Skip Navigation

Federal Communications Commission

English Display Options

Commission Document

Media Ownership Study 7-Further Revision

Download Options

Released: July 27, 2011




Radio Station Ownership Structure and the Provision of Programming
to Minority Audiences: Evidence from 2005-2009

Joel Waldfogel
The Carlson School and Department of Economics
University of Minnesota

July 18, 2011


0


1. Introduction
The goal of this study is to assess recent evidence on the relationship between
ownership structure and the provision of radio programming to minority (African-American
and Hispanic) audiences.1
Ownership structure might affect programming by one (or more) of three basic
mechanisms. First, having multiple stations owned by the same entity can reduce the costs of
operating additional stations – and additional programming formats. A firm facing lower costs
can operate more – and perhaps also more varied – stations within a market. Hence, an owner
of multiple stations might be able operate more minority-targeted stations for a given level of
revenue. A second complementary mechanism concerns the internalization of business stealing
externalities. Suppose that two separately owned stations in a geographic market operate in the
same programming format. If they were instead jointly owned, their owner would have an
incentive to separate them in product space, possibly creating more variety in the local market.
This mechanism could give rise, again, to more variety throughout product space, including
additional minority-targeted stations.2

The identity of owners – for example whether they are ethnic minorities – can also
affect programming. For example, black station owners may be better informed about
programming of interest to their target audiences. If so, then black-owned stations may serve
audience niches that would otherwise be unserved. Related, black owners may derive some
non-monetary benefit from serving black audiences that encourage them to program in ways

1 A companion study examines the relationship between ownership structure and the provision of news
programming. See Waldfogel (2011).
2 These mechanisms are explored in Steiner (1952), Rogers & Woodbury (1996), Alexander (1997), Berry and
Waldfogel (2001), DiCola (2010), and Sweeting (2010).
1


that non-minority owners would not. Under either rationale, the number of black-owned
stations in a market would affect the amount of black-targeted programming available in the
market.3
The amount of variety is potentially important for consumer well-being. Given
consumers’ heterogeneous tastes, adding varieties in a market can attract a greater share
of the population to consumption, in this case radio listening.4 Thus, it is possible to learn
something about how satisfying consumers find their radio options from examining the
determinants of overall radio listening. Of course, the last decade has been a period of
overall declines in radio listening, so the question here is simply whether markets with,
say, higher ownership concentration have more or less radio listening, after accounting for other
determinants, including overall time trends.

It is well documented that blacks and whites – and Hispanics and non-Hispanics – tend
to prefer very different radio programming options (Waldfogel, 2003). Hence, the availability
of options targeting minority preferences may have large effects on their well-being as
consumers of radio programming. It is worth noting at the outset that radio programming is
available free over the air, so we cannot infer the dollar value that listeners attach to the service.
Instead, we can simply document the responsiveness of particular groups’ listening to the
availability of different programming. A large increase in black listening in response to an
additional black-targeted station would be suggestive that the marginal station offers a service
differentiated from other stations on the dial.

3 See Dubin and Spitzer (1993) or Siegelman and Waldfogel (2001) for evidence of the impact of minority
ownership on programming availability.
4 Berry and Waldfogel (1999) estimate a model of radio listening and entry where the listening share has an
explicit utility interpretation.
2


These links between ownership – both ownership concentration and minority ownership
– and programming have been studied empirically in prior work. Using data for 1993, Berry
and Waldfogel (1999) examine the extent of business stealing in radio, concluding that the
status quo produces 2-3 times more stations than would be needed to maximize the welfare of
the buyers and sellers of advertising. Using data from 1993 and 1997, Berry and Waldfogel
(2001) examine the effect of the 1996 Telecommunications Act on programming variety in
radio, finding that markets with larger increases in ownership concentration experienced
smaller growth in the number of stations, larger growth in the number of distinct formats
available in the market, and even larger growth in the number of non-duplicated varieties.
Finally, Siegelman and Waldfogel (2001) examine the link between minority ownership and
minority-targeted programming in 1993 and 1997, finding that markets with more minority-
owned stations offer a greater quantity of minority-targeted programming.
This study revisits these questions with newer data covering the period 2005-2009. In
particular, this study asks the following questions: a) how does minority ownership of radio
stations affect the amount of minority-targeted programming availability in a market? b) how
does ownership structure more generally (i.e. ownership concentration and large-group
ownership) affect the amount of minority-targeted radio programming? c) How do ownership
structure and minority ownership affect radio listening? That is, does increased ownership
concentration promote radio listening overall, or in minority-targeted formats in particular?
Similarly, does diversity of owners by race promote overall radio listening, particularly for
minority audiences?
The questions of interest are all causal questions (e.g. how does ownership structure
affect various outcomes). It is challenging to derive causal answers from observational data in
3


the best of circumstances. Academic studies generally have the luxury of being conducted only
after some policy change or analogous “experiment” gives rise to changes that can be used for
causal inference. For example the Telecommunications Act of 1996 relaxed ownership
restrictions, which unleashed a significant increase in ownership concentration as well as a
decrease in the number of minority-owned stations. Policy changes of this sort present
researchers with auspicious circumstances for causal inference. The period covered in the
present study, while it is a period of change (for example, radio listening is in slow decline), is
not a period of abrupt change. Policy has conducted no “experiments.” This study is not
prompted by a development that makes an answer easy to ascertain; instead, the study is
prompted by policymakers’ interest in the study’s question. My strategy in conducting the
study is to estimate the usual sorts of regression models but to be cognizant of what one can and
cannot say about the results.
The study proceeds in three sections after the introduction. Because the introduction
has already motivated the question and mentioned the relevant prior literature, the document
does not include separate theory or literature review sections. Section 2 describes the data used
in the study. I then turn to three results sections, comparing findings with prior findings in
context. Section 3 provides the station-level analyses, for example comparing the distribution
of broadcast formats for minority and non-minority owned radio stations. Section 4 provides
market-level analyses, for example on the relationship between the number of minority-owned
stations and the number of minority-targeted stations, using both cross sectional and panel
techniques. The conclusion summarizes the findings.

II.
Data
4


The data analyzed in this study (the “GFI”) are station-level data for up to three points
in time (05, 07, 09) indicating station format, station listening (from Arbitron), and station
ownership information. Whether stations are minority-owned is available for 2005 and, in a
different format, for 2007 as well; but this information is not available for 2009. Arbitron
listening data are available for 2005 and 2007 but not, for this study’s purpose, for 2009. The
data also include radio market-level information on demographics (as well as direct measures
of, say, the number of minority-owned stations that could also be calculated from the station-
level data).
I have data on station ownership as of the end of each of the study years (2005, 2007,
2009). To make the Arbitron listening data comparable with the other GFI, I average Fall 2005
with Spring 2006 to create what I term 2005 data, and I average Fall 2007 with Spring 2008 to
create 2007 data.
The overall (all persons aged 12 and over) listening are available for all stations. Data
on black (Hispanic) listening are available for all of the stations in about 85 (62) markets in
2005 and 2007.
The data I employ exist in three separate files, two at the station level and one at the
market level. The two station level data sources are LongitudinalRadio_Long.dta, provided by
the FCC (in the GFI), and various radio listening data sets provided by Arbitron. The GFI
include information on station ownership. In every year the data report the name of the
station’s owner (parent company). The data also include some information on whether the
station is owned by a minority. In 2005, the data include a binary measure, whether a station is
minority-owned. In 2007, the data include a more detailed measure, whether the station is
owned by blacks, Hispanics, Asians, Native Hawaiians, or American Indians.
5


