European release
European release (Photo credit: Wikipedia)

The Scholarly Kitchen has some of the smartest readers around. That’s not self-congratulatory — you read us, so you must be very smart. It seems to be a fact. Well-educated, well-informed, thoughtful, and insightful comments come in every day, and outnumber posts by 12:1. Often, social media notes things like, “The comments are especially useful.” And comments often lead to new ideas for posts. This time, the path is quite direct.

A number of recent comments streaming out of a post about disagreements in the UK about the role of Gold open access (OA) contributed some interesting quantitative assessments, and generated a lot of discussion outside the blog or the comments. This is encouraging, because these are discussions we need to have, as I recently argued in a lengthy post. Highlighting these is a way of further fostering conversation.

In case you missed these data-rich comments, here they are again. The data we accept and the meaning we attach to it are important to how we think about the past, present, and future. I have added one update.

Speculation About PeerJ

Comment by Mike Taylor:

I have just this moment seen this tweet: speaking at the ALPSP conference, Jason Hoyt of PeerJ has said that they spend $600 per month (i.e. $20 per day) on IT services. That’s astonishingly cheap, given the huge scalability and redundancy they’ve built in.

Comment by Thomas Hilock:

Sadly PeerJ doesn’t pass the sniff test. Not even close.

$600 per month on IT services, wow, that’s great. What a headline. That’ll get a retweet from @miketaylor.

Now let’s talk about Alf Eaton, Patrick McAndrew, Jackie Thai, Dave Harrington, and Janos Toth. Who are they you ask? That’s the operations team that runs the PeerJ. How much do you think they cost a month? Let’s average it at an extremely modest $80k each pa, that’s 400k per year, before you house them in one of PeerJ’s two offices, that need to be rented, lit, supplied, managed. And let’s not talk about benefits, or bonuses, or take into account marketing, or the CEO’s salary, or HR, travel expenses, taxes…

Now, let’s talk about incomes. Peer J so far this year has published 157 articles. Let’s average it at 7 authors each, and they all adopt the highest plan of $299. That’s $330k. All they’ll ever get out of these authors. And they can publish as much as they want. BUT they could have bought the cheapest plan, so revenues could be as low as $110k.

Now I’m not a mathematician, but I can tell you this, if I was their accountant, I would be telling Peter to go look for some more venture capital pretty quick.

(Update: PeerJ recently signed deals with UC Berkeley and the University of Cambridge, which will redirect library funds to cover costs for publishing with PeerJ.)

What Is a Profit Margin?

Comment by Mike Taylor:

“The report regularly calls out the excessive profit margins of publishers, yet mentions only one such publisher, Elsevier, who due to its size likely represents an edge case.”

As most readers will be aware, this is not true. Elsevier’s profit margin — 37.3% for 2011 — is right in the middle of those for the Big Four publishers. Springer took 33.9% in 2010, WIley too 42% and the academic division of Informa too 32.4%. Details here. For comparison, Apple’s best ever reported profit margin was 24%. Exxon makes 6.5%.

Comment from me:

A margin of 6.5% for Exxon is US$16 billion in profits. Elsevier is a $3.3 billion company overall — journals, databases, books, legal, etc, etc. At 1/5 the profit margin, Exxon makes 5x the size of Elsevier in its entirety in profit. Given the common 5x valuation for publications as an assumption, Exxon could buy Elsevier with one year’s profit. These comparisons aren’t informative.

The problem with the BIS report is that it tries to extrapolate from an edge case (Elsevier) to an industry in which 81% of the companies make less than $25 million in revenues per year. Clearly, the vast majority of the market consists of small publishers, most of which are society or academic in nature.

Companies like Elsevier and Wiley comprise 0.4% of the STM market. Using them as examples is clearly misleading.

Comment from steelgraham:

“Companies like Elsevier and Wiley comprise 0.4% of the STM market”. Was that a typo ??

Comment from me:

No, that is true. They have 19% of the revenues, but are only 0.4% of the companies. 81% of the companies make $25 million or less in revenues (i.e., they are very small).

Comment from Mike Taylor:

“Companies like Elsevier and Wiley comprise 0.4% of the STM market.”

