A model of scholarly publishing with hybrid academic journals

Abstract

In April 2013, all of the major academic publishing houses moved thousands of journal titles to an original hybrid model, under which authors of accepted papers can choose between an expensive open access track and the traditional track available only to subscribers. This paper argues that authors might use publication strategy as a quality signaling device. The imperfect information game between authors and readers presents several types of Perfect Bayesian Equilibria, including a separating equilibrium in which only authors of high quality papers are driven toward the open access track. The publishing house will choose the open-access publication fee that supports the emergence of the highest return equilibrium. Journal structures will evolve over time according to the journals' accessibility - quality profiles.

Full text

A model of scholarly publishing with hybrid academic
journals
Damien Besancenot, Radu Vranceanu
To cite this version:
Damien Besancenot, Radu Vranceanu. A model of scholarly publishing with hybrid academic
journals. ESSEC Working paper. Document de Recherche ESSEC / Centre de recherche de
l’ESSEC. ISSN : 1291-9616. WP 1406. 2014 2014. <hal-00971541>
HAL Id: hal-00971541
https://hal-essec.archives-ouvertes.fr/hal-00971541
Submitted on 3 Apr 2014
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entific research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destin´ee au d´epˆot et `a la diffusion de documents
scientifiques de niveau recherche, publi´es ou non,
´emanant des ´etablissements d’enseignement et de
recherche fran¸cais ou ´etrangers, des laboratoires
publics ou priv´es.

Research Center
ESSEC Working Paper 1406
2014
Damien Besancenot
Radu Vranceanu
A Model of Scholarly Publishing with
Hybrid Academic Journals

March 23, 2014
A MODEL OF SCHOLARLY PUBLISHING WITH HYBRID
ACADEMIC JOURNALS
Damien Besancenotand Radu Vranceanuy
Abstract
In April 2013, all of the major academic publishing houses moved thousands of journal titles to an
original hybrid model, under which authors of accepted papers can choose between an expensive open
access track and the traditional track available only to subscribers. This paper argues that authors might
use publication strategy as a quality signaling device. The imperfect information game between authors
and readers presents several types of Perfect Bayesian Equilibria, including a separating equilibrium in
which only authors of high quality papers are driven toward the open access track. The publishing house
will choose the open-access publication fee that supports the emergence of the highest return equilibrium.
Journal structures will evolve over time according to the journals’accessibility - quality pro…les.
Keywords: Academic publishing, Open access, Knowledge di¤usion, Imperfect information, Signaling.
JEL Classi…cation: D52, L23, A14
University of Paris 13 and CEPN , 99 rue Jean-Baptiste Clément, 93430, Villetaneuse, France. E-mail:
besancenot.damien@univ-paris13.fr
yESSEC Business School and TH EM A, 105 Av. Bern ard Hirsch, 95021 Cergy, France. E-mail:
vranceanu@essec.fr

1 Introduction
Scholarly publishing of articles in peer-reviewed journals is by far the most important research com-
munication channel. Depending on the underlying economic model, there are two well-established
types of peer-reviewed journals: traditional and open access. Using an economic model established
three centuries ago, a traditional journal restricts its access only to subscribers. With the advent
of the Internet and the ubiquitous portable document format (pdf), traditional journals created
their own websites and, gradually, users shifted from reading the printed edition to downloading
and reading electronic versions of the journal provided that they were subscribers. Traditional
journals will grant free access to a very small number of high impact papers as a teaser for mar-
keting purposes. A small number of journals in Science allow for self-archiving, making the papers
available for free.
At the same time, the Internet supported the emergence of an alternative economic model for
scholarly publishing in so-called open access (OA) journals. OA journals maintain an editorial
board and run a more or less demanding paper selection process. They allow web users to freely
download papers; in turn, to cover the administrative costs and make a pro…t, many open access
journal will charge a publication fee to the authors. In December 2013, the Directory of Open
Access journals recorded 10.000 titles and this number is increasing steadily.1
Until very recently, a signi…cant majority of leading journals relied on the traditional model.
In the last twenty years the academic publishing business underwent a dramatic concentration
process, driven by successive mergers between large players and waves of acquisition of smaller
presses (Bowen et al. 2013). As a result, most academic journals are now owned by large size, for-
pro…t publishing companies. In 2006 - Reed Elsevier,Springer and Taylor and Francis - accounted
for 60% of all journal titles indexed in the ISI Web of Science. If we add the collections by Sage
and John Wiley and Sons (who acquired Blackwell in 2007), it is clear that these …ve Major
Publishing Houses (MPHs) have a tight grip on the traditional journal market.
With their increased market power, MPHs have the ability to implement more aggressive
1Data are available on the website of the D irectory of Open Access journals at www.doa j.org.
1

