revenue or posted spot prices of advertising, are sometimes observable, much of the rel-
evant ad price information is not publicly available. The needs of advertisers give rise to
available data on product characteristics and the number of available products. In media
such as radio broadcasting and newspapers, there are directories listing the available out-
lets along with information on their product targeting (broadcast formats for radio, etc.).
1
Finally, changing technologies—such as Internet radio, television time-shifting devices,
and streaming delivery of various kinds of media products—strain the ability of existing
data infrastructure to describe the performance of media products.
This chapter proceeds in the following sections.
Section 3.2 discusses relevant models
of audience and advertising demand.
Section 3.3 presents a brief discussion of advertiser
demand for media consumers.
Section 3.4 turns to models of entry and equilibrium.
Along the way we make reference to examples in the media and, where appropriate,
more general industrial economics literatures.
Section 3.5 discusses the use of structural
models for policy analysis of media industries.
Section 3.6 concludes and contains a
discussion of future challenges.
3.2. AUDIENCE DEMAND
3.2.1 Introduction
Audience demand is obviously key to the analysis of media markets. For some media
(e.g., newspapers), direct consumers of media are at least one primary source of revenue,
whereas in other markets (traditional radio) audiences are effectively part of the produc-
tion process. In advertising-supported media, audience attention is “sold” to advertisers
and so we can think of audiences as being “produced” for sale to the paying “consumers,”
who are advertisers.
Below we review a set of demand models that are now standard in empirical industrial
organization, but with an emphasis on the features of media markets and on the relevant
policy questions. These models are typically discrete-choice models, although at an
aggregated market level they can also be given an interpretation as involving consumers
with a taste for variety (see
Anderson et al., 1992).
We think of potential audience members who face a set of media choices, described
by various characteristics. Some of these characteristics are observed to the researcher,
and some not. These consumers differ in tastes according to demographic attributes that
we, as researchers, observe and they also vary with unobserved consumer attributes and
tastes. In some cases consumers pay a price for consuming the good, in others not.
These models are broadly consistent with some basic stylized facts about media mar-
kets. Two such interesting stylized facts are that (i) audience size increases in the number
of choices and (ii) different demographics make starkly different choices within the same
1
See Chapters 7–9 of this volume.
93
Empirical Modeling for Economics of the Media
market. Both of these are consistent with heterogeneous preferences, and the first would
also be consistent with a taste for variety.
As an example of the first stylized fact,
Figure 3.1 shows how the number of varieties
and products available varies with the size of the market across US local radio markets.
Larger markets have more radio stations as well as more varieties. As an example of the
second stylized fact,
Figure 3.2 shows how radio choices vary by demographic. A small
number of station formats that are targeted at blacks attract almost two-thirds of black
listening, while these formats attract only a small share (under 3%) of white listening.
A similar pattern holds for Hispanic versus non-Hispanic listening.
Classic discrete-choice models, following on from
McFadden (1974, 1981), allow for
preferences that vary with observed and unobserved demographics, easily accommodat-
ing the pattern seen in
Figure 3.2. In addition, they include an idiosyncratic “match”
component for between each consumer and products. As the number of products
increases, consumers are more likely to find a desirable match with some product and
total audience size increases. In the context of some markets, this is seen as an undesirable
feature of models with idiosyncratic match components. However,
Figure 3.1 suggests
that this is a desirable feature for media markets. Clearly, we would like the data to
50
100
150
5,000,000
10,000,000
15,000,000 20,000,000
Population
Number of stations Number of varieties
US metro areas, 2009
Stations, variety, and market size
Figure 3.1 Stations variety and market size.
94
Handbook of Media Economics
Black-targeted White-targeted
Black listening White listening Hispanic listening Non-Hispanic listening
Black-targeted White-targeted
in English in Spanish in English in Spanish
Figure 3.2 Radio listening and demographics.
somehow determine the relative importance of idiosyncratic match values and more sys-
tematic taste differences depending on observed and unobserved demographics and taste.
3.2.2 Classic Discrete Choice with Observed and Unobserved
Product Characteristics
We think of a set of potential consumers (audience members) in a cross-section of mar-
kets. In each market, consumers make a discrete choice to consume one of the “within-
market” products or else to choose the “outside good.” For example, in
Fan (2013),in
each county consumers can choose to purchase one of a list of available newspapers, or
can purchase no newspaper at all.
