PREreview of Estimating global article processing charges paid to six publishers for open access between 2019 and 2023
- Published
- License
- CC BY 4.0
This study tackles the important but under-investigated question “What is the total global cost of article processing charges (APCs) for open-access publishing?” Using a curated dataset of Article Processing list prices from major publishers and OpenAlex data, the study estimated that global APC spending over 2019-2024 averaged around 8.9 billion USD (prices based on 2023 economics).
Major issues
Introduction: “One increasingly popular approach to OA…”
Inaccurate: APC payment has been the dominant model of open access for a long time. Support the claim of it increasing over time with a citation.
Section 2.3: “$3,496” estimated value
It may be worth explicitly noting that this means that errors in the article counts for earlier years will have more of an effect on the final (monetary) values than later years, since the APC has been adjusted upwards for inflation.
Align result & claim: “Our study shows that the APC model continues to grow”
I don’t think this is correct, or at least not precise enough. What you have shown is that APCs continue to generate ever-increasing amounts of revenue. I’d interpret the conclusion as stated as the APC model having an increasing market share in terms of either revenue or number of published articles, neither of which are studied here.
Overestimated results and uncertainty
Given that waivers and discounts were not accounted for, the values presented here are still likely to be overestimated. The conclusions around the % of revenue growth seem reasonable, but the absolute values still have a large degree of uncertainty.
Definition & operationalization of APC construct
Throughout the text, there is a lack of clarity about whether the payments being referred to are individual article-processing charges or may be part of read-and-publish or transitional agreements, membership schemes, Diamond OA schemes, or other ways that OA may be paid for. These make the calculations of APC charged per article, and in total, much less reliable (in general, a one-off APC is charged at a higher rate than any option that is “bulk buy” by a payer). For example, “publishers, such as Frontiers and MDPI, rely solely on APCs to generate revenue” - I think it means that they rely wholly on OA fees, not on APC payments per se.
Analysis: control for article number growth (Fig. 5)
Discussion of the growth in expenditure is discussed without at the same time mentioning how many articles this applies to. Given that the authors do appear to have data on article numbers (e.g. Figure 5), why don’t they separate the effects of article number growth from the effects of price increases?
I agree with this comment. Multiple factors could individually or collectively contribute to the increase in publishers’ APC revenue, including (1) an increase in the total number of articles published, (2) an increase in the percent of articles being published as OA and requiring APC, (3) an increase in APC/article, etc. Thus, it would be useful to separately present data on how these factors changed over the years.
Clarification
Regarding the number of articles published – some of the publishers examined in this paper have seen exponential growth in the number of articles they publish over recent years (Hansen et al 2024). So, is the increase in APC revenue reported in this article beyond what’s expected from increase in the sheer volume of articles being published? Following this line of logic – when the authors talk about this mode of publication being “unsustainable,” is the underlying issue high APC charges, as the authors seem to imply, or exponential growth of scientific publishing in general?
Citation needed
First sentence of Discussion section: “The combination of a growing volume of gold and hybrid articles and increasing fees leads to global spend almost tripling in 5 year”. What are the data to support “growing volume” (apart from the aggregate % increase in Table 3) and “increasing fees”?
Related comment: “It will be useful to see a version of Table 1 where year-wise numbers for article--years combinations are provided for each publisher. Table 3 provides this as percentages. But actual numbers would be better. Also, year-wise APC fee should be provided (with and without adjusting for inflation).
Balanced Literature Review
The end of the Introduction spends time critiquing the approach of previous studies but is silent on the limitations of these authors’ approach in their own prior and current work (Butler et al). A more balanced consideration of the limitations of all these approaches would be welcome.
Minor issues
Connectivity:
Avoid using keywords that are already in the title.
Clarity:
A more structured abstract with questions, backgrounds, methods, results, interpretations, and implications is desired. Consider the schema:
Rationale (1-2 sentences) - why was the research needed?
Objective (1 sentence)- what were you trying to provide to meet that need?
Method(s) (up to 3 sentences) - briefly summarize what and which parameters were measured.
Results (up to 4-5 sentences)- what did you find? Please add some data to demonstrate the findings.
Conclusions/Recommendations (1 sentence)
a lack of transparency in APC prices in the Abstract
define or explain “APC prices” more. If the authors mean the amount decided by the journal as APC, then this is transparent in the journal guidelines. However, they could mean something else like APC paid including full/partial/no waiver
“OA fees” in Abstract
The fees are relatively transparent, but not the APC revenue.
Align with External Data
MDPI’s annual report from 2020 (https://mdpi-res.com/data/2020_web.pdf) states that the average cost to authors per paper was 1180 CHF and they published 115,174 articles and reviews according to their website. This can be used to estimate a revenue of around 136 million CHF. Can the authors compare that value with the result of their calculations? A similar exercise puts APC revenue at 284.7 million CHF for 2021. Unfortunately, they stopped reporting these numbers from 2022.
Open Code Request
I would urge the authors to provide their analysis code in addition to their dataset (already in Harvard Dataverse, a public repository). Especially since the authors state in the last paragraph of the Discussion section, “The result was a robust and transparent set of calculations”
Figures
1-6 seem blurry/low-res with thin typefaces. Increasing the resolution and thickness of the text would help. Maybe even play with increasing the size of the figures overall to increase readability and accessibility.
Fig. 6: Consider using the same y-axis scale for all the subplots in this figure to aid comparison. It may also be valuable to add median APC prices here, as is shown in Fig. 5.
Bar graphs, density plots, and violin plots can be flipped horizontally to enhance the readability of y-axis labels. If using ggplot2, “coord_flip” will be useful. Vertical text can be tough to read. See https://ggplot2.tidyverse.org/reference/coord_flip.html
Suggestions for future studies
OA switchboard would be a very interesting organization to collaborate with if they were willing to share their data.
This article briefly mentions transformative agreements, so a future study on those agreements between publishers and libraries would be interesting .
The authors could suggest potential applications of their findings, such as informing open access policy development or publisher negotiations.
Authors could survey scholars on how estimates of global APCs impact their perception of open-access publishing and publishing plans.
References
Simard et al., 2024 is cited parenthetically in the body, but not fully cited in References.
Competing interests
Theodora Bloom is employed by BMJ Group which publishes both Gold OA and Hybrid journals and charges APCs. Martyn Rittman is employed by Crossref and previously worked for the open-access publisher MDPI in various roles, including roles in production, editorial, and author services.