Review of Industrial Organization (2005) 26:461–487 © Springer 2005
Airline Code-share Alliances and Costs:
Imposing Concavity on Translog Cost
CHEW LIAN CHUA
, HSEIN KEW
and JONGSAY YONG
MIAESR, L7, 161 Barry Street, The University of Melbourne, Parkville, Vic 3010,
National University of Singapore
Abstract. This paper provides an assessment of how airline code-share alliances affect the
costs of the airline industry. It makes two contributions to the literature. First, it mea-
sures the effects of airline alliances by estimating a translog cost function using a panel
dataset of 10 major U.S.-based airlines over 29 quarters. Secondly, it ensures concavity of
the estimated cost function by using the procedure suggested by Ryan and Wales (2000,
Economics Letters 67, 253–260). A conventional translog cost function is ﬁrst estimated
and scale estimates are computed. Unfortunately, the estimated function fails the curva-
ture requirement, which makes interpreting the estimated effects of alliances somewhat
dubious. Hence, we re-estimate the cost function by imposing local concavity restrictions.
We ﬁnd that large alliance partners have a small negative effect on airlines’ costs, but
small alliance partners’ effect on costs appear to be positive, although the magnitude
is negligible. We also ﬁnd material differences in the estimates of scale economies after
imposing local concavity.
Key words: Airline alliances, concavity, costs, translog.
JEL Classiﬁcations: L00, L12, L93.
A prominent development in the airline industry has been the prolifera-
tion of alliances in recent years. This paper attempts to assess the effect
on costs of a particular type of alliances, namely code-share alliances. In
a code-share arrangement, member airlines can use each other’s designator
code, and often the agreement also calls for coordination of ﬂight schedules
and other services, so as to create the impression that passengers are using
the services of the same airline. These arrangements provide two impor-
tant advantages as (i) most passengers want to avoid inter-line transfers;
Author for correspondence. E-mail: firstname.lastname@example.org