We have two kinds of information on station ownership, whether the station is minority-
owned, and the identity of the station’s owner. The latter information allows me to determine
the number of stations in each station’s group. Because of the importance of these ownership
variables to the study, we characterize the ownership data here.
The minority ownership data are not consistent across years: As Table 1 shows, of 8236
stations in metropolitan area including ownership data in 2005, 260 are coded as minority
owned. Of 6382 metro stations including ownership data in 2007, 278 are coded as black
owned, while another 229 are coded as Hispanic owned, 68 are coded as Asian owned, 11 are
American Indian owned, and 8 are owned by native Hawaiians. There are no data indicating
minority ownership for 2009.
Two features of these ownership data limit the analysis I can undertake with them.
First, the apparent change in meaning between 2005 and 2007 means that it is not possible to
use the difference between a station’s coding in 2005 and 2007 to learn whether the station’s
ownership changed over that period. Accordingly, it is not prudent to use the minority
ownership data for longitudinal analysis. Reinforcing this shortcoming is the absence of 2009
data. Hence, my strategy with these data will be simply to study 2005 and 2007 cross sectional
relationships between minority ownership and minority targeting, etc.
Table 1 also provides information on the distribution of stations across ownership
groups of different sizes. I use all of the stations in the data (14,375) to calculate the size of
each station’s ownership group. Among the stations located inside of a metro area, roughly one
sixth of the stations are singletons (their owners have no additional stations). Just under 30
percent are in groups including 2 to 9 stations, 13-15 percent are in groups of 10 to 24, just
6


under 10 percent are in groups of 25 to 49, and roughly 30 percent of stations are in large (50+)
station groups.
The GFI cover 14,375 stations in each of the three study years. Many of these stations
are located outside of the metro areas I examine in this study. The Fall 2005 Arbitron data
contain listening information for 32,886 station-metro combinations. This exceeds the number
of stations, since a station appears in both its home metro as well as wherever else listeners
report listening to it. In Spring 2006, the Arbitron survey contained 31,854; In Fall 2007,
32,525; and in Spring 2008, 31,667.
Both the GFI and Arbitron datasets contain variables for call sign, and whether the
station broadcasts in AM or FM. However, the data do not match perfectly. When I merge
the BIA and Arbitron data for 2005 and 2007, 10,667 stations in the Arbitron data do not match
with BIA, while 27,370 do. Fewer (622 of 14,375) BIA station observations do not match with
Arbitron. The match rate for 2007 is very similar. Many non-matching stations in 2005 are
Canadian (142 have call signs beginning with “C”). Another 7637 Arbitron stations report zero
or missing AQH listening shares. In 2005, the matching stations account for 88 percent of
Arbitron listening. In 2007, the matching stations account for 93 percent of total listening. For
market-level analyses, I aggregate the FCC and Arbitron data, then merge them by market
rather than by station.


III.
Station Level Analyses
a. Do Minority Listening Preferences Differ?
7


Our goal is to see how extensively minority consumers listen to the radio and, in
particular, to see whether station ownership affects this. If minority and white listeners had
identical tastes in radio programming, then there would be no separate question of whether
minorities are well served. Before proceeding further, it is useful to check whether the rather
stark differences between black and white –and between Hispanic and non-Hispanic – listening
tendencies documented elsewhere are visible in data for a more recent period.5 Table 2 reports
total AQH listening (aggregated across Fall 2005, Spring 2006, Fall 2007, and Spring 2008) to
each of the BIA broad formats (what are termed “Format Categories” in the GFI). Only
listening in markets where Arbitron tallies black listening are included in this table. The
formats are listed in descending order by the amount of black listening. The table shows that a
single format – Urban – attracts half (51.2 percent of black listening), while it attracts less than
five percent of nonblack listening. Urban, along with two more formats – religion and
Contemporary Hit Radio – account for 71 percent of black listening. And five formats – the
five arrived at with the addition of Jazz and Adult Contemporary – collectively account for 84
percent of black listening. These five formats collectively attract a third percent of nonblack
listening (from in the markets where Arbitron tallies black listening separately). Although the
particular format designations differ from the studies of the 1990s, the broad pattern remains
quite similar: blacks and whites listen to stations broadcasting in different formats. Table 2 also
reports the share of each format’s audience that is black. Urban stations’ audiences are 73
percent black, and jazz and religion stations’s audiences are roughly 40 percent black.
Table 3 repeats the exercise, juxtaposing Hispanic and non-Hispanic listening (for
markets where Arbitron separately tallies Hispanic listening). The differences are also large.

5 Waldfogel (2003) documents that in 1993 and 1997, radio broadcast formats attracting two thirds of black
listening collectively attracted 5 percent of non-black listening. Related, Spanish-language programming attracted
roughly half of Hispanic listening and just 2 percent of non-Hispanic listening.
8


Spanish radio stations collectively attract 48 percent of Hispanic listening and only 0.6 percent
of non-Hispanic listening. Spanish-language station audiences are 96 percent Hispanic. This
pattern echoes the pattern observed in the 1990s.
Minority listeners prefer programming in different formats. This suggests that the
availability of programming in these distinct formats affects the well-being of minority
consumers, in their capacity as radio listeners.

b. Minority Ownership and Minority-Targeted Programming
The simplest way to document a relationship between minority ownership and targeting
is to examine the joint distribution of minority ownership and targeting. Table 4 does this. Of
the 8236 DMA stations with ownership data in 2005, 260 (3.1 percent) are minority owned.
Yet, the share of stations owned by minorities differs substantially across formats (one rejects
independence with a chi squared test, at p-values near 0). Minority ownership is relatively
common for Spanish stations (16.6 percent are minority owned), Urban (11.0), and Religion
(4.6). Recalling that these formats attract disproportionate shares of minority listening, it is
clear that formats with relatively high minority ownership are also formats that cater to
minorities.
Of the 260 minority owned stations in 2005, over one third (38 percent) were Spanish,
22 percent were Religious, and 15 percent were Urban. That is, three quarters of minority-
owned stations were in formats that have proportionally large minority audiences. Yet, as
Table 4 indicates, the vast majority of stations in these formats are not minority-owned.
The latter half of Table 4 examines 2007 separately, revealing similar patterns. The
finer ownership variable, which breaks minority ownership into black, Hispanic, and other
9


constituent parts, reveals elevated black ownership among Urban stations (108 of 331 Urban
stations, or 33 percent, were black owned in 2007) and Religious stations (15 percent) .
Operation of Spanish stations is elevated among Hispanic owners: 185 of 548 Spanish stations
(34 percent) were Hispanic owned in 2007.
As in the 1990s, most minority-owned stations broadcast in formats that appeal
disproportionately to minorities. But also as in the 1990s, most stations broadcasting in these
formats are not minority owned. See Siegelman and Waldfogel (2001).
The disproportionate tendency for minority-owned stations to broadcast in formats that
appeal to minority listeners provides suggestive evidence that minority ownership is beneficial
to minority audiences. But the fact that minority-owned stations are likely to broadcast in
formats appealing to minorities does not, by itself, indicate that the presence of minority-owned
stations raises the availability of minority-targeted programming, since non-minority-owned
stations are also active in the provision of minority-targeted programming. (The answer to that
question depends on whether minority-owned stations displace other stations targeting the ames
audiences, and addressing that question requires analysis at the market level, which we
undertake below).

c. Group Ownership and the Availability of Stations by Format
Since the Telecommunications Act of 1996 liberalized ownership, many stations have
come to be owned by large radio station groups.6 The effects of these ownership groups on
programming is unclear a priori. Lower costs could allow groups to operate more stations,
offering more variety. Local ownership concentration might also induce jointly owned local

6 See Berry and Waldfogel (2001) for evidence of the ownership consolidation surrounding the Telecom Act of
1996.
10


stations to spread out in product space, increasing variety. Of course, common ownership of
multiple stations could also promote homogeneity, particularly across geographic areas but
possibly within areas as well. Group ownership could also allow firms to spread investments in
programming quality across more stations.
The station level data allow us, as a first step, to simply examine the relationship
between group ownership and formats. Table 5 summarizes these data. First, stations are
classified according to the size of the groups including them. This classification is done using
all of the active 14,375 stations in the GFI. This table reports the joint distribution for the 8650
metro area stations operating in 2009. As the table shows, 18 percent of stations are singletons,
31 percent are in groups of 2-9, 15 percent are in groups of 10-24, 10 percent are in groups of
25 to 49, and 27 percent are in groups of 50 or more.
Group size is not independent of format, however. Not only can one reject
independence in a chi-squared test, but the deviation from independence is large. For example,
nearly half of jazz stations are singletons, as are 20 percent of Spanish stations. At the other
extreme, nearly half of contemporary hit radio stations – and nearly half of urban stations - are
in groups of 50 or more, as are about a third of country and sports stations. On the other hand,
relatively few (15 percent) of Spanish stations are in groups of 50 or more, as are even fewer (9
percent) of jazz stations. The point emerging from Table 5 is that there is no single relationship
between group size and the provision of programming of interest to minority listeners.
We can also ask how listening varies across stations with the size of the stations’
underlying ownership groups. The latter three columns of Tables 6a and 6b examine this
question, via regression of log AQH listening on broad and narrow format dummies,
respectively (what are termed “Format Categories” and “Format Groups,” respectively, in the
11


