Cost allocation models for OA agreements with UK academic institutions


Since 2013 most journal publishers have received two forms of payments from UK institutions: subscription fees and OA article processing charges (APCs). As a result, Jisc set out requirements for transformative agreements (TAs) which are designed to constrain these costs and accelerate the transition to open access. These agreements also meet the Plan S requirements set out in September 2018, that publishing in hybrid (subscription + OA) journals is only permissible if journals are part of a ‘transformative arrangement’.

Now Jisc negotiates TAs with all journal publishers, insisting that the total fee charged for both access to paywalled content and OA publishing must result in a reduction on existing expenditure These agreements also mean that researchers can continue to publish in hybrid journals and comply with funder OA requirements.

Jisc has asked Information Power to undertake a study to explore the practicality and effect of different models for allocating the cost of TA agreements across UK academic institutions. It  is necessary to look at this because the TA fees now paid to publishers combine the cost to access paywalled content (‘Read’), and the cost of the provision of publishing services (‘Publish’). While all users in all institutions read the content, the amount their researchers publish varies from institution to institution.

Cost distribution models elsewhere in the world

In considering the factors which could form the basis for a new redistribution model, we began by looking at the factors other library consortia use. The Austrian consortium uses two weighted parameters: previous subscription spend and article output. The Dutch consortium uses three equally weighted parameters: institutional funding, student FTEs, and number of articles. The Swedish consortium has a model, not yet implemented, which would be based on every institution paying 10% of the total consortium fee, with the remaining fee distributed across institutions depending on article output.

Principles for cost distribution models in the UK

We interviewed several of the members of Jisc’s negotiation strategy groups to ask them about the principles that should underpin a cost distribution model. We began by asking them about how important it is that a new model marks an effective shift away from historic print spend as a basis. All agreed that it is important to consider an alternative model for cost distribution because it makes no sense to continue using a model which is based on the print subscriptions of over twenty years ago.

While some institutions were keen to see a new model which more closely reflects the value they receive from TAs, they were conscious that care needs to be taken to avoid major increases for some institutions, which could mean they would feel forced to consider stepping away from the Jisc consortium. They were clear that institutions will fare better in the negotiations with publishers if all work together.

It was agreed that the data used to inform parameters for a distribution model should be as transparent as possible. Therefore, we have used data which is in the public domain, such as HESA data for FTEs or data from a trusted Jisc service such as the COUNTER usage data from JUSP.

Which parameters to use in creating a distribution model

We explored a range of factors or parameters that could be used to distribute the costs of transformative agreements. The payments cover a spectrum of services that benefit authors, institutions, and readers.

  • Services for authors – because these agreements include the cost of publication, article output is an obvious factor to be considered. However, our interviewees pointed out that article output can vary from year to year, and so the modelling we have undertaken is based on an average of article output over the last three years.
  • Services for readers – because there will continue to be content behind the paywall for some time and because publishers enhance even OA articles to facilitate discovery and use by readers, it is essential for a Read element to be factored in. We have modelled using two parameters, either FTE or usage. Both FTE and usage figures can fluctuate year by year, so in our modelling we have used an average over three years for each. The three-year average follows the precedent set by the Jisc Band calculations, which average institutional income over a three-year period. There is an argument that only academic FTE should be included, so we have modelled that, but also combined academic and student FTE.
  • Services for institutions – publishers also provide value add services that benefit administrators of institutional services. We have factored in ability to pay in the form of Jisc Banding to take this into account in an equitable way.

In our modelling, we have weighted read and publish price factors equally. We have based these percentages on the Plan S Pricing Transparency Framework, which indicates that about half of publishing activity is focused on services to authors, and around half on services to readers. However, if implemented the Jisc community could agree to different weighting, and this could be amended to reflect different TA agreements. We have also modelled using three equally weighted parameters; read and publish price factors and ability to pay. We have used Jisc banding for this third parameter.

The nature and scale of potential changes

Developing a distribution model for TAs is not a simple case of mapping from one model to another.

Our work indicates that whichever parameters are applied, the change could mean some institutions paying much more or much less than now. The actual nature and scale of changes are not consistent across publishers, and so applying exactly the same parameters to each agreement could mean an institution pays more for some publisher agreements and less for others. This is because the amount currently paid to publishers is based on legacy pricing models for print, and on subscription decisions taken decades ago. Historic print spend does not reflect the size, article output, or financial status of an institution.

Historic print spend is usually based on the print holding of an institution in the early 1990s. This creates strange anomalies with particular journal agreements. For example, as shown in Figure 2, for one publisher agreement, a Band 2 institution is currently paying 60% more than a Band 1 institution. However, for another publisher agreement the Band 2 institution is paying less than half the amount the Band 1 pays. Huge swings are seen as the costs are redistributed using transparent data about publication output, FTE, usage or Jisc Banding.

 Publisher A – current spendPublisher B – current spendPublisher C– current spendPublisher D – current spend
Band 1 Institution£674,222£677,876£8,415£221,241
Band 2 Institution£1,071,157£305,862£6,653£155,429

Figure 1: Example of current spend by a Band 1 and Band 2 institution across four different publisher agreements.

Managed implementation
Wild variations in price would not be manageable for institutions. Therefore, we have added caps and floors to limit variation in price from year to year. This will enable a more managed migration to a new model over the course of several years.

In our modelling, we illustrate a migration over a three-year period, however, the Jisc community could choose a longer or shorter implementation time frame.

Implementation notes

Any new distribution model which Jisc members find broadly acceptable is unlikely to be implemented immediately. It is likely that a managed migration would be implemented over a number of years, so that no institutions are faced with sudden and unaffordable increases in the fees.

Because historic print spend is not a model, but the result of legacy decisions and pricing, it is possible different flavours of a new model could be used for different publisher agreements, in order to smooth the migration.

Next steps

We want to consult with all Jisc member institutions about a distribution model and the parameters that should be used. However, we recognise that it is difficult for institutions to comment in the abstract, without knowing what the possible impact would be on their own institution.

Therefore, we have modelled different scenarios across a range of publisher agreements for each Jisc member library. We will seek your feedback through a webinar and a survey.