Re: Institutional Journal Costs in an Open Access Environment

[Ray English <Ray.English@oberlin.edu>]

I think it's important to read the full study by Bill Walters and 
look at his data closely before deciding what conclusions to 
draw.  I was on a panel with Bill back in March and I saw his 
summary findings at that time.  I also read through his full 
study carefully night before last before formulating these 
comments.

The study looks at journal costs in four scientific fields for a 
representative group of academic and research libraries of 
different sizes.  Specifically it examines three models:  1) what 
those libraries are currently paying for journals in those 
fields, 2) what those institutions (not necessarily just their 
libraries) would pay under a PLOS model, and 3) what those 
institutions would pay under what he calls an equal-revenue open 
access model.

The PLOS model assumes that all journals in those fields would be 
open access, that they would charge author fees of $1,500 per 
article, and that the libraries would maintain print copies of 
all journals at the PLOS annual charge of $160 per title.  Under 
this model all of the institutions studied would pay 
substantially less than they did under the conventional model. 
The cost for the largest research institutions (University of 
Michigan) would be just 51% of what they would pay under the 
conventional model.  Savings for smaller institutions would be 
substantially more (up to 97% in the case of one comprehensive 
university).  All institutions would gain in terms of research 
access, with dramatic gains for the smaller institutions and more 
modest gains for the large research library.)

Bill developed the equal-revenue open access model primarily 
because he felt that publishers would not accept that much of a 
loss in revenue.  The equal revenue model maintains total 
spending at the level of the conventional model (what libraries 
are currently paying), but distributes that spending according to 
the number of first-author articles published at the institutions 
studied.  His model does indeed show that for the largest 
research university (again the University of Michigan) the total 
cost to the institution would be substantially more under this 
model than under the conventional model ($1.14 million versus 
$627 thousand).  All other institutions would have substantial 
savings.

The thing that Bill doesn't focus on, however, is the cost per 
article under his equal-revenue model.  If you assume that the 
cost of providing the print copy remains the same as under the 
PLOS model ($160 per year per title), then the cost per article 
under the equal-revenue model comes out to $6,162.58.  (The per 
article cost increased to $6,491.69, if you assume no print copy 
distribution.)  Either figure is way more than most estimates of 
cost per article that have been cited in the literature.

All of the institutions in the study would actually pay less 
under an open access model if the average cost per article was 
below $3,363. That includes, again, the largest research 
university (Michigan), which comes out at exactly break-even at 
that dollar figure.  (This assumes author fees are fully paid by 
the university and that the journals have no alternative sources 
of revenue.)

Of course the actual average cost per article is in much debate. 
But this clearly shows that if, for example, the Springer Open 
Choice model ($3000 per article author fee) was adopted across 
the board for all journals in these fields, there would be 
substantial savings for all of the institutions in the study. 
Savings would be greatest for the comprehensive universities 
(around 90% or more of current costs), somewhat smaller (but 
still very large) for the liberal arts colleges, smaller still 
for the doctoral universities (about 70% for Brandeis, for 
example), and smallest for the largest research institutions. 
But there would still be significant savings even for the latter 
group.  Michigan, for example, would reduce its journal costs in 
those fields by $67,663 -  more than 10% of current costs.

I would also note that Bill=92s study, like the earlier Cornell 
study, does not take into account the question of grant funds 
supporting authors' fees.  That would likely be an accepted 
practice under the open access model in some of the fields he 
studied.  Those funds should not be counted as institutional 
costs.  The study also does not factor in other possible sources 
of revenue for open access journals, other than authors fees.

So my conclusion is that this study presents some very 
interesting data that deserve close analysis.  The data clearly 
show--for the journals in the fields studied--that most 
universities and colleges would have substantial savings under an 
open access model. The large research university in the study 
were anywhere below $3,363.  It would "pay" more if fees were 
above that level.  (I say "pay" in quotation marks because some 
funding would be supported by research grants. I also note again 
that not all open access revenue has to come from authors' fees. 
To the extent that either of these proves possible, it increases 
the threshold article fee level where all institutions save.)

Given all of the above, I don't think it's justified to conclude 
that the data in the study support the conclusions of the Cornell 
study, which asserted that the largest research universities 
(actually the largest research libraries) would pay more under an 
open access model. Based on Bill's data, that only happens if 
authors' fees exceed a certain level (in the range of $3,363, as 
illustrated by the case of Michigan).

In addition, the model in the Cornell study was far to 
preliminary to justify the broad conclusions that were drawn from 
it.  As I indicated in a lengthy post to the SCHOLCOMM list on 
January 24, 2005, the Cornell study failed to take major factors 
into account.  Some factors that were not built into the Cornell 
model included:  support of authors' fees from grant funds, 
variations in price per article by broad disciplinary categories 
(the study assumed average costs in science fields would also 
apply to humanities and social sciences - which clearly is not 
the case), and the number of articles in broad disciplinary 
categories being generated at each institution.

Ray English
Director of Libraries
Oberlin College
440-775-8287