The Occasional Pamphlet

…provide a transition path… “The Temple of Transition, Burning Man 2011” photo by flickr user Michael Holden, used by permission

David Willetts, the UK Minister for Universities and Research, has written a letter to Janet Finch responding to her committee’s “A Review of Progress in Implementing the Recommendations of the Finch Report”. Notable in Minister Willetts response is this excerpt:

Government wants [higher education institutions] to fully participate in the take up of Gold OA and create a better functioning market. Hence, Government looks to the publishing industry to develop innovative and sustainable solutions to address the ‘double-dipping’ issue perceived by institutions. Publishers have an opportunity to incentivise early adoption of Gold OA by moderating the total cost of publication for individual institutions. This would remove the final obstacle to greater take up of Gold OA, enabling universal acceptance of ‘hybrid’ journals.

It is important for two reasons: in its recognition, first, that the hybrid journal model has inherent obstacles as currently implemented (consistent with a previous post of mine), and second, that the solution is to make sure that individual institutions (as emphasized in the original) be properly incentivized for underwriting hybrid fees.

This development led me to dust off a pending post that has been sitting in my virtual filing cabinet for several years now, being updated every once in a while as developments motivated. It addresses exactly this issue in some detail.

A transitional scholarly journal business model

By way of background, recall that there are the toll-access scholarly journals and the open-access journals. The former comprise the bulk (about two-thirds) of peer-reviewed journals, the latter a growing minority. There is great interest in how a transition from the former to the latter might occur. The most common proposal is the hybrid model, in which a toll-access journal will accept an optional publication fee to make individual articles open access.

I’ve previously argued that the hybrid model does not, despite intentions, provide a transition path. I argue here that an alternative approach addresses many of the failings of the hybrid model.

What do we want of a transitional process? First and foremost, a transitional process should allow for a smooth transition path from toll-access to open-access that is, in the short term at least, revenue-neutral. Intrinsically revenue-reducing transitions will find little support among publishers and revenue-increasing transitions will be appropriately rejected by scholarly institutions. Note that just because a transitional process is revenue-neutral in the short term does not mean that no moneys will be saved in the longer term as the result of the transition; a move to author-side fees from reader-side fees has the potential to be a much more transparent, competitive, and efficient market, which may well lead to overall cost reductions. Secondly, a transitional process should embed incentives for movement along the path. This second issue has rarely been taken into account.

Flaws in the hybrid model

A quick rehearsal of the problems with the hybrid model shows flaws with respect to both aspects.

The hybrid model is revenue-neutral in the short run just in case (i) a journal’s hybrid publication fee is set to be the average revenue per article for the journal under the toll-access model and (ii) the publisher reduces the subscription fee in exact proportion to the percentage of articles for which the hybrid fee is paid. In that case, a publisher receives the same revenue whether no one pays the hybrid fees or everyone pays the hybrid fees (in which case the subscription fee becomes zero). For comparison with later models, we can capture this with some simple arithmetic. Suppose there are \(t\) institutions of which \(s\) are subscribers. For simplicity, assume that there is no price discrimination, so all subscribers pay subscription fee \(f\). Then total revenue is \(s \cdot f\). (I assume here that all revenue of the journal is subscription revenue. If there is other revenue, for instance, advertising revenue, that shouldn’t play a role in setting publication fees, since it doesn’t get displaced by those fees.) Suppose further that there are \(n\) articles published each year. Average revenue per article is thus \(\frac{s \cdot f}{n}\), which we choose for the hybrid fee. Finally, assume \(h\) percent of articles have the hybrid fee paid. Then the total revenue from hybrid fees is \(n \cdot h \cdot \left(\frac{s \cdot f}{n}\right) = h \cdot s \cdot f\) and total revenue from subscriptions is \( s \cdot f \cdot (1 – h) \). Total revenue is therefore \(s \cdot f \cdot (1 – h) + h \cdot s \cdot f = s \cdot f\), independently of \(h\). This is the sense in which the hybrid system is (or at least can be) revenue neutral. In a previous post, I referred to a hybrid model that had these revenue-neutrality properties as a true hybrid.

Of course, there is a serious problem that it is difficult, and perhaps impossible, to determine if the revenue-neutrality conditions are met. It requires knowledge of the average revenue per article, as well as transparency of subscription prices to verify that subscription fees are reduced to \(f \cdot (1-h)\). But let us assume that these issues are surmountable, because there are bigger problems even with true hybrid journals.