marketing strategies, such as journal bundling, which allows them to charge higher prices and
reap higher pro…ts (Edlin and Rubinfeld, 2005; Nevo et al. 2005; Dewatripoint et al. 2007).2In
the last two decades, subscriptions fees to traditional journals have risen dramatically, with a much
steeper increase recorded by for-pro…t journals compared to non-pro…t ones. As a consequence,
the a¤ordability of these journals has gradually declined. Limiting access to knowledge to only
those who can a¤ord to pay a high price has a dubious ethical dimension and can harm long-term
economic growth by setting a cap on human capital accumulation.
To avoid these charges while protecting their high return, in April 2013, the aforementioned
big-5 MPHs operated a coordinated move to replace the traditional model with a "hybrid" or
"two-track" academic journal model. All authors of accepted papers now have the choice between
the "traditional" track and an "open access" track (of the same journal). The traditional track
is free, but only readers who paid the journal subscription fee have access to the paper, either
via the Internet or in the printed version of the journal. If the author chooses the open access
track, he must pay a substantial publication fee, ranging between 500 to 5000 USD (most often
more than 1500 USD). In turn, the paper becomes freely available to all Internet users. Hence,
the composition of traditional journals might gradually evolve toward a hybrid structure, with a
subset of papers published under the traditional, restricted access track, and a subset of papers
published under the OA track. Because of the increased revenue ‡ow from authors, the MPHs have
already announced they will reduce the 2015 subscription fees for journals with a large number of
OA papers. For instance, Taylor and Françis indicates clearly that if "paid OA content makes up
95% of a journal’s content in a volume year, we will convert that title to pure OA in the following
year".3
While hybrid journals existed before 2013, they were seen as a scant curiosity. As note by
Walker (2004), the Entomological Society of America, which runs four important academic jour-
nals, was the …rst to sell Open Access by article in 2000, followed by the American Society of
Limnology and Oceanography, who adopted the hybrid model in 2003. The early experience of
2According to their …nancial statements, op erating pro…t margins of M PHs are close to 30%.
3See www.tandfonline.com/page/ op enaccess/funders# hybrid-pricing as accessed on M arch 1st, 2014.
2

these small presses was positive, with a substantial increase in net revenues, and the proportion
of authors that chose the OA track rose during the …rst years of adoption.
This paper aims to analyze the prospective evolution of scholarly publishing given the rise of
hybrid academic journals. As an original contribution to the literature on scholarly publishing,
it investigates whether authors can use the new publication options to signal the quality of their
papers, taking into account readers’beliefs and the pro…t-maximization behavior of the publishing
house. This research question is motivated by existing empirical analyses of the relationship
between the number of citations, as a proxy for academic impact, and the recourse to OA facilities
as provided by traditional journals. Early empirical analyses have shown that OA papers tend to
be cited 3 to 6 times more than restricted access papers (e.g., Lawrence, 2001; Harnad and Brody,
2004). However, Kurtz et al. (2005), Gaulé and Maystre (2011) and McCabe and Snyder (2013a;
2013b) found that the high academic impact of OA papers is explained by their better intrinsic
quality, and not the higher accessibility of the OA facilities. They conclude that the self-selection
bias can explain most if not all of the positive impact of OA on citations as observed in the early
studies.
The model is cast as an imperfect information game between authors of accepted papers,
readers, and one pro…t-maximizing MPH which manages one representative hybrid journal. The
MPH total revenue comprises publication fees collected on papers registered for the OA track,
plus the subscription fee. Because the OA track provides a better visibility of their work, authors
of high quality papers have an incentive to choose this track even if it means that they have to
pay the publication fee. However, in this case authors of low quality papers might try to bene…t
from the positive signal and follow the same publication strategy. As all signaling games, this
game presents several equilibria in pure and mixed strategies; it will be shown that the type of
equilibrium depends on the OA publication fee.4Hence the MPH can use the fee as an equilibrium
selection device; it will choose the fee that maximizes its total revenue. This adds a screening
level to the traditional signalling model used to analyze the behavior of authors and readers.
In this paper we assume that there is only one MPH, running one journal. Our approach
4We follow the classical methodology pioneered by Spence (1973).
3

is justi…ed since so far the competition among MPHs appears to be relatively weak. Should
competition in the publication market increase in the future, the strategic interaction between
MPHs must be taken into account. We also consider that the publication fee represents a cost for
the author who can a¤ord to pay for it. The degree of complexity of the problem would increase
without changing fundamentally if we assume that some of the authors of the good papers cannot
a¤ord to pay the fee. Furthermore, should authors’employers decide to cover the publication fee
in part or in full, a more sophisticated analysis would be required to account for the strategic
behavior of the employer.
Given that the generalized adoption of the hybrid model by the MPHs is relatively recent, there
are few theoretical papers to which we can directly relate our analysis. One notable set of studies
includes analyses by McCabe and Snyder (2005; 2007) and Jeon and Rochet (2010) who modelled
the academic publication market as a two-side platform with journals playing the mediation role
between authors and readers. Each side of the platform - authors and readers - bene…t from the
positive externalities of the other side: authors enjoy a larger readership and readers like journals
publishing many papers of a high quality. The quality of the papers depends on the resources
spent for paper selection. Editors, who aim at maximizing pro…ts, must arbitrate between a higher
subscription fee (reducing the number of readers) and a higher OA publication fee (reducing the
number of authors). Depending on the parameters of the problem, journals would opt for the
traditional model, for the OA model, or for the hybrid structure. While these studies explicitly
model the process of selecting top quality papers, they simplify the information structure of the
problem, either by considering that authors do not know the quality of their paper or that the
journal can detect this quality without error. Thus, they cannot study the signaling behavior by
informed authors. At di¤erence with them, our paper simpli…es on the paper selection process by
focusing on the accepted papers only, but builds on a more sophisticated, asymmetric information
structure. This context allows us to address the signaling / screening problem as emphasized by
the above mentioned empirical studies.5
5The editorial selection process could b e introduced along the lines of studies by Besancenot and Vranceanu
(2008) or Besancenot et al. (2012).
4

