Transformative agreements: Do they pave the way to open access?

Transformative agreements, also known as ‘offsetting’, ‘read and publish’, or ‘publish and read’ agreements, have shifted the focus of scholarly journal licensing from cost containment towards open access publication. An analysis of 36 full‐text transformative agreements recorded in the ESAC registry shows that ‘transformative agreement’ is an umbrella term that encompasses different kinds of contracts. We differentiate between pre‐transformative, partially transformative, and fully transformative agreements. Pre‐transformative agreements are traditional subscription licences that grant article processing charge (APC) discounts or vouchers for open access publication of a limited number of articles. Partially transformative agreements differentiate between a read fee and a publish fee to cover the processing charges of a certain number of articles. Fully transformative agreements allow unlimited open access publication of the scholarly output of the subscribing institution. In all three categories, some agreements restrict open access publication to hybrid journals, whereas others allow publication in both hybrid and gold journals. Transformative agreements are more transparent than traditional journal licences, allow authors to retain copyright, and make provisions to facilitate the management of open access workflows. It is hard to assess whether these agreements are just a temporary phase in the transition towards open access or will perpetuate the current structure of the scholarly communication system and its associated high costs.


INTRODUCTION
Since the 1970s, when the expression 'serials crisis' was coined (Jurchen, 2020), subscription prices to scholarly journals have regularly increased at a rate higher than inflation. Between 1984 and 2010, the average price of US academic journals increased more than eightfold, while the inflation rate was 110% (Shu et al., 2018). A possible solution to this problem emerged at the end of the 1990s, with the rise of the Internet. The migration of print journals to an electronic format has had limited impact on formal aspects (Herman, Akeroyd, Bequet, Nicholas, & Watkinson, 2020) but has resulted in major changes to how scholarly journals are marketed, their pricing models, and their availability to readers (Anglada & Comellas, 2002). Academic and research libraries began associating in consortia and replacing the subscription of individual titles with big deal licences of journal bundles marketed by large publishers at a discount off the aggregated list price (SPARC, 2020).
has the potential to be a much more transparent, competitive, and efficient market, which may well lead to overall cost reductions'. Shortly afterwards, a report by the Max Planck Society (Schimmer et al., 2015) stated that 'all the indications are that the money already invested in the research publishing system is sufficient to enable a transformation that will be sustainable for the future. There needs to be a shared understanding that the money currently locked in the journal subscription system must be withdrawn and repurposed for open access publishing services'.
The notion of offsetting the costs of OA against the cost of subscriptions was rapidly implemented. In July 2012, the Royal Society of Chemistry (RSC) announced 'Gold for Gold' as 'an innovative experiment to support the funder led evolution to Gold OA'. It provided British institutes that were RSC subscribers with 'credit equal to the subscription paid, enabling their researchers, who are being asked to publish Open Access but often do not yet have funding to pay for it directly, to make their paper available via Open Science, the RSC's Gold OA option' (RSC, 2012a). A few months later, the RSC announced the expansion of the programme worldwide (RSC, 2012b). In February 2014, a group of Austrian institutions and IOP Publishing announced a new pilot programme that would 'provide advance funding for Austrian researchers to publish on a hybrid open access basis in IOP's subscription journals and which will offset that funding against subscription and licence fees paid by the Austrian Academic Consortium for access to IOP's journals' (IOP Publishing, 2014).
At roughly the same time, in November 2014, the Association of Universities in the Netherlands (VSNU) announced the renewal of its subscription to Springer with terms including OA so that 'scientists in the Netherlands will be able to publish in open access format in existing Springer journals, while retaining reading privileges to these journals as well'. At that point in time, the Dutch State Secretary for Education, Culture and Science fixed the target that 'five and ten years from now 60% and 100% of all Dutch academic publications, respectively, should be open access publications' (VSNU, 2014). Similar agreements were reached a year later with Wiley (VSNU, 2015a) and Elsevier (VSNU, 2015b) following a threat of boycott by the consortium if 30% of Dutch papers were not OA in VSNU-subscribed Elsevier journals by 2018 (Butler, 2016). In the UK, Jisc signed a Springer Compact agreement that would run from October 2015 until December 2018, which combined OA publishing and subscription access in one annual fee. A major driving force for the Dutch and British deals 'was to combat the expensive and controversial 'hybrid' business models' (Butler, 2016).
The negotiation of TAs coincided with the cancellation of big deals. At the end of 2016, the Finnish consortium FinELib and Elsevier settled a 1-year extension of negotiations 'in order to find a solution for advancing open access, which is an extremely important goal for the Finnish research community' (Peltonen, 2016). At the same time, it was announced that scientists in Germany, Peru, and Taiwan were about to lose access to Key points • The negotiation of journal licence agreements has shifted its focus from cost containment towards the inclusion of clauses in favour of open access.
• 'Transformative agreement' is an umbrella term that encompasses a continuum of contracts, ranging from traditional subscription licences that grant discounts in publication fees or vouchers to agreements allowing unlimited open access publication.
• Transformative agreements are more transparent than journal licences, allow authors to retain copyright, and make provisions to facilitate open access workflows.
