ROE is a financial performance indicator. They're generally in the single digits, low double digits is considered good. In a competitive market, they should converge to round about the same number on average, even across industries (module riskyness, correlation with market as a whole, and such factors).
The ROE of academic publishers is abnormally high, indicating some sort of market failure.
In this case, scientists do almost all of the work (write, typeset (mostly), referee) for free (or nominal amounts), so expenses are low.
The publishers charge libraries inordinate amounts for subscriptions (dollars per page).
And scientist need to publish, and they need to publish in high prestige/impact journals, and the prestige/impact is tied to the title, and the publishers own the titles, it is hard for competitors to enter.
The ROE of academic publishers is abnormally high, indicating some sort of market failure.
In this case, scientists do almost all of the work (write, typeset (mostly), referee) for free (or nominal amounts), so expenses are low.
The publishers charge libraries inordinate amounts for subscriptions (dollars per page).
And scientist need to publish, and they need to publish in high prestige/impact journals, and the prestige/impact is tied to the title, and the publishers own the titles, it is hard for competitors to enter.