GFI), metro area dummies, and a dummy for 2007, along with dummies for whether the station
is part of a group of 2-9, 10-24, 25-49, or 50+. Standalone stations are the excluded category.
Column (4) shows that overall listening is higher for stations in larger groups, and the size
effect is monotonic in size. Relative to singletons, stations in groups of 2-9 have 11 percent
higher listening. Stations in groups of 50 or more have 48 percent higher listening than
standalone stations, again, compared with stations in the same market and format. Black and
Hispanic listening also bear monotonic, positive relationships with the station’s group size.
Relative to standalone stations, Hispanic listening is 9 percent higher at stations in groups of 2-
9, and it is 58 percent higher at stations in groups of 50 or more. We obtain similar results with
dummies for broad formats (“format categories”), in Table 6b.
If the size of the audience that a station can attract provides an indirect measure of the
station’s appeal, the fact that stations in larger groups attract larger black, white, and overall
audiences than do their geographic and format peers in smaller groups suggests that group
ownership produces some benefit for listeners. It should be noted, however, that stations differ
in the power of their signals. It is possible that stations owned by larger groups have stronger
signals which allow them to reach more listeners. In the absence of data on signal strength, it is
difficult to draw strong conclusions.

d. Station Listening and Minority Ownership
The station-level data also allow us to examine the appeal of minority-owned stations
(relative to non-minority-owned stations) directly in terms of listenership. Do minority-owned
stations attract more listeners than their non-minority-owned counterparts? Because average
station listenership may vary across metro areas and across formats, this comparison is more
12


reasonably done with statistical controls (in this case dummies) for both format and metro area.
See Tables 6a for an analysis controlling for narrow formats, or “format groups” and 6b for an
analysis controlling for broad formats, or “format categories.”
Combining the 2005 and 2007 data, minority-owned stations attract an average of 15
percent less AQH listening overall, after accounting for metro area and narrow BIA format
(standard error = 5.0 percent). If we restrict attention to black listening (in markets with black
listening data), we see that minority-owned stations attract 7.3 percent greater black listening
than their within-market, same-format peers, although this difference is statistically
insignificant (s.e. = 6.2 percent). Restricting attention to Hispanic listening, minority-owned
stations attract 17.8 percent less Hispanic listening, and this difference is statistically significant
(s.e. = 6.7 percent). Regressions using broad format dummies (based on “format categories”)
give similar results. See Table 6b.
The suggestive station-level evidence is thus a bit mixed. On one hand, minority
ownership is positively associated with the provision of minority-targeted formats, which the
minority listeners prefer over other formats. Minority-owned stations attract similar amounts of
listening to non-minority-owned stations in the same narrow format and market. Minority-
owned stations attract substantially less Hispanic listening, compared with other stations in the
same format and metro area.7 If station power varies systematically with owner race, then the
same caveats mentioned above apply here as well. That is, differential listening by owner race
may reflect differential station power, rather than the appeal of the station’s programming per
se.


7 Similar results emerge when we use only 2007 and include separate right-hand side measures for whether a
station is black-owned, Hispanic-owned, or owned by another member of a minority group.
13


IV.
Market-Level Analyses
Our ultimate goal is to determine whether factors that are in principle controllable with
public policy – aspects of ownership structure – affect the well-being of minorities, in their
capacity as radio listeners. The steps along the causal chain for this mechanism include the
following: a) the possible effect of policy on ownership structure, b) the possible effect of
ownership structure on program targeting, and c) the possible effect of the availability of
various kinds of programming on the tendency for consumers to listen (which, we infer,
generates satisfaction for the listeners).
This section attempts to address questions (b) and (c) using cross sectional data at the
metro area level. We begin by simply characterizing the availability of radio programming in
various formats in markets of different sizes, to get a sense of which formats are more prevalent
in their availability. (If every format were available everywhere, then it would be unlikely that
any particular group did not have access to its preferred programming, and there would be little
scope for ownership to affect programming availability). We then ask whether additional
minority-targeted stations attract a greater share of minorities to radio listening, finally turning
to the question of whether additional minority-owned stations in a market raise the total amount
of minority-targeted programming in the market. Throughout, I attempt to be careful about the
assumptions needed for causal inference in this context and explicit about the plausibility of
these assumptions.

a. Availability of Different Programming Formats
The 300 metro areas in the study had an average of 30.4 stations available (according to
FCC definitions) in 2009. Because programming in different formats appeals to different
14


groups, it is useful to characterize the availability of stations in the various formats. We can do
this two ways through numbers (the number of, say, country stations in a market) and presence
(whether a market has a station in a particular format). Table 7 shows the average number of
stations in each format, as well as the share of markets with a station in the format. Religious
stations are the most common: markets have an average of 4.7 stations, and 95 percent of the
markets have at least one. Country and news stations are the next most common. Markets have
an average of 2.9 stations in each of these formats, and 95 and 94 percent of markets have
country and news stations, respectively. Other common formats include album oriented rock,
adult contemporary, contemporary hit radio, oldies, rock, and sports. The least commonly
available formats include middle of the road, easy listening, nostalgia/big band, public, jazz,
urban, Spanish, and classical.8
Figure 1 shows that format presence varies across markets of different sizes, in
particular across population deciles. The upper left depicts the relationship between market
population decile and format presence for the five least commonly available formats: easy
listening, ethnic, nostalgia, middle of the road, and jazz. Moving clockwise, the panels depict
more commonly available formats. Some formats, such as rock, country, news, religion, and
adult contemporary music are available in markets of all sizes, while formats such as those in
the upper left, as well as Spanish, jazz, and classical music, are available in few small markets.
Audiences interested in the formats depicted in the lower right panel face options
virtually everywhere, while audiences interested in other format face fewer targeted options.
The formats appealing most clearly to black and Hispanic audiences – urban, Spanish, and jazz
are among those that are typically unavailable in markets below the median size.

8 “Public” is primarily educational stations. Most NPR-affiliated stations are in the “news” category along with
commercial news stations. See Waldfogel (2011).
15



b. Targeted Programming and Group Listening
It has been clearly documented elsewhere that markets with a greater variety of
programming tend to attract a greater share of consumers in the market to radio.9 Moreover,
there is specific evidence that markets with a larger number of minority-targeted options attract
greater shares of minority consumers to radio listening (Waldfogel, 2003). We now revisit
this question, in particular the relationship between targeted programming and listening, using
recent data.
A natural way to document whether additional variety attracts consumers to radio
listening is to regress the share of a group’s population listening to radio on measures of the
amount of variety available locally. This could be the number of stations, the number of
distinct varieties, or the number of stations in each of a number of varieties (such as the number
of black-targeted stations, etc.). If our goal is to interpret the coefficients on the variety
measures as causal impacts of variety on listening, there is a concern. If local radio markets
were operated “experimentally,” with different numbers of stations – and varieties – available
in different markets for reasons unrelated to audience interest, then inference would be easy.
With station availability determined randomly, any resulting variation in the share of population
listening to radio would reflect the causal impact of differing station configurations.
Real life has different concerns: stations presumably enter markets where they expect
sufficient interest in their programming and advertising to cover their costs. We would then see
a large number of stations in places with large levels of interest in radio listening, but the
relationship would not simply reflect the impact of variety on listening. Rather, the relationship
might reflect the relationship between the appetite for radio listening and both station entry and

9 See Berry and Waldfogel (1999), Rogers and Woodbury (1996), and Alexander (1997).
16


listening. To say this using econometric jargon, one might be concerned that station
availability is endogenous.
A solution to this problem is a source of variation in our variety measure that is not
driven by tastes for radio listening. A natural candidate here is market size, measured by
population. A market with larger population can support more stations for a given per-capita
interest in radio listening (and resulting advertising). As long as markets of different size do
not have different levels of underlying interest in radio programming, market size can serve as
an instrument for a variety measure. Using instrument variables techniques, we can derive an
estimate of the causal impact of additional stations on radio listening.
We are interested not just in the effect of variety on overall listening but also on the
effect of minority-targeted variety on listening. We can use an analogous approach, using the
size of the local minority population as an instrument for the number of minority-targeted
stations. Before proceeding, a note on the description of “minority-targeted” is in order. For
Hispanic targeting I simply classify Spanish-language stations as Hispanic targeted. Black
targeting is less clear cut. Based on the listening data in Table 2, the urban format is more
clearly black-targeted. Two other formats, jazz and religious, have substantial black listening.
I thus employ two different definitions of black targeting. I employ a narrow definition of
black targeting that includes only urban stations as well as a broad definition that includes
urban, jazz, and religious stations.
Tables 8-11 implement these ideas. Table 8a presents regressions of the share of
population listening to radio during an average quarter hour on measures of the numbers of
stations targeting blacks, Hispanics, and others. We include urban, religious, and jazz stations
as a broad group of formats targeted at black listeners, we treat Spanish stations as Hispanic-
17


