for example, people may choose to have stations that are relatively different in order to
cater to different moods at different times of the day (e.g., a favorite News station, a Sports
station, a Lite Adult Contemporary station, and a Top 40 station), while tending not to
pre-program multiple stations that are very similar in nature. In this case, one might
expect that the desire to avoid multi-homing might lead to at least some pairs of com-
peting stations becoming very similar. Unfortunately, this is a conjecture and very little is
known about the empirical relationship between programming variety and multi-
homing, even though the theory clearly indicates that it is important.
32
On the question of whether multi-homing could lead to a common owner choosing
to make its stations more similar than competing owners would, it is true that owners
often choose to operate multiple local stations in similar formats.
Berry and Waldfogel
(2001)
note this empirical pattern in the context of their examination of how consolida-
tion affects product variety, where they find that 27% of randomly drawn pairs of local
sibling (commonly owned) stations are in similar but not identical formats (e.g., Adult
Contemporary and Hot Adult Contemporary), compared with 18% of randomly drawn
local station pairs (ownership ignored). However, at the same time, they are less likely to
operate them in exactly the same format, suggesting that a common owner’s preferred
strategy balances some incentive to cluster stations, which may be generated by econo-
mies of scope or a desire to deter entry into that area of the product space, and incentives
to differentiate in order to avoid excessive cannibalization.
33,34
8.5. EMPIRICAL EVIDENCE ON THE EFFECTS OF OWNERSHIP
CONSOLIDATION IN RADIO
The rapid growth of empirical research in Industrial Organization in the 1990s coincided
with the consolidation of the radio industry following the various rounds of ownership
32
In the context of the local newspaper industry, Gentzkow et al. (2014) estimate a model of demand that
allows for multi-homing and for how multi-homing across newspapers owned by different firms reduces
the amount that newspapers can extract from advertisers, and how this affects incentives for product dif-
ferentiation. To do so, they use data from the 1920s and information from a number of surveys of read-
ership that measure multi-homing. Arbitron’s individual market reports do contain some of this
information in the form of “cume duplication” tables.
33
Sweeting (2004) reports an Infinity Programming Director in Cleveland, quoted in Billboard on October
14, 2000, describing how “I initially made that mistake when I was programming KPNT (The Point) in
St. Louis. We made sure The Point and [sister station] The River were programmed so far away from each
other that you could drop something in the middle of them and that’s what the competition wants you to
do.”
34
Of course, if a common owner does feel threatened by the possibility that other firms will program their
stations too close to one of their existing stations, then this would cast doubt on the argument that com-
peting owners naturally want to differentiate to reduce multi-homing.
359
Radio
deregulation. Radio was therefore a natural place for researchers to look at when trying to
understand several different effects that mergers might have. I have divided this section
into four parts.
Sections 8.5.1 and 8.5.2 consider the effects of consolidation in local mar-
kets on outcomes in the advertising market and programming, respectively.
Sections 8.5.3 and 8.5.4 describe the more limited work that has tried to pinpoint the
effects of consolidation at the national level and the size of cost efficiencies that consol-
idation may have realized.
8.5.1 The Effects of Local Market Consolidation on the Advertising Market
A number of papers, taking markedly different approaches, have tried to test whether
increased ownership concentration in local radio markets has increased or decreased
the price an advertiser must pay to reach a listener. Initially this literature was motivated
by the fact that the antitrust analysis of radio mergers by the Department of Justice was
focused on the welfare of advertisers, as the only customers of broadcast radio stations
who actually make monetary payments (
Klein, 1997). Given the recent theoretical lit-
erature, this empirical work can also be viewed as shedding light on whether models
where listeners single-home, which predict that mergers will lower advertising prices,
or models that assume multi-homing, which can make the opposite prediction, give a
more accurate description of the industry.
Most of the literature has focused on trying to establish the average relationship
between consolidation and prices, although the ambiguity of the theory suggests that
we might well expect to see significant positive effects in some settings and significant
negative effects in others. Distinguishing these cases would potentially be important
given the fact that mergers are analyzed on a case-by-case basis and that the authorities
are often able to negotiate carefully targeted divestitures (e.g.,
United States Department
of Justice, 2000a
), and, as we shall see, two recent structural papers find evidence of dif-
ferent effects.
The approach that most researchers have taken is to use a “reduced-form analysis”
where advertising prices are regressed on measures of ownership concentration.
