FTC Staff Comment Before the FCC In The Matter of Amendment of Section 73.3555 of the Commissions Rules, the Broadcast Multiple Ownership Rules

)
Before the
FEDERAL COMMUNICATIONS COMMISSION
Washington, D.C. 	20554
)
.In the Matter of: 	)
)
Amendment of Section 73.3555 	)
of the Commission's Rules, the 	)
Broadcast Multiple Ownership Rules)
)
MM: Docket No. 87-7
REPL Y COMM:ENTS OF THE BUREAUS OF COMPETITION,
CONSUMER PROTECTION AND ECONOMICS OF
THE FEDERAL TRADE COMMlSSION
July IS, 1987
The 	Federal Trade Commission ("FTC") staff 	is 	pleased 	to 	respond 	to
the· notice of proposed rulemaking ("Notice") in the above-captioned matter. 1
The 	Notice solicits comments on 	proposals 	to relax current 	prohibitions on
(I) the ownership 'of multiple radio stations in the same service (AM or FM)
In 	the 	same 	market, 	and 	(2) 	the 	cross-ownership 	of 	certain 	radio 	and
television stations located in the same market.
In 	this submission, 	we will provide empirical sl.tpport for 	the 	position
that 	easing 	the 	ownership 	rules 	can 	generate 	substantial 	cost 	savings 	and
other benefits. 	These efficiencies may encourage the entry of new broadcast
(stations 	and 	a subsequent expansion in program diversity. 	We also suggest
that 	any 	antitrust 	concerns, 	arising 	as 	a 	result 	of 	the 	consolidation 	of
stations in the same market, are most appropriately appraised on a case-by-
case basis. 	These considerations support adoption of the FCC proposal.
1 These Reply Comments represent the views of the FTC's 	Bureaus of
Competition, Consumer Protection and 	Economics, and 	not necessarily those
of the 	Commission itself or any individual Commissioner. 	The Commission
has, however, voted to authorize us to submit these Reply Comments for the
FCC's 	consideration. 	Inquiries 	regarding 	these 	Reply 	Comments 	should 	be
directed to Keith Anderson, Bureau of E:::onomics (20:-326-3428).

,.
There is one issue in this proceeding that we do not address. 	That is
the question of diversity in editorial viewpoints, and the significance of any
reduction 	in 	that 	diversity 	caused 	by common 	ownership. 	This 	subject 	is
beyond 	our 	institutional 	expertise 	and 	so 	our 	conclusions 	must 	be
appropriately qualified.
Our 	comments 	are 	divided 	into 	four 	main 	sections. 	Section 	I
summarizes 	the 	current 	FCC ownership 	rules and 	the 	changes 	proposed 	in
those 	rules. 	Section 	II 	discusses 	how 	these 	changes 	are 	likely 	to 	foster
technical and 	managerial efficiencies. 	Section 	III 	briefly explores some of
the other 	issues 	involved 	in 	this 	proceeding. 	Section 	IV 	then 	presents our
conc,lusions.
I. THE FCC'S CURRENT AND PROPOSED RULES
The FCC's current "duopoly" rule prohibits any entity from owning two
or more commercial stations in the same service (AM, FM, or televi.sion) that
serve 	the same local area. 2 	Thus, 	no one entity can own 	two 	AM stations
or two FM stations or two television stations in the same broadcast market.
Combinations of different broadcast modes are permitted but are restricted. 3
With 	some 	exceptions, 	a 	single 	entity 	is 	allowed 	to own 	only 	one 	AM-FM
f 	combination, one AM station, one FM station, one televi~ion station, or one
daily 	newspaper 	in 	a 	given 	local 	market. 	In 	general, 	owners 	of 	radio
stations cannot own television stations in the same market or vice versa.·
2
3
See 47 C.F.R. Section 73.3555.
rd.
• 	Upon 	an 	appropriate 	showing, 	the 	FCC 	will 	permit 	AM-UHF
television, FM-UHF television, and AM-FM-UHF television combinations. The
FCC 	does 	not 	permit 	any 	new 	AM- VHF 	television, 	FM- VHF 	television, 	or
2

The recent increase in the number of conventional broadcast (radio and
television) stations 	has 	been substantial, 	rising 	from 	8,127 	such 	stations 	in
1970 to 12,612 in 1986. 5 	The growth in the number of broadcast outlets has
been accompanied by an expansion in the number of non-broadcast methods
of 	electronic 	transmission 	including 	cable 	television 	and 	videocassette
recorders. 6 	In light of this growth, the FCC is now proposing to permit the
common ownership of multiple AM and FM stations in the same market.7 The
FCC is also 	proposing 	to permit, without a special showing, AM-VHF, AM-
UHF, 	and 	AM-FM-UHF 	combinations 	in 	the 	same 	market. 8 	Finally, 	the
Commission 	has 	solicited 	comment 	as 	to 	whether 	FM-television 	st:ltion
combinations should be permitted on the same basis as AM-television station
combinations9 	and 	whether 	the 	ownership 	rules 	should 	vary 	by 	size 	of
market. 10
AM-FM- VHF 	television 	combinations. 	(For 	a 	complete 	discussion 	of 	the
current one-to-a-market rule, see the Notice at paragraphs 8- I I.)
5 	See Notice at note 26.
6 	See Notice at paragraph 15.
t
7 	See Notice at paragraph 26.
8 	See Notice at paragraph 24.
9 	See Notice at note 48.
10 See Notice at paragraphs :4-26.
3

II. THE ECONOMIC EFFICIENCIES FROM COM:MON OWNERSHIP
In 	competitive 	markets, 	a 	move 	from 	a 	state 	of 	highly 	restrictive
regulation 	of 	ownership 	patterns 	to 	a 	less 	restrictive 	regime 	(such 	as 	that
proposed 	by 	the 	FCC) 	can 	generate 	substantial 	consumer 	benefits. 	In
unregulated markets, the profit incentive tends to result in the displacement
of 	less efficient 	ownership 	patterns 	by 	more efficient, and 	therefore 	more
profitable, 	configurations. If common ownership of an FM radio station and
a television station is more efficient than separate ownership, such facilities
will 	tend 	to 	be 	commonly owned 	in 	the 	absence 	of 	regulatory 	restrictions.
Greater efficiency can 	result from more cost effective use of labor. capital,
and 'other 	inputs 	to 	produce 	the 	same 	goods 	and 	services. 	Alternatively,
some ownership patterns may be more conducive to innovation in the kinds
of 	goods 	or 	services 	provided 	to 	consumers. 	As 	compared 	to 	a 	rigid
regulatory 	regime, 	markets 	have an advantage 	in 	generating 	these kinds of
benefits.
The National Association of Broadcasters ("NAB") and other commenters
have 	described 	the 	kinds 	of 	cost 	savings 	that 	may 	flow 	from 	common
ownership of same-market ·stations. 11 These include savings in administrative
or 	overhead 	costs, 	promotional 	costs, 	equipment 	costs, 	and 	programming
costs. 	For example, the engineering staff required for two commonly-owned
broadcast stations may be smaller than the sum of the staffs required if the
stations are sepa.rately owned. 	Based on a comparison between the reported
11 	See Appendix A to Comments of the National Association of Broad-
casters In the Matter of Provision of Improvements and Benefits to the AM
Radio 	Broadcast 	Service, 	August 	1, 	1986. 	See 	in 	this 	proceeding, 	the
Comments 	of 	the 	National 	Association 	of 	Broadcasters 	at 	Appendix 	C
(hereafter 	"NAB"); 	Comments 	of 	CBS, 	Inc. 	at 	9-12 	(hereafter 	"CBS");
Comments of Capital CitieS/ABC, Inc. at Appendix C (hereafter, "ABC).
4

costs 	of 	operating 	AM-FM 	radio 	station 	combinations 	and 	the 	sum 	of 	the
reported costs of AM and FM stations operating on a stand-alone basis, the
NAB suggests that these cost savings can 	be substantial -- savings of up to
35 percent on salary expenses and 	up to 	18 percent on 	technical equipment
costs. 12 	The fact 	that nearly 60 percent of all 	radio stations are accounted
for by AM-FM combinations lends even more credence to the contention that
common ownership results in efficiencies. 13
The 	cost 	savings 	generated 	by 	common 	ownership 	efficiencies 	may
produce significant benefits to consumers. 	Some of the cost savings may be
invested 	to 	produce 	higher 	quality 	programming. 14 	In 	addition, 	the 	cost
sa ,,:ings accompanying common ownership may encourage the construction of
new 	broadcast 	facilities, 	if 	this 	is 	permitted 	by 	the 	FCC's 	spectrum
allocation 	policies. 	As 	we 	discuss 	below, 	this 	new 	entry 	may 	enhance 	the
degree of program diversity in the market.
12 	See Appendix A to Comments of the National Association of Broad-
casters, In the Matter of Provision of Improvements and Benefits to the AM
Radio 	Broadcast Service, August 	I, 	1986. 	If two stations are 	assumed 	to
simulcast the same programming, equipment cost savings are estimated to be
as grea t as 35 percen t.
13 	The 	actual 	percentage 	is 	57.3, 	calculated 	with 	permission 	from
Investing 	in 	Radio. 	1987. The fact 	that a substantial percentage of stations
~ operate as stand-alones, even though the station owners could form 	AM-FM
combinations, 	suggests 	that 	stand-alone 	operation 	is 	more' profitable 	than
combined 	operation 	for 	stations 	with 	some 	characteristics 	even 	though 	the
reverse is true for stations with other characteristics. 	This suggests that if
owners 	are 	free 	to 	combine 	stations, 	they 	will 	tend 	to 	sort 	out 	the
unprofitable co~binations from the profitable combinations.
14 	Some 	commenters 	have 	expressed 	deep 	misgivings 	regarding 	the
FCC's 	expectation 	that 	gre:.:er 	efficiencies 	may 	result 	in 	higher 	Qu:lIity
programming. 	In particular, see the Comments of the Consumer Fcder:ltion
of 	America 	at 	10-12 	(hereafter, 	" 	CFA 	"). 	However. 	for 	evidence 	that
competition 	among 	broadcast 	outlets 	results 	in 	greater 	expenditures 	on
programming, see 	Fournier, " Nonprice 	Competition 	and 	the 	Dissipation of
Rents 	From 	Television 	Regulation," 	Sou/hem 	Economic 	Journal 	(January
1985).
5