targeted, and we classify the remaining formats as “white-targeted.” (Table 8b revisits the
analysis with the narrow definition of black targeting). The first column of Table 8a reports a
regression of overall AQH on the numbers of black, Hispanic, and white-targeted stations.
Both of the statistically significant coefficients (on Spanish-language and other stations) are
positive, indicating that markets with more stations in each of the these categories have more
radio listening. It’s also worth noting that the coefficients are small relative to the constant
term. AQH listening averages roughly 12.3 percent, and the constant term is 12.1, while a
market with an additional white-targeted station has AQH listening that is 0.02 percentage
points higher. That is, markets with an additional white-targeted station have listening that is
0.2 percent higher.
Columns (2) and (3) examine black and non-black listening in the markets with separate
black listening data. All three coefficients are positive and significant in the black regression
(column 2), and the black coefficient is highest. Column (4) examines Hispanic listening. In
this regression, only the Hispanic and white-targeted coefficients are significant, and the
Hispanic coefficient is roughly ten times larger. The picture that emerges from this table is that
a particular group’s listening is larger in markets with more stations targeting that group. In
addition, it’s clear that while overall (and non-black and non-Hispanic) listening is higher in
markets with more stations targeted to non-blacks and to non-Hispanics, the coefficient is
smaller for this majority group than for the minority groups. Presumably, this reflects the fact
that markets always have more majority-targeted stations, and the average marginal listening
increment declines with the number of products targeting each group. Table 8b, using the
narrow definition of black targeting, provides similar results.
18


Table 9 implements the instrumental variables strategy, with first-stage regressions of
group-targeted entry on the sizes of the three population groups. In two of three cases (black
and white-targeted entry), own coefficients exceed cross-group coefficients. Tables 10a and 10b
turn to second-stage IV estimates seeking to obtain the causal relationship between targeted
entry and group listening. Standard errors tend to be larger than for the OLS estimates in Table
8, but the coefficient on the relevant minority group tends to exceed the other coefficients. For
example, in column (2) of Table 10a, the black AQH listening share increases by 0.28 for each
additional black-targeted station, while it increases only 0.13 for each additional Spanish station
and does not vary with the number of other stations. By contrast, non-black listening (in
column 3) is more sensitive to white-targeted stations and less sensitive to the others. Hispanic
listening (in column 4) increases 0.26 in each Spanish station but increases only 0.05 per white-
targeted station and is statistically insignificantly related to black-targeted stations. Results in
Table 10b are similar.
Tables 8-10 – and Tables 10a and b in particular – indicate that members of minority
groups listen more in markets with more stations targeted specifically at their group. Under the
assumptions underlying the IV estimation, we can interpret this estimate as causal. That is, we
can infer that additional minority-targeted stations would raise minority listening. If we take
listening as an indicator of satisfaction that consumers derive from radio broadcast services,
then we can infer that additional minority group-targeted stations would raise the well-being of
minority consumers, in their capacity as radio listeners.

c. Minority Ownership and Minority Targeting
19



The next question is whether ownership structure affects targeting. We begin by
studying the race of the owner. Recall that for 2005 we only have a variable indicating
minority ownership (with no distinction among minority groups). For 2007 we can distinguish
black, Hispanic, and other group ownership.

Our strategy for exploring the effect of minority ownership on the amount of
programming targeting minorities is to ask whether markets with more minority-owned stations
have more minority-targeted programming, after accounting for other determinants of the
amount of programming diversity. This is very similar to strategies pursued in Siegelman and
Waldfogel (2001). As demonstrated above in Table 9, the amount of minority-targeted
programming bears a relationship to the size and mix of local population. Hence, we begin by
regressing, say, the number of broadly black-targeted stations in a market in 2005 on measures
of black, white, and Hispanic population, along with variables for the numbers of minority-
owned and non-minority-owned stations in the market in 2005. Column (1) of Table 11 reports
the result of this regression. The coefficient on the number of minority-owned stations is 0.55
(standard error =0.08). Column (2) repeats the exercise using 2007 data. The coefficient on the
number of black-owned stations is 0.62 (0.06), while the Hispanic and other coefficients are
roughly 0.3. Columns (3) and (4) repeat this exercise with the number of narrowly black
targeted stations. Coefficient patterns are similar.
Column (5) reports the analogous 2005 regression using the number of Hispanic-
targeted stations as the dependent variable. Markets with an additional minority-owned station
in 2005 have an additional 0.62 Hispanic-targeted stations.
Columns (7)-(10) repeat the exercises of columns (1)-(4), adding controls for the total
number of stations as well as the squares of the population terms. Patterns are somewhat
20


attenuated but still reflect a positive impact of minority ownership on minority targeting:
conditional on the number of radio stations in the metro area (and other controls), the markets
with an additional station owned by a black have roughly 0.1-0.4 additional stations targeted at
blacks, while markets with an additional station owned by an Hispanic have roughly 0.2-0.45
additional stations targeted at Hispanics.

These estimates are also similar to, albeit somewhat smaller than, estimates reported in
Siegelman and Waldfogel (2001) for the 1990s. SW run cross sectional regressions of black-
targeted stations on terms in population as well as the number of black-owned stations,
obtaining black ownership coefficients of 0.69 for 1993 and 0.90 for 1997. Their analogous
estimates for Hispanic ownership are 1.14 and 1.52. Because SW have measures of ownership
that are defined similarly over time, they are able to use longitudinal approaches that we cannot
implement here. The period 1993-1997 also surrounded a policy change – the
Telecommunications Act of 1996 – that gave rise to exogenous variation in the number of
minority owners. Making use of that policy “experiment” – and instrumenting for the change
in ownership with measures of market size – SW estimate black and Hispanic ownership
coefficient of 0.99 and 1.14 respectively.

Although the period under study in the current study lacks a policy change or other
plausible source of exogenous variation, the similarity of the estimates of minority ownership
on minority targeting across time periods, approaches, and contexts provides at least suggestive
evidence that the ownership coefficients measured in this study continue to reflect causal
impacts.

d. Ownership Groups, Variety, and Listening
21



We now turn to the relationships between ownership variables (large groups, number of
owners, and the largest local owner) and variety and listening. Here, the question is at the level
of the market rather than the station. We have two possible approaches: we can compare levels
of ownership variables and their outcomes across markets, or we can compare changes in
ownership and possible outcomes across markets. The cross-market approach asks, for
example, whether markets with fewer owners have more variety, given the number of stations
available. This approach is vulnerable to concern of unobserved heterogeneity: markets with
different numbers of owners, given the number of stations operating, may have different
amounts of variety on the dial for reasons unrelated to the effect of ownership concentration.
The use of market-level fixed effects avoids this problem, but the lack of a policy change in the
study period casts doubt on the promise that this approach holds for measuring the relevant
effects.

The first two columns of Table 12a examine the relationship between available varieties
(the number of “format categories” or “format groups” available locally) and ownership
variables, after accounting for the number of stations. The question these regressions seek to
address is whether, with a given number of stations, a market has more or less variety if it has:
fewer owners (and therefore more ownership concentration), a higher share of its stations in
large ownership groups (and therefore possible operating at lower costs), two more measures of
ownership concentration: the average number of commercial stations per, and the size of the
largest local ownership group.

Not surprisingly, markets with more stations have more varieties, as the radio station
coefficients indicate. Markets with a higher share of stations in large groups have more narrow
varieties, and markets whose largest local ownership group is larger have more varieties.
22



Overall listening – in column (3) – is higher in markets with greater ownership
concentration (fewer owners). It is also higher in markets with more of their stations in larger
groups and in markets with more stations per group and a larger largest group.

However, all of these results related to ownership variables disappear with the inclusion
of metro area fixed effects in columns (6)-(9). Moreover, we reject the hypothesis that the
unobservable fixed effect is independent of the explanatory variables for both variety and
listening. This casts doubt on the validity of using the cross-market variation for identifying the
causal impacts of these variables in Table 12a. (Table 12b reports an analogous set of
regressions using the HHI rather than the number of owners as a measure of concentration).
While some variables are significant in the cross market regressions of columns (1)-(3), none of
the coefficients on ownership variables remain significant with the inclusion of the metro area
fixed effects).

This leaves us unable to draw strong conclusions from a study of the relationship
between ownership variables and variety (and between ownership variables and listening) using
data for this recent period. The lack of a finding is not the same as a robust finding that there is
no effect. Rather, it seems reasonable to infer that the variation in explanatory variables that
reflect ownership does not shed particular light on the question. As mentioned in the
introduction to this study, existing studies making use of the policy “experiment” provided by
the Telecommunications Act of 1996, which substantially relaxed ownership restrictions and
unleashed a rapid change in radio station ownership concentration, do find that elevated
ownership concentration raised the number of varieties available, especially conditional on the
number of stations.

