positive externalities—then the more participation there is from both sides. Notice that
for γ < 1 the yen for ads is so strong that the number of advertisers goes beyond 1,
which can be construed as pricing ads below the marginal cost of delivering them
(say, the cost of printing the pages in a magazine). This can make sense in a two-sided
world because when ads are attractive it is possible to charge readers much more for
the media product. For example, some newspapers carry some ads for free (such as
the Hook’s classified ads in Charlottesville), and some vintage car magazines carry free
ads with the consumer footing all platform revenues.
–
1.5 –1.0 –0.5 0.5 1.0
0.2
0.4
0.6
0.8
1.0
N,a
Number of subscribers, N
Nuisance parameter, g
Ad level, a
Figure 2.1 Participation on the two sides of the market.
–
1.5 –1.0 –0.5 0.5 1.0
–1.5
–1.0
–0.5
0.5
1.0
$ consumer
Nuisance parameter, g
Price per ad per consumer, p
Full price, f
Ad nuisance, g a
Subscription fee, s
Figure 2.2 Prices paid on the two sides of the market.
52
Handbook of Media Economics
Figure 2.2 shows the prices faced on both sides of the market (p and s) and furthermore
breaks down the full price paid by the consumers into the subscription fee and the ad
nuisance (or benefit, when this is negative). Notice the negative ad price (noted above)
for γ < 1. For γ < 0, the more consumers like ads then the more ads they are delivered,
even to the point of pricing them below marginal cost. This is the idea from two-sided
market analysis of subsidizing one side of the market to extract more from the other side:
consumers are charged increasingly higher subscription prices as they like ads more.
As γ rises, the price per ad per consumer, p, and the consumer full price, f, rise too
(consistently with the falling participation).
20
Here the breakdown of f is interesting:
because the total ad nuisance rises for γ > 0 then falls as ads disappear, the subscription
fee falls and then rises in order to have the full price rise. We will see more starkly in
the next example the U-shape of s, showing that a particular subscription price can be
consistent with one level of ad loving, and one level of ad annoyance. When ad nuisance
is large, consumers are charged quite a high price to avoid the ads, but when consumers
like ads, they are charged a high price to enjoy a lot of them.
One striking recent development in magazines is the drop in subscription prices and
tilt toward ad finance in the business model.
Figure 2.3 plots the ratio of subscription
revenue to advertising revenue as a function of γ. Above the upper bound γ ¼1 there
is no ad revenue, while below γ ¼1 net ad revenue is negative. The fraction of ad rev-
enue is highest in the middle region around γ ¼0 where ad revenue per consumer is
greatest. For higher levels of ad nuisance there are fewer ads and more is taken from
–
1.0 –0.5 0.5
5
10
15
Nuisance parameter, g
Finance ratio,
sN
paN
Figure 2.3 Ratio of finance from the two sides of the market.
20
Despite rising prices on both sides of the market, platform profits fall consistently with γ—as is clear more
generally from applying the envelope theorem.
53
Two-Sided Media Markets
subscription prices, while for ad loving there are so many ads that little can be charged per
ad to attract so many ads and consequently charge consumers a lot for the benefit
from them.
For the parameters of these first three figures, the subscription price is always positive.
This will not be the case for sufficiently strong ad demand.
Figures 2.4 and 2.5 descri-
be the equilibrium participation and prices respectively for the stronger ad demand
vaðÞ¼3 a (and with the same consumer dema nd as before, N ¼1 f ). These param-
eter s lead to all three regimes being deployed as the eq uilibrium business model, depend-
ing on γ. The new regime—pure ad finance—arises in the middle, when subscription
prices are zero. Consistent with the previous description, equilibrium participation in
–2 –1 1 2 3
1
2
3
N, a
Nuisance parameter, g
Ad level, a
Number of subscribers, N
Figure 2.4 Participation on the two sides of the market with expanded ad market.
–2 –1 1 2 3
–3
–2
–1
1
$ consumer
Nuisance parameter, g
Price per ad per consumer, p
Full price, f
Ad nuisance, g a
Subscription fee, s
Figure 2.5 Prices paid on the two sides of the market with expanded ad market.
54
Handbook of Media Economics
the market for both sides falls with γ and is reflected in rising ad price per consumer an d
rising consumer full price. The kinks in the participation rates occur when s hits zero and
the regime shifts to ad-only finance. Once again, in order to get the consumer full pric e
rising, the subscription price is U-shaped.
2.3.2 Representative Consumer Models
Several authors use a representative consumer approach to modeling the consumer side in
media economics.