However, the evidence provided by the NAB on the magnitude of these
efficiencies is susceptible to some criticism. 	In assessing 	the magnitude of
these 	efficiencies, 	the 	NAB should 	have compared 	the 	costs of an 	AM-FM
combination 	to 	the 	sum 	of 	the 	costs 	of 	those 	same 	stations 	if 	they 	were
operated as stand-alones. 	Instead, the NAB, using broad averages of station
costs across different stations and different markets, compared 	the costs of
AM 	and 	FM 	stations 	that 	are 	jointly 	owned 	to 	the 	sum 	of 	the 	costs 	of
different AM and FM stations that are operated as stand-alones. 1S 	However,
the 	stations 	that 	are 	jointly 	owned 	and 	operated 	may 	possess 	some
combination 	of characteristics (for example, transmitter power and 	antenna
he}ght) 	that 	enable 	them 	to 	attain 	a 	lower 	stand-alone 	cost 	than 	stations
that 	are 	actually 	operated 	as 	stand-alones. 	If 	so, 	then 	the 	NAB's 	cost
savings estimates attributed 	to joint ownership may 	be exaggerated or non-
existent. 16 	Thus, 	the 	question 	of 	whether 	the 	stations 	comprising 	a
.combination would have costs lower than the sum of the stand-alone costs of
those particular stations remains unanswered.
To remedy this deficiency, the FTC staff conducted a statistical study
of 	the 	efficiencies 	associated 	with 	common 	ownership 	of 	AM-FM
combinations in 	the same market. 	We focused 	on 	the 	price 	that 	would 	be
paid 	for 	an 	AM-FM combination and 	the sum of the 	prices that 	would 	be
IS 	See the source cited in note II for a discussion of the NAB methodology.
16 	For exa'mple, two stations that are jointly owned may have costs of
SJOO 	while 	the 	sum 	of 	the 	costs 	of 	two 	stations 	that 	have 	different
characteristics 	and 	operate as 	stand-alones 	may 	be S 150. 	The 	NAB 	would
have concluded that joint operation leads to cost savings of S50. 	However if
the 	two 	jointly 	owned 	stations 	were 	operated 	as 	stand-alones, 	the
characteristics 	of 	those 	stations 	may 	be such 	that 	the 	sum 	of 	their stand-
alone 	costs 	would 	be 	only 	$ 11 O. 	The 	NAB 	in 	this 	example 	would 	h:l ve
overstated the cost savings by $40.
6

paid for the same stations if they were independently owned and operated. If
efficiencies 	result 	from 	combined operation, we would 	expect 	the 	value 	of
the stations to be greater when sold as a combination than 	when sold (and
when 	expected 	to 	remain 	operated) as 	stand-alones. 17 If 	there 	are 	no such
efficiencies, we 	would 	expect the price of the stations as a combination to
be 	the 	same 	or 	lower 	than 	the 	sum 	of 	the 	prices 	of 	the 	stations 	sold 	as
stand-alones.
The 	staff 	analysis 	described 	in 	the 	appendix 	finds 	that 	the 	average
price paid for an existing AM-FM combination exceeds by about 30 percent
the 	price 	that 	would 	be 	paid 	for 	the 	same stations 	if 	those 	stations 	were
opc;r:'J.ted 	(and 	were 	expected 	to 	continue 	to 	operate) 	on 	a 	stand-alone
basis. 18 	This difference is statistically significant. 19 	Our analysis therefore
suggests that the joint ownership of an AM and an FM station in the same
local market has resulted in substantial efficiencies. This docs not mean that
17 	The price of an AM-FM combination might be higher than the sum
of the 	stand-alone 	prices if the 	presence of combina tions 	was accompanied
by market power in the local advertising market. 	As noted below, however,
this 	possibility 	appears 	remote. 	Furthermore, 	one 	of 	the 	factors 	we 	hold
constant in our analysis is the level of concentration of local radio markets.
Therefore, 	to 	the 	extent 	that 	concentration 	can 	be 	regarded 	as 	a 	proxy
(albeit, an imperfect one) for market power, our results are not explained by
. differences in market power.
18 	This estimate is based on a comparison of priceS paid 	for eXIstIng
combinations 	and 	the 	prices 	paid 	for 	stand-alone stations. 	To 	insure 	that
the 	price 	differences 	are 	not 	the 	result 	of 	differences 	other 	than
combination versus stand-alone ownership, we conducted a regression analysis
to control 	for o.ther differences. 	The regression 	analysis 	is similar to that
performed 	in 	Brown, 	"Statistical 	Determinants of 	Radio 	Stations' 	Revenues
and 	Trading 	Prices," 	Working 	Paper 	1':0. 	9, 	Federal 	Communications
Commission, Office of 	Plans and 	Policy (August 	1982) and 	in a 	number of
television studies. 	See, e.g., Fournier, "The Determinants of Economic Rents
in Television Broadcasting," Antitrust Bulletin (Winter 1986).
19 	The 	t-statistic 	for 	the 	difference 	between 	the 	estimated 	ratio 	of
combination to stand-alone prices, 	1.31, and a ratio of 	I, which would exist
if the prices were equal, is 2.42. 	This is significant at a five percent level.
7

all stand-alone stations could achieve these economies if combined with other
stations. 	Some 	stations 	may 	be 	more 	efficiently 	operated 	as 	stand-alones
while other stations may 	be more efficient when 	under common ownership.
Our results do suggest, however, that in a market environment that permits
common 	ownership, 	the 	profit 	incentives 	of 	station 	owners 	will 	generate
combinations of stations that are more efficient than 	if those same stations
were compelled by regulation to operate 3.S stand-alones.
Of course, 	these 	results 	apply 	directly only 	to 	the 	combination of 	an
AM and an FM st3.tion in a local market. 	Since other combinations are not
presently permitted on any widespread basis, it is not possible to empirically
determine 	whether 	there 	are 	significant 	efficiencies 	to 	be 	expected 	from
them. 	The 	efficiencies 	in 	such 	cases 	may 	turn 	out 	to 	be 	different 	from
those 	estimated 	for 	AM-FM 	combinations. 	For 	example, 	if 	a 	second 	AM
station 	is 	acquired 	by 	an 	existing 	owner 	of 	an 	AM-FM 	combination, 	the
"
,additional 	efficiency 	gains 	may 	be 	smaller 	or 	larger 	than 	those' from 	the
combination 	of 	a 	single 	AM 	station 	with 	an 	FM 	station. 	We 	expect,
however, that 	there would be efficiencies from at least some of these other
types 	of 	combinations 	as- well. 	Our 	analysis 	suggests 	that 	when 	station
owners 	are 	free 	to 	combine 	stations, 	station 	owners 	organIze 	ownership
patterns in a way consistent with efficient resource utilizatidn.
III. 	OTHER ISSUES
1. 	Market Power
Concern 	has 	been 	expressed 	by 	some 	commenters 	that 	common
ownership of same-market broadcast facilities would 	le:ld 	to the acquisition
8

of market power in the local advertising market.:w 	We think these concerns
are somewhat unjustified. 	As the FCC indicates in its Notice, and as other
commenters 	have 	discussed, 	there 	has 	been 	a 	substantial 	increase 	in 	the
number 	of 	media 	outlets 	in 	the 	recent 	past. Zl 	This 	should 	substantially
reduce concerns about concentration resulting from common ownership.22
\\!e also disagree with those commenters who argue that the FCC needs
to adopt a different set of rules for common ownership depending upon the
size 	of 	the 	market 	or 	the 	number 	of commercial 	stations 	in 	the 	market.:: 3
These 	kinds 	of' alterna tive 	proposals 	would 	prohibit 	common 	ownership 	in
some 	subset 	of 	markets 	without 	any 	further 	analysis 	of 	the 	particular
economic 	context 	in 	which 	the 	common 	ownership 	may 	arise. 	While 	a
prophylactic rule would provide policy certainty to market participants, it is
likely to prevent some ownership configurations that may promote efficiency
without generating 	market 	power difficulties. 	The 	benefits of such a 	rule
appear 	small 	because, 	at 	worst, 	only 	13.5 	percent 	of 	the 	nation's 	radio
20
21
See the Comments of CFA at 13-14.
See the discussion supra at 3.
ZZ 	The NAB notes in its comments in this proceeding that 47.9 percent
of local broadcasting markets have Herfindahl indexes below 1000, based on a
( market 	definition 	that 	includes 	only 	radio 	stations. 	The 	Department 	of
Justice Merger Guidelines define such markets as being unconcentrated. 	An
additional 38.6 percent of local markets have Herfindahl indexes below 1,800,
i.e., markets that would 	be characterized as "moderately concentrated" under
the 	Merger 	Guidelines. 	(See 	"An 	Updated 	Examination 	of 	Market
Concentration 	in 	Radio 	Markets," 	filed 	as 	Appendix 	E to 	the Comments of
the 	NAB.) 	The 	~AB also 	notes 	that 	there 	are 	a 	variety 	of 	other 	media
serving local 	markets and that in ffi:lny cases the number of such outlets is
considerable. 	(See 	"An 	Analysis 	of 	Media 	Outlets 	by 	Market", 	filed 	:is
Appendix B to the Comments of the NAB.)
::3 	For example, ABC argues that any relaxation of the ownership rules
should be confined to the top fifty radio markets. 	See Comments of ABC at
36-37.
9

markets could 	be categorized 	as 	highly concentrated. 24 	It 	is only 10 these
few 	markets 	that 	the 	prospect 	of 	market 	power 	arising 	from 	common
ownership is likely to be of concern to the FCC.
Precisely because the "problem" markets are likely to be few In number,
however, (about 	35 out of 259 	markets) 	the administrative cost of case-by-
case 	monitoring 	should 	not 	be 	burdensome. 	Furthermore, 	even 	these
concentration calculations are 	likely 	to overstate 	the 	potential 	for 	antitrust
problems. 	The 	calculations 	presume 	that 	the 	services 	provided 	by 	radio
stations constitute a relevant economic market whereas there may in fact be
good substitutes for these services (for example, those provided by broadcast
telev.ision and newspapers). 	The calculations also fail to consider the extent
to 	which 	new 	radio 	station 	entry 	could 	be 	authorized 	by 	the 	FCC 	and
therefore 	they 	ignore 	the constraining 	influence 	of 	this 	potential 	entry on
any attempted exercise of market power. 	We therefore believe that a case-
by-case 	approach 	at 	the 	time 	the 	FCC 	must 	approve 	a 	license 	transfer 	is
most likely to be the appropriate, welfare-maximizing policy.
2. Relaxation of Rules Governing FM Stations
The 	FTC staff also disagrees 	with some commen ters 	who suggest 	that
the FCC should 	limit its proposed relaxation of the radio-television station~
ownership 	rules 	to 	AM 	and 	AM-FM 	combinations. 25 	We 	see 	no 	reason 	to
believe that the economies resulting from common ownership of two or more
broadcast stations will exist only when an AM or an AM-FM combination is
24
25
See note 22 supra.
See, for example, Comments of the NAB at 28.
10