I don’t understand how this can be so. Elsevier’s annual revenue is on the order of £2 billion, in a market which the STM report estimates as being worth about £8 billion in total. Even if only half of Elsevier’s revenue is from journals (which seems likely to be a significant underestimate), its £1 billion in journal revenue surely constitutes 12.5% of the market on its own.

Comment from me:

They are 0.4% of the companies, they hold 19% of the revenues.

If you’re looking for representative companies, you can’t look at them as your sole or even primary or even secondary sources.

Comment from David Wojick:

I think if you look further you will find a lot of corporate level expenses that are not allocated to the operating divisions, especially payments to bondholders, which is debt coverage. I once heard it said that after debt service Elsevier’s profit was more like 3% but have never checked this out. Likewise the operating divisions may not pay for the buildings they operate in, but the corporation does. The people who actually deal with these accounting issues are trying to say that they are not simple, but those who use the big numbers as a political club prefer to ignore them. It is an old trick.

Comment from Mark Danderson:

In 2011 Elsevier’s net profit (after interest, and tax) was £767mn on sales of £6.002bn giving a total corporate profit margin of 12.8%

In 2010 the same figures were £648mn, £6.055bn and 10.7%

But the really important number in business is not profit on sales but return on capital. The return on capital for Elsevier in 2011 was 34.9%.

David, since academic institutions are the sole market for journals and books does it matter which of the two is more profitable? The more important point is that the profit margin from the academic market segment is 37.3%.

I disagree that these numbers are not readily available. The big publicly traded publishers publish their results because it is required by law. The same is true of societies (at least in the US). If you dig hard enough, you can find the numbers you need.

David, you asked about the Univeristy of California Press; here are their numbers according to their annual report:

FY 09-10:
Books Sales $17.322mn; Operating Profit (before overhead) $9.11mn; Operating margin 52.6%
Journals Sales $5.401mn; Operating Profit (before overhead) $2.257mn; Operating margin 41.7%
Overhead expenses $15.375mn
Operating Loss -$2.805mn

Subsidies from:
UC System $2,611mn
UC Press Foundation $888k

TOTAL NET PROFIT $694k (including subsidies) Net Profit Margin 4%

They do not provide balance sheet information so it is impossible to calculate their return on capital.

Comment from Mark Danderson:

PLOS is a not-for-profit entity, so they use different accounting rules than those used by publicly traded companies such as Elsevier. So it is a bit harder to compare financial results (but not impossible).

According to PLoS financial reports here are their results for 2011:

Total revenues $22.276mn (roughly 95% of this comes from publication fees).

Their publication costs were $13.922mn meaning that their equivalent of operating profit is $8.354mn or 37.5% (almost exactly the same as Elsevier).

Their G&A expenses are $4.4mn, leaving them with a surplus (since they are not-for-profit, it cannot be called profit) of $3.954mn or 17.8% – a bit better than Elsevier.

As a not-for-profit, they do not have contributed capital but PLoS had total net assets (assets minus liabilities) of $8.72 at the end of 2011. Therefore their rate of return on net assets (which can be roughly compared to contributed capital in a publicly traded company) was 45.3% (which is considerably better than Elsevier).

You are right David, you cannot extrapolate the results from one company to an entire industry. As my brief and incomplete analysis shows, some companies perform better than others.


There are high-profile firms in our industry, and they are too often used as representatives of the entire industry, which is diverse and composed mainly of small publishers that make less than $25 million each year in revenues (PLOS may still be one of these, but probably matriculated from this group in 2012). The generally small scale of most scientific publishers has created major challenges for them in a digital age that rewards scale with economic benefits. It currently ensures diversity and deep involvement from scholars, as most companies are integral to their communities. Moving away from this diverse and integrated arrangement should be done with care.

As one comment noted, arriving at a “profit margin” is fraught because the term can refer to a number of related but differently encumbered measurements. There is margin on sales, gross margin, net margin, return on equity, return on assets, and so forth. Gross margins tend to be the most misleading, because they measure a subset of an organization’s expenditures against revenues — often, corporate expenses like legal, human resources, and other administrative expenses come later.

