subscription demand becomes less sensitive to advertising and as a result, the equilibrium
level of advertising will be higher with price-setting than with quantity-setting. The
reverse holds if consumers like advertising. Hence when γ > 0, price competition for
advertisers leads to a more competitive outcome for advertisers.
A similar intuition applies for the effect of the subscription price on demand. When
platforms fix the price of advertising and advertising is a nuisance, the subscription demand
is less price elastic than when platforms fix the amount of advertising. In this case,
Armstrong (2006) and Crampes et al. (2009) conclude that equilibrium subscription
prices will be higher. Hence when γ > 0, price competition for advertisers leads to a less
competitive outcome for consumers. In both models, the equilibrium level of advertising
with mixed financing is not affected by the nature of competition (price or quantity) for
advertisers.
29
2.3.8 Consumer Information
The other assumption that has been questioned and analyzed is that of the observability of
advertising levels by consumers before they patronize a media platform. Non-
observability may apply better in some contexts than others.
Gabszewicz and Wauthy
(2004, 2014)
consider equilibria where each side
30
does not observe the other side’s price
and activity levels, and holds fixed beliefs about the other side’s behavior (whose beliefs
are, however, consistent in equilibrium). These are referred to as passive beliefs.
31
To see
how this works, suppose that in our base model, consumers do not observe the amount of
advertising a
i
before they decide whether and which platform to consume. Then they
expect some level a
i
e
that would not vary if the platform actually decided to choose some
different a
i
. In this case of passive beliefs, the subscription demand is not responsive to the
actual choice of advertising and the profit is
Ra
i
ðÞN
i
γa
e
i
;γa
e
i
:
As demand is fixed from the platform’s perspective, the equilibrium level of advertising
maximizes the revenue per consumer, R(a), and so is set at the monopoly level a
m
sat-
isfying R
0
a
m
ðÞ¼0. Thus, under passive beliefs we obtain larger levels of advertising than
29
The reason is that for any given residual demand, it is optimal to set the advertising price so that the quan-
tity maximizes the joint surplus with the consumers R
0
aðÞ¼γ that is independent of the consumer market
share, under the assumption that advertisers’ value is linear in audience.
30
The sides are advertisers and consumers in the media context: Gabszewicz and Wauthy consider bilateral
positive externalities in the context of buyers and merchants holding and accepting credit cards. Their later
paper assumes single-homing while the earlier one considers multi-homing. See also
Section 2.4.3.
31
See also Ferrando et al. (2008) and Gabszewicz et al. (2012). The concept was introduced by Katz and
Shapiro (1985)
. Hagiu and Halaburda (2014) provide a general treatment in a two-sided
market allowing a mix between the two types of consumers’ expectations.
Hurkens and Lopez (2014)
look at belief formation in two-sided market in the context of telephony.
62
Handbook of Media Economics
when ad levels are observed. As emphasized by Anderson et al. (2013b, 2015b), this leads
to a hold-up problem. Once the consumer is committed to a platform, the platform sets
the monopoly advertising level.
Consumers rationally anticipate this high level of advertising and adjust demand to
N
i
(γa
m
, γa
m
). In particular, the aggregate participation of consumers will be lower under
passive beliefs, while there will be more advertisers.
2.3.9 Nonlinear Tar iffs and Insulated Equilibrium
In most industries involving some form of network externalities, firms develop business
strategies aiming at facilitating consumers coordination, raising the benefit of positive
network externalities or reducing the cost of negative externalities. The ability of firms
to cope with externalities and to capture the value created depends in particular on the
pricing instruments (see the discussion by
White and Weyl (2015)).
With two-sided externalities, it is common that platforms offer complex tariffs,
including tariffs that are (at least implicitly) contingent on the level of participation on
the other side of the market. Examples include click-through rates which condition
the total price paid by an advertiser on consumer participation, or credit card fees which
condition the merchant payment to the value of transactions. As pointed out by
Armstrong (2006), allowing for pricing contingent on the other side’s actions results
in the existence of multiple equilibria of the pricing game.
For example, suppose that in the advertising-financed regime, platforms offer two-
part tariffs to advertisers so that an advertiser pays P
i
¼f
i
+ p
i
N
i
. Then for given tariffs,
the advertising levels solve the system of equations
f
i
+ p
i
N
i
γa
i
;γa
i
ðÞ
¼va
i
ðÞ
N
i
γa
i
;γa
i
ðÞ
; i ¼1, 2, …: (2.3)
Let us fix all tariffs except for platform i, which has profit Ra
i
ðÞN
i
γa
i
;γa
i
ðÞwith the
constraints
va
j
¼
f
j
N
j
γa
j
;γa
j
+ p
j
, j 6¼i: (2.4)
Notice that only a
i
enters into the profit and the constraints. In particular, any combina-
tion of fixed fee f
i
and p
i
that satisfies condition (2.3) is a potential best-response of
platform i to the tariffs of competing platforms. With this degree of liberty, we can find
multiple equilibria.
To see this, suppose that all other platforms except i choose a zero fixed fee. Then for
any platform j 6¼i, a
j
¼v
1
p
j
is independent of platform i’s strategy. The problem of
platform i is to maximize va
i
ðÞN
i
γa
i
;γa
i
ðÞunder constraint (2.3). Any combination
of f
i
and p
i
that yields the desired level of advertising is privately optimal. Therefore plat-
form i may choose f
i
¼0. Thus there is an equilibrium where all platforms set zero fixed
63Two-Sided Media Markets
fees. Because in this case choosing p
i
is equivalent to choosing a
i
, this equilibrium coin-
cides with the equilibrium of the quantity-setting game discussed above.
