and ballet (segment B). A critical assumption, which will be relaxed in Section 6.3.4, is that
viewers are single-homing. Now, suppose that 90% of viewers want to watch football pro-
grams and 10% want to watch ballet. If their favorite genre is not aired, then they do not
watch any TV at all. It is then clear that with only two channels, both will offer segment
F (football) and get 45% of the viewers each (assuming equal split between identical chan-
nels). The ballet lovers will be left unserved. If the number of channels increases from two
to three, we still have duplication of genres. All deliver football, now to 30% of the viewers
each. If we continue this line of reasoning, we find that we need nine independent single-
channel platforms before one of them will deliver ballet. Then, they will be indifferent
between football and ballet since both choices provide them with 10% of the viewers.
Steiner argues that a change from two competing single-channel providers to one
multiple-channel provider increases product variety. The story is simply that a multi-
channel platform is concerned about the total number of viewers, and so has no incen-
tives to air two identical channels.
14
Accordingly, when considering consequences of
mergers (or entries), we need to make a distinction between the number of channels
and the number of competing firms.
If we return to the case with two single-channel platforms, the lesson from
Steiner
(1952)
is that duplication prevails as long as
v
A
2
> v
B
,
where v
i
, i¼A,B, is the fraction of consumers in segment i. Waldfogel (2009) terms this
concept “preferences externalities” which refer to the fact that the majority overrides the
minority when there are few alternatives (see
Anderson and Waldfogel, 2015, this volume).
To compare with more recent contributions, it is straightforward to reformulate
Steiner’s model into a standard Hotelling framework with uniform consumer distribu-
tion. If consumers single-home, two ad-financed rivals will locate in the middle of
the line (genre duplication) when consumers are ad-neutral. This was labeled the prin-
ciple of minimum differentiation by
Boulding (1955).
In an extension to more than two firms,
Eaton and Lipsey (1975) find that the results
of
Boulding (1955), and other extensions of Hotelling (1929), might be sensitive to the
number of firms, distribution of consumers, and changes in conjectural variations. Eaton
and Lipsey show that in a one-dimensional model, the principle of minimum differen-
tiation does not survive if we have more than two firms. However, when they extend
their model to a two-dimensional space they arrive at the principle of local clustering;
an entrant locates as close as possible to another firm. Thus, the principle of minimum
differentiation in the two-firm case is considered by Eaton and Lipsey as a special case
of the principle of clustering (or principle of pairs as they also label it).
15
14
Other early contributions that indicate such a duplication result are Rothenberg (1962) and Wiles (1963) .
15
See Hannesson (1982) for a critical assessment of the assumption made in Eaton and Lipsey (1975) on
conjectural variations.
238
Handbook of Media Economics