size—would be expected to change media markets. In Section 1.3 we turn to “results,”
i.e., empirical evidence on the questions illuminated by the theoretical models. We dis-
cuss empirical results on entry and preference externalities that speak to the predictions,
we discuss the effects of technological change, and we suggest directions for future work.
1.2. FIXED COSTS AND HETEROGENEOUS PREFERE NCES
1.2.1 Fixed Costs
Media outlets, such as newspapers and radio and television stations, have cost structures
that are predominantly fixed. The cost of putting together a daily newspaper is based
mostly on the staff of reporters and editors, and this cost does not vary directly with
the number of copies produced (although a newspaper with more content might attract
more readers). To a varying degree across different sorts of media products, these fixed
costs are large, in the sense that markets can support few, or sometimes only one,
product(s). In the late 1990s the Columbus Dispatch, the major newspaper serving a met-
ropolitan area of roughly 3 million people, had 69 reporters and editors. At the same time,
the New York Times, serving the 22 million person metro area around New York, had
about 300 reporters and editors.
2
According to the Bureau of Labor Statistics, reporters
and editors earn an average of $44,000 per year.
3
Hence, the annual fixed cost of putting
together content at these two papers was roughly $3 million and $13 million respectively.
Radio stations have a similar cost structure, although their absolute level of fixed costs is
much lower than for newspapers. One cost estimate for a rudimentary religious radio
station puts the annual cost of operation at $142,000 per year.
4
Typical radio stations have
more employees, including at least six on-air personalities, as well as other managers, sales
staff, and engineers, bringing their costs of operation to about $650,000 per year accord-
ing to one estimate (without interest service on a license).
5
The budget of a public radio
station serving Garden City, Kansas (population 26,000) was reportedly $1 million in
2014.
6
The marginal cost of serving an additional listener to a radio station is, of
course, zero.
These estimates are of course rough, but they provide clear substantiation of high
fixed costs: it’s clear that the availability of these media products depends on many others
also wanting them.
2
See http://www.census.gov/population/cen2000/phc-t3/tab03.txt for 2000 population and Burrelle’s
Media Directory for information on newspaper staff size.
3
See http://www.bls.gov/oes/current/oes273022.htm.
4
http://www.christianradiohome.com/operating_costs.asp.
5
http://en.allexperts.com/q/Radio-Industry-2499/2008/10/radio-station-budget.htm.
6
Brad Cooper. “State Subsidy to Kansas Public Broadcasting Could Disappear.” Kansas City Star, April 29,
2014
http://www.kansascity.com/news/government-politics/article347706/State-subsidy-to-Kansas-
public-broadcasting-could-disappear.html.
6 Handbook of Media Economics
1.2.2 Preference Heterogeneity
Theoretical characterizations of preferences commonly represent them as smooth distri-
butions, such as consumers whose preference for some one-dimensional characteristic
(such as the sweetness of cider) is distributed uniformly between, say, a and b. Such char-
acterizations are of course useful for the development of tractable models, but they seem
to miss much of the nature of preference heterogeneity in reality, particularly for media
products.
For example, in the US, blacks and whites have starkly different preferences over most
sorts of media products. The broadcasting industry tends to divide its radio stations into
about 30 station types targeting different sorts of consumers. These formats include cat-
egories like “top 40” (or “contemporary hit radio”), “album-oriented rock,” “adult
alternative,” “country,” and so on. Some formats, such as those with “urban” or
“black” in their names, are targeted explicitly at African American listeners. Racial dif-
ferences in listening by format are stark. The most popular format overall is country
music, which attracts 12% of listening among non-blacks. Yet, country music attracts
only 1.5% of black listeners, so that the audience for country music is 97% non-black
(see
Waldfogel, 2003). At the other end of the spectrum, stations whose format names
include the word “black” (such as “black/adult contemporary” or “black/oldies”) attract
less than 3% of non-black listening and almost 60% of black listening, producing audi-
ences that are about 90% black. Blacks and whites have starkly different preferences in
radio programming.
7
Hispanics and non-Hispanics, too, have starkly different preferences in radio pro-
gramming. Stations broadcasting in Spanish attract roughly 50% of Hispanic listeners
in the US markets with substantial Hispanic populations. These same stations attract only
trace amounts of non-Hispanic listening.
Preference differences are not limited to radio. Newspaper and television preferences
also differ by race. In markets with two daily papers—typically a broadsheet and a
tabloid—the market share of the broadsheet is generally much larger in heavily white
neighborhoods, which is suggestive of preferences that differ by race. While black and
white television viewers both rank some television programming highly—such as
football—many television shows that are top-rated among whites are often bottom-rated
among blacks. For example, in 1998 when Seinfeld was the top-rated show among whites,
it was ranked 50th among black viewers, while the comedy “Between Brothers” was ranked
first among blacks and 112th among whites.
8
Preferences also differ by gender and age. In 1993 less than 1% of listeners below age
45 listened to stations in the big band/nostalgia format, while these stations attracted 14%
of listeners over age 65. By contrast, top 40 stations attracted 38% of under-18 listening,
7
See also Aldrich et al. (2005).
