Over the last few years, development in open access publishing has increasingly led to Gold Open Access based on article-specific publication fees. In many countries, the development has been accelerated by the transition to nationwide transformative open access agreements. Many universities have also created centralised funds to pay article processing charges (APCs). Because of the above, the monitoring of article payments has become pivotal. To be able to outline the total cost of scientific publication, it should be clear what is being paid and to whom. A better and wider knowledge base on the cost of publishing has been identified as an important goal, especially in scientific libraries (see e.g. Lawson, 2015; Pinfield et al., 2016). However, estimating the total cost has proven difficult.

In Finland, 12 universities (86 percent of all universities) followed APCs in 2019 (Vilén et al., 2021). In recent years, the motivation of Finnish universities to monitor the costs of open access publications has been partly influenced by a national policy. One main goal of the policy is to make the total cost of scholarly publishing transparent and publicly available (Open Science Coordination in Finland, Federation of Finnish Learned Societies, 2020).

As the first university library in Finland, Turku University Library started the systematic collection of APCs in co-operation with the university’s financial services in the spring of 2017. Monitoring was organised on two levels: the processing of purchase invoices centrally in the financial services and a more detailed analysis of payments in the library. It is noteworthy that the model is not based on a centralised fund, but invoices are paid from unit-specific budgets. Payments made with credit cards were retroactively included in the data collection.

At the University of Turku (UTU), invoice processing is centralised in the financial services. Financial secretaries identify the APC payments from amongst all invoices received by the university and allocate them to the correct general ledger account. Before the organisation-level monitoring started, the library arranged a training session for financial secretaries on open access publishing and APCs to help them identify the correct invoices.

The library receives a compilation of the APC invoices based on the data of the APC general ledger account as an Excel file three times per year. The original invoices are available as PDF files in the local accounting system where an information specialist has a user account. However, library access is limited to the specific general ledger account only. With authorised access to the APC account, it is possible to view the original invoices more closely and identify the publication, Digital Object Identifier (DOI), journal title, publisher and whether the publication channel is full open access or hybrid. After identification, the information specialist creates an Excel table where the identifying data of the publication and publication channel are recorded. If the fee is not APC but instead, for example, a submission fee, it will be deleted from the report.

In addition, the library checks whether and under what conditions the publisher allows the publication to be self-archived, as well as whether the information about the publication has been deposited in the university’s current research information system UTUCRIS. If not, the publication and APC information are recorded to UTUCRIS, and the full article is self-archived and made openly available via the institutional repository. Finally, the compilation of the APC data is submitted to the international Open APC service (University of Turku APC data available at https://treemaps.openapc.net/apcdata/turku-u/).

Open APC is a service managed by the Bielefeld University Library. Open APC releases data sets on fees paid for open access journal articles by universities and research institutions (Pieper & Broschinski, 2018). In the future, the separate data transfer from Turku University Library to Open APC will become unnecessary because APC data can be transferred from UTUCRIS to the national VIRTA publication service¹ automatically. After this, VIRTA delivers the data received from the Finnish organisations centrally to Open APC.

There have been a few attempts to study the monitoring of article fees by research organisations. The latest review can be found from Andre Bruns and Niels Taubert (2021). They have reviewed Gold Open Access payment data compiled from German universities into the Open APC data collection. In their study, Bruns and Taubert (2021) found two significant problems in tracking publication fees. They call them blind spots. Comprehensive data are not available for organisations, as they do not have a centralised fund to facilitate the monitoring of payments. In these organisations, fees are paid in different units, and collecting centralised data has not been simple. There may also be a complex set of funding sources. In addition, even if an organisation has a centralised fund, monitoring payments can still be challenging, as payments are made by different entities of a research organisation and processed in a variety of ways.

To look at the blind spots they perceived, Bruns and Taubert (2021) used, in their own words, an indirect approach to identify APC-specific publications. They used bibliometric data sources (Web of Science and SSN-Gold-OA-list) for the identification instead of relying on data collected by organisations. As a result, for a total of more than 40 German universities, the authors concluded that the data reported in the Open APC database was notably incomplete. In universities that transferred data to Open APC, more than half of the estimated costs were not covered. However, the reasons for the incompleteness of the data were beyond the scope of their study.

