FTC Staff Comment Before the FCC In the Matter of Amendment of Part 74 of the Commissions Rules Concerning FM Translator Stations

COMMISSION AUTHORIZED
UNITED STATES OF AMERICA
FEDERAL COMMUNICA nONS COMMISSION
V890015
)
In the Matter of Amendment of 	)
Part 74 of the Commission's 	)
Rules Concerning FM Translator)
Stations 	)
)
MM Docket No. 88-140
RM-5416
RM-5472
January 23, 1989
COMMENTS OF THE STAFF OF THE BUREAU OF ECONOMICS OF
THE UNITED STATES FEDERAL TRADE COMMISSION

I. 	Introduction and Summaryl
In a Notice of Inquiry (NOO released on June 2, 1988, the Federal
Communications Commission (FCC) solicited comment on its regulations
governing the use and licensing of low-power FM commercial translators
( lt translators").2 	The staff of the Bureau of Economics of the Federal
Trade Commission submitted a comment on this NOI suggesting that granting
increased flexibility in the use of translators may benefit consumers by
permitting greater competition and increasing listening options. s 	On
November 4, 1988, the National Association of Broadcasters (NAB) submitted
to the FCC a Supplement to its Reply Comments.· 	The Supplement focuses on
asserted weaknesses in the analytical and empirical analyses offered in
the FTC staff comments. 	This comment is submitted by the staff of the
Bureau of Economics of the Federal Trade Commission in response to the
FCC's invitation for' comments on the NAB's submission,s and it examines
1 	These comments are the views of the staff of the Bureau of
Economics of the Federal Trade Commission. 	They are not necessarily the
views of the Federal Trade Commission or of any individual Commissioner.
Please contact staff economist Robert P. Rogers at (202) 326-3382 should
you have any questions regarding our comments.
2 	Federal Communications Commission, In the Matter of Amendment of
Part 74 of the Commission's Rules Concerning FM Translator Stations, MM
Docket No. 88-140, RM-5416, and RM-5472 (released June 2, 1988).
3 	Federal Trade Commission, Bureau of Economics, Co'mments of the
Staff of the Bureau of Economics of the United States Federal Trade
Commission, MM Docket No. 88-140, filed August IS, 1988.
" 	National Association of Broadcasters, Supplement to Reply
Comments, MM Docket No. 88.;140, filed November 4, 1988, and National
Association of Broadcasters, Reply Comments, MM. Docket No. 88-140, filed
September IS, 1988.
5 	Federal Communications Commission, Order Reopening the Period for
Fjling Comments, MM Docket No. 88-140, December 5, 1988 in 47 CFR part
74, Federal Register, Vol 53, No. 250, December 29, 1988.

three points of the NAB critique to show how each is potentially
misleading.
The NAB asserts first that the FTC staff, by using erroneous data,
overestimated the impact of increasing the number of stations on the
number of formats and on total radio audiences. 	Second, NAB claims to
show that most new FM translators are not directed at low population areas
but rather at high population areas which already have adequate radio
service. 	Third, NAB asserts that FM translators are likely to drive out
established full power stations and thereby lessen the consumer benefits
derived from radio service.
In this comment, the staff will demonstrate that the NAB critique
does not, in fact, invalidate the FTC staff's estimate of the impact of
the number of stations on format diversity and listenership. 	Moreover, a
reconsideration of the issue suggests that, although the estimate in the
original comment was the be~t possible given the nature of the available
data, the comment probably understated, rather than overstated, the
relationship that staff sought to uncover. 	We also show that contrary to
NAB assertions, most new FM translator applications are targeted to low
population areas. 	Finally, we demonstrate that the hypothesized demise of
a full power radio station due to the entry of an FM translator would be
unlikely to reduce and may increase total consumer benefits.
II. 	The NAB and the FTC Statistical Analyses
A. 	TALC Survey Data and the FTC Format Model
The NAB Supplement criticizes the data used by the FTC staff to
uncover the impact of additional stations on the number of programming
2