Suppose instead that all platforms j 6¼i choose fixed fees with p
j
¼0. The problem of
platform i is to maximize va
i
ðÞ
N
i
γa
i
;γa
i
ðÞ
under constraints
(2.3) and (2.4). Again, any
combination of f
i
and p
i
that yields the desired level of advertising a
i
is optimal (and gives
the same values for a
i
). Thus platform i may choose p
i
¼0. When all platforms choose
p
i
¼0, the situation corresponds to the price-setting game discussed above.
As shown by
Armstrong (2006) for the case of pay media, there is a continuum of
equilibria that can be indexed by the slopes of the per-consumer tariff p
i
of each plat-
form.
32
This raises two issues for applications. In some contexts, the choice of tariff is
naturally guided by observation of business practices, as for credit cards or click-through
rates. In this case, one would like to understand the reasons that motivated platforms for
their choice of tariff, whether they are issues of implementation or more strategic con-
siderations. In other contexts, there is little guidance and we need some theory to
proceed.
To address this problem,
Weyl (2010) and White and Weyl (2015) propose the con-
cept of insulated equilibria. In their approach, a motivation for the choice of tariff is to
improve coordination between sides by offering tariffs that offset the effect of other side’s
participation on the decisions of agents. They refer to this concept as insulation.
To see the implications, consider the case of a monopoly platform. The participation
of consumers depends on their expectation of the advertising level. We saw above that
the outcome depends on whether consumers observe or not the advertising levels. The
monopoly would then want to achieve a given level of participation in a manner that is
robust to the nature of the coordination process. In the above model, this is simple to
achieve: the platform just has to offer to consumers a tariff contingent on the ad level,
taking the form f
i
γa
i
. In this case, the full price of watching the channel is f
i
indepen-
dent of the level of advertising so that the platform can anticipate demand N
i
(f
i
) irrespec-
tive of what happens on the advertising side. The advertisers’ participation can also be
“insulated” by setting a price p
i
per consumer (see below). Weyl (2010) refers to such
a tariff as an insulating tariff and shows that, for a monopoly, the choice of insulating tariffs
is equivalent to the choice of quantities on each side of the market.
White and Weyl (2015) extend the concept to oligopoly. Here platforms try to secure
their demand by “insulating” the demand on each side from the other side, which intu-
itively means that demand remains constant when demand on the other side changes (see
below for a precise statement). The difficulty faced by a platform is twofold. First, the
platform must account for the tariffs offered by competitors so that insulation can only
be for given competitors’ tariffs on the same side of the market (consumers here). Thus
32
Reisinger (2014) proposes a solution based on the heterogeneity of agents with respect to their trading
volumes.
64
Handbook of Media Economics
the platform can only insulate consumer demand against changes in tariffs to advertisers.
Second, the value of outside options for a consumer depends on the level of advertising
on all the channels. This means that the platform may need to make prices contingent on
all advertising levels, although we will see that this is not always the case. Formally, they
suppose that each platform can offer advertisers a tariff P
i
N
i
, N
i
ðÞcontingent on the
consumer allocation. On the other side, each platform can offer consumers a tariff
S
i
a
i
, a
i
ðÞcontingent on the vector of advertising levels on all channels.
33
They say that
platform i’s tariffs are insulating for given competitors’ tariffs if the following two con-
ditions hold:
(i) the tariff P
i
N
i
, N
i
ðÞis such that the advertising demand a
i
is independent of
N given the other advertising tariffs P
i
and
(ii) the tariff S
i
a
i
, a
i
ðÞis such that the consumer demand N
i
is independent of a given
the other consumer tariffs S
i
.
They show that an insulating best-reply exists for any tariffs of competitors and then
define an insulated equilibrium as an equilibrium of the competition game in tariffs such
that all platforms offer insulating tariffs.
Insulating tariffs have the property that the equilibrium is robust to assumptions about
the coordination of the two sides and formation of expectations. They can thus be viewed
as an appropriate tool for situations where platforms have enough instruments at their
disposal to overcome any coordination problem and implement any desirable allocation
on their residual demand curve.
To illustrate the concept, let us consider the mixed-finance regime above. Consider
first the advertisers. As the benefits of the marginal advertisers for a mass a
i
is v(a
i
)N
i
,itis
immediately apparent that an insulating tariff takes the form P
i
N
i
, N
i
ðÞ¼p
i
N
i
. Thus the
media can set a price per user p
i
(per reader/per viewer/per click). The amount of adver-
tising is then a
i
¼v
1
p
i
ðÞ, and it is independent of consumer demand. As advertisers have
a constant value per consumer, choosing the price per user or choosing the quantity a
i
are
equivalent.
34
Consider now the consumer side. The demand faced by platform i is then
N
i
S
i
a
i
, a
i
ðÞ+ γa
i
;S
i
+ γa
i
ðÞ:
Intuitively if all competitors propose a tariff S
j
¼f
j
γa
j
, the insulating tariff for platform i
is also of the form S
i
¼f
i
γa
i
because then the demand is N
i
f
i
; f
i
ðÞindependent of
advertising levels. It follows that the equilibrium that we derived before with platforms
33
Whether such contingencies are feasible may be debatable, but they argue that it can be seen as a reduced
form of a dynamic adjustment process.
34
This is valid only if the value of users of one platform is independent of other platforms, and thus if con-
sumers single-home.
65
Two-Sided Media Markets
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.