A local knowledge gap in monitoring publication fees was also noticed by Stephen Cramond et al. (2019) a couple of years earlier. In 2018, the Council of Australian University Librarians (CAUL) carried out a project to study the financial impact of APC payments on universities in Australia and New Zealand. One of the main results of the pilot project was how little is actually known about the nature and amounts of publication fees in universities. An important finding was the variance of different sources of funding such as research funders, endowment funds, departmental funds, transfers from research partners or researchers’ personal funds. They also found that only a few universities had a centralised APC fund, and only some of the universities collected data on publication fees at all.

Cramond et al. (2019) reported that, due to the limitations of centrally reported figures, the Australian project decided to look at established APC payments in an indirect way. They evaluated the volumes of open access articles and potential APC payments on the basis of publication data in the Web of Science and Scopus and compared the data obtained with data from Unpaywall. The methodology used was the same as in this article.

APC payments that are paid from different funds and from different sources outside the centralised organisational fund are often referred to as ‘APCs in the wild’ (see e.g. Andrew, 2016; Monaghan et al., 2020). Theo Andrew (2016) noted in a review of the University of Edinburgh’s ‘APCs in the wild’ that there was a significant amount of APC fees, of which the university was unaware. He estimated that ‘APCs in the wild’ could account for up to 20 percent more in the total cost of publication than previously thought.

In Sweden, the monitoring of APC payments in research organisations has also been examined (Gustafsson et al., 2017; Lovén, 2019). Stockholm University Library and Gothenburg University Library have been tracking open access publishing costs since 2016. Sweden has a national centralised service run by the National Library of Sweden to collect APC fees from universities, called Open APC Sweden² (Gustafsson et al., 2017).

In her article, Lisa Lovén (2019) has described the process of collecting fees at Stockholm University. The process within the university works similar as at the University of Turku. However, Stockholm University seems to have a more transparent accounting system, in which all payments are accessible for all authorised users. This makes it much easier to track invoices from the system if they were not initially recorded for the correct APC account. In her review, Lovén (2019) has looked more closely at the publishers to whom the largest amount of APC fees is paid, as well as the sources of funding.

As of 2018, Stockholm University Library managed a significant centralised fund for APC payments (Lovén, 2019). At that time, Lovén (2019) noted that Sweden decided to withdraw from the agreement with Elsevier, and the saved funds were used to finance article fees for full open access journals. Many benefits were identified by the centralised fund, including easier tracking of article fees, greater understanding of the open access publishing landscape and clear benefits for researchers who no longer had to worry about paying article fees themselves.

In a report from the University of Gothenburg, Gustafsson et al. (2017) noted how difficult it was to find out the exact amount of APC payments. In Gothenburg, publication fees for 2016 were analysed. Due to the monitoring problems, two different methods were used in this analysis to identify the payments. First, the authors reviewed the organisation’s invoices in the financial management system, and second, they requested information directly from the publisher. Excluding Elsevier, data were obtained from several significant publishers. Combining these methods, the aim was to estimate the minimum APC costs for the University of Gothenburg in 2016. The authors concluded that at least SEK 5.2 million (approximately EUR 490,000) was paid in 2016 for the openness of articles.

According to the Gustafsson et al. (2017) report, APC payments had been credited with a total of 18 different accounts in the university’s financial management system. APCs paid with a credit card were also a problem in terms of tracking. Additionally, there were major shortcomings in the publishers’ invoices from the perspective of identifying the article.

When it comes to tracking APC fees at the organisational level utilising financial management systems, all the articles and reports above (Bruns & Taubert, 2021; Cramond et al., 2019; Gustafsson et al., 2017; Lovén, 2019) recognise the great difficulty of collecting data. To improve the quality of the data collected by organisations, further consideration needs to be given to the reasons why the coverage of the data is largely poor. To get to the root causes, it is necessary to continue to observe the data collected by organisations and compare it with data from external sources.

The objective of this study was to identify APC-liable publications from bibliographic information sources and to compare the results with centrally collected APC data from the University of Turku. The initial assumption was that the university’s own bookkeeping has managed to collect APCs comprehensively. The analysis is based on comparison of in-house APC reports and harvested University of Turku-affiliated open access articles from the bibliographic information sources Web of Science and Scopus. Unpaywall and DOAJ are used for identifying the open access statuses of articles more closely and to exclude articles from so-called Diamond publication channels, which do not charge APCs. The analysis focuses on open access articles whose corresponding authors are affiliated with the University of Turku or Turku University Hospital and which are not reported in internal APC bookkeeping at the university or hospital.