The rest of the paper is organized as follows. The …rst section introduces the main assumptions.
Section 3 presents the equilibria of the game. In section 4, we determine the MPH’s optimal pricing
and preferred equilibrium. The last section concludes.
2 Main assumptions
The model features three types of agents: a publishing house that manages one hybrid journal,
authors of accepted papers and readers.
The journal has accepted for publication a continuum of papers of mass one. Authors (of these
accepted papers) know the quality of their paper. They chose a publication strategy such as to
maximize the impact of their work.
Accepted papers are heterogenous with respect to their intrinsic quality. There are out-
standing quality papers (with a quality index Hfor "high quality")and (1 )standard quality
papers (with a quality index Lfor "low quality");where H> L. De…ning the quality of a paper
is a challenging task. In a broad perspective, a high quality paper is probably both rigorous and
relevant, i.e., its …ndings are useful to someone - peers in academia, decision makers or society at
large, be it today, or in the years to come. To acknowledge that high quality research is a scarce
good, we impose that  < 1=2:
An author can choose to have his accepted paper published either under the traditional, re-
stricted access track (Tstrategy) or in the new, open access track (Astrategy):The frequency
of authors of a high quality paper (H)who play the Astrategy is denoted by : The frequency
of authors of a low quality paper (L)who play Ais denoted by : Figure 1 presents the author
Decision Tree.
If the author opts for the open access track, he must pay an OA publication fee denoted by c:
There is no publication fee for the traditional track.
At the outset of the game, the MPH must post the OA publication fee (per article) and
the subscription fee; given the standard legal constraints of the publication market, it cannot
renegotiate the fee after observing authors’ strategy. The MPH chooses the pricing rule that
5

maximizes its income.6
Readers have only an imperfect information about the quality of a paper. They update their
beliefs about a paper quality after observing the publication strategy of the author and they will
learn the quality after reading the paper. The two dotted lines in Figure 1 indicate that, out of
the special case of a separating equilibrium, readers cannot unambiguously infer the true quality
of the paper from observing the publication strategy.
Figure 1: The Author’s Decision Tree
The author’s payo¤.
We assume that the scholar’s reward from a published paper depends on the impact of his
work, that is the ability of his research to have a bearing on fellow scholars, decision makers or
6We implicitly assume that the cost of pro cessing a pap er do es not vary from one track to another; we can set
it to zero without a substantial loss in generality.
6

society at large. Impact is assumed to depend on the paper’s accessibility, and its intrinsic and
perceived quality. More in detail, for a paper to have an impact, it must …rstly reach a wide
readership (Gaulé and Maystre, 2011). All things equal, potential readers have an easier access to
papers published in the OA track as compared to papers published in the subscriber-only section
of a journal. Next, high quality papers should have a larger social and academic impact than low
quality papers just because they address a more important topic, with a better methodology. But
since readers cannot gauge the quality of a paper before they read it, the impact of a paper also
depends on their beliefs about the quality of the paper. For instance, if readers believe that a
paper is of a high quality they will read it more carefully than if they believe the paper is of a
low quality; thus, taking the same low quality paper, should the readers believe the paper is of
high quality, the impact of the paper will be higher than if readers believe the quality is low (and
vice-versa).
The former assumptions about the determinants of "impact" can be captured by a simple payo¤
function. We de…ne the utility of an author of type i2 fH; Lgwho plays strategy S2 fA; T gas:
Ui(S) = S[(1 )i+E[jS]] 1Sc: (1)
The …rst term is the product between a measure of accessibility (S)and a measure of quality.
We assume that A> Tto acknowledge that the open access track facilitates readers’access to
the paper as compared to the traditional track. The measure of quality weights intrinsic quality
(i)and perceived quality (E[jS]);with parameter 2[0;1] shifting the balance between the
two terms. We assume that intrinsic quality contributes to impact more than perceived quality,
i.e.  < 0:5:Notice that expression (1 )i+E[jS]can be written in the equivalent form:
i+(E[jS]i);where grasps the "reputation e¤ect". Finally, the indicator function 1S
takes the value 1 if S=Aand 0 if else.
As already mentioned in the Introduction, one imperfect measure of a paper’s impact is the
number of citations it has received within a given period. Whether this accessibility e¤ect is strong
or weak, the topic is still open to debate. Using a sample of 100 journals in science and controlling
for quality, McCabe and Snyder (2013b) found a net impact of OA publishing on citations of 8%.
7