formats and audience size. 	The NAB claims to show that the FTC staff used
incomplete data for its variable depicting the number of stations (NS)
available to the average listener in given radio markets. 	The NAB
Supplement describes a survey of radio listeners in every county in the
continental United States and in the metropolitan areas of Anchorage,
Alaska, and Honolulu, Hawaii; it is called the Total Audience Listening
Output (T ALO) Survey. 	The survey defines the number of radio stations in
a given county to include any radio station to which any survey respondent
listened during the survey period. 	The NAB asserts that this number is a
better indicator of the number of stations available to the average
listener than the number used by the FTC staff. 	The NAB asserts that the
number used by the FTC staff understates the number of available radio
stations in a market and therefore overstates the impact of the number of
stations on listenership.
The variable (NS) used by the FTC staff is the number of stations
listed in the Arbitron Market Survey. 	To be included in this list, a
radio station must fulfill the following three minimum standards: (1) be
reported as having been heard for at least five minutes by 10 or more
Survey participants; (2) have been heard by at least" 0.05 per cent
(unrounded) of the persons twelve years old and over in the market (as
shown by the survey); and (3) have garnered an average quarter-hour share
of 0.05 per cent tluring that portion of the day the station is on the
air. 6 -In other words, the station must have a minimally significant
market share; from this information, the staff inferred that most
6 	For Arbitron rating purposes, the day is divided into 15 minute
time periods. 	The Average Quarter-Hour Metro Rating is an average of the
station market shares for these fifteen minute periods over the day.
3

listeners in a given area could receive most if not all of these stations.
In contrast, the TALO sample for a given county includes any radio station
to which any survey participant listened for five minutes or more during
the survey periqd. 	Thus, it is not surprising that the T ALO estimate of
the number of stations available in a given area is often much larger than
the number of stations with enough listeners to get an Arbitron rating. 1
To put this issue in perspective, the methodology the FTC staff used would
count a station as present in the market if only one out of 2000 persons
(e.g. 0.05 "per cent of the total) in the market listened to it. 	In th"e
NAB's view this approach is underinclusive. 	The NAB would count a station
even if only one person in a sample of 6200 persons listened to it during
the survey petiod. 8 	Indeed, under the NAB's approach, if only one person
in a given market was able to receive and listened to a particular radio
station, that station would be counted as available to all listeners in a
market area. 	While not all the stations in the" Arbitron list are received
by all the listeners in a market, the number of these stations seems to be
a better indicator of the number of stations available to the average
listener than the number of stations on the T ALO list.
A brief review of the FTC staff's analysis will help the reader to
understand why the Arbitron Market Survey was used. 	The FTC staff first
1 	Exacerbating this tendency is the T ALO Survey requirement that the
respondents repOrt any radio station to which they listened during the
survey period even when they were traveling outside of the market area.
Given the size of the sample, it is unlikely that none of the respondents
tra veled during the survey period. 	When traveling, the respondent could
listen to radio stations that could never be heard by listeners in the
home market.
8 	The largest sample in the Arbitron survey (for the New York City
area) has 6200 respondents.
4

