call a differentiated good: one article is not a perfect — or often even good — substitute for another. If articles were perfect substitutes, then if one among several publishers, if one tried to charge an above-competitive article price (to first order, the average cost of reviewing, editing, reproducing and distributing it), readers would switch to a different provider who competed by offering a lower price, with the competition maintaining low, competitive prices — as we see for corn flakes, for example. But since articles are (often quite) imperfect substitutes, when one publisher charges above-competitive prices, readers don’t simply switch to a different article from a competing publisher. There will be some limit to how much readers (or their agents, such as libraries) will pay, but it will be well above the low incremental cost of an article.

Thus, articles that people want to read have scarcity value — you can only get them from one publisher, generally (this is why publishers care so much about obtaining and protecting copyrights). Whoever has copyright on an article that readers want to read can charge a scarcity rent (price > cost).

The market for publishing has evolved so that a small number of organizations control copyright on the most valuable articles (e.g., Elsevier, Springer, Wiley, Taylor & Francis, the American Chemical Society). They are able to charge prices well above incremental and average cost, so they are are earning above-competitive profit margins. In recent years the profit margins of the largest for-profit scholarly publishers have been around 35% or higher; a competitive, risk-adjusted profit margin is probably closer to about 10%. (See Caveat below.[2]) So, on the order of 25% of what we’re paying is not for the cost of publishing value added, but for excess (above-competitive, or monopolistic) profit.

Back to “flipping”. People concerned about the Expression of Interest or other discussions of “gold” OA almost always assert that we will simply be reinforcing, even strengthening the publishing monopoly.

I am unaware of any cogent economic argument that leads to this conclusion.

To strengthen the existing monopolists, something about the APC (article processing charge) model for paying the costs of publishing (instead of subscription) charges would need to raise the barriers to competition: that is, to give the existing firms increased market power. I don’t see anything about gold OA that does this. It may leave market power unchanged (I don’t think so; see below).

But I don’t see any way in which it increases market power. The primary sources of market power for the large monopoly publishers are three:

  1. They have journals that have a reputation for prestige, and so authors want to submit their articles to be published in those journals, rather than in journals published by less monopolistic organizations.
  2. They have many such journals, which they can sell in “big deal” bundles that make it very difficult for purchasers (mostly libraries) to put competitive pressure on the publishers by dropping subscriptions to their weaker journals (that is, those for which there are reasonably good substitutes).
  3. (This is a big one, and will come back below.) The decision to commit resources to purchase a journal is for the most part made by someone different (usually a librarian) than the decision about where to submit an article for publication (made by the author). Even if authors realize in the abstract that by submitting to publishers that charge monopoly prices they are reinforcing the power those publishers have, which results in their university or research lab having to spend too much on subscriptions, we have a classic collective action problem: the decision of each individual author about where to publish does not directly affect the amount the author’s institution spends on subscriptions, but does affect his or her readership and prestige, so authors (for the most part) quite rationally ignore the monopoly power of the publishers to whom they submit.

Changing to a funding model in which (monopolistic) publishers get paid in the form of pre-publication APCs rather than post-publication subscription payments does not change the first two sources of market power in any way I can see. It may change the third source of market power, but in a way that weakens, not strengthens the monopolies. (I’ll discuss this below).

So why do librarians (and others) fear that flipping to gold OA will increase the power of the monopoly publishers? The only reason that might make sense is that they think there is an alternative path in the future — neither the current market structure, nor gold OA — that will be more competitive than either, and thus that committing to gold OA means we lose the opportunity for something better.

What might that better world be? The only model that I see actively discussed is “green OA”: a model in which the current publishing model continues, but in parallel authors submit the article to an open access repository (typically with an embargo period). Funding agencies in the US have been driving towards this model, with NIH being the most aggressive to date. As a result, most recently published biomedical articles are now made available through PubMed Central, though typically with a six to twelve month lag.

Is the green OA model going to reduce market power (more than gold OA might, see below)? I haven’t seen any research evidence to date suggesting that it has yet (green advocate Stevan Harnad agrees, “no evidence yet of … subscription cancellations … even in fields where it has already attained 100% Green OA for more than two decades”). The pressure on publisher prices would be this: if enough readers were willing to wait the six to twelve embargo months for access, enough subscribers might start cancelling their subscriptions, inducing publishers to lower prices. We’d expect to see this first and strongest among publishers of biomedical journals, since the most successful green OA effort to date is PubMed Central. I haven’t seen a slow-down in the above-inflationary rate of price increase in biomedical journals (much less a decrease in prices, towards competitive levels; see Caveat below[2]).

There are serious limitations on the green OA model, and its ability to create this hypothetical pressure on publishers to lower prices.

