A study of institutional spending on open access publication fees in Germany

General Background

The rise of open access journals has been matched by the increasing relevance of publication fees in academic publishing (Davis & Walters, 2011; Laakso & Björk, 2012; Pinfield, 2015). To cover these fees, also referred to as article-processing charges (APCs), authors tend to make use of funding that grant agencies or academic institutions provide (Suber, 2012). However, the question of how and to what extent these research support activities are effective in terms of the number of supported articles and their associated costs remains under debate.

The study of institutional spending on open access journal articles has been limited for several reasons. The first is that payment of these charges is fragmented across the budgets of grant agencies, research institutions, and libraries, or is covered by personal budgets. A comprehensive 2010 survey asked 9,645 authors from various disciplines how they financed publication fees, and it revealed that the majority of the respondents had access to research funding or institutional support to cover these charges. By contrast, 12% paid publication fees individually (Dallmeier-Tiessen et al., 2011). These results are consistent with similar findings from other studies: Previous studies also found that funding sources exist primarily in higher-income countries, mainly to support research articles in the biological and physical sciences (Solomon & Björk, 2011). Personal budgets, however, are likely used to cover lower publication fees (Björk, 2015; Solomon & Björk, 2011).

Another key problem in this regard is that funding for open access journals using publication fees lacks transparency because the parties involved—authors, universities, funders, and publishers—do not release information about who pays for what or the costs of publishing (Björk & Solomon, 2014); this situation is similar to the lack of transparency regarding journal subscriptions (Lawson & Meghreblian, 2015). To date, empirical studies examining publication fees have obtained price estimates by surveying authors (Dallmeier-Tiessen et al., 2011) or obtained them from journal websites. Using the latter method, two studies investigating journals across a broad range of disciplines calculated similar price averages that ranged between $904 (Solomon & Björk, 2012) and $923 (Walters & Linvill, 2011), as well as considerable price variation across journals and publishers. Accordingly, Solomon & Björk (2012) suggested using publication fees to cluster fully open access journal into several groups. In descending order, these are high-impact journals, followed by biomedicine journals from commercial publishers, large multi-disciplinary journals, and mid-price journals from commercial publishers covering a large spectrum of disciplines. Lower-priced journals are those published by academic societies and by publishers from low-income countries.

Nevertheless, it remains unclear which factors contribute to pricing in academic publishing. Generally, these might include article processing, impact, rejection rates, management and investment, and profit margins (Noorden, 2013). While fixed prices for individual articles are common, agreements between publishers and institutions can lead to discounts, and publishers sometimes waive publication fees for authors from low-income countries (Björk & Solomon, 2012; Lawson, 2015c). Other factors leading to variable pricing schemes include submission or page charges (Björk & Solomon, 2012).

Hybrid journals substantially add to the complexity of open access funding (Björk & Solomon, 2014; Kingsley, 2014; Pinfield, Salter & Bath, 2015). These journals, which allow articles to be published immediately as open access after a charge is paid, rely on both subscriptions and publication fees as revenue sources. Although the uptake of open access through hybrid journals was described as lower and more expensive than that of fully open access journals (Björk & Solomon, 2014; Solomon & Björk, 2012), this model has gained increasing attention because of recent open access science policies, notably from the UK (Pinfield, 2015).

To address the problems of fragmented spending on publication fees and the lack of transparency about what is being paid, some European research funders and research-performing institutions have recently begun to disclose their expenditures for publication as open data. To the best of our knowledge, the first research funders to provide such data were the Wellcome Trust (Kiley, 2014) and the Austrian Science Fund FWF (Reckling & Kenzian, 2014). The not-for-profit company Jisc followed this example by collecting data from UK universities (Lawson, 2015b). Disclosed as publicly available spreadsheets, these data-sets self-report expenditures along with bibliographic information, including title, journal and publisher, and a persistent identifier for the publisher’s version. Curatorial efforts focused on the disambiguation of publisher and journal titles as well as on detecting duplicates and persistent identifiers for the full text including the Digital Object Identifier (DOI) (Neylon, 2014; Woodward & Henderson, 2014). A preliminary version of Jisc’s cost data was examined by Pinfield, Salter & Bath (2015). Although the average spending on publication fees remained stable across universities, the authors found large price variations, as well as a varying number of articles supported by UK universities between 2007 and 2014, findings which confirm earlier studies that collected price information from journal websites (Solomon & Björk, 2012).

