in all contexts, and leads the discussion into the equilibrium effects of viewer multi-
homing, which is continued in the next section. These anomalies were first pointed
out by
Ambrus and Reisinger (2006) and these “puzzles” are further discussed in
Anderson et al. (2012b).
The first puzzle is that if a public broadcaster is allowed to carry ads then the private
broadcasters will be better off because the private broadcasters pick up some of the con-
sumers diverted by the ad nuisance on the public channel. With SHCs, there is no direct
competition for advertisers, who multi-home.
Anderson et al. (2015b) give some anec-
dotal evidence that instead private broadcasters do not relish there being ads on the public
channel.
24
Second, entry of an additional commercial broadcaster raises the equilibrium ad price
per consumer, although we would usually expect more competition to reduce prices.
This effect, already discussed above, stems from competition for consumers being in
ad levels, which are the effective prices paid by consumers. These “prices” do indeed
go down (just like prices go down with more competition in standard one-sided mar-
kets), but when they do so we move up the advertiser demand curve. The opposite direc-
tion of the change in prices is an example of a “see-saw effect” in two-sided media
markets, as we discuss further below (
Section 2.3.4).
Third, and related to the previous one, a merger reduces ad prices per consumer
because the merged entity raises the ad level (analogous to raising the price paid by con-
sumers in standard markets). Moreover, when the merged entity then gets a smaller mar-
ket base, its price falls a fortiori. Evidence on this effect is mixed (see
Chapters 8 and 9, this
volume). Mergers in media markets are discussed further in
Chapter 6 (this volume). One
theme that comes up in thinking about these puzzles in the single-homing case is the
realization that two-sided markets are quite different from one-sided markets. Even
though the equilibrium reaction to a change (such as a merger) has the “expected” change
on one side of the market (the consumer side in the above examples), it may have the
opposite effect on the other side. Anderson and Peitz (2014) christen this the “see-
saw effect” (see Section 2.3.4) and illustrate several instances when it occurs when con-
sumers single-home.
A final puzzle noted by Ambrus and Reisinger (2006) is the “ITV premium” that
programs with more viewers have ad prices that are more than proportionately higher.
The idea is that such programs are likely delivering viewers who are hard to pick up on
other programs, so this (as with the other puzzles) can be explained by the existence of
MHCs, to which we turn in
Section 2.4.
24
Chapter 7 (this volume) gives a spirited discussion of public broadcasting in television markets. We do not
enter here into the complex objective function of a public broadcaster and how to model its behavior: for
the point of the current comparison, we simply assume that it moves from carrying no ads to carrying some
ads.
57Two-Sided Media Markets
2.3.4 See-Saw Effects in Media Markets
A “see-saw effect” (following Anderson and Peitz, 2014) arises when a change in market
fundamentals causes one group of agents served by platforms to be better off while the
other group is worse off. In the media context, advertisers can be better off and consumers
worse off, so giving a conflicted interest to changes or legislation concerning platforms.
In standard markets, there is typically a conflict that a change (say, firm exit) can make
firms better off and consumers worse off. In two-sided markets, there are three groups of
agents who interact. Surprisingly, perhaps, what is good for platforms may also be good
for one group (typically the advertisers, which are therefore counted on the side of plat-
forms). The consequences for merger analysis are treated in
Chapter 6 (this volume).
Anderson and Peitz (2014)
apply an aggregative game framework to a two-sided
media market context with competitive bottlenecks to address who are the winners
and losers from changes in market circumstances (such as mergers, etc.).
25
The motivat-
ing research question was to uncover when there can be “see-saw” effects that (say)
advertisers are better off while consumers are worse off from some change. This speaks
to the heart of two-sided interactions.
The fundamental trade-off is seen in an elementary fashion in the advertiser demand
curve for reaching consumers. The price per advertiser per viewer is on the vertical axis,
while the number of advertisers is on the horizontal one. But the number of ads also rep-
resents the “price” in terms of ad nuisance that is paid by consumers in free-to-air broad-
casting. So any move down the ad demand curve tends to make advertisers better off and
consumers worse off.
Consider first a merger between platforms. As long as ad levels (the strategic variables)
are strategic complements, such a merger makes all platforms better off, including the
merged entity (so there is no “merger paradox”—this is true by analogy to models of
Bertrand competition with substitute products in standard one-sided markets). What
happens to advertisers depends on whether ads are a nuisance to consumers or they
are desired. In both cases, ad levels move closer to the “monopoly” levels, meaning that
they rise for γ > 0 and fall for γ < 0. Consumers are worse off in both cases. For γ > 0,
advertisers are better off (and so there is a see-saw effect in evidence) from the fact that
prices per ad per viewer fall. But this effect is mitigated by the lower audiences for the ads.