23


Conclusion

We have examined the relationship between various aspects of radio station ownership
and the provision – and consumption – of minority-targeted programming, with the following
findings:

1) As in the 1990s, blacks and nonblacks – and Hispanics and non-Hispanics – have
starkly different preferences in radio programming. Urban stations collectively attract
half of black listening but only four percent of nonblack listening. Spanish stations
attract nearly half of Hispanic listening and negligible amounts of non-Hispanic
listening.
2) Some minority-targeted formats – urban, jazz, and Spanish – are among the formats that
are less commonly available.
3) As in the 1990s, most minority-owned stations target minority listeners, but – also as in
the 1990s – most minority-targeted stations are not minority-owned.
4) Stations in large groups tend to attract more listeners – overall, as well as among blacks
and Hispanics – that do stations in smaller ownership groups, after accounting for metro
area and programming format.
5) The availability of minority-targeted stations attracts more minorities to radio listening.
This result emerges in OLS cross sectional investigations and cross sectional IV
approaches but not from the within-market variation between 2005 and 2007.
6) The presence of minority-owned stations in market appears to raise the amount of
minority-targeted programming.
24


7) The recent period provides mixed evidence on the relationship between ownership
concentration and variety: cross sectional regressions suggest that higher concentration
promotes variety, but longitudinal exercises (using within-market variation) produce no
statistically significant relationships. It should be noted, again, that the period 2005-
2009 contains no “policy experiments” so that the absence of detected relationships may
owe as much to the absence of an “experiment” as it does to the lack of a detected
effect.


25



References


Alexander, Peter, "Product Variety and Market Structure," Journal of Economic Behavior
& Organization
XXXII, (1997), 207-214.

Berry Steven and Joel Waldfogel, "Free Entry and Social Inefficiency in Radio
Broadcasting," RAND Journal of Economics XXX, (1999a), 397-420.

Berry Steven and Joel Waldfogel, "Public Radio in the United States: Does it Correct
Market Failure or Cannibalize Commercial Stations?" Journal of Public Economics LXXI,
(1999b), 189-211.

Berry., Steven T. AND Joel Waldfogel. “Do Mergers Increase Product Variety? Evidence from
Radio Broadcasting.” Quarterly Journal of Economics (2001): 1009-1025.

DiCola, Peter. “FCC Regulation and Increased Ownership Concentration in the Radio
Industry.” Northwestern University School of Law, July 16, 2010.

Dubin, Jeffrey A. and Matthew L. Spitzer. “ Testing Minority Preferences in Broadcasting.”
USC Law Center Working Paper No. 94-8, 1993.

Rogers, Robert and John Woodbury, “Market Structure, Program Diversity, and Radio
Audience Size,” Contemporary Economic Policy, XIV (1996), 81-91.

Siegelman, Peter and Joel Waldfogel , Race and Radio: Preference Externalities, Minority
Ownership, and the Underprovision of Programming to Black and Hispanic Listeners, in
Advertising and Differentiated Products (Michael R. Baye and Jon P. Nelson, eds., 2001)

Steiner, Peter, “Program Patterns and the Workability of Competition in Radio
Broadcasting,” Quarterly Journal of Economics LXVI, (May 1952), 194-223.

Sweeting, Andrew. “The Effects of Horizontal Mergers on Product Positioning: Evidence from
the Music Radio Industry.” RAND Journal of Economics, 41(2), 372-397, Summer 2010.
Waldfogel, Joel. "Preference Externalities: An Empirical Study of Who Benefits Whom in
Differentiated-Product Markets." RAND Journal of Economics, vol. 34(3), pages 557-68,
Autumn 2003.
Waldfogel, Joel. “Station Ownership and the Provision and Consumption of Radio News.”
Study prepared for the FCC, 2011.



26




Table 1: Minority and Group Radio Station Ownership, 2005 -2009

2005
2007
2009
2005
2007
2009
No Controlling Interest
0
480
na
0.0%
7.5%

Non-Hispanic White
0
5,308
na
0.0%
83.2%

Non-minority (2005 only)
7,976
0
na
96.8%
0.0%

Minority (2005 only)
260
0
na
3.2%
0.0%

Asian
0
68
na
0.0%
1.1%

American Indian or Alaskan
0
11
na
0.0%
0.2%

Black or African American
0
278
na
0.0%
4.4%

Hispanic or Latino,
0
229
na
0.0%
3.6%

Native Hawaiian
0
8
na
0.0%
0.1%








total
8236
6382
na










Group size
2005
2007
2009
2005
2007
2009
singleton
1,507
1,519
1,518
18.3%
17.7%
17.5%
2 to 9
2,328
2,548
2,646
28.3%
29.7%
30.6%
10 to 24
1,043
1,284
1,295
12.7%
15.0%
15.0%
25 to 49
739
816
860
9.0%
9.5%
9.9%
50 and up
2,620
2,404
2,331
31.8%
28.0%
26.9%







total
8,237
8,571
8,650






27


Table 2: Black and White Listening Patterns (BIA Broad Formats)
Format Category

Total

%black
cumul
cumul
listening

B

NB

Urban
66507
75.7%
54.4%
3.9%
Religion
24322
40.1%
64.9%
7.4%
Contemporary Hit Radio/Top 40
43750
19.9%
74.3%
15.9%
Jazz/New Age
14861
39.1%
80.6%
18.1%
Adult Contemporary
65016
7.4%
85.8%
32.7%
News
47023
7.6%
89.7%
43.2%
Sports
14322
10.5%
91.3%
46.3%
Talk
16409
8.3%
92.8%
50.0%
Oldies
21166
5.6%
94.0%
54.8%
Miscellaneous
9678
11.5%
95.2%
56.9%
Rock
40627
2.7%
96.4%
66.5%
Country
45120
2.3%
97.5%
77.1%
Classical
12942
4.6%
98.2%
80.1%
Album Oriented Rock/Classic Rock
25224
1.9%
98.7%
86.1%
Spanish
44073
1.0%
99.2%
96.7%
Public/Educational
4994
7.2%
99.6%
97.8%
Ethnic
1600
13.6%
99.8%
98.1%
Nostalgia/Big Band
3575
3.1%
99.9%
99.0%
Easy Listening/Beautiful Music
3370
2.0%
100.0%
99.8%
Middle of the Road
938
2.1%
100.0%
100.0%
Notes: total black and nonblack listening to stations in markets with separate black listening
data. Surveys for Fall 2005, Spring 2006, Fall 2007, and Spring 2008 are included. This table
includes only stations that match with the BIA stations, and the listed formats are BIA’s Format
Category variable.
28


Table 3: Hispanic and non-Hispanic Listening Patterns (BIA Broad Formats)
Format Category

TOTAL

%Hispanic
cumul H
cumul NH
Spanish
52079
96.4%
51.3%
0.6%
Contemporary Hit Radio/Top 40
37851
27.4%
61.9%
9.3%
Adult Contemporary
51956
15.4%
70.1%
23.2%
Urban
44745
15.2%
77.0%
35.1%
Rock
32198
11.6%
80.8%
44.1%
Oldies
17366
14.5%
83.4%
48.8%
Country
31107
7.7%
85.9%
57.9%
Album Oriented Rock/Classic Rock
20298
11.6%
88.3%
63.5%
News
38228
6.1%
90.7%
74.9%
Religion
15703
13.0%
92.8%
79.2%
Jazz/New Age
13455
13.2%
94.6%
82.9%
Talk
14146
8.9%
95.9%
86.9%
Miscellaneous
8648
14.1%
97.1%
89.3%
Sports
12547
8.7%
98.2%
92.9%
Classical
10679
5.7%
98.9%
96.1%
Ethnic
1604
23.9%
99.3%
96.5%
Public/Educational
4647
6.2%
99.6%
97.8%
Easy Listening/Beautiful Music
3485
6.0%
99.8%
98.9%
Nostalgia/Big Band
3019
6.7%
100.0%
99.8%
Middle of the Road
750
2.8%
100.0%
100.0%
Notes: total Hispanic and non-Hispanic listening to stations in markets with separate Hispanic
listening data. Surveys for Fall 2005, Spring 2006, Fall 2007, and Spring 2008 are included.
This table includes only stations that match with the BIA stations, and the listed formats are
BIA’s Format Category variable.
29