Central funding for publication fees in Germany

This paper focuses on how much German universities and research organisations spend on open access publication fees. In Germany, the Deutsche Forschungsgemeinschaft (DFG), the largest German research funder, has strongly influenced how universities manage institutional support for these charges. Before the DFG started to support centrally funded publication fees through its “Open-Access Publishing” programme in 2011, only a few central funds existed (Eppelin et al., 2012). This is similar to the situation described in Canada (Hampson, 2014) and the UK (Pinfield & Middleton, 2012). The DFG has enforced a set of criteria with which grantees have to comply and which has resulted in similar policies regarding support for publication fees across German universities (Fournier & Weihberg, 2013). These criteria exclude the sponsorship of articles in hybrid journals and the funding of articles for which the publication fee exceeds € 2,000.¹ Grantees agree not only to pay for APCs, but also to find ways to improve the handling of those financial transactions. These ways include central invoicing schemes and memberships that are agreed upon by university libraries and publishers and that often lead to a discount on publication fees (Fournier & Weihberg, 2013).

Non-university research organisations, i.e., institutes organised in the Fraunhofer-Gesellschaft, Helmholtz-Gemeinschaft, Leibniz-Gemeinschaft, and Max-Planck-Gesellschaft, are not eligible for this DFG funding programme. However, in response, some organisations have adopted similar processes to support authors. The Max-Planck-Gesellschaft operates its long-standing open access activities, including handling spending and publisher agreements centrally, through the Max Planck Digital Library (MPDL) (Schimmer, Geschuhn & Palzenberger, 2013; Sikora & Geschuhn, 2015), while the Leibniz-Gemeinschaft set up a dedicated open access fund in 2016.

The evolving institutional support for covering open access publication fees has led to calls for a unified approach towards open access funding in Germany. The MPDL called for re-allocating subscriptions in favour of open access journals in 2015 (Schimmer, Geschuhn & Vogler, 2015). At the same time, the Allianz der Wissenschaftsorganisationen, a science policy board representing all major research organisations in Germany, marked price transparency as one way to sustain an “adequate open access publication system” (Bruch et al., 2015). Reflecting Austrian and UK initiatives to share institutional spending on open access publication fees as open data, as well as professional discussions on open access publishing, Bielefeld University Library began to openly share its payment of publication fees in May 2014. After engaging with the working group “Electronic Publishing” of the Deutsche Initiative für Netzwerkinformation (DINI), other German institutions joined under the umbrella of the Open APC initiative soon after (Apel et al., 2014–2016).

Research question

The aim of this study was to examine how much German universities and research organisations spent on open access publication fees. Using self-reported cost data from the Open APC initiative, the analysis focused on the amount that was being spent on publication fees, and compared these expenditure with data from related Austrian and UK initiatives in terms of both size and the proportion of articles being published in fully and hybrid open access journals. We also investigated how thoroughly self-reported articles were indexed in Crossref, a DOI minting agency for scholarly literature, and analysed how the institutional spending was distributed across publishers and journal titles.

Cost Data

After excluding payments for non-journal articles as well as articles invoiced in 2016, we retrieved information on 7,417 open access journal articles that 30 German universities and research institutions supported financially between 2005 and 2015. As illustrated in Fig. 1, payments made for open access journal articles increased over the years. While one institution supported five articles in 2005, most institutions included in our study shared their expenditure from 2013 onwards. The best represented year was 2015, with 1,999 articles. However, at the time of analysis only 27 institutions had contributed cost information for 2015, suggesting a lag between the time that payments are made and expenditures are reported to the Open APC initiative.

The fees for all of the articles amounted to € 9,627,537, including value-added tax; the average payment was € 1,298 (median = € 1,231, SD = € 486). Figure 2 presents the distribution of institutional spending on publications. We observed that 6,996 (94%) of the publication fees were paid in accordance with the DFG price cap of € 2,000. Most payments for publications ranged from € 1,000 to € 1,250.

Figure 3 presents institutional spending per article and year. Large price variations can be observed. Publication fees that were paid by German universities and research organisations ranged from € 40 to € 7,419. However, the average price paid varied somewhat during the period from 2011 to 2015 (€ 1,239–€ 1,423).