As
Anderson and Peitz (2014) show, advertiser surplus rises on the platforms that are not
party to the merger, but the effect on the advertiser surplus that accrues on the merged
platforms is more delicate. Nonetheless, there are central cases in which it rises, so then
25
An aggregative game, following Selten (1970), has the simplifying structure that platforms’ payoffs can be
written as functions simply of own actions and an aggregate which is the sum of all platforms’ actions.
Judicious choice of an aggregator can render a large class of games as aggregative ones. One benefit of
the approach is to deal cleanly with heterogeneity across platforms, for example in intrinsic program qual-
ity, and thus be able to give a clean cross-sectional characterization of equilibrium. See
Acemoglu and
Jensen (2013)
and Anderson et al. (2013a) for more details on aggregative games.
58
Handbook of Media Economics
the total effect is unambiguous and there is a see-saw effect in operation. For γ < 0, the
effects go in the opposite directions, and so merger is bad for all the participants on both
sides of the platform. These results serve to highlight the idea that the traditional conflict
between firms and those they serve are much more intricate in two-sided markets: the
see-saw effect draws this out.
Entry also involves see-saw effects for γ > 0. Consumers are better off (with more
variety and more competition) but advertisers can be worse off if the entrant platform is
a small player and the market is close enough to being fully covered. However, for γ < 0
(ad appreciation), the interests of consumers and advertisers are aligned in relishing entry.
Restricting the ad level of a platform (for example, a public broadcaster) has an unam-
biguous see-saw effect for γ > 0. Consumers are happier with diminished ads across the
board (as the private companies respond with lower ad nuisance in the face of tougher
competition for viewers). Advertisers though are worse off for the twin reasons of
demand diversion to the public channel where they are denied communication with
viewers, and the higher ad prices elsewhere, coupled to lower audiences.
For a mixed-finance market,
Anderson and Peitz (2014) consider asymmetries
between platforms to show that platforms with higher equilibrium consumer numbers
have higher subscription prices because they correspond to higher “qualities” to con-
sumers.
26
All platforms with positive subscription prices set the same advertising level,
as per our earlier results to this end. Platforms with too low qualities set zero subscription
prices (it is assumed negative prices are infeasible, for reasons previously discussed), and
the lower the quality, the lower the ad level that is supported.
Suppose now that platforms are financed by a mix of subscription prices and ads, and
consider a merger. Then, as we had for pure ad finance, ad levels on all platforms rise for
γ > 0 and fall for γ < 0. Consumers in both cases are worse off, while all platforms are
better off. The consequences for advertisers are also as before: there is a see-saw effect
(advertisers are better off ) for γ > 0 if all platforms have the same quality, for example.
See-saw effects are also apparent in other approaches. For example,
Kind et al. (2007)
(who consider γ > 0) find that entry raises consumer surplus but decreases advertiser
surplus.
27
There is not much work on see-saw effects with consumer multi-homing. However,
one example gives the opposite effect from the single-homing case.
Anderson and Peitz
(2015)
in their model with advertising congestion (described below) find that advertisers
lose from merger while consumers gain. This is because the merged entity internalizes ad
congestion to a greater degree and so places fewer ads.
26
Anderson and de Palma (2001) derive analogous results for differentiated product oligopoly.
27
Kind et al. (2009) do not consider a welfare analysis, being a Marketing Science paper, although one would
expect similar see-saw effects to arise in that context. (Recall it differs from
Kind et al. (2007) because it
considers a mixed-finance regime.)
59
Two-Sided Media Markets
2.3.5 Heterogeneous Ad-Nuisance Costs, Price Discrimination, and TiVo
Ad nuisance varies across consumers. This raises two issues: platforms might offer differ-
ent combinations of subscription price and ad nuisance, and consumers might be able to
access technology to strip out the ad nuisance. In both cases, in equilibrium it will be
those consumers most annoyed by ads who consume fewer of them.
The first of these issues is a second-degree discrimination problem. Platforms can offer
different “contracts” and invite consumers to choose between them.
Tag (2009) analyzes
a monopoly allowing two choices, which he restricts to being an ad-only program and an
ad-free pure subscription option. As he shows, allowing for the subscription option
induces a higher level of advertising for those remaining on the ad-only option, in a clas-
sic illustration of the idea (going back to
Dupuit, 1849) of “frightening” those with high
ad nuisance to self-select into the lucrative subscription segment. While aggregate con-
sumer surplus falls in his parameterized example, advertiser surplus rises through lower ad
prices per consumer, despite a lower consumer base.