one of 	the 	entities 	involved. 	Common ownership of an 	FM station 	and 	a
VHF or 	UHF television station 	is 	likely 	to generate 	efficiencies similar to
those that result from common ownership of an AM station and a VHF or a
UHF television station.
While the FCC has endorsed the proposed change in the radio-television
ownership rules with respect to AM and A~1·F~1 ~ombinations and television
stations, it is apparently not yet willing to do the same with regard to Fyl-
television station 	combinations. 	To 	the extent 	that 	the 	FCC perceives 	that
the 	gains 	from AM-television 	station 	combinations 	outweigh 	any 	losses 	in
viewpoint 	diversity, 	we 	fail 	to 	see 	how 	the 	FCC 	could 	reach 	a 	different
conclusion 	with 	respect 	to 	FM-television 	station 	combinations. 	We 	would
urge the FCC to consider expanding its endorsement to FM-teJevision st:ltion
combinations as welI.26,27
We 	are 	particularly 	concerned 	that 	a 	rule 	excluding 	FM-television
station 	combinations 	may 	artificially 	encourage 	AM-FM-c'elevision
26 	The 	NAB 	argues 	that 	a 	relaxation 	of 	the 	ownership 	regulations
should 	be confined 	only 	to 	AM stations 	in order 	to 	enhance 	the 	faltering
financial 	fortunes of AM radio. See the Comments of the NAB at 23-26. If,
in fact, FM stations are prospering while AM stations are suffering because
consumers 	prefer 	to 	listen 	to 	FM 	stations, 	we 	believe 	the 	FCC 	should
consider changing 	the amount of spectrum allocated 	to 	FM stations, if that
.. is 	feasible. 	Consumer 	welfare 	may 	be 	enhanced 	if 	more 	spectrum 	was
allocated to FM broadcasting and less ~o AM. 	' .
21 	The" FCC does 	not 	propose to permit, nor does 	the 	Notice discuss,
the ownership of radio stations by newspapers in the same market. While the
evidence provided in 	the Appendix to these Reply Comments does not bear
directly 	on 	any' efficiencies 	that 	may 	be attained 	through 	newspaper-radio
cross-ownership 	in 	the 	same 	market, 	the 	evidence 	nonetheless 	is 	consistent
with 	the 	observation 	that 	media 	combinations 	may 	create efficiencies. The
possibility 	that 	common 	ownership 	of 	radio 	and 	newspapers 	in 	the 	same
market 	would 	engender 	market 	power 	can 	be 	policed 	by 	the 	case-by-case
application of conventional antitrust tools. We would therefore urge the FCC
to 	consider 	commencing 	a 	rulemaking 	proceeding 	to 	reassess 	the 	rule
prohibiting newspaper ownership of co-located radio stations.
II

combinations. 	That 	is, 	if 	FM-television 	station 	combinations 	generated
substantial efficiencies, an entrepreneur desiring to capture those efficiencies
would 	have 	to acquire an AM station in addition to the FM and television
stations. 	For 	any 	given 	level 	of 	broadcast 	services, 	the 	total 	cost 	of
operating 	the 	three 	stations combined 	may 	be 	higher 	than 	the 	sum 	of 	the
costs 	for 	operating 	the 	FM-television 	station 	combination 	only, 	plus 	the
costs of the 	AM station 	in 	its most efficient use. If so, the 	FCC will 	have
encouraged 	ownership 	configurations 	that 	are 	less 	efficient 	than 	could 	be
attained with a less restrictive rule.
3. ProQramming Varietv
Some 	commenters 	have 	expressed 	concern 	regarding 	the 	impact 	of
common ownership on program (or format) diversity.28 	It is well known that
advertiser-supported programming may not result in the mix of programming
that would best satisfy consumer preferences.~9 	This is because advertisers
care only about how many people watch a program rather than the ,value (as
measured 	by 	willingness 	to 	pay) 	consumers 	place 	on 	the 	program. 	The
divergence 	between the mix of programs preferred by advertisers and those
preferred 	by 	consumers 	depends 	on 	(among 	other 	things) 	the 	structure 	of
~ consumer 	preferences and the number of stations in 	the 	market. 	However,
28 	See, for example, the Comments of CFA at 	7-13. 	By "format," we
mean 	the 	kinds 	of 	programs 	offered 	by 	a 	radio 	station, 	such 	as 	easy
listening, 	classical 	rock, 	or 	rhythm 	and 	blues. 	We 	are 	not 	addressing 	the
issue of viewpoint diversity.
29 	See Samuelson, "The Pure Theory of Public Expenditure," Rel'jew of
Ecollomics and Statistics (1954).
12

as 	the 	number 	of 	stations 	in 	the 	muket 	increases, 	it 	is 	more 	likely 	that
minority tastes will be satisfied via an increase in program variety.30
The 	FCC 	and 	other 	commenters 	have 	noted 	that 	in 	the 	recent 	past
there has been a considerable increase in 	the number of radio stations and
other media outlets. 3l 	As a consequence of this growth, it seems likely that
any gap between the mix of programs that consumers are willing to pay for
and the mix actually broadcast is far less significant today than at the time
the 	FCC 	adopted 	the 	rules 	in 	the 	1940's. 	Further, 	if 	relaxation 	of 	the
common ownership rules permits the attainment of greater efficiencies, and
if this in turn encourages the construction of additional radio and television
stations, 	it 	could 	enhance 	the 	FCC's 	ability 	to 	satisfy 	the 	twin 	goals 	of
program diversity and competition. 32
IV. CONCL USION
The 	FTC 	staff 	and 	other 	commenters 	have 	provided 	the 	FCC 	with
evidence 	that 	relaxation 	of 	these 	restrictions 	will 	produce 	increased
efficiencies 	in 	the 	provision 	of broadcast services. 	In 	particular, 	the 	FTC
staff 	analysis 	suggests 	that, 	when 	station 	owners 	are 	free 	to 	combine
stations, 	they 	tend 	to 	do 	so 	when 	that 	is 	the 	most 	efficient 	use 	of 	the
stations. 	However, 	where 	there 	appear 	to 	be 	no 	gains Ito "joint 	operation,
stations 	continue 	to 	operate on 	a 	stand-alone 	basis. 	The 	fact 	that station
30 	See Steiner, "Program Patterns and Preferences and the Workability
of Competition in Radio Broadcasting: Quarterly JouTllal of Ecollomics (May
1954).
31 	See note 22 and the discussion at 3, supra.
32 	The discussion here is analogous to that in Network Inquiry Special
Staff, Final Report. Volume I (1980) at 475·505.
13

ownership is organized in a way consistent with efficient resource utilization
is further evidence that the market, rather than governmental rules, can 	be
relied 	upon 	to 	maximize consumer 	welfare. 	Relaxation 	of 	the 	FCC's 	rules
would 	permit 	these 	economies 	to 	be 	achieved, 	and 	such 	an 	action 	could
benefit consumers. 33
We 	do 	not 	believe 	that 	the 	FCC should 	condition 	a 	more 	permissive
policy on the size of the local market or on the number of commercial radio
stations 	in 	that 	market. 	Rather, 	we 	would 	urge 	the 	FCC 	to 	apply
conventional 	antitrust 	tools 	to 	investigate 	any 	apparent 	market 	power
problems on a case-by-case basis. 	Finally, if the 	FCC chooses 	to relax 	the
ownership rules with respect to AM-television station combinations, we would
also 	urge 	the 	FCC 	to consider including' FM·television station 	combinations
within its proposed rule relaxation.
33 	As 	noted 	above, 	we 	have 	not 	considered 	all 	the 	issues 	raised 	in
this 	proceeding. 	In 	particular, 	we 	have 	not 	considered 	claims 	that 	a
relaxation 	in 	the 	common 	ownership 	rules 	will 	result 	in 	an 	unacceptable
decline 	in 	viewpoint diversity. 	Our support for 	the proposed action of the
. FCC must be qualified accordingly.
14

Appendix
Efficiencies in the Joint Ownership of Radio Stations
by
Keith B. Anderson and John R. Woodbury·
In 	this 	appendix, 	we 	report 	on 	an 	empirical 	analysis 	of 	possible
efficiencies from owning more than one radio station within the same local
market. 	That is, we seek to determine whether the costs of providing radio
services 	are 	reduced 	if stations 	are 	jointly 	owned 	and 	operated. 	We 	will
also 	consider 	whether 	joint 	operation 	of 	radio 	stations 	always 	leads 	to
increased 	efficiency 	or 	whether 	in 	some 	cases 	stand-alone 	stations 	are 	as
efficient or more efficient than combinations.
The 	analysis 	focuses 	on 	the 	prices 	paid 	for 	radio stations. 	For some
set of radio stations, 	the expected 	future costs of operating a 	radio station
may 	be 	lower 	if 	a 	station 	is 	part 	of 	a 	joint 	operation. 	If 	so, 	then 	the
operating 	profits 	of 	the 	station 	will 	be 	greater 	than 	if 	the 	station 	were
forced by regulation to be independently owned and operated. 	As a result, a
buyer 	will 	be 	willing 	to 	pay 	more 	for 	such 	a 	station 	that 	is 	part 	of 	a
combination 	than 	would 	be 	the case 	if· the station 	were 	compelled to 	be a
stand-alone. 	Since 	the 	Federal 	Communications 	Commission 	currently
prohibits all common 	ownership 	within 	the same 	local 	market except for a
combination of qne AM and one FM station,l we compare the price paid for
pairs 	of 	AM 	and 	FM stations 	that 	are 	operated 	as 	combinations 	with 	the
prices paid for the same AM and FM stations operated as separate entities.
Our statistical analysis shows 	that 	when 	stations 	are 	in 	fact operated
as AM-FM combinations, the price of the stations as a combination exceeds
the 	price 	that would 	have been 	received if the stations 	had 	been sold and
operated 	independently. 	This 	result 	is 	statistically 	significant 	at 	the 	five
percent level. 	For those stations that are operated as stand-alones, we find
that 	the 	price 	these 	stations 	could 	command 	as 	part 	of 	an 	AM-FM
combination is statistically no different (rom the stand-alone prices.
We 	therefore 	conclude 	that 	there 	are 	in 	some 	cases 	efficiencies
associated 	with 	joint ownership of an 	AM and 	an 	FM stat-ion 	in 	the same
market. 	In 	other 	cases, 	there 	appear 	to 	be 	no 	economies 	and 	no 	dise-
conomies from joint operation. 	While these results do not demonstrate that
there are economies associated with the 	joint ownership of multiple AM or
multiple FM stations in 	the same market or with 	the common ownership of
• We would like to thank Alan Mathios, Paul Pautler. Dan Sherman, and
Douglas Webbink for helpful comments on earlier versions of this paper, Tom
Buono of Broadcast Investment Analysts. Inc .. for permission to use the data
from Investing in Radio (1987), The Arbitron Company for permission to use
their market share data, and 	Delores \1unson and 	David 	Eaton for 	research
assistance.
See 47 CFR 73.3555(a).