Since the price reduction for each hybrid fee paid is distributed over all subscribing institutions, the net effect on each institution of paying a hybrid fee is a large increase in fees paid. The institution pays \(\frac{s \cdot f}{n}\) but recovers just \(\frac{s \cdot f}{n \cdot s} = \frac{f}{n}\) subscription fee reduction; it recoups just \(1/s\) of each hybrid fee paid. That is, for every hybrid fee it pays, it recoups a share inversely proportional to the number of subscribers to the journal. The reduction is minimal even for journals that are poorly subscribed. If the journal has 100 subscribers, the author institution recoups just 1% of each hybrid fee it pays. Institutions would be unlikely to pay hybrid fees under these conditions. Indeed, this is what we see empirically. The setup induces a coordination problem: If all institutions paid the hybrid fees, all would be better off – collectively paying the same amount but achieving open access to all articles. But no institution has an incentive to be the “first mover”.

A preferable alternative

To solve this prisoner’s dilemma, a better method would allocate all subscription fee benefit of a hybrid payment to the author’s institution. Rather than each subscribing institution seeing a reduction of \(f/n\), the author’s institution would see a reduction of \(f\). I call this business model the transitional model, to distinguish it from the hybrid model and also the pure models – the subscription model and the publication fee model. To the journal, there is no difference in revenue in the short run between these two approaches, hybrid and transitional. Instead of reducing the subscription fee of \(n\) institutions by \(f/n\) as in the hybrid model, the transitional model reduces the subscription fee of 1 institution by \(f\), in either case for a total reduction of \(f\), thereby exactly recouping the publication fee and retaining revenue neutrality. To the author’s institution, however, there is a big difference. Under the transitional model, it pays \(f\) in publication fees but recoups \(f\) in subscription fees; it’s a wash, so any institution should happily underwrite such a fee. It costs no more than the institution is already paying, and provides open access to the article.¹ All of the institutions other than the author’s no longer receive a lower subscription fee, but the marginal reduction to them was trivial anyway, so their behavior wouldn’t be expected to change. I’ll refer to articles where the publication fee leads to a commensurate reduction in subscription to the institution as a fully recovered article.

Parenthetically: What happens once an institution has paid so many publication fees that it has fully covered the entire subscription price? After an institution has published \(s/n\) fully covered articles, it has paid \(s\) in hybrid fees and its subscription fee is reduced to zero. At that point, there are several approaches that can be taken. Here’s one: Since there’s no more subscription fee to reduce, to preserve revenue neutrality, the publisher could then reduce the subscription fee of all other institutions by \(f/n\) (or a bit more actually, to account for the subscribers who have bottomed out on their subscription fee). Articles from that point on are partially recovered articles; they work like hybrid journals currently do.

There needs to be a way to complete the transition from a transitional subscription journal (with this kind of publication fee) to a fully open-access journal. The natural approach is a commitment on the part of the journal to switch models once a sufficiently high fraction of articles are paying hybrid fees. This transition threshold should be determined by where the publication fee model becomes independently viable. I don’t know where this threshold should be, but presumably quite high, perhaps 75% or more. One interesting point to examine is the percentage of articles published that would fall under the full recovery limit. If the percentage of fully recovered articles is greater than the transition threshold, the journal is quite likely to complete the transition.

I would appreciate suggestions as to how to estimate these quantities: the transition threshold and especially the fully recovered percentage. The latter is dependent on how skewed the publication behavior of institutions is. The more skewed, the lower the fully recovered percentage.

One way of thinking about the transitional model is that subscription fees are merely reenvisioned as prepaid publication fees. In that sense, the transitional proposal looks a bit like the SCOAP3 model, except that crucially unlike SCOAP3 (but like the hybrid model), there is no free-rider problem. Institutions can’t immediately drop their subscriptions and free ride on the availability of all articles. Rather they would gain access only to the open-access subset of the articles, and once the transition to fully open access occurred their authors would have to pony up publication fees (perhaps subsidized by the institution, and of course waivable in case of exigency so as not to disenfranchise necessitous authors).

In fact, since I first began thinking about this issue, a journal publisher has actually implemented just such a system. The Royal Society of Chemistry has developed what they call the RSC Gold program, in which institutional subscribers to their journals receive vouchers that they can use to cover a certain number of OA fees. To the extent that the value of the vouchers is identical to the subscription cost of the set of subscribed journals (which is not made clear in their literature), this would constitute an instance of the transitional model. Unfortunately, the RSC Gold program is limited to subscribers that take the entire bundle of RSC journals, furthering the market problems of bundling.

Implementing the transitional model

The transitional model relies on reduction of the subscription fee in lock-step with payment of publication fees. There are several ways this could be achieved. Let’s assume subscription fees are paid in advance annually, for simplicity on a calendar year basis. One way is to perform the accounting of subscription fees and publication fees on a same-year basis. The institution pays the subscription fee for a given year and then pays no publication fees for articles published in that year until the subscription fee is fully recovered. Thereafter, the institution (or its authors or authors’ funders) can pay further publication fees, which will be only partially recovered.