• It is hard to assess whether transformative agreements are transitory or will perpetuate the current structure of the scholarly communication system. Elsevier journals. In Germany, the DEAL consortium of universities and research organizations complained that the contract proposed by Elsevier cost too much and failed to include an OA clause (Schiermeier & Rodríguez Mega, 2017). Negotiations went on for a year (Schiermeier, 2017a), with German universities asking for a 'deal that would give most scientists in Germany full online access to […] Elsevier journals, at about half the price that individual libraries have paid in the past' (Schiermeier, 2017b). At this stage, OA proved 'to be the sticking point in the talks: under the deal sought, all corresponding authors affiliated with German institutions would be allowed to make their papers free to read and share by anyone in the world at no extra cost' (Schiermeier, 2017b). Finally, in July 2018, Elsevier cut off access to German and Swedish researchers when negotiations on new contracts broke down (Else, 2018a German negotiations with other publishers proved more successful than those with Elsevier. At the beginning of 2019, DEAL reached its first TA with Wiley (Else, 2019a), and at the beginning of 2020, it reached another with Springer Nature. Through these two agreements, the number of articles by German authors published OA will be around 10,000 and 13,000 per year, respectively.
Although the agreement with Springer Nature included journals published under several imprints, Nature and its sister journals were excluded. However, on 20 October 2020, the Max Planck Digital Library announced a new agreement covering reading and OA publishing in 34 Nature-branded journals plus access to articles in 21 Nature Reviews journals on the basis of a price of 9,500 euros per article (Van Noorden, 2020).
Elsevier, which had trouble reaching TAs in Germany and Sweden, struck a national deal in Norway at the beginning of 2019 (Elsevier, 2019;Qureshi, 2019;Unit, 2020). Under the agreement, Norwegian scientists gained access to Elsevier journals and were able to publish 1,850 OA articles annually.
Based on historical data, this would cover about 90% of output by Norwegian authors in Elsevier journals. According to a negotiator for the Norwegian consortium, the deal was 'cost neutral' compared with the previous agreement, which did not include OA fees. The consortium had paid Elsevier 9 million euros in subscription costs in 2018, plus an estimated 1 million in APCs (Else, 2019b). The consortium of Swiss universities also reported, at the end of 2019, that they had signed memorandums of understanding with Elsevier and Wiley to include OA publications in their licences (swissuniversities, 2019). The agreement with Elsevier was announced in May 2020, enabling Swiss researchers 'to publish Open Access across the majority of Elsevier's Gold andHybrid journals, scaling up to 100% by 2023' (swissuniversities, 2020).
Other consortium agreements started to include OA clauses too. In Greece, in the context of negotiations for the 2019-2021 renewal of licences, provisions were made to combine access with OA publication. The website of the HEAL-Link consortium (2020) lists information about several agreements with publishers, including either the publication of a certain number of articles in gold and/or hybrid journalsas in the agreements with Cambridge University Press or the Royal Society of Chemistryor a 10% discount on the APCas in the case of Elsevier, Emerald, IEEE, or Wiley. Similarly, the agreements negotiated in Portugal for the period between 2019 and 2021 included either APC waivers for a certain number of articles or APC discounts ranging from 5% to 20% (J. Novais, personal communication, August 5, 2020). The consortium of Irish higher education institutions reached an agreement with Elsevier in February 2020, which in addition to giving them access to its journals enabled 'Irish researchers to publish more than 70 percent of their research without having to pay an Article Publishing Charge' (Irish, 2020). South Africa is also considering a similar national shift (Bawa, 2020). By August 2020, Springer listed 13 countries where 'Read and Publish (Springer Compact)' agreements had been signed, all of them in Europe except for two in Asia (Springer, 2020).
Contrary to what other large European consortia were doing, in April 2019, the French Couperin consortium announced the renewal of its bundled journal subscription with Elsevier (Clavey, 2019). The agreement included a 13.3% discount on access over 4 years and a 25% APC discount for French researchers in Elsevier gold and hybrid journals. The noveltyand the controversyof the agreement lay in the clauses related to the provision of green OA (Bourrion, 2019). The deal envisaged that, 12 months after publication, the postprint (author-accepted manuscript) would be automatically uploaded by Elsevier to ScienceDirect with a notice in HAL (the French national repository) linking to the file. Later, 24 months after publication, this manuscript would be available in HAL.
This procedure would relieve French researchers of the task of depositing their articles, but it would do so with an embargo period longer than the one provided for in French law (6 months in science and technology and 12 months in social sciences and humanities).
So far, TAs have essentially been a European phenomenon (Else, 2018b). A survey by the European University Association (Morais, Bauer, & Borrell-Damián, 2018) reported that, in 2017, only 11% of consortia in Europe made deals that included OA, but 63% planned to do so in the future. Possibly, one of the reasons for this interest is the fact that funding institutions increasingly mandate OA for the results of the research they fund. In 2018, cOAlition S (www.coalition-s.org), a consortium of national research agencies and funders from 12 European countries, launched Plan S, which requires researchers who benefit from grants to publish their work OA by 2021. The ESAC initiative (https://esac-initiative.org), currently managed by the Max Planck Digital Library, maintains a registry of TAs to which libraries and consortia are invited to submit their agreements.