It should be noticed that empirical studies use citations data from the ISI Web of Science, which
covers only academic journals. In some …elds, such as economics and management, researchers also
seek to have a direct impact on practitioners and policymakers; this type of in‡uence is probably
not well captured by the number of citations in academic journals. In these special …elds, the
impact of OA can be much stronger.7
The payo¤ of the MPH
The MPH’s income is made up of publication fees collected on OA articles, and the subscription
fee.
The frequency of authors that choose the open access track is given by '= + (1 ):
For every paper published in the open access track, the journal will collect a fee c(and then the
paper can be downloaded for free). The total amount collected as a publication fee is 'c:
What can be the amount of the subscription fee? Since the open access track is available for
free for all Internet users, the subscription fee should be related to the number and quality of the
papers available in the traditional track. In order to keep the analysis simple but realistic, we
assume that the subscription fee is the product of the number of papers and their average quality,
i.e.: (1 ')E[jT]:
Thus the payo¤ function of the MPH can be written as:
R='c + (1 ')E[jT](2)
It is important to remark that the only strategic decision of the MPH is the choice of the OA fee.
As will be shown latter on, this variable determines the frequency 'and the average quality of
the papers available in the traditional track, E[jT].
3 Equilibria of the game
A Perfect Bayesian Equilibrium equilibrium of the sub-game between authors and readers is de…ned
as a situation where authors choose their optimal publication strategy given readers’beliefs, and
readers’ beliefs are correct given the optimal strategy of the authors. Furthermore, the MPH
7The number of citations reported by Google Schola r, a database that covers a much broader range of publication
outlets, would provide a better measure of such "so cietal" impact.
8

chooses the publication fee such as to maximize its revenue, given the equilibrium strategies of
authors and readers.
3.1 Conditional probabilities and authors’payo¤s
Before solving for the equilibria of the game between readers and authors, we use Bayes rule
to determine the conditional probability that a paper has a high quality given the publication
strategy S2 fT; Ag:
Pr[HjA] = Pr[AjH] Pr[H]
Pr[AjH] Pr[H] + Pr[AjL] Pr[L]=
 +(1 )(3)
Pr[HjT] = Pr[TjH] Pr[H]
Pr[TjH] Pr[H] + Pr[TjL] Pr[L]=(1 )
(1 )+ (1 ) (1 )(4)
Thus, the expected quality of a paper given S2 fT; Agis :
E[jA] = H+ (1 )L
 + (1 )(5)
E[jT] = (1 )H+ (1 ) (1 )L
(1 )+ (1 ) (1 )(6)
Introducing the expected quality in the general payo¤ de…nition (Eq. 1), after some calculations
we obtain the payo¤s for each type of author and each strategy:
UH(A) = A2
4H
(1 )HL
 + (1 )
3
5c(7)
UH(T) = T2
4H
(1 ) (1 )HL
(1 )+ (1 ) (1 )
3
5(8)
UL(A) = A2
4L+
 HL
 + (1 )
3
5c(9)
UL(T) = T2
4L+
(1 )HL
(1 )+ (1 ) (1 )
3
5(10)
We can study now the various equilibria of this game.
3.2 Separating equilibrium
We study …rstly the existence of a separating equilibrium where all authors of high quality papers
play the Astrategy and all authors of standard quality papers adopt the traditional Tstrategy.8
9

The equilibrium frequencies are = 1;  = 0 and conditional expectations (as de…ned here above)
become: E[jA] = Hand E[jT] = L:
Thus the equilibrium requires two necessary conditions:
UH(A)> UH(T),c < c1=AHT[HHL](11)
UL(T)> UL(A),c > c2=A[L+HL]TL:(12)
It can be easily checked that A> T)c2< c1. The equilibrium exists if c2[c2; c1]:
Given that '=and E[jT] = L;in this equilibrium the MPH payo¤ can be written as a
function of c:
Rsep(c) = c + (1 )L:(13)
3.3 Pooling Equilibrium A
We study the existence of a pooling equilibrium where all authors choose the OA track. In this
case, = 1;  = 1:If readers are rational, equilibrium beliefs must respect Pr[HjA] = ;we
further assume that should one author decide to deviate from the equilibrium strategy and play
the Tstrategy (refuse to pay the open access fee), readers will consider that the paper has a
low quality: Pr[HjT]=0:With this system of beliefs, the expected quality of the papers is:
E[jA] = H+ (1 )Land E[jT] = L:
For this equilibrium to occur, both types of authors must prefer the Astrategy to the
Tstrategy: This requires two following necessary conditions:
UH(A)> UH(T),c < A[H(1 )HL]T[H+LH](14)
UL(A)> UL(T),c < A[L+ HL]TL:(15)
We can easily check that the …rst condition implies the second one. Thus the necessary condition
for this equilibrium to exist is:
c < c3=A[L+ HL]TL:(16)
8It can b e shown th at the opp osite separating equilibrium do es not exist.
10