used regression analysis to estimate a formula for predicting the number
of programming formats in a market: the goal of the analysis being to
estimate the impact of increasing the number of stations on the
programming diversity available to the average listener. 	In the
regression formula, the (logarithm of the) number of stations is an
independent variable which is multiplied by a positive constant (which
gives information on the impact of the number of stations on formats),
and the (logarithm of the) number of formats is the dependent variable.
The results of this analysis indicate that increasing the number of radio
stations would increase the number of programming formats. 	In a second
regression analysis, the (logarithm of the) number of programming formats
is an independent variable that has a positive impact on the dependent
variable, total radio Iistenership.9 	Accordingly, increasing the number
9 	See Federal Trade Commission, Bureau of Economics, Comment's of
the Staff of the Bureau of Economics of the United States Federal Trade
Commission, MM Docket No. 88-140, filed August 15, 1988, for a fuller
description of the data. 	The equation used to determine the number of
programming formats present" in a market area is the following log-log
function:
InFOR 	= 	Inb o + b l InNS + b 2 InPOP + b g InX I + u
where FOR equals the number of formats, NS equals the number of stations
in an Arbitron-defined market, POP is the population in a radio market,
and Xl is a vector of other market characteristics that might affect the
number of formats.. The regression coefficients, bo- b l , b 2, and .b g are
constants that reflect the impact of each independent variable on the
dependent variable, and u is a residual. 	The regression coefficient for
NS, bI , is positive as predicted.
The regression equation for listenership is as follows:
In [PI (l - P)] 	•
where P equals the average proportion of the population listening to radio
over the day, and P is transformed into a log of the odds for the purpose
5

of radiQ stations in a market would result in an increase in the number of
programming formats. 	The increase in the number of programming formats,
in turn, would have a positive impact on total radio listenership.lO
Consequently, increasing the number of radio stations would indirectly
lead to increases in total listenership. 	The liberalization of the FM
translator rules, by increasing the number of stations, would lead to an
increase in the number of formats and therefore increase listenership.
As can be seen, the dependent variable in the format regression
analysis is the (logarithm of the) numb"er of formats present in an
Arbitron-defined market. 	Included as independent variables in this
regression are the number of stations in a market, the population in the
radio market, and a group of other variables reflecting demographic
characteristics that might affect the number of formats.
In" conducting the regression analysis to estimate the formula for
predicting the number of formats, our focus is upon the number of formats
of optimal regression analysis. 	ao, at. and a 2 are constants that reflect
the impact of each independent variable on the dependent variable, and v
is a resid ual. 	The regression coeff icien t for FO R, a l' is posi ti ve as
predicted.
10 	Our concept of the relevant listenership variable is different
from that of NAB. 	Essentially we define listenership as the average over-
the-day proportion of the population listening to the radio for any given
time period. 	NAB's def ini tion is the proportion of the population tha t
listened to the radio at some time during the day. 	Since nearly all the
population listen to the radio at some time during the day, it is almost
impossible to increase the NAB listenership total. 	Nevertheless, anyone
person who listens at some point during the day could increase her
listening time significantly thereby increasing the FTC staff listenership
figure. 	If a person increases her listening in response to an increase in
the available formats and/or stations resulting from increased
flexibility in the FM translator rules, then, consumer welfare could very
well have been increased. 	However, the proportion of the population
having listened to the radio at some point during the day (the NAB
listenership variable) may not have increased at all.
6

available to the average listener in a market area. 	We could have
computed the number of formats by using all of the stations included in
the TALO total, as the NAB suggests; however, it is not clear that all the
stations in the TALO survey are available to the majority or even to a
large plurality of the listeners in a given market area. 	Therefore, in
our view, a better indicator of the number of formats is likely to be
derived from the Arbitron-listed stations only -- that is, those having at
least a minimally significant listener share as defined by Arbitron. 	Even
though some listeners in a given market area may not receive all the
Arbitron-listed stations, these stations as a whole have such a large
share of the audience that we can plausibly infer that most listeners are
able to receive all or nearly all of the formats offered by the listed
sta tions. 11
In determining the number of stations in a market area for purposes
of the format regression analysis, there is a related reason to rely on
the number of sta·tions in the Arbitron list rather than the T ALO number.
The basic inquiry concerns not so much the number of stations in a
particular market but rather the number of stations having an impact on
the format decisions of the owners of the stations capable of reaching all
or most of the listeners in the market. 	We have defined this set as the
stations listed by Arbitron having at least a minimally significant
listener share in'the particular market. 	The NAB suggests that we expand
this set to include the T ALO stations -- that is, stations that were each
heard by at least one survey respondent in the market, regardless of
11 	As discussed in detail in footnote 12, the Arbitron-listed
stations in our sample always have over 85 percent of the audience.
7