  1. It needs to be adopted aggressively by government funding agencies in most or all major research producing countries, if enough content is to be available in green OA to create enough subscription-cancellation risk to force publishers to lower prices.
  2. Author deposit compliance must be high, again, for enough content to be available. Experience to date is that voluntary author submission to OA repositories is very low, and that stands to reason: it’s an extra work-flow for them, on top of the already time-consuming and irritating journal submission process. The exception to low compliance rates is in biomedical, with the difference being that NIH (alone among US funding agencies, I believe) enforces deposit by holding up subsequent funding if an author has not deposited previously-funded research outputs into PubMed Central. But…
  3. Even if all funding agencies (in all research-producing countries) implemented an NIH-like policy with enforcement, much scholarly output is not directly tied to extramural funding. The very strong and specific dependence of identifiable biomedical projects on specific grants makes this model work in that set of disciplines. It would be very difficult to use this compliance enforcement mechanism in many other fields where the link between grant and article is much less clear. Even more so because many scholars in social sciences and humanities (in particular) are not continuously funded, so threatening to withhold the next grant may not have much effect. Indeed, much research receives no extramural funding at all.

I’m not against green OA where it is reasonably efficient. But I think we’ve been trying it long enough now that we can see plenty of (informal) evidence that it is not going to be effective for much scholarly research.[2] And I don’t see any evidence yet that it will reduce publisher market power enough to lower prices even where it is effective.

How might gold OA reduce market power?

Gold OA doesn’t increase the power of the dominant publishers: they are already extracting profit-maximizing monopoly revenues from us, and switching to gold won’t increase the amount they extract. And articles will be available open access to everyone. So we’ll be better off than we are today in the subscription world.

But I think it is likely to be better still, given some time for participants in the market to adjust. Leap ahead some years and imagine the world in which most significant journals are published gold OA, with article pre-payments and no subscription charges. And imagine that authors are at least somewhat involved in the financial decision of where to submit their articles: that is, the submission decision has an economic effect on their research budgets. While authors will still have a strong incentive to publish in high-prestige, widely-read journals, rarely is there only a single good journal suitable for a given article. In this world, authors will have an incentive to also consider the article fee price for submitting to various good journals, and will start to exert competitive pressure on publishers to lower those prices. And it is exactly that competition which has the best — indeed, I think probably the only — chance of leading us to a new market equilibrium in which the amount we spend on scholarly publishing is reduced to the resource cost — that is, the amount it takes to actually provide the peer review management, administrative, copy-editing and marketing services that publishers provide. The obscenely high monopoly profit margins will be whittled down as journals compete to attract submissions of the best articles from the best scholars.

There are some simple and lovely reasons why competition would work in a gold OA world when it doesn’t in a subscription world. The most important is that the person deciding where to submit an article will have some reason to care about the price the publisher is charging. The second is that the first two huge market power advantages I enumerated above are both deflated when journals aren’t sold by subscription. When subscription purchasing decisions are made at the level of the journal (bundle of articles) — and increasingly the “big deal” (bundle of journals) — the only possibility for competition is between journals or bundles of journals, and as I pointed out, different articles published in different journals are not very good substitutes for each other. But in a gold OA world there is competition at the level of the article submission, and before the article is accepted for publication, multiple journals can be very good substitutes for each other (for example, an economist generally gets as much readership and prestige whether her article is published in the American Economic Review, the Journal of Political Economy, or the Quarterly Journal of Economics).

This is the key: in a gold OA world with authors bearing part of the article fee, journals and publishers compete on price, not just quality.

Authors pay? How will they afford this?

Much of the concern about gold OA seems to be about authors who are not funded by research grants to pay article fees. This is “merely” a problem of the distribution of scholarly communication funds. We — society — are paying $X billions currently to sustain the scholarly communications enterprise. Gold OA doesn’t increase the resource cost of commnications: it doesn’t require that we employ more people or more servers or more kWhs of electricity. It shifts the payments from post-publication to pre-publication, but as I argued above, gold OA doesn’t increase market power, and in fact is likely the best path to decreasing market power, and so the total amount paid to publishers should, if anything, fall. So, the problem is “merely” one of getting the money from subscription budgets into APC budgets. (I keep scare quoting “merely” because I am aware that changing administrative flows of funds will take some time, but it’s an administrative problem and can be resolved — the path through which various activities often changes.)

Right now most of the money flowing to publishers is paid by research libraries, who receive their funds from their central campus budgets. For the most part, central campuses (at least in the US) receive these funds through overhead or indirect cost payments by external research funders. This flow of funds suggests one straightforward way in which funds could be re-directed in a gold OA world (there are others). First, central campuses could reduce the budgets of libraries by the amount that we save by not having to pay for journal subscriptions. Then, the central campus could provide authors with research funds to offset that part of their APCs that are not covered by research grants.[3]

I did say that to get the benefits of article submission competition we need authors to bear a meaningful share of the cost. If campuses simply reimburse them for APCs, no matter how high they are, then authors don’t have an incentive to force publishers to compete on price. So, to get competition, campuses might offer only a fixed reimbursement, equal say to a reasonable estimate of the current resource cost (not price) of publishing an article — today around, say, $2000 — and if the author wants to publish in a journal with an above-competitive (monopolistic) APC she will have to come up with additional funds herself. (An elegant alternative that may seem less harsh to authors, suggested to me by Mark McCabe via his collaborator Mackenzie Smith, is to give authors a somewhat higher amount — say, average cost plus $1000 — per article, and let the author bank the difference between that amount and the APC in a research account.)