Given that all authors publish in the open access track ('= 1) in this equilibrium the MPH payo¤
function is
RP A(c) = c: (17)
3.4 Pooling Equilibrium T
In the opposite pooling equilibrium all authors would choose the traditional track, that is = 0;
= 0:A rational reader would consider that Pr[HjT] = and Pr[HjA] = 1 : We further
assume that an author who deviate and choose the A-strategy (pays the fee) would be perceived
as the author of a high quality paper: Pr[HjA]=1:According to this system of beliefs, the
expected quality of the papers are E[jT] = H+ (1 )Land E[jA] = H. This equilibrium
exists if authors prefer the traditional track whatever the quality of their paper:
UH(A)< UH(T),c > AHT[H(1 )HL](18)
UL(A)< UL(T),c > A[L+HL]T[L+ HL]:(19)
It can be checked that the …rst condition implies the second one. Thus the necessary condition
for this equilibrium to exist is:
c > c4=AHT[H(1 )HL]:(20)
In this equilibrium, the MPH payo¤ is a constant (it is independent of the publication fee):
RP T =E[jT] = H+ (1 )L:(21)
3.5 Hybrid Equilibrium 1
We consider the equilibrium where all authors of high quality papers play A (= 1), the others
randomize between the open access and the traditional strategy (0<  < 1):Given that the
traditional strategy signals that the paper is of a low quality, it will never be played by the author
of a high quality paper. The open access strategy can be adopted by both types. The expected
quality is E[jT] = Land E[jA] = H+(1)L
+(1):
There are two equilibrium conditions. It can be shown that UH(A)> UH(T)is always true.
The second one states that author of a low quality paper should be indi¤erent between playing
11

the Aand Tstrategy:UL(A) = UL(T):This allows us to determine the equilibrium frequency
of authors of low quality papers that will register for the open access track:
=
1
2
4HL
c(AT)LA13
5:(22)
It turns out that = 0 for c=c2and = 1 for c=c3as de…ned by Equations (12) and
(16). It can be also be observed that c3< c2:Thus this hybrid equilibrium exists for c2[c3; c2]:
Furthermore, in this equilibrium d=dc < 0: the proportion of authors of low quality papers who
adopt the Open Access strategy is declining with the open access fee.
The MPH payo¤ is:
RH1(c)=[+ (1 )]c+ [(1 )(1 )] L
=LAHL2
4Lc
c(AT)L
3
5:(23)
This gain depends on cas indicated by the derivative
dRH1
dc =AHLL[1 (AT)]
hc(AT)Li2(24)
For c2[c3; c2], there are two distinct cases:
If A>1 + Tthen dRH1=dc < 0:In this case, the MPH makes the highest pro…t if cis as
low as c3;but at c3the hybrid equilibrium is nothing else than the Pooling Equilibrium A, and
the payo¤ of the MPH is the payo¤ in the Pooling Equilibrium A as evaluated for c=c3.
If T< A<1 + T;then dRH1=dc > 0:In this case, the MPH should better push the fee up to
c2; but for c=c2;the Hybrid equilibrium takes the extreme pro…le of the Separating equilibrium,
and the MPH payo¤ is the same one obtained in the Separating Equilibrium as evaluated for
c=c2.
Thus, if the MPH aims at maximizing its payo¤, it will never aim to create the conditions of
existence of the Hybrid equilibrium 1 by choosing c2]c3; c2[. We can thus eliminate the Hybrid
equilibrium 1 from the subsequent analysis of the income maximization decision of the MPH.
3.6 Hybrid Equilibrium 2
In this case, all authors of low quality papers adopt the Tstrategy (= 0) and authors of high
quality papers randomize between the traditional and the open access track; 0<  < 1. Since
12

only high quality papers are presented in open access, this publication strategy reveals the quality
of the paper. The traditional publication strategy can be adopted by both authors of high and
low quality papers. The expected quality is E[jA] = Hand E[jT] = (1)H+(1)L
(1)+(1):
There are two equilibrium conditions. The …rst UL(A)< U L(T)is ful…lled whatever the
parameter values. The second indi¤erence condition UH(A) = UH(T)de…nes the equilibrium
proportion of high quality papers that is presented to the open access track:
= 1 (1 )

2
4THL
c(AT)H13
5:(25)
We can check that for c=c4we get = 0;and for c=c1we get = 1 with the two cuto¤ costs
as de…ned by Equations (11) and (20). It can also be noticed that c4< c1. Thus, this hybrid
equilibrium is feasible for c2[c4; c1]:
Furthermore, it turns out that d=dc > 0 : The proportion of authors of high quality papers
who go for the open access track is increasing in the OA publication fee.9
In order to study the MPH income maximization problem, we …rstly determine E[jT]replacing
by its expression in Eq. (25):
E[jT] = (1 )H+ (1 )L
(1 )+ (1 )=Hc[AT]H
T
:(26)
The MPH revenue is:
RH2(c) = 'c + (1 ')E[jT] = c+ (1 )HL2
4THc
c(AT)H13
5:(27)
It can be shown that, depending on parameters, this revenue is monotonically increasing or de-
creasing with respect to c:
dRH2
dc = 1 + T(1 )HLH(AT)1
hc(AT)Hi2(28)
There is no internal solution to the revenue maximization problem in the Hybrid equilibrium 2.
Should the market settle to this equilibrium, the MPH would either try to obtain the Separating
equilibrium, by pushing the fee up to c1, or would seek to obtain the Pooling Equilibrium T by
9This prop erty p oints out that this hybrid equilibrium is unstab le: any min or expectation shock would push
auth ors of high quality pap ers to shift massively either to the Tstrategy or to the Astrategy:
13