Publications were included in the analysis based on the institutional affiliation of the corresponding author (Scopus) or the reprint author (Web of Science) to the University of Turku or the Turku University Hospital. Although there is disciplinary variation in determining the roles and responsibilities among authors, including the role of the corresponding or reprint author, they are generally likely to be the principal authors and therefore most likely to be responsible for paying APCs (Cramond et al., 2019; González-Alcaide et al., 2017). The analysis in this study is based on this assumption. As the reprint or corresponding author needs to remain locatable for requests as far to the future as possible, the role is typically assigned to an author with a stable position in the organisation (Weiss, 2012).

However, even though the corresponding author is usually the person responsible for assuring the publication costs, the APC invoice may be distributed to a co-author in another institution or split by several authors (Lovén, 2019). For this study, it was reasonable to assume that the corresponding or reprint author was most likely to be a person responsible for submitting the APC fee, even though there could be no absolute certainty about it. This is one of the limitations of the study.

The first research question of this study was to find out whether it is possible to obtain information on paid APCs by using external information sources. Based on the analysis, and paying attention to certain limitations, it seems possible to harvest and identify APCs from external sources. External data sources have also been utilised previously, for example at Stockholm University where the Web of Science is used to complete the annual institutional APC report by identifying open access articles in which University of Stockholm researchers are identified as corresponding authors (Lovén, 2019).

However, a few limitations must be remembered when using bibliographic information sources. Payments cannot be clearly defined from them. A particular problem is caused by the roles of the corresponding author and publication fee payer. In this study, it has been assumed that the corresponding author is also the payer of the article fee, but in fact, this is not always the case. The paying party cannot be found clearly from any available source other than the organisation’s own monitoring. In addition, when using external sources, it is impossible to find out the payment year of the publication fee. The publication fee may have been paid in a different year than the publication has been published.

The email address used by the author can also cause problems. If the researcher is not using the organisation’s email domain when submitting the article to the journal, it can be challenging to identify the affiliation to a certain organisation. Consequently, articles of this type may be completely excluded from the review, as decided in this study.

Using external sources to monitor publication fees is possible but rather laborious and contains the above-mentioned limitations. The most ideal situation would be if the payments could be identified within the paying organisation. Although the internal monitoring of organisations does not seem very accurate in light of this and previous research, it is possible to develop it.

Although in this study we did not focus on the reasons why the organisation’s own monitoring is so inaccurate, some possible causes can be presented. Currently in Finland, the organisations’ own monitoring is largely based on the manual work of the financial secretaries. They are usually not familiar with open access terminology. If the APC is not clearly mentioned in the invoice, it is possible that it is not registered. Publication fees paid by credit card are particularly difficult because there is very little information on the credit company’s invoice.

The second research question concerned how information retrieved from other information sources compares with the organisation’s internal accounting. Based on information received from Web of Science, Scopus, Unpaywall and DOAJ, it can be assumed that the University of Turku has paid APCs for more than 500 articles published in 2017–2018 (see Figure 5). Most of these articles were indexed in Web of Science (444 articles, 86 percent), and only 34 (7 percent) articles were published in journals that were indexed only in Scopus. In this context, it is important to note that Web of Science was assessed as the primary data source, so if the same article was found in both databases, only Web of Science data were taken for analysis. In the university’s own bookkeeping were also 29 articles that were not indexed by either Web of Science or Scopus.

The third and final research question concerned whether open access APC-liable publications identified from external bibliographic sources share some common characteristics (e.g. a publisher). A significant number of articles (51 percent) reported in the organisation’s own bookkeeping had been published in Springer, Wiley and Elsevier journals (see Figure 3). When comparing the same distribution with publishers of non-reported open access articles, one can observe that, although Springer is the largest of these, MDPI rises second (Figure 2). Both statistics also highlight the large deviation of articles among different publishers, which suggests that in 2017–2018, open access publishing was not centralised for journals of large commercial scientific publishers. In fact, the data from DOAJ demonstrate that there were 32 articles that were published in open access journals that did not charge APCs, so they were removed from the final analysis for that reason.

The final result shows (Figure 5) that the University of Turku in-house bookkeeping has succeeded in registering 52 percent of APC invoices, while 48 percent have not been identified. The results show that the number of unreported publications that have been openly published and whose corresponding author is affiliated with the university is almost equal to those registered in the university’s institutional APC report. Based on the result, it can be assumed that the APC monitoring for the years 2017–2018 has not produced a particularly reliable result, and based on that, the APC statistics collected by the organisation should not be used exclusively as a basis for evaluating the total costs.