Meanwhile, in the USA, a growing number of libraries were critically reappraising big deals, resulting in numerous cancellations tracked by SPARC (2020). In June 2018, Schonfeld (2018) suggested that TAs, which were becoming quite popular in Europe, would have much less of an impact in the USA. Making a comparison with the flu pandemic of 1918 (sadly, a timely analogy in light of recent developments), he stated that 'while the germs are circulating, at least in the near term, publishers are unlikely to face a global pandemic'. The diversity of the American market seems to be to blame for the delay in adopting TAs (Machovec, 2019). However, in the past 2 years, there has been a growing number of TAs in the USA since the first one agreed on between the Massachusetts Institute of Technology (MIT) and the Royal Society of Chemistry (Fay, 2018).
At the end of 2018 and into the first quarter of 2019, the University of California and Elsevier ran into difficulties in reaching an agreement to renew the contract that expired in December 2018. The University of California was 'not primarily protesting rising prices, though it would still like to see its fees reduced', but it was seeking 'to fundamentally alter how it pays for journal content from publishers like Elsevier and to accelerate open-access publishing in the process' (McKenzie, 2018). The breakdown in negotiations was announced in February 2019 (Fox & Brainard, 2019;Gaind, 2019). The dispute was significant because the University of California system accounts for 10% of scholarly output in the USA, and it was paying more that 10 million dollars for access to Elsevier journals. The difficulties arose because the University of California wanted to roll access and publication into one annual fee, whereas Elsevier did not want to shift to a fixed annual rate for an unknown number of articles. As the first TAs in the market began to expire, subscribing institutions conducted evaluation studies to assess their performance. In brief, although it was found that TAs had slowed down increases in costs and could help institutions in the administration and implementation of OA, they were also 'flawed through their implicit acceptance and strengthening of the current costly and opaque market for journal subscriptions' (Earney, 2017). When Jisc assessed the results of the offset agreements it had signed over the period 2015-2017with an estimated combined value of £19.5 millionit observed that the Springer Compact agreement was the biggest contributor to subscription cost avoidance while also providing the largest proportional reduction in publishing costs. It concluded that offsetting had increased the value of journal licences and had raised the number of articles published OA. However, 'the approach also has significant drawbacks, notably the risk of entrenching the existing structure of the journals market and locking up even more money in big deals rather than reducing overall costs' (Lawson, 2019). When assessing the results of the first TA signed in Sweden with Springer for the period from 2016 to 2018, Olsson, Francke, Lindelöw, and Willén (2020) reached similar conclusions. The advantages of TAs would include the increase in OA publishing, improved workflows, and ease for researchers. However, TAs also posed the risk of perpetuating the current system and its associated high costs. More recently, in the UK, given the sharp increase in publishing during the first 9 months of the TA with Wiley, Jisc decided to implement controls on articles being published OA (Vernon, 2020). Gutknecht (2020) recently provided an initial assessment of the Swiss TA with Elsevier. Although the first year of the agreement had not yet expired, he found that the APC quota included would hardly be exhausted before the end of the year. On the other hand, there were many articles from the institutions involved that were not published via the agreement and articles with a 'Swiss' co-authorbut not a corresponding authorthat were therefore excluded from the agreement. Based on this analysis, Herb (2020) compared the agreement between Elsevier and the Swiss consortium with the TA that Projekt DEAL reached with Springer Nature and Wiley.
An additional source of concern is free riders. If more institutions shift towards OA embedded in TAs, the share of OA will dramatically increase. As a result, less research-oriented institutions will be tempted to cancel their subscriptions (Esposito, 2018). In the case of consortia, this behaviour may unbalance internal agreements for cost distribution. The Swedish Bibsam consortium has already produced a report on the topic that envisions three possible scenarios (van der Vooren, 2019).
The COVID-19 pandemic has introduced more uncertainty into this evolving landscape. The need for better and faster research has given a new boost for OA. However, research institutions face economic constraints, with projected library budget cuts of up to 40% (RLUK, 2020) that will require flexibility from publishers (ICOLC, 2020), no price increases (Spanish Board of Consortia and Purchase Groups, 2020), or fee reductions that Jisc aims to set at 25% (Jisc, 2020a (now referred to as Subscribe-to-Open); Libraria; publish-andread agreements; read-and-publish agreements; SCOAP3; and Subscribe to Open developed by non-profit publisher Annual Reviews. Some of these models refer to specific projects, such as Knowledge Unlatched (https://knowledgeunlatched.org/ku-journals/), Libraria (http://libraria.cc), or SCOAP3 (https://scoap3.org). Similarly, although other publishers can apply it, Subscribe to Open is a model developed by a particular non-profit publisher, Annual Reviews. In another, simpler classification, Hinchliffe (2020) distinguishes four categories of OA publishing models: transformative agreements; pure publish agreements; subscribe to open; and membership models.
Finally, a Delphi study commissioned by the European University Association (van Barneveld-Biesma et al., 2020) concludes that current 'read and publish' or 'publish and read' contracts are just an 'intermediary phase' on the way towards a different scholarly publishing market based on OA platforms that are either publisher-or community-owned.
To some extent, 'transformative agreement' has become an umbrella term that encompasses a variety of deals. Our purpose is to shed light on the main features of these agreements. In this study, we focus on TAs between traditional publishers and libraries and consortia, that is, what is commonly known as 'read and publish' and 'publish and read' agreements. More specifically, we analyse the features of the agreements listed in the ESAC Transformative Agreement Registry. The research is underpinned by three questions: • What types of TAs are in the marketplace?
• What are their features?
• To what extent do TAs pave the way towards OA?