Following
McFadden (1981), consider a utility function for potential consumer i for
product j in market t of
u
ijt
¼δ
jt
+ μ
ijt
, (3.1)
where δ
jt
is the “mean utility” of product j in market t and μ
ijt
captures the distribution of
tastes about that mean. For each consumer, the tastes μ
ijt
can be correlated across products
in ways that depend on the attributes of the consumer and the characteristics of the
product.
Following
Berry (1994), we model the mean utility as depending on a set of observed
characteristics, x
jt
, as well as an unobserved (to the researcher) characteristic ξ
jt
. The
unobserved characteristic captures elements of quality that are not measured in our data
and its presence also helps to fix ideas about the econometric endogeneity of demand.
The vector x
jt
includes observed data on product quality, but also measures of horizontal
differentiation such as type of product—for example, a dummy variable for whether a
paper is an urban daily or whether a radio station plays country music. Depending on
the media market, the vector x
jt
sometimes includes price and sometimes does not.
For convenience, we assume that
δ
jt
¼x
jt
β + ξ
jt
: (3.2)
A classic random coefficients (e.g.,
Hausman and Wise, 1978) formulation for the distri-
bution of consumer tastes would be
μ
ijt
¼
X
k
x
kjt
β
ikt
+ ε
ijt
: (3.3)
The term ε
ijt
is consumer i’s match value for product j, usually assumed to be i.i.d. across
products. Most commonly ε
ijt
is assumed to be distributed i.i.d. standard normal or else to
be distributed according to the “double exponential” extreme value distribution that
generates the familiar logit model of choice probabilities. The scale of utility is often
normalized by holding the scale of ε
ijt
fixed as other parameters change, although other
normalizations are possible.
96 Handbook of Media Economics
The summation term then introduces cross-product correlation in tastes. The term
β
ikt
is consumer i’s marginal utility for characteristic k, often modeled as a function of
the observed demographics and unobserved “tastes” of the consumer:
β
ikt
¼
X
k
x
kjt
υ
ik
σ
k
+ γ
k
z
i
½: (3.4)
The observed demographics are the vector z
i
, and the utility interaction between product
characteristic k and the consumer demographics is parameterized by γ
k
z
i
. In empirical
practice, the number of such interactions generally has to be limited. Unobserved tastes
for characteristic k are captured by the scalar υ
ik
, often assumed to be normally distributed.
The variance of unobserved tastes for characteristic k is increasing in the parameter σ
k
.
The combination of unobserved product characteristics (as in
3.2) with a normally dis-
tributed random coefficients logit (as in
3.4) is introduced in Berry et al. (1995), referred
to as “BLP.”
The distribution of marginal tastes for characteristics, β
ikt
, drives the cross-product
substitution patterns implied by the model. A higher variance of marginal tastes for prod-
uct characteristic k implies, ceteris paribus, that consumers will more closely substitute
between products with similar values of x
kjt
. For example, in radio markets a larger σ
k
on the unobserved taste for country music will imply that consumers will be more
“loyal” to the country music format as market conditions change. The simple reason
is that when σ
k
is large, those consuming country music are disproportionately those with
a systematic taste for country music, whereas when σ
k
is small many consumers listening
to a given country music station are doing so simply because of a large idiosyncratic match
value ε
ijt
.
The idiosyncratic match value is almost always present in a classic discrete-choice
demand model, in part because it “smooths out” market shares and avoids kinks in the
demand function (
Berry and Pakes, 2007). However, the economics of the idiosyncratic
match value is sometimes questioned, as it implies that each new product is “born” with
some fresh component that brings value to some subset of consumers. This assumption
seems more reasonable in media markets than in some other markets. Further, the i.i.d.
match value implies that the share of media consumers increases in the number of product
choices, an assumption that matches our stylized fact. The degree to which the market
expands is governed by the relative importance of the systematic β
ikt
as opposed to the
idiosyncratic ε
ijt
and this relative importance is determined by the parameters σ
k
and γ
k
.
There are several variations and classic special cases. The nested logit (
Cardell, 1991;
McFadden, 1978
) assumes that the only random coefficients are those on a set of “nested”
product group dummies. With one level of product nests, the unobserved random taste
component of utility for a product in nest k is given by
2
2
Arbitrarily rich observed interactions with consumer demographics are easily added to the nested logit
model; they are omitted here only for exposition.
97
Empirical Modeling for Economics of the Media
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.