Another commentator argued there is nothing immoral about a certain profit margin. I tend to agree, as the role of any enterprise is to maximize its value while minimizing its costs. The difference between the two is profit. Companies that deliver a lot of value at relatively low costs can teach us things, yet are usually anomalies. But the very idea of a “margin” can also mislead. Comparing Elsevier’s margins to Exxon’s was particularly enlightening here, as a higher margin from Elsevier still amounts to a pittance compared to Exxon’s lower-margin business. Or Apple’s. Or Google’s. Size also plays a role. Scale matters. PLOS has a potentially higher-margin business than Elsevier, and Hindawi definitely has a higher-margin business. Those businesses are comparatively small. Are those margins immoral? Are larger companies immoral because they make more cash at a lower margin?

Keep those comments coming. Sweeping statements and isolated cases don’t help us figure out the overall trend or direction of the industry. Reading these comments, and others like them, is perhaps the most interesting facet of this blog, and where a lot of the conversations we need to have are occurring.

Kent Anderson

Kent Anderson

Kent Anderson is the CEO of RedLink and RedLink Network, a past-President of SSP, and the founder of the Scholarly Kitchen. He has worked as Publisher at AAAS/Science, CEO/Publisher of JBJS, Inc., a publishing executive at the Massachusetts Medical Society, Publishing Director of the New England Journal of Medicine, and Director of Medical Journals at the American Academy of Pediatrics. Opinions on social media or blogs are his own.


43 Thoughts on "Quantitative Comments — Highlighting Some of the Math About Finances and Profits Contributed by Readers"

Kent, come on. You simply typed the wrong sentence when you wrote: ‘Elsevier and Wiley comprise 0.4% of the STM market’.

‘The STM market’ means – to everyone – the financial market for STM publishing. Correct would have been ‘Elsevier and Wiley comprise 0.4% of STM [publishing] companies.’

You might as well try to defend the statement that ‘Google comprises less than 1% of the search engines market’ because there are more than 100 companies out there.

But everyone reading this blog understands this anyway. Making sense of the financials and trying to compare like-with-like in the STM publishing market is a difficult job, as you note. It’s not helpful to sprinkle these extra confusions into the mix.

The context of the initial response was to the fact that some were consistently using one or two companies to stand in for the entire industry. In that context, saying that Elsevier and Wiley comprise less than 1% of the industry made sense. It immediately became clear that the response was open to being interpreted as a market share value, which led to my later clarification that these large companies account for 19% of the market, but are only 0.4% of the companies. There was nothing “wrong” per se, just a small context and communication issue.

I also recreated the string pretty completely above, so any careful reader would see exactly what happened.

Indeed, for example Mike gives Elsevier a profit margin of 37% while Mark gets 10-12%, which adding principal payments will make even less. Clearly confusion is rampant.

Hey, no need to take my word for it. I got 37% direct from Reed Elsevier’s own 2011 annual report, which you can read for yourself. Skip down to page 9 for their figures for Elsevier: £768M adjusted operating profit on revenue of £2058M. If they want to report a different banner figure in their next annual report, I’ll happily switch to that.

You miss the point Mike. When it comes to profit there is a confusion of concepts. Different people are using wildly different, inconsistent concepts. The one you are using, while technically correct, is probably the most misleading. As a small business my concept of profit is what is left when I pay all my business bills. Operating profit of a division is far from that.

Then you are deliberately missing the point even after it has been carefully explained. The operating profit of a division is not the profit of the business. I cannot say it any simpler.

Actually for the purposes of comparing Elsevier’s pub business to say PLoS or Springer, Mike is correct. It’s why companies report unit level results – by breaking out results by division, it allows investors (and managers within the company) to have a clearer sense of the performance (accountability) of each unit. Consequently companies must then allocate all direct and indirect expenses accordingly so that investors can value each business unit vs. comparable companies in the industry, and if there’s synergy, then each unit should perform better, and get a better multiple, than their respective peers. If not, then they are more like a holding company (i.e. Berkshire). Investors value each business unit separately, then add them up to create an overall enterprise value, and triangulate that valuation with an overall enterprise discounted cash flow model, among other techniques.