8
See James Sterngold. “A Racial Divide Widens on Network TV.” New York Times, December 29, 1998.
7
Preference Externalities in Media Markets
while such stations attracted less than 2% of over-65 listening. Album-oriented rock sta-
tions attracted 20% of male listening but only 10% of female listening.
9
Two points bear
emphasis, however. First, these preference differences tend not to be as stark as those dif-
ferences by race and Hispanic status. Second, while preferences differ across groups, the
share of population does not differ much across markets according to age or gender, while
the black and Hispanic shares vary greatly. This is important because the mechanisms
we’ll study—entry and consumption in response to population—are visible only by
way of comparisons across markets.
While the shares of population by gender and age are quite similar across metropolitan
areas, the shares of population in minority groups with different preferences for media
content vary substantially. Across the top 100 US metro areas in 1993, the median black
share was 6.3%, while the median Hispanic share was 2.1%. The metro area at the 90th
percentile of the black share distribution was just under a quarter black, while the metro
area at the 10th percentile was just under 1% black. The 90th percentile Hispanic share
was 21%, while the 10th percentile Hispanic share was under 1%. These magnitudes of
variation in the mix of consumers with different media product preferences raise the pos-
sibility of detecting impacts of preference group sizes on product targeting and
consumption.
Just as preferences vary across demographic and ethnic groups within the United
States, preferences for some kinds of media products vary across national groups. While
German, French, and American consumers have access to almost all of the same music
and movies, their consumption patterns are starkly different. Music consumption in each
country exhibits a substantial amount of home bias: 29% of French consumption is
domestic, while only 3% of German consumption is French; 21% of German consump-
tion is domestic music, while less than 2% of French consumption is German. Much of
the difference in consumption patterns stems from language: Austrians consume German
music at elevated rates, Belgians listen to French music, and vice versa. Two repertoires
that all destinations consume at elevated rates are those of Great Britain and the United
States, which make up 10–30% of consumption and 30–60% of consumption outside of
their home markets.
10
We see very similar patterns in cross-national tastes for movies.
11
It is a widespread phenomenon in media markets that preferences differ, sometimes
starkly, across groups. And the mix of different preference groups differs across place, rais-
ing the possibility of both the operation and detection of preference externalities.
9
See table 5, p. 40, Joel Waldfogel, 1999. “Preference Externalities: An Empirical Study of Who Benefits
Whom in Differentiated Product Markets.” NBER Working Papers 7391, National Bureau of Economic
Research, Inc.
10
See Aguiar and Waldfogel (2014) and Ferreira and Waldfogel (2013).
11
See Ferreira et al. (2013).
8
Handbook of Media Economics
1.2.3 Willingness to Consume Second-Choice Products
While consumers from many groups prefer one sort of media product to another, there is
also fairly clear evidence that consumers are, in many cases, willing to consume a second-
choice product—rather than forgoing consumption in the category altogether—when a
more desired product is not available. Consider again the example of race and radio in the
US. Waldfogel (1999) documents that while black listeners clearly prefer black-targeted
programming, blacks continue to listen to the radio nearly as much in markets that lack
black-targeted programming. While 18.4% of blacks listened to radio (for at least 5 min
during an average quarter-hour in 1997) in markets with four or more black-targeted
stations, 17.3% of blacks listened to the radio in markets with 0 or 1 stations in black-
targeted formats. Patterns were similar for Hispanics. This strongly suggests that blacks
and Hispanics are willing to consume second-choice alternatives when most preferred
alternatives are unavailable.
12
1.2.4 Advertiser Finance
Some media products are financed by users directly; others are financed by advertisers.
Television and radio have traditionally been financed entirely by advertisers. Newspapers
and many magazines have been predominantly advertiser-financed. In 1999, for example,
roughly 80% of newspaper revenue was derived from advertisers. Magazines populate a
spectrum from those that are predominantly financed by advertisers—such as bridal and
photography magazines—and those that are mainly financed with subscription revenue,
such as US Weekly or Scientific American.
13
Since the advent of cable television in the 1970s, user finance has grown in promi-
nence. Many channels—such as HBO and Showtime—are financed entirely by users and
carry no advertising. The financial model for radio has also changed over time. While
radio programming was user-financed at the dawn of the industry in the 1920s, it was
entirely advertiser-financed for most of the twentieth century. Satellite radio, which
emerged in the US in 2001, is user-financed and carries no ads.
14
Internet radio is
financed with a mix of advertiser and user finance.
When newspapers placed their content online, most initially relied only on revenue
from advertisers. While the Wall Street Journal instituted a paywall in 1997, most news-
papers did not immediately follow. Beginning around 2010, many newspapers began to
put their content behind paywalls, shifting back toward a mix of user and advertiser
finance.
15
12
See Waldfogel (1999), table 7.
13
Authors’ calculation using data from the Publisher’s Information Bureau for 2000.
14
http://en.wikipedia.org/wiki/Satellite_radio.
15
See Chiou and Tucker (2013).
9
Preference Externalities in Media Markets
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.