round-number pricing and shows that quarterly data from six large US newspapers are
generally consistent with the predictions of the model.
Asplund et al. (2005) examine how likely newspapers are to vary prices on the
circulation versus the advertising side in response to financial constraints. They argue
that newspaper advertisers are not locked in to any given newspaper and therefore will
exhibit a much more elastic demand. Readers, on the other hand, develop tastes and pref-
erences for certain newspapers and therefore face high switching costs and are less likely
to switch papers in the event of a price increase. They examine Swedish newspapers in
the midst of the deep recession of 1990–1992 and show that publishers, who were faced
with liquidity constraints during this period, were far more likely to raise subscription
rather than advertising prices.
9.5. ADVERTISING IN NEWSPAPERS AND MAGAZINES
We devote this section to a discussion of the special role of advertising in newspapers and
magazines. While the role of advertising is, of course, important in all advertising-
financed media, it is particularly interesting, from a research perspective, in the case of
print media. Unlike in the radio and television industries, it is not obvious that advertising
imposes a cost on the circulation side. There are two reasons for this: first, print media do
not necessarily involve the platform having to trade off between advertising and content;
second, consumers may be able to skip advertising more easily in print media.
The question of whether consumers value advertising positively or negatively is
therefore important in newspapers and magazines. We will review the theoretical liter-
ature on this question in
Section 9.5.1 and the empirical literature in Section 9.5.2.We
will address the related issue of multi-homing by readers in
Section 9.5.3. Finally, in
Section 9.5.4, we discuss how audience characteristics determine advertising rates in
print media.
9.5.1 Reader Valuation of Advertising: Theory
We have laid out the effects of the interrelatedness between advertising and circulation in
Section 9.4. The network effects exist no matter whether readers like or dislike adver-
tising. While it is clear that advertisers appreciate higher circulation, at least given reader
characteristics (
Thompson, 1989), it is not obvious whether the same is true in reverse;
i.e., whether readers like advertising in print media, or if advertising constitutes a nui-
sance. A positive mapping between advertising and circulation implies that publishers
have incentives to subsidize circulation through advertising revenues, as in
Kaiser and
Wright (2006)
. If readers appreciate advertising, they simultaneously have incentives
to subsidize advertising. The degree to which each market side subsidizes the other
depends on the relative appreciation of each market side. If readers dislike advertising
422 Handbook of Media Economics
(but advertisers appreciate circulation), the reader side of the market will not subsidize
advertising (but subsidies will still flow from advertising to circulation).
Reader preferences with respect to advertising are therefore central to the analysis of
pricing structures in print media markets. Most theoretical studies of media markets—of
which we shall review some below—assume that readers dislike advertising (
Ambrus and
Reisinger, 2005; Ambrus et al., 2012; Anderson, 2005; Anderson and Coate, 2005;
Gabszewicz et al., 2001; Ha
¨
ckner and Nyberg, 2008; Jullien et al., 2009; Kind et al.,
2003, 2007; Kohlschein, 2004; Kremhelmer and Zenger, 2004; Reisinger et al., 2009
).
Reader neutrality with respect to advertising is assumed by
Gabszewicz et al. (2001),a
paper we briefly discussed in
Section 9.4.2 and to that we shall return to below, who
argue that it is simple to avoid advertisements in newspapers since it is easy to get past
them. Empirical studies that assume advertising neutrality include
Gentzkow et al.
(2012)
and Fan (2012). This shuts down the network effect from advertising to readers
and collapses the two-sided market into a market with network externalities from readers
to advertisers, thereby simplifying the analysis and avoiding fixed-point problems. By
contrast, the models in
Kaiser and Wright (2006) as well as Chandra and Collard-
Wexler (2009)
do not impose a priori restrictions on reader’s attitude toward advertising.
The importance of the degree to which readers dislike advertising is highlighted by
Anderson and Coate (2005), who study the under- or over-provision of ads in a two-
sided TV market setting. In an extension to previous work,
Anderson and Coate
(2005)
allow the degree to which viewers dislike advertising to vary. They assume that
viewers are distributed on a Hotelling line and that platforms are located at each end
of the line. Viewers are allowed to watch a single channel, while advertisers can
multi-home. They show that advertising volume may be too high or too low, depending
on how much viewers dislike advertising.