whether they attain a minimal listener share. 	Because many of the
stations included in the T ALO Survey are heard by very few members of the
listening population in the particular Arbitron market, we believe that
these stations would not have a significant impact on the station owner's
format choice or at least would have less impact per station than the
stations contained in the Arbitron list. I2 	Thus, the exclusion of the
T ALO number does not appear to be a serious error in the statistical
analysis.
B. 	Specification ErrQr in the FTC Staff MQdel and its Meaning
AlthQugh nQt discussed by the NAB, there is yet another set Qf
stations that arguably could be relevant tQ the analysis 	-- thQse
stations that may Qr may not be heard in a particular Arbitron market but
that compete with some of the ArbitrQn-listed stations in other markets.
If a station listed by ArbitrQn in a particular market and included in NS
can be heard in other ArbitrQn markets, then its format decisions depend
in part on the number of competing stations in these other markets. 	For
example, a clear channel AM station based in Chicago may be regularly
heard in Charlotte, North Carolina, but no Charlotte station is regularly
listened to in Chicago. 	The cQmpetition of Charlotte stations can attract
listeners in Charlotte away from the Chicago station. 	Such audience
shifts may induce the Chicago station to compete for listeners in all
12 	To be sure, there are many markets in which the audience share
of all the stations listed in an Arbitron market is substantially less
than 100 percent. 	This indicates that much Qf the audience in these
markets listens to stations that are outside Qf the Arbitron Market area.
To reduce the errQrs associated with extensive listening of "Qutside"
stations, the FTC staff's empirical analysis' focused Qnly on those markets
in which the listed stations in a market accQunted for at least 85 percent
of the total listenership.
8

markets by changing its format. 	Therefore, even though they are not heard
in Chicago, these competing Charlotte stations can affect the number of
formats in Chicago.
The numb~r of these competing stations outside the market (here
called NC) should be included as a variable in the format equation.
Because NC is not included in addition to NS in the FTC staff's format
equation, our statistical analysis may be biased. 	Even though these
outside stations may not be heard in our sample market, they may affect
the format decision of the station owners in the sample market. 13 	As
discussed below, the bias was unavoidable given the available data, and in
any event, we doubt that, had the data been available, it would have led
us to conclude that the results we reported overstate the effect of the
number of stations on format diversity and listenership.
Stations located outside a given market and not included on the
Arbitron list for the area can play an important role in a given station's
format decisions. 	Nevertheless, in the great bulk of the sample (87 out
of a total 115 markets), this was not a problem because there were no
stations with a significant market share in more than one market. 	In four
of the remaining markets, the station overlap was with another market in
the sample. 	In these markets, the overlap is only one station, and the
market share of the station is very small in either one or both markets.
In the remaining 24 sample markets, there is an overlap with a market
not in the sample. 	For each of these out-of -sample markets, the share of
13 	We do not know whether NC would have the same effect on the
number of formats in our sample market as NS. 	Had we accurate data on NC,
we would have introduced NC and NS as distinct independent variables in
the regression analysis and explicitly tested statistically whether. the
format impacts of both were identical.
9