decreasing the fee to c4;and obtain the (constant) revenue RP T as de…ned in Eq.(21). In the
subsequent analysis of the optimal decision of the MPH, we can thus discard any payo¤ speci…c to
the Hybrid equilibrium 2 and focus on the highest payo¤ obtained in the pure strategy equilibria.
Finally, it can be shown that an equilibrium where both type of authors randomize between
the open access and the traditional track is impossible; conditions UH(A) = UH(T)and UL(A) =
UL(T)cannot be ful…lled simultaneously.
4 The MPH optimal pricing policy
4.1 The OA publication fee and the equilibrium existence
As already mentioned, the publishing house uses the OA publication fee (c)as a screening device;
more precisely, it will tune the fee such as to create the conditions to obtain the type of equilibrium
entailing the highest payo¤. Recall that the payo¤ comprises the journal subscription and the
collected OA fees.
Given that the equilibrium of the game depends on the OA publication fee, in order to solve
the MPH payo¤ maximization problem we need (1) to see what equilibria are feasible when c
takes various positive values, and then (2) to compare the maximum payo¤s across the feasible
equilibria. In the previous section we have shown that the set of "relevant" equilibria comprises
the Separating Equilibrium and the two Pooling Equilibria. Indeed, the two Hybrid equilibria are
payo¤-dominated by the Pooling Equilibrium T or the Separating Equilibrium.
We have also de…ned four threshold values of the OA fee, delineating the boundaries between
equilibria:
c1=AHT[HHL]
c2=A[L+HL]TL
c3=A[L+ HL]TL
c4=AHT[H(1 )HL]
We have already checked that for  < 1and A> T;the following inequalities always hold:
c1> c2(see Separating equilibrium), c2> c3(see Hybrid equilibrium 1) and c4< c1(see Hybrid
14

equilibrium 2). Moreover, we can show that c3< c4(for  < 1=2):Thus the thresholds can be
ranked as:
c3<fc2; c4g< c1:
It can be shown that c4> c2if A> Th1 + 
(1)iand c4< c2if T< A< Th1 + 
(1)i:10
Figure 2 displays the equilibrium frequencies and as a function of the OA fee, in the
case where c4> c2. For a OA fee 0< c < c3;only the Pooling Equilibrium A would prevail
(= 1;  = 1); in this equilibrium, the MPH would maximize its payo¤ by choosing c=c3:
For c3cc2Hybrid equilibrium 1 would prevail (= 1;  2[0;1]); yet we have shown that
the MPH obtains the highest income at the extremes of the interval, that is either obtaining the
payo¤of the Pooling Equilibrium A or the Separating Equilibrium. For c2< c < c4the Separating
equilibrium is obtained (= 1;  = 0). For c4cc1the game presents a typical situation of
multiple equilibria: any of the Separating, Hybrid 2 of Pooling T equilibria can exist. Which one
will actually materialize depends on readers’beliefs and how the MPH has managed the OA fee.
We have shown that in the Hybrid equilibrium 2, the MPH would prefer a fee at the boundaries
of the interval; the actual choice is between c1implementing the Separating equilibrium, or c4
implementing the Pooling Equilibrium T. If the Separating equilibrium brings the highest income,
the MPH can make sure that readers’beliefs support this equilibrium by setting …rstly a fee lower
than c4;then rising it gradually to c1. Finally, by setting the OA fee above c1the MPH is sure to
obtain the Pooling Equilibrium T and its constant subscription fee.
So in the Pooling Equilibrium A the MPH obtains the maximum payo¤ if it pushes the OA
fee up to c3:According to Eq. (17) this highest payo¤ is:
RP A(c3) = A[L+ HL]TL:(29)
According to Eq.(13), in the Separating Equilibrium the MPH gets the highest payo¤ if it brings
the OA fee up to c1:Its expression is:
RSep (c1) = c1+ (1 )L=AHT[HHL]+ (1 )L:(30)
10 G iven that = (1 )<1;a su¢ cient condition for c4> c2is A> T(1 + ):Thus chances th at c4> c2are
high if the proportion of outstanding papers ()is small.
15

Figure 2: Equilibrium frequencies and 
Finally, the highest payo¤ the MPH can get in the Pooling Equilibrium T is RP T as de…ned in
equation (21). To obtain it, the MPH should push the OA fee above c1;in which case no author
will resort to the OA facility, and the journal can ask for the maximum subscription fee:
RP T =H+ (1 )L:(31)
Notice that in the opposite case where c4< c2the region where the Hybrid equilibrium 2 can exist
would overlap with the separating equilibrium, but this does not modi…es the decision problem of
the MPH, nor the optimal fees. For the sake or parsimony we do not display this graph.
4.2 The MPH preferred equilibrium
In order to determine the global optimum, we need to compare two by two the highest payo¤ in
each equilibrium as de…ned in Equations (29), (30), (31). Of course, the payo¤ ordering depends
on all the parameters of the problem. In order to guide intuition, we need to restrict the parameter
space to a tractable number. We thus choose to emphasize the quality gap and the accessibility
gap. By quality gap we understand the relative position of Hwith respect to L:The accessibility
gap refers to the relative position of Awith respect to T.
Table 2 presents the payo¤ dominance conditions as imposed on accessibility; we must distin-
guish between the case of high quality gap and a low quality gap. The payo¤ inequalities depend
on the position of Arelative to a term that is a linear function in T;these functions will be
represented graphically as the AA0,BB0and CC 0lines (as denoted in the last column of Table
16

