structures are restricted to MHAs together with either SHC or MHC. The former we
have already described, and it involves rational anticipation of monopoly ad levels.
The MHA–MHC equilibrium is the most intricate. One result (as follows from Foot-
note
45 above) is that there can be no symmetric equilibria even though the model is sym-
metric. Asymmetries are somewhat to be expected given that the vertical differentiation
model gives rise to asymmetric quality choices in its standard incarnation (e.g.,
Shaked and
Sutton, 1983
). These results above are analogous to those for the case of the model of
Gabszewicz and Wauthy (2004) for bilateral positive participation externalities.
The insight that the interaction between multi-homing and vertical differentiation
may be a source of asymmetry is nicely illustrated by
Calvano and Polo (2014).In their
duopoly model, as in
Ambrus et al. (2014), homogeneous advertisers choose advertising
intensity with decreasing returns to impressions, parameterized by the effectiveness of
each single impression. Consumers devote more or less time to each platform, with large
consumers multi-homing. They show that when each impression is very effective, and
thus multiple impressions not valuable, one platform chooses to be purely ad-financed
(and thus free to consumers) while the other charges consumers for a service without
advertising. The reason for this asymmetry is that once one platform proposes advertising,
the risk of multiple impressions and low incremental ad price deters the other from doing
so. According to the same logic as
Gabszewicz et al. (2012), the ad-financed platform is a
low-quality/negative-cost platform that sets a zero subscription price. The duopoly then
generates the same pattern of services as would a monopoly screening ad-averse con-
sumers from others with a free-of-ads service (
Tag, 2009).
Athey et al. (2014) point out that in practice platforms have limited ad capacity. Their
model differs from above as they focus on the issue of tracking (the ability to follow a
consumer’s behavior on the platform) and account for the fact that attention of viewers
is a scarce resource. In their model, MHCs spread their attention over the two platforms
(they switch), so that they are less likely to be reached than single-homers, and they com-
bine this effect with advertiser value from multiple impressions. Limited attention implies
that the media platforms are constrained in their supply of advertising slots (or impres-
sions) per consumer. With fixed SHC and MHC demands, they examine the game where
each platform chooses first advertising intensity and then prices adjust. Their asymmetric
equilibrium exhibits the same MHA–MHC pattern as described above, with low valu-
ation advertisers single-homing and high valuation advertisers multi-homing. In their
analysis, as more consumers switch between platforms, advertising capacity expands
and the platforms’ revenue decreases.
2.4.4 Information Congestion and MHCs
Information congestion in ads constitutes another channel through which multi-homing
can impact market performance. The congestion idea is that a consumer is less likely to
76 Handbook of Media Economics