listeners captured by the Arbitron-listed stations in those markets is
less than 85 percent, our criterion for including a market in our sample.
The failure of these out-of-sample markets to satisfy this criterion
suggests that much of the listening in these markets is attributed to
other unlisted stations. 	As a result, attempting to include these out-of-
sample listed stations as NC in any regression analysis would likely
result in a substantial nonsystematic under-measurement of that variable
and thus lead to biases in the regression coefficients.
However, the exclusion of NC from the staff's regression model could
also result in biased regression coefficients. 	One way of gauging the
extent of this bias is to reestimate the format regression excluding from
the sample the 28 markets with station overlaps. 	The results of this
reestimation do not differ from those of the original staff analysis,
suggesting that exclusion of NC did not lead us to conclude erroneously
that increasing the number of stations (as a consequence of greater
. programming flexibility for FM translators) would increase listenership.H
Nonetheless, some might still contend that even for the remaining 87
sampled markets, there may be other competing stations outside the sampled
markets that we have not considered. 	By not including these outside
stations, our analysis may still be biased.
If an important and relevant factor is omitted from a regression
analysis, the estimated effect of an included variable on the dependent
variable is related to the effect of the omitted variable on the dependent
14 	A statistical procedure called the Chow test indicates that the
hypothesis that the regression model for the attenuated 87 observation
regression sample is the same as that for the complete 115 observation
sample can not be rejected. 	See J. Kmenta, Elements of Econometrics
(1986), pp. 420-421.
10

variable. 15 	Thus, the estimated impact of NS, an included variable, on
the number of formats is related to the effect of NC, the omitted
variable, on formats. 16 	One can think of the estimated effect of NS on
formats (and therefore listenership) as a weighted average of the "true"
effect of NS on formats and the "true" effect of NC on formats. 	Whether
the format effect that the staff has attributed to NS is magnified or
reduced as a result of omitting NC depends on two factors. 	The first
factor is the effect of NC on the number of formats in the market. 	The
second factor is the correlation between NS and Nc. 17 	If both of these
factors are positive or negative, the estimated effect of NS on the number
of formats will be overstated. 	If one factor is positive and the other
negative, the estimated format effect of NS will be understated.
We expect that the effect of increases in NC on the number of formats
offered in the market would be positive. 	Other things equal, the presence
of a greater number of competing stations inside (NS) and outside (NC) any
particular Arbitron market would likely result in more formats being
broadcast by the inside stations. 	The greater the number of competing
stations, the smaller the potential audience that any station could expect
15 	A complete discussion of the effects of omitting relevant
variables can be found in J. Johnston, Econometric Methods (1984), pp.
260·261. 	For purposes of the discussion in the text, we have simplified
the exposition but without any loss in generality.
16 	While we regard exclusion of the T ALO stations as appropriate,
their inclusion as part of NC would not affect our discussion of the
understatement of the staff -estimated effect of NS on the number of
formats offered in a particular Arbitron market. 	The T ALO stations can be
regarded as another component of NC.
17 	More precisely, the relationship between NC and NS is measured by
the partial correlation coefficient, which holds other factors affecting
the relationship (such as market population and demographics) constant.
11

to garner from more popular formats. 	Therefore, more stations will tend
to choose specialized formats when the number of competing stations is
larger (NS plus NC) than when the number of competing stations is smaller
(NS).
In contrast, the correlation between NS and NC is likely to be
negative: the higher is NC, the lower is NS likely to be. 	Other things
being equal, increasing the number of outside stations competing with the
stations inside the Arbitron market will reduce the number of inside
stations that are financially viable because the total potential audience
that can be garnered by the inside stations is reduced by the presence of
greater outside competition. 	If some, or many, of the inside stations
have significant audiences outside the sampled Arbitron market, then an
increase in the competing outside stations will reduce the potential
outside audience that the inside station could capture and therefore
reduce the number of financially viable inside stations (most of which are
licensed to communities in the Arbitron market).18
Thus, we expect that NC and the number of formats will be positively
correlated while NC and NS will be negatively correlated. 	As a result,
the "true" effect of an increase in the number of stations on formats (and
18 	That is, the total number of economically viable full-power
stations in such overlapping markets will be determined by the
characteristics of the two markets (e.g., population and demographics) and
the FCC's spectrum allocation policies for full-power stations. 	In
markets in which the number of full-power stations is less than that
permitted by the FCC's spectrum policies, another full-power station may
not be profitable although an FM translator might be. 	Those markets in
which the FCC constraint on the number of full-power stations is effective
(and thus in which licensees are earning above-competitive returns) would
not admit an additional full-power station (because of the FCC's spectrum
allocation pOlicies) but could financially support one or more FM
translators.
12