METHODS
On 25 April 2020, the ESAC Registry contained 98 transformative agreements, although 21 had expired before 31 December 2019, leaving 77 transformative agreements active at the time of data collection. Since we aimed to analyse the content of the agreements, we were forced to remove 40 agreements that had not been disclosed and 1 agreement only available in Hungarian. The remaining 36 agreements are the ones analysed in this article, which are listed in the Annex to this paper. This approach has three limitations. First, subscribing libraries voluntarily register their agreements in ESAC, so this source does not necessarily provide a full picture of the population of TAs that are currently active. Second, ESAC describes the fundamental guidelines of a TA, but it does not check whether the agreements registered fulfil these requirements. Third, although disclosure is one of the features required of a TA for it to be acknowledged as such, the full text of many agreements was not available. Despite these limitations, the number of agreements (36) and the diversity of publishers and libraries involved (see Table 1) are sufficient to suggest that they provide a reliable picture of the landscape of TA in the marketplace.
The 77 agreements initially identified in ESAC had been signed by 23 publishers. There was at least one agreement disclosed for 19 of these publishers. The initial 77 agreements covered a total of 17 different countries. There were full-text agreements available for nine of these countries.

RESULTS
As noted above, 'transformative agreement' is an umbrella term that encompasses different kinds of agreements. Based on the analysis of the 36 TAs disclosed in the ESAC registry, we distinguished three categories of TA. The classification was determined by several features: how agreements grant OA publication (discounts, vouchers, average APC rates, or unlimited publication), how the cost is balanced between read-and-publish fees, and whether they include OA publication only in hybrid journals or in both hybrid and gold journals ( Table 2).
The boundaries between the different types of agreements are blurred. Rather than closed categories, there is a continuum of contract types ranging from traditional subscription licences that grant APC discounts or vouchers for a certain number of articles to unlimited OA publishing agreements, with a wide range of options in between. In order to set limits within this continuum, we distinguished between pre-transformative, partially transformative, and fully transformative agreements.

