Open Access and Academic Journal Quality.pdf

(219 KB) Pobierz

American Economic Association

Open Access and Academic Journal Quality

Author(s): Mark J. McCabe and Christopher M. Snyder

Source:

The American Economic Review,

Vol. 95, No. 2, Papers and Proceedings of the One

Hundred Seventeenth Annual Meeting of the American Economic Association, Philadelphia,

PA, January 7-9, 2005 (May, 2005), pp. 453-458

Published by:

American Economic Association

Stable URL:

http://www.jstor.org/stable/4132864

Accessed: 05/02/2015 15:47

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Economic Association

is collaborating with JSTOR to digitize, preserve and extend access to

The

American Economic Review.

http://www.jstor.org

This content downloaded from 193.0.118.39 on Thu, 5 Feb 2015 15:47:28 PM

All use subject to

JSTOR Terms and Conditions

Open Access and Academic JournalQuality

By MARK J. MCCABE AND CHRISTOPHER M. SNYDER*

Scholars and librarianshave grown increas-

ingly dissatisfied with the marketfor academic

journals. New technologies might be expected

to lower journals' production and distribution

costs, and for these reduced costs to factor into

reduced prices; but library subscription prices

remain high (Theodore Bergstrom, 2001) and

indeed have continued to rise faster than infla-

tion (McCabe, 2002). This dissatisfaction has

led to the proposal of a new business model for

academicjournals, open access. In contrastto a

traditionaljournal, which generates most of its

revenue with subscriptionfees, an open-access

journal makes its articles freely available on the

Internet, generating revenue with author fees.

As of October 2004, the Directory of Open

Access Journals (www.doaj.org) listed over

1,300 open-access journals across academic

fields. Perhapsthe most famous of these are the

biomedical journals published by the Public

Library of Science, founded by Nobel-prize-

winning biologist Harold Varmus to compete

against the top-tier journals in the field. These

journals charge substantialauthor fees, $1,500

per accepted paper.'

The fee structureof journals has important

consequences for social welfare. Subscription

prices have risen to the point that librarieshave

begun to cancel significant titles, potentially

harming scholars in their roles both as readers

and authors.Research-fundinginstitutions such

as the U.S. National Institutesof Health and the

Wellcome Trust have considered policies rang-

* McCabe: Departmentof Economics, Georgia Institute

of Technology, 781 Marietta Street, N.W., Atlanta, GA

30318; Snyder: Departmentof Economics, DartmouthCol-

lege, Hanover, NH 03755. The authors are grateful for

helpful comments from seminar participantsat Dartmouth,

George Washington,Michigan State, and Tulane, and to the

Open Society Institute and the NET Institutefor funding.

1 In

economics, the adoption of open access has been

relatively slow, with only nine refereed open-access jour-

nals listed on the Directory of Open Access Journals.None

of these charges author fees, operating on donated institu-

tional support.

453

ing from increasingthe amountthat is budgeted

in grantsfor authorfees to requiringthat funded

authorspublish in open-access journals.

Much of the previous theoretical work on

academicjournals (see e.g., Doh-Shin Jeon and

Domenico Menicucci, 2003; McCabe, 2004)

cannot be used to study open access because it

only considers one side of the market, library

subscriptions, and does not endogenize author

fees on the other side of the market.In an earlier

paper (McCabe and Snyder, 2004a), we con-

structed a two-sided-market model with bilat-

eral externalities (readers obtain benefits from

reading, and authorsobtain benefits from being

read). We showed that a commercial journal is

more likely to adopt open access the lower its

market power, the lower its marginal cost of

serving readers, and the greaterare authorben-

efits relative to reader benefits. The last result

holds because, if authors have relatively high

benefits per reader,the journal will try to recruit

more readers by lowering subscription fees,

thus increasingthe revenue thatcan be extracted

from authors. We proved a series of "anything

is possible" results, providing cases in which

even a monopoly journal would adopt open

access in equilibrium and other examples in

which open access is not socially efficient even

if the marginal cost of serving readers were

zero.

