Sauropod Vertebra Picture of the Week

Last time, we looked at the difference between cost, value and price, and applied those concepts to simple markets like the one for chairs, and the complex market that is scholarly publication. We finished with the observation that the price our community pays for the publication of a paper (about $3,333 on average) is about 3–7 times as much as its costs to publish ($500-$1000)?

How is this possible? One part of the answer is that the value of a published paper to the commnity is higher still: were it not so, no-one would be paying. But that can’t be the whole reason.

In an efficient market, competing providers of a good will each try to undercut each other until the prices they charge approach the cost. If, for example, Elsevier and Springer-Nature were competing in a healthy free market, they would each be charging prices around one third of what they are charging now, for fear of being outcompeted by their lower-priced competitor. (Half of those price-cuts would be absorbed just by decreasing the huge profit margins; the rest would have to come from streamlining business processes, in particular things like the costs of maintaining paywalls and the means of passing through them.)

So why doesn’t the Invisible Hand operate on scholarly publishers? Because they are not really in competition. Subscriptions are not substitutable goods because each published article is unique. If I need to read an article in an Elsevier journal then it’s no good my buying a lower-priced Springer-Nature subscription instead: it won’t give me access to the article I need.

(This is one of the reasons why the APC-based model — despite its very real drawbacks — is better than the subscription model: because the editorial-and-publication services offered by Elsevier and Springer-Nature are substitutable. If one offers the service for $3000 and the other for $2000, I can go to the better-value provider. And if some other publisher offers it for $1000 or $500, I can go there instead.)

The last few years have seen huge and welcome strides towards establishing open access as the dominant mode of publication for scholarly works, and currently output is split more or less 50/50 between paywalled and open. We can expect OA to dominate increasingly in future years. In many respects, the battle for OA is won: we’ve not got to VE Day yet, but the D-Day Landings have been accomplished.

Yet big-publisher APCs still sit in the $3000–$5000 range instead of converging on $500-$1000. Why?

Björn Brembs has been writing for years about the fact that every market has a luxury segment: you can buy a perfectly functional wristwatch for $10, yet people spend thousands on high-end watches. He’s long been concerned that if scholarly publishing goes APC-only, then people will be queuing up to pay the €9,500 APC for Nature in what would become a straightforward pay-for-prestige deal. And he’s right: given the outstandingly stupid way we evaluate reseachers for jobs, promotion and tenure, lots of people will pay a 10x markup for the “I was published in Nature” badge even though Nature papers are an objectively bad way to communicate research.

But it feels like something stranger is happening here. It’s almost as though the whole darned market is a luxury segment. The average APC funded by the Wellcome Trust in 2018/19 was £2,410 — currently about $3,300. Which is almost exactly the average article cost of $3,333 that we calculated earlier. What’s happening is that the big publishers have landed on APCs at rates that preserve the previous level of income. That is understandable on their part, but what I want to know is why are we still paying them? Why are all Wellcome’s grantees not walking away from Elsevier and Springer-Nature, and publishing in much cheaper alternatives?

Why, in other words, are market forces not operating here?

I can think of three reasons why researchers prefer to spend $3000 instead of $1000:

1. It could be that they are genuinely getting a three-times-better service from the big publishers. I mention this purely for completeness, as no evidence supports the hypothesis. There seems to be absolutely no correlation between price and quality of service.
2. Researchers are coasting on sheer inertia, continuing to submit to the journals they used to submit to back in the bad old days of subscriptions. I am not entirely without sympathy for this: there is comfort in familiarity, and convenience in knowing a journal’s flavour, expectations and editorial board. But are those things worth a 200% markup?
3. Researchers are buying prestige — or at least what they perceive as prestige. (In reality, I am not convinced that papers in non-exceptional Elsevier or Springer-Nature journals are at all thought of as more prestigous than those in cheaper but better born-OA journals. But for this to happen, it only needs people to think the old journals are more prestigious, it doesn’t need them to be right.)

But underlying all these reasons to go to a more expensive publishers is one very important reason not to bother going to a cheaper publisher: researchers are spending other people’s money. No wonder they don’t care about the extra few thousand pounds.

How can funders fix this, and get APCs down to levels that approximate publishing cost? I see at least three possibilities.

First, they could stop paying APCs for their grantees. Instead, they could add a fixed sum onto all grants they make — $1,500, say — and leave it up to the researchers whether to spend more on a legacy publisher (supplementing the $1,500 from other sources of their own) or to spend less on a cheaper born-OA publisher and redistribute the excess elsewhere.

Second, funders could simply publish the papes themselves. To be fair several big funders are doing this now, so we have Wellcome Open Research, Gates Open Research, etc. But doesn’t it seem a bit silly to silo research according to what body awarded the grant that funded it? And what about authors who don’t have a grant from one of these bodies, or indeed any grant at all?

That’s why I think the third solution is best. I would like to see funders stop paying APCs and stop building their own publishing solutions, and instead collaborate to build and maintain a global publishing solution that all researchers could use irrespective of grant-recipient status. I have much to say on what such a solution should look like, but that is for another time.

We have a tendency to be sloppy about language in everyday usage, so that words like “cost”, “value” and “price” are used more or less interchangeably. But economists will tell you that the words have distinct meanings, and picking them apart is crucial to understand economic transaction. Suppose I am a carpenter and I make chairs:

• The cost of the chair is what it costs me to make it: raw materials, overheads, my own time, etc.
• The value of the chair is what it’s worth to you: how much it adds to your lifestyle.
• The price of the chair is how much you actually pay me for it.

In a functioning market, the value is more than the cost. Say it costs me £60 to make the chair, and it’s worth £100 to you. Then there is a £40 range in which the price could fall and we would both come out of the deal ahead. If you buy the chair for £75, then I have made £15 more than what it cost me to make, so I am happy; and you got it for £25 less than it was worth to you, so you’re happy, too.

(If the value is less than the cost, then there is no happy outcome. The best I can do is dump the product on the market at below cost, in the hope of making back at least some of my outlay.)

So far, so good.

Now let’s think about scientific publications.

There is a growing consensus that the cost of converting a scientific manuscript into a published paper — peer-reviewed, typeset, made machine-readable, references extracted, archived, indexed, sustainably hosted — is on the order of $500-$1000.

The value of a published paper to the world is incredibly hard to estimate, but let’s for now just say that it’s high. (We’ll see evidence of this in a moment.)

The price of a published paper is easier to calculate. According to the 2018 edition of the STM Report (which seems to be the most recent one available), “The annual revenues generated from English-language STM journal publishing are estimated at about $10 billion in 2017 […] collectively publishing over 3 million articles a year” (p5). So, bundling together subscription revenues, APCs, offsets deals and what have you, the average revenue accruing from a paper is $10,000,000,000/3,000,000 = $10,000/3 = $3,333.

(Given that these prices are paid, we can be confident that the value is at least as much, i.e. somewhere north of $3,333 — which is why I was happy earlier to characterise the value as “high”.)

Why is it possible for the price of a paper to be 3–7 times as high as its cost? One part of the answer is that the value is higher still. Were it not so, no-one would be paying. But that can’t be the whole reason.

Tune in next time to find out the exciting reason why the price of scholarly publishing is so much higher than the cost!