Likewise some business (ex. Amazon) choose to not break out results by segment so that competitors cannot see how one unit compares to others. It frustrates investors and can lead to a lower valuation if investors mis-value the parts – in which case, businesses then may decide to show segment results.

That would be fine if they were breaking out all of the corporate costs by segment, but as Mark’s numbers show they are not. Many costs are only being reported at the corporate (or business) level. This makes operating units look far more profitable than they really are. It is a standard trick that Mike has fallen for. For example, if increased revenue is due to acquisition then the cost of that acquisition is part of the cost of that revenue, but that is not how it gets reported.

You are missing the point. Elsevier may report one profit concept to reassure investors, while their net profits after all the costs Mark Danderson noted yield another profit metric that actually means “extra money to spend.”

You can use “the one Elsevier are reporting” but if you don’t understand what that is, you are using a number without comprehension, which means you can’t easily compare it to other measures of “profit” other companies may be using.

Elsevier may report one profit concept to reassure investors, while their net profits after all the costs Mark Danderson noted yield another profit metric that actually means “extra money to spend

While it is no doubt possible to imagine all sorts of mathematical contortions, Elsevier is a publicly traded company which is required by law to file the report referred to above. It is precisely the point of generally accepted accounting standards that one company’s measure of “profit” will indeed be congruent with another’s.

Comment by Thomas Hilock: “Sadly PeerJ doesn’t pass the sniff test. Not even close.”

Well, of course, what do you expect? PeerJ is barely 6 months old. Look at typical journal lag times submission to publication.

For PeerJ to be financially successful it needs the economy of scale from publishing thousands of articles a year. Is that a stretch long term? It is the exact same model as PLoS One which published over 26,000 articles last year and PeerJ is run by at least two of the people who worked on developing PLoS One.

Publishing in PeerJ costs $500 for a typical sized group of 5 authors and gives them the ability to publish once a year for the rest of their life. It cost $1,350 to publish in PLoS One. PeerJ has a slick intuitive journal management system and a very impressive group of over 800 academic editors. I’ve both published in PeerJ and acted as a academic editor and the review process is no different than other journals run by volunteer editors backed up by a managing editorial team. The only difference is you are not stuck trying to use some clunky outdated journal management system like Scholar One.

Will PeerJ be able to scale up and publish thousands of articles a year? It really up to biomedical researchers but I pray every day researchers start doing the sensible thing because this is a model that provides all the aspects of basic quality scientific publishing at a price most any researcher even those in developing countries can afford without charging grants or getting into the whole RCUK debate.

There’s more to it than that though–if PeerJ is able to scale up to those levels, is it still a viable business model? How many of the authors in those 26,000 papers are first time authors? And then in the next year, the same question applies. And the next, and the next.

You figure there’s a new crop of graduate students coming in every year, but how many of them are included on papers, and even so, once they’ve been included, they now become a cost center rather than a revenue stream. Can the influx of new authors pay for the already existing members? Is there a tipping point?

It seems something of a front-loaded business model which, perhaps is appropriate for venture capitalists looking for a short term profit from establishing the journal then selling it off.

All fascinating questions, and perhaps tomorrow’s Scholarly Kitchen post, a podcast interview with Peter Binfield will shed some light on them.

Unless the can publish papers for ten dollars the present pricing scheme cannot survive steady state. They are playing the bubble.

David has it spot-on. It’s comparable to opening an all-you-can-eat restaurant where you pay once, for life. It’s not a business model I’m familiar with.

I’d also say that my expenses in my original post are extremely modest. I would predict their fixed costs run beyond $1m per year. I would be happy to break that down.

As a transient, startup model an example would be a country club selling very cheap memberships to get people in, then raising rates to a sustainable level. I have seen this done.

Agreed, but you rarely see them give cheap memberships for life, even for early adopters. As you point out, it would require a change to their business model and the fundamental message of their business.

As it has been mentioned often on this blog, the average scientific author publishes just one paper in their career. The long author lists in biomedicine are usually mostly composed of graduate students and postdocs, so I think the fraction of new PeerJ authors will be significant in the long term as well, even if all the PIs publish for free.