listenership) is likely greater than the staff-estimated effect. 19 	Our
analysis, then, understates the format and listenership effect of the
increase in the number of stations resulting from permitting FM
translators greater programming f1exibility.20 	The seriousness of this
underestimate cannot be determined without accurate data on NC, which are
unavailable. 	Nonetheless the staff estimate of the effects of NS on the
number of formats and listenership can be considered a lower bound for the
likely larger "true" effect of NS on these variables. 	Simply put, the
staff estimate is likely to be conservative.
19 	The discussion in the text of the bias from excluding NC would be
strictly true only if our original format regression analysis had included
as independent variables the population and demographics of the outside
markets. 	Because these variables were omitted, our estimated impacts of
NS on formats would contain an additional bias. 	It seems likely, however,
that the effect of these additional omitted outside-market variables on
the number of formats in the sampled market is smaller than that of the
sampled market's own population and demographics. 	Further, the effect of
the sampled market's own population and demographics on the number of
formats in that market is itself small relative to the impact of NS. 	For
example, a doubling of the number of stations in the sampled market (NS)
has nearly twice the impact on the number of formats as a doubling of the
sample market's population. 	Because the regression results for the sample
excluding the outside-market overlaps are statistically identical to the
regression results for all sample markets, the biases from these omitted
variables are likely small. 	In the smaller sample, one would expect the
effect of excluding the outside demographics to be minimal because of the
lack of any obvious market overlaps. 	The lack of difference between the
two regression analyses suggests that the impact of these variables is
small in all markets. 	Thus, any additional bias is likely to be trivial.
20 	The format impact of NS in the regression analysis for the
sample without the outside market overlaps is larger than that of NS in
the regression analysis for the whole sample. 	The regression coefficient
for NS in the attenuated model is 0.251, while the comparable coefficient
for NS in the model with the whole sample is 0.196. 	This suggests that
the bias is negative as our analysis indicates.
13

III. 	The FM Translator Audiences - High or Low Population Areas
The second position by the NAB which warrants further scrutiny is the
claim that most.of the proposals for new FM translators are not in low
population areas where only a few radio stations exist but in high
population areas where the NAB asserts too many radio stations already
operate. 	To illustrate its point, the NAB presents in its Appendix B a
list of the most recent applications for FM translators in addition to the
population and T ALO total number of stations for the counties in which the
applicants plan to locate. 	An examination of this list, however, does not
indicate that translator entrepreneurs are inclined to serve only highly
populated, well-served markets. 	The bulk of the applicants (over 55
percent) plan to locate in counties with less than the average county
population for the United States of 71,465 people. 	The Census Bureau
categorizes counties by population size, and over 64 percent of the
applications listed in NAB Appendix B were for location in counties with
fewer than 100,000 people. 	According to the Census Bureau, only 28.6
percent of the population live in counties with fewer than 100,000
people. 21 	Thus, contrary to the NAB position, most FM translators are
targeted to areas with relatively small populations. 22
21 	For population totals, see U. S. Bureau of Census, Statistical
Abstract of the United States: 1988, (I08th Edition.) Washington D. c.,
1987, p. xvii and p. 275.
22 	Part of the NAB argument is that the areas in the Appendix B
sample are already well served by the present set of radio stations and
therefore do not need FM translators. 	They list the number of stations in
the T ALO list for the counties in each of the areas; the T ALO count
averages 35 stations over the sample in Appendix B. 	However, as
indicated above, it is clear that the T ALO number does not reflect the
number of stations that are available to the average listener. 	We do not
14