Another possibility is to recover in arrears. The institution pays the subscription fee for a given year and then during that year whatever publication fees it chooses to. The publication fees paid in that year get recovered in the following year by a reduction in (or rebate of) the subscription fee for that year. This method has some important advantages. In particular, it is easier for the institution to decide on an article by article basis whether or not to actually pay the publication fee or to have a funder pay it instead or even to refuse to pay it based on whatever criteria it uses. I’ve argued elsewhere that the ability to control which publication fees get underwritten by an institution is an important check on the publication fee market; it allows the construction of a market not subject to the moral hazards of the subscription model. On the other hand, the arrears approach has the disadvantage that the institution is always floating the future reimbursement. This means that it will incur a big increase in cost, perhaps a doubling or more, in the first year, after which time things will settle back to roughly the status quo ante. Institutions can at their control spread that increase over several years by covering an increasing percentage of articles over time.

A third method solves this problem. Publication fees are recovered immediately, not in arrears, but a “fee” needs to be sent for every article, whether recovered or not. (The fee for a fully recovered article would be 0, so really this amounts to just a notification to the publisher that this article is one for which recovery is being claimed.) This allows the institution to refrain from paying (and hence recovering) an article’s publication fee if that would violate whatever constraints it puts on underwriting fees. This is, in essence, the approach taken by the Royal Society of Chemistry’s RSC Gold voucher program.

Update January 9, 2015: JISC and Wiley have entered into an experimental agreement along these lines, as announced December 19,2014. Thanks to Thomas Munro for the pointer.

There are still problems with this transitional model. It only works for what I called true hybrids, that is, the OA fee must be set at no more than the true average revenue of the journal. There are still all the problems of knowing what that true average revenue is and of monitoring subscription decreases in the face of price discrimination and for the partially recovered articles. And finally, there is a greater complexity to the system. This final problem might be mitigated by a clever reimagining of the transitional model, just as eBay managed to get across the idea of second-price auctions to its customers by reimagining them as a traditional (first price) auction with proxy bidding. The RSC voucher approach may be a good first step in that direction.

Would publishers approve?

A few months ago, I spoke to a group from a major commercial publisher about this business model. (The topic came up in a question about why Harvard’s open-access fund doesn’t cover hybrid fees.) The reaction to this kind of proposal – which was not news to them because of the RSC program – reveals a deep problem in how this publisher thinks about the OA transition. The problem with this approach, I was told, was that as a larger percentage of articles became available open access, libraries may start to cancel their subscriptions, reducing revenues to the publisher in a way that is not made up for by the OA fees.

This is, of course, true. (It would hold also for the hybrid approach, except for the fact that uptake is so low that there is essentially no incentive to cancel subscriptions merely because of hybrid OA articles, and there is unlikely ever to be.) It is because of this possibility — that over time as the transition happens that subscriptions may be cancelled — that I refer to revenue neutrality in the short term. Examining revenues related to the marginal article, the scheme I described is revenue neutral, but overall as the larger-scale transition starts to occur, aggregate phenomena can change the revenue neutrality.

In the face of these changes, publishers have choices. If a publisher wants to achieve revenue neutrality in the face of subscription cancellations, it could raise its OA fee accordingly. The higher fee might have the effect of reducing the attractiveness of the journal to authors as they compare the fee against that of other journals, but that must be traded off against the attempt to maintain revenue. Setting prices is a business decision, a decision that should be made by the publisher to maximize its revenue. The fact that that’s harder to do in the transitional model as the anticompetitive features of the subscription market are reduced is an advantage of the model, not a flaw.

This publisher claimed that their concern was that the transitional model could substantively affect their bottom line. But what they were really admitting is that open access could substantively affect their bottom line. If uptake on the transitional model could induce cancellations that could not be recouped by increases in article fees, then the same is true for the hybrid model. Why is this publisher (like many others) an enthusiastic supporter of the hybrid model? I’m guessing it’s because they know that the hybrid model will never have substantial uptake. Since the transitional model might, they oppose it.

The point of the open-access journal model is not to maintain publishers’ revenues at the current levels made possible by the dysfunctional journal market. It is to provide publishing services without using access limitation to fund them. If doing so also introduces a competitive free market mechanism that saves money – as this publisher implicitly corroborates – so much the better.

Perhaps many current publishers, seeing the likelihood that any realistic approach to an OA transition would harm their revenues in the long term, would avoid a model like the one discussed here that has a real possibility of navigating the transition. But there may well be forward-thinking publishers (society publishers perhaps), who would honestly like to make the transition if it could be done in an appropriately gradual manner. For them, this transitional open-access model may be just the thing. If so, they should be supported in taking it up.