In the present paper, we extend our previous

model to allow articles and journals to vary in

quality. Good articles provide a reader benefit;

bad articles do not. Readers cannot tell the

quality of articles prior to reading them, and

reading an article requires an effort cost. Jour-

nals' quality differences emerge endogenously

through the talent of their editors, where more

talented editors can distinguish between good

and bad articles with more precision. High-

qualityjournals thus publish more good articles.

This extension is useful because it allows for a

more realistic depiction of journals. It also al-

lows us to answer new questions of policy

relevance about open access. First, should we

This content downloaded from 193.0.118.39 on Thu, 5 Feb 2015 15:47:28 PM

All use subject to

JSTOR Terms and Conditions

454

AEA PAPERSAND PROCEEDINGS

MAY2005

expect to see open access being more likely to

emerge at the high-qualityor low-quality end of

the journal spectrum?HaroldVarmussuggested

that open access should emerge at the high end:

"The most importantthing is that we, as pub-

lishers of open access journals, want our jour-

nals to be high quality. It is the only way we are

going to succeed" (House of Commons, 2004 p.

80). The House of Commons Science and Tech-

nology Committeeconcluded otherwise:"There

is a risk that some partsof the marketwould be

able to producejournalsquickly, at high volume

and with reduced quality control and still suc-

ceed in terms of profit, if not reputation.Such

journals would cater for those academics for

whom reputationand impact were less impor-

tant factors than publication itself' (House of

Commons, 2004 p. 80). Second, should we ex-

pect open-access journals to lower their quality

standardsin orderto boost revenue from author

fees? Such is the attack often leveled against

open access by commercial publishers, for

example the CEO of Elsevier: "If you are re-

ceiving potential payment for every article

submittedthere is an inherentconflict of interest

that could threatenthe qualityof the peer review

system" (House of Commons, 2004 p. 81).

I. Model

There are three sorts of agents in the model: a

unit mass of authors, unitmass of readers, a

and

single commercial(i.e., profit-maximizing)

jour-

nal. Authors submit articles of varying quality

to the journal. The journal (more precisely, the

journal'seditor;the two labels will be used inter-

changeablyhere)judges the qualityof the submit-

ted articlesand acceptsa subset.Acceptedarticles

are bundledtogetherin an issue and distributed

readers.

Each authoris endowed with a single article.

Authors obtain a benefit

ba ?

0 per reader,

derived in part from the fact that having more

readers increases the article's impact and

chances of being cited and thus improves the

author's career prospects. The analysis is con-

siderably simplified without much loss of in-

sight with the assumptionthat all authors have

the same benefit

ba.

Articles are of random

quality. A fraction y E [0, 1] of them are

"good" and 1 - y are "bad." Readers only

obtain a benefit from reading good articles.

Since there will be a cost per article of reading,

readers will prefer journals that have a high

percentage of good articles. To simplify the

model, assume authorsdo not know their own

article's quality prior to submission.2

The editor can only imperfectly determinean

article's quality, dependingon his talent,t E [0,

1]. The editor can perfectly identify good arti-

cles as being good. With probabilityt, he cor-

rectly identifies a bad article as being bad. With

probability 1 - t, he mistakenly judges a bad

article to be good. Assume that t is public

information.

Readers obtain no benefit from reading bad

articles. Reader k obtains benefit

brk

0 per

good article read. Assume

brk

is a randomvari-

able with cumulative distribution function F

and densityf. Reading an article requireseffort,

which costs the readerp 0. Hence the reader

wishes to avoid readingbad articles,which pro-

vide no benefit but are costly to read.The reader

cannot determine the quality of an articleprior

to reading it.

Let

- 0 be the journal's cost of handlinga

submitted article up through and including the

process of judging its quality, reflectingthe cost

of referees' and editor's time and any adminis-

trativecosts of processing the author'saccount.

Let ca - 0 be the cost of processing an accepted

article, reflecting copyediting, typesetting, and

administrativeexpenses. The cost of distribut-

ing the articles to a single reader includes a

fixed cost

0 for the bundle of articlesin the

journal (reflectingthe cost of servicing the read-

er's account and any fixed shipping and han-

dling costs) plus a variable cost c per article

(reflecting remaining variable shipping costs,

including the cost of bandwidth in the case of

Internetdistribution).