21
There are pros and cons to proceeding thus. First, it is not always
clear what is obscured by aggregating explicitly heterogeneous individuals and their
choices. This issue is particularly germane given we have drawn strong differences in out-
comes when some consumers multi-home from the case where all single-home. The rep-
resentative consumer consumes some of all platforms’ offerings, and on the advertiser’s
side it is assumed that ads are valued on each platform the same way regardless of whether
the ad(s) are seen on other platforms. The latter is consistent with the single-homing dis-
aggregated approach. Thus, one way to interpret the representative consumer is not as
“representative” in the traditional sense of aggregating disaggregate behavior, but instead
more as a “typical” consumer who watches multiple channels. This brings up a benefit of
the approach, which is that it can deliver a multi-homing model for the consumer side,
and allow for allocation of time/attention across platforms. However, it must also be
assumed that any multi-homing advertiser (MHA) puts ads on all channels at the same
time so that the problem of what happens when a consumer sees more than one ad from
the same advertiser does not arise. (Note that these issues are common to the advertising
congestion approach deployed by
Anderson and Peitz (2014), and described further
below.) The model therefore may fit TV (where one can only reasonably watch at most
one channel at any given time) rather than magazines, where ads are not ephemeral.
Alternatively, one might indeed assume that the consumer’s response to any ad is inde-
pendent of how many times she sees it. Another point on the plus side is that it is impor-
tant to explore alternative settings, and to check for robustness of findings.
There are two further issues. First, it is typically assumed that consumption of a
medium entails a constant money price per unit of time. However, most TV channel
subscriptions are “all-you-can-watch” after paying a term subscription. A second issue
concerns the utility/demand functions frequently used, which are based on the
Shubik and Levitan (1980) linear demand system. This particular demand system has been
criticized for some perverse comparative static results in the way it deals with entry of new
products: the point becomes apparent in the implicit representative consumer approach
that generates it, having extra interaction effects through the number of products per se.
22
21
See, for example, Dewenter et al. (2011), Cunningham and Alexander (2004), and Godes et al. (2009),in
addition to the other papers cited below.
22
Nonetheless, Kind et al. (2009, footnote 9) note that their results are robust to the exact specification of
consumer preferences.
55
Two-Sided Media Markets
Kind et al. (2007) consider a quadratic representative consumer utility function à la
Shubik and Levitan (1980). They assume a three-stage game structure whereby platforms
first set advertising space
23
; then advertisers choose how many ads to place on each plat-
form; then the consumer makes her viewing choices. Advertisers are potentially hetero-
geneous, but each advertiser has a constant return per ad aired per viewer hour on a
channel (regardless of how many ads are seen and whether or not they are on other chan-
nels). Advertisers are finite in number so, because they act before consumers observe their
choices, each internalizes the negative effect of its ad levels on viewers’ consumption
levels. One result from the analysis is that advertisers make less profit the more platforms
there are. This result broadly concurs with the analysis of the disaggregated models: plat-
forms raise prices per viewer to reduce ad nuisance, which is how they compete for
viewers. This effect reduces advertiser surplus per viewer (although the effect may be
offset if the market is not covered because more viewers can be reached, and depending
on the ad demand function form). Although the mechanism is a little different in the
representative consumer context with the (partial) internalization of own ad nuisance,
the effect is still there that set lower ad levels when there are more platforms.
Kind et al. (2009) extend the model to consider mixed finance. One other difference
from
Kind et al. (2007) is that platforms set (linear) ad prices (per ad) in the first stage,
along with prices to the consumer per hour of watching. Because of their game timing
(whereby advertisers internalize the effect of their ads on the viewer’s choice), they break
the R
0
a
ðÞ
¼γ relation that comes from the usual disaggregated model. This changes some
characterization results. For example, the volume of advertising is larger the less differ-
entiated are the platforms (with the standard model, ad levels would be unaffected per
platform, although indeed with partially covered markets the total level of consumer-
ad-minutes would typically rise as more of the market gets covered).
Kind et al.
(2009)
also address how the number of competitors affects the equilibrium financing bal-
ance. While more competition decreases revenues from both sources (consumers and
advertisers), the share underwritten by consumers rises with the number of platforms.
2.3.3 Competitive Bottlenecks
The single-homing assumption used so far implies that each platform has a monopoly
position over delivering its exclusive viewers to advertisers. While there is competition
for viewers through ad nuisance, there is no direct competition for advertisers. This
assumption gives rise to so-called competitive bottlenecks in the market (
Armstrong,
2002, 2006, Armstrong and Wright, 2007
), as evinced by the price per ad per viewer
exceeding the “monopoly” rate against the advertiser demand curve (i.e., v(a
m
) where
a
m
solves R
0
a
m
ðÞ¼0). This in turn generates several strong predictions that may not hold
23
They note their main results still hold if instead ad prices were chosen here.
56
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.