Pre-transformative agreements
In our proposed list of categories, pre-transformative agreements are traditional subscription licences which, besides granting online access to the publisher's bundle of journals, make a provision for OA publication. In addition to access to paywalled content, publishers grant subscribers either APC discounts for OA publication or a set of vouchers for the OA publication of a limited set of articles.

APC discounts
In the introduction, we described examples of TAssuch as those of HEAL-Link in Greece and those of the Biblioteca do Conhecimento Online (b-on) in Portugalthat grant subscribers discounts on APC rates for OA publication. We have not found many examples of this kind of contract in our sample, although a similar clause is present in the agreement between VSNU Netherlands and the Royal Society of Chemistry, which offers a number of vouchers to some subscribing institutions, while others are granted a 15% discount on APC rates. APC waivers and discounts are relatively common through institutional membership programmes or to individual authorsto reviewers for instance but may not be formalized in subscription agreements. It is also possible that some libraries do not register these agreements in ESAC because it may seem doubtful whether this practice can be qualified as 'transformative'.
In the agreement between VSNU Netherlands and the Royal Society of Chemistry, only corresponding authors affiliated with the subscribing universities qualify as eligible authors to publish the fixed amount of OA articles. The same criterion is employed throughout all the agreements, which always use nearly identical definitions of eligible authors.

OA vouchers
Some TAs, in addition to setting conditions for online access to the publisher's journals, grant the subscriber a set of vouchers for OA publication. In most cases, the number of vouchers is relatively smalla few dozensuggesting that they only cover a fraction of the subscriber's scholarly output. In other cases, however, the number is much larger, and for small institutions or specialized publishers, they may cover the OA publication of a large share of the subscriber's output. This is the case, for example, of the agreement between EISZ Hungary and the American Chemical Society, which grants the annual OA publication of 61 articles. Similarly, the agreement between VSNU Netherlands and Emerald grants the annual OA publication of 55-57 articles, and the agreement between VSNU Netherlands and Karger includes the OA publication of 125 articles (AE10%). On a different scale, the agreement between EISZ Hungary and Elsevier provides for the yearly OA publication of 1,000 articles in both hybrid and gold OA journals.
In these agreements, the whole cost of the contract is intended to pay for access to the publisher's journals, whereas discounts or vouchers are not assigned a specific cost. Obviously, the fact that the price is not disaggregated does not mean that the total cost has not increased with respect to previous licences to cover the additional OA publication option.

Partially transformative agreements
We classify as partially transformative those agreements whose cost differentiates between a read fee and a publish fee to cover the APCs of a certain number of articles. Compared to pretransformative agreements, based on discounts or vouchers,  for what seems to be a relatively large share of the output of the subscribing institution. The difference between pretransformative and partially transformative agreements lies in the fact that the former include discounts or vouchers in addition to a subscribing licence, whereas the latter set an average APC rate for the OA publication of a certain number of articles specified in the agreement.
The distinction between pre-transformative and partially transformative agreements is to some extent artificial. However, the inclusion of OA publishing fees in partially transformative agreements suggests a transition towards a new kind of contract focused on OA publication. The distinction between partially and fully transformative agreements is also vague. Multi-year contracts progressively increase the publish fee against the read fee, suggesting that they will end up being fully transformative agreements intended to cover the OA publication of the complete scholarly output of the subscribing institution. Most contracts that grant the OA publication of a limited number of articles do not state what share of the scholarly output of the institution is expected to be published OA under the agreement. Therefore, some agreements classified here as partially transformative could become fully transformative if the number of articles in the quota approaches 100% of the institutional output.
Finally, another feature that merits some comment is the fact that some agreements restrict OA publication to hybrid journals, whereas others include OA publication in both hybrid and gold journals.

