Review of Industrial Organization 14: 321–335, 1999.
© 1999 Kluwer Academic Publishers. Printed in the Netherlands.
Upstream-Downstream Specialization by Integrated
Firms in a Partially Integrated Industry
Département de sciences économiques and Centre de recherches sur les transports, Université de
Montréal, C.P. 6128, Succursale Centre-ville, Montréal, Québec, Canada H3C 3J7
NGO VAN LONG
Department of Economics, McGill University and CIRANO, Montréal, Québec, Canada H3A 2T7
GREQAM and Faculté des sciences économiques, Université de la Méditérranée, 13290 Les Milles,
Abstract. We propose a simple model of a partially integrated industry which explicitly takes into
account persistent production cost differences across upstream ﬁrms, such as one might observe
in natural resource industries. The model allows us to highlight the respective roles of strategic
considerations and of cost considerations in the determination of an integrated ﬁrm’s interaction with
the non-integrated sector of the industry and, in the end, on its relative upstream-downstream spe-
cialization. Some crude stylized facts from the world oil industry are used to motivate and illustrate
Key words: Oligopoly, vertical integration, specialization, oil industry
Studies of vertical integration have typically modeled the integration decision as a
dichotomic decision where, although some ﬁrms may be integrated and some not,
all integrated ﬁrms behave in the same way, as do all non-integrated ﬁrms
many industries this is a reasonable simpliﬁcation, since there is usually no reason
to believe that either upstream or downstream cost asymmetries can persist in the
long run and there is therefore no a priori justiﬁcation for the integrated ﬁrms not
to behave identically.
We thank the participants at the Ecole de printemps, GREQAM, Universit
e d’Aix-Marseille II,
May 25-27, 1995, for their comments on a preliminary version of this paper. Gaudet and Long wish
to acknowledge the ﬁnancial support of the Fonds FCAR du Gouvernement du Qu
ebec and the Social
Science Research Council of Canada.
See for instance the papers by Bonanno and Vickers (1988), Gaudet and Long (1996), Hart and
Tirole (1990), Ordover, Saloner and Salop (1990), Salinger (1988), as well as the survey by Perry