Table 4: Minority Ownership and Programming Format, 2005 and 2007

2005
2005
2005
2007
2007
2007
2007
2007

non-minority
minority
total
white
Hispanic
Black
other
total
Adult Contemporary
766
7
773
607
4
5
54
670
Album Oriented Rock/C
402
2
404
314
0
7
39
360
Classical
190
1
191
20
0
0
1
21
Contemporary Hit Radi
402
1
403
334
4
6
41
385
Country
793
5
798
720
3
7
77
807
Easy Listening/Beauti
35
1
36
24
1
0
0
25
Ethnic
84
9
93
49
4
4
27
84
Jazz/New Age
125
4
129
46
1
6
3
56
Middle of the Road
30
0
30
22
0
0
0
22
Miscellaneous
415
2
417
139
1
3
55
198
News
739
7
746
525
4
11
43
583
Nostalgia/Big Band
210
2
212
147
1
3
9
160
Oldies
399
8
407
321
2
6
27
356
Public/Educational
97
0
97
4
0
0
0
4
Religion
1,209
58
1,267
473
6
84
17
580
Rock
619
2
621
456
0
5
42
503
Spanish
499
99
598
325
185
10
28
548
Sports
351
4
355
347
4
4
50
405
Talk
287
8
295
248
7
9
19
283
Urban
324
40
364
186
2
108
35
331









Total
7,976
260
8,236
5,307
229
278
567
6,381


0



Table 5: Group Size and Station Format, 2009

singleton 2 to 9
10 to 24
25 to 49
50 and
total
up
Adult Contemporary
49
203
103
74
221
650

7.54
31.23
15.85
11.38
34
100

3.23
7.67
7.95
8.6
9.48
7.51







Album Oriented Rock/C
27
72
73
42
151
365

7.4
19.73
20
11.51
41.37
100

1.78
2.72
5.64
4.88
6.48
4.22







Classical
27
127
14
9
2
179

15.08
70.95
7.82
5.03
1.12
100

1.78
4.8
1.08
1.05
0.09
2.07







Contemporary Hit Radi
36
89
62
38
198
423

8.51
21.04
14.66
8.98
46.81
100

2.37
3.36
4.79
4.42
8.49
4.89







Country
65
258
140
99
259
821

7.92
31.43
17.05
12.06
31.55
100

4.28
9.75
10.81
11.51
11.11
9.49







Easy Listening/Beauti
6
9
1
1
1
18

33.33
50
5.56
5.56
5.56
100

0.4
0.34
0.08
0.12
0.04
0.21







Ethnic
34
21
7
29
9
100

34
21
7
29
9
100

2.24
0.79
0.54
3.37
0.39
1.16







Jazz/New Age
48
30
10
2
9
99

48.48
30.3
10.1
2.02
9.09
100

3.16
1.13
0.77
0.23
0.39
1.14







Middle of the Road
4
8
1
2
3
18

22.22
44.44
5.56
11.11
16.67
100

0.26
0.3
0.08
0.23
0.13
0.21







Miscellaneous
285
135
59
67
96
642
0



44.39
21.03
9.19
10.44
14.95
100

18.77
5.1
4.56
7.79
4.12
7.42







News
96
303
103
74
234
810

11.85
37.41
12.72
9.14
28.89
100

6.32
11.45
7.95
8.6
10.04
9.36







Nostalgia/Big Band
28
45
17
14
21
125

22.4
36
13.6
11.2
16.8
100

1.84
1.7
1.31
1.63
0.9
1.45







Oldies
70
186
89
56
127
528

13.26
35.23
16.86
10.61
24.05
100

4.61
7.03
6.87
6.51
5.45
6.1







Public/Educational
50
66
12
1
1
130

38.46
50.77
9.23
0.77
0.77
100

3.29
2.49
0.93
0.12
0.04
1.5







Religion
260
474
180
130
338
1,382

18.81
34.3
13.02
9.41
24.46
100

17.13
17.91
13.9
15.12
14.5
15.98







Rock
157
112
78
41
158
546

28.75
20.51
14.29
7.51
28.94
100

10.34
4.23
6.02
4.77
6.78
6.31







Spanish
147
227
154
92
110
730

20.14
31.1
21.1
12.6
15.07
100

9.68
8.58
11.89
10.7
4.72
8.44







Sports
32
133
91
47
167
470

6.81
28.3
19.36
10
35.53
100

2.11
5.03
7.03
5.47
7.16
5.43







Talk
47
93
48
23
72
283

16.61
32.86
16.96
8.13
25.44
100

3.1
3.51
3.71
2.67
3.09
3.27







Urban
50
55
53
19
154
331

15.11
16.62
16.01
5.74
46.53
100

3.29
2.08
4.09
2.21
6.61
3.83







1


Total
1,518
2,646
1,295
860
2,331
8,650

17.55
30.59
14.97
9.94
26.95
100

100
100
100
100
100
100



2




Table 6a: Station Listening and Ownership Type (Broad Format Category Dummies)

(1)
(2)
(3)
(4)
(5)
(6)

Log AQH
Log Black
Log Hisp
Log AQH
Log Black
Log Hisp
Listening
AQH
AQH
Listening
AQH
AQH
Minority Owned
-0.1111
0.2645
-0.2055




(0.0489)*
(0.0614)**
(0.0659)**



group 2-9



0.1440
0.0579
0.1034




(0.0316)**
(0.0542)
(0.0603)
group 10-24



0.3988
0.1549
0.4542




(0.0358)**
(0.0612)*
(0.0654)**
group 25-29



0.3860
0.0700
0.3977




(0.0394)**
(0.0686)
(0.0690)**
group 50+



0.5322
0.3625
0.6069




(0.0307)**
(0.0505)**
(0.0561)**
Constant
1.8067
0.6865
1.1207
1.4397
0.4429
0.6848

(0.0277)**
(0.0468)**
(0.0495)**
(0.0382)**
(0.0648)**
(0.0696)**
Observations
24325
6230
5965
26986
6777
6471
Number of Metro
306
135
101
306
135
101
areas
Standard errors in parentheses. * significant at 5% level; ** significant at 1% level. All regressions include dummies for year, metro area and broad
format.


3


Table 6b: Station Listening and Ownership Type (Narrow Format Group Dummies)

(1)
(2)
(3)
(4)
(5)
(6)

Log AQH
Log Black
Log Hisp
Log AQH
Log Black
Log Hisp
Listening
AQH
AQH
Listening
AQH
AQH
Minority Owned
-0.1521
0.0729
-0.1781




(0.0499)**
(0.0616)
(0.0666)**



group 2-9



0.1087
0.0253
0.0480




(0.0321)**
(0.0540)
(0.0620)
group 10-24



0.3539
0.1312
0.3762




(0.0365)**
(0.0611)*
(0.0675)**
group 25-29



0.3757
0.1570
0.3755




(0.0405)**
(0.0697)*
(0.0716)**
group 50+



0.4802
0.3403
0.5239




(0.0318)**
(0.0516)**
(0.0588)**
Constant
1.9625
1.0670
1.7813
1.6070
0.8216
1.3022

(0.2254)**
(0.3073)**
(0.3730)**
(0.2232)**
(0.3062)**
(0.3677)**
Observations
24325
6230
5965
26986
6777
6471
Number of Metro
306
135
101
306
135
101
areas
Standard errors in parentheses. * significant at 5% level; ** significant at 1% level. All regressions include dummies for year, metro area and narrow
format.



4


Table 7: Metro Area Station Availability, 2009
Format
presence number
Adult Contemporary
91.7%
2.3
Album Oriented Rock
79.3%
1.3
Classical
52.0%
0.6
Contemporary Hit Radio
81.0%
1.5
Country
95.0%
2.9
Easy Listening/Beautiful
5.7%
0.1
Ethnic
12.7%
0.4
Jazz/New Age
28.3%
0.3
Middle of the Road
6.7%
0.1
Miscellaneous
71.7%
2.3
News
93.7%
2.9
Nostalgia/Big Band
32.3%
0.4
Oldies
84.0%
1.8
Public/Educational
32.0%
0.5
Religion
95.3%
4.7
Rock
86.0%
2.0
Spanish
49.7%
2.6
Sports
82.3%
1.6
Talk
53.3%
1.0
Urban
48.3%
1.1

Total


30.4

Note: Calculations include 300 Arbitron metro areas.

