Submission fees: a means of defraying costs for OA journals?

I was struck by a recent report on a talk by Harold Varmus that

The [PLoS] peer reviewed journals now had high reputations, and rejected some 90% of submissions, but had needed to raise fees beyond his forecast of five years ago to cover costs. While the regular journals were still making losses, and these losses were still being covered by philanthropic foundation support, he was confident that the ongoing growth of PLoS One would see the whole of PLoS cover its costs from its own trading operations next year.

There are two points explicitly mentioned there, and two more implicit in the broader discussion; respectively:

• PLoS journals reject many more articles than they accept. (N.B. Only accepted articles are levied the article-processing charges that are the backbone of PLoS’ business model.)
• Not all PLoS journals are self-financing yet. (N.B. Few start-ups in any industry are completely self-financed in their first five years; and in addition to PLoS’ publishing function, it also has a philanthropic function that its revenues help underwrite.)
• There are still questions about the effects of article-processing charges as a business model for publishing: namely, does it unfairly exclude (or deter) authors who can’t afford them, and who will pay for them.
• There is still doubt about the financial sustainability of the article-processing charges model.

The APC model isn’t the only model for financing OA journals (a majority of OA journals use other models), but it is a significant one. Scholarly publishing has always been a field where different business models flourished, so I wouldn’t expect any less of OA scholarly publishing. The APC model is compatible with other revenue streams, such as underwriting by sponsoring organization (e.g. a scholarly society, research institution, library, government agency, etc.), donations, philanthropic grants, advertising, sales of print subscriptions, sales of merchandise, etc. So let me touch a third rail and suggest another method of defraying costs for OA journals: submission fees.

Defenders of the APC model are quick to point out that no author is charged until their article is accepted for publication. It’s not pay to play, the argument goes: articles are accepted on the basis of their quality, without regard for the author’s willingness or ability to pay; only after refereeing is any invoice sent. That’s all well and good, but look back at the numbers from PLoS: they’re rejecting 9 out of 10 papers submitted. The 9 rejected papers require editorial resources, but they don’t pay any of that cost. It’s an externality: if I have a paper I know isn’t up to a PLoS journal’s standards, I can submit it on a lark anyway, at no cost to me. The journal still has to locate reviewers and shepherd the paper through the review process. So, what do you do when demand for editorial resources is outstripping the supply of said resources? You raise the price.

One of two things happens when you raise the price of access to editorial resources, depending on how elastic the demand is:

• If demand is elastic — I didn’t seriously think PLoS would publish my paper, I was only going to submit it on a whim — then demand tamps down, reducing the pressure on referees and editors.
• If demand is inelastic — everybody really wants their paper in a PLoS journal — then demand isn’t reduced, but you generate more revenue, which you can use to increase the supply of editorial resources (e.g. by hiring more editors, investing more in technology to better manage to editorial process, spending more money to recruit referees, providing perks for reviewers to motivate and retain them, etc.).

Let’s address first what the effects of those outcomes would be, then speculate on which is more likely.

If demand is elastic: There’s some argument that reducing the number of submitted papers would be bad. We’d rather get more submissions and have to filter them, a journal might say, than have authors second-guess their paper’s suitability and potentially miss out on a gem. There might be something to this, but I don’t think it’s a significant concern. Authors are smart people, most of them university faculty with terminal degrees, who are familiar with their field: they know roughly how strong a paper is and where it’s likely to be accepted. Even if they didn’t, there’s a strong pressure to be ambitious — you want a paper in a prestigious publication so you can get tenure, that next grant, etc. — so you’ll aim high in where you submit. But not unreasonably high: even without submission fees, there’s already a significant cost to submitting a paper: time. If you know Nature isn’t going to publish your paper, you won’t submit it, because you’ll be forgoing the weeks (or longer!) it takes for them to reject your paper. Since authors shop their papers serially, not concurrently, every rejection means a longer wait until the paper is finally published. So there’s already a strong incentive to get it right the first time — which is the same thing a submission fee does.

If demand is inelastic: I can’t see how anyone would argue with the effect of a prestigious OA journal raising more revenue to fund its editorial processes, other than with one of these claims:

• Authors won’t be able to afford the submission fees; or it will be an undue burden on researchers.
• The journals will enrich themselves, at the expense of the research enterprise.

Notably, both those claims are already levied against the APC model (and, it must be said, the subscription model). As to the latter, the only defense is market pressures, consumer action, or regulatory intervention: in other words, if a journal is abusing the system, either the market will dispatch with them, or else consumers (researchers, libraries, universities) must organize against it, or government must intervene (e.g. a funding agency capping the available funding to cover fees). As to the former, it’s a question of how expensive submission fees are. I would propose a moderate amount, in the range of $20-$100 per submission. I find it hard to believe that an author who feels strongly that her paper is suitable for a certain publication will be unable or unwilling to contribute $50 to that publication’s continued operation, or that $50 per submission will be an undue burden*. (* At least in rich countries; it will be appropriate to continue the APC system of discounts and waivers to accommodate developing countries. There is also some argument for trying to accommodate independent and unsupported researchers, especially in fields like the humanities which are less likely to be grant-funded.)

Keep in mind that in the U.S., most universities charge application fees in the range of $20-100 (with some waivers available). If an 18-year-old can be expected to chip in $30 on his college application, surely tenure-track faculty can find $30 to help ensure the sustainability of the OA journal which they hope will publish their paper. (In fact, most university-bound Americans apply to more to one college, so the student might have to pay six application fees. If a paper has been rejected from six journals, well — the author might have a bigger problem than paying for submission fees.)

So now to the question of which is more likely: I expect demand would be fairly inelastic. Here’s why: all the authors submitting to an APC-levying journal already know that if accepted, they’ll have to pay the APC (subject to discounts and waivers). So they’re willing and able to pay. If the 90% of rejected authors are willing and able to pay an APC (which often runs more than $1,000), I can’t imagine that many would be unwilling or unable to pay a submission fee in my suggested range of $20-100. (If you like, you could deduct the submission fee from the APC for accepted papers.)

So would it actually generate revenue for the journal? Well, let’s take the example of a fairly selective journal with a fairly high APC: a 90% rejection rate and a $2,000 APC. A $100 submission fee would generate $900 of additional revenue for each accepted article. That’s a 45% increase in revenue, which is not chump change. Even a journal with only a 50% rejection rate, a $2,500 APC, and a $100 submission fee could result in a 20% increase in revenue. (Of course, this is assuming perfect inelasticity, so probably adjust the figures down a bit.)

In summary, I speculate that submission fees could generate significant revenue for APC-model OA journals without significantly decreasing submissions or unduly burdening authors. Worth a shot, no?

(Bonus: Waiving or discounting submission fees could be another perk for publisher institutional memberships, improving the value proposition there. Plus, the usual APC gimmicks are applicable: waiving the fee for a first-time author, waiving the fee for the inaugural issue or the first year, etc.)

(P.S. For a publisher concerned with the stigma of pay-to-play, or of squandering goodwill with authors, call the submission fee a “recommended donation” and make it clear it doesn’t affect editorial outcomes. You’ll get less than 100% compliance, but I bet at least a few authors chip in, especially as more funds spring up to support OA publishing.)