choosing advertising levels and a subscription price (as in Anderson and Coate, 2005)isan
insulated equilibrium. Under some generic conditions, it can be shown to be the unique
insulated equilibrium.
35
This shows that insulated equilibria may have a simple and attractive structure in some
cases. In particular when consumers single-home, it provides some support to the
quantity-setting model as the price-setting game would not yield an insulating
equilibrium.
2.4. MULTI-HOMING VIEWERS/READERS
The models described in the previous section suppose that each media consumer chooses
only one platform. However, advertisers choose to place their ads on multiple channels.
The behavior of the consumers is known as single-homing while that of the advertisers is
called multi-homing. This (unfortunate) nomenclature comes from common usage in
the context of Internet Service Providers. The assumption that consumers single-home
gives rise to the competitive bottleneck property of the equilibrium (as discussed in
Section 2.3.3), which has several strong implications that may not hold in practice.
The competitive bottleneck and the ensuing predictions can change quite radically if
viewers watch several channels over the course of the relative product choice span. Or,
indeed, if readers subscribe to several magazines, or web-surfers go to several sites. Put
simply, if ad prices are high on one platform, advertisers can avoid it by reaching viewers
elsewhere, which just is not possible under single-homing. The competitive bottleneck is
defanged by competition for advertisers. Several different variants on this theme are
described below.
The presence of multi-homing viewers would not alter the analysis if the return to an
ad on one platform were independent of whatever ads are on the other platform. How-
ever, there are several reasons why a multi-homing viewer differs from a single-homing
viewer for advertisers.
First, when a consumer watches several channels or reads several newspapers, the level
of attention devoted to each platform may be lower, which may reduce the efficacy of
advertising. Following this logic,
Ambrus and Reisinger (2006) allow for the possibility
that an ad seen on a platform has diminished value if the consumer seeing it is multi-
homing. They assume that the value to an advertiser of such a consumer is lower than
for a consumer who single-homes.
36
Athey et al. (2014) point out that when multi-
homing viewers switch between channels, a single ad may reach a multi-homing viewer
35
A caveat is that if aggregate demand is fixed,
X
i
N
i
¼1, as in the Hotelling or Salop model, there is a
continuum of insulated equilibria. Thus the conclusion requires some aggregate demand elasticity.
36
They assume that the valuations are independent of how many ads are seen overall, so there is no
“information congestion” per se. A broader treatment might determine endogenously the value of an
impression on a multi-homer through a more explicit model of information congestion.
66
Handbook of Media Economics
with a smaller probability than a single-homing viewer who consumes all her content on
the same platform.
Second, the return on an ad placed on one platform may depend on whether the
advertiser is already reaching the consumer through another platform. While for
single-homers an ad placed on two platforms is guaranteed only unique impressions, this
is not so for multi-homing viewers. A second impression might have a lower value for the
advertiser than the first impression. This implies that the additional value for placing an ad
on a second platform will be lower if there are overlapping viewers.
We investigate below the consequences of consumer multi-homing when the value
of a second impression is less than the value of the first one. A first consequence is that the
price of advertising is depressed as advertisers worry that the ads they put on some plat-
form generate second impressions rather than exclusive ones (the latter is the case with
only SHCs). Building on
Anderson et al. (2015b), we start with a simple presentation of
the incremental pricing principle that underlies the analysis of MHCs by
Ambrus and
Reisinger (2006)
and subsequent work. We then discuss how this affects equilibrium
advertising levels depending on the context.
When demand is affected by advertising, we have the additional effect that if ad levels
are low enough, multi-homing is attractive to consumers. However, low ad levels also
entail high ad prices, and herein lies the confound. In particular, with high ad prices,
advertisers are less likely to want to pay for ads on several channels because adding a sec-
ond channel delivers some viewers they already reach on the first one. This effect fosters
competition in the advertising market.
The viewer side in the duopoly analysis of Anderson–Coate was drawn from the clas-
sic
Hotelling (1929) single-homing setup, whereby viewers watch the “closer” channel,
corrected by ad-nuisance costs (which play the role of prices to viewers).
Ambrus and
Reisinger (2006)
and other contributors assume that viewers who get a positive net
utility from both channels (the consumers “in the middle” of the Hotelling line)
multi-home and are exposed to ads from both. Thus the marginal consumer is indifferent
between one of the channels alone and multi-homing.