Brown and Williams (2002) use a panel of data from 1996 to 2001, a period when local
concentration, ownership concentration at the national level, and real radio advertising
prices all increased substantially. They regress the market-level, SQAD-estimated adver-
tising price per 1000 listeners aged 18–49 on measures of local concentration and own-
ership by national radio firms, market fixed effects and proxies for local advertising
demand, such as market population and real income, and, to capture changes in national
demand, either time effects or national GDP growth. They find that increases in local
market concentration, as measured by revenue-based Herfindahl–Hirschmann indices
(HHI), are positively correlated with changes in local advertising prices, but that these
changes only explain around 5% of the large increase in advertising prices during the
360 Handbook of Media Economics
period of their data.
35
They find that greater ownership by large national radio companies
is associated with lower advertising prices. However, as they note, to interpret this second
correlation, it is important to recognize that SQAD-estimated prices are largely based on
prices charged to regional and national advertisers who may be simultaneously buying
commercials in several markets. The fact that national radio firms reduce prices to these
buyers, with whom they may enjoy some economies of scale or scope by selling com-
mercials in many markets simultaneously, does not necessarily imply that they also reduce
prices to local advertisers. This matters because local advertisers account for the majority
of station revenues (2006 BIA data would put the average at around 70%, and it has sub-
sequently risen to around 75%)
36
, and one might imagine that local advertisers are less
able to substitute to other media than national advertisers.
Chipty (2007) also estimates reduced-form regressions to examine the relationship
between concentration and advertising prices, using a cross-section of data from 2006,
but also using a wider range of SQAD prices (for example, for different dayparts, and
measures based on both costs per thousand listeners and costs per share point) than Brown
and Williams. Chipty finds no significant relationship between local concentration and
her measures of advertising prices but, like Brown and Williams, she finds a weak neg-
ative correlation between ownership by national radio firms at the local level and prices
once she controls for market demographics. However, this carries the same caveat about
interpretation as the Brown and Williams study, and in fact, using a sample of data on
programming content, she shows that there is no significant relationship between the
national ownership and the quantity of advertising on the radio, whereas a general decline
in advertising prices would have led one to expect an increase in the amount of adver-
tising.
Sweeting (2008) does find a positive effect of ownership by large, national radio
companies on the number of minutes of advertising using a panel of playlist data from
relatively large music stations during the time period 1998–2001. The effect is of mod-
erate size: around 0.6 more minutes of commercials per hour, or roughly 5% of the aver-
age commercial load for one of the stations in the sample.
37
Consistent with the lack of
price effects of changes in local concentration in the other reduced form papers, Sweeting
35
As part of its investigation into the merger between Global Radio and the Guardian Media Group’s radio
business, the
United Kingdom Competition Commission (2013) performed a detailed price-
concentration analysis, and found that “the presence of fewer good radio alternatives, and/or where
the radio alternatives are not as good, is associated with higher advertising prices” (Appendix I, p. I1).
Unfortunately, the magnitudes are not disclosed in the published report for confidentiality reasons.
36
Conversation with Mark Fratrik of BIA/Kelsey, February 13, 2015.
37
Sweeting’s analysis is based on imputing the number of minutes of commercials using gaps between songs
when some commercials were being played. It is therefore possible that this result instead reflects the fact
that national owners tend to insert more non-commercial talk programming (e.g., promotions, sponsor-
ship information) around commercial breaks rather than increases in the length of breaks themselves.
Sweeting also looks at hours outside the morning drivetime period, which has been the focus of other
studies.
361
Radio
finds no significant effects of changes in local concentration on how many commercials
are played. Therefore, one can summarize the reduced-form literature as finding no sig-
nificant effects of local consolidation, and some evidence that national consolidation
raised advertising quantities and reduced prices to national advertisers.
Two recent papers,
Mooney (2010b) and Jeziorski (2014a), have taken structural
approaches to understanding the relationships between local concentration and market
power in the advertising and listening markets. The potential advantage of structural
approaches is that they may be able to disentangle complicated relationships in the limited
available data by imposing the structure of an economic model and assumptions about
firm behavior, and they can also allow us to translate changes in the amount or price
of advertising into effects on welfare. On the other hand, the conclusions may depend
on how modeling assumptions made by the researcher interpret the data. This may be
particularly true in this case, as, for example, the models used both assume that listeners
single-home, by estimating discrete-choice models of listener demand, rather than
explicitly modeling the possibility of multi-homing. Unlike in standard Anderson and
Coate-style theoretical models with single-homing listeners, however, they allow for
the price of advertising on a particular station to depend on the quantities of commercials
that are aired on all stations so that, potentially, a common owner might restrict the quan-
tity of commercials to raise advertising prices. Therefore, even though listeners single-
home, it is not imposed by construction that station mergers should lead to more com-
mercials being aired.