The journal charges submission fee ps and,

conditional on acceptance, accepted-paperfee

This assumption is consistent with a numberof other

recent papers involving quality certificationby an interme-

diary (Josh Lerner and Jean Tirole, 2004; Alan Morrison

and Lucy White, 2004). It serves to simplify the analysisby

abstracting from complicated signaling behavior by in-

formed authors. Lerner and Tirole show that adding up-

streamprivateinformationdoes not altertheirbasic analysis

(their Proposition 4).

This content downloaded from 193.0.118.39 on Thu, 5 Feb 2015 15:47:28 PM

All use subject to

JSTOR Terms and Conditions

VOL.95 NO. 2

Pa.

COMPETITION

POLICYFOR JOURNALS

455

The journal charges subscription fee

readersfor the bundle of articles in the journal.

We will constrain prices

ps, Pa,

and

to be

nonnegative.3 Assume that an article's quality

cannot be verified ex post, so in particular,the

journal's pricing scheme cannot be conditioned

on realized quality (althoughin equilibriumfees

will depend on editorial talent).

The timingof the model is as follows. First,the

journalchooses prices. Then authorsand readers

simultaneouslymake their submission and sub-

scription decisions. Finally, the journal decides

which articlesto acceptor reject.We will look for

a subgame-perfect,

rational-expectations

equilib-

riumin which outcomeson any subgameinvolv-

ing the infinitesimal players (authors and/or

(RobertAu-

readers) a strongNash equilibrium

mann,

1959).4

JournalsAdopt Open

II. Will High-Quality

Access?

Taking the case of a monopoly commercial

journal,we will performthe comparative-statics

exercise of examining the effect of a change in

the editor's talent t on the equilibriumsubscrip-

tion price with the goal of determiningwhether

a high- or low-quality journal would be more

likely to adopt open access.

In this section, we will maintainthe assump-

tion that the editor can commit to a policy of

only accepting articles believed to be good.

Under this editorial policy, the probability of

acceptance,denoted a, is a = y + (1 - y)(1 -

t). Journalprofit is

(1)

psna +

pana

nr)

+ prnr - TC(na, nr)

(2)

TC(na, nr)

nacs

anaca + nrCr

ananrC.

Aggregate author demand is inelastic be-

cause authors are homogeneous. The number

of submissions is positive, equal to the unit

mass of authors, if and only if net author

surplus,

(3)

a(nrba - Pa) - Ps

is nonnegative. Reader k's expected net surplus

from subscribing to the journal is

(4)

ynabrk

anap - Pr,

Reader k will subscribeto the journal if expres-

sion (4) is nonnegative, implying that aggregate

reader demand is

(5)

nr = 1 - Fp

naP

)

yna

Conditionalon the level of the total expected

payment from an authorto the journalp, +

acpa,

the particulardivision into subscription fee p,

and acceptance fee

is irrelevant. (This divi-

sion will become relevant in the next section.)

Without loss of generality, we will set the equi-

librium submission fee, p,* to 0. Then the equi-

librium acceptance fee will be the highest value

subject to author demand being positive. From

equation (3), the equilibriumacceptancefee and

author demand satisfy p* =

n*ba

and

The equilibrium subscription fee maximizes

journalprofit,which upon substituting =

nrba

and

= 1, as well as equations (2) and (5) into

(1), becomes

(6)

II(pr) = (aba + Pr -

- ac)

where

TC(na,

is the total cost function,

subsidize authors and readers, in that

be set below marginal cost, but journals are

prices may

assumed not to make explicit cash transfers to authors or

readers. The restriction of cash transfers appears to be

nearly universal among scholarlyjournals.

4 A strong Nash equilibriumrequires the outcome to be

immune to profitable deviations by any coalition of the

infinitesimal players. See the more detailed theoretical pa-

per on which this article is based, McCabe and Snyder

(2004b), for a discussion of the role of this refinement.

Journals may

F(PrY)

-cs

aca-

Applying the implicit function rule to equa-

tion (6) yields the following proposition.

PROPOSITION

1. Assume

> p + c. Assume

the second-order condition from maximization

of profit in equation (6) holds. The equilibrium

subscription fee p* charged by a commercial

This content downloaded from 193.0.118.39 on Thu, 5 Feb 2015 15:47:28 PM

All use subject to

JSTOR Terms and Conditions

456

AEA PAPERSAND PROCEEDINGS

MAY2005

journal is weakly increasing in journal quality/

editorial talent, t.