2).
MPH preferred
payo¤
High quality gap
H> L1
 
Small quality gap
L< H< L1
 
RP A > RP T A>(HL)+L
(HL)+L+TL
L+(HL)A>(HL)+L
(HL)+L+TL
L+(HL)(BB0)
RSep > RP T A>1 + Th11L
Hi A>1 + Th11L
Hi (AA0)
RSep > RP A always true A<(1)L
((HL)H+L)+T(CC0)
Ordinates and coe¢ cients
1<(HL)+L
(HL)+L1<(HL)+L
(HL)+L<(1)L
((HL)H+L)
11L
H>L
L+(HL)11L
H<L
L+(HL)
Table 2. Payo¤ dominance conditions
Figures 3 and 4 present a map of the equilibria that outline regions where one equilibrium
dominates the two other equilibria in terms of payo¤s. This equilibrium brings the highest payo¤
as compared with the other relevant equilibria and should be the preferred choice by the MPH.
To obtain this equilibrium the MPH must chose the submission fee that supports it. For instance,
if the highest payo¤ is brought by the Separating equilibrium, the MPH will set c=c1:
The "high quality gap" case is presented in Figure 3. Because A> T;the area below the
45line is not relevant. The Separating equilibrium is always to be preferred to the Pooling
Equilibrium A (the frontier separating the the two regions is below the 45line). The AA’ line
delineates the frontier between the Separating and the Pooling Equilibrium T. Above this line,
the separating equilibrium dominates both pooling equilibria. Below this line, we know that the
Pooling Equilibrium T is to be preferred to the Separating Equilibrium. A quick look to the BB’
line shows that the Pooling Equilibrium T should also be preferred to the Pooling Equilibrium
A. So, in the OAA’area, the preferred equilibrium is the Pooling Equilibrium T. Above the AA’
frontier, the preferred equilibrium is the separating one.11
For relatively small values of T, if the accessibility gap between Tand Ais relatively narrow,
11 T he slope of BB’is lower than the slop e of AA’. It can be checked analytically that they cannot cross ab ove
the 45line (which would imply a logical contradiction, w here RP A > RP T > RSE P > RP A).
17

Figure 3: Equilibrium map: The high quality gap case
the MPH would prefer an equilibrium where all authors follow the traditional track, and they keep
the "historical" subscription fee. If the gap widens, it might become interesting for the journal
to seek the Separating equilibrium. If Tis large enough, the Separating equilibrium would be
preferred whatever the accessibility gap.
In the same way, Figure 4 analyzes the "low quality gap case"; in this con…guration, for a
relatively low T;a high accessibility gap supports the Pooling Equilibrium A (above the CC’line);
an intermediate accessibility gap entails the Separating equilibrium, and, for a low accessibility
gap, we get the Pooling Equilibrium T. If Tis large enough, the Pooling Equilibrium T is no
longer possible.
Thus far, we have considered that the MPH is managing one representative journal. In reality
journals cover a di¤erent content and target a di¤erent readership. If MPHs tune the OA fee to
18

Figure 4: Equilibrium map: the low quality gap case
maximize the income per journal, their publication patterns may evolve toward di¤erent equilibria
depending on the speci…c parameters of each journal. Our analysis can help in understanding the
future change in journal structures.
For instance, if we agree that "leading journals", i.e. journals unanimously recognized as the
best in a given …eld, have both a low quality gap because of the high standards of the editorial
screening process, and a low accessibility gap because experts interested in these journals have
easy access to the traditional track (Gaulé and Maystre, 2011), then our analysis suggests that
these journals will evolve towards the Separating equilibrium (Figure 4, Tlarge, and Aclose to
A).
Specialized expert journals also have a low quality gap (Figure 4), but not all libraries have a
subscription; thus their open access track should improve paper accessibility in a signi…cant way.
19

With a large accessibility gap, these journals would either move toward the Separating equilibrium,
or, for a very strong accessibility gap, toward the Pooling equilibrium A.
Finally, second-tier journals are characterized by a high quality gap (Figure 3) and a high
accessibility gap, for a relatively small T; these would move toward the Separating equilibrium
(or the Pooling Equilibrium T).
Table 3 summarizes these scenarios, which are not exhaustive of the whole range of journal
pro…les.
Journal types T
accessibility gap
(AT)
quality gap
(HH)
MPH preferred equilibrium
leading high low low (Fig. 4) Separating Equilibrium
specialized (good) high high low (Fig. 4) Separating Eq.; Pooling Eq. A
second-tier low high high (Fig. 3) Separating Eq.; Pooling Eq. T
Table 3. Journal types and preferred equilibria
Considering these scenarios, it is quite reasonable to expect that a large number of academic
journals will move away from the traditional economic model toward a hybrid form (separating
equilibrium) or genuine open-access. In these two cases, the income of the MPH will increase, as
will increase the accessibly of scholarly publishing, with an overall positive impact on knowledge
dissemination.
5 Conclusion and welfare implications
In April 2013 all of the major publishing houses moved thousands of academic journal titles away
from the traditional economic model to a hybrid economic model, allowing for an expensive OA-
per-article option. This paper aims to analyze the consequences of the rise of the hybrid economic
model on scholarly publishing. Empirical studies that analyzed the impact of OA facilities as
provided by the "old" traditional journals on citations have already revealed that high quality
papers tend to self-select for the open access model. Taking stock of this empirical …nding, this
paper investigates under which conditions authors can use the publication strategy as a signaling
20