Balance between read and publish fees
In the agreement between VSNU Netherlands and Oxford University Press, the read fee represents 88% of the cost. Similarly, in the agreement between Bibsam Sweden and Oxford University Press, the read fee represents 77% of the total cost (a figure based on the cost of the previous subscription licence), while the remaining 23% corresponds to the publish fee (a figure also based on previous APC expenditure). In some cases, there is a nearly perfect balance between read and publish fees. This is the case of the agreement between EISZ Hungary and Wiley, which sets the read fee at 52% of the cost in the first year of the agreement, declining to 47% 3 years later.
Similarly, the agreement between Unit Norway and Wiley has an initial read fee that represents 70% of the total cost but gets balanced (50%-50%) between read and OA publishing in hybrid and

Fully transformative agreements
We define as fully transformative those agreements that cover unlimited OA publication of the whole scholarly output of the subscribing institution. In some cases, the agreement includes a read fee, while in other cases, it does not. As mentioned above, it is hard to mark a clear distinction between partially transformative agreements where most of the cost is assigned to the publish fee and fully transformative agreements that include a read fee.
This category includes the agreements signed by DEAL Germany with Springer Nature and Wiley. The former is described as 'publish and read' and covers the publication of the whole output of DEAL institutions. The cost for the initial year is calculated by setting an expected output of 9,500 articles at an APC rate of 2,750 euros (917 euros for non-research articles), making for a total amount of 26.1 million euros. In the following years, publication fees are calculated based on the actual number of articles published, although the APC expenditure varies within percentage limits set in the agreement. The agreement does not include any access fee, although there is a fee for 'journal archive collections'. Compared to all other Springer Nature agreements (which are classified as 'partially transformative'), in this case, there is not a number of articles to be published each year but just an initial figure that seems to have been established simply to calculate the cost.
DEAL Germany's agreement with Wiley is similar to the one with Springer Nature. Again, the initial cost is calculated by setting an expected output of 9,500 articles at an APC rate of 2,750 euros. In the following years, publication fees are calculated based on the actual number of articles published. In addition, the subscribers pay a 'consolidated access fee' of two million euros.
The agreements signed by Sage with Unit Norway and Bibsam Sweden grant unlimited publication in hybrid journals and a 20% discount for publication in gold journals, in addition to access to the publisher's journals. Meanwhile, the agreement between Bibsam Sweden and Elsevier grants OA publication of an unlimited number of articles in both hybrid and gold journals.
Numerous agreements signed by Jisc with learned societies and small-and medium-size publishers adopt this strategy and provide for the publication of an unlimited number of articles ( Table 2). In some cases, such as the agreement with the Association for Computer Machinery, the contract provides for a decrease in access fees for institutions that wish to access content but do not publish.
IWA Publishing's agreement with the Universities of Delft and Wageningen provides for the publication of articles submitted by corresponding authors affiliated with both institutions in the 14 journals of the publisher, 3 of which are OA. Although the total cost of the agreement is not disclosed, it is based on half the historic subscription cost plus half of the historic APC expenditure. The expected change in the annual article output is predicted based on the figures for the three preceding years.

Additional features of TAs
The option to publish an OA article in the framework of a TA depends on the affiliation of the corresponding author. Some agreements stipulate that publishers should make efforts to inform potential authors about the details of the arrangements. In turn, authors should identify themselves as possible beneficiaries, generally by using an institutional e-mail address when submitting their manuscripts. Subscribing institutions are usually requested to validate this information. Although some agreements mention the possibility of using ORCID for these purposes, the wording suggests that this is a desirable future improvement rather than a current course of action.
In addition to defining 'eligible authors', some TAs also list the types of 'eligible articles' that can be published OA under the agreement. Most agreements provide for the publication of original articles and reviews, but other types of articles may be excluded from the contract. For instance, the VSNU Netherlands agreement with Emerald does not include 'brief communications', 'continuing education', 'case reports', 'letters to the editor', or 'invited letters'.
In managerial terms, publishers have a commitment to report to subscribersusually on a monthly or quarterly basishow many articles have been published, names of authors, affiliations, titles of articles and journals, DOIs, etc. If an agreement is based on an APC fund, the publisher must also inform the subscriber when the fund is due to run out. Frequently, the agreements also state that publishers should make efforts to upload metadata to Crossref and expose articles to discovery services.
One of the characteristics of a TA to be recognized as such is the fact that authors retain copyright, that is, full use and reuse rights of their work. The right of authors to deposit a copy of a published article in a repository depends on the journal policy, the copyright transfer, or the licence under which the article has been published. Nevertheless, some agreements make specific provisions regarding the green OA route. For instance, the agreement between VSNU Netherlands and the American Chemical Society states that the publisher agrees 'to the supply of the final published version of the article to a repository specified by the author's funding agency'. The agreement between VSNU Netherlands and Emerald indicates that 'there is a zero embargo policy across all journals to support Emerald's Green Open Access policy'. In other cases, there is mention of a specific repository to which the article is uploaded. The agreement between EISZ Hungary and the American Chemical Society indicates that 'ACS will deposit the final version of record under a CC-BY license into the US Pub Med Central repository' (Agreement 5 in annex,p. 22). Similarly, the agreement between VSNU Netherlands and Karger states that 'the publisher will provide articles to PubMed Central'.
Finally, most agreements stipulate the editorial independence of the publishers and state that they are not obligated to publish an article submitted by an eligible author.

