mixed financing, equilibrium profits are a hump-shaped function of the strength of
advertising demand. A weak advertising demand is bad news for platforms, but so too
is a strong one because then profits are dissipated through high investments in quality
to try and attract consumers. Another intriguing result (reminiscent of
Grossman and
Shapiro, 1984
) is that platform profits are increasing in the marginal cost of quality invest-
ment. This comes from the strategic effect of softening competition.
Anderson (2005) looks at an asymmetric model of quality investment in which
one platform has a central role (like a “hub” or a Lowest Common Denominator)
and competes in local markets with local platforms. The central platform (think Clear
Channel radio, or Hollywood movies) competes in all local markets but a local platform
(Welsh language radio, or Bollywood) is also present in each local market. The structure
in each local market is like the
Armstrong and Weeds (2007) setup, i.e., “Hotelling” seg-
ments with ad levels being set in local markets by both the local and the global compet-
itor, but all such local segments are effectively connected through the hub. The global
producer here has an economy of scale in quality provision because its quality is
“one-size-fits-all” and applies to all the local markets in which it competes. However,
the local market decisions (advertising levels) are tailored to each market. Through
the quality choice of the global platform, there are externalities between local producers
even though they do not interact directly. In equilibrium, the large platform chooses
higher quality than the local ones because it can spread its costs of providing quality over
all the local markets. Each local producer’s quality and audience share is larger in larger
local markets, and so the disparity is largest in the smallest markets.
Kerkhof and Munster (2015) analyze a different “quality” margin. They assume that
advertisers’ willingness-to-pay to contact consumers is decreasing in a quality variable
that consumers find desirable. For example, consumers may appreciate a serious docu-
mentary, but the framing effect of embedding ads may make them less willing to buy
frivolous products. The platforms then face a classic sort of two-sided market trade-off
that what is good for extracting revenue from advertisers (here “low” quality) is bad
for delivering the viewers’ eyeballs.
Kerkhof and Munster (2015) argue that a cap on
advertising in this context can be welfare improving. Their mechanism is thus different
from the standard arguments about ad caps (e.g.,
Anderson, 2007).
59
The models so far discussed assume single-homing by media consumers. As we argued
in the previous section, MHCs can be worth substantially less (in a platform’s consumer
portfolio) than SHCs. This effect can lead to bias in platform positions to favor catering to
SHCs and against catering to MHCs. Since the single-homers are worth more, platforms
will strive to deliver such exclusive viewers while avoiding overlapped consumers. The
theme is developed in
Anderson et al. (2015b) in a spatial duopoly model. As noted
59
The latter paper also analyzes the effects of ad caps on platform quality choice.
83
Two-Sided Media Markets
earlier, in the context of the Hotelling model, the MHCs will be those in the middle of
the market. The less valuable these are (e.g., the lower the value of a second ad impres-
sion), the further apart will platforms locate in equilibrium, and the worse off are the
multi-homers.
Athey et al. (2014) discuss the supply of multiple content in a model of
decreasing return per impression and imperfect tracking of viewer behavior. They show
that a reduction of supply by one platform may lead the other platform to expand its sup-
ply. This points to a potential free-rider issue insofar as investment by one platform to
reduce the multi-homing demand benefits all platforms.
2.5.1 Free-Entry Analysis
Classic analysis of long-run equilibrium with oligopoly or monopolistic competition
closes the model with a free-entry condition, which is often taken as a zero-profit con-
dition for symmetric firms. Equilibrium product variety is then described by the number
of products in the market. This can be compared to optimal product variety to discern
market failures in the overall range of diversity provided by markets. Following
Spence
(1976)
, the market delivers excessive product variety when the negative externality on
other firms of entry (the “business-stealing” effect) dominates the positive externality on
consumers from having better-matched products and lower prices.