Even when FM translators are broadcasting to high population areas,
however, they may be enhancing the benefits of radio service to consumers.
In even the largest population areas, there may be audiences with
specialized programming tastes that are ignored by full power radio
stations. 	By providing service to these audiences, an FM translator can
meet an unfilled need and thereby advance the interests of consumers.
IV. 	Will the Entry of FM Translators Decrease Social Welfare?
The third point in the Supplement that warrants attention is the NAB
assertion that the addition of an FM translator will lower consumer
welfare if it results in a full power radio station going out of business.
The NAB scenario suggests that an FM translator enters the market and
takes away a suffi'ciently large portion of the full power station's
audience that the latter station is no longer profitable. 	When the full
power station leaves the market, its old listeners who either cannot get
the translator's signal or do not like the translator programming, are
worse off than before. 	Thus, while those who listen to the translator
gain (obviously because they prefer the translator over the other
station), the other listeners lose.
There are two flaws with this analysis. 	First, when the full power
radio station leaves the market, the listeners who are either not
have available a more plausible indication of the number of stations heard
by most listeners for all the counties in their sample. 	Since most of
these areas are not included in the Arbitron market area sample, we have
no information from which to make inferences on the number of stations
available to the average listener in those areas. 	Thus, we do not
directly address the Question of the number of radio stations except to
show that many translators are going into low population areas where there
is not likely to be an excess of radio listening options.
15

attracted to or unable to get the new FM translator present an opportunity
for other FM translator entrepreneurs. 	In many parts of the country,
there seems to be enough spectrum for a number of new translators to
enter. 23 	Thus, the former listeners of the now defunct station may find a
desirable alternative to the old station. 	Second, even where the spectrum
is not available, welfare is not necessarily lessened by the demise of the
full power station. 	It is quite possible that the gain to the translator
listeners may be greater than the loss to the other listeners.
In view of the excess of spectrum in most areas, one would expect the
case where welfare is lessened to be unusual. 	Thus in an environment with
imperfect information and mobility of resources, the free unimpeded
operation of the market will more likely achieve greater consumer
satisfaction than an administrative allocation procedure. 	Giving the
owners of FM translators an opportunity to sell their product with minimal
restrictions will go far toward achieving this result by increasing the
probability that more specialized formats will be offered, thereby
increasing radio listenership.
23 	FM translators are allocated the spectrum space between the
points where full-service FM stations operate. 	Not all the FCC spectrum
allotments for FM full service stations are being used; for instance,
there are excess FM station allotments in 95 of the 115 areas in the
sample used in our regression analysis. 	Appendix B of the N AS submission
lists the most recent FM translator applications; if there was no spectrum
space, these applications would not have been made. 	All these facts
indicate that there are wide areas of the country where there is probably
enough spectrum to accommodate more FM translators.
16

\ '
v. 	Conclusion
We do not believe that the NAB critique invalidates the FTC staff's
estimates of the relationship between station numbers and audience size.
The TALO Survey variable suggested by the NAB appears less appropriate
than the Arbitron data relied upon by the original FTC staff comment.
Further, when those sampled markets which may experience significant
competition from stations in other markets are excluded from the
regression analysis, the results are virtually identical to those in the·
original FTC staff analysis. 	This result suggests that any bias in our
results from omitting a variable measuring this outside competition is
trivial. 	To the extent that any bias remains from excluding competing
stations outside the market from the or.iginal study, an analysis of the
effect of this exclusion indicates that the impact of the number of
stations on the format diversity· and therefore on listenership may have
been underestimated. 	The conclusion that the availability of additional
stations would increase the number of formats and indirectly increase the
total radio audience has not been refuted. 	Relaxation of the FCC's FM
translator rules may thus provide benefits to consumers. 	Second,
contrary to the NAB's position, most of the most recent translator
applications were for locations in relatively low population counties.
Finally, even if some full power radio stations exit the market due to the
entry of an FM translator, this result generally would not be expected to
reduce consumer benefits from radio service.
17