DISCUSSION
The transition from print to electronic journals has occurred in the background of two independent trends: big deals and OA. Although they began independently, they are now developing 'hand-in-hand' (LIBER, 2017). As a result, the negotiation of new licensing agreements has shifted its focus from cost containment towards the inclusion of clauses in favour of OA.
Since the publication of the Finch Report (2012), there is a consensus on the idea that the main barrier to reaching full OA is the management of a transition period that should introduce adjustments in the scholarly communication market to reach a 'new normal' where OA is the rule. TAs should be seen as part of this 'intermediary phase on the way to a different scholarly publishing marketnot as an endpoint' (van Barneveld-Biesma et al., 2020).
ESAC defines TAs as agreements that meet five criteria: they are temporary and transitional; authors retain copyright; they are transparent; they aim to constrain costs of scholarly communication and foster equity in scholarly publishing; and they govern service and workflow requirements for publishers to ensure that the needs of authors and administrators are addressed. To what extent do the agreements currently in the marketplace conform to these definitions?

TAs are temporary and transitional
TAs are designed for the transition period from journal subscription licences towards OA publication. In detailing the requirements for negotiating TAs, Jisc (2020b) insists that publishers must 'demonstrate a commitment to an Open Access transition through the conversion of subscription expenditure to support immediate open access publication'. Actually, Plan S already imposes a deadline to the transition by setting 2024 as the last year that funding will be provided to support publication fees of journals covered by TAs. Afterwards, any agreements with hybrid journals will not be transformative under Plan S terms.
Hinchliffe (2019) reminds us that, at that time, 'the same contract might be listed as transformative by ESAC but considered not transformative by cOAlition S'.
Despite the diversity of TAs, their various features all suggest a progression towards OA. We use the term 'pre-transformative' here to describe licences that include an OA publication option in the form of discounts or vouchers. It is doubtful whether these agreements can be classified as 'transformative' since discounts are also offered to institutions that do not sign TAs. Springer Nature, for instance, is an example of a publisher that uses both market approaches simultaneously. Its TA option is 'Springer Compact'that is, the kind of agreement signed with DEAL or Jiscwhereas traditional subscribers are offered 'Open Membership' deals, which grant APC discounts. A second stage in the evolution towards OA is represented by those agreements that differentiate between a read fee and a publish fee. The gradual increase of the publish fee against the read fee suggests a progression towards fully transformative agreements, intended to cover OA publication of the complete scholarly output of the subscribing institution or consortium.
The diverse nature of TAs itself supports the conclusion that  (2020), which asserts that, to be truly transformative, these agreements must include two conditions that are not currently met: '(1) guarantee the full transition to 100% OA within a defined, short timeframe and (2) guarantee that the process cannot be easily reversed or cancelled at the end of the contractual period'.
Some agreements make explicit, usually in their preamble, the intention of subscribers and publishers to pilot a new type of contract that should facilitate the transition towards OA. However, in order to guarantee this transitory nature, it would be useful to set measurable objectives to assess this progression towards OA within a specified temporal framework.

Authors retain copyright
All TAs in the sample allow authors to retain copyright. Most OA journals, whether gold or hybrid, publish articles under Creative Commons licences. This issue does not seem to raise much concern since TAs do not have an impact on the publishers' workflow. However, authors may opt out of retaining copyright (Gutknecht, 2020). TAs should guarantee copyright retention by authors, and subscriber institutions should consider the advisability of centrally defining the type of licence for all articles published under the agreement.

TAs must be transparent
A requirement for a TA to be recognized as such is that its content be disclosed. However, as shown in Table 1, the full text for only 37 of the 77 active agreements listed in ESAC was available at the time of data collection. It is possible, however, that the content of some agreements has been disclosed at a later stage.
In some consortium agreements, the distribution of payments among individual institutions is crossed out, but the total cost is always visible. The number of agreements whose content is made public represents an important step forward compared to the traditional secrecy of journal subscription licences, but it is necessary that TAs be systematically open and accessible.