device.
In our model, because paper accessibility is stronger in the OA track than in the traditional
track, authors of high quality papers would pay the high OA fee only to signal their type. Whether
authors of low quality papers follow them or not, depends on readers’beliefs and the publication
fee as decided by the MPH. If many authors of standard quality papers copy the strategy of the
high quality paper authors, the incentive of the latter group to choose the OA track declines.
The MPH uses the publication fee as an equilibrium selection device tuning it to maximize its
income. We solved the game for the various equilibria and determined the optimal publication
fee in each equilibrium. The relevant equilibria include a Separating equilibrium –where authors
of high quality papers self-select to the OA track and authors of low-quality papers self-select for
the traditional track –and two Pooling equilibria, where all authors choose one of the two pure
strategies (either OA or traditional). Finally, we present an equilibrium map depending on the
quality gap and accessibility gap. This allows us to anticipate the possible evolution of a journal
structure, depending on its pro…le.
One important …nding of this analysis is that, with the exception of second-tier journals, many
homogenous quality journals would have an incentive to move toward the hybrid model; and
perhaps even toward the more extreme full open access structure. This move will come with
higher revenues for the MPHs, a reduced burden for libraries, and high costs for authors.
Raising income for the MPHs and o¤-setting raising publication costs for authors and their host
Institutions is the mark of the excessive market power of MPHs; this trend can be interpreted as
a negative for knowledge dissemination. Nonetheless, the fact that leading journals might extend
their OA track can be seen as good news for knowledge dissemination, given the improved access
for scholars working in poorer regions.
Furthermore, the emergence of the Separating equilibrium where high quality papers will select
for the OA track can also contribute to increase the overall e¢ ciency of research, given that readers
of journals evolving in this direction will now have a powerful tool to detect quality, and can avoid
investing time and e¤ort in reading low quality papers.
21

References
Bowen, Elizabeth A., Mark A. Mattaini and Sandra L. De Groote, 2013. Open access for
social work research and practice, Journal of the Society for Social Work and Research 4
(1): 31–46.
Besancenot, Damien, Kim V. Huynh, and Joao R. Faria, 2012. Search and research: the
in‡uence of editorial boards on journals’quality, Theory and Decision 73 (4): 687–702.
Besancenot, Damien, and Radu Vranceanu, 2008. Can incentives for research harm research?
A business schools’tale. Journal of Socio-Economics 37 (3): 1248–1265.
Dewatripont, Mathias, Ginsburgh, Victor, Legros, Patrick, and Alexis Walckiers, 2007. Pric-
ing of scienti…c journals and market power, Journal of the European Economic Association
5 (2-3): 400–410.
Edlin, Aaron S., and Daniel L. Rubinfeld, 2005. The bundling of academic journals. American
Economic Review 95 (2): 441–446.
Gaulé, Patrick and Nicolas Maystre, 2011. Getting cited: does open access help?, Research
Policy 40 (10): 1332–1338.
Harnad, Steven and Tim Brody. 2004. Comparing the impact of Open Access (OA) vs.
non-OA articles in the same journals, D-Lib Magazine 10 (6).
Jeon, Doh-Shin and Jean-Charles Rochet, 2010. The pricing of academic journals: A two-
sided market perspective, American Economic Journal: Microeconomics 2 (2): 222–255.
Kurtz, Michael J., Eichhorn, Guenther, Accomazzi, Alberto, Grant, Carolyn, Demleitner,
M., Henneken, E. and Stephen S. Murray, 2005. The e¤ect of use and access on citations,
Information Processing and Management 41 (6): 1395–1402.
Lawrence, Steve, 2001, Free online availability substantially increases a paper’s impact,
Nature 411: 521.
McCabe, Mark J., and Christopher M. Snyder, 2013b. Identifying the e¤ect of open
access on citations using a panel of science journals, mimeo, University of Michigan,
http://mccabe.people.si.umich.edu/McCabe_Snyder_EI_2013.pdf.
McCabe, Mark J., and Christopher M. Snyder, 2013a. Does online availability increase
citations? Theory and evidence from a panel of economics and business journals, Review of
Economics and Statistics, In print.
McCabe, Mark J., and Christopher M. Snyder, 2007. Academic journal prices in a digital
age: A two-sided market model, BE Journal of Economic Analysis and Policy 7.(1).
McCabe, Mark J., and Christopher M. Snyder, 2005. Open access and academic journal
quality, American Economic Review 95 (2): 453–459.
Nevo, Aviv, Daniel L. Rubinfeld, and Mark McCabe, 2005, Academic journal pricing and
the demand of libraries, American Economic Review 95 (2): 447–452.
Spence, Michael, 1973. Job market signaling, Quarterly Journal of Economics 87 (3): 355–
374.
Walker, Thomas, 2004. Open access by the article: an idea whose time has come? Nature
Web Focus, 15 April, 1004, http://www.nature.com/nature/focus/accessdebate/13.html.
22

Alison Bougi
+33 (0)1 34 43 33 58
bougi@essec.fr
www.essec.fr
research.center@essec.fr
ISSN 1291-9616