37
For simplicity, most authors
have assumed model setups such that the marginal consumer is indifferent between
multi-homing and single-homing, and we follow this approach because it appears to give
the strongest difference from the case of pure single-homing described in the previous
section.
38
37
One might think of preference configurations in which the marginal consumer is indifferent between
single-homing on two options, in which case the analysis would essentially be that of the SHC model
modulated by the competition for advertisers described in the next section, or indeed there might be mar-
ginal consumers on all markets.
38
Doganoglu and Wright (2006) consider various permutations of marginal (and infra-marginal) consumers
in a two-sided platform context with bilateral positive externalities.
67
Two-Sided Media Markets
2.4.1 MHCs and Incremental Pricing of Ads
To see most starkly the importance of MHCs on competition in the advertising market,
we start by assuming that the allocation of consumers to platforms is fixed. This would
hold true when consumers are indifferent to the presence of ads. We then indicate how
the results can be extended to include consumer preferences about ad content.
The role of multi-homing is brought out quite immediately in this context, and shows
quite transparently how it alters market equilibrium characteristics as regards the impact
of public platforms, entry, merger, etc. The analysis, which follows
Anderson et al.
(2015b)
, highlights the incremental pricing principle at work in the Ambrus and
Reisinger (2006)
analysis and subsequent papers.
There are n media platforms accessible for free (i.e., without a subscription price) to a
population of consumers (readers/viewers/listeners/surfers). Each platform is financed by
advertising. The key modification is to allow for consumers to watch more than one
channel/buy more than one magazine.
39
There is some heterogeneity in the population
so that some individuals may choose to consume from only one platform (single-homing)
while others choose to consume from both platforms. Let N
i
E
denote the number of
exclusive consumers that platform i has and let N
i
S
be the number of consumers i shares
with one other platform. Notice that we do not specify with which other platform they
are shared: we shall see that this does not matter (modulo the exception for a public
broadcaster not carrying ads) given the rest of the model setup. The total number of con-
sumers on platform i is N
i
N
E
i
+ N
S
i
.
40
The advertiser side is like Ambrus–Reisinger, extending to more than two platforms.
As is assumed in all the papers discussed below, advertiser valuations are independent
of how many ads are seen overall, so there is no “information congestion” per se. Sup-
pose that there is a mass of A advertisers, all with the same valuation for reaching
consumers, so each of the A advertisers is willing to pay v to contact a consumer.
Furthermore, a consumer reached more than once on a different platform is worth
v + βv so βv is the incremental value of a second impression. Impressions beyond two
have no further incremental value.
41
An ad seen on one platform is worth vN
i
because
it is viewed exclusively by all the viewers of the platform on which it is aired. An ad that
is seen on all platforms is worth v + βvN
S
, which is the full value of everyone reached
once plus the incremental value of those reached twice (recalling that the mass of
consumers is normalized to unity, and where N
S
¼ 1=2ðÞ
X
i
N
S
i
is the fraction
of viewers in the population shared one time). If an ad is placed on all platforms except
platform i, the ad on platform i generates two benefits. It brings a unique impression
to the exclusive consumers of platform i, and a second impression on consumers that
39
See also Kim and Serfes (2006) for a multi-homing demand model.
40
The difference between the LHS and the RHS corresponds to consumers shared with more than one
platform.
41
Anderson et al. (2015b) allow for further impressions to have value, but this is suppressed here.
68
Handbook of Media Economics
platform i shares with only one other platform. The incremental value of an ad on
platform i is the sum of these two benefits, vN
E
i
+ βvN
S
i
, which we shall shortly see in
the equilibrium ad price.
Assume that platforms first simultaneously set prices per ad, P
i
, i ¼1,…,n. Advertisers
then observe these ad prices, and then choose where to buy ads. Note first that it is an
equilibrium for all platforms to price at incremental value, and for advertisers to then
place ads on each platform. If all other platforms set their prices at their incremental
values, then any platform would increase profit by raising its price up to its own incre-
mental value, but would get nothing by pricing above its incremental value. Further-
more, this is the unique equilibrium: First note that all advertisers are on all platforms
in equilibrium, for if a platform had no adherent advertisers it would get no profit
and could certainly get something by pricing at the value of its exclusive viewers. But
then, if all advertisers are to be on all platforms, prices must be at the incremental values
that each platform delivers. Therefore, at the unique equilibrium, each platform sets a
price per ad P
i
¼vN
E
i
+ βvN
S
i
, i ¼1,…,n, so each advertiser places an ad on each plat-
form. This is the principle of incremental pricing, whereby each platform prices at the value of
its exclusive consumers plus the incremental value of those shared with just one other
platform. Note that any consumer shared with more than one other platform has no value
because they are already delivered to the advertisers at least twice elsewhere.