In an extension to
Anderson and Coate (2005), who do not allow for multi-homing
viewers,
Gabszewicz et al. (2004) allow viewers to mix their time between channels.
They set up a sequential Hotelling game where channels first choose programming
and subsequently determine the ratio of ads to content. They show that when adver-
tising is a nuisance, programs will be differentiated, whi ch contrasts with the very early
finding of
Steiner (1952), who predicts duplication of content among competing
channels.
That the degree of advertising nuisance is important for market outcomes is also
underscored by
Peitz and Valletti (2008), who analyze optimal locations of stations in
terms of programming. They show that if viewers strongly dislike advertising, the ratio
of advertising to content is larger in free-to-air TV, where all revenues are from adver-
tising, than in pay TV. They predict that free-to-air TV comes with less differentiated
content whereas program differentiation is maximal for pay-TV stations. The analogy
to the business model of newspapers and magazines here is that free-to-air TV is com-
parable to free newspapers while pay TV compares to paid newspapers and magazines.
423Newspapers and Magazines
Free newspapers and their impact on market structure have, however, not been system-
atically studied so far.
Choi (2006) also compares free-to-air and pay TV and studies the types and extent of
market failures under the two regimes under free entry. Similarly,
Jullien et al. (2009)
investigate the effects of free platform entry where platforms are financed both from
ad revenues and subscriptions. They predict excessive entry and too-low ad levels com-
pared to the social optimum. Both papers assume that advertising is a nuisance.
Finally, the importance of nuisance costs is highlighted by
Reisinger (2012), who
shows that profits may increase the more viewers dislike advertisements and that channels
make revenues from advertising despite price competition for advertisers. In his model,
there are single-homing advertisers and channels are differentiated from a viewer’s
perspective.
9.5.2 Reader Valuation of Advertising: Empirics
Our discussion of the theoretical literature has shown that reader’s attitude toward adver-
tising is important for market outcomes. We now review the empirical literature that has
dealt with this topic. This literature has arrived at very mixed results. The older papers on
two-sided market by
Bucklin et al. (1989), Dertouzos and Trautman (1990), and
Thompson (1989) for US and British newspapers respectively find that readers appreciate
advertising.
Sonnac (2000) conducts a cross-country descriptive analysis and finds that
readers’ attitudes toward advertising vary across countries.
The more recent literature tends to find positive effects of advertising pages on print
media demand.
Kaiser and Wright (2006) show that readers appreciate advertising for
their sample of German magazines. Their data set is, however, not representative of
the German magazine market since it comprises only magazine markets with duopoly
competition.
Filistrucchi et al. (2012) also find evidence for readers of Dutch newspapers
appreciating advertising while
van Cayseele and Vanormelingen (2009) show that Bel-
gian newspaper readers are advertising-neutral.
In their analysis of the entire German magazine market,
Kaiser and Song (2009) find
evidence that advertising is valued positively by readers. They estimate logit demand
models (with and without random effects), finding little evidence for advertising being
a nuisance to readers. On the contrary, in markets where there is a close relationship
between advertising and content, such as in Women’s magazines, Business and politics
magazines as well as car magazines, readers in fact have a clear appreciation of advertise-
ments. To study the role of informative vs. persuasive advertising more closely,
Kaiser
and Song (2009)
categorize each magazine in terms of advertisement informativeness.
They subsequently link the degree of informativeness to reader’s perception of advertis-
ing and demonstrate that there is a positive link between informative advertising and
reader’s appreciation of advertising.
424 Handbook of Media Economics
Depken and Wilson (2004) use data on 94 US consumer magazines to study the
effect of advertising on advertising rates and advertising demand. They define advertising
as “unambiguously good” if advertising increases both sales and prices, and as
“ambiguously good” if it decreases sales but increases ad rates (and vice versa for
“bad” effects of advertising on sales and ad rates). The main finding of the paper is that
advertising tends to be “unambiguously good” for 45 magazines and “ambiguously
good” for 19 magazines. For 31 magazines it is an ambiguous bad.
9.5.3 Multi-Homing
A central aspect of theoretical work on two-sided markets apart from the nuisance cost to
readers or viewers is multi-homing by readers and advertisers.