Is gold OA bad for research-production-intensive institutions?

Some are concerned that if the costs of scholarly communications are paid through pre-publication fees those institutions that produce relatively more publishable research will be disadvantaged because they will pay more in APCs than they will save in subscriptions. It’s almost surely true that there will be some redistribution of where the funds are flowing. However, again, this is “merely” an administrative flow of funds issue and with some time our research support systems can adjust. Most of the funds supporting the costs of scholarly communications (currently mostly subscription charges) are, ultimately, coming from government funding agencies (through indirect cost payments). As the subscription expenditures of institutions that consume (subscribe) relatively more than they produce (publish) fall, indirect cost reimbursements to those institutions will (and should) decline correspondingly. And as the publication costs of institutions that produce relatively more increase, indirect cost payments (or direct cost APC payments through grants) will (and should) increase. As long as the scholarly communications industry doesn’t cost more to support (and when competition heats up in a mostly gold OA world, we will be spending less), it’s just a matter of adjusting the routing of research cost reimbursements across institutions, which is something funding agencies already address continuously.

(I am not advocating that “the rich get richer” — if publication costs go up and they are reimbursed through direct or indirect funding agency payments, institutions aren’t getting richer, they are just being held harmless. I’m not suggesting the Harvards of the world should make a profit by being given more government reimbursement than they need to pay publishing costs!)

Is gold OA bad for under-resourced institutions (e.g., those in the “global south”)?

A variant on the argument that gold OA is bad because authors can’t afford to pay APCs is that it is bad because poor institutions can’t afford to pay APCs. But the point I just made about research-producing institutions demonstrates that generally just the opposite should be true. Under-resourced institutions generally are below average in their ratio of research published to subscriptions obtained. That is, the savings to them on subscription payments should be larger than the APCs they have to pay. (If cost reimbursements are adjusted as I predict, they may not get to keep these savings, but they won’t be made worse off.)

I imagine some will argue that poor institutions, say, in the “global south”, can’t afford either subscriptions or APCs, and should get both for free. OK, that’s a fine policy recommendation, but it doesn’t change the fact that the amount of subsidy they need will be less in a pre-publication payment world (gold OA) than in a post-publication payment world (subscriptions). If publishers, or wealthier nations want to offer subscription subsidies, those subsidies could be provided to cover (the lower total amount of) APCs instead.

Will the transition to gold OA be slow and costly?

Slow? Maybe, though if most research institutions come together behind the Max Planck Society’s Expression of Interest and collaborate to negotiate with the dominant publishers, the tipping point may be near. I don’t see it as any slower than the progress of green OA has been: in fact, since gold OA requires implementation by a modest number of professional organizations (publishers), whereas universal green OA requires implementation by more or less every individual author, I think transition to universal gold is much more feasible than to universal green.

Costly? Maybe. Opponents quickly assert we’ll be paying double while the transition is occurring: paying for all of our subscriptions while publishers still require subscriptions to journals, but also paying APCs for the articles that are published by these journals on an OA basis. This fear is primarily about the so-called “hybrid” path to gold OA: that subscription journals accept and publish a subset of articles OA when the author pays an APC, but that the rest of the content in the journal is only available by subscription. (The journal-by-journal “flip” model has entire journals switching to accept only APC-paid OA articles, and thus eliminating their subscription fees.)

Some advocates for gold OA argue that libraries — especially consortia — will be able to negotiate “offsets”, that is, reducations in subscription payments to offset the amount of revenue the publishers collect in APCs. There has been some progress in this regard, for example in the UK, Austria, and the Netherlands.

I suspect that offsets will only partially cover the costs of transition to a gold OA world. Even where progress is being made, offset savings are lagging behind growing APC payments. And my economic arguments above suggest why offsets are unlikely to fully finance the costs of transition: the dominant publishers have substantial market power, and they are going to use that power to resist the transition to gold OA, trying to make sure we (research-producing institutions) find the transition to be costly.

However, it’s important to recognize that however costly the transition is (and as the consortia demanding gold OA increase in size and experience, offsets will become increasingly helpful), these are transition costs. If the resulting outcomes — open access to scholarly research, and lower prices due to competition — are good for society, then we should be willing to invest something in the transition to get there.



This article is based on a talk I gave at a recent ACRL Workshop on Scholarly Communications, hosted at UC Berkeley on 15 April 2016.

1. Most of the discussion these days focuses on journals and articles, not monographs. I’m going to write using that terminology. But generally everything I say below applies equally to the economics of monograph publishing, as well.

2. Caveat. This is a thought piece. I haven’t had the opportunity yet to do more than casual empirical analysis to back up the empirical claims. So I mention “caveat” above in a few key places where I am making empirical claims that I believe, but which I realize could turn out to be incorrect. I’ve been following these issues long enough to be reasonably certain that there is not compelling empirical evidence to refute my claims, however.

3. Yes, I am a university librarian saying that library budgets should be cut if we no longer have to pay subscription fees for scholarly journals. Our goal as a campus library is not to maximize our budget, but to support the research and education mission of the university, and I think both will be better off if we move from libraries paying for subscriptions to authors paying for article processing charges.