The canonical models that are usually analyzed are the CES representative consumer
model, the
Vickrey (1964) –Salop (1979) circle model, and “random utility” discrete-
choice models such as the logit. We concentrate on the latter two because they derive
from explicit micro-underpinnings for individual consumers.
Consider first the mixed-finance context, whereby both subscription prices and
advertising are used. Then, the characterization of the start of
Section 2.3 applies so that
platforms’ ad choices satisfy R
0
aðÞ¼γ. As we noted earlier, the implication for subscrip-
tion pricing is analogous to there being a negative marginal cost. Therefore, for the class
of models that assume fully covered markets and symmetric firms, because cost levels do
not affect equilibrium profits, the market equilibrium is fully independent of the adver-
tising demand. The implication is that equilibrium product variety is not impacted by the
advertiser demand strength. This strong decoupling result implies that the standard
Vickrey–Salop analysis goes through: there is excessive variety in equilibrium (see
Choi, 2006, for the statement in the media context). The same remark applies to other
covered market models (e.g., discrete-choice random utility models with covered mar-
kets). This decoupling is somewhat disconcerting for both the positive and the normative
analysis, but the problem is really the assumption that the market is fully covered. While
the circle model cannot be easily relaxed (apart from the trivial expedient of introducing
low consumer reservation prices and hence local monopolies), the discrete-choice model
can allow for uncovered markets through “outside” options, and this reconnects equi-
librium variety to advertising demand strength.
84 Handbook of Media Economics
The canonical model also assumes that the revenue per consumer R(a) is independent
of the audience, which is questionable. For instance, access to a large customer base helps
Internet platforms improve the efficiency of their advertising services. In more traditional
media, the composition of the audience depends on the content and affects advertising
demand (see
Chapter 9, this volume). A large audience with very heterogeneous con-
sumers may not be attractive for specialized advertising. Crampes et al. (2009) point
out that allowing R to depend on the consumer base is tantamount to considering var-
iable returns to scale in the audience, with less (more) entry when the revenue R increases
(decreases) with N
i
.
60
Now consider a regime of pure advertising finance. Here the advertising side is recon-
nected to equilibrium diversity. Under the fully covered market specification (e.g., circle
and logit), equilibrium ad levels and profits are decreasing in the number of platforms (as
long as R ðÞis log-concave: see the analysis in
Anderson and Gabszewicz, 2006). The
important point is that a weak advertising demand delivers low equilibrium diversity
because the economic impetus for entry is absent through low profitability from
advertising. This indicates that market failures through underprovision of variety can
be especially severe in such circumstances, for example in less-developed nations. With
strong advertising demand, the classic over-entry result still attains:
Choi (2006) notes the
strong disconnect between optimal variety and advertising levels in this case, albeit for the
first-best optimum.
61
Indeed, the disconnect holds whenever markets are fully covered:
the first-best optimal ad level of ads satisfies va
0
¼γ regardless of the number of
platforms.
Accounting for the endogeneity of the business model (free vs. pay) leads to slightly
different conclusions as the free-media business model emerges only if the advertising
demand is strong enough. As shown by
Crampes et al. (2009), imposing a non-negativity
constraint on consumer prices raises equilibrium prices (to zero) when they would be
negative. This relaxes competition for consumers, raises profits, and reduces advertising
and consumer surplus. As a result, there is more entry and total welfare is lower when
platforms are free than if they could subsidize consumers’ participation. Moreover, plat-
forms rely more extensively on quality improvement as an indirect form of subsidy, as
well as tying (
Amelio and Jullien, 2012).
62
60
They also show that price competition on the advertising side delivers more entry in the media market
than quantity competition.
61
Considering a constrained optima, such as with a zero-profit constraint or the constraint that platforms
choose ad levels non-cooperatively, would alleviate this separability result.
62
Dukes (2004) analyzes free entry with advertising for competing products. Lowering product differenti-
ation reduces entry by media, thereby intensifying their use of advertising. High media diversity (due to
easy differentiation) results in excessive advertising.
85
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.