Armstrong (2006) coins
the term “competitive bottleneck” model where readers single-home and advertisers
multi-home, a model that found widespread use in the theoretical literature, e.g., by
Anderson and Coate (2005) that we discussed above. Apart from the results already dis-
cussed,
Armstrong (2006) shows that there is under-advertisement compared to the social
optimum since platforms operate as monopolists on advertising market.
The other main puzzle that the early theoretical study by
Steiner (1952) generated
apart from content duplication is that advertising levels unambiguously increase with
competition.
Ambrus and Reisinger (2005) were the first to notice that both anomalies
may be reversed if the models allow for multi-homing viewers. In their follow-up paper,
Ambrus et al. (2012) find that advertising levels can go up or down depending on how
viewer tastes are correlated. In their setting, viewers are allowed to use multiple plat-
forms, platforms do not steal viewers from one another but competition changes the
composition and hence the value that advertisers attach to consumers—multi-homing
viewers are worth less to advertisers than single-homers.
Ambrus et al. (2012) coin this
type of competition “either or both competition” (viewers watch either or both chan-
nels). They find that entry increases ad levels if viewer preferences are negatively corre-
lated. The paper also comes with an empirical analysis that is based on 68 cable channels
received by a viewer on a base lineup observed between 1989 and 2002 that validates
theoretical implications.
Ambrus et al. (2012) regress the number of supply choices on
the number of channels in each market segment (news, sport, infotainment) and find
large positive effects of competition on the number of avails.
In a related paper,
Anderson and McLaren (2012) also demonstrate how to resolve
the early
Steiner (1952) puzzles by allowing for viewer multi-homing.
10
In their model,
platforms become more differentiated when faced by competition in order to attract
exclusive (single-homing) viewers. They show how multi-homing viewers affect plat-
form differentiation, finding that platforms have incentives to make content unattractive
to multi-homing viewers, which works against Steiner’s duplication result.
Anderson and
10
This issue is also discussed in Chapter 6 .
425Newspapers and Magazines
McLaren (2012) allow, in contrast to Ambrus et al. (2012), for both multi-homing
viewers and endogenous platform quality, which is assumed to be fixed in
Ambrus
et al. (2012)
.
Multi-homing also is a major issue in
Athey et al. (2011) , a paper that discusses alter-
native viewer-tracking technologies and the implications they have for pricing and
advertising demand. They consider multi-homing viewers and let advertising effective-
ness differ between single- and multi-viewing viewers by allowing viewer tracking to be
imperfect. They implicitly assume exogenous ad levels and allow for heterogeneous
advertiser demands.
A key assumption of the theoretical literature is that multi-homing viewers have
lower value for advertisers than single-homers have. This assumption is questioned by
Chandra and Kaiser (2014), who show that contacting potential consumers via alternative
channels may actually increase the value that multi-homing readers generate.
9.5.4 The Determinants of Advertising Rates
The determinants of advertising rates have long been studied in empirical papers, while
there is no related theoretical treatment. The study by
Thompson (1989), which was
mentioned in
Section 9.2, was among the first to recognize the importance of certain
types of readers for advertisers. He finds that high-income readers are more valuable
to advertisers than are low-income readers in his sample of 34 British and Irish quality
and tabloid newspapers. This leads him to conclude that there exists a tradeoff between
circulation and a newspaper’s ability to target certain types of readers. He estimates a four-
equation structural model for circulation, cover price, and two types of advertising rates.
As in other earlier studies, the demand for copies and the (inverse) demand for advertising
space are linear and are not dependent on prices of competing newspapers.
Thompson
(1989)
estimates a simultaneous equation model and backs out own-price elasticities of
demand.
The importance of reader characteristics has subsequently been studied by
Koschat
and Putsis (2000a,b)
in their analysis of 101 US magazines. They essentially estimate a
linear hedonic pricing equation where all variables are assumed to be exogenous. They
find that young and affluent readers have a positive effect on advertising rates. In
Koschat
and Putsis (2000a)
, it is argued that advertiser’s preferences for particular types of readers
may induce publishers to—as they phrase it—“skew” content toward those types of
readers.
Koschat and Putsis (2000b) additionally conduct a counterfactual analysis where
they analyze the returns to fully target (which they refer to as “unbundling”) the most
relevant audiences from an advertiser’s perspective. They find that if publishers indeed
targeted the most profitable audiences, they would be able to considerably increase
profits. A critical assumption of their paper is, however, that targeting has no effects
on circulation.
426 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.