co-located 	radio 	and 	television 	stations, 	they 	suggest 	that 	such 	efficiencies
may well be present.
1. Efficiencies and the Prices of Radio Stations
There are several ways in which any efficiencies associated 	with joint
operation of an AM and an FM radio station could manifest themselves. 	If
there are efficiencies from joint operation, the cost of operating both an AM
and an FM station would be less than the sum of the costs of operating an
AM station and the costs of operating an FM station. 	That is, there may be
economies 	of 	scope 	resulting 	from 	the 	joint 	operation 	of 	more 	than 	one
station.::! 	The attainment of such economies could 	result in lower costs and
In higher revenues.
Theoretically, AM-FM combinations could always be more efficient than
stand-alone operations or they could always be less efficient. 	However, we
observe 	that 	large numbers of stations operate as combinations at the same
time 	that 	large 	numbers 	operate 	as 	stand-alones. 3 	This suggests 	that 	it 	is
unlikely that AM-FM combinations are always more efficient or are always
less. efficient 	than 	stand-alone 	operation. 	Rather, 	it 	is 	likely 	that 	if
efficiencies exist, 	they are 	found 	only 	for 	a subset of stations, 	e.g., those
with 	particular 	characteristics 	or 	those. located 	in 	markets 	with 	particular
characteristics. 	However, there are still a number of possibilities. 	Combina-
tions 	could 	be 	more 	efficient 	than 	stand-alones 	in 	some 	cases 	and 	less
efficient in others. 	Combinations could be more efficient in some cases and
equally efficient 'in others; or they could be equally efficient in some C:lses
and less efficient in others. 	Finallv, combinations and stand-alones could be
equally efficient in all C:lses. 	We ~ill attempt to determine which of these
possibilities is consistent with the data.
There 	are 	a 	variety of 	ways 	in 	which 	the 	costs 	of operating a 	radio
station could be reduced as a result of joint ownership. 	For example, there
may be economies in selling advertising. 	A firm operating both an AM and
an 	FM station 	in 	the 	same 	local 	market 	may 	be 	able 	to 	utilize 	the 	same
sales force for both of its stations. 	Since it would not be necessary to have
a separate salesman 	representing 	each s'tation call 	on 	each 	potential adver-
tiser, the sales force required for the combined stations may be smaller than
2 Economies of scope are said (0 exist if the cost of providing a given
level 	of 	output 	of 	two 	or 	more 	products 	is 	greater 	if 	the 	products 	are
supplied 	by separate firms 	than 	if a single firm 	produces both 	products in
common. 	(See ·Baumol, 	Panzer, 	and 	Willig 	{l982), 	p. 	72) 	In 	the 	current
context, 	broadcasting 	on 	multiple 	frequencies 	can 	be 	thought 	of 	as 	the
provision of different products.
3 As 	of 	the 	end 	of 	1986, 	approximately 	60 	percent 	of 	stations 	were
operated as part of AM-P"1 combinations and roughly 40 percent operated as
stand-alones. 	(Derived from data in Invesling in Radio (1987 ).)
.,

the total number that would be required if the stations operated separately.4
Economies may also be present in the provIsIon of services that can be
utilized 	by 	both 	stations. 	For 	example, 	the 	FCC 	notes 	in 	its 	Notice 	that
there 	may 	be 	economies 	in 	the 	provision 	of 	news 	services 	since 	the 	same
news 	gathering 	and 	production 	staff 	could 	produce 	news 	broadcasts 	for
several stations 	under common ownership.5 	Similar economies may exist in
the 	production of other program material 	that can be used on two or more
stations. 	In 	addition, 	common 	ownership, 	particularly 	if 	both 	stations
opera te 	from 	the 	same 	location, 	may 	permit 	more 	efficient 	utilization 	of
personnel and capital equipment. 6
If 	there 	are 	cost 	savings 	resulting 	from 	common 	ownership, 	these
savings will 	be reflected in the price that a potential 	buyer will be willing
to pay for a set of stations. 7 	A buyer will be willing to pay a price equal
to 	the 	discounted 	value 	of 	the 	future 	profit 	stream 	generated 	by 	the
station. 8 	Since 	profits 	in 	any 	period 	of 	time 	are 	simply 	the 	difference
between 	the 	revenues 	received 	from 	the 	sale 	of 	advertising 	and 	the 	costs
incurred in operating the station, a reduction in 	the costs of operation will
increase the price a buyer is willing to pay.
4 Because a VISIt to a customer may take somewhat longer if the sales
person 	is 	representing 	more 	than 	one 	station, 	the 	sales 	force 	may 	be
somewhat 	larger 	than 	that 	which 	would 	be 	employed 	by 	anyone 	of 	the
stations if it were operated separately. 	However, there are scope economies
.in 	the 	sale 	of 	advertising 	provided 	that 	the 	sales 	force 	of 	the 	combined
operation 	is 	less 	than 	the 	sum 	of 	those 	that 	would 	be 	employed 	by 	the
various stations if they operated independently.
5 See 	Federal 	Communications 	Commission, 	Amendment 	of 	Section
73.3555 of the Commission's Rules, the Broadcast Multiple Ownership Rules,
Notice of Proposed Rulemaking (MM Docket No. 87-7), Adopted January 	IS,
1987, paragraph 22.
6 See Appendix A to Comments of the National Association of Broad-
casters, In the Matter of Provision of Improvements and 'Benefits to the AM
Radio Broadcast Service, August 1, 1986.
7 This will 	be true provided there is a limited 	number of stations for
which 	combination 	operation 	is 	more 	efficient. 	The 	FCC's 	allocation 	of
broadcast spectrum of course 	limits 	the 	number 	of 	radio stations that 	may
exist 	in 	any 	locality. 	Further, our 	results suggest 	that 	combination opera-
tion 	may be more efficient than operating stations separately only for some
subset of the stations in operation.
8 The 	prospective purchaser 	will also consider 	the 	price for 	which 	he
can sell 	the station at some 	point 	in the 	future. 	However, 	that 	price will
simply reflect the future profit stream from the time of th:!t sale.
3

A 	buyer 	would 	also 	be 	willing 	to pay a 	higher 	price 	the 	greater 	the
revenues 	he 	expects 	to 	receive. 	The 	efficiencies 	resulting 	from 	joint
operation 	of 	radio 	stations 	could 	lead 	to 	increased 	station 	revenues 	by
increasing the number of listener minutes of advertising it is profitable for
the 	station 	to 	carry.9 	For 	example, 	if 	a 	station 	that 	is 	part 	of 	a 	joint
operation 	is 	able 	to 	attract 	more 	listeners, 	e.g., 	because 	it 	provides 	better
news service, the revenues received for each minute of advertising sold will
increase. 	Similarly, if the marginal cost of selling an additional minute of
advertising is lowered because of increased efficiency in the sale of adverti-
sing, 	the station 	may sell additional advertising time and 	thus earn greater
revenues.
Station profits would also increase if the price of advertising rose with
common ownership. 	That is, profits would rise if common ownership led to
the creation of 	market 	power in 	the local advertising market. 	If increased
advertising rates resulting from the exercise of market power were the cause
of 	increased 	profits 	and 	therefore 	increased 	sales 	prices, 	the 	higher 	sales
prices could 	not 	be 	taken 	as evidence that 	there 	were 	necessarily efficien-
cies 	resulting 	from 	combinations. 	However, 	if 	we 	can 	eliminate 	the 	pos-
sibility 	that 	increases 	in 	sales 	prices 	are 	the 	result 	of 	increased 	market
pow,er, evidence of higher prices being paid for combina tions would provide
ev.'idence of the increased efficiency from join t opera tion.
If, on average, the price paid for existing combinations is greater than
the price that wQuld have been paid for the same stations if they had been
sold 	separately 	and 	forced 	to 	operate 	independently, 	this 	indicates 	the
presence of economies associated with common ownership of stations within
the same market. 	An average combination price lower than the sum· of the
'single station prices would suggest that common ownership is less efficient. lO
Finally, 	if 	the 	two 	prices 	are 	the 	same, 	there 	would 	appear 	to 	be 	no
efficiencies or 	inefficiencies. 	Below we compare the prices of stations that
are 	part 	of 	a 	combination 	to 	the 	prices 	that 	these 	same 	stations 	would
command if operated on a stand-alone basis.
II. The Price of Radio StatioQs: A Regression Analysis
Our approach requires that we have estimates of the price for AM·FM	~
combinations 	if 	those 	stations 	had 	been 	sold 	and 	operated- as 	stand-alones
and of the price of stand-alone stations if those stations had been sold and
operated 	as ·combinations. 	We 	use 	the 	prices 	we 	observe 	for 	stations 	that
were indeed sold and continue to operate as stand-alones or as combinations,
respectively, 	to 	infer 	these 	prices. 	However, in 	order to 	isolate 	the effect
9 This 	will 	result 	in 	higher 	revenues 	for 	the 	station 	even 	though
advertising rates, measured in price per listener minute, are unchanged.
10 One would expect to find such a result only for st::ttions th:lt are in
fact 	operated on 	a 	stand-alone 	basis. 	We 	would 	not 	expect 	to see stations
operated 	as 	combinations 	if 	stations 	were 	more 	efficiently 	operated 	on 	a
stand-alone basis.