Maybe it is worth noticing that if you pay after your manuscript has been accepted the charge is 139$/author instead of 99$.

I have no idea how the math works out, or what percentage of authors on a typical paper are “new” authors. If one assumes some sort of saturation point for authors in a field, then the only revenue a journal would get is from these new authors. So can you at least break even receiving only $139 for some percentage of papers that you publish?

I have worked the math a lot of ways and it never works for any reasonable assumptions that I can find. If someone has a math model that works let’s see it, especially the assumptions.

As for one time authors they are probably using their dissertation and there will usually be only one per paper, plus their adviser and maybe some other high-repeat authors. Nor are one time authors likely to buy life memberships.

Authorship practices vary highly between fields. What you describe would not be typical of the life sciences (which is the only field I know). Often the list looks more like this:

The student/postdoc who did most of the work. A bunch of undergraduates or junior graduate students who have carried out sub-projects. A number of postdocs/senior graduate students who have provided technical assistance. A senior postdoc/junior faculty member who supervised the project day-to-day. Big name PI who got the money and initiated the study.

I had no idea most life science authors were undergrads or junior grads, most of whom do not go into research. If the business model depends on them all buying life memberships I question that assumption.

I think the point is that you will not really reach a saturation point. The vast majority of graduate students and postdocs do not end up in a permanent position in academic research. Effectively, 90% of the population of potential authors thus turns over every five years. The appropriate analogy is thus not a restaurant selling a life-time membership, but rather a college-cafeteria, knowing that most people will be gone after a couple of years.

No one knows what the fraction of new authors would be in an equilibrium situation. My guess based on biomedicine as the primary subject area would be 30-50%. Assuming on average 7 authors as Kent did that works out as 300-500$/paper. Whether you can operate a profitable business on that kind of money is another question, but given PLOS ONE’s profitability at 1350$/paper, it doesn’t seem that ridiculous.

Also, we should acknowledge that the “99$ for a lifetime of publishing” was a brilliant marketing strategy. Who would have heard of PeerJ if they had set their APC at 2/3 of PLOS ONE’s?

As I said, I can’t begin to process the numbers in my head. But I’m not sure that you’d really have 30-50% of the authors on a given paper being first (and only) publication graduate and undergraduate students. I’d guess closer to 1 or 0.5 such students per paper at least in the fields where I used to work, where a lot of the labs are postdoc heavy.

But I wholly agree that camouflaging the costs in this manner is indeed brilliant. The people behind PeerJ that I know are all really smart, and I assume they’ve done the projections and are satisfied with the numbers.

Good points and it will be interesting to see what he has to say. I expect Pete and his colleagues used PLoS One data to actually model the assumptions they made about publishing behavior unlike the people who commented below.

I doubt PLoS One has that kind of data as it would be labor intensive. Nor would it likely be useful since the business models are vastly different, hence targeting different demographics.

So as the person who chaired that particular session at #alpsp13 and invited Jason to speak ( and full,disclosure, sits on the ALPSP awards judging panel) I can directly comment on the intent of that $600PM number. It’s about the cost of running a Development, Test and Live server environment on Amazon’s web services, including access to the content delivery network, and accounting for upshifts and downshifts in the bandwidth requirements of PeerJ.

It’s an impressive number full stop. Does it cover techies and what not? Well yes and no. yes in that Amazon do all the back end heavy lifting. No in the sense that somebody at the PeerJ end would have to swap to back-ups etc if the Amazon network went down. They have a plan for that by the way.

Developer costs and the rest… Not part of that number. but it’s an impressive number and should give folks pause for thought – there can be other ways of doing things.

The point of that number is that it speaks to a very admirable focus on spending money on what are truly core activities for an organisation… Wrangling the information is what PeerJ are about, not mucking about with recalcitrant servers.

FWIW PeerJ have have publicly stated that they are on track to be operationally self sustaining next year. that’s a statement that they can be tested against.