The number of APC payments per institution varied considerably (see Table 1). With 2,856 reported articles, the Max Planck Society contributed 39% of the overall article volume. By contrast, we observed a lower number of supported open access journal articles for several universities that had only recently begun to set up centrally managed open access funds to cover publication fees.

Comparison of related cost data-sets

Table 2 compares the Open APC spending data with that of the Austrian FWF, as well as with Jisc’s and Wellcome Trust’s expenditures. Prices were converted according to the average Euro exchange rate during the examined periods and were gathered for both fully open access journals and hybrid journals. The comparison revealed that the Open APC initiative lacked cost information about hybrid journals, whereas the related Austrian and UK open data initiatives reported a large share of spending on these journals between 2014 and 2015. Over the years 2005–2015, three out of 30 German universities and research institutions reported 60 hybrid journal articles to the Open APC initiative, representing 0.81% of all articles included in the data-set. In contrast, in terms of the number of supported articles and the amount spent on publication fees, the Open APC data-set provided more comprehensive price information for fully open access journals than did the Austrian and UK initiatives.

A comparison of average prices revealed that publishing in hybrid journals was, on average, more expensive than publishing in fully open access journals. Price differences between these two categories were also reported earlier, indicating that prices for fully open access journals were lower on average (Pinfield, Salter & Bath, 2015; Solomon & Björk, 2012). In 2014 and 2015, the mean price for fully open access journals calculated from all data-sets was below the DFG price cap of € 2,000.

Crossref indexing

To identify publication fee spending at article level and to gather bibliographic metadata, DOIs were a mandatory part of the Open APC initiative’s data collection activities. The participating institutions reported DOIs for 7,373 out of 7,417 articles. Using these DOIs, we retrieved additional metadata from Crossref for 7,346 publications, representing 99% of the total article volume. Articles for which no metadata could be obtained, were registered with the DOI agency DataCite (ten articles) or Medra (two articles). For eight articles, our parser could not gather the XML resource, although these publications were registered with Crossref at the time of our study. Seven DOIs reported to the Open APC initiative could not be resolved.

Cost data by publisher and journal

We used the DOI to automatically fetch publisher and journal names for each article from the Crossref REST API. Table 3 shows the top ten publishers in terms of the number of financially supported articles. These publishers represented 92% of all articles included in our data-set. In total, payments were made to 139 publishing houses. Comparing these data with data from the UK, we observed that a greater share of total spending was directed to some open access publishers. Pinfield, Salter & Bath (2015), for instance, reported remarkably lower proportions for the open access publishers MPDI, Copernicus, and Hindawi.

Most of the publication fee spending in Germany was on articles published in Springer Nature journals, which likely reflects the results of mergers with the open access publisher BioMed Central in 2008 and between the well-established publishers Springer Science + Business Media and Nature Publishing Group in 2015. Using the Crossref-Member-ID instead of the publisher name, we were able to differentiate between journals formerly published by Springer Science + Business Media and Nature Publishing Group. In terms of articles, the majority of payments made were for publications in journals formerly associated with Springer Science + Business Media. Springer Science + Business Media journals accounted for 2,027 articles, representing 94% of the overall Springer Nature article volume recorded by the Open APC initiative and 92% of the amount that was spent. Median publication fee spending differed slightly between Springer Science + Business Media (€ 1,355 €) and Nature Publishing (€ 1,386). However, the price variation was higher for Nature Publishing journals (SD = € 848) than for the former Springer Science + Business Media titles (SD = € 313).

In contrast to Springer Nature, other well-established publishing houses such as Elsevier and Wiley-Blackwell ranked lower in our analysis.

Prices also varied within journals. Table 4 illustrates the top ten out of 732 journals based on the number of supported articles. We normalised PLOS journal titles because the name change from “PLoS” to “PLOS” was only partially represented in the Crossref metadata at the time of our study. Articles published in the top ten journals represented 45% of the overall article volume. The multidisciplinary journal PLOS ONE ranked highest. In addition, the journals New Journal of Physics, Atmospheric Chemistry and Physics Discussions and Frontiers in Psychology, all of which publish contributions from all branches of their respective discipline, were also well represented in our study. In the case of Atmospheric Chemistry and Physics Discussions, the large price range can be explained by the fact that this journal charges per page and takes the submission’s file format into consideration.