of joint operation, it is necessary to estimate the prices of stations that are
alike in all other ways.
The price of a 	radio station 	will 	depend 	on 	many 	factors 	other 	than
just whether or not the station is part of a combination. 	Stations in 	large
markets 	are 	apt 	to 	sell 	for 	more, 	ceteris 	paribus, 	than 	stations 	located 	in
small markets since they will reach more listeners. 	Similarly, station prices
should be higher where future growth is expected to be high. 	Stations with
better 	technical 	characteristics, 	e.g., 	more 	power, 	should 	sell 	for 	more
because 	they 	reach 	more 	listeners, 	though 	this 	advantage 	is 	likely 	to 	be
reduced 	if competing stations are also very powerful. 	In addition, stations
located in 	markets where radio station ownership is more concentrated may
bring 	higher 	prices, if 	the 	high concentration allows 	stations to collusively
raise the price of advertising.
Finally, a 	station 	that 	has a 	low 	market 	share 	prior 	to sale is apt 	to
sell 	for a 	rel:Hively low price, even if it has good technical characteristics.
While the presence of good technical properties should indicate the potential
to 	attract 	a 	large 	3udience 	and 	thereby 	generate 	large 	revenues, 	a 	low
current 	market 	share 	would 	indicate 	that 	the 	station 	is 	not 	currently
real.izing 	its 	potential. 	Further. 	it 	is 	likely 	to 	take 	some 	period 	of 	time
before a new management can achieve that potential. 	Thus, the revenues of
a 'station that is not well managed may remain low for some period of time
after 	a 	transfer 	of 	ownership. 	In 	addition, 	costs, 	particularly 	for 	st3tion
promotion, 	are 	likely 	to 	be 	high 	as 	the 	new 	management 	seeks 	to 	aUr3ct
more 	listeners 	and 	more 	advertising 	dollars. 11 	Thus, 	profits 	will 	rem3in
below 	their 	potential 	for 	a 	period 	after 	the 	station 	is 	sold; 	and 	this 	will
lower the price a buyer will pay.
In order to adjust observed prices for differences other than combina-
tion 	operation, a 	regression analysis 	was 	performed 	in 	which 	station 	prices
were 	regressed 	on 	variables 	posited 	to 	affect sales 	price. 	Separate regres-
sions 	were run 	for 	pairs of stations that were in 	fact sold as combinations
and 	for 	stations 	that 	were 	sold 	on 	a 	stand-alone 	basis. 12 	The 	effects 	of
combination operation wer.e then inferred by comparing the predicted prices
of 	a 	pair 	of 	stations 	with 	a 	certain 	set 	of 	characteristics 	from 	the
11 For example, WLUP-AM in Chicago recently launched a 	two-month,
S I million advertising campaign to introduce a new format. 	("Riding Gain,"
Broadcasting, May 11, 1987.)
12 Because the regressions were 	run 	in a 	log-log 	form 	(see 	the discus-
sion 	on 	page 	II, 	below.), 	statistical 	hypothesis 	testing 	required 	the 	use 	of
the 	price 	of "pseudo-combinations," which 	were 	the 	combination 	of 3n 	AM
station and an 	FM station 	both of which 	were sold on a stand-::l1one basis,
as the dependent variable in the stand-alone regression.
5

combination equation with the 	predicted 	price of the same pair of stations
from the stand-alone equation. 13
The 	price 	regressions 	were 	run 	in 	a 	log-log 	form 	-- 	that 	is 	the
variables used in the regressions are the logarithms of the sales price and of
the 	various 	factors 	that 	affect 	price. 	Such 	a 	form 	seems 	preferable 	to 	a
linear 	one 	in 	which 	price 	is 	regressed 	on 	the 	actual 	value 	of 	the 	various
independent 	variables 	because 	the 	independent 	variables 	in 	our 	equation
should 	have 	the 	implied 	multiplicative 	effect 	on 	the 	price 	paid 	for 	radio
stations. 	For example, the 	benefit of higher power should 	be greater in 	a
large 	market 	than 	in 	a 	small 	one. 	Similarly, 	the 	effect 	of a 	high 	future
growth 	rate 	should 	be 	greater 	in 	large 	markets 	than 	in 	small. 	Using 	a
logarithmic 	form 	of 	the 	regression 	permits 	us 	to obtain 	this 	multi plica tive
effect; a linear regression would no£.14
13 This technique is very similar to one that has been used in estimat-
ing 	the 	extent 	of 	discrimination 	in 	wage 	markets. 	(See 	Blinder 	(1973),
:-"1alkiel and Malkiel (1973), and Oaxaca (1973).)
In the context of discrimination analysis, it has been suggested that the
technique may provide biased estimates of discrimination as a result of the
need to use proxies for productivity and the inability to include all relevant
variables 	in 	the 	regression 	equations.·For 	example, 	Conway 	and 	Roberts
state:
"In 	regression 	studies 	of 	discrimination, 	not 	all 	pertinent 	job
qualifications 	are 	available 	to 	the 	statistician. 	Indeed, 	the 	job
qualifications actually available typically comprise a 	very 	incom-
plete listing of pertinent qualifications for any job. 	Rarely is any
measure of performance included among available qualificatio.ns ....
One 	may 	therefore 	ask 	whether 	there 	are 	other 	legitimate 	job
qualifications, 	not 	captured 	in 	the 	available 	[independent]
variables, 	that 	ought 	to 	be 	used 	in 	statistical 	adjustment."
(Conway and Roberts (1983), p. 78)
While 	we 	have 	included all 	of 	the 	measures 	for 	which 	we 	have data 	that
should 	affect 	the 	price 	of a 	station, 	the 	possibility 	remains 	that 	we 	may
ha ve omitted one or more relevant variables.
14 Beyond suggesting that a linear form is not appropriate, theory does
not 	suggest 	much 	about 	the 	correct 	functional 	form. 	Therefore, 	we 	also
experimented 	with 	other 	functional 	forms 	of 	the 	regression 	equation,
although these experiments were not performed using precisely the approach
described 	in 	this 	appendix. 	For 	example, 	in 	one 	early 	specification 	the
independent variables were multiplied 	by market size. 	We ultimately .:hose
the log-log specification because it W:lS most successful in explaining prices.
(For example, with the specification involving multiplication by market size
described above. our equation was un:lble to expI:lin a signifi:ant :lmount of
the 	variation 	in 	the 	prices of 	AM stand-alone stations -- 	the 	F-statistic for
that equation 	was 	1.43. For an explanation of why we did not use separate
equations 	to 	explain 	the 	prices 	of 	AM stand-alones 	and 	the 	prices of 	FM
(continued ... )
6

A. The Variables Used in the Analysis
The 	following 	independent 	variables are 	included 	in 	the 	price 	regres-
sions: 15 	To control for market size and future growth, we used:
LRET =
GROW = the log of 1985 retail sales in the local market 16; and
the projected rate of growth in local market retail sales over
the period 1985 to 1990.
Larger local retail sales and higher anticipated future growth should lead to
increases 	in 	the 	price 	of 	a 	radio 	station. 	Therefore, 	the 	coefficients 	on
both LRET and GROW should be positiveY
As 	measures 	of 	a 	station's 	technical 	characteristics, 	the 	following
variables were included:
LAMF =
LAMD =
LAMN =
LFMP =
. the log of the frequency of an AM station;
the 	log 	of 	the 	power 	with 	which 	an 	AM station 	broadcasts
during daylight hours;
the log of the power with which an AM station broadcasts at
night;
the 	log of the 	power 	with 	which 	an 	FM station 	broadcasts;
and
H(... con tin ued)
stand-alones, see the discussion below in the text.) 	The specification finally
decided 	upon 	also 	parallels 	that 	of 	other 	studies 	of 	broadcast 	prices. 	See
note 	16 	in 	the 	text. 	Our 	results 	might, 	of 	course, 	have 	been 	.somewhat
-different if we had used a different functional form.
There 	is 	one 	difficulty 	in 	using 	the 	double-log 	specification. - While 	it
seems reasonable to assume, for example, that the 	benefits of higher power
depend 	upon 	the 	size 	of 	the 	market, 	the 	costs 	of 	attaining 	higher 	power
levels 	is 	not 	likely 	to 	be 	dependent 	on 	market 	size. 	To 	account 	for 	this
possibility 	would 	have 	required 	entering 	some 	of 	the 	variables 	(such 	as
power) both linearly and multiplicatively. (Even here, the correct specific:lt-
ion 	of 	the 	equation 	could 	be 	far 	more- complex 	than 	this. 	While 	the 	cost
effect of higher power may be linearly related 	to a station's sales price, it
( 	is 	not 	likely 	to 	be 	linearly 	related 	to 	the 	log of 	the 	price.) 	Because 	there
were 	relatively 	few 	observations 	on 	AM stand-alones 	and 	FM stand-alones
sold 	in 	the .same 	market, 	we 	would 	not 	have 	been 	able 	to 	so 	specify 	the
equations estimated without an unacceptable loss in degrees of freedom.
15 All data-are from Imesting in Radio (1987), unless otherwise noted.
16 Local market definitions are those used in Illvestillg ill Radio.
11 The variable GROW is not entered in a logarithmic form since it is
already expressed as a percentage and we expect that a one percent incre:lse
in 	growth 	should 	have 	the 	same 	effect 	on 	price 	independent 	of 	the 	level
from which the increase occurs.
7

LHAA T = the 	log of the 	height of an 	FM station's antenna above the
surrounding terrain.
Each 	of 	these 	variables 	measures 	an 	aspect 	of 	a 	station's 	power. 	The
greater 	a 	station's 	power, 	the 	more 	potential 	listeners 	a 	station 	is 	able 	to
reach, and hence the higher the station's potential revenues.
Many 	AM stations 	are 	required 	to 	operate 	at 	reduced 	power 	during
nighttime 	hours 	in 	order 	to 	avoid 	interfering 	with 	other 	stations. 	As 	a
result, it is necessary to include two measures of an AM station's power. 	In
addition, a given level of AM daytime power allows a station's signal to be
received over a greater distance if the station is located in the low frequen-
cy end of the AM broadcast band. 	Thus, frequency is another measure of an
AM sta tion's power. 18
FM stations 	broadcast at the same level of power throughout the day.
However, since FM signals can be received only 	within the line of sight of
the transmitter, the height of the antenna may be an important determinant
of 	a 	station's 	potential 	audience. 	We 	therefore 	expect 	the 	coefficient 	on
LAMF to be negative, while LAMD, LAMN, LFMP, and LHAA T should each
ha ve a positive effect on the price of a station.
To control for 	the "quality" of the competition faced 	by a station, we
included as variables the average value of each of our technical characteris-
tics variables for all stations in the market. 	These variables are LAMFMN,
LAMDMN, LAMNMN, LFMPMN, and LHAA TMN.19 	Since an increase in the
power of a statio"n's competitors should reduce the value of that station, we
expect a 	positive sign on the coefficient on 	LAMFMN, and 	negative coeff-
icients on the other variables.
Two 	other 	variables 	were 	also 	included 	to 	reflect 	the 	amount 	of
competition faced by a station:
LHERF = the 	log 	of 	the 	Herfindahl 	index 	of 	concentration 	based on
radio listenership in the local market;20 and
PFM = 	the 	percent' of the stations in a 	market that operate on the
FM band.n
18 See Brown (1982), p. 12. "
19 Each 	variable is the logarithm of the average across all AM or FM
stations located in the 	local market, depending on 	the characteristic. 	Thus,
LAMFMN is the log of the average frequency for AM stations located in the
local market, while LFMPMN is the log of the a verage power of FM stations
in the market.
20 The 	Herfindahl 	index 	is 	equal 	to 	the 	sum 	of 	the 	squares 	of 	the
market shares of firms owning radio st:ltions in a local market.
21 Again, PFM is not included in a logarithmic form as it is already a
percen ta ge.
8

If an 	increase in 	the 	concentration of 	radio station ownership 	leads 	to an
increase 	in 	the 	price 	of 	advertising 	charged 	to 	local 	advertisers, 	the
coefficient on LHERF would 	be expected to 	be 	positive. 	PFM is included
because another AM station may be a closer competitor for an 	AM station
than 	another 	FM station. 	Thus, ceteris paribus, we 	would 	expect 	the price
of 	an 	FM station 	to 	fall 	with 	an 	increase 	in 	PFM and 	would 	expect 	the
price of an 	AM station 	to rise. 	For a 	combination of an 	AM and an 	FM
station, the coefficient on PFM will represent the sum of the effect on the
price of the AM station and the effect on the FM price. 	We do not have a
clear expectation for the sign of this net effect.
Two 	variables 	were 	included 	in 	the 	regression 	model 	to 	proxy 	costs
incurred by station owners:
L WAGE = the log of an index of local clerical wage rates;22 and
RA TE = 	.the prime rate of interest at the time the station was sold. 23
Higher local wage rates reflect higher costs of operating a radio station and
therefore, 	ceteris 	paribus, 	lower 	profits. 	Higher 	interest 	rates 	at 	the 	time
of sale indicates greater expenses in financing the purchase, which may cut
the 'price a 	buyer is willing to pay. 	Thus, we expect the signs on both of
these variables to be negative.
Finally, the market share of the stations being sold was included. 	This
variable is denoted LMRKSH and is equal to the logarithm of the sum of the
market shares of' the 	included 	AM and 	FM stations during 	the 	last spring
rating period prior to the sale. 	Since a low market share indicates reduced
revenues and increased costs during the first years under new ownership, we
expect this variable to have a positive sign. 	.
B. The Data Used
We estimate this equation for combination sales using data on 39 sales
of AM and FM combinations. 	To estimate the price relationship for stations
sold 	and 	operated 	on 	a 'stand-alone 	basis. 	a 	sample 	was 	constructed 	by
f 	22 The 	index 	comes 	from 	the 	Bureau 	of 	Labor 	Statistics. 	"Wage
Differences Among Metropolitan Areas, 1985," Summary 86-5, May 1986, and
"Wage 	Differences Among Selected 	Areas, 	1985," Summary 86-6, May 	1986.
Where 	the 	index 	was 	not 	available 	for 	a 	particular 	market, 	a 	value 	was
extrapolated from the values for other markets in the same state.
23 This 	variable 	is 	taken 	from 	The 	Economic 	Report 	of the 	President
1987, p. 325. For this analysis, we 	used the nominal 	rate of 	interest, which
is appropriate if expectations of future 	inflation are relativeiy constant. To
the extent 	that this is not the case, the interest rate used should 	have been
the 	nominal 	rate 	less 	the 	expe:::ed 	inflation 	rate, 	that 	is, 	the 	real 	interest
rate. 	Time 	did 	not 	permit 	us 	to 	estimate 	the 	expected 	rate 	of 	inflation,
although 	the 	high 	explanatOry 	power 	of 	the 	equations 	suggests 	that 	our
results would not have been tOo different had we used the real interest rate.
9

combining AM stations that continued to be operated separately after their
sale 	with 	FM 	stand-alone 	stations. 	The 	sample 	constructed 	in 	this 	way
contained 34 observations.
Both samples were limited to sales that occurred during the 	19 month
period 	between 	June 	1, 	1985, and 	December 	31, 	1986. 24 	There 	were 	two
reasons for doing this. 	First, using a sample of sales that occurred during a
short time period allows us to avoid adjusting for inflation. 	Second, we can
avoid 	the effects of other regulatory changes which may h:lve affected the
value of radio stations. 	In particular. in May 1985, the Federal Communica-
tions Commission relaxed its rules concerning ownership of multiple stations
located in different markets, permitting one firm to own 12 AM, 12 FM, :md
12 television stations as opposed to the previous limit of seven of e3.ch type
of facility.~s
Our 	samples 	were 	also 	limited 	to 	sales 	in 	markets 	for 	which 	the
stations 	located 	in 	that 	market 	account 	for 	at 	least 	75 	percent 	of 	total
listeners 	in 	the 	area. 	This 	was 	done 	in 	order 	to 	insure 	that 	the 	market
average 	characteristics 	and 	the 	Herfindahl 	index 	of 	concentration 	are
re3.sonably accurate measures of the actual conditions in the market. 	If only
half 	of 	all 	listeners, 	for 	example, 	listen 	to 	local 	stations. 	it 	is 	clear 	that
sultions in that market are likely competing with many other stations located
in other communities in addition to those in the identified market. While our
75 percent criterion appears reasonable, it obviously will still not provide a
completely accurate measure of market concentration.
It 	is 	not 	difficult 	to 	generate a sample of sales of 	AM-FM combina-
tions. 	Between June 	1985 and 	December 1986, ownership of more than :00
pre-existing 	AM-FM 	combinations 	located 	in 	the 	259 	largest 	local 	radio
. markets changed hands. 26 	Approximately 	150 of these transfers 	involved a
single 	pair 	of stations; and 	we 	began 	with 	a 	one-third 	random 	sample of
24 Because our sample of stations is drawn 	solely from 	those stations
sold between June 1985 and December 1986 (rather than all stations, sold or
not), 	we 	do 	not 	have 	a 	random 	sample 	of 	radio 	stations. 	This 	lack 	of
randomness 	can 	introduce 	biases 	into' the 	results 	(see 	Heckman 	(1979)),
although a study similar to the one here suggests that the bias is 	not large
~ (see 	Brown 	(1982». 	The 	sample 	of sold 	AM-FM combinations 	is 	random,
while the sample of paired stand-alone AM and FM sales represents virtually
the universe .. of all stand-alone sales for those markets in which both an AM
station 	and 	an 	FM station 	were 	sold 	as 	stand-alones during 	the 	18 	month
period. 	While 	we 	have 	no 	reason 	to 	believe 	that 	the stand-alone 	sample 	is
not representative, it is obviously 	not a random sample of stand-alone s:!les
because 	there 	were some 	markets 	in 	which 	no stand-alone stations of each
type (AM or FM) were sold.
:::5 See Memorandum Opinion and Order 10 General Docket No. 83-1009,
100 FCC :d (1985).
26 See Inl'esling in Radio ( 1987).
10

these transacticins. 21 	After deleting sales that did not satisfy the 75 percent
market coverage criterion discussed above and deleting three observations for
which sales prices could not be verified 28, we had a final sample of 39 sales.
Developing a sample of stand-alone sales was a bit more difficult. 	In
order to estimate a stand-alone price equation in the logarithmic form and
to 	be 	able 	to 	compare 	the 	prices 	from 	this 	equation 	with 	those 	from 	the
combination 	equation, 	it 	is 	necessary 	to 	estimate 	the 	prices 	for 	pairs 	of
stations that were not sold as combinations. 	One could, of course, estimate
a price equation for 	AM stand-alone stations and another equation 	for 	FM
stand-alones. 	However, 	given 	the 	logarithmic 	form 	in 	which 	we 	are 	esti-
mating 	the 	equation, 	the statistical 	properties of 	the 	sum 	of 	the 	predicted
prices would be unknown; and we would be unable to determine whether any
differences 	in 	predicted 	prices 	was statistically significant. 	By 	estimating
the sum of the prices for a stand-alone AM and a stand-alone FM station in
one equation, we were able to 	perform statistical tests on 	the 	ratio of 	the
price 	of 	a 	combination 	and 	the 	sum 	of 	the 	prices 	of 	the 	stations 	sold
independently.
Since 	we 	have several 	variables 	representing 	market 	characteristics 	in
our .regression equations, we needed both an AM and an 	FM station in the
same 	local 	market. 	Further, in 	order 	to 	have 	a sample 	of stations 	whose
apparent 	best 	use 	was 	as stand-alones,' we 	eliminated 	stations 	that 	became
27 The 	other 	50 	transfers 	were 	parts 	of 	group 	sales 	where 	stations
located 	in 	a 	nu~ber 	of 	markets 	were sold 	in 	a 	single 	transaction. 	As 	a
result 	we 	have only a single sales price for all of the stations involved in
the 	transaction. 	In 	order 	to 	use 	a 	combination 	that 	was 	part 	of 	one 	of
'these group sales, we would have had to estimate a sales price for that pair
of stations alone. 	We chose not to make such estimates and simply deleted
such stations from our sample.
28 Sales prices for influential observations were confirmed in telephone
conversations 	with 	Mr. 	pave 	Schutz 	of 	ComCapital, 	Inc. 	In 	two 	cases
(WMSQ/WCPQ in Havelock, North Carolina, and KOOZ/KQDI in Great Falls,
Montana), observations were deleted because we were informed that the sale
involved a sale of the firm's stock rather than a sale of the firm's assets.
~ Since 	a 	stock 	sale 	can 	involve 	the 	acquisition 	of 	a 	firm's 	liabilities 	in
addition 	to 	its 	radio 	facilities, 	such' transfers 	may 	not 	reflect 	the 	future
discounted value of radio station profits. 	In a third case (KFRE/KFRE in
Fresno, 	California), 	our 	two 	sources 	disagreed 	on 	the 	sales 	price 	by 	S I
million.
We sought· to 	verify sales prices for 	observations 	that appeared 	to 	be
highly 	influential 	in 	determining 	the 	price 	regression. 	We 	considered 	an
observation to 	be 	influential 	if it 	had 	a strong effect on 	predicted values.
The 	statistic 	used 	to 	identify 	influential 	observations 	was 	the 	DFFITS
statistic 	discussed 	by 	Belsley, 	Kuh, 	and 	Welsch 	(1980). 	Since 	we 	did 	not
confirm the prices of all observations, it is possible that there are problems
with 	other 	observations, 	in 	particular 	observations 	that 	lie 	close 	to 	the
fitted 	regression 	equation. 	Consequently, 	it 	is 	possible 	that 	the 	estimated
standard errors are biased downwards.
I I

part 	of 	a 	combination 	after 	they 	were 	sold. 	There 	were 	numerous 	sales
involving 	AM or FM stations which continued to operate on a stand-alone
basis 	after 	they 	were 	sold. 	Between 	June 	1985 	and 	December 	1986, 	148
stand-alone 	AM 	stations 	were 	sold 	that 	continued 	to 	be 	independently
operated 	as 	of 	December 	1986. 	During 	the 	same 	time 	period, 	158 	FM
stations 	were 	sold 	and 	continued 	to 	operate 	on 	a 	stand-alone 	basis. 29
However, in only 34 cases could an AM stand-alone station be combined with
an 	FM stand-alone 	in 	the 	same 	market.3o 	These 34 observations 	make 	up
the data set used to estimate the non-combination price equation.
C. Regression Results
Using 	these 	two 	data 	sets, 	the 	price 	of 	combination 	and 	stand-alone
stations 	was 	estimated 	using 	the 	regression 	equation 	described 	above. 	The
regression 	results 	are 	listed 	in 	Table 	I. 	Table 	2 	provides 	a 	summary
identification 	of 	the 	variables 	included 	in 	the 	regressions. 	In 	terms 	of
overall explanatory power, our price regressions worked very well. 	The R 2'S
are in excess of 0.9 in each case; and the F-statistics for overall explanatory
power are significant at the 0.1 percent level.
1. 	Combination Results
Looking at the individual coefficients in the combination equation, four
of the five 	measures of station power have the correct sign. 	The exception
is 	FM 	antenna 	height 	(LHAA T) 	which 	has 	a 	negative 	sign 	that 	is 	not
significantly different from zero. 	Three of the coefficients -- LAMF, LAMN,
and LFMP -- are significantly different from zero at the one percent level.
29 Investing in Radio (1987). 	An additional 28 AM stations and 49 FM
stations were sold to 	parties who combined them with another station they
already owned or subsequently acquired to form a combination. 	Since our
data 	source 	does 	not 	show 	purchases 	after 	the 	end 	of 	1986, 	it 	is 	possible
that 	some 	stations 	that 	.were 	independently 	operated 	as 	of 	December 	31,
1986. 	have 	become 	combinations 	through 	the 	acquisition 	of 	other stations
since tha t date. 	.
There 	also 	were 	8 	AM 	and 	27 	FM 	stand-alone 	stations 	that 	were
transferred 	as 	part 	of 	a 	sale 	involving 	stations 	in 	more: 	than 	one 	local
market. 	As noted above, these sales could not be used in our analysis.
,-
so In those few markets in which there were multiple sales of both AM
and 	FM stand-alones, observations were created by randomly combining AM
stand-alone 	stations 	and 	FM stand-alone 	stations 	in 	the 	same 	market 	that
had 	been 	sold 	between 	June 	1985 	and 	December 	1986 	that 	continued 	to
operate independently.
As with the combination data set, we checked the prices of influential
observations 	with 	Mr. 	Dave 	Schutz. 	In 	the 	case 	of 	the 	non-combination
sales, only one price -- that for WLEQ an AM station located in Ft. Meyers,
Florida 	-- 	could 	not 	be 	verified. 	Another 	AM 	station 	in 	Ft. 	Meyers--
WWWQ -- was substituted for WLEQ in our data set.
12

Table 1
Regression Results: The Price of Radio Stations 31
Combination Sales 	Stand-alone Sales 	Expected Sign
Constant 	6.347 	28.392 	?
(0.47) 	(1.23 )
Market Size Variables
LRET 	0.298 	0.985 	+
(1.60) 	(3.54) ...
GROW 	26.769 	36.715 	+
(2.76) .. 	(2.77) ..
Station Power Variables
LAMF 	-1.184 	0.354
(-3.53) ••• 	(0.60)
LAMD 	0.0069 	0.138 	+
(0.06) 	(1.39 )
LAMN 	0.249 	0.0382 	+
(3.42) ... 	(0.44)
'LFMP 	0.290 	0.251 	+
(3.61) ... 	(2.93) ..
LHAAT 	-0.087 	0.0018 	+
(-0.86) 	(0.01)
Competitive Conditions Variables
LHERF 	-1.068 	0.981 	+
t 	(-2.63) .. 	(l.40)
LAMFMN 	2.137 	-4.114 	+
(1.47) 	(-1.44 )
LAMDMN 	-0.456 	1.219
(-1.24) 	(3.00) ...
31 The 	dependent 	variable 	in 	the 	regressions 	is 	the 	log 	of 	price.
Figures in parentheses are t-ratios. 	• denotes a t-ratio that is significant at
the 	10 	percent 	level;·· 	significant 	t-ratio 	at 	the 	5 	percent 	level; 	•••
significant at 1 percent.
13

Table 1 (continued)
Combination Sales 	Stand-alone Sales 	Expected Sign
LAMNMN 	0.458 	-0.366
(1.86) • 	(-1.35)
LFMPMN 	-0.330 	-1.310
(-1.73) • 	(-3.90) ...
LHAATMN 	0.060 	-0.612
(0.26) 	(-1.60)
PFM 	3.870 	-4.400 	?
(2.19)** 	(-1.50)
Stadon Cost Variables
LWAGE 	-0.795 	-1.111
(-0.60) 	(-0.85)
RATE 	0.146 	12.194
(0.01) 	(0.52)
Marketshare Variable
LMRKSH 	0.176 	0.346 	+
(2.50) .. 	(3.76) ...
R2 	0.938 	0.926
F • Statistic 	18.631 ••• 	11.794 ...
.- n = 39 	n = 34
14

LPRICE
LRET
GROW
LAMF
LAMD
LAMN
LFMP
LHAAT
LHERF
LAMFMN
LAMDMN
LAMNMN
LFMPMN
t LHAATMN
PFM
LWAGE
RATE
LMRKSH
Table 2
Variables in Price Equations
the log of the stations' combined sales prices
the 	log 	of 	retail 	sales 	in 	the 	local 	market 	in 	which 	the
sta tions are loca ted in 1985
the 	predicted growth 	rate of retail sales 	between 	1985 and
1990
the log of the frequency at which an AM station operates
,the log of the daytime power of an AM station
the log of the nighttime power of an AM station
the log of the power of an FM station
the log of the antenna height for an FM station
the 	log of the 	Herfindahl 	measure of listener concentration
in, the local market
the log of the average value of AM frequency for stations in
the local market
the log of average daytime power of AM stations in' the local
market
the 	log of average 	nighttime power 	for 	AM stations in 	the
local mark~t
the log of average power'for FM stations in the local market
the log of the average antenna height fon FM stations in the
local market
the percent of local stations that are FM stations
the log of the wage index for the local market
the prime interest rate at the time the stations were sold
the log of the market share of the stations
15

The market share variable (LMRKSH) has a positive and significant coeffi-
cient as expected. 	The measures of market size -- retail sales (LRET) and
projected 	future 	growth 	(GROW) 	both 	have 	positive 	signs 	as 	expected,
though only GROW is statistically significant.
The variables 	representing 	the competition 	faced 	by a 	station 	do 	not
perform particularly well. 	The Herfindahl 	measure of concentration has a
surprising negative and significant coefficient. 32 	Of the variables measuring
average power of stations in the market, only the average level of FM power
(LFMPMN) is statistically significant with the correct sign. 	The coefficients
on LAMFMN and LAMDMN have the predicted signs, but are not statistically
significant. 	Finally, neither the interest rate variable (RATE) nor the wage
variable is significantly different from zero.
2. 	Stand-alone Results
In 	the 	stand-alone 	price 	equation, 	both 	measures 	of 	market 	size--
LRET and GROW -. have positive coefficients that are significant at the one
and 	five 	percent 	levels 	respectively. 	The 	measures 	of 	individual 	station
power do not work as well as in the combination equation. 	While all except
LAMF 	have 	the 	correct 	sign, 	only 	LFMP 	is 	significant. 	As 	with 	the
combination equation, market share has a significant positive coefficient.
The Herfindahl index has a positive, though insignificant, sign in this
equation. 	Three 	of 	the 	market 	average 	power 	variables 	-- 	LAMNMN,
LFMPMN, 	and 	LHAATMN 	-- 	have 	the 	expected 	negative signs. 	However,
only the coefficient on LFMPMN is significantly different from zero. 	While
the coefficient of LAMDMN 	is also significant, it is 	unexpectedly positive.
Finally, 	the 	coefficients 	on 	LWAGE 	and 	RATE 	are 	not 	statisti·cally 	sig-
'nificant. 	.
At 	least 	in 	part, 	the 	lack 	of 	statistical 	significance 	of 	some 	of 	the
individual 	coefficients 	may 	be 	the 	result 	of a 	high 	degree 	of cOllinearity
among variables in both our combination and stand-alone data sets. 33 	It is
32 In 	part 	this 	may 	be 	the 	result· of 	the 	high 	de£ree 	of 	cOllinearity
between 	market size and 	market 	concentration. 	Given 	current 	restrictions~
on 	the common ownership of more than one AM and 	one .FM station in a
market, concentration will be high only in small markets where the number
of 	stations 	is 	small. 	As 	a 	result, 	the 	negative 	sign 	on 	LHERF 	may 	be
partially proxying the higher sales price effect of larger markets.
It 	is 	also 	possible 	that 	our 	definition 	of 	the 	relevant 	market, 	i.e.,
including only radio stations, is too narrow; and this may contribute to the
unexpected 	behavior 	of the 	coefficient. 	We 	note 	that 	similar 	unexpected
signs on concentration indexes were found 	by Fournier and Martin in their
study of television advertising rates. 	(See 	Fournier and Martin (1983). 	As
we have, Fournier and Martin used a media specific market definition.)
33 Belsley, 	Kuh, 	and 	Welsch 	(1980) 	suggest 	a 	statistic 	they 	call 	a
condition 	index 	as 	a 	measure 	of 	the 	degree 	of cOllinearity 	in 	a 	data 	set.
(con tin ued ... )
16

well known 	that high collinearity can reduce the significance of particular
coefficients. 	However, collinearity problems should not pose a problem for
predictions based on all of the coefficients in a m·ode1. 34 	Since comparison
of 	the 	predicted 	prices 	from 	our 	two 	equations 	is 	the 	main 	goal 	of 	our
stud y, we proceed to a discussion of those results.
III. A Comparison of Combination and Stand-alone Prices
Using the regression results discussed above, we can 	predict the log of
the 	price 	that 	would 	have 	been 	received 	if 	the 	stations 	that 	make 	up 	an
AM-FM 	combination 	had 	been 	sold 	and 	operated 	as 	stand-alone 	stations.
This value together with the predicted log of the price from the combination
equation provides an estimate of the log of the ratio of the combination and
non-combination 	prices 	for 	these 	stations. 35 	We 	can 	also 	estimate 	the
variance of this statistic. 36 	The average of this ratio for 	the 39 combinati-
on 	observations 	in 	our 	data 	set 	provides 	an 	estimate 	of 	the 	effect 	of
combination operation on sales price. 37 	The average ratio for the 34 stand-
33( ... continued)
The¥ 	further suggest 	that 	potentially significant collinearity 	problems exist
a~"y time a condition index has a value greater than 30 (p. 157). 	In our data
set of combination sales, 10 of 18 condition indexes have values in excess of
30. 	The 	largest 	is 	900. 	With 	the 	stand-alone 	data, 	11 	of 	18 	condition
indexes exceed 30, and one has a value of almost 1600.
34 See Theil'(l97l), pp. 147-154.
35 For each pair of stations, the values of the independent variables in
'the regression analysis were used in the stand-alone equation to estimate the
expected 	value 	of 	the 	log of 	the 	sum of 	the 	prices 	that 	would 	have 	been
paid for these stations if they had been sold and operated as separate stand-
alone stations. 	The predicted value of the log of the price of the stations
as a combination was obtained by using the same values in the combination
equation. 	Extrapolation .beyond 	the 	observed 	range 	of 	the 	independent
variables 	in 	the 	either 	equation 	was 	not 	a 	serious 	problem 	because 	the
ranges of the variables in the combination equation substantially overlapped
with the ranges of the variables in the stand-alone equation.
We 	use 	the 	predicted 	(log 	of 	the) 	combination 	price 	rather 	than 	(the
log 	of) 	the 	observed 	price 	in 	order 	to 	a void 	ascribing 	the 	unexplained
portion of the. price to the fact that the· stations are indeed part of a combination.
36 We assume that the two estimates are uncorrelated and therefore the
variance of the -difference in the logs is merely the sum of the variances of
the two mea:: predictions ..-
37 Because 	we 	take 	the 	average 	of 	the 	differences 	In 	the 	log 	of
predicted 	prices, we have a geometric mean of the ratio of the prices. 	We
believe that a geometric mean is the correct statistic to use in this case for
two reasons. 	First, given the form of our regression equations, we are only
able 	to 	perform 	a 	test 	for 	statistical 	significance 	if 	a 	geometric 	mean 	is
(continued ... )
17

alone or non-combination observations in our sample provides a 	measure of
the 	difference 	between 	stand-alone 	and 	combination 	sale 	prices 	for 	the
sample of stations that are, in fact, operated as stand-a Iones.
We estimate that, on average, the prices paid for AM-FM combinations
were 	31.7 	percent 	greater 	than 	they 	would 	have 	been 	if 	the 	stations 	had
been sold separately. 	A 95 percent confidence interval on this estimate runs
between 3.6 percent and 65.4 percent. 38
Thus, 	we 	have 	evidence 	that 	operating 	as 	an 	AM-FM 	combination
results 	in 	lower 	operating 	costs 	and/or 	larger 	revenues 	for 	those 	stations
that 	are 	combinations. 	Lower 	operating 	costs 	are 	clearly 	evidence 	of
efficiencies from combination operation. 	Further, increased revenues are, as
we 	noted 	earlier, 	also 	evidence 	of 	efficiencies 	unless 	they 	result 	from
increased 	market 	power. 	Since 	we 	have 	controlled 	for 	the 	level 	of 	con-
centration in the market in our regression equations, our results suggest that
any price premium paid for stations as part of combinations as compared to
the stand-alone value of those stations is the result of increased efficiencies,
not 	the profits that 	may flow 	from increased 	market power. 	As a result,
our analysis 	provides evidence that ownership of a combination 	of an AM
and· an FM station in the same local market can generate increased efficien-
cies.
The average price ratio for stations that continued to operate as stand-
alones is 0.955. 	That is, the predicted value of the sum of the prices of an
AM and 	an 	FM ,station 	as 	stand-alones exceeds 	the 	predicted 	price 	if 	the
stations had been sold as a combination. 	This result 	would be expected 	if
stations 	that 	continue 	to 	operate 	as 	stand-alones 	are 	more 	efficient 	when
31( .•• continued)
used. 	Secondly, a geometric mean appears to be a better measure of central
tendency for a ratio that cannot take on negative values. 	(See Croxton and
Cowden (I955), 198-203.)
We 	note 	that 	this 	technique 	assumes 	that 	the 	expected 	value 	of 	the
ratio of prices is the same for all observations. 	That is, it assumes that the
value of the ratio is not systematically larger or smaller where, for example,
the stations are located in larger markets. 	We see no reason to believe that	~
this assumption is incorrect. 	While it is possible to test this assumption, we
have 	not attempted 	to 	do so because of the 	limited degrees of freedom 	in
our regressions. 	(While it would have been possible to increase the number
of observations in our combination data set, it was not possible to increase
the size of the data set used 	to estimate the stand-alone equation since we
had already inclUded all possible pairs of stations.)
38 This is an approximate five percent confidence interval obtained by
exponentiating 	the 	upper and 	lower 	values 	of 	the 	five 	percent 	confidence
interval on the average log of the price ratio.
A t-test on 	the ratio of the prices confirms 	that 	the mean estimate is
significantly different from one. 	The t-statistic for 	the difference between
the 	estimated 	value 	of 	1.31 	and 	a 	ratio of 	1 is 	2.42, 	which 	is statistically
significant at a five percent level.
18

used in that ~ay. 	However, the difference between this ratio and one is not
statistically 	significant. 39 	Thus, 	we 	cannot 	reject 	the 	hypothesis 	tha t
stations that operate as stand-alones are, on average, neither more efficient
nor less efficient than they would be if operated as part of a combination.4o
IV. Conclusion
This appendix has described an empirical analysis of possible efficien-
cies in group ownership of radio stations. 	If such efficiencies are present,
we 	expect 	them 	to 	be 	reflected 	in 	the sales prices of stations. 	Examining
the prices paid for combinations of an AM and an FM station located in the
same 	market 	-- 	the 	only 	type 	of 	co-located 	common 	ownership 	currently
permitted 	by 	the 	Federal 	Communications 	Commission, 	we 	find 	that. 	on
average, the prices paid exceed the prices that would have been paid if the
stations 	had 	been 	sold 	separately 	and 	operated 	as 	stand-alones 	by 	ap-
proximately 	30 	percent.41 	This 	difference 	is 	statistically 	significant,
suggesting 	that 	there 	are 	economies 	associated 	with 	joint 	ownership 	of
combinations in those cases where joint ownership is observed.
It 	is 	interesting 	to 	note 	that 	our 	results 	suggest 	that, 	where 	station
owners are 	free to combine stations together, they tend to do so when that
is.· the 	most 	efficient use of 	the stations. 	However, where 	there 	appear to
be 	no 	gains 	to 	combination 	operation,· stations 	continue 	to 	operate 	on 	a
stand-alone 	basis. 	That 	station 	owners 	organize 	station 	ownership 	in 	the
most efficient way is further evidence that the market rather than govern-
mental rules can be relied on to maximize consumer welfare.
39 The t-ratio for the significance of the difference in the logs is
-0.60. 	The approximate 95 percent confidence interval for 	the ratio of the
prices is from 0.802 to 1.130.
40 It 	is, 	of 	course, 	possible 	that 	there 	are 	substantial 	economies
resulting from stand-alone operation for some subset of the stations operat-	, 	,
ing as 	stand-alones 	but 	that 	these 	effects are 	being 	lost 	in 	our estimation
because 	they 	are 	combined 	with 	other 	stations 	for 	which 	there 	is 	no
significant difference between combination and stand-alone operation.
41 This 	estimate 	is, 	of 	course, 	based 	on 	the 	particular 	regressions 	we
ran. 	We 	believe 	that 	the 	regressions 	used are the 	best 	that could be done
with 	the 	data 	we 	had 	available. 	However, as 	we discussed 	above (see, 	in
particular, 	notes 	13 	and 	14), 	it 	is 	possible 	that 	the 	results 	would 	differ
somewhat 	with 	a 	different 	sample, 	with 	different 	variables, 	or 	with 	a
different 	functional 	form. 	In 	addition, 	the 	30 	percent 	figure 	is 	only 	a
statistical estimate. 	As noted previously, the 95 percent confidence interval
on this estimate runs from 3.6 percent to 65.4 percent.
19

Bibliography
Baumol, William J., John C. Panzar, and Robert D. Willig (1982), Contestable
Markets and The Theory of Industry Structure, Harcourt Brace Jovanov-
ich, Inc.
Belsley, David A., Edwin Kuh, and Roy E. Welsch (1980), Regression Diagnos-
tics: Identifying Influential Data and Sources of Col!inearity, John Wiley
& Sons.
Blinder, Alan S. (I973),"Wage Discrimination: 	Reducte Form and Structural
Estimates," Journal of Human Resources, 8 (Fall 1973), pp. 436-455.
Brown, 	James 	A., 	Jr. 	(1982), 	"Statistical 	Determinants 	of 	Radio 	Stations'
Revenues 	and 	Trading 	Prices," 	Working 	Paper 	No.9, 	Federal 	Com-
munications Commission, Office of Plans and Policy, August 1982.
Conway, 	Delores 	A., 	and 	Harry 	V. 	Roberts 	(1983), 	"Reverse 	Regression,
Fairness, 	and 	Employment 	Discrimination," 	Journal 	of 	Business 	&
. Economic Statistics, 1 (January 1983), pp. 75-85.
Croxton, 	Frederick 	E., 	and 	Dudley 	J.. Cowden 	(1955), 	Applied 	General
Statistics, 2nd Edition, Prentice-Hall, Inc.
Fournier, Gary M, and Donald L. Martin (1983), "Does Government-Restricted
Entry 	Produ'ce 	Market 	Power?: 	New 	Evidence 	From 	the 	Market 	for
Television Advertising, The Bell Jounral of Economics, 	14 (Spring 1983),
pp. 44-56.
Heckman, James J. (1979), "Sample Selection Bias as a Specification Error,"
Econometrica, 47 (January 1979), pp. 153-161.
Malkiel, Burton G., and judith A. Malkiel (1973), "Male-Female Pay Differen-
tials 	in 	Professional. 	Employment," 	American 	Economic 	Review, 	63
(September 1973), pp. 693-705.
. Oaxaca, 	Ronald 	(1973), "Male - Female Wage Differentials 	in 	Urban Labor
t 	Markets," 	International 	Economic 	Review, 	14 	(October 	1973), 	pp. 	693-
709.
Theil, Henri (1971), Principles of Econometrics, John Wiley & Sons.2
20