Recall the excellent Keynote that Larry Schwartz gave at an SSP annual meeting a few years back on running Newstex via a whole slew of SaaS business services designed to provide the ultra low cost functionality that a business needs. That was an impressive talk by an impressive company.

Running an agile business takes a cultural mindset that some new orgs have. Perhaps that’s why those orgs get bought for circa $100 million. The value of the mindset that can run a company via skype, and just do it.

I don’t know if PeerJ will be successful. The truth is, the odds are against them. Most innovative companies fail, for all sorts of reasons. But we can still learn useful things by observing their approach to business. Bearing down on costs and a relentless focus on the users, are worthy traits. Full stop.

FWIW PeerJ have have publicly stated that they are on track to be operationally self sustaining next year. that’s a statement that they can be tested against.

Sustainable over the short term may not necessarily mean sustainable over the long term. And VC’s tend to play more of a short term game.

From the distant past of my finance classes I remember that you can’t evaluate profitability without analyzing risk.

Companies like Elsevier reduce risk by borrowing to buy mature (low risk) publishing enterprises. That’s why most of their “profits” are absorbed by interest payments.

New ventures like Hindawi and PeerJ are high risk and therefore justify and need much higher profits.

In aggregate, successful enterprises need to “pay” for all the failures. A “cost plus” view of publishing ignores the impact of risk in the economic equation. Maybe that’s a realistic assumption in an OA context?

Kent said the STM publishing industry is “composed mainly of small publishers that make less than $25 million each year in revenues (PLOS may still be one of these, but probably matriculated from this group in 2012).” PLoS CERTAINLY matriculated from that group in 2012: their revenue was over $37M if they collected full price for every article they published ($31M from P£o$ On€). In 2013 they will collect over $40M from P£o$ On€ alone (if they realize full payment for every article published).

You’re probably right, but until I see their 990 and actual evidence, I’m still listing it as “probable.” Their 990 should be up in the next few weeks, so we’ll have confirmation then.

I often wonder where everyone gets their numbers from and how they are interpreted. Having tracked this area for nearly forty years, let me share some STM data with everyone. There are about 16,000 publishers in this marketplace. However there are 30 publishers that makeup 90% of the total sales certainly to the library marketplace. And 40 publishers make up 95% of the sales. The interesting thing is even with 10 years of OA talk and preaching, most publishers including the top 40 obtain 90% of their revenue from the traditional subscription model. The trend out there is still libraries subscribing to their favorite journal. Find a publisher that obtains more than 10% of their revenue from OA activity. Tipping point! Not even close.

It’s true that 90% of publishing is still subscription-based by revenue. But the proportion by volume is much smaller (and that of open-access publishing correspondingly larger) precisely because open access is cheaper. And that of course is why publishers have resisted it.

It’s much simpler than whether more is better. This is about whether cheaper is better. And the answer is obvious: yes for customers, no for suppliers. Is it any wonder that you, a supplier, and I, a customer, disagree over what is to be done?

“Cheaper” often leaves things out, like editors-in-chief, statistical review, marketing, audience development, etc. Do you really think cheaper = better if important things are left out in order to achieve the price reduction? More and cheaper. Sounds like a great formula for scientific progress. How about better and valid?

And there is no evidence OA is cheaper. In fact, it seems to centralize payments around governments and academic centers, making taxpayers and students pay more. That *seems* cheaper to users, but the costs of diverted research funds and additional fees for students are real, and they can have long-term implications. Studies comparing what universities pay for subscriptions vs. what they’d pay for OA APCs have consistently shown a 3-10x increase in costs at the university level. Subscriptions actually make things more affordable for academic centers and governments, while preserving incentives for quality.

All of this is exactly what I’d expect a for-profit barrier-based publisher to say.

We all know perfectly well that open access is much cheaper. As this post shows, the average Gold APC comes to about $450 (£280), while the average barrier-based article costs the community of subscribers about $5000. Even if some of those figures are a bit off, there’s no escaping that there’s about an order of magnitude of difference.

With the top 40 publishers the paid subscription model is about 95% of revenue and of paid units. Again I am only taking about the top 40 STM publishers which account for 95% of the sales. OA in this community is more about testing and experimentation.

Comments are closed.