38
Jeziorski (2014a) estimates an equilibrium model of a two-sided market using a panel
of data from 1996 to 2006. The data includes market-level SQAD advertising price esti-
mates, Arbitron data on station audiences, and BIA data on station formats, ownership
and BIA’s estimates of station revenues. Jeziorski estimates that local consolidation led
to quite large, 17%, reductions in the amount of advertising heard by the average listener,
with a corresponding 6.5% increase in per-listener advertising prices. These changes are
estimated to be largest in smaller markets, where the advertiser demand for radio adver-
tising is estimated to be less elastic, reflecting the fact that there may be more limited alter-
natives to radio for advertisers in smaller cities. As I will return to
Section 8.5.2, the effects
on listeners can be quite complicated because of the way that a common owner may
redistribute commercials across stations, and the way that changes in advertising quanti-
ties interact with changes in station formats.
Mooney (2010b) estimates a similar type of model using data from 1998 and 2003, and
reaches the conclusion that, on average, the effect of increased concentration was to
increase the quantity of advertising. The overall conclusion about how advertising chan-
ged is therefore the opposite of Jeziorski. The difference may be due to the fact that the
38
On the other hand, a valid objection is that these models do not provide an explicit rationale for why
stations would be substitutes for advertisers.
362
Handbook of Media Economics
papers treat what is known about the quantity of commercials quite differently. Jeziorski
assumes that the quantity of advertising on each station can be calculated using BIA’s esti-
mates of station revenues, SQAD advertising prices (per-share point), and station ratings.
Mooney assumes that it is not observed. Instead, she matches a moment based on the
average amount of advertising reported in
Sweeting (2008). Sweeting’s measure is based
on a sample of successful music stations, so it may not be representative of the industry as a
whole.
39
Jeziorski’s approach has the potential advantage that it creates an advertising
quantity variable for almost all stations (BIA does not provide revenue estimates for some
smaller stations, but for this exercise this problem should not be too important) that can
be used in estimation of listener and advertiser demand. In contrast, Mooney’s estimates
are likely to be more dependent on the assumed structure of the model to give predictions
about how the quantity of commercials varies across stations and over time. However,
Jeziorski’s approach obviously depends on any systematic errors in the revenue estimates
and SQAD prices not being correlated with changes in consolidation.
40
Future work
using better data on the quantity of commercials could clearly help us to understand what
the true relationships are with more confidence.
While they may disagree on the direction of the average effects, both papers empha-
size that the effects of consolidation may be heterogeneous, causing significant increases
in advertising prices in situations where advertisers’ demand is less elastic. Mooney finds
that this is likely to be the case for minority populations, which radio might be more
effective at reaching than other media. In
Mooney (2010a), she finds additional support
for this conclusion using the advertising quantity data reported by
Chipty (2007), which
suggests that when a single firm owns a group of Urban stations, which appeal to black
audiences, they tend to restrict how many commercials are played. Jeziorski estimates that
advertiser demand is less elastic in smaller markets, where there may be fewer media alter-
natives. Both papers therefore make the plausible point that some mergers may be much
more troubling from an antitrust perspective than others, and they give some guidance on
where (smaller markets or minority-focused stations) the antitrust authorities should look
for problems or require stronger evidence of efficiencies.
39
Jeziorski estimates that stations played, on average, 37.5 min of commercials per day between 1996 and
2006. This estimate is much lower than estimates based on airplay data (e.g.,
Sweeting, 2010) or contem-
poraneous industry reports (Radio and Records, April 21, 2000 cited by SchardtMedia,
http://
schardtmedia.org/?page_id¼80, accessed February 21, 2015) that indicate that stations played around
12 min of commercials per hour on average in 2000 and 2001. Of course, for the conclusions of the study,
what would matter is if differences between imputed and actual quantities vary systematically with station
ownership or over time.
40
As discussed in Section 8.3, SQAD prices may be more reflective of the prices paid by national advertisers,
who disproportionally advertise on stations owned by large firms, rather than those paid by local
advertisers.
363
Radio
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.