5


Table 8a: Listening and Station Targeting (Broad Definition of Black-Targeted)

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)

Overall
Black
Nonblack
Hisp AQH
NonHisp
Overall
Black
Hisp AQH
AQH
AQH
AQH
AQH
AQH
AQH
Number of Urban,
-0.0090
0.1185
-0.0338
-0.0795
0.0371
-0.0034
0.1345
-0.1701
Religious, and Jazz
Stations

(0.0124)
(0.0327)**
(0.0191)
(0.0520)
(0.0234)
(0.0330)
(0.0955)
(0.1956)
Number of Spanish
0.0648
0.0936
0.0762
0.2105
-0.0272
-0.0675
-0.0432
-0.0278
Stations

(0.0059)** (0.0279)**
(0.0305)*
(0.0441)**
(0.0265)
(0.0489)
(0.1458)
(0.2552)
Number of other
0.0217
0.0223
0.0291
0.0275
0.0214
-0.0172
-0.1684
0.0387
Stations

(0.0040)**
(0.0091)*
(0.0065)** (0.0098)** (0.0058)**
(0.0246)
(0.0680)*
(0.1285)
Constant
12.0917
12.0636
11.5436
13.0986
11.7556
13.3423
18.0564
14.6896

(0.1157)** (0.3190)** (0.1695)** (0.4200)** (0.2165)** (0.6665)** (2.1347)** (4.2534)**
Metro FE
No
No
No
No
No
Yes
Yes
yes
Observations
598
263
263
186
186
598
263
186
Standard errors in parentheses. * significant at 5% level; ** significant at 1% level. Regression includes observations for both 2005 and 2007, as well as
an unreported year dummy. Standard errors are clustered at the metro area level in rows (1)-(5).


6


Table 8b: Listening and Station Targeting (Narrow Definition of Black-Targeted)

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)

Overall
Black
Nonblack
Hisp AQH
NonHisp
Overall
Black
Hisp AQH
AQH
AQH
AQH
AQH
AQH
AQH
Number of Urban
-0.0228
0.3991
-0.1402
-0.3103
0.0547
0.0422
0.1688
0.0690
Stations

(0.0335)
(0.0885)** (0.0463)**
(0.1433)*
(0.0767)
(0.0750)
(0.1999)
(0.4530)
Number of Spanish
0.0648
0.0813
0.0804
0.2149
-0.0222
-0.0613
-0.0447
0.0217
Stations

(0.0059)** (0.0254)**
(0.0313)*
(0.0446)**
(0.0266)
(0.0498)
(0.1483)
(0.2721)
Number of other
0.0207
0.0331
0.0268
0.0264
0.0254
-0.0121
-0.1196
0.0171
Stations

(0.0033)** (0.0067)** (0.0047)** (0.0094)** (0.0057)**
(0.0260)
(0.0724)
(0.1306)
Constant
12.0895
11.8460
11.6478
13.0083
11.7754
13.1263
17.3744
13.9345

(0.1147)** (0.3307)** (0.1611)** (0.4136)** (0.2240)** (0.7571)** (2.5660)** (4.7980)**
Metro FE








Observations
598
263
263
186
186
598
263
186
Standard errors in parentheses. * significant at 5% level; ** significant at 1% level. Regression includes observations for both 2005 and 2007, as well as
an unreported year dummy. Standard errors are clustered at the metro area level in rows (1)-(5).










7


Table 9: Station Targeting and Demographic Mix (First Stage Regressions)

(1)
(2)
(3)
(4)
(5)

Number of
Number of
Number of
Number of
Number of
Urban,
Urban Stations
Spanish
other Stations
other Stations
Religious, and
(narrowly black
Stations
(neither Urban,
(neither Urban
Jazz Stations
targeted)
Religious, Jazz,
nor Spanish)
(broadly black
nor Spanish)
targeted)
Black Pop(mil)
12.6717
4.8841
-17.1000
0.1524
7.9400

(3.9142)**
(1.0954)**
(13.9938)
(8.4954)
(9.9652)
Hispanic
-3.4367
-0.4989
0.8565
-4.7648
-7.7027
Pop(mil)

(1.4442)*
(0.4262)
(4.9768)
(3.4773)
(3.5225)*
Pop Not Black
1.1780
-0.2001
6.6367
13.3335
14.7116
nor
Hispanic(mil)

(0.6537)
(0.1626)
(5.4617)
(3.2604)**
(3.6559)**
Constant
4.7161
1.0213
0.0312
14.3304
18.0252

(0.3296)**
(0.1064)**
(1.0423)
(0.7473)**
(0.8760)**
Observations
598
598
598
598
598
R-squared
0.39
0.30
0.22
0.65
0.66
Standard errors in parentheses. * significant at 5% level; ** significant at 1% level. Regression includes observations for both 2005 and 2007, as well as
an unreported year dummy. Standard errors are clustered at the metro area level.


8


Table 10a: IV Estimates of Listening and Station Targeting (Broad Definition of Black-Targeted)

(1)
(2)
(3)
(4)
(5)

Overall AQH
Black AQH
Nonblack
Hisp AQH
NonHisp AQH
AQH
Number of Urban,
-0.5958
0.2803
-0.0820
-0.1800
0.1025
Religious, and Jazz
Stations

(1.1211)
(0.0659)**
(0.0353)*
(0.1656)
(0.0594)
Number of Spanish
-0.3969
0.1324
0.0856
0.2595
-0.0294
Stations

(0.8438)
(0.0630)*
(0.0338)*
(0.0918)**
(0.0329)
Number of other
0.2926
0.0010
0.0402
0.0511
0.0107
Stations

(0.4983)
(0.0170)
(0.0091)**
(0.0361)
(0.0130)
Constant
11.0993
11.1586
11.8915
13.0355
11.7260

(1.8536)**
(0.4119)**
(0.2209)**
(0.5336)**
(0.1914)**
Observations
598
263
263
186
186
Standard errors in parentheses. * significant at 5% level; ** significant at 1% level. Regression includes observations for both 2005 and 2007, as well as
an unreported year dummy. Standard errors are clustered at the metro are level. Dependent variable is the AQH share, the percent of population listening
to radio during an average quarter hour.




9


Table 10b: IV Estimates of Listening and Station Targeting (Narrow Definition of Black-Targeted)

(1)
(2)
(3)
(4)
(5)

Overall AQH
Black AQH
Nonblack
Hisp AQH
NonHisp AQH
AQH
Number of Urban
0.8944
0.7752
-0.2986
-0.6518
0.2901
Stations

(0.7824)
(0.1987)**
(0.1069)**
(0.5701)
(0.2102)
Number of Spanish
0.2653
0.0372
0.1272
0.3122
-0.0504
Stations

(0.1840)
(0.0590)
(0.0317)**
(0.1059)**
(0.0391)
Number of non-
-0.0691
0.0356
0.0251
0.0383
0.0157
Urban, non-Spanish
Stations

(0.0798)
(0.0099)**
(0.0053)**
(0.0259)
(0.0095)
Constant
12.7942
10.9288
11.9921
12.5859
11.9047

(0.6959)**
(0.4444)**
(0.2391)**
(0.5554)**
(0.2048)**
Observations
598
263
263
186
186
Standard errors in parentheses. * significant at 5% level; ** significant at 1% level. Regression includes observations for both 2005 and 2007, as well as
an unreported year dummy. Standard errors are clustered at the metro are level. Dependent variable is the AQH share, the percent of population listening
to radio during an average quarter hour.




10


Table 11: Minority Ownership and Minority Targeting

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)

Number of
Number of
Number of
Number of
Number of
Number of
Number of
Number of
Number of
Number of
Number of
Number of
Urban,
Urban,
Urban
Urban
Spanish
Spanish
Urban,
Urban,
Urban
Urban
Spanish
Spanish
Religious,
Religious,
Stations
Stations
Stations
Stations
Religious,
Religious,
Stations
Stations
Stations
Stations
and Jazz
and Jazz
and Jazz
and Jazz
Stations
Stations
Stations
Stations

2005
2007
2005
2007
2005
2007
2005
2007
2005
2007
2005
2007
Minority-Owned
0.5451

0.1596

0.6295

0.1780

0.0907

0.2213

Stations

(0.0849)**

(0.0298)**

(0.1587)**

(0.0767)*

(0.0298)**

(0.0436)**

Black-Owned

0.6224

0.3020

0.0344

0.3617

0.2414

0.0357
Stations


(0.0687)**

(0.0217)**

(0.1331)

(0.0617)**

(0.0217)**

(0.0341)
Hispanic-Owned

0.3164

0.0434

0.8993

0.0741

0.0181

0.4967
Stations


(0.0731)**

(0.0231)

(0.1416)**

(0.0733)

(0.0258)

(0.0405)**
Black Pop(mil)
9.2266
5.6695
4.0061
1.4774
-22.5473
-16.9557
23.9098
16.6673
12.2171
7.7053
-4.7122
-3.8644

(1.8199)**
(1.6416)**
(0.6397)**
(0.5186)**
(3.4015)**
(3.1804)**
(2.7301)**
(2.8659)**
(1.0590)**
(1.0082)**
(1.5508)**
(1.5828)*
Hispanic Pop(mil)
-4.8804
-3.7372
-0.9085
-0.0163
0.3915
-5.7485
-8.0873
-4.5678
-1.4661
0.0624
14.8602
8.5601