As pointed out by
Athey et al. (2014), the principle of incremental pricing may explain
why larger platforms charge higher prices per consumer, P
i
=N
i
¼vN
E
i
=N
i
+ βvN
S
i
=N
i
.
Indeed, this is the case if larger platforms share a smaller percentage of their demand
with other platforms so that the ratio N
i
E
/N
i
is higher.
Consider now (in this framework) the presence of a public broadcaster newly allowed
to air ads. Then consumers shared between the public broadcaster and one private one are
effectively converted from being exclusively deliverable to advertisers by the private
broadcaster, so devaluing them in the ad price and hence private profit. Similarly, those
shared three ways in a combination including the private broadcaster are reduced to zero
value. These effects are nuanced, as described below at the end of this section, when con-
sumer demand is ad-sensitive.
The effects of entry in this model depend on where the entrant picks up its consumer
base. If all its consumers are new consumers, entry does not affect existing platforms’
profits. If (as might be expected) a platform’s exclusive and shared viewers both fall with
entry because the market is more crowded, then ad prices fall with entry.
42
The effect on
ad prices per consumer is more subtle because it depends on changes in the composition
of consumers. This price goes down if there are proportionately more shared-once con-
sumers, or if it does not fall as much as the number of exclusives (which might be the
expected impact).
42
That is, if N
i
E
(A) and N
E
i
AðÞ+ N
S
i
AðÞboth decrease with entry.
69
Two-Sided Media Markets
Merger in this framework is quite easy to deal with. The idea is that a merged entity
can still put A ads on each platform so that the situation facing other platforms is
unchanged. However, the merged platform can now charge advertisers for access to con-
sumers that are now exclusive to the merged pair but were shared-once between them
before, so it can now charge v + βv for these, up from β v each before. It can also now
charge βv for any consumers attending the two merged platforms plus one other, for these
were worth nothing before.
2.4.1.1 Endogenous Viewer Choices
The results above readily extend to when consumers care about the advertising levels, but
do not observe them before choosing which platforms to attend. Recall that we argued in
Section 2.3.8 that, in this case, consumer demand depends only on expected quantities
and not realized quantities, so that with only single-homers, the platforms would choose
the monopoly quantity, here a
i
¼A (with monopoly price P
i
¼vN ).
In the presence of multi-homing demand, platforms will still choose the maximal
level of advertising a
i
¼A, so each advertiser places an ad on each platform.
43
Rational
consumers then expect this maximal level of advertising on each platform and choose
platforms accordingly. At the (unique) equilibrium, each platform sets a price per ad
P
i
¼vN
E
i
+ βvN
S
i
, i ¼1,…,n, where N
i
E
and N
i
S
correspond to demands at the expected
level of advertising A per platform. Thus the principle of incremental pricing still applies,
and each platform prices at the value of its exclusive consumers plus the incremental value
of those shared with just one other platform.
44
Notice that the price per ad per consumer, p
i
¼P
i
=N
i
, may increase or decrease with
the strength of advertiser demand, A, depending on the effect of ad nuisance on the share
of single-homers. Along similar lines, the price per ad per consumer decreases if the num-
ber of shared consumers goes up with entry, or if it falls less in percentage terms than the
number of exclusive consumers (
Anderson et al., 2015b).
The model with endogenous consumer choices also ties together the single-homing
and multi-homing c ases . For example, if ads are a nuisance to consumers, allowing a
public platform to air ads has two contradicting forces. First, if ads are a nuisance,
the public broadcaster will tend to lose consumers, and some will be picked up by
the private platfo rms. If there is little multi-homing going on, this effect helps the
private ones by expanding the base of consumers they can deliver to advertisers.
Conversely though, before allowing ads, any private broadc aster sharing consumers
43
For any consumer expectations of a, platforms will price at incremental values and so all advertisers will
advertise on all platforms. Thus the rational expectation is a ¼A .
44
The result readily extends along the same lines when impressions beyond the second have value.
70
Handbook of Media Economics
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.