Anderson and Gans (2011) allow instead for consumers to choose a costly ad-stripping
technology (such as TiVo). They find an analogous result to
Tag (2009). Selection into
TiVo by ad-averse consumers leads a non discriminating platform to set higher ad levels
on those (less ad-averse) consumers remaining because of their lower sensitivity to ad
nuisance. Ad stripping (“siphoning” off content without paying the “price” of consum-
ing the advertising) may reduce welfare and program quality.
28
The theme of ad avoid-
ance is further developed in
Stu
¨
hmeier and Wenzel (2011).
Various authors also consider mixed markets where different platforms concentrate
on serving different consumers as a function of their ad sensitivity.
Anderson (2003)
and Lin (2011) look at a mixed duopoly with a free-to-air broadcaster and one using sub-
scription prices only.
Weeds (2013) extends the analysis to look at quality competition in
a mixed duopoly with competition between a free-to-air broadcaster and one using
mixed finance.
2.3.6 Market Failures in Advertising Finance
Market failures from market power are well established for oligopoly in one-sided mar-
kets. Thus, the subscription-only regime suffers from prices that are too high and the
markets served are too small. Now consider the mixed-finance regime. Again, oligopoly
pricing leads to insufficient site visitors. The level of ads provided under a mixed-finance
system solves R
0
a
s
ðÞ¼γ. However, the social benefit of an extra ad is the demand price,
vaðÞ> R
0
aðÞðÞ, plus the extra advertiser surplus due to the effect of the extra ad reducing
the market price of ads. This inclusive social benefit should be equal, at the optimum, to
the nuisance cost, γ. This implies that the market provision of ads is below optimal
28
See also Shah (2011) for more on ad avoidance as well as time-shifting of consumption.
60
Handbook of Media Economics
because the social benefit exceeds the private benefit. Again, overpricing, this time in the
ad market, is the usual concern with market power.
Lastly, consider pure ad financing where the upshot—too many or too few ads?—is
ambiguous. To see this, first suppose that γ ¼0, so readers are ad-neutral. Then, the opti-
mum allows all ads with a positive benefit to advertisers, so vaðÞ¼0. The market solution
instead delivers a lower ad level, where R
0
aðÞ¼0. At the other extreme, if γ > v 0ðÞ, the
optimum has no ads at all by dint of the nuisance exceeding the maximal ad demand price.
However, under pure ad finance, the market will always deliver some ads because they
are the only source of profit. Between these extremes, there will be too little advertising
for low values of γ, and too much for high enough levels of γ (
Anderson and
Coate, 2005
).
2.3.7 Alternative Equilibrium Concepts: Price Versus Quantity
The model presented so far follows the convention adopted by most articles that the
media platforms choose the amount of advertising (time length for TV, pages for news-
papers, etc.) and that this choice is observed by consumers. As already pointed out, this
assumption implies that we may view the choice of a as similar to choosing a vertical
quality dimension. While the assumption that the media directly choose a might make
sense for traditional media, it may be questionable for modern media. For instance, the
amount of display on a web page may vary with the short-term fluctuations of demand for
advertising slots. Auctions for ad slots often impose a reservation price so that some slots
remain unfilled.
Armstrong (2006) and Crampes et al. (2009) consider an alternative sce-
nario where the platform fixes the price P
i
for ads and lets the quantity a
i
adjust to clear the
advertising market.
If platforms choose their P
i
values, the situation becomes more similar to a standard
two-sided market, where the demand on one side depends on the demand on the other
side. While platform i’s profit is still π
i
¼P
i
a
i
, determining the quantity a
i
requires solving
a complex fixed-point problem to obtain the advertising levels as a function of prices. For
instance, in our base model this requires inverting the system of equations:
P
i
¼va
i
ðÞN
i
γa
i
;γa
i
ðÞ; i ¼1,…,n;
to obtain ads quantities as function of prices a
i
¼A
i
P
i
; P
i
ðÞ. Whether price-setting for
ads results in lower or higher levels of advertising than quantity-setting depends on the
sign of the nuisance term γ. To see this, consider the impact on demand of a marginal
increase in the amount of advertising. In the quantity-setting game, the impact on sub-
scription is just γN
0
i
γa
i
;γa
i
ðÞ, where N
i
0
denotes the derivative with respect to the first
argument. In the price-setting game however, there is a transfer of demand across other
platforms. When the nuisance cost (γ) is positive, this transfer of demand attracts more
advertisers to competing platforms. This effect attenuates the negative impact of platform
i’s advertising on its subscriptions. Thus when consumers dislike advertising, the
61Two-Sided Media Markets
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.