TAs must constrain the costs of scholarly communication
It is hard to determine whether TAs are cost neutral, that is, not adding costs to previous investments in scholarly communication.
Although the costs of TAs have been disclosed, it is not known how much institutions were previously paying for journal licences, let alone APCs.
In recent years, two elements have increased the amounts that institutions invest in scholarly communication: annual increases in the cost of journals, constantly above the inflation rate, and the introduction of APCs. If the expression 'cost neutral' excludes APCs, there is little evidence that TAs are constraining costs. In fact, it is quite possible that some institutions are signing TAs to ensure a contained increase in APC rates in the short term. There is evidence of APC hyperinflation, with publication fees increasing at a rate three times higher than what would be expected according to inflation (Khoo, 2019).
The addition of APCs to their expenditures is a source of tension for many libraries. Even if the cost of subscription licences declines, it will be challenging for libraries to cover this additional cost because APCs have, until now, been paid mostly with researchers' grants. A possible solution to this problem is to partner with authors in paying publication fees. At the University of California, the library contributes 1,000 dollars per article, and grant-funded authors are requested to pay the remaining portion of the APC. The library covers the APC in full for unfunded authors (University of California, 2020). Tennant (2019) points out that, despite their apparent benefits, TAs mean that a lot of public resources continue to be paid to private companies in exchange for publishing. If commercial publishers are making profits at around one-third of their income, this means that one-third of the 24-26 million euros paid by DEAL for German TAs are going directly to publishers' profits. In the long term, this may lead to more tensions as there are already studies warning of APC hyperinflation that call to mind the serials crisis in subscription prices. Similarly, Crotty (2019) believes that the APC model for OA is 'an evolutionary dead end' since it improves access for readers but shifts the inequity of the system onto authors. Further studies by subscribing institutions will be necessary to determine to what extent TAs contain costs compared to current expenditure on journal licences and APCs.

TAs should address the needs of authors and administrators
TAs have an impact on how libraries work with researchers and publishers to transition towards OA. In order to facilitate workflows, most agreements detail how eligible authors are identified and verified. It is probably too early to determine the amount of resources necessary to implement the everyday routines associated with TAs, including invoicing and reporting.
According to the assessment of their offset agreements produced by Jisc (Lawson, 2019), administration costs are hard to calculate but appear to make up less than 1% of the total cost of publishing. Most TAs foresee that publishers will provide monthly or quarterly reports to their subscribers with detailed information about the number of articles published, their authors, APCs, etc.
In the medium term, it may be necessary to normalize how publishers provide this information to libraries in a manner similar to how Project COUNTER standardized the provision of usage statistics of electronic resources in library settings.
Based on its experience, the Vienna University Library has already reviewed the benefits and pitfalls of various systems in place and has warned that 'trying to build an OA workflow by adjusting current subscription-based methods will not yield satisfactory results' (Pinhasi, Blechl, Kromp, & Schubert, 2018). Schönfelder (2020) has suggested the need for a new kind of TA since the current models have 'unfavorable features regarding coordination costs, disruptive workflow adjustments, and timing'.
Finally, Machovec (2020) has reviewed a variety of tools to track authors and APCs and provide underlying analytics.

CONCLUSION
Negotiations between libraries and publishers have shifted their focus from cost containment towards the inclusion of OA clauses.
Despite the diversity of TAs in the marketplace, contracts are increasingly transparent, allow authors to retain copyright, and make provisions to facilitate OA workflows. However, it is hard to assess whether they only represent a temporary phase in the path towards OA or will perpetuate the current state of affairs in the scholarly communication market with its associated high costs.
The big deal model was based on 'previous expenditure' in print subscriptions. The concept was clearly inadequate because the cost of digital acquisitions was calculated based on the size of print collections. This gave rise to substantial differences in digital subscription prices for institutions with different previous print collections. However, it has been impossible to change this pricing model because no one wants to pay more if that is what a new model entails. Something similar may be happening with TAs. Earney (2018) has highlighted the need to define an 'acceptable offsetting agreement', particularly in relation to aspects that demonstrate it is supporting the transition to OA. This should mean 'the reduction of the 'reading' element of an agreement and ultimately, the removal of historical expenditure as the basis of pricing for journal agreements'.
The landscape of scholarly communication is characterized by increasing costs and limited access to research output. It is doubtful whether TAs are helping to solve the cost problem, although they may represent a step forward in removing access restrictions on scholarly information. The main concern lies in its supposed transitoriness. Although TAs should establish a time horizon for the transition towards OA, there are serious doubts about whether they are actually doing so. As long as this requirement is met, TAs can make a substantial contribution to changing the scholarly communication model from journal licence subscription towards OA.