(0.8618)**
(0.9323)**
(0.3029)**
(0.2945)
(1.6108)
(1.8062)**
(1.3950)**
(1.6573)**
(0.5411)**
(0.5830)
(0.7924)**
(0.9153)**
Other Pop (mil)
1.8694
1.6932
0.0101
-0.0493
7.1949
7.7622
-1.4933
-1.2519
-2.0948
-1.6668
0.1987
0.6771

(0.4913)**
(0.4369)**
(0.1727)
(0.1380)
(0.9183)**
(0.8465)**
(0.8802)
(0.7826)
(0.3414)**
(0.2753)**
(0.5000)
(0.4322)
Black Pop Sq'd






-8.2602
-5.7862
-6.9154
-5.0199
2.5109
1.7103







(2.3540)**
(2.1506)**
(0.9131)**
(0.7566)**
(1.3371)
(1.1877)
Hisp Pop Sq'd






1.2632
0.7210
-0.0576
-0.2131
-2.1139
-1.5182







(0.3945)**
(0.3714)
(0.1530)
(0.1306)
(0.2241)**
(0.2051)**
Number of Varieties






0.2148
0.1671
0.1122
0.1005
0.0809
0.0325







(0.0898)*
(0.0812)*
(0.0348)**
(0.0286)**
(0.0510)
(0.0449)
Other Pop Sq'd






0.1603
0.0642
0.4910
0.3653
-0.3709
-0.1837







(0.2277)
(0.1833)
(0.0883)**
(0.0645)**
(0.1293)**
(0.1012)
RadioStations






0.1273
0.1252
0.0112
0.0057
0.0379
0.0390







(0.0215)**
(0.0195)**
(0.0083)
(0.0069)
(0.0122)**
(0.0108)**
Constant
3.8454
3.5495
0.7503
0.5416
-0.8385
-0.9773
-0.8334
-0.5197
-0.4139
-0.4189
-1.1802
-0.7411

(0.2834)**
(0.2592)**
(0.0996)**
(0.0819)**
(0.5296)
(0.5022)
(0.7673)
(0.7220)
(0.2976)
(0.2540)
(0.4359)**
(0.3988)
Observations
298
300
298
300
298
300
297
299
297
299
297
299
R-squared
0.49
0.55
0.40
0.57
0.25
0.33
0.70
0.71
0.57
0.66
0.79
0.83
Notes: Standard errors in parentheses, clustered on metro area. * significant at 5% level; ** significant at 1% level. Regressions include observations for
2005 in odd-numbered columns and observations for 2007 in even-numbered columns. The 2005 minority ownership variable does not distinguish
different groups.


11


Table 12a: Ownership Structure, Variety, and Listening (REVISED)

(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)

Number of
Narrow
Overall
Black
Hisp AQH
Number of
Narrow
Overall
Varieties
Varieties
AQH
AQH
Varieties
Varieties
AQH
(revised)
(revised)
Radio Stations
0.1570
0.4135
0.1870
0.2040
0.1344
0.0586
0.2617
0.0344

(0.0264)**
(0.0331)**
(0.0351)**
(0.0638)**
(0.0686)
(0.0439)
(0.0560)**
(0.0356)
Radio Parents
-0.0914
-0.1641
-0.2243
-0.2004
-0.1302
0.0152
0.0010
-0.0542

(0.0473)
(0.0613)**
(0.0597)**
(0.1068)
(0.1110)
(0.0612)
(0.0781)
(0.0507)
Share in 25+
0.3826
1.6266
0.4038
3.9085
3.8304
0.1786
0.9774
0.0116
Groups

(0.4913)
(0.6381)*
(0.6229)
(1.4720)**
(2.0868)
(0.4983)
(0.6363)
(0.3339)
Avg Comm'l
-0.1845
-0.6111
0.6583
0.1637
0.8401
-0.0187
-0.0789
-0.1598
Stns per Owner

(0.2061)
(0.2910)*
(0.2419)**
(0.4789)
(0.6994)
(0.1902)
(0.2428)
(0.1391)
Largest Local
0.3003
0.2366
0.2457
0.4926
0.4436
-0.0115
-0.1026
0.0152
Owner Group

(0.1017)**
(0.1324)
(0.1025)*
(0.2108)*
(0.2094)*
(0.0569)
(0.0727)
(0.0397)
Metro FE
No
No
No
No
No
Yes
Yes
Yes









Constant
7.1385
7.2256
4.1278
2.2707
1.8968
9.9814
10.1382
9.5851

(0.6784)**
(0.8907)**
(0.8445)**
(1.6800)
(2.1767)
(0.9894)**
(1.2634)**
(0.8088)**
Observations
897
897
596
263
186
897
897
596
Hausman test





28.19
18.93
61.83
p-val





0.0002
0.0084
0.0000
Robust standard errors in parentheses.* significant at 5% level; ** significant at 1% level. Hausman test compares fixed effects estimates in columns (6)-
(8) with random effects estimates (not shown).



12



Table 12b: Ownership Structure, Variety, and Listening (HHI Measure of Ownership Concentration) (REVISED)


(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)

Number of
Narrow
Overall
Black
Hisp AQH Number of
Narrow
Overall
Varieties
Varieties
AQH
AQH
Varieties
Varieties
AQH
(revised)
(revised)
Radio Stations
0.0905
0.2978
0.0312
0.0495
0.0299
-0.0030
0.1724
0.0057

(0.0145)** (0.0215)** (0.0094)** (0.0112)**
(0.0120)*
(0.0424)
(0.0534)**
(0.0245)
HHI
-0.0013
-0.0017
-0.0021
-0.0036
-0.0038
-0.0001
0.0000
-0.0001

(0.0003)** (0.0005)** (0.0005)** (0.0006)** (0.0006)**
(0.0002)
(0.0002)
(0.0001)
Share in 25+ Groups
0.1942
1.4822
0.3717
2.5722
1.0411
0.0411
0.8715
-0.0075

(0.4510)
(0.5792)*
(0.5202)
(1.2171)*
(1.6987)
(0.5784)
(0.7285)
(0.3345)
Avg Comm'l Stns per
-0.1376
-0.4792
0.8704
0.0770
0.3003
0.1554
0.1201
-0.0689
Owner

(0.1633)
(0.2173)*
(0.1817)**
(0.3676)
(0.5770)
(0.1913)
(0.2409)
(0.1106)
Largest Local Owner
0.1658
0.0890
0.0881
0.2446
0.0876
-0.0731
-0.1857
0.0268
Group

(0.0940)
(0.1262)
(0.0942)
(0.1640)
(0.1874)
(0.0660)
(0.0832)*
(0.0382)
Metro FE
No
No
No
No
No
Yes
Yes
Yes









Constant
9.7609
10.4300
7.7910
9.5909
11.2744
12.1123
12.7523
9.3684

(0.8949)** (1.3145)** (1.2346)** (1.6860)** (2.3518)** (1.3368)** (1.6837)** (0.7732)**
Observations
596
596
596
263
186
596
596
596
Hausman test





42.61
47.91
237.99
p-val





0.000
0.0000
0.0000
Robust standard errors in parentheses.* significant at 5% level; ** significant at 1% level. Hausman test compares fixed effects estimates in columns (6)-
(8) with random effects estimates (not shown).


13


Figure 1: Format Category Presence
Rare Format Presence
Moderately Rare Format Presence
by Population Decile
by Population Decile
.8
1
.8
.6
.6
.4
.4
.2
.2
0
0
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Easy Listening
Middle of the Road
Public/Educational
Urban
Ethnic
Jazz/New Age
Spanish
Classical
Nostalgia/Big Band
Talk


Moderately Common Format Presence
Common Format Presence
by Population Decile
by Population Decile
1
1
.8
.8
.6
.6
.4
.4
.2
.2
0
0
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Miscellaneous
Album Oriented Rock
Rock
Adult Contemporary
Contemporary Hit Radio
Sports
News
Country
Oldies
Religion


14



15


Note: We are currently transitioning our documents into web compatible formats for easier reading. We have done our best to supply this content to you in a presentable form, but there may be some formatting issues while we improve the technology. The original version of the document is available as a PDF, Word Document, or as plain text.

close
FCC

You are leaving the FCC website

You are about to leave the FCC website and visit a third-party, non-governmental website that the FCC does not maintain or control. The FCC does not endorse any product or service, and is not responsible for, nor can it guarantee the validity or timeliness of the content on the page you are about to visit. Additionally, the privacy policies of this third-party page may differ from those of the FCC.