Proposition 1 states that, under the specified

conditions,5the subscriptionfee is increasing in

editorial talent, implying that if a high-quality

journal adopts open access in equilibrium, a

lower-qualityjournal would as well. Hence, un-

der the conditions of the proposition, a low-

quality journal would be more likely to adopt

open access.6 The intuition for the result is that

as journal quality (equivalently,editorial talent)

increases, authorssuffer a direct loss and read-

ers enjoy a direct benefit. The direct loss to an

authoris that his article is published with lower

probabilitysince his articlemay be bad, and bad

papers are more likely to be rejected.The direct

benefit to a readeris that his cost of reading the

journal falls because the journal contains fewer

bad articles. The journal optimally responds to

the relative changes in surpluses by reducing

authorfees and increasing readerfees.

Some caveats regardingProposition 1 are in

order. The result can be reversed (thus, high-

quality journals can be more likely to adopt

open access) if the conditions underlying the

propositiondo not hold. Furthermore, shown

in McCabe and Snyder (2004b), the results for

nonprofit journals are less clear-cut. As dis-

cussed in the Introduction,many other supply

and demand parametersenter into the decision

to adopt open access besides journal quality.

JournalsCheaton

III. Will Open-Access

Quality?

As noted in the Introduction,critics of open

access suggest that open access may lead to a

corruptionof the editorial process. Because an

open-access journal obtains its revenue from

authors rather than readers, it may have to

charge high authorfees to be viable. Once high

authorfees are in place, the journalmay have an

incentive to publish many articles to boost rev-

enue, lowering editorial standardsif need be.

To address this issue of possible "overpub-

lishing" by open-accessjournals,we will exam-

ine a model in which the journal cannotcommit

to abide by the editorial standardof accepting

only those articles believed to be good. Rather,

the journal makes its editorial decision after

pricing, submission, and subscriptiondecisions

have been sunk. A commercial journal would

then make the acceptance/rejection decision

solely to maximize ex post profit. To make the

commitment problem as stark as possible, we

maintain a static model, abstractingfrom any

long-runconcerns for reputationthatmight mit-

igate the commitment problem.

The next propositionstates that, even in this

stark model, the commitment problem has no

bite: the journal can obtain the same profitif it

is not able to commit to an editorialpolicy as it

could if it were able to commit.

PROPOSITION 2. Let p*, p*, and p* be the

price scheme for a journal that can commit to

an editorial policy of only accepting articles

believed to be good, where without loss of gen-

erality, p* = 0. Let n* be the equilibriumnum-

ber of readers [i.e.,

= 1 - F([p* + ap]/y)].

, and

p*r

be the optimalprices

LettingPs

in the no-commitmentcase, a journal that can-

not commit to an editorial policy can obtain the

same profit as the journal that can commit by

= n*c +

setting p** =

a[nr(ba

ca],

pa**

Ca,

and pr* = Pr

As opposed to the case in the previoussection

in which the journal could commit to an edito-

rial policy (a case in which there were a whole

range of combinationsof submissionand accep-

tance fees thatcould provide an optimumfor the

journal), when the journal cannot commit to an

5 A few remarksaboutthe conditions behind Proposition

1 are in order. The condition

> p + c implies that an

author's benefit from having his article read exceeds the

generalized marginalcost of reading it, including the mar-

ginal cost of shipping the article to a reader, c, and the

reader's marginaleffort cost p. The second-ordercondition

associated with the objective function (6) holds if the slope

off is not too negative at an optimum. The condition holds

for variousdistributionsincluding the uniform.See McCabe

and Snyder (2004b) for details.

There exist examples in which a profit-maximizing

journal adopts open access if it has a low t but does not if it

has a high t. In particular,it can be shown that, assuming F

is the uniformdistributionon [0, 1], the journal adopts open

access if and only if

p + ba - c,

(

+ pc)(

c -

y) "

(ba

+ p - c)(1 -

This content downloaded from 193.0.118.39 on Thu, 5 Feb 2015 15:47:28 PM

All use subject to

JSTOR Terms and Conditions

Open Access and Academic Journal Quality.pdf

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: