Aw

din. :

. .. . ,
. 4 in... 3, .1 .
{www.nwfln.‘ .5?
a .. .
$4,??? .

_ . ‘ .r x

3w .

... .Sfiqrmmflrw ...

have,» sfi.&fi..§.f V

. ......,.a..,n.s ‘ Aflwar: ..

. .33....9 .nfllri ‘ ’ .
flmndmmumf. vflfiunm....

r

3
l

101.!

87")!
‘

.1
l
.1»

o;
to. :a x}
1.3:... u.

{hurt-F: I!) y:

.1. £925.: a; J... l.” a.

. dunk»? £1§$fl§t1~eéyfi§fl
511...?! . ¢ ,t-muflrgont‘tlihpll
.3. HR“). . .. are. t...

. .gt -311

ll ‘Otlln.

3x 4!.
.13. «v:
. gig}...

 

 

‘ . '24 irrith
l. .1-Iz.l. i .Y.I;.i.i..)
>......:.I~I? ‘ eiui . ; ‘

.fl figswvfiwhwnxfimm.

. 5:1 ... 5...: «.

 

(3&0 7

This is to certify that the
dissertation entitled

SEARCHING FOR IMPLICIT MARKET PRICES

FOR KOREAN BROADCAT ADVERTISING TIME:

3]

o
‘3 A HEDONIC APPROACH
>‘ Pa" ..
8
gig presented by
H
a :3 s
2
I MYENG JA YANG

has been accepted towards fulfillment
of the requirements for the

Ph. D. degree in Mass Media

fig” 4/”

f Major Professor’s Signature
07/; 2/47

Date

 

MSU is an affinnat‘ive-action, equal-opportunity employer

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/DateDue.indd-p.1

 

SEARCHING FOR IMPLICIT MARKET PRICES
FOR KOREAN BROADCAT ADVERTISING TIME:
A HEDONIC APPROACH
By

Myeng Ja Yang

A DISSERTATION

Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Telecommunication, Information Studies and Media

2007

ABSTRACT

SEARCHING FOR IMPLICIT MARKET PRICES
FOR KOREAN BROADCAST ADVERTISING TIME:
A HEDONIC APPROACH
By

Myeng J a Yang

Korea Broadcast Advertising Corporation (KOBACO), the public media
representative established by the Korean government in 1981, monopolizes the sale of
broadcast advertising time for all terrestrial TV and. radio stations in South Korea. In
setting prices for advertising time, the public media representative has been alleged by
industry participants to have behaved as an industry coordinator subsidizing public
interest programs carried by TV and radio broadcasters. Hence, it is suspected that
KOBACO has tied under-priced advertising time in regular programs with over-priced
advertising time favored by regulators (Jung, 2005; Kim, 2006). However, discussions on
issues of price distortion and subsidy in the Korean broadcast advertising market have not
been supported by well-designed empirical studies.

Using actual transaction data, this study estimates implicit market prices for
Korean broadcast advertising time. Those prices were estimated using the hedonic
regression approach where price was assumed to be determined by the characteristics of
advertising time including audience Size, demographic compositions, and parts of day
when programs were broadcast (i.e., daypart). This dissertation’s empirical challenge is to
find and construct market valuations of advertising spots in the Korean broadcast
advertising market where published rate cards might not reflect market values. Drawing

upon the bundling literature, the proposed model assumed that package prices reflect

market values for advertising Spots although individual spot prices might not. However,
actual package data were not available for the study. Alternatively, package prices were
constructed in terms Of advertising budgets spent by an advertiser in a network through a
particular advertising agency.

This study found evidence for price distortion and subsidy in the Korean
broadcast advertising market. However, the Size of distortion or subsidy was smaller than
estimations made by previous studies. Implicit market prices for most types of
advertising time in the major TV programs were estimated to be higher than the
KOBACO set prices while implicit market prices for most types of advertising time in
religious broadcasters’ programs were estimated to be lower than KOBACO set prices. In
terms of overall revenue, major networks should earn 1.42 % more, while religious
broadcasters should earn 36.36% less, at implicit market prices, than they earned at
KOBACO prices.

AS the first econometric analysis using actual transaction data, this study
contributes to KOBACO’S exploration to improve its pricing formula. In addition, the
reform of Korean broadcasting policy and efficient decision-making by industry
participants will be assisted by the study. Beneficiaries may include content providers,
broadcasters, advertisers, and other media firms. The study also contributes to the design
of an explicit subsidy system Should policymakers decide such a policy is appropriate.
Particularly, estimated market value deviations from corresponding entries in published
rate cards provide estimates of the size of rate restructuring that will occur when

competition iS introduced.

Copyright by
Myeng J a Yang
2007

This dissertation is dedicated to
my husband, Michael Lee, for his unconditional love and support,
my son, Hyun-seok Lee, for his patience awaiting my return from the library,
and God, for making all this possible.

ACKNOWLEDGMENTS

Not all which exists is visible. Some recognize it but some not. Some develop it
but some not. Therefore, some ask for help and some give help to fill the gap. I could
complete my doctoral program as I was able to meet those who provided to me what I did
not have. In this space, I express my sincere appreciation to them.

First, I thank members Of my dissertation committee for their constant
encouragement and guidance. I am extremely grateful for the invaluable support I
received from my dissertation director and committee chair, Dr. Steve Wildman. He
provided me with the necessary intuition, instruction and innovation to design and draft
my dissertation and he gave those fine elements abundantly, as fathers give to their
children. Dr. Kurt DeMaagd has been a superb counselor to me. Especially, his
suggestion on research models was a major force for me and for my committee members
to have confidence in my embarking on this dissertation journey. Also, his statistical
support was substantially helpful. Dr. Steve Lacy always has been a haven I visited
whenever I had concerns. Thanks to his comments and ideas, I could have, did have, the
courage to take the final model whose merits I had resisted seeing. Dr. Johannes Bauer
instructed me what to see broadly and what to see narrowly. His comments were always
pinpoint—precise. I am indebted to him for much.

Dr. Dennis Gilliland at the Center for Statistical Training and Consulting was
extremely helpful. He reviewed every detail of my dissertation and provided highly
specific solutions to concerns. Serendipitously, both he and I settled independently on the

same model to Solve the multicollinearity problem. His kindness is unforgettable.

vi

Additionally, I thank Jung-Eun Lee for developing programs needed to conduct
data analysis. First, She developed a program counting the number of ad contracts using
data of contract durations and aired dates. Second, she developed a program calculating
weighted averages of the SBS’ regional audience ratings. Without her assistance, the time
required to complete dissertation would have been much longer.

My parents in law, Sang Won Lee and Kyoung Ae Lee, my mother, Bong Choon
Nam, my sisters and brothers, Wha Ja Yang, Yeon Soon Yang, Kyoung Whan Yang, and
In J a Yang and My lovable niece, Hye Kyoung Park deserve for special thanks for their
love and encouragement. My friend, Eunsun Lee, showed me what generosity is for. She
has been there for me whenever I need someone to talk to and need hands. My friends in
Prayers’ Meetings also deserve sincere thanks for their prayers and encouragements. Eun
Sil Lee, Joung Wha Choi, Miran Kim, Mira Yoo, Yoomi Chin, Hyunjoo Ha, Hyunjoo
Choi, Mi Kyoung Kim, Jee Young Bang, Myung Sun Huh, Jeonghee Noh, Jin Young
Choi, and Christina Dokter. For the past four years, they have been my family.

I offer special thanks to Hyung Bae Park of Korea Broadcast Advertising
Corporation (KOBACO) for affording me access to data. This study was possible thanks
to his cooperation, and was initiated and proceeded via discussions with him. Extensive
interactions with members of the KOBACO staff, especially with Yong Ho Lee, have
been helpful in clarifying concepts and understanding the environment they work in.

My doctoral study was funded by Korea Telecom (KT). Also, my research was
funded partially by a dissertation fellowship from the Michigan State University Graduate
School. I am grateful to those institutions for their support.

Most of all, I thank God. He is the Only One who made all this possible.

vii

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... x
LIST OF FIGURES ........................................................................................................ xii
CHAPTER 1 ...................................... ' ............................................................................. 1
INTRODUCTION .......................................................................................................... 1
Market Structure and Pricing Mechanism .............................................................. 4
Rationale for the Study ........................................................................................... 15
CHAPTER 2 ................................................................................................................... 19
LITERATURE REVIEW ............................................................................................... 19
Studies on Prices for Korean Broadcast Advertising Time .................................. 19
Hedonic Price Estimation Studies and Advertising Time Prices ............................ 24
Implicit Market Prices for the Bundled Products ................................................... 30
CHAPTER 3 ................................................................................................................... 32
EMPIRICAL MODEL OF IMPLICIT MARKET PRICES ........................................... 32
Estimation of Implicit Market Prices ...................................................................... 32
Estimation of Price Distortion ................................................................................ 43
Estimation of Subsidies .......................................................................................... 43
CHAPTER 4 ................................................................................................................... 45
METHODOLOGY ......................................................................................................... 45
Data ......................................................................................................................... 45
Measurements ......................................................................................................... 47
Package Price .................................................................................................. 47
Audience Size per Demographic Characteristic ............................................. 49
Daypart ............................................................................................................ 51
Construction of Packages ................................................................................ 52
CHAPTER 5 ................................................................................................................... 53
RESULTS ....................................................................................................................... 53
Model Re-specification ........................................................................................... 55
Estimation of Implicit Market Prices .................................................................... 60
Implicit Market Prices .................................................................................... 66
Revenues estimated with Implicit Market Prices ........................................... 68
Further Analysis ...................................................................................................... 72
Package Size Effect ........................................................................................ 72
Effect of Demographic Composition .............................................................. 73
Selling Major Radio Time in Package with Major TV and Religious
Radio Time ..................................................................................................... 76
Significance of Difference Between KOBACO Prices and Implicit
Market Prices .................................................................................................. 81
Comparison of present findings to those of the other studies ......................... 81

viii

Diagnostic Analysis ................................................................................................ 90

Normalin ........................................................................................................ 90

Linearity and Homoscedasticity ..................................................................... 94
CHAPTER 6 ................................................................................................................... 96
DISCUSSION AND CONCLUSIONS .......................................................................... 96
Implications ............................................................................................................ 101
Implications for Media Policy and Industry .................................................. 101
Implication for the Studies of Advertising Prices ......................................... 104
Limitations .............................................................................................................. 1 05
Conclusions ............................................................................................................. 107
APPENDIX ............................................................................................................. 108
REFERENCES ............................................................................................................... 119

ix

LIST OF TABLES

Table 1. Variables and Data .......................................................................................... 38
Table 2. Descriptive Statistics _ Before Outlier Removed ........................................... 54
Table 3. Descriptive Statistics __ After Outlier Removed ............................................. 56
Table 4. Pearson Correlations of Package Data ............................................................ 57
Table 5. Pearson Correlations of Audience Ratings in the Major TV Networks. ........ 59
Table 6. Regression Estimates for Average Package Price .......................................... 61
Table 7. Regression Estimates for Average Audience Seconds of TV Block Ads 63

Table 8. Hierarchical Regression of Average Audience Seconds of TV Block Ads:
Daypart of TV block ads in the 2" Stage ....................................................... 64

Table 9. Hierarchical Regression of Average Audience Seconds of TV Block Ads:

Other than TV block dyparts in the 2nd Stage ................................................. 65
Table 10.1. Implicit Price Estimation for the Maj or TV Networks. ............................. 67
Table 10.2. Implicit Price Estimation for the Religious Broadcasters .......................... 69
Table 11.1. Revenue Estimation for the Major TV Networks ...................................... 70
Table 11.2. Revenue Estimation for Religious Broadcasters ....................................... 71
Table12.1. Package Size Effect on APPU .................................................................... 74
Table12.2. Package Size Effect on Average Price per Audience Second .................... 74
Table 13. Demographic Composition Effect ................................................................ 75
Table 14. Regression Output Including Major Radio ................................................... 77
Table 15.1. Price Estimation for the Major TV Including Major Radio. ..................... 78
Table 15.2. Price Estimation for the Religious Broadcasters Including Major
Radio. ........................................................................................................ 79
Table 15.3. Price Estimation for the Major TV Including Major Radio. ..................... 80

Table 16.1. Revenue Estimation for the Major TV Networks Including Maj or

Radio .......................................................................................................... 82
Table 16.2. Revenue Estimation for the Religious Broadcasters Including Major
Radio ......................................................................................................... 83
Table 16.3. Revenue Estimation for the Maj or Radio Broadcasters Including
Major Radio ............................................................................................... 85
Table 17.1 Mean Difference Test for Major TV Program Ads .................................... 86
Table 17.2. Mean Difference Test for Major TV Block Ads ....................................... 87
Table 17.3. Mean Difference Test for Religious Radio Program Ads ......................... 88
Table 17.4. Mean Difference Test for Religious Radio Block Ads .............................. 89
Table 18. Comparison of Findings with those of Previous Studies .............................. 91
Table 19. Residual Statistics. ........................................................................................ 92
Table 20. Skewness and Kurtosis . ............................................................................... 93
Table 21.1. Comparison between the Estimators of Implicit Market Prices and
KOBACO Set Prices for Major TV. ......................................................... 99
Table 21.2. Comparison between the Estimators Implicit Market Prices and
KOBACO Set Prices for Minor Radio and Major Radio. ........................ 99
Table 21.3. Estimator Comparison. .............................................................................. 100
Table 22. The Comparison of Price Ranges. ................................................................ 103

xi

LIST OF FIGURES

Figure 1. Transactional Relationships between Organizations ........................................ 8
Figure 2. Types of Advertising Time ............................................................................... 10
Figure 3. Dayparts for TV and Radio Programs .................. 13
Figure 4. Histogram of Regression Standardized Residual ............................................. 92
Figure 5. Normal P-P Plot of Regression Standardized Residual. .................................. 95

xii

CHAPTER 1

INTRODUCTION

In the Korean broadcasting market, the system for selling advertising time is
unique. The Korea Broadcasting Advertising Corporation (hereafter, KOBACO), the
public media representative established by government in 1981, is a monopoly
intermediary coordinating the sales of broadcast advertising time for all terrestrial TV and
radio stations. Officially, the reason for founding KOBACO was to protect public interest
and ensure content diversity, likely to be suboptimal in a competitive ad-supported
market for audience with limited competition. This is consistent with the predictions of
numerous program choice studies (e.g., Beebe, 1977; Owen and Wildman, 1992; Spence
and Owen, 1977; Steiner, 1952). It also has been said by researchers that regulators have
considered a balanced allocation of advertising budgets throughout the media to be an
important policy goal. Restrictions on the pricesiof broadcast advertising time have been
perceived as contributing to financial stability Of other competitive media such as
newspapers (Choi, 2005; Jung, 2005).

In addition to establishing a monopoly in the sale of broadcast advertising time,
the Korea Broadcast Law has regulated quantity as well as format and frequency of

broadcast advertising (Clause 73). Advertising time per program cannot exceed 10%I of

 

' The 10% rule applies to the program advertising broadcast right before or right afier programming. Other
types of ads are also available, including station break ads, caption type ads, and ads associated with time
announcements (see Page 11 for details). In sum, 16.7% of program airtime is available for advertising.

TV program time and it is not permissible to insert commercials into programs (The
Order of Korea Broadcast Law, 2006, Clause 59).2

Under this monopoly system, KOBACO has been at the center of criticism
regarding the issues of packaged sale of advertising time and an inefficient pricing
formula (Park and Lee, 2007). As KOBACO sells multi-entity and multi-program
advertising time concurrently, the organization has been alleged to tie under—priced
regular program spots with over-priced spots in programs favored by the regulator (Choi,
2005; KFTC, 2004; Jung, 2005 ; Yoo, 2005). Summarizing the problems ascribed to the
monopolized broadcast advertising sales market, Yoo (2005), a director of the Korean
Fair Trade Commission (KF TC), identified tying as one of KOBACO’S monopolistic
behaviors likely to distort advertising prices and to cause inefficient allocation of
resources. Proponents of the current monopoly system, such as non-governmental
organizations (N GOS), newspapers, and financially weak broadcasters, have argued that
abolishing the current KOBACO system would increase the prices for ad time on the
major TV networks while causing a financial crisis for minor broadcasters (Choi, 2005;
Jung, 2005). Previous studies (Jeon, 2004; Jung, 2005) claimed that KOBACO’S
packaging practice has been tolerated or, in a sense, facilitated by the regulators to '
promote their policy goals: subsidizing financially weak broadcasters and “public
interest” programming within the major networks (Park and Lee, 2007). Packaging can
be a mechanism for implementing a subsidy as it forces advertisers to purchase ad time
for public interest programs at a higher price than its value while purchasing ad time in

regular programs at a lower price than its value.

 

2 In many countries, TV and radio commercials are delivered within and between programs. Those
broadcast between programs are called “adjacencies.” If this terminology is used, only “adjacencies” are
permitted in Korea.

More clearly, a recently published paper (Park and Lee, 2007) describes the
existence of a packaging system in the Korean broadcast ad time market as a fact. In the
paper (of which the first author is a KOBACO researcher) the authors argue that,
although there is a tendency for the prices not to be determined in the market, demand
and supply are equated in the market through the packaging practice where ad time in
programs whose values are higher than KOBACO set prices is sold in combination with
ad time in programs whose values are lower than KOBACO set prices. So, prices for
packages are market prices but component prices may not be.

Despite all the policy and academic debates, however, there has been no explicit
evidence of such subsidies in the market except a recent paper (Park and Lee, 2007).
Officially, KOBACO has set prices for ad time based on a published formula and each
contract is recorded as if individual spots were sold independently at the price specified
by the formula. What has been conducted in the market is the profit regulation of the
major networks3 and the implicit obligation imposed on KOBACO to sell advertising
time for programs believed to provide an important public service, such as traffic safety,
education, religious or local programming, regardless of technology. As the government
granted TV or radio licenses for the public broadcasting, it required KOBACO to secure
the minimum revenues necessary for public interest programs to be produced
continuously (Park and Lee, 2007). In a Situation where there is no explicit subsidy
system available for the public broadcasting services, an implicit subsidy mechanism

tends to be accepted as reasonable by the industry as well as the public. The implicit

 

3 Profits of the major networks have been regulated in many countries as their advertising prices have
significant impacts on those of other firms within or outside of broadcasting industry (Lee, 2000). For
example, in the U.K., Channel 4 revenues are regulated not to exceed 14% of total advertising revenue, and
50% of excessive revenue is to be remitted to [TV (Min, 1995). In Netherlands, public broadcaster’s
advertising revenues are regulated to maintain balance with other media (Lee, 1995*).

subsidy system is believed to operate by forcing advertisers to buy packages of both
commercial and religious slots whose prices are arguably biased downward for the
commercial programs but upward for the public interest programs (Park, 2005; Kim,
2006). However, KOBACO’S pricing mechanism has never been tested formally. For in-
depth discussion, however, it is necessary to understand the Korean broadcast market

structure and pricing mechanism.

Market Structure and Pricing Mechanism

The Korean terrestrial broadcast market is composed of three national
broadcasters, ten local broadcasters networked with each other for national broadcasting,
and eight independent broadcasters established to support religious and local
programming. Each of the national and networked local broadcasters owns both TV and
radio stations while the independent broadcasters Operate only radio stations,4 with the
exception of iTV from 1997 to 20045. The three national broadcasters including KBS,
MBC, and EBS are non-profit entities owned by govemment-created public institutions.
KBS owns and operates 18 local stations and MBC has 19 local subsidiaries. EBS utilizes
KBS facilities to transmit its signal nationally. Ten networked local broadcasters are all
privately owned and one of them, SBS, is licensed to cover the Seoul region, produces
and provides programs to the other nine local broadcasters while each produces its own
local programming. Despite the gap in the number of local stations, the disadvantage of

SBS in audience reach has been mostly offset by cable carriage of over-the-air

 

" Whereas some public broadcasters have cable TV channels, they are not licensed for OTA TV
broadcasting. Hence, OTA radio stations are included in my analysis.

5 iTV is the only independent broadcaster operating both over-the-air TV and radio stations. It covers
lncheon, the western area of Gyounggi, and the Gangseogu district of Seoul, the capital of Korea. Its
license had been suspended fi'om 2004 to 2007 due to failing to fulfill requirements set by the government.

broadcasters. Cable penetration in Korea is approximately 80%. Regarding the number of
channels, KBS has two channels, KBSl and KBSZ, while the other broadcasters have
only one channel.

To fund programming, KBSl levies an obligatory signal reception fee6 but the
others, including KBSZ, MBC, SBS, EBS, and religious broadcasters, sell advertising
time. For this reason, KBS] is more strictly regulated by the government than are other
major broadcasters. KBSZ programming is similar to that of MBC and SBS. On the other
hand, EBS, established exclusively for the TV delivery of educational content, receives
3% of the obligatory signal reception fee collected by KBS (The Order of Korea
Broadcast Law, 2006, Clause 49). At the same time, EBS is supported by the Korea
Broadcast Development Fund (KBDF) which the Korea Broadcast Commission (KBC)
levies on commercial broadcasters. In 2004, approximately 15% of the fund (KW 17.3 of
116.2 billion) was spent to support EBS (KBC, 2005). In addition, EBS sells commercial
time to advertisers through KOBACO, but its dependency on ad revenues is not as strong
as that of the religious broadcasters due to explicit support from government.

While the costs of KBS] and EBS have been supported explicitly via the
obligatory signal reception fee (The Korea Broadcast Law, 2006, Clause 64) and the
KBDF, it is a general interpretation that the losses of religious broadcasters have been
implicitly subsidized by KOBACO through the packaged sale of their ad time with the
major TV networks’ ad time. This implies that an important policy goal of diversity,

arguably represented by religious programming, has been left in the hands of KOBACO

 

6 Clause 64 of the Korea Broadcast Law (2006) reads: “Anyone who owns a TV set to receive broadcasting
signal should register its ownership of TV set and pay the signal reception fee to KBS.” However, referring
to the clause 67 (2) of the same law, KBS has appointed Korean Electronic Power Company (KEPCO), the
government owned monopoly, to collect the fee on behalf of KBS, which adjusts the fee with the
permission of House of Representatives.

which may have no means to support them other than to package commercial programs in
the major TV networks with public interest programs in the religious broadcasters.
Until recently, however, KOBACO did not distinguish public interest programs

7 or in constructing packages (Park and

from commercial programs in calculating prices
Lee, 2007, p. 338). Instead, KOBACO has set target revenues assumed to be sufficient
for each broadcaster to sustain its operation. Broadcasters typically identified as subsidy
recipients are religious broadcasters including BBS (Buddhism, FM Radio), F BBC
(Christian, AM Radio), CBS (Christian, AM and FM Radios), PBC (Catholic, FM
Radio), and WBC (Traditional Korean Religion, FM Radio). Despite the roles of EBS in
delivering educational TV content to audience, it has not been mentioned in the literature
as subsidized via packaged sale of ad slots. Exclusion of EBS from the list of subsidy
recipient seems to be because EBS has other sources of subsidy such as the obligatory
signal reception fee and the KBDF. EBS’S need for packaged sale of ads might not be as
strong as religious broadcasters. On the other hand, the omission of independent local TV
or radio stations such as iTV and SunnyFM covering Incheon area and KF M covering
Suwon area from the list of presumed subsidy recipient seems to be attributed to their
ownership structure. They were established by private companies with the purpose of
profit maximization by delivering commercial content to the particular areas.

In addition to these, public service obligations have been undertaken by the major
TV networks as well. The Korea Broadcast Law (2006, Clause 50) requires the major TV

networks to maintain the proportion of entertainment programming under 50% of their air

time. The regulated networks are required to fill the rest of their airtime with educational

 

7 KOBACO introduced a public interest program index to its pricing formula in 2005. This gives 5%
premium to the ad time in the designated public interest programs.

and cultural programming. Hence, the public broadcasting services should be defined in a
rather broader term in Korea to include the public service programming of the major TV
networks as well as the religious broadcasters.

Regarding sales Of broadcast ad time, four institutions are involved in ad time
transactions: advertisers, ad agencies, KOBACO, and broadcasters (see figure 1).
Advertisers buy ad time through ad agencies which in turn go to KOBACO to buy the
time. KOBACO represents all the terrestrial TV and radio stations. But KOBACO
organizations are divided into three subunits to represent one Of the three major TV
networks, its radio subsidiary, and its associated minor broadcasters including religious
and local broadcasters. In order for an ad agency to buy ad time in programs of three
different major TV networks, it must buy it separately through each KOBACO subunit.
An advertiser can work with one or more ad agencies to buy ad time.

Two advertising sales markets exist in Korea, defined in terms of contract
duration: upfront8 and scatter. The longer-term contracts (generally, longer than six
months but shorter than a year) are made in the upfront market while the shorter-term
contracts (generally, shorter than six months) are made in the scatter markets. The
upfront market is held twice a year, April and October. Spots sold in the upfront market
begin to air in May and November respectively. Among slots sold in the April upfront
market, those contracted for greater than six months are excluded from sales in the
October upfront market as they are still on the air. KOBACO generally sets a target of

selling 50% of ad time inventory in the upfront market. Ad time not sold in the upfront

 

8 The term of “upfront” is used differently in Korea than in the US. Upfront in USA is associated with sales
that take place substantially in advance of a television season, while scatter market sales can occur any
time. However, the distinction between upfront and scatter in Korea is more related to contract duration
than to timing.

Ems—.0 —. Hnuumwomoaa waanoamrmcm wagon: mewumuuaosm

 

 

>Q<a3maw >9 arm‘s—.3. WOw>OO 32:»

\iii ir/ 2&2. HAS
.9 ”iii.” WEE—E. /.

 

 

 

 

 

 

       

 

:32. H<N

 

 

 

 

2:52. ”a

......... '\ \ ........................ :“bO-l H<w
4 ....... .v 0 Av mic—E:
........ . . fl-

 

 

 

 

 

 

 

 

 

 

 

market is sold in the scatter market. Depending on length of contract, the scatter market is
divided into two sub-markets: one for commercials being aired for three to six months
and one for commercials airing less than three months.

In general, there are two typespof ad slots available in the market: slots for
program ads and for block ads. The ad break has been divided into three parts. The part in
the middle of the ad break is a block ad, and the two separate parts on each side of the ad
break are program ads (see figure 2). As Korean broadcast regulation prevents
broadcasters from inserting ads within a program, all the ads are broadcast between two
programs. However, while the reported audience ratings are those for the adjacent
programs, the values of ad time to advertisers are different depending on their placement
within the ad break. The gap between the size of a program’s audience and the size of the
audience during adjacent ad time is largest in the middle of the break. Hence, KOBACO
has divided the ad break into three parts and charged lower prices for block ads than
program ads.

Program ads are the most abundant and most preferred advertising resource in
Korea. While there is a restriction on total program ad minutes per program (10% of
program duration), there is no regulation on per spot ad length. However, the most
popular are 15-second commercials for TV and 20-second commercials for radio. Within
the time allowed for program ads, half (5% of program minutes) are delivered before the
program (i.e., pre-program ads), and the remainder after the program (i.e., post-program
ads).

Block ads are further divided into three sub-types. Station break (SB) refers to ad

slots placed between pre-program and post-program ads. Seemingly, this type of ad is

Ems—d N. .5694. a. >m<eamm=m .558

 

 

ween—.95 H >9 Wane—n Fem—.3: N

—k
1

M

’
IIIIIIIIIIII III-IIIIII‘

W

         

7:552. ow >=Ew=oo 2:552. 3. >=Eo=nm
E: 13955 — .. m: wan—.3: N

I
. Eco—a >3 _ _
mom? 5.?
Tennis >3 Fem—.3: >3

  

 

 

 

23:51» EU 56ng H5»
Enos—.852: bang—.852:

 

10

 

indistinguishable from program ads. Technically, SB is separated from the post-program
ads of the preceding program by a network ID announcement and from the pre-program
ads of the following program by its title announcement. Five percent of program airtime
is allowed for SB ads. Specifically, they are limited to two broadcasts an hour and four
commercials per SB within 1.5 minutes. Radio can air SB commercials up to four times
an hour. Each radio SB is allowed to have four commercials within 1.2 minutes. SB ad
slots tend to have less value than program ads as audience Size decreases by the distance
from a program, having the lowest point in the middle of the commercial break.

Caption-type ads refer to text or still picture ads delivered during the
announcement of network ID or program titles. This type of advertising is allowed up to
six times an hour only on TV. Each ad should last less than ten seconds. The size of this
type of advertisment cannot be bigger than a quarter of screen size.

Finally, there are commercials associated with time announcements. Broadcasters
announce the time at every hour or every half hour while inserting ads during the
announcement. The Korea Broadcast Act (2006, Clause 59) limits this type of ad to not
more than two an hour. Each can last no longer than ten seconds, ten times a day. In sum,
maximum advertising time including program ads, SB ads, caption type ads, and ads
associated with time announcements, comprises 16.7% of total program airtime.

All individual advertising units are sold at prices determined by a formula
published by KOBACO no matter whether they are sold in the upfront market or in the
scatter market. Park and Lee (2007, pp. 337-338) describe the formula as follows

(footnotes added):

11

Standard price for an advertising spot = required revenues per second of a

network9 x length of an advertising spot10 x index of an expected audience rating

of a programll x index of the program’s environment12

As presented above, the pricing formula for TV during these periods incorporates
several factors such as spot length, expected audience rating, and market environment.
Expected audience rating is a product of genre and past performance of daypart,l3 time of
day divided into four categories reflecting the size of overall television audience
predicted for those times (see figure 3). Scholars accept these factors as important
determinants of ad time prices (Fournier and Martin, 1983; Park, 1996) and they are

applied widely in the industry. Prices set by the formula become the bases from which

rate adjustments are calculated.

 

9 Required revenues per second for each network are calculated by dividing total required revenues for a
network by its total commercial time allowed.

'0 The length of an advertising spot varies depending on demand from advertisers. However, 15 seconds are
most common for TV commercials and 20 seconds for radio commercials.

” Expected audience rating of a program is a product of the expected audience rating of a daypart and an
index of genre of the program. The first term represents the simple average of long-term (past two years)
and short-term (past 3 months) audience ratings of a daypart in each network. The multiplier index for a
genre reflects the relative average of past ratings in a time period for different genres. The genre is assigned
one of seven index levels. Regardless of the network, drama/movie/entertainment receives the highest
multiplier index while composites and unclasifieds receive the lowest. Expected audience rating is then
transformed into the multiplier index.

'2 The index of expected audience rating of a program, in turn, is multiplied by the index of market
environment, which represents the demand/supply ratio (D/S ratio) for advertising spots of a specific
program. The D/S ratio assigned one of 13 index levels within a range of 150% to 40% where 150%
indicates that demand for ad time is 1.5 times of its supply while 40% indicates that demand for ad time is
40% of its supply. While all other factors considered are not program specific, the index of market
environment represents expectations of the performance of a particular program. The D/S ratio is
influenced by the advertisers’ preferences for a particular program. What advertisers consider most is the
actual audience rating of the program. Thus, this factor reflects the gap between audience ratings calculated
by the formula and those anticipated by advertisers at contract time. In this sense, the index provides
feedback from the audience rating of a program to its net profits, and therefore creates incentives for the
networks to invest in program production budgets. In reality, the ratio is subjectively determined by
KOBACO sales personnel using intuitive calls based on years of experience.

'3 A television day is divided into 4 dayparts based on the size of the overall audience. Dayparts are
categorized as SA, A, B, and C in TV depending on advertisers’ preference, where daypart SA is equivalent
to prime time in the US broadcasting system, while daypart C is the time slot in the early morning and late
night and is least desirable. Daypart A is in the early or late fringes of daypart SA while daypart B is
between daypart A and C (KOBACO, 2006). Radio divides dayparts into three: A, B, and C in order of
most valued to least valued.

l2

~3de w. 5368.? 3.. H< 35 ”name wanna—um

 

 

H<

€8.33.

 

31.8 ‘

Sue . 8%

3:5

5:5

. 8

,. 8. N98,.

was: A .N............_..

. A... . c 1%. .._

 

 

O

 

>

 

 

w

 

 

 

>

 

m>

 

>

 

w

 

n

 

 

mans—dew.

 

. 3.25

. , as... as A ,

as _ ..

_. ..wa..‘e..__._

No.8:

  as Fe. ‘

 

 

O

 

>

 

 

 

 

m>

 

>

 

w

 

O

 

 

m==Q~<

 

)

. Sam

33 ‘

ewe

\\\\\

 

 

 

w

 

 

m>

 

>

 

w

 

O

 

 

”an?

iaarawwgaarga

 

as: E...

_ as

..__....s.‘

...,. s...

s

 

 

6 w

 

 

 

 

 

 

 

 

 

 

l3

Examples of such adjustments include preemption”, designation for ad
placement15 , seasonal premiums or discounts”, and discounts for venture firms eligible
for daytime advertising spots at around 30% of formula prices (KOBACO, 2006).

The pricing formula used for TV block ads is similar to that for TV program ads
but uses different indexing factors. The formula for radio ad slots is basically the same
except that the audience rating factor is excluded from the formula because radio
audience ratings are not reported as comprehensively as or as regularly as TV ratings.
While TV audience ratings are reported daily by two well-known audience research
companies (Nielsen and TNS), there is no formal audience research company for radio
programs. KOBACO subscribes to HRC radio audience reports published quarterly. To
fill the gap, research companies such as Lee’s PR and ACR irregularly survey radio
audiences at the request of broadcasters, advertisers, or other interest groups (Park, 2005).
Due to the lack of organizational support and low reliability, radio audience ratings have
been disregarded in KOBACO’S ad pricing formula (Kim, 1999).

It has been known that there is no volume discount in practice, or published on the
rate cards. This fact indicates that calculated spot prices are applied to spots purchased in

the package. Informal interviews with advertisers reveal that package price is simply the

 

‘4 Preemption offers advertisers the right to acquire certain previously sold ad spots if they pay higher
prices than the previous buyers. Preemptible spots are generally announced in advance. All spots in
preemptible programs are sold only on a preemption basis. For example, if 24 spots are available for a 60-
minute program, all 24 spots are available for preemption if any are. Preemption contracts generally last for
a month but are guaranteed up to six months if the offered prices are higher than thrice the formula price
(KOBACO, 2007).

'5 Advertisers can designate the location of ad slots purchased at extra costs. Concerning the placement of
the commercial, a slot is selected from the slots available before or after programs (Note that ad messages
cannot be inserted within a program in Korea.) As audience size tends to decrease closer to the middle of
the commercial break, advertisers generally prefer placement closer to a program.

‘6 The formula prices are adjusted seasonally to take into account the seasonal fluctuations of audience size
and demand for ad time. At least, 10% of the formula prices are discounted for the purchase of ad time
during the off-peak season while premiums of around 10% of are applied during peak season. In 2006, peak
season included September, October, May, and June while off-peak season included July, August, January,
and February.

14

sum of individual spot prices in a package. Many industry observers believe that the
formula itself is biased downward for programs with high commercial value and upwards
for programs with low commercial value. If, as alleged, prices on rate cards were
distorted, it is natural for the previous studies not to find empirical evidence of price

efficiency in the broadcast advertising market as presented in the literature review.

Rationale for the Study

These pricing practices raise two interrelated issues to be investigated. One is that
KOBACO cannot know how well its posted prices compare to implicit market prices and
no one knows whether KOBACO has approximated market prices well. This study tests
if there are distortions built into their pricing formula, more specifically, whether the
factors they employ to determine prices appropriately reflect true market values. The
second issue is whether there are subsidy flows from other programs to programs favored
by regulators. Selling ad time in favored programs in a package with other programs’ ad
time and setting prices for the other programs’ ad units lower than their market values
while selling time in favored programs at above market rates is the mechanism whereby
subsidies could be implemented.

Because ad units are sold in packages, market prices for ad time in different
programs from different providers cannot be observed directly. If package prices are
determined by market forces, however, we can assume market values for the ad units
comprising packages are implicit in the package price. Given the lack of empirical
knowledge of implicit market prices, regulatory bodies must sit as puzzled arbiters of

conflicting claims made by various private and public interest groups. For example, Park

15

and Lee (2007) argue that the current KOBACO pricing system has been improved to
reflect the market values of ad time, while others claim that the distortion is still
significant. Expanding the pricing issues to the policy concerns, some (i.e., Park, 2004;
J eon, 2004) state that policy change from monopoly to competition will cause a
Significant price increase for major networks and financial distress for public
broadcasters while Jung (2005) and others claim that there will be little price impact from
regulatory regime change. Depending on the methods employed, predictions of likely
price increases range from 10% (Jung, 2005) to 400% (Park, 2004) for the major
networks.

In either case, however, previous studies have substantial shortcomings. Studies
using published rate cards (e. g., Lee, 1995; 1997; 2004; Kim and Lee, 2004; Park, 2004)
are flawed because advertising spots are sold in packages, and there is no way to know
whether advertisers value the spots in the packages at their published rates. Even the
study employing a survey method to estimate advertisers’ willingness to pay (Jung, 2005)
contains limitations: (1) surveys may systematically under-estimate market prices as
advertisers report their willingness to pay more conservatively in response to a survey
than their behavior in the market would indicate (Monroe, 1990, p. 107—112); at the same
time, (2) translating survey results into demand and supply curves requires a number of a
priori assumptions such as frequency of advertising purchase by advertisers and the
number of spots supplied to these survey participants. The assumptions tend to render the
results arbitrary and unreliable.

Hence, this study develops a model to estimate implicit market prices for Korean

broadcast advertising time using actual transaction data, which enables estimation of the

16

level of price distortion under the KOBACO system and of the cross-subsidies alleged to
exist due to packaging practices. Implicit market prices are estimated using the hedonic
approach where price is determined by characteristics of component advertising time.
Factors assumed to influence market value include audience size, demographic
composition, and daypart (F oumier and Martin, 1983; Napoli, 2003; Wildman, 2003).

The empirical challenge of this study is to find and construct market valuations of
ad spots in the Korean broadcast market where published rate cards may not reflect
market values. Drawing on bundling theory see c. g. Venkatesh and Kamakura, 2003;
Adams and Yellen, 1976; Schmalensee, 1984; McAfee, McMillan, and Whinston, 1989;
the proposed model assumes that package prices reflect market values for the ad spots
although the individual spot prices may not. This assumption implies that a buyer’s
reservation price for the package is equal to the sum of his or her separate reservation
prices for component ad spots. However, actual package data were not available for the
study. Therefore, composite package prices were constructed from the ad budgets spent
by an advertiser on a network through a particular ad agency. Constructed composite
packages for individual advertisers were comprised of all ad time all packages purchased
by an advertiser through a specific ad agency. Composite package prices were measured
as each advertiser’s purchases from the agencies. This approach reflects the actual
transaction practices where individual packages purchased. by advertisers did not mix ad
slots across agencies or networks.

This dissertation is organized as follows: Chapter I introduces the background and
goals for the study. Chapter II reviews related literature. First, the studies of prices for

Korea broadcast advertising time are reviewed. Second, the theoretical and empirical

17

aspects of the hedonic model are reviewed to develop the model for this study. Third,
three different viewpoints on the valuation of packages are reviewed briefly from the
perspective of the bundling literature. Chapter III describes a hedonic model developed to
estimate implicit market prices for Korean broadcast advertising time. Chapter IV
describes the data and methodologies for measuring individual variables. Chapter V
reports regression results. Chapter VI discusses the dissertation’s contributions and

limitations and its implications for future studies.

18

CHAPTER 2

LITERATURE REVIEW

Studies on Prices for Korean Broadcast Advertising Time

The transaction system for Korean broadcast advertising time has been studied
mostly from the perspective of policy change (Kim, 2004; Kim, 2006; Yang, 2007; Jung,
2007). These studies equate the monopoly transaction system with inefficient ad time
pricing in the market and they assume that the most popular programs are under-priced to
sell protected types of programs at higher prices than their market values. However, these
assumptions have never been empirically tested with sufficient evidence. Nevertheless,
studies undertaken by Lee (1995; 1997; 2004) and Jung (2005) Should be acknowledged
for their contributions to the understanding of the Korean broadcast ad pricing system
and exposition of the level of possible price distortion, which also is a goal of this
dissertation.

Lee has conducted several empirical studies to investigate relationships between
broadcast ad time prices and program performance. In a study published in 1995, he
analyzed 391 television programs aired in the first half of 1995 and found that ad time
prices were not determined by program performance. For example, while ad time prices
for the programs delivered during the same daypart were fixed at KW 4,788,000 in a
particular week of May 1995, audience ratings ranged from 5.8% to 32.5% for these
programs.

In an extended study of the four major media, TV, radio, newspapers and

magazines, Lee (1997) stated that the study confirmed his previous findings that Korean

19

broadcast ad time prices are not well explained by audience ratings. In 2004, the author
replicated his 1995 study to evaluate the reform of KOBACO’S pricing system in 2000
when the “global standard (GS)” ad pricing system was introduced with the goal of
setting prices based on audience ratings, market environment, and required ad revenues
for the networks. The author concluded that audience ratings were still not reasonably
taken into account for ad time pricing. His conclusion was based on the facts that only
55% of price variation was explained by audience ratings and that comparative ratios of
average price to average audience rating were inconsistent across the dayparts and
networks.

However, Lee does not explain why he sees 55%, the variation in rate card price
explained by audience seconds, as too low to interpret KOBACO prices as efficient.
Generally, ad time prices are not fully explained by audience seconds, as uncertainties are
involved with the prediction of audience ratings at the point of contract (F oumier and
Martin, 1983). Only 57% of USA ad time prices, believed to be efficient market prices,
was explained by the number of viewers, the number of viewing males aged 18-49, and
the number of viewing females aged 18-49 exposed to the 30-second spot (Fournier and
Martin, 1983, p.49). More fundamentally, however, studies such as Lee’s (1995; 1997;
2004) using published rate cards are flawed because advertising spots are sold in
packages, and whether advertisers value spots in the packages at their published rates is
indetenninable. His studies have been cited without criticism in subsequent discussions
of KOBACO’S pricing system (e.g., Kim and Lee, 2004).

While attempts are made by Lee (1995; 1997; 2004) to find evidence for

inefficient ad time pricing in the Korean broadcast ad market, others explain causes and

20

effects of inefficient ad time pricing. For example, Kim, et a1. (2003) argued that, due to
packaging practices, KOBACO’S pricing system has caused loss of social surplus in both
types of programs, subsidizing and subsidized. Kim and Lee (2004) also claim that
investment efficiency and program quality cannot be improved until a strong relationship
is established among investment, audience rating, and return on program production.
Studies indicating packaging practices as a reason for price distortion include Choi
(2005), Jeon (2004) and Jung (2005). Especially, Park and Lee (2007), clearly posit the
existence of a packaging transaction system in Korean broadcast ad time market. In the
paper, the authors argue that although there is a tendency for prices not to be determined
in the market, demand and supply are cleared in the market through the packaging
practice where ad time in high valued programs is sold in combination with ad time in
low valued programs.

Accepting KOBACO’S packaging practice as a fact, Jung (2005) attempts to
estimate the implicit market prices for Korean broadcast advertising time. In a survey of
the 100 largest advertising agencies, the author found that current, posted prices deviated
significantly from willingness to pay for the programs showing the equivalent
performance. He found that willingness to pay for programs with a 25% audience rating,
typical for popular primetime dramas assigned to daypart SA, was KW 20-million, which
was 26% higher than MBC’S published price, 34% higher than SBS’S published price,
and 42% higher than KBS’s published price. As for prices for ad time in daypart A,
estimated market prices are KW 8-million for MBC and KW 7-million for SBS and for
KBS. These estimates are higher than the published price by 22%, 25%, and 10%,

respectively. On the other hand, willingness to pay for ad time in dayparts B and C was

21

estimated to be lower than the published prices by 13% to 23% depending on the
networks. On average, ad revenue was estimated to increase by 14.8%. The author also
estimated implicit market prices for religious broadcasting while admitting limitations of
prediction. Employing the same method used in the prediction of TV ad slots, he claimed
that religious broadcasters’ revenues would decrease by at least two thirds.

The findings, if the employed methodology is deemed reasonable, provide
evidence of the existence of price distortions caused by packaging practices. Otherwise, it
is difficult to explain KOBACO’S charging lower prices than those advertisers are willing
to pay for time in popular programs. Also, it would be impossible for KOBACO to sell
time in dyparts B and C as well as in the religious broadcaster’s programs at prices above
what advertisers would be willing to pay if purchasing time in these programs on a stand-
alone basis.

Nonetheless, the survey method using advertisers’ self-stated willingness to pay
(Jung, 2005) has the limitation that advertisers tend to report their willingness to pay
more conservatively in surveys than in the market (Monroe, 1990, p. 107—112). Moreover,
assumptions made for the study contain serious limitations. Jung (2005) administered a
survey to the 100 largest advertisers and drew a demand curve by quoting the willingness
to pay indicated by 82 subjects who returned questionnaires. Then he adjusted the supply
for the ad time in each daypart referring to the average program hour, the number of
respondents, and the probability that the 100 largest advertisers would buy more than one
ad slot and that other advertisers would purchase the particular ad slot. The study
estimated prices for the ad time in daypart SA by assuming that the supply for the ad time

on a one-hour program was 24, the total number of ad slots allowed for a one hour

22

program (24 = 10% of 60 minutes / 15 seconds). However, the study increased the supply
of ad slots to 42 in daypart A, to 72 in daypart B, and to 72 in daypart C, assuming an
increasing probability that the 100 largest advertisers would buy more than one ad slot
and that advertisers would also purchase the particular ad time. But he did not provide
any supporting arguments why these particular assumptions are employed. Small changes
in these assumptions could produce significant changes in the estimated market price.

Hence, it is meaningful to explore implicit market prices using actual transaction
data reflecting advertisers’ purchasing patterns. Consumers, including advertisers,
basically react to the market based on their demands for products and substitutes for them.
In the advertising market, it is reasonable to assume that advertisers would pay no more
for ad time than its value to them as buyers are well-informed by advertising agencies
about prices and substitute products. Even when ad time is sold in packages due to
regulation or for other reasons, there should be indicators that reflect market demand.
Hence, the present study proposes a hedonic model to investigate the following research
questions:

RQl. What are the implicit market prices for advertising Spots?

RQ2. Are there cross-subsidies flowing from some types of programs to other

types of program?

RQ3. If subsidies exist, how large are they?

23

Hedonic Price Estimation Studies and Advertising Time Prices

Significant research has been conducted to estimate ad time prices for different
media. Especially, the hedonic approach has provided an empirical framework for studies
assessing the relationship between ad time prices and the characteristics of media
audiences upon which ad prices are based (Fisher, McGowan, and Evans, 1980; Fournier
and Martin, 1983; Waterman and Yan, 1999; Koschat and Putsis, 2002). In the hedonic
model, measures of audience demographics have been treated in the same way product
attributes are treated in studies of demand for other products and services. This logic is
stated most explicitly, perhaps, by Fournier and Martin (1983) in their article on the price
of station-sold time in local television markets, who characterize their econometric model
as a hedonic model.

The theoretical exploration of the relationship between product characteristics and
the demand for a product was pioneered by Lancaster (1966, 1971). Different from
traditional economic theory of consumer demand, the author assumes that utility or
satisfaction from goods is derived from product characteristics, not from the product
itself (Ratchford, 1975). While it has been acknowledged as a “revolutionary” change of
perspective, Lancaster's assumption of infinite divisibility makes his model difficult to
apply to some goods, especially, to “expensive and infrequently purchased goods”
(Ratchford, 1975, p.71). A rather convenient solution for empirical modeling was
developed by Rosen (1974) following initial modeling by Griliches (1971). Rosen
assumes that “the various goods sold in a market are indivisible, and that a consumer
buys only one brand of the goods per year” (Ratchford, 1975, p.71). Hedonic price

functions suggested by Rosen have been acknowledged as “empirical summaries Of the

24

relationship between the prices and the characteristics of goods sold in differentiated
product markets” (Pakes, 2003, p. 1580).

While a series of papers investigated the theoretical issues of the relationship
between characteristics and prices for the differentiated products (Rosen, 1974, Epple,
1987, Anderson et al., 1992, Berry et al. 1995; F eenstra, 1993; Pakes, 2003), hedonic
analysis achieved its popularity by providing a convenient empirical framework for
industry members and researchers. To design an empirical model based on the hedonic
approach, it is therefore important to define characteristics correctly, although, as
Lancaster himself admits (1971, Ch. 9-10), there may be “severe difficulties in defining
characteristics operationally”.

In the context of ad time pricing, Webster and Lichty (1991) suggest numerous
factors affecting broadcasts’ advertising revenues. Included are audience Size,
demographic composition, and daypart. Gensch and Ranganathan (1974) found
significant mean differences among demographic groups in their preferences for program
types and it is generally believed that advertisers value age group 18-49 more than others
(CAB, 1989, p 22; Barnes, 1990, p. 32). Napoli (2003) also asserts that firms advertising
products for a target group aged 18-49 pay relatively higher prices for spots reaching a
higher proportion of viewers in the category. Fournier & Martin (1983) and Park (1996)
also assumed that the market equilibrium value of audiences exposed to commercial
messages will differ between age groups and found evidence supporting this assumption.

Even among the same age group, however, a study by Koschat and Putsis (2002,
p. 267) found that increasing the fraction of women has a negative effect on the price.

This indicates that advertisers prefer male audience members to female audience

25

members and therefore a relative increase in the female proportion of the audience will
reduce price after controlling for other factors. Fournier and Martin, (1984, p. 49) also
found that the fraction of an audience comprised of women aged 19-49 had smaller effect
on price than the fraction of men aged 19-49. Hence, including demographic
characteristics of audience, such as age and gender, should improve the accuracy of
predictions of ad time prices. However, the effects of audience segments on prices for
broadcast ad time can differ depending on cultural context. Difference in viewing
patterns among demographic groups should be larger in countries where individualism
dominates than in countries where family-oriented norms dominate. If all family
members watch the same programs, then demographic segmentation does not have much
impact on in ad time prices.

In an empirical study, Fisher, McGowan, and Evans (1980) also found that
daypart has an effect on audience revenues in the broadcasting programs. This seems to
be a reflection of reality (Waterman and Yan, 2001; Lee, 2004). In the context of the
Korean broadcast market, KOBACO defines daypart as “a central factor determining
prices for the broadcast ad time” as it “reflects life style and size of audience in specific
hours of broadcasting” (Lee, 2004). Currently, the broadcast day in Korea are divided
into four parts for TV: SA (8pm-11pm), A (8:30am-9z30am, 7:00pm-8z00pm, 11pm-
midnight), B (7:00am-8:30am, 9:30am-noon, 5:00pm-7z00pm, midnight-0:30am), and C
(noon-5:00pm, 0:30am-7z00am). The time slots mentioned above apply to weekdays.

Weekends have different time schedules. '7 Radio divides its broadcast day into three

 

‘7 On Saturday it is SA (7:00pm-11pm), A (8:30am-9z30am, 5:00pm-7:00pm, llpm-122pm), B(7:00am-
8:30am, 9:30am-5:00pm), C (0:30am-7200am); On Sunday it is SA (7:00pm-1 1:30pm), A (8:30am-7zpm,
11:30pm-midnight), B (7:30am-8z30am, midnight-0:30am), C (0:30am-7z30am)

26

components regardless Of day Of the week: A (7am-4pm, 7pm-9pm), B (6am-7am, 4pm-
6pm, 9pm-midnight), and C (midnight-6am).

While audience size and dayparts tend to co-vary, they have been considered as
different variables. While dayparts are fixed before contracts are made, knowing the
exact size and the composition of the subset of an audience composed of an advertisers’
potential customers is extremely difficult or impossible as no one in the industry can
make perfect predictions. Even the viewers themselves do not know exactly what they
will watch or listen to before they actually choose (N apoli, 2003). Furthermore, even
afier media consumption is complete, audience research firms, and thus media operators
and advertisers, do not know the exact composition of an audience due to statistical
problems of audience measurement. The uncertainty involved in estimating audience size
affects pricing decisions and thus the revenues of a media firm. As such, daypart has
merit in determining ad time prices but does not incorporate the unique characteristics of
programs. The performance of a program differs depending on genre, plot, scenario,
director, actor/actress, and many other factors even within the same daypart. Despite
difficulty in prediction, audience rating captures the variance which is not captured by
daypart.

Although the hedonic model is acknowledged to be a convenient method to
estimate valuation of ad time attributes, its theoretical and methodological weaknesses
have been addressed by many researchers. Especially, Garrod and Willis (1999) point out
that the hedonic model only can estimate consumption benefits, while Garrod’s previous
work (1994) addresses more analytical problems such as omissions of important

characteristics and incorrect mathematical specification of the model. However,

27

Vanslembrouck et al. (2003, p.19) claim these limitations can be taken into account
adequately. Referring to Palrnquist (1991) and Freeman (1993), they argue that “ignoring
the producer side does not create theoretical or econometric problems” when the focus is
on buyers’ valuation of product characteristics. This is applicable to the Korean broadcast
ad time market as the supply of the ad slots is fixed by regulation. This implies that all
advertisers make their profit-maximizing choices given the prices of ad time (Wildman,
2003) as prices reflect the existing market given the supply of ad time and its
characteristics.

Despite the abundance of empirical studies utilizing the hedonic method, few
studies have been conducted to explore empirically the implicit market prices for a
bundle of products. However, the pervasive use of bundling strategy in reality to sell
products, specifically ad spots (Park, 2004; Tankard and Henry, 1993), increases the need
to explore implicit market prices for component products sold in bundles. Meeting the
need with the hedonic approach is the central focus of this study. Closest to the present
study is research conducted by Koschat and Putsis (2002) who interpret the advertising
space as a product that bundles different audience demographics and attempts to find
implicit market prices for each unbundled (targeted) audience, by decomposing the value
of each demographic group from the bundled audience. In a strict sense, however, every
hedonic model for advertising price has a bundling component in it. For example,
Fournier and Martin (1983) looked at bundled prices Of audience demographics, since
they viewed the price for a television spot as the sum of the values of the audience

segments sold. However, Koschat and Putsis (2002) advance the logic to estimate the

28

implicit market prices for component demographics. Their argument is Stunmarized in the
following statement (Koschat and Putsis, 2002, p. 264):
Because market price is expressed as a function of a product’s characteristics, the
availability of objective measures of a product’s characteristics combined with
Observations on market prices enables a researcher to estimate the implicit price of

each characteristic. Therefore, market price can be broken down into the components
that correspond to the characteristics of the product.

However, the present study differs from the work by Koschat and Putsis (2002) in
terms of research purpose and methods employed: they explored the bundling of audience
segments, not bundling of different ad units. They intended to explore networks’
revenues expected to be generated by the targeting strategy given that current spot prices
are the market price for an audience given its demographic composition. This study,
however, explores implicit market prices for the individual ad spot comprising a package
where package prices, defined as the ad budgets spent by an advertiser through a
particular ad agency in a network, are assumed to reflect market values. Despite the lack
of literature exploring prices for the bundled ad spots from the hedonic perspective, the
prior studies using the hedonic model provide a theoretical and empirical foundation for
the present study.

In order to use the hedonic approach, it is critical to have market prices as a
dependent variable in the model. However, published rate cards in the Korean broadcast
ad market are known not to reflect market values. Hence, it is important to find an
alternative variable reflecting market values for advertising slots. The proposed model
assumes that package prices are the sums of true market values for each component ad
slot in the package regardless of prices on the published rate card. This is an especially
salient issue for this study because, if the package price is the sum of market prices for

the component programs, it should respond to the sum of their quality characteristics as

29

well. While the mathematical presentation of this argument is reserved for Chapter III,

the bundling literature provides theoretical support for the rationale.

Implicit Market Prices for the Bundled Products

Regarding relationships between the value of the bundle and the stand-alone
values of the components, bundling literature, Specifically the theory of reservation prices,
suggests three different possibilities: super-additive, sub-additive, and strictly-additive
(V enkatesh and Kamakura, 2003). Super-additivity occurs when a consumer’s valuation
for a bundle is larger than the sum of the stand-alone values for the component products,
while sub-additivity describes the opposite condition where a consumer’s valuation is
smaller than the sum of the stand-alone values for the component products. Strict-
additivity refers to the middle position where a consumer’s valuation for the bundle is
equal to the sum of his or her separate valuations for the component products.

The difference in the consumer’s valuation for the bundle compared to the stand-
alone products is attributed to the unique relationships among bundled products. Supper-
additivity arises when products complement each other in the bundle while sub-additivity
arises when products substitute for each other in the bundle as Venkatesh and Kamakura
(2003) explain:

When products are complements, a consumer’s reservation price for the bundle is
superadditive in those for the component products. Guiltinan (1987) suggests that
complementarity arises because of search economies (e. g., oil and filter changes at
the same gas station), enhanced customer satisfaction (e.g., Ski rental accompanied by
a lessons package), and improved total image (e.g., offering lawn care and shrub care
services). Alternatively, when the products are substitutes, a consumer’s reservation
price for the bundle would be subadditive in those for the components. This is likely

when the products offer (some) overlapping benefits (e. g., “Coke” and “Pepsi”) or
when they compete for similar resources such as a consumer’s time.

30

On the other hand, strict additivity arises when the component products are
independent of each other (V enkatesh and Kamakura, 2003). Venkatesh and Kamakura
(2003) conclude that many bundling articles develop their arguments based on this strict
additivity assumption after surveying key papers including Adams and Yellen (1976),
Schmalensee (1984), McAfee, McMillan, and Whinston (1989). The empirical analysis is
Simplified considerably by strict additivity. Hence, this paper will follow convention and

rely on the strict additivity assumption.

31

CHAPTER 3

EMPIRICAL MODEL OF IMPLICIT MARKET PRICES

Estimation of Implicit Market Prices

A hedonic model is developed to predict implicit market prices on the basis of
audience demographics and other characteristics of programs to which advertising
messages are attached. Past hedonic studies have suggested many factors affecting rates
that broadcast television networks and stations charge advertisers (Webster and Lichty,
1991; Fournier and Martin, 1983; Levin, 1980; Poltrack, 1983; Wirth and Bloch, 1985).
Among variables found to affect ad time prices, the proposed model includes audience
size, demographic composition, and daypart as independent variables.

While the proposed model is constructed drawing upon previous studies, it is
not relevant to directly apply previous approaches to the Korean market as there is no
evidence that published rate cards approximate market prices (Lee, 1995; 1997; 2004;
Kim and Lee, 2004; Jung, 2005). Hence, it is important to find a reasonable, alternative

variable that reflects market price. A model is constructed based on the assumption that
the package price (Pi) is the market price for the bundle Of ad units in package i
(P1. = pl x xi" + p2 x x; + p3 x x9"), where p j and x j represent market values and ad

units sold respectively for advertising units in program j , for j = 1, 2, 3, , J where

J is the number of programs in the bundle as well as the last program listed.
This assumption is reasonable because advertisers would not pay more than

their willingness to pay for the packaged products. According to Venkatesh and Mahajan

32

(1993), an individual who expects to buy J products is likely to buy the bundled products
if the bundled price (P,- ) or the mean price per product (P,- /J ) is less than or equal to
his/her total or mean reservation price. Bundling articles commonly assume that a
consumer’s reservation price for the bundle is equal to the sum of his or her separate
reservation prices for the component products.

In the context of the Korean broadcast ad market, the package price is the sum
of the products of the prices set by KOBACO and the corresponding quantities of ad
units in the package (P,- = kl x xi + k2 x x3 + k3 x x; + ). The necessary equality of the
two sums of prices and quantities can be represented as follows:

Pi =p1xxli+p2 xx; +p3 xp§.+..=k1xx1i+k2 xx; +k3 xx; +... (3.1)
where KOBACO set prices (kl , k2 , k3 ) can differ from their corresponding implicit

market values ( p1, p2 , p3 ...) if KOBACO’S prices are determined by the factors other

than market value.

If the package price is the sum of market prices for the ad units in the component
programs indexed by j , it should respond to the sum of their quality characteristics. To
illustrate the argument, a simple formula is created where the market price for a spot in

program j in package ( p ij ) is a linear fimction of audience size (A (j) and a dummy for
the program’s daypart (Dij ). Audience size is the total number of viewers watching a

particular program while dummies for daypart indicate the categorical variables for the

dayparts, SA, A, B, and C where daypart B is treated as a base case. Thus, we have:

A

SA A C
pi]. =a+,6’1Aij +,6sz- +,6’3Dij +,B4Dij (3.2)

33

If individual market prices are summed up across ad units in a package, the left
side of the equation becomes the package price (Pi) and the right side of the function
becomes the sum of the products of the coefficients and the independent variables which

now are the sum of the viewing audiences (ZAz‘j ) and the sum of ad spots in each
J

daypart (2 D5” , 2 D51, 2 D5 ) in the package. The sum of ad spots in each daypart is
J J J

the total number of ad spots the advertiser purchased in each daypart, which could
include multiple spots in some programs. As the sum of dummy values in each daypart

becomes the number of ad spots in each daypart, the notation of the sum of daypart

dummy, “ 2D!" ”, is transformed into the number of ad spots in each daypart, “ Z N ij ”,
J J

to minimize confusion. While the function with dummy variables (3.2) has the base case

as daypart B and the coefficient for each dummy is interpreted as each daypart’s

incremental effect relative to the base case, an equation with Z N ij does not need to
J

have a base case and the coefficients for Z N 5A ,ZNJI , ZNUB , 2N5 indicates their
1' J J J

own effects on the package price. Thus, interpretation of the constant term (or ) also

changes. While or in equation (3.2) represents the original intercept and includes the

effect of the base case, the corresponding term (a) in an equation with Z N ij
J

represents the old intercept term without the effect of daypart B. (a) is thus replaced by

((51):: N; ). Thus, we have:
1

34

21:19,]- =(-Z;NiJT- +IB1§J,:Az'j +51%:NgA +522;le

B C
”32% ”42% :1), (3.3)
J J

However, reality is more complex than this. Advertisers buy advertising spots
reflecting demographic characteristics of consumers of their products and the effects of

spot duration on consumer behavior. Hence, the preferred specification is where P, is the

m2049
Air

package price, is the number of viewing males aged 20-49 in the package,

A If 2049 is the number of viewing women aged 20-49 in the package, and A (JO-”7” is the

number of male and female audience aged other than 20-49 weighted by the length Of ad
spots in the package. For simplicity, the number of viewers weighted by the spot seconds
is called audience seconds. Total audience seconds are calculated by multiplying the ad

time purchased in each program times the size the program's audience size and then

summing the resulting products over all programs in the package. N 5A is the number of
ad spots in dapyart SA, Nil/.1 is the ntunber of ad spots in daypart A, N 15'} is the number of

ad Spots in daypart B, N If is the number of ad spots in daypart C weighted by the
lengths of individual ad spots in the package. The weighted number of ad spots is called

the number Of ad units. Hence, 2 N ”T is now the total number of ad units in the package.
j

The relationship can be represented as follows:

_ ‘ T \ 2722049 f2049 other
13.—aim]. +fl1;.4,j +62%:Aij 1563;.40.
J

35

(3.4)

SA A B C
+61;Nij +52;sz +53;Nij +64%:Nij +eij

Where P- is a nonnegative dependent variable and e -- are the sum of independent

I 11
random errors with a mean of zero.

Reality adds more complexity to the model. Packages are constructed with
various types of ad slots including program ad slots and block ad slots for the major TV
networks (KBS-TV, MBC-TV, and SBS-TV) as well as for the subsidized radio
broadcasters (FEBC-AM, CBS-AM, CBS-FM, PBC-FM, WBS-FM). Hence, the product
should be specified further by identifying TV-program ads, TV-block ads, radio-program

ads, and radio block ads where radio indicates those subsidized. In the final model, the

package price (Pi) is modeled as a function of TV audience seconds for each ad type and

PG PGf2049 PG
each demographic specification( A ”’2049 , A , A 01h” ;
Z]: TVij Z TVij 21:. TV!)-

ZTV AB Lm2049 ’ZTV A BLf2049 ,ZAfVij LOW” ,where superscript “ PG ” and “ BL ” represent
”’7 VJ

program ads and block ads respectively), radio audience seconds described by the ad

. PG 7 PGf2049 PG 1 BL 20 9
types and demographics ( A ”171-049 , A __ , A .. 0’ 7"r; A ”m 4 ,
BLf2049 BL (h) - - PGSA
EAR” ,2 ARHO " ), number of TV ad units In each daypart (Z NTV.. ,
J ’ J ’J j U

PG
ZNT (:A’ENTVy i3 ZNTVif ZNfVSA ZNTV; ’ZNTVf ZNTVS ),and

PC PC
number of radio ad units in each daypart (2 NR] A ,Z NR1] B ,Z N; 0": ,2 N11; A,

36

BL B BLC . . .
Z N R.. , Z N R.. ). There 15 no commonly accepted, complete or accurate Information
J' ’1 J ’1

on radio audience size available. Hence, the audience second variables related to the
radio ad time are removed from the proposed model. A complete listing of the variables

with their definitions appears in Table 1.The model can be represented as follows:

— PG 130
_. T ‘PC5n2049 .f2049 cnher
Pi "aZNij +IBIZATV1“ +flQZATV~ +fl3ZATV“
J J 1 J '1 J ’1
+,B ZABLm2049 +16 XABLf2049 "I'fl ZABLot/ter
4 TV,” 5 , TV.. 6 TV..
J IJ IJ
J J J
PG PG PG PG
SA A B C
+512NTVIH +522:an +53ZNTV__ +54ZNTV”
J U U U
J J J J
BL BL BL BL
SA A B C
“552qu 4462an +67ZNTV” +582NTVH
J ’J J ’1 J ’1 J ’1

PGA PGB PGC
+692NR” +5IOZNR” “SHEA/R.-
J ” J ’1 J ’1

BLA BLB BLC
+612;NRIJ mug“? MHZ/1%,] +ey. (3.5)

37

Table 1. Variables and Data

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Definition Source
P2 ‘k' 0 P"‘ : S f KOBACO .‘ t f ‘d‘l‘fe‘e tzd 1' ":h' ."d b ‘d t th‘ Uh ' d :0
Pubr 1c age rice um 0 so prrccs or 1 crcn‘ 1 Irme lel asc y a vcr rser a rout, a tgency b KOBACO
from network (
’)
ZAT vzzdmg Total TV audience seconds for the male 20—49 group revealed to program ads in a package abt‘ The Korea Stigfiglcgearbook 2004
PG .
f 4049 . . . . KOBACO
A . , 4.. . . . 7_(o. -. fl) 0.. ‘2"‘0 . .. s
2; Tvabq Total TV audience seconds for the female 20 4) group rcVe Ilcd to program ads In a p tckage abc The Korea Statistlcal Yearbook 2004
A other 2 I ,7 .. .‘ I .,‘. I 9_ O, -, ”t _OII( I .' I "'0 _ KOBACO,
g: T (1!ch Tot 11 TV audience seconds for other than male/female -0 49 grouI reve fled to progr rm ads In a package abc The Korea Statistical Yearbook 2004
AB Lm2049 I I . I . . . . I 7 _ O “I 2 . . 2 I I 0 . KOBACO,
ZTV Tvabq' Total TV audrencc seconds for the male 20 49 group revelled to block Ids 1n 1 package abc The Korea Statistical Yearbook 2004
Lf2049 I I -. ,~ I 7 _ II , . . .- I .., . KOBACO,
gAii/Z’JCJ Total TV audience seconds for the female 20 49 group rev.aled to block ads In a package abc The Korea Statistical Yearbook 2004
AB L'othel I 4 . I I - I a _ 0, _ I . 7 I . - I I I a . KOBACO,
ZTV Vabcj Total TV audience seconds for other than male/female 20 49 t,Ioup revealed to block ads In a package abc The Korea Statistical Yearbook 2004
Z N:l:j Total number Of ad units among programs ( j ) in a package abc KOBACO
ZN TVGS 17A Total number of TV program ad units in daypart SA among programs ( j) in a package abc KOBACO
a cj
Z NTVA b Total number of TV program ad units in daypart A among programs (j) in a package abc KOBACO
a Cj
Z NTVB b Total number of TV program ad units in daypart B among programs (j) in a package abc KOBACO
a Cj
Z NTV Wabcj Total number of TV program ad units in daypart C among programs (1) in a package “/99 KOBACO
21(13ijthch Total number Of TV block ad units in daypart SA among programs (1) in a package ab" KOBACO
Z N ijbcj Total number of TV block ad units in daypart A among programs (j ) in a package abc KOBACO
ZN TVaBbcj Total number of TV block ad units in daypart B among programs ( j ) in a package abc KOBACO
BL .
ZNTVC Total number Of TV block ad units in daypart C among programs (1 ) in a package abc KOBACO

j abcj

 

 

 

 

 

38

 

 

 

 

 

Table l (cont’d)

 

 

 

 

 

 

 

 

 

 

ZNR C

 

 

Total number of religious broadcasters’ block ad units in daypart C among programs (j ) in a package abc

 

PG ’
Z N 1!) , Total number of relrgrous broadcasters program ad units In daypart A among programs ( _] ) In a package abt KOBACO
j c ('j
PGB . . .
Z NR I , Total number 01 religious broadcasters‘ program ad units in daypart B among programs ( j ) in a package (the KOBACO
j a )(J
PGC
Z N I _ Total number of religious broadcasters’ program ad in daypart C among programs ( j) in a package abr- KOBACO
J a )(J
BLA . , _ .
ZNR I _ Total number of relIgious broadcasters’ block ad umts 1n daypar= A among programs (j ) In a package (the KOBACO
j a 7‘7
BL}, . . . . . 7
NR I , Total number Ofrel1gious broadcasters’ block ad umts 1n daypart B among programs (1 ) In a package abc KOBACO
j a )1]
BL
KOBACO

 

 

 

 

; j abcj

39

 

 

 

thett

toaH

incre

XX

,1

dhid

pace.

 

- I?

-F;;

 

However, this model could suffer from multicollinearity among variables because

the total number of ad units in the package (2 N 5 ) is correlated, with a varying degree,
1'

to all other variables in the model. That is, adding another unit into the package also

increases audience size as well as ad units in one of dayparts. It is possible that

Z N UT swamp everything else in the regression with this model. Hence, it is necessary to
J

divide the formula by z N UT to focus on per commercial unit price rather than package
1'

price.

This treatment produces a model where average price per unit in the package

(PI- /Z N ”T ) is explained by average audience seconds per unit for each demographic

characteristic In TV ([ZT AP VGI’”2049 / ZN; ], [2A7]; (32049 / Z N UT l,
J

PG
A other / NIT AB 51/722049/ N-T A B/If2049/ NIT
I; WV :1] 'IZT Z; 1 [2 TI 21
[Z Aféf’he’ /Z N VT ]) and the fraction of each daypart in TV and radio
I V J
PG PG PG T PG T
([gINTVI;9A /;NIJT- ], [ZNTVI (1 /ZNI.IT ] @NW; /;NV ], [§I_‘INTVIIC /;Nij ],
[ZNTVLISA/ZN ], [ZNfVIA/ENVI, [ZNffié/ZNVT. ], [ZNTVC/ZN.-],

PG PG BL
[Zr/.4IZNILIMP/2mIZNR..CIZNIII2~.AIZNIL
J V J J V J J U j J V J

40

BL BL
[X NR..B /ZNyT-], [Z NR..C /2 N5] ). As the sum ofthe fraction terms equals to 1,
J' ’1 J' J ’1 I

however, one needs to remove a fraction term (say, fraction of TV program ads in

daypart B) from the following model as a base case.

T _ A A PC 2049 T A PG 2049 T
RIZNV —a+fll(ZATVIf” /ZNy-)+fl2(ZATV_If /ZNI.J-)
J J ’1 J J ’1 J
" PG T A BL T
+ .642 Angst” IZNV )+/3.(Z ATVI’T’2049 IZNV- )
J J J J J J
A BL A BL
2049 T -T
“LANE/INA /ZNIJ) +fl6(ZATV?ther/ZNIJ)
J ’1 J J ’1 J
A PC A PG
SA T ,4 T
+al(;NTVII /;NV)+52(;NTVII lgNy)
A PG A BL " BL
C T SA T ,4 T
+53(;NTVII /;NU)+64(;NTVIJ_ /;Nij)+65(;NTVIj lgNy)
A BL A BL A m
B T C’ T A T
+a6(;NTVII /§NN+67(;NTVIJI /21:NI-j)+58(;NRIj /;Nij)
A PG A PG A 3L
B T C T A T
+59(;NRII /ZJZNII.)+5IO(;NRIJI /;NV)+5“(;NRII /;Nij)

" BL A BL
B T C T
+512(§IINRII /2jINy-)+613(ZJINRIJI /§j:NI-j)+eij (3.6)

Once estimation is complete, implicit market prices for individual ad units can be
computed by using regression coefficients. As per equations (3.2) and (3.3), coefficient
estimates are the same for the individual spot prices and the package prices and they
remain the same for the average package prices per unit in formula (3.6) where both sides

of the package price equation are divided by the total number of ad units in the package.

41

The coefficients of average audience seconds per spot in formula (3.6) can be interpreted
as the unit price increase attributable to a one unit increase of audience seconds in the
particular ad slot. The coefficients for the fraction terms of dayparts can be interpreted as
if they are coefficients for dummy variables for the respective dayparts in the pacakge.
To explain the logic, let us assume one of the fraction terms turns out to be 1. This means
the package is composed of a single product (say, TV program spots in daypart SA) and
the average spot price of the package is merely the average spot price of that particular
type of spot (i.e., TV program spots in daypart SA). The constant term in the model can
be interpreted as the coefficient of the base case.

Hence, the implicit market price for an individual ad unit can be estimated by

applying the coefficients estimated with the package prices to the individual

A A A

characteristics of the purchased ad time: ,6] , ,62 , and ,63 respectively are multiplied

times the male 20-49 audience seconds, female 20-49 audience seconds, and audience

seconds of male and female aged other than 20-49 for the individual spot of TV program

A A A

ads. ,64 , ,B5 , and 166 are multiplied times the male 20-49 audience seconds, female 20-

49 audience seconds, and audience seconds of male and female aged other than 20-49 for
the individual spot of TV block ads. As the coefficients for the fraction terms of dayparts

can be interpreted as if they are coefficients for dummy variables for the respective

A A A A A A

dayparts in the package, as noted above, 6 , 2,63,64,65,66,67,68,69,510,611,612 ,

—

A

and 513 , respectively are multiplied to the dummy value for each daypart. To the dummy

value for daypart B, which is the base case, constant term is multiplied.

42

Estimation of Price Distortion

Price distortion is defined as the gap between estimated implicit market price and
the KOBACO set price for the particular ad slot. Hence, it is calculated by subtracting the
estimated implicit market price from KOBACO-set price. For presentational purposes,
however, average differences between KOBACO set prices and estimated implicit market
prices are compared for each category (TV program ads in daypart SA; TV program ads
in daypart A; TV program ads in daypart B; TV program ads in daypart C; TV block ads
in daypart SA; TV block ads in daypart A; TV block ads in daypart B; TV block ads in
daypart C; radio program ads in daypart A; radio program ads in daypart B; radio
program ads in daypart C; radio block ads in daypart A; radio block ads in daypart B;
radio block ads in dayaprt C) and for each broadcaster (BBS-F M; F BBC-AM; CBS-AM;
CBS-FM; PBC-FM; WBC-FM). A comparison is also made between expenditures on ad
units in each daypart calculated with the implicit market prices and those calculated with

the KOBACO set prices.

Estimation of Subsidies
There can be two types of subsidy flows in the market. First, the external subsidy
flowing among broadcasters, and they we estimated by subtracting each broadcaster total
expenditures with KOBACO set prices from the estimates of what expenditures for the
same units would be, with the implicit market prices estimated for each broadcaster. The
internal subsidies flowing from commercial programs to public interest programs within
a major TV network should be estimated by comparing expenditures for each program

type. However, the estimation of internal subsidies cannot be completed unless public

43

interest programs are clearly identified. No one has undertaken this exercise and I will
follow precedent and ignore this type of subsidy in this study. However, the method

employed here could be used to examine internal subsidies.

44

CHAPTER 4

METHODOLOGY

Data

To test the proposed model, actual transaction data for the 100 largest Korean
broadcasting advertisers were obtained from KOBACO for the one-year period May
2003 to April 2004. The large number of small advertisers (yearly, more than 6,500
advertisers bought at least one ad spot in terrestrial broadcasting programs) makes the
task of gathering and processing data for all of them infeasible. However, expenditures by
the largest 100 advertisers account for 64.3% of all terrestrial broadcast advertising
revenues and dominate in the determination of ad time prices. So it is unlikely that
restricting the sample in this way is a source of significant bias in the estimates.
Furthermore, I assume this introduces no bias to the estimates as I can’t think of reasons
why advertisers smaller than the 100 largest should value the ad slots differently at the
margin that do major advertisers. Plus, if there is competitive bidding among all
advertisers for access to ad time slots, then the same competitive implicit market prices
should apply to all of them.

The sample period is concurrent with the upfront markets where contracts for ad
slots with duration longer than six months, airing May 2003 to April 2004, were
negotiated.

From the data, advertisers and advertising agencies were identified by randomly
allocated numeric codes preserving anonymity. Other information in the data included

spot prices set by KOBACO, airing seconds of the ad slots, and contract duration of each

45

ad slot. The data also included the audience ratings of programs to which the purchased
ad slots were attached. Audience ratings of TV programs were subdivided into those for
males aged 20-49, females aged 20-49, and males and females of ages other than 20-49.
In the sampling period, Nielsen Media Research (NMR) and Taylor Nelson Sofres (TNS)
conducted audience research on 1550 and 1200 TV household panels respectively,
employing people meter technique (Lee, 2004). NMR audience ratings were used for this
study. However, the reporting format of SBS audience ratings differed from those of
KBS and MBC audience ratings. While audience ratings for KBS and MBC programs
were averaged nationally, those for SBS programs were reported separately for each
region covered by networked local broadcasters. Hence, locally surveyed audience
ratings of SBS programs were averaged by weights of audience size in each region, to
calculate national average audience ratings for networked programs.

The original data contained 102,080 contracts made with major TV networks
(23,681 with KBS TV; 38,099 with MBC TV; 40,300 with SBS TV) and 16,246
contracts made with religious broadcasters (2,658 with BBS-F M; 4,214 with CBS-AM;
1,975 with CBS-FM; 1,680 with FEBC-AM; 2,925 with PBC-FM; 2,794 with WES-FM).
The data did not include ad contracts for “yearly sports” programs that presented the
major sporting events held through the year. Advertisers deposited certain amount of
their ad budgets in advance for the ad time attached to those programs. Due to this
transaction pattern, ads on the yearly sports programs were regarded as not packaged with
other programs. In addition, data did not include local ad spots as they tend not to be

packaged with ads broadcast nationwide.

46

However, the original data contained errors or missing values. A major source of
errors is incorrect record-keeping of contract periods. The data included starting and
ending dates for the supply of commercial time in a program, but in some instances, end
dates were recorded prior to starting dates. I acquired the number of ad slots purchased by
an advertiser for a particular program by counting the number of aired days for the ad slot
within the contract period. Hence, errors in contract dates made it impossible to count the
number ad slots during the contract periods. For example, if an ad slot in a program was
filled by a commercial message every Monday for a three-week period, the particular
advertiser was assumed to buy three ad slots for the program. Also, audience ratings were
missing in some instances. Those observations with obvious errors or missing values
were removed from the data.

Afier excluding all problematic observations the data set included 98,558 contract
cases, made with major TV networks (22,864 with KBS TV; 36,946 with MBC TV;
38,748 with SBS TV) and 10,585 contract cases with religious broadcasters (1,978 with
BBS-FM; 3,162 with CBS-AM; 1,714 with CBS-FM; 1,264 with FEBC-AM; 1,432 with
PBC-FM; 1,035 with WBS-FM).

Measurements

Package Price: Package price is operationalized as the amount an advertiser spent
on ad time sold by a network (and its associated religious broadcasters), through a
specific ad agency, during the sample period. As each of the KOBACO sub-units18 sold
ad time for one of the three major TV networks and its associated religious radio

broadcasters to ad agencies (see figure 1), it is reasonable to assume that packages were

 

'8 Three KOBACO subunits exist: one to represent KBS-TV, KBS radio stations, EBS-TV, FEBC-AM, and
Sunny FM; one to represent MBC-TV, MBC radio stations, CBS-FM, CBS-AM, PBC, and KP M; one to
represent SBS-TV, SBS radio stations, BBS-F M, WBS-F M.

47

constructed per ad agency per advertiser for broadcasters represented by each of the
KOBACO sub-units. In figure 1, the separate arrows represent individual packages. This
implies that a package created for an advertiser by one agency was not combined with the
package created for the same advertiser by the second agency. Also, this implies that a
package was not created by combining ad slots across the major networks (and their
associated religious broadcasters) represented by each of the KOBACO sub-units.

While this method is a way of dealing with the absence of data identifying
individual packages, it has the limitation of treating all purchases by an ad agency for an
advertiser as a single package, when multiple packages may have been purchased.
However, this does not violate the logic of the package-price regression approach
outlined above, as KOBACO prices are the same for all packages. The downside of
treating all purchases by an advertiser through a particular ad agency as the purchase of a
single package is that it reduces the number of observations. Nevertheless, if there is
enough variation among the constructed packages bought by different advertisers through
different ad agencies, it becomes possible to estimate significant coefficients identifying
impacts of audience demographics and other characteristics of component programs on
the package prices which, in turn, makes it possible to estimate implicit market prices for
individual programs. While it is desirable to use observed market prices according to
economic theory, using a proxy variable for prices causes no problem from an estimation
perspective.

Package prices were calculated by adding all the KOBACO prices for ad spots

purchased by an advertiser (a) through a particular ad agency (b) from a network (and

48

its associated religious and local radio stations) (c ). This definition can be presented via

the following formula:
Pabc : gijabcj (4.1)

Where a is advertiser, b is advertising agency, c is network (henceforth, I will use

“network” to include a network’s affiliated religious broadcasters), and j is a program

on the network. k j is a KOBACO set price for an ad spot attached to program j and

x a bcj is the number of ad spots associated with program j, purchased by advertiser a

through advertising agency b from network c. While the data set contained prices set by
KOBACO, the numbers of ad slots purchased at these prices were not provided. Hence I
have assumed the number of ad spots aired during the contract period to be the same as
the number of days a commercial was aired during the contract period in the network or
the associated religious broadcasters whose time was purchased.

Audience Size per Demographic Characteristic: Audience size of a package is
operationalized in terms of total audience seconds of the component programs in a

package. Individual audience seconds for TV program ads in a particular demographic

PG PG PG
(A ”’2049 for males 20-49, A f2049 for females 20-49, A 0m” for males and
TV TV TV
abcj abcj abcj

females with ages other than 20-49) were calculated by multiplying the audience rating of

PG PG PG , ,
the program (R j "’2049 , R j f2049 , R j 0m” ) times the total population of the

corresponding demographic group (Pfi,2049 , FTC/2049 , PTof/her ), and the length of a

commercial measured in seconds of each ad slot purchased by an advertiser for a

49

particular program (S is the number of ad slots attached to program j

abcj )' xabcj
purchased by advertiser a through advertising agency b from broadcaster c during the

sample period. Hence, total audience seconds of program ads per demographic group in a

acka e AP Gm2049 A PGf2049 AP 00"” is the sum of audience seconds
p g (ZTV ZTV TVabcj ZTV TVabcj )

calculated for the component programs, as the following formula shows.

:14]? m2049_ =Z(Rf Gm2049 x P75132049 XS

Vabcj abcj X xabcj)

2049 2049 2049
ZATVf =Z(Rf Cf x1374) xS xx

- )
a bq €1ij

abcj

G_.0ther _ Z( R Gather XPotherx S

4.2
Vabcj ( )

ZAPV

However, as no published data are available for TV population specified by the

abcj x xabcj )

demographic characteristics, the calculation was made with the assumption that 100% of
the population in each demographic group watched TV. The Korea Statistical Yearbook
2004 provides the number of Koreans by gender and age. All the measures for males 20-
49, females 20-49, and the others are counted as the potential viewers in each

demographic category. S a be j is the length in seconds of every ad spot purchased by an
advertiser. xabcj is the number of ad slots attached to program j purchased by an

advertiser through a particular advertising agency from a broadcaster during the sample
period.
Audience seconds for TV block ads were calculated following the same process

used for TV program ads except the audience ratings of the program

50

BL BL BL . _
(R j ”’2049 , R j ”049 , R j 0m” ) were measured d1fferently. As block ads are located in

the middle of two programs, it was hard to associate the block ads with the audience
ratings of either program. Hence, audience ratings of two adjacent programs are averaged
for the block ads.

Daypart: TV dayparts are classified as SA, A, B, and C as described above. SA is
equivalent to the primetime daypart in the USA broadcasting system, A to primetime
fringe period, C to early morning and late night, and B to other times. For program ads,

dayparts were first dummy-coded for daypart SA, A, B and C and then weighted by the

PG PG B

PGS A A PGC _
length of ad 3 ot (D , D , D , and D ). By settin a 15 second ad spot
P TV j TV j TV j TV J. g

as a base case for TV ads, weights were calculated by dividing total ad seconds for each
spot by 15. Then, the weighted dummy values were multiplied by the number of ad slots

.) and summed for each package bought by an advertiser through a

in program j (xabcj

particular ad agency. As the sum of weighted dummies are the same as the total number
of ad units, as discussed above, notation for dummy “ D ” was replaced with the number

of spots “ N ” here, as shown by the following formula:

PG PG
N SA = (D SA x x )
2]: TV ; TV a bcj £2ij

abcj
PG PG
N A = (D A xx .)
; TVabcj ; TVabcj abcy

PG PG
N B = (D 8 xx .)
g: TVabcj ; TVabcj (7ij

PC PC
N C = D C X x . 4.3

51

This process was applied in the same way to TV block ads and radio ads for both
program ads and block ads. Weights were calculated by dividing ad seconds of each spot
by 20 for radio ads as radio ad time was generally sold in 20 second units.

Construction of Packages: Using the pivot function in Microsoft Office Excel
2003, individual contracts are summed over the advertisers and ad agencies. It returned
467 package cases. This means that the 100 largest advertisers bought ad time through
three different KOBACO sub-units and that some advertisers made deals through more
than one ad agency. Among the 467 cases, 151 cases were made through the KOBACO
subunit representing KBS-TV and its associated religious broadcasters, while 161 cases
were made through the subunit representing MBC-TV and its associated broadcasters and
155 cases were made through the subunit representing SBS-TV and its associated

broadcasters.

52

CHAPTERS

RESULTS

This chapter reports regression results obtained using package prices, defined as
advertisers’ annual ad budgets spent in a particular network (and its associated
broadcasters) through a particular advertising agency. By this definition, 467 package
cases originally were constructed. However, due to missing values in the data, 12 cases
were automatically dropped from the analysis and 455 cases were used for estimation
where the average price per unit in the package (APPU) was KW 3,172,547 million, and
each component of the advertising types comprises packages in proportions shown in
Table 2. The columns values of minimum, maximum and mean for the fraction variables
represent the proportions composing the package. On average, program advertising in
daypart A with religious broadcasters comprised the largest proportion, 21% of the total
spots in package, while program advertising in dayparts A and B with the major TV
networks comprised 18% each. Program advertising in daypart SA and daypart C in the
major networks comprised 14% and 9% each. Program advertising in daypart B in
religious broadcasters comprised 7%. In sum, 86% of the total package spots comprised
these six types of advertising and the remaining 14% were comprised of block
advertising in the major TV networks and religious broadcasters plus program advertising
in daypart C in the major TV networks. Average audience seconds per unit (AASPU) for
TV program advertising were 14,709,000 while those for TV block advertising were

2,698,000. AASPU for both program and block advertising on radio were not taken into

53

Table 2. Descriptive Statistics _ Before Outlier Removed

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N Minimum Maximum Mean Std. Deviation
Average Package Price Per Spot 455 18,627 10,005,000 3,172,547 1,501,459
Average F2049 Audience Seconds of TV Program Ads 455 0 21,416 4,787 2,395
Average M2049 Audience Seconds of TV Program Ads 455 0 10,341 3,090 1,538
Average Other’s Audience Seconds of TV Program Ads 455 0 26,205 6,826 3,480
Fraction ofTV Program Ads in Daypart SA 455 0 1.00 0.14 0.11
Fraction of TV Program Ads in Daypart A 455 0 0.83 0.18 0.12
Fraction ofTV Program Ads in Daypart B 455 0 0.83 0.18 0.14
Fraction of TV Program Ads in Daypart C 455 0 0.73 0.09 0.10
Average F2049 Audience Seconds ofTV Block Ads 455 0 15,310 880 1,287
Average M2049 Audience Seconds of TV Block Ads 455 0 13,147 559 941
Average Other Audience Seconds of TV Block Ads 455 0 25,802 1,259 1,878
Fraction of TV Block Ads in Daypart SA 455 0 0.91 0.04 0.07
Fraction of TV Block Ads in Daypart A 455 0 0.48 0.02 0.05
Fraction of TV Block Ads in Daypart B 455 0 0.60 0.03 0.05
Fraction of TV Block Ads in Daypart C 455 0 0.20 0.01 0.02
Fraction of Religious Radio’s Program Ads in Daypart A 455 0 1.00 0.21 0.20
Fraction of Religious Radio’s Program Ads in Daypart B 455 0 0.63 0.07 0.10
Fraction of Religious Radio’s Program Ads in Daypart C 455 0 0.14 0.01 0.02
Fraction of Religious Radio’s Block Ads in Daypart A 455 0 0.90 0.03 0.08
Fraction of Religious Radio’s Block Ads in Daypart B 455 0 0.16 0.01 0.02
Fraction of Religious Radio’s Block Ads in Daypart C 455 0 0.11 0.00 0.01

 

54

 

 

 

 

acco

01‘ dz

1101'

1356

to [1

101:

regi

\l‘t’r

06$

SUC

CU

 

 

 

account in this analysis due to lack of data. However, advertisers work with the same lack

of data. Therefore, this treatment is believed not to cause a bias.

Model Re-speeification

Before the main analyses were conducted, data were screened for outliers.
However, outlier presence is unrelated to sampling error or incorrect data entry by a
researcher, as population data provided by KOBACO were used in the analysis. Referring
to the Tabachnick and F idell’s (1996) definition of outlier, two strategies were employed
to remove potential outlier impacts: remove cases more than 3 standard deviations from
the mean in the dependent variable and another ten cases having highest residuals in the
regressions from the visual inspection of the residuals plot. With this measure, 12 cases
were removed from the data and 443 cases were left out of 455.19 A complete data
description is presented in Table 3.

After outliers were removed, Pearson correlations among the variables were
checked, as Table 4 showed. It was found that, while correlations between other
independent variables are .79 at the highest, correlations between the audience seconds of
demographic groups were extremely high, .92 to .98. Three possible reasons exist for
such high correlations. First, this could be the nature of the original data. Due to certain
cultural reasons, Koreans may watch TV with little variation across demographic groups,
hence, audience ratings reported by Nielsen Media Research may have originally high
correlations. If this is the case, the logic of employing demographic composition as a

variable affecting advertising prices does not apply to the Korean broadcast market and

 

'9 I also ran a regression with outliers and found that exclusion of outliers did not change the qualitative
nature of my findings.

55

 

Table 3. Descriptive Statistics N After Outlier Removed

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N Minimum Maximum Mean Standard Deviation

Average Package Price Per Spot 443 18,627 7,275,000 3,078,250 1,378,650
Average F2049 Audience Seconds of TV Program Ads 443 0 1 1,437 4,655 2,095
Average M2049 Audience Seconds of TV Program Ads 443 0 8,276 3,017 1,423
Averag e Other’s Audience Seconds of TV Program Ads 443 0 18,104 6,671 3,204
Fraction of TV Program Ads in Daypart SA 443 0 0.58 0.13 0.10
Fraction ofTV Program Ads in Daypart A 443 0 0.83 0.17 0.12
Fraction of TV Program Ads in Daypart B 443 0 0.83 0.18 0.14
Fraction of TV Program Ads in Daypart C 443 0 0.73 0.09 0.10
Average F2049 Audience Seconds of TV Block Ads 443 0 9,534 841 1,086
Average M2049 Audience Seconds of TV Block Ads 443 0 7,266 529 733
Average Other Audience Seconds of TV Block Ads 443 0 12,1 15 1,193 1,464
Fraction of TV Block Ads in Daypart SA 443 0 0.65 0.04 0.05
Fraction of TV Block Ads in Daypart A 443 0 0.48 0.02 0.05
Fraction of TV Block Ads in Daypart B 443 0 0.60 0.03 0.06
Fraction of TV Block Ads in Daypart C 443 0 0.20 0.01 0.02
Fraction of Religious Radio’s Program Ads in Daypart A 443 0 1.00 0.21 0.19
Fraction of Religious Radio’s Program Ads in Daypart B 443 0 0.63 0.07 0.10
Fraction of Religious Radio’s Program Ads in Daypart C 443 0 0.14 0.01 0.02
Fraction of Religious Radio’s Block Ads in Daypart A 443 0 0.90 0.03 0.08
Fraction of Religious Radio’s Block Ads in Daypart B 443 0 0.16 0.01 0.02
Fraction of Religious Radio’s Block Ads in Daypart C 443 0 0.11 0.00 0.01

 

 

56

 

 

 

 

Table 4. Pearson Correlations of Package Data

 

 

Ix)

5

6

7

8

9

10

ll

20

 

Average Package Price Per Spot

 

1.
2. Average F2049 Audience
Seconds ofTV Program Ads

 

3. Average M2049 Audience
Seconds ofTV Program Ads

0.96

 

4. Average Other’s Audience
Seconds of TV Program Ads

0.92

0.95

 

5. Fraction of TV Program Ads in

Daypai SA

0.80

0.69

0.70

0.70

 

6. Fraction of TV Program Ads in
Daypart A

0.69

0.70

0.68

0.70

0.27

 

7. Fraction of TV Program Ads in
M

0.17

0.12

0.20

0.05

0.00

0.04

 

 

 

8. Average F2049 Audience
Seconds of TV Block Ads

0.07

—0.03

—0.01

—0.05

-0.03

—0.10

—0.01

 

9. Average M2049 Audience
Seconds of TV Block Ads

0.09

-0.02

0.02

-0.04

0.00

—0.10

0.00

0.97

 

10. Average Other Audience
Seconds of TV Block Ads

0.10

—0.01

0.02

—0.02

—0.01

—0.09

—0.01

0.98

0.96

 

11. Fraction of TV Block Ads in
Daypart SA

0.13

—0.01

0.03

—0.01

0.08

—0.07

—0.03

0.77

0.79

0.78

 

 

 

 

12. Fraction of TV Block Ads in
WA

0.01

—0.02

—0.02

—0.04

-0.04

—0.07

0.03

0.64

0.59

0.63

 

13. Fraction of TV Block Ads in
Daypart B

—0.08

-0.10

-0.11

—0.10

—0.16

—0.08

-0.02

0.55

0.45

0.53

0.33

 

 

14. Fraction of TV Block Ads in
Daypart C

—0.16

—0.16

—0.17

—0.19

-0.16

—0.15

0.01

0.27

0.21

0.25

0.24

0.54

 

 

15. Fraction of Religious Radio’s
wAds in Daypart A

-0.62

—0.60

—0.58

-0.61

—0.35

—0.47

—0.31

—0.28

—0.26

-0.29

-0.20

—0.22

-0.23

—0.09

 

 

16. Fraction of Religious Radio’s
Program Ads in Daypart B

—0.39

—0.38

—0.36

—0.36

—0.25

-0.23

-0.19

—0.21

-0.19

-0.22

—0.16

-0.15

-0.14

-0.05

0.15

 

 

17. Fraction of Religious Radio’s
Program Ads in Daypart C

-0.18

—0.18

—0.17

—0.16

—0.05

—0.15

-0.10

-0.09

—0.08

—0.09

—0.04

—0.06

—0. 10

—0.06

0.12

0.19

 

18. Fraction of Religious Radio’s
Block Ads in Daypart A

-0.29

—0.28

-0.28

-0.29

-0.20

—0.20

—0.13

—0.07

—0.07

—0.09

—0.09

—0.05

—0.04

0.00

0.01

—0.05

—0.04

 

 

19. Fraction of Religious Radio’s
Block Ads in Daypart B

—0.20

—0.18

-0.17

-0.20

-0.15

—0.14

-0.08

—0.05

—0.04

—0.06

—0.06

—0.04

—0.05

0.01

0.00

—0.05

—0.01

0.49

 

 

20. Fraction of Religious Radio’s
Block Ads in Daypart C

—0.11

 

 

—0.10

 

—0.10

 

-0.10

 

-0.08

 

—0.06

 

—0.07

 

—0.04

 

—0.04

 

-0.05

 

—0.04

 

~0.02

 

—0.05

 

—0.02

 

-0.07

 

-0.06

 

—0.04

 

0.44

 

0.57

 

 

 

57

 

 

[ECO

gror

they

 

 

 

the three different audience-demographics should be combined into one for the purpose
of regression estimation. Otherwise, they will generate a multicollinearity problem in
estimation. Second, this might be attributed to the measurement process of summing
multiple spots into one package. In this case, how to measure audience size should be
reconsidered. Third, it may just reflect possibility that the relative sizes of demographic
groups in audiences don’t change nearly as much as their absolute sizes, even though
they do vary.

Table 5 shows that high correlation between demographic groups is not due to the
measurement process. Correlations between individual programs’ audience ratings for
each demographic group were high (for program ads, .93, .91, .91 in KBS; .96, .97, .95 in
MBC; .98, .92, .90 in SBS between female 20-49 and male 20-49, female20-49 and other,
male 20-49 and other respectively; for block ads, .95, .93, .93 in KBS; .89, .94, .87 in
MBC; .90, .94, .91 in SBS respectively). This implies that audience segmentation in
Korea does not have such critical weight as it does in USA. Reflecting on these findings,
the regression model proposed in chapter 3 was re-specified as presented in equation
(5.1). R 2 remained almost the same when audience demographic groups were combined
into one (.96 in the model having three specified demographic groups; .95 for the model

with them combined).

T A A PG T A BL T
1)./2N.j =a+131<ZATV§UM/ZN,)+P.(ZATV§UM /ZN,-,)
J J ’1 J J ’1 J
A PG A PG A PG
SA T A T C T
+6I(ZJ;NTVU /;NU)+52(;NTVU /;N,.j) ”AZ/ZN“? /;N,j)

A BL A BL A BL
SA T A T B T
+54(;NTVI_]_ /;Nij)+55(;NT,/ij /;N,j)+56(;NTVij /;NU)

58

 

Table 5. Pearson Correlations of Audience Ratings in the Major TV Networks

 

 

Program Ads

 

Female 2049

Male .3049

 

Program Ads

KBSTV
Block Ads

Female 2049
Male 2049
Other
Female 2049
Male 2049
Other

1
0.93
0.91

 

Block Ads

 

Other

Female 2049

Male 2049

Other

 

0.95
0.93

0.93

 

Program Ads

MBCTV
Block Ads

Female 2049
Male 2049
Other
Female 2049
Male 2049
Other

0.96
0.97

l
0.89
0.94

0.87

 

Program Ads

 

Block Ads

 

SBSTV

Female 2049
Male 2049
Other
Female 2049
Male 2049
Other

 

0.98
0.92

 

0.90

 

 

0.90
0.94

 

0.91

 

 

 

59

{f/

 

 

A BL T A PG T A pa T
+57(ZNTV€ /§:N,j ) +§8(ZNRHA /ZNU)+69(ZNR__B /ZN,-j)
J ’1 J J ’1 J J ’1 J
A PG A BL A BL
C T A T B T
+510®NR11 /;NU)+511(;NRg /;N,.j)+512(;1v,,le /;Nij)

A BL P
c T
+513jSNRi-j /§j:1v,.j)+e,.j (5.1)

Estimation of Implicit Market Prices

Table 6 shows regression results where average price per unit (APPU) is a
dependent variable and average audience seconds per unit (AASPU) for program and
block ads in TV, and fractions of ad units in each daypart for TV and radio programs
comprising a package, are independent variables. All variables except fraction terms of
TV block advertising in daypart C, and radio block advertising in dayparts A, B, and C
were statistically significant at .05 level. Consistent with other literature, the estimate for
daypart SA had the highest value (KW 6,325,037, which as for the other dayparts is the
sum of the regression constant term and the daypart’s program ads coeficient) followed
by daypart A (KW 3,535,512), daypart C (KW 1,708,524), and daypart B (KW 516,695)
for program ad slots. As audience seconds increase, the value of a TV program ad slot
should also increase as much as KW 84 per audience second. On the other hand, daypart
indicators for block ads generally commanded negative values (KW -1,176,806 for
daypart SA, KW -2,117,404 for daypart A, KW -858,411 for daypart B, and KW 396,266
for daypart C). Negative coefficients for dayparts for block advertising should be
interpreted as indicating block advertising was valued less than was program advertising

even during the same dayparts.

60

 

Table 6. Regression Estimates for Average Package Price

 

 

Unstandardized

 

Standardized

 

 

(Constant)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Coefficients Standard _C_ocfficients Correlations Collinearity Statistics
B Error Beta t Sig. Zero—order Partial Part Tolerance VIF
516,695 112,693 4.58 0.00
Average Audience Seconds of TV Program Ads 84 6 0.40 14.82 0.00 0.94 0.58 0.14 0.12 8.66
Fraction of TV Program Ads in Daypart SA 5,808,342 221,365 0.41 26.24 0.00 0.80 0.79 0.24 0.35 2.86
Fraction of TV Program Ads in Daypart A 3.018.817 175,502 0.26 17.20 0.00 0.69 0.64 0.16 0.38 2.65
Fraction of TV Program Ads in Daypart C 1,191,829 161,627 0.08 7.37 0.00 0.17 0.34 0.07 0.65 1.53
Average Audience Seconds of TV Block Ads 102 15 0.24 6.77 0.00 0.09 0.31 0.06 0.07 15.00
Fraction of TV Block Ads in Daypart SA -1,693,501 672,775 —0.07 —2.52 0.01 0.13 —0.12 —0.02 0.12 8.56
Fraction of TV Block Ads in Daypart A -2,634,099 598,750 —0.09 —4.40 0.00 0.01 —0.21 —0.04 0.22 4.62
Fraction of TV Block Ads in Daypart B —1,375,106 368,221 —0.05 —3.73 0.00 —0.08 -0.18 —0.03 0.39 2.57
Fraction of TV Block Ads in Daypart C —l20,429 785,860 0.00 —0.15 0.88 —0.16 —0.01 0.00 0.66 1.53
Fraction of Religious Radio's Program Ads in Daypart A ~393,467 125,738 —0.06 —3.13 0.00 —0.62 —0. 15 —0.03 0.27 3.76
Fraction of Religious Radio's Program Ads in Daypart B -425,339 167,343 -0.03 —2.54 0.01 —0.39 -0.12 —0.02 0.55 1.80
Fraction of Religious Radio's Program Ads in Daypart C —1,215,291 629,873 —0.02 —1.93 0.05 -0.18 —0.09 —0.02 0.90 1.11
Fraction of Religious Radio's Block Ads in Daypart A -345,956 209,995 —0.02 -1.65 0.10 —0.29 —0.08 -0.02 0.54 1.86
Fraction of Religious Radio's Block Ads in Daypart B 476,343 714,969 —0.01 —0.67 0.51 —0.20 -0.03 —0.01 0.59 1.70
Fraction of Religious Radio's Block Ads in Daypart C 372,072 1,461,498 0.00 0.25 0.80 —0.1 1 0.01 0.00 0.63 1.59

 

Dependent Variable: APPU
R2 (15,427) = .96 (Sig. F Change = .00)

61

 

 

 

1161‘

ime

rais

TI)
(’9
(‘4

()1.
ofi

€XC

bet
AA
blo
reg
ad)
to 1

Edi

C0:

COr

da

(71..
r.

 

The marginal value per audience second of block advertising in major TV
networks was estimated higher (KW 102) than that for program ads. However,
interpretation of block advertising estimates requires special caution, as the VIF indicator
raised concerns regarding multicollinearity, especially for average block advertising
seconds in TV where its VIF value is 15, which is higher than the cutoff value of 10
(Montgomery et al., 2001). The coefficient of VIF is obtained when a predictor variable
of interest is regressed on the remaining predictor variables. Any variable whose VIF
exceeds 10 is recommended to be deleted, or an alternative method should be used
instead of OLS. However, it is important to diagnose where and why the problem arose
before the variable is excised. When other independent variables were regressed on
AASPU of TV block advertising, it was found that, as Table 7 shows, AASPU of TV
block advertising were affected mainly by block advertising dayparts. Hence, stepwise
regression was run to see the pure effect of daypart on the value of AASPU of TV block
advertising. Of 95% of the variation explained by the model, 56% was attributed purely
to its TV block advertising dayparts (see Table 8) while other independent variables
added only 0.3% in R2 (see Table 9). As noted above, it is natural for the audience size
of a program to covary with its daypart. As long as the covariation does not affect
coefficient estimates for other types of ad time, however, it seems safe to use these
coefficients for the price estimation.

The finding that daypart C has a larger coefficient than does daypart B in TV ad
time differs from the KOBACO formula where daypart B had a higher weight than did
daypart C. Although surprising, this may be a reflection of the fact that the programs in

daypart C attraCt a larger than average fraction of a particular demographic group valued

62

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7. Regression Estimates for Average Audience Seconds of TV Block Ads
Unstandardized Standard
Coefficients Standard Coefficients t Si0 Correlations Collinearity Statistics
B Error Beta Zero—order Partial Part Tolerance VIF

(Constant) —682 416 —1.64 0.10

Average Audience Seconds of TV Program Ads 0 0 0.18 5.02 0.00 —0.02 0.24 0.06 0.12 8.18
Fraction of TV Program Ads in Daypart SA —3,324 852 —0.10 —3.90 0.00 —0.01 —0.19 -0.05 0.24 4.09
Fraction of TV Program Ads in Daypart A —l,931 713 —0.07 —2.71 0.01 —0.09 —0.13 —0.03 0.24 4.23
Fraction of TV Program Ads in Daypart C —604 519 —0.03 —1 .16 0.25 —0.05 -0.06 —0.01 0.31 3.21
Fraction of TV Block Ads in Daypart SA 40,503 886 0.68 45.71 0.00 0.78 0.91 0.57 0.70 1.44
Fraction of TV Block Ads in Daypart A 33,293 1,064 0.46 31.29 0.00 0.63 0.83 0.39 0.71 1.41
Fraction of TV Block Ads in Daypart B 12,353 1,019 0.21 12.12 0.00 0.52 0.51 0.15 0.53 1.90
Fraction of TV Block Ads in Daypart C 3,416 2,547 0.02 1.34 0.18 0.25 0.06 0.02 0.65 1.55
Fraction of Religious Radio's Program Ads in Daypart A 545 457 0.03 1.19 0.23 —0.28 0.06 0.01 0.21 4.81
Fraction of Religious Radio's Program Ads in Daypart B 639 581 0.02 1.10 0.27 —0.21 0.05 0.01 0.47 2.11
Fraction of Religious Radio's Program Ads in Daypart C —1,404 2,023 —0.01 —0.69 0.49 —0.09 —0.03 —0.01 0.91 1.10
Fraction of Religious Radio's Block Ads in Daypart A 790 712 0.02 1.1 1 0.27 —0.08 0.05 0.01 0.48 2.07
Fraction of Religious Radio's Block Ads in Daypart B 2,390 2,322 0.02 1.03 0.30 —0.05 0.05 0.01 0.58 1.73
Fraction of Religious Radio‘s Block Ads in Daypart C —l ,635 4,690 —0.01 —0.35 0.73 —0.04 —0.02 0.00 0.63 1.58

 

 

Dependent Variable: AASPU of TV Block Ads
R2 (14,428) = .93 (Sig. F Change 2 .00)

63

 

 

 

 

Table 8. Hierarchical Regression of Average Audience Seconds of TV Block Ads: Daypart of TV block ads in the 2"d Stage

 

 

 

 

 

j (Constant)

 

 

 

 

 

 

 

 

 

 

 

 

 

_

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Unstandardized Standard Collinearity
Coefficients Standard Coefficients Correlations Statistics
B Error Beta t Sig. Zero—order Partial Part Tolerance VIF
12.765 718 17.77 0.00
Average Audience Seconds of TV Program Ads 0 0 —0.62 —6.28 0.00 —0.02 —0.29 -0.24 0.15 6.69
Fraction of TV Program Ads in Daypart SA 1,185 2,106 0.04 0.56 0.57 —0.01 0.03 0.02 0.37 2.68
Fraction of TV Program Ads in Daypart A —5,212 1,698 —0.19 —3.07 0.00 —0.09 —0.15 0.12 0.39 2.57
Fraction of TV Program Ads in Daypart C —10.379 1,449 -0.31 —7.16 0.00 —0.01 —0.33 -0.27 0.79 1.27
Fraction of Religious Radio's Program Ads in Daypart A -12,689 922 —0.76 -13.77 0.00 —0.28 -0.55 —0.52 0.48 2.09
Fraction of Religious Radio's Program Ads in Daypart B —13,324 1,410 -0.42 —9.45 0.00 —0.21 —0.41 —0.36 0.75 1.33
Fraction of Religious Radio's Program Ads in Daypart C —15,1 14 6,108 —0.10 —2.47 0.01 —0.09 -0.12 -0.09 0.93 1.08
Fraction of Religious Radio's Block Ads in Daypart A —1 1,627 1,918 —0.29 -6.06 0.00 —0.08 —0.28 —0.23 0.62 1.61
Fraction of Religious Radio's Block Ads in Daypart B —8,054 6,980 —0.06 —l.15 0.25 —0.05 —0.06 —0.04 0.60 1.68
_ Fraction of Religious Radio's Block Ads in Daypart C —17,356 14,303 —0.06 —1.21 0.23 —0.04 —0.06 -0.05 0.63 1.58
(Constant) —1,285 357 —3.60 0.00
Average Audience Seconds of TV Program Ads 0 0 0.18 5.02 0.00 —0.02 0.24 0.06 0.12 8.18
Fraction of TV Program Ads in Daypart SA —2,721 700 -0.08 —3.89 0.00 -0.01 -0.18 —0.05 0.36 2.76
Fraction of TV Program Ads in Daypart A —1,327 561 —0.05 -2.37 0.02 —0.09 —0.11 -0.03 0.38 2.61
Fraction of TV Program Ads in Daypart C 604 519 0.02 1.16 0.25 —0.01 0.06 0.01 0.66 1.52
Fraction of Religious Radio's Program Ads in Daypart A 1,149 401 0.07 2.87 0.00 -0.28 0.14 0.04 0.27 3.69
Fraction of Religious Radio‘s Program Ads in Daypart B 1,243 535 0.04 2.32 0.02 —0.21 0.11 0.03 0.56 1.78
Fraction of Religious Radio's Program Ads in Daypart C —800 2,025 —0.01 —0.40 0.69 —0.09 —0.02 0.00 0.90 1.11
Fraction of Religious Radio's Block Ads in Daypart A 1,394 672 0.04 2.08 0.04 —0.08 0.10 0.03 0.54 1.84
Fraction of Religious Radio's Block Ads in Daypart B 2,994 2,295 0.02 1.30 0.19 —0.05 0.06 0.02 0.59 1.69
Fraction of Religious Radio's Block Ads in Daypart C —1,031 4,700 0.00 —0.22 0.83 —0.04 —0.01 0.00 0.63 1.59
Fraction of TV Block Ads in Daypart SA 41,107 856 0.69 48.03 0.00 0.78 0.92 0.60 0.75 1.34
Fraction of TV Block Ads in Daypart A 33,897 1,011 0.47 33.52 0.00 0.63 0.85 0.42 0.78 1.27
Fraction of TV Block Ads in Daypart B 12,957 1,005 0.22 12.89 0.00 0.52 0.53 0.16 0.54 1.85
g Fraction of TV Block Ads in Daypart C 4,020 2,520 0.02 1.60 0.11 0.25 0.08 0.02 0.66 1.52

 

 

Dependent Variable: Average Audience Seconds of TV Block Ads

Model 1: R2(10,432) = .39 (Sig. F Change = .00)
Model 2: R2(4,428) = .93 (Sig. F Change 2 .00)

 

 

Table 9. Hierarchical Regression of Average Audience Seconds of TV Block Ads:

of TV Block Ads in
of TV Block Ads in
ofTV Block Ads in
ofTV Block Ads in

of TV Block Ads in

of TV Block Ads in
ofTV Block Ads in

of TV Block Ads in
Audience Seconds of TV
of TV Ads in
ofTV Ads in
of TV Ads in

Stand.
Coefficients
Beta

Unstandardized
Coefficients

B Std. Error
5
0.6

0.
0.2
0.0

Ads
SA
A
C

 

of Radio's Ads in
of Radio's Ads in
of Radio's Ads in
of Radio's Block Ads in A
of Radio's Block Ads in B
2 of Radio's Block Ads in C
Dependent Variable: Average Audience Seconds of TV Block Ads
Model 1: R2(4,438) = .93 (Sig. F Change = .00)

Model 2; R2 (10,428) = .93 (Sig. F Change = .00)

65

 

Other than TV block dyparts in the 2"d Stage

Correlations

Zero—order Partial Part Tolerance VIF

0. 0.

O\ D—I

 

 

 

by advertisers not taken into account by the formula used by KOBACO.

Once estimation was complete, implicit market prices for the individual ad times
and types were computed using the coefficient estimates presented in Table 6. As per
equations (3.2) and (3.3), coefficient estimates are the same for individual prices and
package prices. Hence, implicit market prices for individual ad times (and types) were
estimated by applying coefficients estimated with package prices.

Implicit Market Prices

Implicit market prices for program ads in the major TV networks were estimated
to be higher for dayparts SA, A, and C but lower for daypart B than KOBACO set prices.
The only exception was daypart A for KBS where the estimated market price was 0.96%
lower than KOBACO set prices. Block advertising time on KBS and MBC showed a
similar pattern: implicit market prices for dayparts SA, A, and B were estimated to be
lower than the KOBACO set prices while implicit market prices for daypart C were
estimated to be higher than KOBACO prices. For SBS, the implicit market prices for
dayparts SA, B, and C were estimated to be higher than the KOBACO set prices, the
implicit market prices for daypart A were estimated to be lower than the KOBACO set
prices. Details are in Table 10.1.

For religious broadcasters, implicit market prices for program advertising showed
two different patterns: in BBC-FM, CBS-AM, and PBC-FM, the implicit market prices
were estimated to be lower for all three dayparts while in the CBS-FM, FEBC-AM, and
WBS-FM, implicit market prices were estimated to be higher for dayparts A and B but
lower for daypart C. On the other hand, implicit market prices for block advertising were

estimated to be lower for daypart B while higher for daypart C. For daypart A, the

66

Table 10.1. Implicit Price Estimation for the Major TV Networks
(Unit: KW/Setrond)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads
Daypart C Daypart B Daypart A Daypart SA Total Daypart C Daypart B Daypart A Daypart SA Total
(1) Average Household Audience
Rating (%) 3.31 5.99 8.64 13.76 8.18 2.92 4.91 8.33 14.04 9.44
(2) Average Unit Price_KOBACO 150,323 200,244 380,804 633,235 347,777 43,442 77,567 206,600 368,306 227,217
KBSTV (3) Average Unit Price_Estimated 165,520 127,898 377,160 669,487 334,637 80,866 35,727 22,632 224,385 121,538
(4) Gap of Unit Price (3) - (2) 15,197 —72,346 —3,644 36,253 —13,139 37,424 —41,840 —183,968 —143,920 —105,680
(5) % of Gap (4)/(2) 10.11% —36.l3% —0.96% 5.72% —3.78% 86.15% —53.94% —89.05% —39.08% -46.51%
(1) Average Household Audience
Rating 2.76 6.03 9.76 19.39 8.95 2.54 5.32 9.56 13.20 7.75
(2) Average Unit Price_KOBACO 105,209 171,643 355,593 723,980 319,040 28,263 52,338 134,388 263,145 117,334
MBCTV (3) Average Unit Price_Estimated 156,079 127,584 402,083 780,430 346,632 70,758 41,166 44,222 204,229 82,787
(4) Gap Unit Price (3) — (2) 50,869 —44,059 46,490 56,450 27,592 42,495 —11,172 -90,166 —58,916 —34,547
(5) % Gap (4)/(2) 48.35% —25.67% 13.07% 7.80% 8.65% 150.36% —21.35% —67.09% —22.39% —29.44%
(1) Average Household Audience
Rating 2.91 5.34 9.73 12.55 7.66 1.98 4.20 8.44 10.79 6.34
(2) Average Unit Price_KOBACO 141,701 160,949 379,798 644,113 329,970 8,483 16,494 58,115 128,511 53,378
SBSTV (3 ) Average Unit Price_Estimated 159,607 1 18,665 403,531 645,673 329,100 61,118 21,335 15,574 153,518 64,955
(4) Gap of Unit Price (3) — (2) 17,906 -42,284 23,733 1,560 —871 52,634 4,841 —42,541 25,007 11,577
(5) % of Gap (4)/(2) 12.64% 26.27% 6.25% 0.24% —0.26% 620.43% 29.35% —73.20% 19.46% 21.69%

 

 

 

67

 

 

 

 

implicit market prices for block advertising in BBS-FM and PBS-FM were estimated to
be lower but higher in CBS-AM, CBS—FM, and FEBC-AM. There were no data for
dayparts A and C for WBS-FM. Details can be found in Table 10.2. It should be noted
that difference in audience size for radio programs was not taken into account in the
model. Audience size not taken into account in the model seems to cause inconsistency
between broadcasters.
Revenues estimated with Implicit Market Prices

Overall, revenues from program advertising in the major TV networks were
estimated to be larger at implicit market prices than at KOBACO set prices. The
difference accounted for 3.28% (KW 46,026 million) of the total major TV networks’
program advertising revenue covered by the study (KW 1,403,615 million). Meanwhile,
revenues at implicit market prices from block ads in the major TV networks were
estimated to be 25.94% lower than revenues earned at KOBACO prices (KW 95,634
million versus KW 70,828 million). In sum, it was estimated that revenues in the major
TV networks should be 1.42% larger at the market prices (KW 1,520,470 million) than at
the KOBACO prices (KW 1,499,249 million). On the other hand, religious broadcasters’
revenues estimated with implicit market prices were reduced by 36.36% (reduced to KW
29,338 million from the KW 46,103 million which was covered by the study). The
findings imply that religious broadcasters were subsidized by the major TV networks,
primarily by the program advertising in the major TV networks. Details are shown in

Table 11.1 for major TV networks and in Table 11.2 for religious broadcasters.

68

 

Table 10.2. Implicit Price Estimation for the Religious Broadcasters
(Unit: KW/Second)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads
Daypart C Daypart B a, Daypart A Total Daypart C Daypart B Daypart A Total

(1 ) Average Unit Price A": 3,948 5,504 9,585 8,614 3,220 5,540 9,032 7,819

(2) Average Unit Price 13* 34,930 4,568 6,161 5,118 44,438 2,018 8,537 8,261

BBC_FM (3) Gap of Unit Price (2) — (1) 38.878 —936 3,424 3,496 41,218 3,522 495 443
(4) % of Gap (3)/( 1) —984.64% 47.01% 35.72% 40.59% 1280.07% 63.58% 6.48% 5.67%

(1 ) Average Unit Price A* 6.768 13,426 21,008 19,007 7,733 7,831 6,381 6,620

(2) Average Unit Price B“ —34,930 4,568 6,161 5,155 44,438 2,018 8,537 8,301

CBS_AM (3) Gap of Unit Price (2) — (1) —4l,697 -8858 -14,847 —13,851 36,705 -5,813 2,156 1,681
(4) % of Gap (3 )/(1 ) 616.13% 65.98% -70.67% —72.88% 474.65% —74.24% 33.78% 25.39%

(1) Average Unit Price A* 817 2,088 3,853 3,529 1,135 2,795 3,782 3,134

(2) Average Unit Price B** -34,930 4,568 6,161 5,509 44,438 2,018 8,537 11,746

CBS—FM (3) Gap of Unit Price (2) — (1) 35.747 2,480 2,308 1,980 43,303 -778 4,755 8,612

(4) % of Gap (3)/(1) —4373.63% 118.74% 59 89% 56.10% 3815.27% —27.82% 125.73% 274.84%

(1) Average Unit Price A* 850 1,991 3,954 3,695 1,156 2,312 3,630 3,114

(2) Average Unit Price B** 34,930 4,568 6,161 5,848 44,438 2,018 8,537 8,676

FEBC—AM (3 ) Gap of Unit Price (2) — (1 ) 35,780 2,577 2,208 2,153 43,282 —294 4,907 5,563
(4) % of Gap (3)/(1) —4209.39% 129.47% 55.85% 58.28% 3743.32% 42.73% 135.19% 178.66%

(1 ) Average Unit Price A* 2,850 5,400 10,311 9,313 3,420 5,516 11,030 8,992

(2) Average Unit Price 13M 34,930 4,568 6,161 5,317 44,438 2,018 8,537 10,404

PBC—FM (3 ) Gap of Unit Price (2) — (1) -37,780 —832 4,150 3,996 41,018 -3,498 -2,493 1,412

(4) 0;, 0f Gap (3)/(1) —l325.61% 45.41% 40.25% 42.91% 1199.37% 63.42% —22.60% 15.70%

(1) Average Unit Price A* 3,405 3,450 6,034 5,057 3,176 3,176

(2) Average Unit Price B** 34,930 4,568 6,161 832 2,018 2,018

WES—FM (3) Gap of Unit Price (2) — (1) 38,335 1,118 128 4,225 —1,159 —1,159

(4) % of Gap (3)/(1) —1125.75% 32.40% 2.12% 83.55% 36.48% 36.48%

 

 

 

 

 

 

* Average Unit Price A was estimated with KOBACO set prices

 

 

 

 

 

 

 

** Average Unit Price B was estimated with implicit market prices for the programs provided only by the major TV and religious radios

69

 

 

 

 

 

Table 11.1. Revenue Estimation for the Major TV Networks
(Unit: KW, Million/Year)

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads
Daypalt C Daypart B Daypart A Daypart SA Total Daypart C Daypart B Daypart A Daypart SA Total SUM

(1) Revenues at Price A* 19,500 57,027 154,636 171,469 402,633 369 3,372 5,631 24,923 34,295 436,928

KBS (2) Revenues at Price BM 20,545 40,664 149,883 187,940 399,032 713 1,157 552 15,478 17,901 416,933
TV (3) Gap of Revenues (2) — (1) 1,045 —16,364 —4,753 16,471 —3,601 344 —2,215 —5,079 —9,445 —16,394 -19,995

(4) % Gap (3)/(1) 5.36% —28.69% —3.07% 9.61% —0.89% 93.26% -65.68% —90. 19% -37.90% —47.80% —4.58%

(1) Revenues at Price A* 30,545 64,894 195,296 245,834 536,568 1,472 7,562 12,683 24,275 45,992 582,560

MBC (2) Revenues at Price B“- 46,721 53,691 223,186 247,644 571,243 3,935 5,562 6,035 20,334 35,866 607,109
TV (3) Gap of Revenues (2) — (1) 16,177 -1 1,203 27,890 1,811 34,675 2,463 —2,000 —6,648 -3,941 -10,125 24,549

(4) % Gap (3)/(1) 52.96% -17.26% 14.28% 0.74% 6.46% 167.32% —26.45% —52.41% —16.23% —22.02% 4.21%

(1) Revenues at Price A* 31,929 62,726 144,163 225,596 464,414 184 1.159 3,334 10,670 15,347 479,761

SBS (2) Revenues at Price B** 37,795 52,587 157,338 231,646 479,367 1,388 1,075 763 13,836 17,061 496,428
TV (3) Gap of Revenues (2) — (1) 5,866 —10,138 13,174 6,050 14,952 1,204 ~85 2,571 3,166 1,714 16,667

(4) % Gap (3)/(1) 18.37% —16.16% 9.14% 2.68% 3.22% 654.91% —7.31% —77.1 1% 29.67% 11.17% 3.47%

(1) Revenues at Price A* 81,974 184,647 494,096 642,899 1,403,615 2,025 12,093 21,648 59,868 95,634 1,499,249

(2) Revenues at Price B** 105,062 146,942 530,406 667,231 1,449,641 6,036 7,793 7,351 49,648 70,828 1,520,470

SUM (3) Gap of Revenues (2) — (1) 23,088 —37,705 36,31 1 24,332 46,026 4,01 1 —4,299 —14,297 —10,220 —24,805 21,221
(4) % Gap (3)/(1) 28.17% —20.42% 7.35% 3.78% 3.28% 198.08% —35.55% —66.04% —17.07% -25.94% 1.42%

 

 

 

 

 

 

* Price A is KOBACO set prices
** Price B is the implicit market prices estimated with the model including only the major TV and religious radios

70

 

 

 

 

Table 11.2. Revenue Estimation for Religious Broadcasters
(Unit: KW, Million/Year)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads

Daypart C Daypart B Daypart A Total Daypart C Daypart B Daypart A Total SUM
(1) Revenues at Price A* 75 1,266 6,395 7,736 17 93 466 576 8,311
(2) Revenues at Price B** —624 1,024 4,215 4,614 230 26 383 638 5,253
BBS—FM (3 ) Gap of Revenues (2) — (1) 699 —242 —2,180 3,121 213 67 —83 62 3,059
(4) % Gap (3)/(1) —931.91% 49.13% 34.09% 40.35% 1280.07% 42.13% 47.88% 10.85% 36.80%
(1) Revenues at Price A* 23 3,272 16,183 19,479 152 796 3,281 4,229 23,708
(2) Revenues at Price 8“ -120 1,124 4,734 5,738 976 166 4,095 5,237 10,975
CBSJAM (3 ) Gap of Revenues (2) — (1 ) 443 —2,148 41,450 43,741 823 630 815 1,008 42,733
(4) % Gap (3 )/(1 ) 617.23% 65.65% 40.75% 30.54% 540.22% —79.13% 24.84% 23.84% —53.71%
(1) Revenues at Price A* 8 237 1,687 1,932 24 83 208 315 2,247
(2 ) Revenues at Price B** 320 541 2,859 3,080 904 62 445 1,411 4,491
CBS—FM (3 ) Gap of Revenues (2) — (1 ) 328 303 1,172 1,148 879 —21 238 1,096 2,244
(4) % Gap (3 )/(1) 4224.97% 127.99% 69.50% 59.43% 3592.48% —25.16% 114.41% 347.37% 99.86%
(1) Revenues at Price A* 0 221 1,811 2,033 23 78 337 438 2,471
(2) Revenues at Price B** —14 507 2,884 3,377 811 60 758 1,630 5,007
FEBC—AM (3 ) Gap of Revenues (2) — (1 ) 44 286 1,073 1,345 789 48 421 1,192 2,536
(4) % Gap (3)/(1) —4209.39% 128.99% 59.26% 66.14% 3470.74% -23.11% 124.87% 271.90% 102.63%
(1) Revenues at Price A* 51 923 5,724 6,698 13 71 323 406 7,104
(2) Revenues at Price BM" —621 780 3,420 3,579 163 22 250 435 4,014
PBC—F M (3 ) Gap of Revenues (2) — (1 ) 672 442 —2,305 3,119 150 48 —73 29 3,090
(4) % Gap (3)/(1) 4325.61% 45.41% 40.26% 46.56% 1199.37% 68.26% -22.60% 7.09% 43.49%
(1) Revenues at Price A* 247 414 1,570 2,231 30 30 2,262
(2) Revenues at Price B“ -2,582 549 1,603 —431 29 29 —401
WES—FM (3 ) Gap of Revenues (2) — (1) -2,829 134 33 —2,662 —1 —1 —2.663
(4) % Gap (3)/(1) 4143.54% 32.40% 2.10% 419.30% 3 01% 3.01% 417.75%
(1) Revenues at Price A* 404 6,333 33,370 40,108 229 1,152 4,614 5,995 46,103
(2) Revenues at Price B** 4,281 4,524 19,714 19,957 3,084 367 5,931 9,381 29,338
SUM (3) Gap of Revenues (2) — (1) 4,685 4,809 43,656 —20,151 2,855 —786 1,317 3,386 46,764
(4) % Gap (3)/(1) 4158.56% —28.57% 40.92% -50.24% 1247.75% —68.17% 28.54% 56.48% —36.36%

 

 

;
* Price A is KOBACO set prices

** Price B is the implicit market prices estimated with the model including only the major TV and religious radios

71

 

 

 

However, as the revenues covered by the study are 64.3% of the total market size,
the estimate of the total subsidy amounts has to be adjusted to reflect ad time purchases
by advertisers not included in the sample. The total revenues predicted for increase in the
major TV networks is KW 33,003 million at implicit market prices while the total

revenue reduction predicted for the religious broadcasters is KW 26,072 million.

Further Analysis

Package Size Eflect

To design a model based on the assumption that the package price is the sum of
components' market prices, no package discounts should be available. Otherwise, any
package discount becomes part of the error term. The concern might be that this violates
the independence of errors if discounts are not distributed randomly among independent
variables. Hence, it is necessary to check that there are no explicit package discounting
practices in the Korean advertising market. While there is no evidence that KOBACO
provides higher discounts for larger packages, it is possible for KOBACO to construct
packages in a way to favor bigger packages (i.e., constructing packages with better
performing programs for bigger packages). If it does, a package size variable should
affect average package price per spot negatively. Package size effect was tested by
including a package size dummy (1 for those packages with higher than average package
price, 0 for the others) into the regression where the dependent variable is average
package price and independent variables are average audience seconds per spot and the
fractions of each daypart for the package. Table 12.1 shows that the coefficient for the

package size dummy is negative but not statistically significant (significance p=. 39).

72

Also, as Table 12.2 shows, the effect of package size on the price per audience second
(equivalent to CPM) was checked with the regression where price per audience second is
a dependent variable and package price is an independent variable. Again, the correlation
was close to zero and its estimator was insignificant (significance p=.93). The results
provide evidence for no package discounts available in the market.

Efi’ect of Demographic Composition

Although audience size variables specified by demographic characteristics were
dropped from the model estimating implicit market prices due to multicollinearity
concern, it is still possible to check the effect of demographic composition by adding a
proportional term for gender in the analysis. However, data did not include gender
fraction across all demographics. Only the 20-49 age groups are divided into gender.
Hence, assuming audiences other than 20-49 age groups are also composed in the same
gender proportions, the fraction variable of female 20-49 audience seconds out of total
20-49 audience seconds were added to the model (3.5).

As Table 13 shows, the female fraction has a negative effect (KW -1,873,068) on
the average package price. This implies that the advertising slot in a program whose
audience contains only females, commands a KW 1,873,068 lower price than one whose
audience has only males. That is, advertisers prefer male viewers to female viewers at
least among those aged 20-49. This finding is consistent with previous finding by
Fournier and Martin (1983) and Koschat and Putsis (2002) where female viewers were

less valued in terms of advertising price.

73

 

TablelZJ. Package Size Effect on APPU

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Unstandardized Standardized Collinearity
Coefficients Standard ifoefficients Correlations Statistics
B Error Beta t Sig. Zero—order Partial Part Tolerance VIF

(Constant) 1,706,] 14 129,925 13.13 0.00

Average Audience Seconds of TV Program Ads 84 6 0.40 14.81 0.00 0.94 0.58 0.14 0.11 8.80
Fraction of TV Program Ads in Daypart SA 4,601,530 270,168 0.32 17.03 0.00 0.80 0.64 0.16 0.24 4.25
Fraction of TV Program Ads in Daypart A 1,818,243 223,918 0.15 8.12 0.00 0.69 0.37 0.07 0.23 4.31
Fraction of TV Program Ads in Daypart C —1,194,645 161,71 1 —0.12 —7.39 0.00 0.14 —0.34 -0.07 0.31 3.22
Average Audience Seconds of TV Block Ads 102 15 0.24 6.77 0.00 0.09 0.31 0.06 0.07 15.00
Fraction of TV Block Ads in Daypart SA —2,889,073 668,496 —0.12 —4.32 0.00 0.13 —0.20 —0.04 0.12 8.45
Fraction of TV Block Ads in Daypart A —3,824,825 600,1 15 —0.13 —6.37 0.00 0.01 -0.30 —0.06 0.22 4.64
Fraction of TV Block Ads in Daypart B -2,567,538 367,455 —0. 10 —6.99 0.00 —0.08 —0.32 —0.06 0.39 2.56
Fraction of TV Block Ads in Daypart C —1,264,626 795,754 —0.02 —1.59 0.11 —0. 16 —0.08 -0.01 0.64 1.56
Fraction of Religious Radio’s Program Ads in Daypart A —l,571,981 143,339 —0.22 -10.97 0.00 —0.62 —0.47 -0.10 0.20 4.88
Fraction of Religious Radio’s Program Ads in Daypart B —1,612,818 181,181 —0.12 —8.90 0.00 —0.39 -0.40 —0.08 0.47 2.11
Fraction of Religious Radio’s Program Ads in Daypart C —2,314,537 638,997 —0.04 —3.62 0.00 -0. 18 —0.17 -0.03 0.88 1.14
Fraction of Religious Radio’s Block Ads in Daypart A —1,520,729 222,675 —0.09 -6.83 0.00 —0.29 -0.31 —0.06 0.48 2.09
Fraction of Religious Radio’s Block Ads in Daypart B —1,599,893 727,667 —0.03 —2.20 0.03 —0.20 —0.11 -0.02 0.57 1.76
Fraction of Religious Radio’s Block Ads in Daypart C —734,397 1,462,598 —0.01 —O.50 0.62 —0.11 —0.02 0.00 0.63 1.59
Package Size Dummy (1 for lager than Average) —25,349 29,629 —0.01 —0.86 0.39 —0. 12 —0.04 —0.01 0.88 1.13
Dependent Variable: APPU

R2 (16,426) = .96 (Sig. F Change = .00)

Table12.2. Package Size Effect on Average Price per Audience Second
Unstandarized Standardized Collinearity
Coefficients Standard Coefficients Correlations Statistics
B Error Beta t Sig. Zero—order Partial Part Tolerance VIF

(Constant) 185 2 83.88 0.00

Total Package Price 0 0 0.00 -0.09 0.93 0.00 0.00 0.00 1.00 1.00

 

 

 

 

Dependent Variable: Average Price per Audience Second
R2 (1,436) = .00 (Sig. F Change = .93)

74

 

 

 

Tablel3. Demographic Composition Effect

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5 Unstandardized Standardized
Coefficients Standard Coefficients Correlations Collinearity Statistics
B Error Beta t Sig. Zero—order Partial Part Tolerance VIF

(Constant) 1,819,948 382,987 4.75 0.00

Average Audience Seconds of TV Program Ads 81 6 0.39 14.41 0.00 0.94 0.57 0.13 0.12 8.33
Fraction of TV Program Ads in Daypart SA 5,685,741 222,104 0.40 25.60 0.00 0.79 0.78 0.24 0.35 2.87
Fraction of TV Program Ads in Daypart A 2,930,960 175,671 0.25 16.68 0.00 0.68 0.63 0.16 0.38 2.63
Fraction of TV Program Ads in Daypart C 874,591 183,218 0.06 4.77 0.00 0.15 0.23 0.04 0.51 1.98
Average Audience Seconds of TV Block Ads 97 15 0.23 6.47 0.00 0.07 0.30 0.06 0.07 15.03
Fraction of TV Block Ads in Daypart SA —1,855,840 668,620 —0.08 —2.78 0.01 0.11 —0. 13 —0.03 0.12 8.57
Fraction of TV Block Ads in Daypait A —2,552,323 593,766 —0.09 -4.30 0.00 0.00 -0.21 -0.04 0.22 4.61
Fraction of TV Block Ads in Daypart B —1,261,480 366,584 —0.05 —3.44 0.00 —0.10 —0. 17 -0.03 0.39 2.58
Fraction of TV Block Ads in Daypart C —269,645 780,598 0.00 -0.35 0.73 —0. 18 —0.02 0.00 0.65 1.53
Fraction of Religious Radio’s Program Ads in Daypart A —559,153 133,515 —0.08 —4.19 0.00 —0.61 -0.20 —0.04 0.25 3.96
Fraction of Religious Radio’s Program Ads in Daypart B —610,3 10 173,633 -0.05 —3.51 0.00 -0.40 -0.17 —0.03 0.51 1.95
Fraction of Religious Radio’s Program Ads in Daypart C —1,326,924 626,921 —0.02 —2.12 0.03 —0.19 —0.10 —0.02 0.90 1.12
Fraction of Religious Radio’s Block Ads in Daypart A ~341,038 270,631 —0.02 —1.26 0.21 -0.27 —0.06 -0.01 0.53 1.90
Fraction of Religious Radio’s Block Ads in Daypart B —723,578 724,382 -0.01 —l.00 0.32 -0.19 —0.05 -0.01 0.58 1.71
Fraction of Religious Radio’s Block Ads in Daypart C —188,744 1,554,273 0.00 —0. 12 0.90 -0.11 —0.01 0.00 0.55 1.83
Female Percentage in the Audience —1,873,068 527,127 0.04 -3.55 0.00 —0.20 -0. 17 —0.03 0.67 1.50

 

 

 

Dependent Variable: APPU
R2 (16,421) = .96 (Sig. F Change 2 .00)

75

 

 

Selling Major Radio Time in Packages with Major TV and Religious Radio Time

Although previous literature mainly focuses on major TV and religious
broadcasters in the context of price distortion attributed to packaging or other factors
such as lack of information, it is also meaninng to test how well the other advertising
resources sold by KOBACO approximate implicit market prices in the framework of
package price. However, this approach has both pros and cons. If they were sold as part
of packages, including them in the packages fixes the problem regarding model
misspecification. Conversely, if not, including them in the packages could cause
additional bias when there is correlation between the number of ad units for other media
programs and the number of ad units for major radio programs. Hence, results should be
interpreted cautiously.

Table 14 shows regression results where the program and block advertising for

major radio broadcasters (major TV networks’ radio subsidiaries) was added to the model

(3.5). By this addition, the variation explained by the model (R2) increases by .02

from .96 to .98 and major radio’s total ad revenues estimated at implicit market prices
remained about the same (-0.76%) although the budget estimation for the major TV
networks decreased by a little (0.02%) compared to the model without major radio. By
including major radio in the model, the revenues for the religious broadcasters were
estimated to be KW 21,977 million, which is a 52.33% reduction from KW 46,103
million at KOBACO prices. The revenues estimated at the implicit market prices with the
model before major radio was added to the model were KW 29,338, which is a 36.36%

reduction from the corresponding revenues. at KOBACO prices. Tables 15.1 to 15.3 show

76

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 14. Regression Output including Major Radio
Unstandardized Standard
Coefficients Standard _floefficients Correlations Collinearity Statistics
B Error Beta t Sig. Zero-order Partial Part Tolerance VIF

(Constant) 589,514 1 11,272 5.30 0.00

Average Audience Seconds of TV Program Ads 84 6 0.40 15.01 0.00 0.96 0.59 0.11 0.08 12.87
Fraction of TV Program Ads in Daypart SA 5,692,204 220,406 0.36 25.83 0.00 0.82 0.78 0.19 0.29 3.40
Fraction of TV Program Ads in Daypart A 2,859,017 168,161 0.23 17.00 0.00 0.74 0.64 0.13 0.32 3.17
Fraction of TV Program Ads in Daypart C 1,096,662 152,352 0.07 7.20 0.00 0.37 0.33 0.05 0.54 1.86
Average Audience Seconds of TV Block Ads 88 16 0.17 5.57 0.00 0.25 0.26 0.04 0.06 15.82
Fraction of TV Block Ads in Daypart SA -1,216,548 766,459 -0.03 —1.59 0.11 0.28 -0.08 001 0.13 7.72
Fraction of TV Block Ads in Daypart A —2,369,850 614,693 -0.07 —3.86 0.00 0.15 —0.18 p —0.03 0.19 5.38
Fraction of TV Block Ads in Daypart B —882,507 373,618 —0.03 —2.36 0.02 0.06 0.11 —0.02 0.34 2.91
Fraction of TV Block Ads in Daypart C —853,774 789,265 —0.01 -1.08 0.28 —0.05 —0.05 —0.01 0.57 1.76
FraCtion of Major Radio's Program Ads in Daypart A —347,043 128,783 —0.04 —2.69 0.01 0.38 -0.13 —0.02 0.24 4.14
Fraction of Major Radio’s Program Ads in Daypart B -241,385 170,498 —0.01 —1.42 0.16 -0.24 —0.07 —0.01 0.56 1.79
Fraction of Major Radio's Program Ads in Daypart C —875,815 355,874 -0.02 -2.46 0.01 —0.19 —O.12 —0.02 0.79 1.26
Fraction of Major Radio's Block Ads in Daypart A -366,279 207,633 __ —0.02 —1.76 0.08 —0.18 —0.09 —0.01 0.55 1.82
Fraction of Major Radio's Block Ads in Daypart B —372,642 462,202 __ —0.01 -0.81 0.42 -0.11 —0.04 -0.01 0.78 1.28
Fraction of Major Radio's Block Ads in Daypart C —434,901 782,373 0.00 —0.56 0.58 —0.10 —0.03 0.00 0.76 1.31
Fraction of Religious Radio's Program Ads in Daypart A -506,061 125,443 -0.07 —4.03 0.00 —0.53 —0.19 —0.03 0.21 4.72
Fraction of Religious Radio's Program Ads in Daypart B -581,535 167,194 —0.04 —3.48 0.00 —0.34 —0.17 —0.03 0.55 1.83
Fraction of Religious Radio's Program Ads in Daypart C —1,104,008 603,632 —0.01 —1.83 0.07 —O.17 —0.09 —0.01 0.90 1.11
Fraction of Religious Radio‘s Block Ads in Daypart A —286,847 248,613 —0.01 -1.15 0.25 —0.23 —0.06 —0.01 0.48 2.09
Fraction of Religious Radio's Block Ads in Daypart B —832,808 721,777 -0.01 —1.15 0.25 —0.23 —0.06 -0.01 0.49 2.02
WM;— 298,793 1,598,865 0.00 0.19 0.85 —0.08 0.01 0.00 0.45 2.21
Dependent Variable: APPU

R2 (21,421) = .98 (Sig. F Change = .00)

77

 

 

 

 

 

 

Table15.l. Price Estimation for the Major TV Including Major Radio
(Unit: KW/Second)

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads
Daypart C Daypart B Daypart A Daypart SA Total Daypart C Daypart B Daypart A Daypart SA Total

(1) Average Household Audience Rating 3.31 5.99 8.64 13.76 8.18 2.92 4.91 8.33 14.04 9.44

(2) Average Unit Price A* 150,323 200,244 380,804 633,235 347,777 43,442 77,567 206,600 368,306 227,217

(3) Average Unit Price B** 165,520 127,898 377,160 669,487 334,637 80,866 35,727 22,632 224,385 121,538

KBS (4) Average Unit Price C*** 164,249 133,148 371,961 667,649 333,673 29,497 60,900 23,022 220,243 124,137
TV (5) Gap of Unit Price (3) - (2) 15,197 —72,346 -3,644 36,253 —13,139 37,424 —41,840 —183,968 —143,920 —105,680
(6) Gap of Unit Price (4) - (2) 13,926 —67,096 —8,843 34,415 —14,104 —13,946 -16,667 —183,578 —148,063 —103,081

(7) % Gap (5)/(2) 10.11% —36.13% -O.96% 5.72% -3.78% 86.15% —53.94% -89.05% 69.08% —46.51%

(8) % Gap (6)/(2) 9.26% —33.51% —2.32% 5.43% —4.06% —32.10% -21.49% 88.86% —40.20% -45.37%

(1) Average Household Audience Rating 2.76 6.03 9.76 19.39 8.95 2.54 5.32 9.56 13.20 7.75

(2) Average Unit Price A* 105,209 171,643 355,593 723,980 319,040 28,263 52,338 134,388 263,145 117,334

(3) Average Unit Price B** 156,079 127,584 402,083 780,430 346,632 70,758 41,166 44,222 204,229 82,787

MBC (4) Average Unit Price C*** 154,767 132,833 396,989 779,063 345,631 20,750 65,606 41,721 202,801 80,282
TV (5) Gap of Unit Price (3) — (2) 50,869 —44,059 46,490 56,450 27,592 42,495 —11,172 —90,166 —58,916 —34,547
(6) Gap of Unit Price (4) — (2) 49,558 —38,810 41,396 55,082 26,591 —7,513 13,268 -92,667 —60,343 -37,053

(7) % Gap (5)/(2) 48.35% —25.67% 13.07% 7.80% 8.65% 150.36% -21.35% —67.09% —22.39% —29.44%

(8) % Gap (6)/(2) 47.10% —22.61% 11.64% 7.61% 8.33% —26.58% 25.35% —68.95% -22.93% —31.58%

(1) Average Household Audience Rating 2.91 5.34 9.73 12.55 7.66 1.98 4.20 8.44 10.79 6.34

(2) Average Unit Price A* 141,701 160,949 379,798 644,113 329,970 8,483 16,494 58,115 128,511 53,378

(3) Average Unit Price B** 159,607 118,665 403,531 645,673 329,100 61,118 21,335 15,574 153,518 64,955

SBS (4) Average Unit Price C*** 158,311 123,877 398,443 643,734 328,438 12,408 48,447 16,932 158,922 68,282
TV (5) Gap of Unit Price (3) - (2) 17,906 —42,284 23,733 1,560 -871 52,634 4,841 —42,541 25,007 11,577
(6) Gap of Unit Price (4) - (2) 16,610 -37,072 18,645 —379 —1,533 3,925 31,953 -41,183 30,411 14,904

(7) % Gap (5)/(2) 12.64% —26.27% 6.25% 0.24% —0.26% 620.43% 29.35% —73.20% 19.46% 21.69%

(8) % Gap (6)/(2) 11.72% -23.03% 4.91% -0.06% —0.46% 46.26% 193.72% —70.86% 23.66% 27.92%

 

 

 

 

 

* Average Unit Price A was estimated with KOBACO set prices . _ . .
** Average Unit Price B was estimated with implicit market prices for the programs provided only by the major TV and religious radios

*** Average Unit Price C was estimated with implicit market prices for the programs provided by the major TV, religious radios, and major radios

 

 

 

Table15.2. Price Estimation for the Religious Broadcasters Including Major Radio

(Unit: KW/Second)

 

BBS_FM

 

 

 

 

 

 

CBS_AM

 

 

CBS_FM

 

FEBC~AM

 

PBC_FM

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads
Daypart C Daypart B Da ypart A V Total Daypart C Daypart B Daypart A Total
(1) Average Unit Price Ai‘< 3,948 5,504 9,585 8,614 3,220 5,540 9,032 7,819
(2) Average Unit Price B“ —34,930 4,568 6,161 5,118 44,438 2,018 8,537 8,261
(3) Average Unit Price C*’* —25,725 399 4,173 2,853 44,415 —12,165 15,133 8,829
(4) Gap of Unit Price (3) — (2) —38.878 —936 -3,424 —3,496 41,218 —3,522 —495 443
(5) Gap of Unit Price (4) — (2) —29,673 —5,105 -5,412 —5,761 41,195 -17,704 6,102 1,010
(6) % Gap (4)/( 1) —984.64% —17 01% -35.72% —40.59% 1280.07% —63.58% -5.48% 5.67%
(7) % Gap (5)/( 1) —751.51% —92.75% 66.47% —66.88% 1279.36% —319.60% 67.56% 12.92%
(1) Average Unit Price A* 6,768 13,426 21,008 19,007 7,733 7,831 6,381 6,620
(2) Average Unit Price BM —34,930 4,568 6,161 5,155 44,438 2,018 8,537 8,301
(3) Average Unit Price CW -25,725 399 4,173 2,826 44,415 —12,165 15,133 11,725
(4) Gap of Unit Price (3) — (2) —41,697 -8,858 — 14,847 —13,851 36,705 —5,813 2,156 1,681
(5) Gap of Unit Price (4) — (2) —32,492 —13,027 46,835 —16,181 36,682 —19,996 8,752 5,104
(6) % Gap (4)/(1) -616.13% 65.98% —70.67% —72.88% 474.65% ~74.24% 33.78% 25.39%
(7) % Gap (5)/(1) —480.1 1% —97.03% —80.14% -85.13% 474.35% —255.34% 137.15% 77.10%
(1) Average Unit Price A* 817 2,088 3,853 3,529 1,135 2,795 3,782 3,134
(2) Average Unit Price B** —34,930 4,568 6,161 5,509 44,438 2,018 8,537 11,746
(3) Average Unit Price C*** —25,725 399 4,173 3,260 44,415 —12,165 15,133 11,580
(4) Gap of Unit Price (3) — (2) —35,747 2,480 2,308 1,980 43,303 -778 4,755 8,612
(5) Gap of Unit Price (4) — (2) —26,542 —1,689 319 —269 43,280 —14,960 11,351 8,447
(6) % Gap (4)/(1) -4373.63% 118.74% 59.89% 56.10% 3815.27% —27.82% 125.73% 274.84%
(7) % Gap (5)/(1) —3247.39% —80.90% _ 8.28% —7.62% 3813.25% -535.18% 300.14% 269.57%
(1) Average Unit Price A* 850 1,991 3,954 3,695 1,156 2,312 3,630 3,114
(2) Average Unit Price B** —34,930 4,568 6,161 5,848 44,438 2,018 8,537 8,676
(3) Average Unit Price C*** -25,725 399 4,173 3,611 44,415 -l2,165 15,133 8,945
(4) Gap of Unit Price (3) - (2) —35,780 2,577 2,208 2,153 43,282 —294 4,907 5,563
(5) Gap of Unit Price (4) — (2) —26,575 —1,592 219 —84 43,259 —14,477 11,504 5,831
(6) % Gap (4)/(1) -4209.39% 129.47% 55.85% 58.28% 3743.32% —12.73% 135.19% 178.66%
(7) % Gap (5)/(1) -3126.44% —79.96% _ 5.54% —2.26% 3741.33% —626.19% 316.92% 187.28%
(1) Average Unit Price A" 2,850 5,400 10,31 1 9,313 3,420 5,516 11,030 8,992
(2) Average Unit Price B** —34,930 4,568 6,161 5,317 44,438 2,018 8,537 10,404
(3) Average Unit Price C*** —25,725 399 4,173 3,080 44,415 -12,165 15,133 11,423
(4) Gap of Unit Price (3) — (2) -37,780 —832 —4,150 —3,996 41,018 -3,498 —2,493 1,412
(5) Gap of Unit Price (4) — (2) —28,575 —5,001 —6,139 —6,233 40,995 —17,681 4,103 2,430
(6) % Gap (4)/(1) —1325.61% —15.41% —40.25% -42.91% 1199.37% 63.42% —22.60% 15.70%
(7) % Gap (5)/(1) Mw#—5953% —66.93% M -320.54% 37.20% 27.02%

 

3“

 

 

 

Table15.2. (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(1) Average Unit Price A* 3,405 3,450 6,034 5,057 3,176 3,176
(2) Average Unit Price BM —34,930 4,568 6,161 832 2,018 2,018
(3) Average Unit Price C*** —25,725 399 4,173 —374 -12,165 —12,165
WBS_FM (4) Gap of Unit Price (3) - (2) —38,335 1,118 128 —4,225 —1,159 4,159
(5) Gap of Unit Price (4) — (2) -29,130 —3,051 —1,861 —5,432 -15,341 45,341
(6) % Gap (4)/(1) —1125.75% 32.40% 2.12% —83.55% —36.48% -36.48%
(7) % Gap (5)/(1) -855.43% —88.44% —30.85% ~107.40% -482.99% —482.99%
* Average Unit Price A was estimated with KOBACO set prices
** Average Unit Price B was estimated with implicit market prices for the programs provided only by the major TV and religious radios
*** Average Unit Price C was estimated with implicit market prices for the programs provided by the major TV, religious radios, and major radios
Table15.3. Price Estimation for the Major Radio Broadcasters Including Major Radio
(Unit: KW/Second)
Program Ads Block Ads
Daxpart C Da art B Daypart A Total Dagpart C Daypart B Daypart A Total
(1) Average Unit Price A* 6,530 12,923 __. 20,057 17,947 3,601 5,739 10,640 8,828
(2) Average Unit Price C** —14,315 17,406 _fl_’ 12,124 12,605 7,731 10,844 11,162 10,949
KBS—AM (3) Gap of Unit Price (2) — (1) —20,845 4,484 __ -7933 —5.343 4,130 5,105 522 2,120
(4) % Gap (3)/(1) —319.23% 34.70% _ 89.55% -29.77% 114.68% 88.96% 4.90% 24.02%
(1) Average Unit Price A* 985 3,100 __ 5,611 4,752 1,478 3,014 3,852 3,315
(2) Average Unit Price C** —14,315 17,406 __ 12,124 10,884 7,731 10,844 11,162 10,557
KBS—FM (3) Gap of Unit Price (2) — (1) —15,301 14,306 _fl_ 6,513 6.131 6,253 7.830 7,310 7,241
(4) % Gap (3)/(1) —1552.64% 461.50% 116.07% 129.02% 423.05% 259.80% 189.78% 218.41%
(1) Average Unit Price A* 9,600 20,881 _ 27,255 24,916 12,835 15,193 14,972 14,766
(2) Average Unit Price C** —14,315 17,406 _ 12,124 13,017 7,731 10,844 11,162 10,675
MBC—AM (3) Gap of Unit Price (2) _ (1) 23,915 —3,474 _15,132 —11,900 —5,104 4,350 -3,810 4,091
(4) % Gap (3)/(1) -249.12% —16.64% _ —55.52% 47.76% -39.77% -28.63% -25.45% —27.70%
(1) Average Unit Price A* 9,870 15,581 _, 17,397 16,107 10,081 14,213 9,022 9,827
(2) Average Unit Price C** —14,315 17,406 _ 12,124 10,666 7,731 10,844 11,162 10,744
MBC—FM (3) Gap of Unit Price (2) - (1) -24,185 1,825 _~ 5274 —5,442 2,350 -3,370 2,140 917
(4) % Gap (3)/(1) —245.04% 11.71% __ —30.31% —33.78% -23.31% -23.71% 23.71% 9.33%
(1) Average Unit Price A* 1,200 2,767 _ 6,055 4,921 1,487 3,016 5,109 4,263
(2) Average Unit Price C** -14,315 17,406 ___ 12,124 11,586 7,731 10,844 11,162 10,809
SBS—AM (3) Gap of Unit Price (2) — (1) —15,515 14,639 6,069 6,665 6,243 7,828 6,053 6,546
(4) % Gap (3)/(1) W 529.00% _ 100.24% 135.45% 419.77% 259.56% 118.49% 153.55%
MEAL—’— 2,906 4,785 _ 8,250 7,076 1,292 2,301 3,580 3,061
MW —14,315 17,406 ___ 12,124 11,010 7,731 10,844 11,162 10,822
SBS_FM (3) Ga of Unit Price (2) _ (1) 47,221 12,622 _ 3,874 3,934 6,439 8,543 7,582 7,761
(4) % Ga (3)/(1) “MM—MW 498.28% 371.30% 211.82% 253.54%
* Average Unit Price A was estimated with KOBACO set prices
**Average Unit Price C was estimated with implicit market prices for the programs provided by the major TV, religious radios, and major radios {
1
1.

80

 

price estimation results for major TV networks, religious broadcasters, and major radio
broadcasters respectively, while Tables 16.1 to 16.3 show revenue estimates.
Significance of Diflerence Between KOBA C 0 Prices and Implicit Market Prices

Tables 17.1 to 17.4 show the results of testing for thesignificance of differences
between KOBACO prices and implicit market prices. Due to the large number of cases,
most tests were conducted twice with total cases and randomly sampled cases with
around 200 cases in the test as a large number of cases tends to render non-significant
differences significant. In some dayparts whose number of cases in the data was less than
200, sampled tests were not conducted. For the major TV networks, the differences
between KOBACO prices and implicit market prices were statistically significant except
for KBS’s daypart A program ads, SBS’s daypart SA program ads, and SBS’s daypart B
block ads. For the religious radio broadcasters, implicit market prices for both program
ads and block ads whose sample sizes were larger than 100 were statistically different
from KOBACO prices in all the dayparts. However, daypart C for religious broadcasters’
block ads contained too few cases to have statistical meaning.
Comparison of present findings to those of the other studies

Previous studies have attempted to estimate implicit market prices for broadcast
ad time in two ways: comparing ratios of TV ad CPMs to newspaper ad CPMs in Korea
and in other countries (Joo, et al., 2001; Park, 2004; Shin, 2002), or measuring the
advertisers’ willingness to pay for broadcast ad time (Jung, 2005). While the studies
comparing CPMs found that overall 1evels of TV ad time prices in a competitive market
ranged up to 200% (Shin, 2002) to 400% (Park, 2004) of the current prices, Jung (2005)

estimated implicit market prices to be 114.8% of the current prices. Jung’s (2005)

81

 

(1) Revenues at Price A"
(2) Revenues at Price 8’”
(3) Revenues at Price C*’4
(4) Gap of Revenues (3)—(2)
(5) Gap of Revenues (4)—(2)

 

KBS
TV
(6) % Gap (5)/(2)
(7) % Gap (6)/(2)
(1) Revenues at Price A*
(2) Revenues at Price BM
(3) Revenues at Price C
MBC
TV

_ (7) % Gag (6)/(2)

SBS
TV

844*

(4) Gap of Revenues (3)—(2)
(5) Gap of Revenues (4)—(2)
(6) % Gap (5)/(2)

(1) Revenues at Price A*
(2) Revenues at Price B**
(3) Revenues at Price C***
(4) Gap of Revenues (3)—(2)
(5) Gap of Revenues (4)—(2)
(6) % Gap (5)/<2)

(7) % Gap (6)/(2)

 

 

 

SUM

(1) Revenues at Price A*
(2) Revenues at Price B**
(3) Revenues at Price C’**
(4) Gap of Revenues (3)—(2)
(5) Gap of Revenues (4)-(2)
<6) % Gap (5)/<2)

Table16.1. Revenue Estimation for the Major TV Networks Including Major Radio
(Unit: KW, Million/Year)

Daypart C
19,500
20,545
20,387

1,045
887
5.36%

4.55%

30,545
46,721
46,326

22,217

 

 

(7) % Ga (6)/(2)

* Price A is KOBACO set prices

** Price B is the implicit market prices estimated
*** Price C is the implicit market prices estimate

 

Daypart B
57,027
40,664

—7,649
—16. 16%

—12.19%

184,647
146,942
153 ,621
—37,705
—31,025

28.17% —20.42%

27.10% -16.80%
with the model including only the major TV and religious radios
d with the model including the major TV, religious radios, and major radios

 

  

 

 

 

 

 

 

 

 

 
 
 
  
 
 
 
 

 

 

  
 
 
 
 
  
 

 

 

 

  
    
 
 

 

 

 

 

 

 
   
 
 

 

 

Program Ads ,2 Block Ads
Daypart A Daypart SILK Total Daypart C Daypart B Daypart A Daypart SA Total SUM

154,636 171,469 402,633 369 3,372 5,631 24,923 34,295 436,928

149,883 187,940 399,032 713 1,157 552 15,478 17,901 416,933

147,908 187,475 398,329 284 2,224 570 15,115 18,193 416,522
—4,753 16,471 —3,601 344 —2,2 15 —5,079 -9,445 ~16,394 —19,995
-6,728 16,006 —4,303 —85 —1,147 —5,062 9,808 —16,102 —20,406
—3.07% 9.61% —0.89% 93.26% 65.68% —90. 19% —37.90% -47.80% 4.58%
—4.35% 9.33%; —1.07% —23.05% —34.03% —89.88% —39.35% —46.95% —4.67%

195,296 245,834 536,568 1,472 7,562 12,683 24,275 45,992 582,560

223,186 247,644 571,243 3,935 5,562 6,035 20,334 35,866 607,109

220,537 247,055 569,904 1,188 9,337 5,594 20,196 36,316 606,219
27,890 1,81 1 34,675 2,463 —2,000 —6,648 —3,941 —10,125 24,549
25,241 1,221 33,335 —284 1,776 —7,089 -4,079 —9,676 23,659
14.28% 0.74% 6.46% 167.32% —26.45% —52.41% —16.23% -22.02%

12.92% 0.50%_ 6.21% —l9.29% 23.48% —55.89% —16.80% -21.04%

144,163 225,596 464,414 184 1,159 3,334 10,670 15,347

157,338 231,646 479,367 1,388 1,075 763 13,836 17,061 496,428

155,361 230,939 478,855 360 3,001 836 14,389 18,586 497,440
13,174 6,050 14,952 1,204 —85 —2,571 3,166 1,714 16,667
11,197 5,343 14,440 176 1,842 —2,498 3,719 3,239 17,679
9.14% 2.68% 3.22% 654.91% —7.31% —77.1 1% 29.67% 11.17%
7.77% 2.37%4fl 3.11% 95.82% 158.84% -74.94% 34.86% 21.10%

494,096 642,899 1,403,615 2,025 12,093 21,648 59,868 95,634

530,406 667,251 1,449,641 6,036 7,793 7,351 49,648 70,828 1,520,470

523,806 665,469 1,447,088 1,832 14,563 6,999 49,700 73,094 1 520,182
36,311 24,332 46,026 4,01 1 -4,299 —14,297 -10,220 —24,805 21,221
29,711 22,570 43,472 —193 2,470 -14,648 —10,168 —22,539 20,933
7.35% 3.78% 3.28% 198.08% —35.55% —66.04% —17.07% —25.94% 1.42%
6.01% 3.51%” 3.10% —9.52% 20.42% —67.67% —16.98% -23.57% 1.40%

  
  

 

Table16.2. Revenue Estimation for the Religious Broadcasters Including Major Radio
(Unit: KW, Million/Year)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

83

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads

Daypart C Daypart B Daypart A Total Daypart C Daypart B Daypart A Total SUM
(1 ) Revenues at Price A* 75 1,266 6,395 7,736 17 93 466 576 8,311
(2) Revenues at Price B*’ —624 1,024 4,215 4,614 230 26 383 638 5,253
(3) Revenues at Price C*** —459 89 2,854 2,484 230 -157 678 751 3,235
BBS_FM (4) Gap of Revenues (3)—(2) —699 —242 —2,180 —3,121 213 —67 —83 62 —3,059
(5) Gap of Revenues (4)—(2) 634 —1,176 —3,541 6,251 213 —250 212 175 6,076
(6) % Gap (5)/(2) —93 1 91% —19.13% 64.09% —40.35% 1280.07% —72.13% —17.88% 10.85% —36.80%
(7) % Gap (6)/(2) —712.67% —92.94% 65.37% 67.89% 1279.36% —268.03% 45.57% 30.42% —61.08%
(1) Revenues at Price A* 23 3,272 16,183 19,479 152 796 3,281 4,229 23,708
(2) Revenues at Price B** —120 1,124 4,734 5,738 976 166 4,095 5,237 10,975
(3) Revenues at Price Ct“ —88 98 3,206 3,216 975 —1,002 7,260 7,233 10,449
CBS_AM (4) Gap of Revenues (3)-(2) —l43 —2,148 —11,450 —13,741 823 —630 815 1,008 —12,733
(5) Gap of Revenues (4)—(2) —1 12 ~3,174 —12,978 -16,263 823 —1,798 3,979 3,004 —13,259
(6) % Gap (5)/(2) —617.23% 65.65% —70.75% -70.54% 540.22% —79.13% 24.84% 23.84% 63.71%
(7) % Gap (6)/(2) —480.93% -97.00% 80.19% —83.49% 539.89% —225.86% 121.30% 71.03% 65.93%
(1) Revenues at Price A* 8 237 1,687 1,932 24 83 208 315 2,247
(2) Revenues at Price B** —320 541 2,859 3,080 904 62 445 1,411 4,491
(3) Revenues at Price Ct“ —236 47 1,936 1,748 903 —376 789 1,316 3,064
CBS_FM (4) Gap of Revenues (3)-(2) -328 303 1,172 1,148 879 -21 238 1,096 2,244
(5) Gap of Revenues (4)—(2) —243 -190 249 —184 879 —460 581 1,001 817
(6) % Gap (5)/(2) -4224.97% 127.99% 69.50% 59.43% 3592.48% —25. 16% 114.41% 347.37% 99.86%
(7) % Gap (6)/(2) —3137.91% -80.09% 14.79% —9.52% 3590.57% 651.24% 280.08% 317.15% 36.35%
(1) Revenues at Price A* 0 221 1,811 2,033 23 78 337 438 2,471
(2) Revenues at Price B** -14 507 2,884 3,377 811 60 758 1,630 5,007
(3) Revenues at Price C*** —10 44 1,953 1,987 811 —363 1,344 1,791 3,779
FEBCfiAM (4) Gap of Revenues (3)—(2) -14 286 1,073 1,345 789 —18 421 1,192 2,536
(5) Gap of Revenues (4)—(2) —11 —177 142 -45 788 —442 1,007 1,353 1,308
(6) % Gap (5)/(2) —4209.39% 128.99% 59.26% 66.14% 3470.74% —23.11% 124.87% 271.90% 102.63%
(7) % Gap (6)/(2) -3126.44% -80.00% 7.86972 —2.24% 3468.89% 663.58% 298.63% 308.76% 52.92%
(1) Revenues at Price A* 51 923 5,72 6,698 13 71 323 406 7,104
(2) Revenues at Price B** 621 780 3,420 3,579 163 22 250 435 4,014
(3) Revenues at Price C*** —457 68 2,316 1,927 163 —136 443 470 2,396
PBC FM (4) Gap of Revenues (3)—(2) —672 —142 -2,305 -3,119 150 —48 —73 29 —3,090
— (5) Gap of Revenues (4)-(2) —508 854 6,408 -4,771 150 —207 120 64 —4,707
(6) % Gap (5)/(2) -1325.61% —15.41% -40.26% —46.56% 1199.37% 68.26% —22.60% 7.09% —43.49%
(7) % Gap (6)/(2) —1002.62% —92.61% 69.54% -71.23% 1198.69%__—2fl.3_8%____37_.20%__’15._6§7g___£6_.27;%

 

r"
1

 

 

 

Tablel6.2. (cont’d)

 

 

 

 

 

 

 

 

 

 

 

(l) Revenues at Price A* 247 414 1.570: 2.231 30 30 2,262

(2) Revenues at Price 8"“ —2.582 549 1,603} —431 29 29 —401

(3) Revenues at Price CWWW“ —1,902 48 1,085 —768 ~177 -177 -946
WBS_FM (4) Gap of Revenues (3)—(2) —2,829 134 33 ~2,662 —l —1 —2,663
(5) Gap of Revenues (4)—(2) —2,149 —366 —484 —3,000 -208 —208 —3,207

(6) % Gap (5)/(2) —1143.54% 32.40% 2.10% —119.30% 301% ~3.01% -117.75%

(7) % Gap (6)/(2) 868.53% 88.44% —30.85% —134.43% 684.75% —684.75% —141.81%

(1) Revenues at Price A* 404 6,333 33,370 40,108 229 1,152 4,614 5,995 46,103

(2) Revenues at Price BM -4,281 4,524 19,714 19,957 3,084 367 5,931 9,381 29,338

(3) Revenues at Price C**W —3,153 395 13,351 10,593 3,082 —2,212 10,513 11,384 21,977

SUM (4) Gap of Revenues (3)—(2) -4,685 -1,809 —13,656 —20.151 2,855 -786 1,317 3,386 —16,764
(5) Gap of Revenues (4)—(2) —3,557 —5,938 —20,019 —29,515 2,853 —3,364 5,900 5,389 -24,126

(6) % Gap (5)/(2) ~1158.56% —28.57% 40.92% —50.24% 1247.75% -68.l7% 28.54% 56.48% -36.36%

(7) % Gap (6)/(2) —879.59% —93.76% —59.99% —73.59% 1247.05% —291.90% 127.87% 89.89% 62.33%

 

* Price A is KOBACO set prices

** Price B is the implicit market prices estimated with the model including only the major TV and religious radios

*** Price C is the implicit market prices estimated with the model including the major TV, religious radios, and major radios

84

 

 

 

 

 

(Unit: KW, Million/Year)

Tablel6.3. Revenue Estimation for the Major Radio Broadcasters including Major Radio

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Program Ads Block Ads
Daypart C Daypart B Daypart A Total Daypart C Daypart B Daypart A Total SUM
(l) Revenues at Price A* 201 2,465 9,120 1 1,786 46 384 1,662 2,092 13,878
(2) Revenues at Price C** —473 3,660 5,788 8,975 87 652 1,819 2,559 11,533
KBS—AM (3) Gap of Revenues (2) — (1) —674 1,195 -3,333 —2,811 41 268 157 466 —2,345
(4) % Gap (3)/(1) —335.75% 48.49% —36.54% —23.85% 88.66% 69.92% 9.46% 22.30% -16.89%
(1) Revenues at Price A* 23 229 1,542 1,794 11 59 212 282 2,076
(2) Revenues at Price C** —305 1,465 3,703 4,862 57 210 610 876 5,738
KBS—FM (3) Gap of Revenues (2) — (1) —328 1,235 2,161 3,068 46 151 397 594 3,662
(4) % Gap (3)/(1) —1436.09% 538.66% 140.17% 171.03% 403.94% 257.70% 187.34% 210.66% 176.41%
(1) Revenues at Price A* 113 3,751 12,673 16,536 219 478 1,133 1,830 18,367
(2) Revenues at Price C** -168 2,828 5,394 8,054 127 297 1,051 1,475 9,529
MBC—AM (3) Gap of Revenues (2) — (1) —281 —923 —7,279 8,482 —93 -181 —82 —356 —8,837
(4) % Gap (3)/(1) -249.12% -24.60% —57.43% —51.29% —42.19% —37.95% —7.20% -19.42% —48.12%
(1) Revenues at Price A* 449 2,098 8,307 10,855 92 133 992 1,217 12,072
(2) Revenues at Price C** -722 2,659 5,946 7,884 71 192 1,430 1,693 9,577
MBC~FM (3) Gap of Revenues (2) — (1 ) —1,171 561 -2,361 —2,971 —22 59 439 476 _2,495
(4) % Gap (3)/(1) —260.70% 26.73% —28.42% -27.37% -23.37% 44.43% 44.23% 39.12% —20.67%
(1) Revenues at Price A* 105 587 2,679 3,370 35 163 474 671 4,042
(2) Revenues at Price C** —1,247 3,656 5,408 7,816 158 561 976 1,695 9,511
SBS~AM (3) Gap of Revenues (2) - (1) —1,352 3,070 2,728 4,446 123 399 502 1,023 5,469
(4) % Gap (3)/(1) —1292.92% 523.27% 101.83% 131.91% 350.66% 245.19% 106.04% 152.49% 135.33%
(1) Revenues at Price A* 189 718 3,935 4,842 28 143 434 604 5,446
(2) Revenues at Price C** —949 2,676 5,755 7,482 174 606 1,307 2,087 9,569
SBS—FM (3) Gap of Revenues (2) — (1) ~1,l38 1,958 1,820 2,640 147 463 873 1,483 4,123
(4) % Gap (3)/(1) —601.17% 272.72% 46.26% 54.52% 532.73% 324.49% 201.27% 245.47% 75.70%
(1) Revenues at Price A* 1,079 9,848 38,256 49,183 432 1,358 4,907 6,697 55,880
(2) Revenues at Price C** —3,865 16,944 31,993 45,073 673 2,517 7,194 10,384 55,457
SUM (3) Gap of Revenues (2) - (1) —4,944 7,097 —6,263 —4,1 10 242 1,159 2,287 3,688 —423
(4) % Gap (3)/(1) —458. 10% 72.06% —16.37% 8.36% 56.02% 85.30% 46.61% 55.07% —0.76%

 

* Price A is KOBACO set prices

** Price C is the implicit market prices estimated with the model inclu

85

ding the major TV, religious radios, and major radios

 

 

  
   
  
  

Table 17.1 Mean Difference Test for Major TV Program Ads
(Unit: KW/Second)

    
     

  

95% C1
Lower
14

    
 

 
 
  
  
 
  
  

Mean
Differences

Mean Prices
KOBACO Market
633 669 487
77
377 160

      
      
       
     

         

     

    
  

91

    

 

740 15

   

   

   

           
    

775
17 138
73

l 13
1 4
733

 

     

 

389

    

    

      
   
   
      

127
l
1
l

 

199
1
149 771
723

  

    
  
   

  

430

    
  
    
   
 
   
 
  
  

 
 

791

 

110
355
361

1
1
103 425
644 113

 
   
  
 
   
   
   
  

    
   

. 143
117 778

379 798
387
1
1
141 701
1

 
    
   

Table 17.2. Mean Difference Test for Major TV Block Ads
(Unit: KW/Second)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7 Block Ads
Mean Prices Mean 95% Cl Sig.
KOBACO Market Differences _ SD SE Lower Upper t df (2-tailed)

Total 368,306 220,243 148,063 _ 157,552 6,831 134,644 161,482 21.68 531 0.00
DP_SA Random Sample (30%) 354,410 221,162 133,248 _ 153,451 11,839 109,874 156,621 11.25 167 0.00
KBS Total 207,471 21,007 186,464 1 1 1,034 7,972 170,741 202,187 23.39 193 0.00
TV DP_A Random Sample N/S N/S N/S 7 N/S N/S N/S N/S N/S N/S N/S
Total 77,567 35,727 41,840 48,557 2,51 1 36,903 46,778 16.66 373 0.00
DP_B Random Sample (50%) 78,327 34,679 43,648 _ 50,845 3,632 36,486 50,811 12.02 195 0.00
DP_C Total 43,442 80,866 (37,424) 45,198 4,991 (47,355) (27,493) (7.50) 81 0.00
Total 263,145 204,229 58,916 156,154 5,744 47,639 70,193 10.26 738 0.00
DP_SA Random Sample (30%) 252,571 196,997 55,574 144,474 9,547 36,762 74,385 5.82 228 0.00

Total 134,388 44,222 90,166 99,212 3,163 83,959 96,372 28.51 983 0.00 J
MBC DP_A Random Sample (20%) 122,721 34,878 87,844 87,431 6,106 75,804 99,884 14.39 204 0.00
TV Total 52,338 41,166 11,172 39,934 1,224 8,771 13,573 9.13 1,064 0.00
DP_B Random Sample (20%) 55,247 39,747 15,500 41,496 2,736 10,108 20,891 5.66 229 0.00
Total 28,263 70,758 (42,495) #17834 716 (43,900) (41,090) (59.38) 620 0.00
DP_C Random Sample (30%) 28,387 71,347 (42,960) 18,790 1,342 (45,607) (40,313) (32.01) 195 0.00
Total 128,51 1 153,518 —25,007 125,479 4,594 -34,026 —15,988 (5.44) 745 0.00
DP_SA Random Sample (30%) 124,102 149,093 —24,991 111,666 7,363 -39,499 —10,483 (3.39) 229 0.00
Total 58,115 15,574 42,541 _ 50,881 2,489 37,649 47,433 17.09 417 0.00

SBS DP_A Random Sample (50%) 58,195 16,106 42,088 _ 47,982 3,242 35,698 48,479 12.98 218 0.00 ’

TV Total 16,494 21,335 —4,841 42,759 1,352 —7,494 —2,188 (3.58) 999 0.00
DP-B Random Sample (20%) 16,678 20,062 -3,384 _ 41,415 2,858 —9,018 2,250 _flL 209 0.24
Total 8,483 61,1 18 —52,634 _ 17,763 818 —54,243 —51,026 (64.31) 470 0.00
DP_C Random Sample (50%) 8,260 59,476 —51,216 15,176 986 —53,158 -49,274 (51 96) 236 0.00

 

 

Note: Radom sampling was not made in case the number of observation is smaller than 260

87

 

 

 

 

 

 

 

 

 

Table 17.3. Mean Difference Test for Religious Radio Program Ads

(Unit: KW/Second)

 

Program Ads

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mean Prices Mean 95% C1 ' —

KOBACO 1 Market % Differences SD SE Lower 1 Upper 1 t 1 df Stalligleilz)

DP A Total 9,585 6,161 3,424 645 17 3,391 1 3,457 1 204.61 1 1,483 1 0.00

7 Random Samp1e(12%) 9,587 1 6,161 3,425 586 46 3,334 3,516 1 74.42 1 161 1 0.00

BBS_FM DP B Total 5,504 4,568 936 1,174 58 823 1,050 1 16.22 1 413 1W

7 Random Sample (50%) 5,608 4,568 1,040 1,317 89 864 1,216 1 11.63 216 1 0.00

DP_C Total 3,948 (34,930) 38,878 1,212 21 1 38,448 39,308 184.20 32 1 0.00

DP A Total 21,008 6,161 14,847 1,686 41 14,765 14,928 358.62 1,657 1 0.00

7 Random Sample (13%) 20,889 6,161 14.727 1,810 121 14,489 14,965 122.06 224 1 0.00

CBS_AM DP B Total 13,426 4,568 8,858 2,232 98 8,666 9.050 90.49 519 1 0.00
7 Random Sample (35%) 13,575 4,568 9,007 2,493 184 8,643 9,371 48.87 182 1 0.00

DP_C Total 6,768 (34,930) 41,697 154 26 41,644 41,751 1577.01 33 1 000

DP A Total 3,853 6,161 (2,308) 687 19 (2,345) (2,271) (121.93) 1,316 1 0.00

7 Random Sample (15%) 3,833 6,161 (2,329) 741 51 (2,430) (2,228) (45.43) 208 0.00

CBS_FM DP_B Total 6,648 4,568 2,080 4,594 106 1,872 2,289 19.56 1,865 0.00
Random Sample (10%) 6,328 4,568 1,760 4,519 332 1,105 2,416 5.30 184 0.00

DP c Total 3,737 (34,930) 38,666 2,138 142 38,387 38,946 272.49 226 0.00

7 Random Sample N/S N/S N/S N/S N/S N/S N/S N/S N/S 1 N/S

DP A Total 3,954 6,161 (2,208) 178 6 (2,219) (2,197) (387.88) 974 1 0.00

7 Random Sample (20%) 3.947 6,161 (2,214) 179 13 (2,239) (2,189) (176.85) 203 0.00

FEBC_AM DP_B Total 1,991 4,568 (2,577) 66 5 (2,588) (2,566) (470.32) 142 000
DP c Total 850 (34,930) N/C N/C N/C N/C N/C N/C 3* N/C

7 Random Sample N/S N/S N/S N/S N/S N/S N/s N/S N/S N/S

DP A Total 10,311 6,161 4,150 1,017 30 4,090 4,209 137.42 1,133 0.00

PBC FM 7 Random Sample(18%) 10,262 6,161 4,101 982 67 3,969 4,232 61.51 216 0.00
7 DP_B Total 5,400 4,568 N/C N/C N/C N/C N/C N/C 258* N/C
DP_C Total 2,850 (34,930) N/C N/C N/c N/C N/C N/C 19* N/C

DP A Total 6,034 6,161 (128) 205 8 (144) (112) (15.77) 640 0.00

WBS FM 7 Random Sample (30%) 6,037 6,161 (124) 183 13 (150) (99) (9.66) 201 0.00
7 Total 3,450 4,568 N/C N/C N/C N/C N/C N/C 263* N/C
Total 3,405 (34,930) 38,335 2,004 181 37,977 38,693 212.18 122 0.00

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Note: Radom sampling was not made in case the number of observation is smaller than 260
N/C: Not Computable because the standard error of the difference is 0.
* total number of cases in the data

88

 

 

 

Table 17.4. Mean Difference Test for Religious Radio Block Ads

(Unit: KW/Second)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FM Block Ads
Mean Prices Mean SD SE 95% C1 t1 df Sig. (2-

KOBACO Market Differences Lower Upper tailed)
DP_A Total 9,032 8,537 495 3,01 1 532 (591 ) 1 1,580 0. 93 1 31 1 0.36
BBS_FM DP_B Total 5,540 2,018 3,522 3,545 983 1,380 1 5,664.58 1 12 1 0.00
1311C Total 3,220 44,438 N/C N/C N/C N/C 1 N/C N/C 1 2* 1 N/C
DP A Total 6,381 8,537 (2,156) 7,262 258 (2,662) 1 (1,649) (8 35) 1 791 1 0.00
CBS AM 7 Random Sample (25%) 4,906 8,537 (3,631) 5,704 407 (4,434) (2,827) (8 91) 1 195 1 0.00
‘ DP_B Total 7,831 2,018 5,813 6,587 559 4,709 6,918 1041 1 138 1 0.00
DP_C Total 7,733 44,438 (36,705) 4,100 941 (38,681) (34,729) (3902) 1 18 1 0.00
DP_A Total 3,782 8,537 (4,755) 350 43 (4,841) (4,669) (110.37) 65 1 0.00
CBS_FM DP_B Total 2,795 2,018 778 890 157 457 1,098 4.95 31 1 0.00
DP_C Total 1,135 44,438 (43,303) 412 103 (43,523) (43,084) (420.47) 15 1 0.00
DP_A Total 3,630 8,537 (4,907) 309 32 (4,970) (4,844) (154.10) 93 1 0.00
FEBC_AM 13sz Total 2,312 2,018 294 804 126 40 548 2.34 40 1 0.02
DP_C Total 1,156 44,438 (43,282) 414 146 (43,628) (42,936) (295.95) 7 1 0.00
DP_A Total 1 1,030 8,537 N/C N/C N/C N/C N/C N/C 14* 1 N/C
PBC_FM DP B Total 5,516 2,018 3,498 3,859 1,726 (1,294) 8,291 2.03 4* 1 0.11
DPWC Total 3,420 44,438 N/C N/C N/C N/C N/C N/C 2* N/C
15pr Total N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A
_ Total 3,176 2,018 1,159 2,521 891 (949) 3,266 1.30 7 0.23
WBS_FM DP_B Random Sample N/C N/C N/C N/C N/C N/C N/C N/C N/C N/C
DP c Total N/A N/A N/A N/A N/A N/A N/A N/A N/A N/A

 

Note: Radom sampling was not made in case the number of observation is smaller than 260

N/C: Not Computable because the standard error of the difference is 0.

N/A: Not Available

* total number of cases in the data

89

 

“1%

estimation is specified by the dayparts. Table 18 compares my findings to those of other

studies.

Diagnostic Analysis

Reliable estimates from the regression can be obtained only via close examination
of data from the perspective how well the data fit the assumptions of regression analysis.
Regression is fairly robust, but the level of robustness varies with the assumptions.
Normality, linearity, homoscedasticity and independence of residuals are assumptions to
be examined.
Normality

Multivariate normality is the underlying assumption of statistical tests of multiple
regression. This assumption maintains that “each variable and all linear combinations of
the variables are normally distributed.” The first check of normality assumption is to
“investigate residuals of the analysis and to determine whether they are normally
distributed and independent” (Tabachnick and Fidell, 1996, p.70). For the present
regression results, as Figure 4 shows, the multivariate residual was normally distributed,
and minimum and maximum residuals were within 3.35 standard deviations of the mean
(see Table 19). According to Tabachnick and Fidell (1996), it is acceptable not to further
screen individual variables for normality if the residual plot looks normal. Appendix A
shows the Cook’s distance of the model, which “measures the effect of deleting a given
observation. Observations with larger Cook’s distance (D) values than the rest of the data
are those which have unusual influence” (Garson, 2007, p. 30). Referring to F ox (1991),

Garson suggests as “a cut-off for detecting influential cases, values of D greater than 4/(n

9O

 

Table 18. Comparison of Findings with those of Previous Studies

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. Religious
Major TV Broadcasters
Program Ads Block Ads Total
SA_25% SA_15% SA_10% SA_Avg. A_Avg. BfiAvg. CfiAvg. SUM Total TOtal
Shin (2002) %Gaps ALL N/E N/E 200% N/E
Park (2004) %Gaps ALL N/E N/E 400% N/E
Current Prices 9,293,000 6,382,000 2,591,000 1,559,000
Implicit Market
Prices KBS 20,000,000 15,000,000 10,000,000 13,182,000 7,000,000 2,000,000 1,000,000
Gaps TV 3,889,000 618,000 591,000 559,000
%Gaps 41.85% 9.68% —22.81% -35.86% 13.00%
Current Prices 10,571,000 6,575,000 2,376,000 1,266,000
Implicit Market MBC
Jung(2005) Prices TV 20,000,000 15,000,000 10,000,000 13,333,000 8,000,000 2,000,000 1,000,000 14.80% N/E N/E —66.60%
Gaps 2,762,000 1,425,000 —376,000 -266,000
%Gaps 26.13% 21.67% —15.82% —21.01% 13.60%
Current Prices 9,320,000 5,606,000 2,391,000 1,1 13,000
Implicit Market SBS
Prices TV 20,000,000 15 ,000,000 10,000,000 12,500,000 7,000,000 2,000,000 1,000,000
Gaps 3,180,000 1,394,000 -391,000 —113,000
%Gaps 34.12% 24.87% —16.35% —10.15% 18.90%
Current Prices 10,579,873 10,191,196 9,199,721 9,498,518 5,712,059 3,003,655 2,254,840 5,216,653
Implicit Market KBS
Prices TV 14,697,105 11,500,460 9,537,183 10,042,307 _5,657,399 1,918,465 2,482,797 5,019,562
Gaps 4,117,232 1,309,264 337,461 543,789 _ —54,660 —1,085,191 227,958 ~197,091
%Gaps 38.92% 12.85% 3.67% 5.72% ~0.96% —36. 13% 10.11% —3.78%
Current Prices 12,380,398 9,825,620 10,491,609 10,859,703 5,333,902 2,574,638 1,578,139 4,785,603
Implicit Market
Present Prices MBC 19,723,644 11,322,113 9,514,379 11,706,457 6,031,248 1,913,757 2,341,179 5,199,477 328% 35.94% 142% 36.36%
Study Gaps TV 7,343,245 1,496,493 —977,230 846,754 697,347 —660,882 763,039 413,874
%Gaps 59.31% 15.23% —9.31% 7.80% 13.07% —25.67% 48.35% 8.65%
Current Prices 10,078,085 10,991,486 9,568,651 9,661,695 5,696,977 2,414,237 2,125,521 4,949,557
Implicit Market SB S
Prices TV 16,989,982 11,655,460 9,548,748 9,685,097 6,052,966 1,779,980 2,394,111 4,936,493
Gaps 6,911,897 663,974 -19,903 23,401 355,989 -634,258 268,590 —13,063
%Gaps 68.58% 6.04% —0.21% 0.24% 6.25% -26.27% 12.64% —0.26%

 

 

N/E: No Equivalent Data

 

 

 

 

 

 

 

- k - 1), where n is the number of cases and k is the number of independent variables”

(Garson, 2007, p. 30).

Figure 4. Histogram of Regression Standardized Residual

Histogram

Dependent Variable: AvgBugtTVMinor

 

a:
O
l

Frequency
5:

20"

 

-2 .5

0.0
Regression Standardized Residual

  

2.5

Note: AvgBugtTVMinor stands for Average Unit Price

Table 19. Residual Statistics

 

Mean =1 .22E-14
Std. Dev. =0.983
N=443

 

 

 

 

 

Standard

Minimum Maximum Mean Deviation
Predicted Value 1 14,268 7,617,768 3,078,250 1,353,698 443
Residual -940,296 1,059,599 0 261,1 10 443
Std. Predicted Value -2 3 0 1 443
Std. Residual -4 4 0 1 443

 

 

 

 

 

 

92

 

Although not required, it is meaningful to check distributions of individual
variables as “inconsistent direction of skewness among variables raises concerns”
(Tabachnick and F idell, 1996, p.73). Frequency statistics showed that variables in this
study were either normally or positively distributed (see Table 20). Variables which
turned out to be normal include average package price (.66 for skewness, .00 for kurtosis),
average audience second of TV’s program advertising (.68 for skewness, 0.40 for
kurtosis), fraction of TV’s program advertising in daypart SA (1.31. for skewness, 2.47
for kurtosis), fraction of religious radio’s program advertising in daypart A (.77 for
skewness, 0.64 for kurtosis), fraction of TV’s program advertising in daypart A (1.51 for
skewness, 3.39 for kurtosis), fraction of religious radio’s program advertising in daypart
B (1.68 for skewness, 3.68 for kurtosis). All other variables are skewed positively. This
positive skewness is mainly attributed to a stack of data points at zero. This means that
many packages were sold without the ad units in those dayparts. Overall, it seems that

data have no significant non-normality problem.

Table 20. Skewness and Kurtosis

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Skewness Kurtosis
Average Package Price 0.66 0.00
Average Audience Seconds of TV Program Ads 0.68 0.40
Fraction of TV Program Ads in Daypart SA 1.31 2.47
Fraction of TV Program Ads in Daypart A 1.51 3.39
Fraction of TV Program Ads in Daypart B 1.68 3.68
Fraction of TV Program Ads in Daypart C 2.22 6.96
Average Audience Seconds of TV Block Ads 3.51 20.36
Fraction of TV Block Ads in Daypart SA 4.47 37.91
Fraction of TV Block Ads in Dag/part A 4.53 31.36
Fraction of TV Block Ads in Daypart B 4.02 28.42
Fraction of TV Block Ads in Daypart C 4.69 31.12
Fraction of Minor Radio Program Ads in Daypart A 0.77 0.64
Fraction of Minor Radio Program Ads in Daypart B 2.43 7.50
Fraction of Minor Radio Program Ads in Daypart C 3.92 15.82

 

93

Table 20 (cont’d)

 

 

 

 

 

 

 

Fraction of Minor Radio Block Ads in Daypart A 5.39 40.76
Fraction of Minor Radio Block Ads in Daypart B 4.36 20.54
Fraction of Minor Radio Block Ads in Daypart C 7.64 58.92

 

Linearity and Homoscedasticity

The assumption of linearity means the effects of independent variables on the
dependent variable are constant. Linearity is important practically because “Pearson’s r
only captures linear relationships among variables.” On the other hand, the assumption of
homoscedasticity is that “the variability in scores for one continuous variable is roughly
the same at all values of another continuous variable” (Tabachnick and F idell, 1996, p.
70). Homoscedasticity is related to the assumption of normality because “relationships
between variables are homoscedastic when the assumption of multivariate normality is
met” (Tabachnick and Fidell, 1996, p. 70).

Hence, both linearity and homoscedasticity are diagnosed from residuals plots in
analyses involving a predicted variable. “In plots where standardized residuals are plotted
against predicted values, nonlinearity and heteroscedasticity is indicated when most of
the residuals are above the zero line on the plot at some predicted values and below the
zero line at other predicted values” (Tabachnick and Fidell, 1996, 70). Figure 5 (Normal
P-P Plot of Regression Standardized Residual) shows there is no indication for

nonlinearity or heteroscedasticity.

94

Figure 5. Normal P-P Plot of Regression Standardized Residual

Normal P-P Plot of Regression Standardized Residual

Dependent Variable: AvgBugtTVMinor

 

1.0—

o
in
1

Expected Cum Prob
i ‘1’

.0
n
l

 

 

 

0-3 T l l I
0.0 0.2 0.4 0.6 0.8 1 .0

Obsorvod Cum Prob

Note: AvgBugtTVMinor stands for Average Unit Price

95

CHAPTER 6

DISCUSSION AND CONCLUSION

This study found evidence fer price distortion and subsidy in the Korean
broadcast advertising market. In general, implicit market prices for major TV networks
were estimated to be higher than KOBACO set prices, while implicit market prices for
religious broadcasters were estimated lower than KOBACO set prices. Within the major
TV networks, implicit market prices for program ads were estimated higher than
KOBACO set prices while block ads’ implicit prices were estimated lower than
KOBACO set prices. Even within program ads, dayparts SA, A, and C were under-priced
by KOBACO while daypart B was over-priced. Revenues for major radio broadcasters
remained the same overall, although their implicit program ad time prices were estimated
to be lower and implicit market block ad spot prices were estimated to be higher than
KOBACO set prices.

Table 21.3 shows that price distortion was attributed mainly to weight difference
assigned to pricing formula factors. KOBACO generally assigned a higher weight to
daypart than to audience size. Viewed by daypart, dayparts SA, A, and C for major TV
networks were under-priced while daypart B for major networks was over-priced by
KOBACO. At the same time, religious broadcasters were over-priced by KOBACO. To
approximate implicit market prices, KOBACO should assign higher weight to dayparts
SA, A, and C but a lower weight to daypart B for major TV networks. At the same time,
KOBACO may need to charge less for ad time in religious broadcasters’ programs to sell

all their ad time without resorting to packaging. It seems evident that religious

96

broadcaster programs were subsidized by major TV network programs through packaging.
Otherwise, over-priced ad slots in religious broadcaster programs might not have been
sold.

It remains unclear whether ad time in major radio broadcaster programs was sold
as a part of packages (Informal interviews with advertisers indicate that major radio ad
time is infrequently packaged with major TV ad time but is more frequently packaged
within major radio programs). There was only a 0.76% difference between estimated
advertiser payment at implicit market prices and advertiser expenditures at KOBACO
prices. Nevertheless, there were significant differences in average prices between
KOBACO set prices and implicit market prices depending on dayparts and ad types.
Dayparts A and C for program ads were over-priced while daypart B was under-priced.
On the other hand, time for block ads on major radio broadcasters was under-priced
throughout dayparts.

It was also found that, while coefficients for radio block ad dayparts explaining
package prices were not statistically significant, estimates of KOBACO set prices for

radio ad time were not explained by factors employed in the implicit market price

estimation (i.e., dayparts) either. Table 21.2 shows that the R2 ’s of the regressions,
where KOBACO prices were dependent variables and dayparts were independent
variables, were extremely low for both program ads and block ads for radio programs.
Dayparts explained only 0% to 7% of variation in KOBACO prices. Conversely, more
than 81% of KOBACO set prices for TV program ads were explained by factors in the

model (81% for KBS-TV and SBS-TV; 84% for MBC-TV) while 40% (SBS-TV) to 71%

97

(KBS-TV) of KOBACO set prices for TV block ads were explained by the factors (see
Table 21.1).

In addition, no evidence was found of discounts for larger packages in the Korean
broadcast ad time sales market. There is literature on advertising prices indicating the
existence of volume discounts and some suspect price discrimination effect against small
advertisers (Blake and Blum, 1965; Blank, 1968). However, in the Korean advertising
sales market, data shows that no such practice was undertaken. The package size variable
turned out to be non-significant in models where either the package price or the price per
audience second (a concept equivalent to CPM) was a dependent variable.

The current study also confirmed previous studies finding that advertisers prefer
male viewers to female viewers. When the female fraction variable was included in the
model, its coefficient had a statistically significant negative effect on average package
price.

As the first econometric analysis using actual transaction data, this study is
believed to contribute to KOBACO’s exploration of ways to improve its pricing formula.
Reform of Korean broadcasting policy and efficient decision-making by industry
participants can be assisted by the study. Beneficiaries may include content providers,
broadcasters, advertisers and other media firms. The study also contributes to the design
of an explicit subsidy system should policymakers decide such is appropriate. Particularly,
estimated market value deviations from corresponding entries in the published rate card
provide estimates of the size of rate restructuring that will occur when competition is
introduced. Cross-subsidies flowing from major TV networks to religious radio

broadcasters'were explored also. Specific implications of the study are discussed below.

98

 

Table 21.1. Comparison between the Estimators of Implicit Market Prices and KOBACO Set Prices for Major TV
(Unit: KW/Secontl, KW/Unit)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Major TV
Program Ads! Block Ads
Implicit Market Implicit Market KOBACO KOBACO KOBACO Implicit Market Implicit Market KOBACO KOBACO KOBACO
Price 1 Price 2 (KBS—TV) (MBC-TV) (SBS—TV) Price 1 Price 2 (KBS—TV) (MBC—TV) (SBS-TV)
(Constant) 516695” 589,514*** 2,088,034*** 2.238,967*** l,573,595*** 516,695*** 589,514*** 1.3 17,969*** (729,696)*** 416740***
Audience
Second 84*** 84*“ 54*‘7'W 20*“< 55*** 102*” 88*** (~8)*** 3** (-10)***
Daypart SA 5,808,342*** 5,692,2()4*** 5,()00,755*** 7,334,590*" 5,860,9104Wi‘ (—],693,501)*** (-l,216,548)* 4,683,284*** 3,063,694*** 2,003,444***
Daypart A 3,018,817*** 2,859,017*** 2,248,128*** 2,499,018*** 2,452,107*** (—2,634,099)*** (—236,985)*** 2,032,074*** 1,184,291 *** 791,439***
Daypart B (-1,375,106)*** (—882,507)**
Daypart C 1,191,829*** 1,096,662*** (353,048)*** (—814,699)*** 90,3()8*** _ -120,429 —853,774 (~567,143)*** (-331,279)*** (-221,522)***
R squared 096 0.98 0.81 0.84 _ 0.81 i 0.96 0.98 0.71 0.57 0.40
Implicit Market Price 1: Estimated with the Model constituted only with Major TV and Minor Radio
Major Radio

Implicit Market Price 2: Estimated with the Model constituted with Major TV, Minor Radio, and
KOBACO: Estimated with KOBACO prices for each of broadcasters’ programs (i.e., KBS—TV, MBC—TV, and SBS—TV)

Table 21.2. Comparison between the Estimators Implicit Market Prices and KOBACO Set Prices for Minor Radio and Major Radio

(Unit: KW/Second, KW/Unit)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

___ . _ _ _ _—/_ _ . _ _ ._ _ __
Minor Radio Major Radio
Prooram Ads Block Ads Program Ads Block Ads
Implicit Market Implicit Market KOBACO Implicit Market Implicit Market KOBACO Implicit Market KOBACO Implicit Market KOBACO

____ Price 1 Price 2 (Minor Radio) Price 1 Price 2 (Minor Radio) Price 2 (Major Radio) Price 2 (Major Radio)
M 516,695*** 589,514*** 56,049 516,695*** 589,514*** __ 59,879*** 589,514*** 77,191*** 589,514*** 83,942***

Audience

Second 7__

Daypart SA _

Dayglrt A (-393,467)*** (—506,061)*** 96,980 (—345,956)* —286,847_ 31,628** (~347,043)*** 152,768*** (-366,279)* 45,829***

Daypart B (-425,339)*** (—581,535)*** 43,669 —476,343 832,808 28,214* -241,385 97,106*** —372,642 9,060

Daypart C (~121,5291)** (—1,104,008)* 372,072 298,793_ (-875,815)*** -434,901

R s uared 0.96 0.98 0.07 0.96 0.98 0.00 0.98 0.10 0.57 0.03

del constituted only with Major TV and Minor Radio

Implicit Market Price 1: Estimated with Mo
Implicit Market Price 2: Estimated with Mo
KOBACO: Estimated with KOBACO prices

del constituted only with Major TV, Minor Radio, and Major Radio
for each of broadcasters’ programs (i.e., Minor Radio and Major Radio)

 

99

 

 

 

 

 

 

 

 

 

 

 

 

 

100

Audience Second 54

KBSTV Daypart SA 7.088.789

Program Ads Daypart A 4‘336‘162

b Daypart B 2,088,034

Daypart C 1,734,986

Audience Second 20

MBCTV Daypart SA 9,573,559

Procram Ads Daypart A 4’737‘987

D Daypart B 2,238,969

Daypart C 1,424,270

Audience Second 55

SBSTV Daypart SA 7,434,505

Prowram Ads Daypart A 4,0-5,703

D Daypart B 1,573,595

Daypart C 1,663,903

Minor Radio Daypart A 153,029

Program Ads Daypart B 99,718

D Da art C 56,049

Major Radio Daypart A 77:38:;
Program Ads Daypart B ’

Da art C 77,191

Audience Second -8

Daypart SA 6,001,253

KBSTV Daypart A 3,350,043

81”" Ads Daypart B 1,317,969

Daypart C 750,826

Audience Second 3

Da art SA 3,793,391

MBCTV Dazgart A 1,913,987

BIOCk Ads Daypart B 729,696

Da art C 398,417

Audience Second —10

Da art SA 2,420,184

SBSTV Dajgart A 1,208,179

BIOCk Ads Daypart B 416,740

Daypart C 195,219

Minor Radio Daypart A 3938::
Block Ads Daypart B ’

Daypart C 59,879

Major Radio Daypart A 129,772

Block Ads

Table 21.3. Estimator Comparison
(Unit: KW/Second, KW/Unit)

6,325,037
31,535,512
516,695

84
6,325,037

......

516,695

84
6,325,037
3,535,512

516,695
123,228
91,356

ft

—1,176,806
-2,1 17,404
—858,41 1

102

—1 , 176,806
-2,1 17,404
-858,41 1
396,266
102
—1,176,806
—2,1 17,404
—858,41 1
396,266

___/’4

40,352

 

If,”
Daypart B 93,002
Day art C 83,942 'fl_

Esimatorsglmplicit Market Price 1

1,708,524

1 ,708,524

1,708,524

698,596

102

396261

' 170,739

M

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Esimators_1mplicit Market Price 2 GAP l GAP 2

84 30 30

6,281,718 ~763,752 -807,071

3,448,531 800,650 887,631

589,514 —1,571,339 -1,498,520

1,686,176 -26,462 —48,8 1 0

84 64 64

6,281,718 -3,248,522 -3,291,841

3,448,531 -1,202,475 -1,289,456

589,514 -1,722,274 - 1,649,455

1,686,176 284,254 261,906

84 29 29

6,281,718 -1,109,468 -1,152,787

3,448,531 -490,191 -577,172

589,514 -1,056,900 —984,081

1,686,176 44,621 22,273

83,453 -29,801 -69,576

7,979 —8,362 —91,739

—514,494 -754,645 —570,543

242,471 —229,960 12,511

348,129 —174,298 173,831

-286,301 -77,191 —363,492

88 1 10 96

—627,034 -7,178,059 -6,628,287

352,529 —5,467,447 -2,997,514

—292,993 —2,176,380 —1,610,962

-264,260 —354,560 -1,015,086

88 99 85

-974,077 4,970,197 4,767,468

5,486 4,031,391 -1,908,501

640,036 -1,588,107 -1,369,732
A 2,151 —1,009,720
88 1 12 98

864,019 —3,596,990 -3,284,203

1 15,544 -3,325,583 -1,092,635

-529,978 — 1 ,275,151 -946,718
A 201,047 —696,463
302,667 79,233 211,161

—243,294 —47 ,741 —331,387
A 828,888 828,428 _
223,235 —129,772 93,463

216,872 -93,002 123,870
L/fflg 83,942 70,671

 

Implications

Implications for Media Policy and Industry. Since 1998, the Korean government
continuously has tried to introduce competition into the market. However, it has been
delayed by conflicting claims on the effect of market liberalization. The claims are
twofold: one, that there will be a price increase by major TV networks and, two, there
will be financial distress for minor radio stations. On one hand, some argue there will be
significant price increase by major TV networks while causing financial crisis for minor
broadcasters. On the other hand, opposite claims are made. Disagreement on the effect of
market liberalization is attributed mainly to a paucity of empirical evidence on how much
price regulation and the KOBACO monopoly system distorts prices from market values.
Were a study to provide estimates of implicit market prices based on a theoretically and
empirically reasonable model, discussion on policy change could proceed on the basis of
sound empirical knowledge.

The current study was designed to address this need. By providing reasonably
estimated implicit market prices, this study contributes to faster market liberalization and
the establishment of alternative mechanisms such as rate-restructuring schedules and
fundraising methods to support minor broadcasters. The current empirical study furnishes
answers for the following two questions raised by Kim (2001): one, would market
liberalization cause rapid increases in advertising prices?; two, would revenue gaps
between major TV networks and minor broadcasters increase?

Regarding the first question, the findings described above indicate the possibility
of price increase in major TV networks in average terms. However, individual prices

could increase as well as drop, depending on performance. Table 22 shows the ranges of

101

price variation comparing implicit market prices against KOBACO set prices. Ranges for
implicit market prices became larger than those of KOBACO set prices in all ad time
types except those in major radio. For TV program ad time, the highest implicit market
prices were estimated 20% (KBS-TV) to 60% (MBC-TV) higher than KOBACO
maximtun prices. For major TV block advertising, maximum implicit market prices were
estimated 24% to 5 5% higher than KOBACO maximum prices. Regarding minima,
implicit market prices for major TV program ads also were estimated higher than
KOBACO set prices. But price ranges of implicit market prices were still larger than
those for KOBACO set prices, as the maximum price increased even further. For major
TV block advertising, minimum prices were estimated to have negative values. This
means that advertising spots existed that advertisers would never want to buy. Were they
bought, it would imply an opportunity to buy other ad time units whose prices were set
lower than their values.

For religious broadcasters, ranges of implicit market prices were estimated to be
larger than those of KOBACO set prices. However, their minima and maxima all were
estimated lower than KOBACO prices. As is the case of block ads in TV, minimum
prices for religious broadcasters had negative values and the ad slots having negative
prices appear sold by being packaged with others. The same logic applies to interpreting
negative prices for major radio. Unique to major radio’s block advertising time is
estimated minimum ad time prices that were higher than KOBACO set prices. Therefore

price ranges were smaller than those for KOBACO set prices.

102

 

Table 22. Comparison of Price Ranges

(Unit, KW/Second)

 

 

 

  
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

: Range Minimum Maximum Mean SD Variance

Unit Price _ KOBACO 856,000 41,000 897,000 350,040 194,514 37,835,816,616

KBSTV Unit Price _ Implicit Market Price 1 1,033,897 44,893 1,078,791 336,643 213,391 45,535,543,939

Program Ads Unit Price _ Implicit Market Price 2 1,028,895 49,792 1,078,687 335,666 211,225 44,615,842,973

Unit Price _ KOBACO 966,854 3,146 970,000 320,134 238,670 56,963,162,621

MBCTV Unit Price _ Implicit Market Price 1 1,507,336 45,485 1,552,821 347,752 269,466 72,612,039,300

Program Ads Unit Price _ Implicit Market Price 2 1,504,339 50,386 1,554,726 346,745 268,185 71,923,247,628

Unit Price _ KOBACO 995,366 26,634 1,022,000 331,260 229,525 52,681,887,123

SBSTV Unit Price _ Implicit Market Price 1 1,206,074 48,490 1,254,564 330,046 222,273 49,405,241,388

Program Ads Unit Price _ Implicit Market Price 2 1,201,801 53,404 1,255,205 329,382 220,168 48,474,032,484

Unit Price _ KOBACO 25,990 360 26,350 9,312 6,286 39,519,312

Minor Radio Unit Price _ Implicit Market Price 1 41,091 34,930 6,161 4,840 6,317 39,906,072
Program Ads Unit Price _ Implicit Market Price 2 29,897 25,725 4,173 2,685 4,736 22,431,799 ‘

Major Radio Unit Price _ KOBACO 50,300 900 51,200 13,595 8,716 75,976,755

Program Ads Unit Price _ Implicit Market Price 2 31,722 14,315 17,406 11,794 7,042 49,589,753

Unit Price 3 KOBACO 363,850 19,800 383,650 222,044 137,123 18,802,820,496

KBSTV Unit Price _ Implicit Market Price 1 579,829 59,264 520,565 122,683 129,968 16,891,750,706

Block Ads Unit Price _ Implicit Market Price 2 524,350 47,825 476,526 125,355 117,322 13,764,519 207

Unit Price _ KOBACO 362,650 21,000 383,650 129,785 112,134 12,574,123,628

MBCTV Unit Price _ Implicit Market Price 1 652,406 62,890 589,516 83,346 87,368 7,633,243,862

Block Ads Unit Price _ Implicit Market Price 2 587,151 —50,962 536,189 80,983 83,710 7,007,298,313

Unit Price _ KOBACO 194,850 10,200 205,050 52,646 42,708 1,824,012,568

SBSTV Unit Price _ Implicit Market Price 1 404,868 87,091 317,776 66,759 78,480 6,159,153,176

Block Ads Unit Price _ Implicit Market Price 2 372,957 -71,904 301,054 72,923 73,690 5,430,200,705

Unit Price _ KOBACO 194,850 10,200 205,050 52,646 42,708 1,824,012,568

Minor Radio Unit Price _ Implicit Market Price 1 404,868 —87,091 317,776 66,759 78,480 6,159,153,176

Block Ads Unit Price _ Implicit Market Price 2 372,957 -71,904 301,054 72,923 73,690 5,430,200,705

Major Radio Unit Price _ KOBACO 18,350 435 18,785 6,360 4,298 18,472,706‘

Block Ads Unit Price _ Implicit Market Price 2 3,431 7,731 11,162 11,025 437 190,540

 

Implicit Market Price 1: Estimated with the Model constituted only with Major TV and Minor Radio
Implicit Market Price 2: Estimated with the Model constituted with Major TV, Minor Radio, and Major Radio

KOBACO: Estimated with KOBACO prices for each of broadcasters’ programs (i.e., KBS-TV, MBC—TV, and SBS-TV)

 

 

  
 
   
  
  
 
  
   
 

 

Regarding the second question whether revenue gaps between major TV networks
and religious broadcasters would increase, it is expected that major TV networks would
earn more while religious broadcasters would earn less at implicit market prices. The
estimates predict that major TV networks would earn 1.4% more with the implicit market
prices than under the current KOBACO system. Specifically, MBC-TV and SBS-TV
would earn more (4.2% and 3.5% respectively) while KBS-TV would earn less (4.6%),
compared to what they earned under the current KOBACO system. On the other hand,
religious broadcasters’ revenues at implicit market prices fell by 36% to 52% compared
to their values with KOBACO prices. It seems necessary to establish an explicit funding
mechanism to support religious broadcasters if the government decides on such support.

Implication for the Studies of Advertising Prices. Previous studies, despite their
contribution to understanding Korean broadcast ad time pricing, contain serious
limitations. Studies using published rate cards are flawed because advertising spots are
sold in packages and it is indeterminable whether advertisers value spots in the packages
at their published rates. Even the study employing a survey to estimate advertisers’
willingness to pay contains limitations: surveys may systematically under-estimate
market prices as advertisers report their willingness to pay more conservatively in
response to a survey than their behavior in the market would indicate; at the same time,
translating survey results into demand and supply curves requires a number of a priori
assumptions such as frequency of advertising purchases by advertisers and behaviors of
market actors other than survey participants.

The present study achieved a significant improvement for studies of broadcast

advertising'time in Korea. By estimating implicit market prices for broadcast ad time

104

using actual transaction data and a theoretically-driven hedonic regression, the study
generates more reliable predictions for prices expected to prevail in a competitive market

system.

Limitations

This study has several limitations. First, real package data were not used. Instead,
package price was operationalized in terms of advertiser ad budgets spent in a network
(and its associated religious broadcasters), through an ad agency, during the sample
period. This method has the limitation of treating all ad agency purchases for an
advertiser as a single package, when multiple packages may have been purchased.
However, this does not violate the logic of the package-price regression approach
outlined above, as KOBACO prices are the same for all packages. The downside of
treating all purchases by an advertiser through a particular ad agency as the purchase of a
single package is that it reduces the number of observations. Future studies using actual
package prices would improve estimate precisions.

Second, a critical proportion (50% - 70%) of ad contracts with religious
broadcasters and major radio broadcasters were removed from the analysis due to
unidentified information on dayparts. It was possible to match programs with dayparts for
regular programs because their histories were posted on the website. However, it was not
possible to obtain schedules for special programs broadcast on holidays or for sporting
events. Future studies can improve prediction precision by obtaining broadcast times for
special programs on religious and major radio stations. In addition, if audience ratings for

radio programs were available, estimation would improve significantly.

105

Third, a major concern of this study is to predict prices at the edge, or to predict
prices outside the boundary of data. The study estimates for dayparts’ influence on
implicit prices from the regression where average package price per spot is a dependent
variable and average audience seconds per spot and fractions of dayparts for several types
of programs comprising the package are independent variables. Then, the coefficients for
daypart fractions were interpreted as those for dummies because fraction term 1 is
theoretically the same as the dummy for the daypart. However, there were few such cases
comprising packages with only one type of advertising spot in the data. Hence, prediction
of implicit market prices by projecting beyond the range of the data, a method
statisticians suggest should be avoided. Future studies including samples, some of which
are made up of only one type of ad, would reduce this concern.

Although it was not taken seriously in this study, defining public interest
programs is a critical part of package studies. Theoretically, subsidies flow from regular
programs to programs favored by regulators. Hence, to estimate subsidy size,
categorization of programs with respect to public interest should be precise. The current
study approached public interest programs through institutions. That is, all religious
broadcaster programs were considered to be public interest programs. To estimate the size
of the internal subsidy flowing from commercial programs to public interest programs
within a network, a clear definition and categorization of programs, in terms of public

interest, is critically important.

106

Conclusions

Despite the limitations, the present study contributes to KOBACO’S efforts to
improve its pricing formula. In addition, the reform of Korean broadcasting policy and
efficient decision-making by industry participants will be assisted by the study.
Beneficiaries may include content providers, broadcasters, advertisers, and other media
firms. The study also contributes to the design of an explicit subsidy system should
policymakers decide such a policy is appropriate. Particularly, estimated market value
deviations from corresponding entries in published rate cards provide estimates of the

size of rate restructuring that will occur when competition is introduced.

107

Appendix A. Outlier Statistics

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Cook's Distance (CD) Cut-off Value 1 if CD is bigger than Cut-off Value
0.00013 0.00937 0
0.00090 0.00937 0
0.00090 0.00937 0
0.00001 0.00937 0
0.00003 0.00937 0
0.00001 0.00937 0
0.02375 0.00937 1
0.00052 0.00937 0
0.00026 0.00937 0
0.00005 0.00937 0
0.00006 0.00937 0
0.00053 0.00937 0
0.00008 0.00937 0
0.01527 0.00937 1
0.00012 0.00937 0
0.00001 0.00937 0
0.00005 0.00937 0
0.00000 0.00937 0
0.05667 0.00937 1
0.00487 0.00937 0
0.00013 0.00937 0
0.00000 0.00937 0
0.00231 0.00937 0
0.00873 0.00937 0
0.00000 0.00937 0
0.80796 0.00937 1
0.03865 0.00937 1
0.00010 0.00937 0
0.00000 0.00937 0
0.00025 0.00937 0
0.05632 0.00937 1
0.00515 0.00937 0
0.00113 0.00937 0
0.00002 0.00937 0
0.00357 0.00937 0
0.06992 0.00937 1
0.01400 0.00937 1
0.00007 0.00937 0
0.00003 0.00937 0
0.00018 0.00937 0
0.02803 0.00937 1
0.00237 0.00937 0

 

108

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00039 0.00937 0
0.00247 0.00937 0
0.00042 0.00937 0
0.00041 0.00937 0
0.00034 0.00937 0
0.00013 0.00937 0
0.00099 0.00937 0
0.00156 0.00937 0
0.00029 0.00937 0
0.00001 0.00937 0
0.00016 0.00937 0
0.00001 0.00937 0
0.00024 0.00937 0
0.00016 0.00937 0
0.00504 0.00937 0
0.00886 0.00937 0
0.00011 0.00937 0
0.01081 0.00937 1

. 0.00937 1
0.00090 0.00937 0
0.00152 0.00937 0

. 0.00937 1
0.00458 0.00937 0
0.00015 0.00937 0
0.00016 0.00937 0
0.00001 0.00937 0
0.00016 0.00937 0
0.00330 0.00937 0
0.00121 0.00937 0
0.00715 0.00937 0
0.00001 0.00937 0
0.0001 1 0.00937 0
0.00134 0.00937 0
0.00087 0.00937 0
0.00256 0.00937 0
0.02048 0.00937 1
0.00020 0.00937 0
0.00004 0.00937 0
0.00027 0.00937 0
0.00160 0.00937 0
0.01101 0.00937 1
0.00791 0.00937 0
0.00381 0.00937 0

 

 

 

109

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00067 0.00937 0
0.00014 0.00937 0

. 0.00937 1
0.03071 0.00937 1
0.00622 0.00937 0
0.00011 0.00937 0
0.00120 0.00937 0
0.02393 0.00937 1
0.00043 0.00937 0
0.00004 0.00937 0
0.00001 0.00937 0
0.00023 0.00937 0
0.00282 0.00937 0
0.00014 0.00937 0
0.00012 0.00937 0
0.00071 0.00937 0

. 0.00937 1
0.08404 0.00937 1
0.00034 0.00937 0
0.00008 0.00937 0

. 0.00937 1
0.00044 0.00937 0
0.00041 0.00937 0
0.00025 0.00937 0
0.00171 0.00937 0

. 0.00937 1
0.00660 0.00937 0
0.00082 0.00937 0
0.00129 0.00937 0
0.00105 0.00937 0
0.00575 0.00937 0
0.00039 0.00937 0
0.00001 0.00937 0
0.00001 0.00937 0
0.00004 0.00937 0
0.00037 0.00937 0
0.00037 0.00937 0
0.00027 0.00937 0
0.00012 0.00937 0
0.00029 0.00937 0
0.00012 0.00937 0
0.00003 0.00937 0
0.00033 0.00937 0

 

 

 

110

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00446 0.00937 0
0.00007 0.00937 0
0.00163 0.00937 0
0.00110 0.00937 0
0.00056 0.00937 0
0.08325 0.00937 1

. 0.00937 1
0.00071 0.00937 0
0.00000 0.00937 0
0.16525 0.00937 1
0.00104 0.00937 0
0.00305 0.00937 0
0.00000 0.00937 0
0.03915 0.00937 1
0.00084 0.00937 0
0.00003 0.00937 0
0.00013 0.00937 0
0.00014 0.00937 0
0.00061 0.00937 0
0.02975 0.00937 1
0.00213 0.00937 0
0.00001 0.00937 0
0.00021 0.00937 0
0.00027 0.00937 0
0.00002 0.00937 0
0.00005 0.00937 0

. 0.00937 1
0.00048 0.00937 0
0.00049 0.00937 0
0.00154 0.00937 0
0.00010 0.00937 0
0.00041 0.00937 0
0.00007 0.00937 0
0.00003 0.00937 0
0.00001 0.00937 0
0.00023 0.00937 0
0.00011 0.00937 0
0.00081 0.00937 0
0.00030 0.00937 0
0.00000 0.00937 0
0.00016 0.00937 0
0.00006 0.00937 0
0.00085 0.00937 0

 

 

 

111

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00013 0.00937 0
0.00002 0.00937 0

. 0.00937 1
0.00481 0.00937 0
0.00045 0.00937 0
0.00000 0.00937 0
0.00411 0.00937 0
0.00137 0.00937 0
0.00000 0.00937 0
0.00013 0.00937 0
0.00028 0.00937 0
0.00001 0.00937 0
0.00009 0.00937 0
0.00018 0.00937 0
0.00104 0.00937 0
0.00081 0.00937 0
0.13491 0.00937 1
0.00136 0.00937 0
0.00151 0.00937 0
0.00170 0.00937 0
0.00016 0.00937 0
0.00048 0.00937 0
0.00036 0.00937 0
0.00295 0.00937 0
0.00003 0.00937 0
0.00014 0.00937 0
0.00014 0.00937 0
0.00048 0.00937 0
0.00022 0.00937 0
0.00028 0.00937 0
0.00001 0.00937 0
0.00240 0.00937 0
0.00004 0.00937 0
0.00005 0.00937 0
0.00008 0.00937 0
0.00001 0.00937 0
0.00001 0.00937 0
0.00014 0.00937 0
0.00331 0.00937 0
0.00003 0.00937 0
0.00016 0.00937 0
0.00002 0.00937 0
0.16209 0.00937 1

 

 

 

112

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00896 0.00937 0

. 0.00937 1
0.00027 0.00937 0
0.00000 0.00937 0
0.00038 0.00937 0

. 0.00937 1
0.00122 0.00937 0
0.01037 0.00937 1
0.00013 0.00937 0
0.00029 0.00937 0
0.00027 0.00937 0
0.00022 0.00937 0
0.00001 0.00937 0
0.01513 0.00937 1
0.00003 0.00937 0
0.00005 0.00937 0
0.00021 0.00937 0
0.00002 0.00937 0
0.00077 0.00937 0
0.00307 0.00937 0
0.00000 0.00937 0
0.00064 0.00937 0
0.00036 0.00937 0
0.00015 0.00937 0
0.00000 0.00937 0
0.00183 0.00937 0
0.00006 0.00937 0
0.00009 0.00937 0
0.00149 0.00937 0
0.00109 0.00937 0
0.00001 0.00937 0
0.00259 0.00937 0
0.00020 0.00937 0
0.00071 0.00937 0
0.00793 0.00937 0
0.00202 0.00937 0
0.00005 0.00937 0
0.00037 0.00937 0
0.00006 0.00937 0
0.00005 0.00937 0
0.00001 0.00937 0
0.00062 0.00937 0
0.00574 0.00937 0

 

 

 

113

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00011 0.00937 0
0.00225 0.00937 0
0.00010 0.00937 0
0.00030 0.00937 0
0.00013 0.00937 0
0.00036 0.00937 0
0.00007 0.00937 0

. 0.00937 1
0.00088 0.00937 0
0.00046 0.00937 0

. 0.00937 1
0.00633 0.00937 0
0.00104 0.00937 0
0.00000 0.00937 0
0.00000 0.00937 0
0.00065 0.00937 0
0.00001 0.00937 0
0.00001 0.00937 0
0.00035 0.00937 0
0.00119 0.00937 0
0.00004 0.00937 0
0.00011 0.00937 0
0.00001 0.00937 0
0.00008 0.00937 0
0.00003 0.00937 0
0.00023 0.00937 0
0.00052 0.00937 0
0.00000 0.00937 0
0.00927 0.00937 0
0.00168 0.00937 0
0.00054 0.00937 0
0.00019 0.00937 0
0.00212 0.00937 0
0.00026 0.00937 0
0.00005 0.00937 0

. 0.00937 1
0.00100 0.00937 0
0.00001 0.00937 0
0.00000 0.00937 0
0.00134 0.00937 0
0.00205 0.00937 0
0.00606 0.00937 0
0.00137 0.00937 0

 

 

 

114

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00001 0.00937 0
0.00460 0.00937 0
0.00009 0.00937 0
0.00007 0.00937 0
0.00006 0.00937 0
0.00004 0.00937 0
0.00056 0.00937 0
0.00001 0.00937 0
0.00025 0.00937 0
0.00010 0.00937 0
0.00011 0.00937 0
0.00003 0.00937 0
0.00001 0.00937 0
0.00010 0.00937 0
0.00068 0.00937 0
0.00005 0.00937 0
0.00011 0.00937 0
0.00120 0.00937 0
0.00009 0.00937 0
0.00073 0.00937 0
0.00008 0.00937 0
0.00033 0.00937 0
0.00008 0.00937 0
0.00010 0.00937 0
0.00051 0.00937 0
0.00000 0.00937 0
0.00006 0.00937 0
0.00022 0.00937 0
0.00013 0.00937 0
0.00004 0.00937 0
0.06400 0.00937 1
0.00149 0.00937 0
0.00078 0.00937 0
0.01259 0.00937 1
0.01249 0.00937 1
0.00018 0.00937 0
0.00000 0.00937 0
0.00000 0.00937 0
0.00317 0.00937 0
0.00000 0.00937 0
0.00000 0.00937 0
0.00170 0.00937 0
0.00078 0.00937 0

 

 

 

115

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00010 0.00937 0
0.00124 0.00937 0
0.00005 0.00937 0
0.00239 0.00937 0
0.01435 0.00937 1
0.00693 0.00937 0
0.00201 0.00937 0
0.00498 0.00937 0
0.00032 0.00937 0
0.00004 0.00937 0
0.00009 0.00937 0
0.00039 0.00937 0
0.00074 0.00937 0
0.00006 0.00937 0
0.00077 0.00937 0
0.00078 0.00937 0
0.00000 0.00937 0
0.00131 0.00937 0
0.00000 0.00937 0
0.00232 0.00937 0
0.00067 0.00937 0
0.00665 0.00937 0

. 0.00937 1
0.00006 0.00937 0
0.00005 0.00937 0
0.00269 0.00937 0
0.00021 0.00937 0
0.00000 0.00937 0
0.00000 0.00937 0
0.01324 0.00937 1
0.00059 0.00937 0
0.00336 0.00937 0
0.00911 0.00937 0

. 0.00937 1
0.00379 0.00937 0
0.00141 0.00937 0
0.00441 0.00937 0
0.00501 0.00937 0
0.00046 0.00937 0
0.00010 0.00937 0
0.00085 0.00937 0
0.00037 0.00937 0
0.00125 0.00937 0

 

 

 

116

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00003 0.00937 0
0.00002 0.00937 0
0.00012 0.00937 0
0.00488 0.00937 0
0.00181 0.00937 0
0.00103 0.00937 0
0.00022 0.00937 0
0.00092 0.00937 0
0.00000 0.00937 0
0.00001 0.00937 0
0.00000 0.00937 0
0.00017 0.00937 0
0.00001 0.00937 0
0.00139 0.00937 0
0.00609 0.00937 0
0.00109 0.00937 0
0.00012 0.00937 0
0.00220 0.00937 0
0.00345 0.00937 0
0.00065 0.00937 0
0.00602 0.00937 0
0.00638 0.00937 0
0.00015 0.00937 0
0.00000 0.00937 0
0.00006 0.00937 0
0.00002 0.00937 0
0.00383 0.00937 0
0.00020 0.00937 0
0.00010 0.00937 0
0.00166 0.00937 0
0.00665 0.00937 0
0.00088 0.00937 0
0.00122 0.00937 0
0.01119 0.00937 1
0.00107 0.00937 0
0.02639 0.00937 1
0.00019 0.00937 0
0.00014 0.00937 0
0.00222 0.00937 0
0.00016 0.00937 0
0.02902 0.00937 1
0.00006 0.00937 0
0.00040 0.00937 0

 

 

 

117

 

Appendix A (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0.00003 0.00937 0
0.00049 0.00937 0
0.00013 0.00937 0
0.00012 0.00937 0
0.00003 0.00937 0
0.00001 0.00937 0
0.00045 0.00937 0
0.00001 0.00937 0
0.00079 0.00937 0
0.00003 0.00937 0
0.00023 0.00937 0
0.00003 0.00937 0
0.00006 0.00937 0
0.00057 0.00937 0
0.00031 0.00937 0
0.00009 0.00937 0
0.00119 0.00937 0
0.00003 0.00937 0
0.00080 0.00937 0
0.00449 0.00937 0
0.00063 0.00937 0
0.00277 0.00937 0
0.00016 0.00937 0
0.00023 0.00937 0
0.00113 0.00937 0
0.00291 0.00937 0
0.00099 0.00937 0
0.00015 0.00937 0
0.02236 0.00937 1
0.00003 0.00937 0
0.00030 0.00937 0
0.00007 0.00937 0
0.00079 0.00937 0
0.00028 0.00937 0
0.00017 0.00937 0
0.00003 0.00937 0
0.00057 0.00937 0
0.00004 0.00937 0

 

 

 

118

 

REFERENCES

Adams, A. W. and Yellen, J. (1976), Commodity bundling and the burden of monopoly,
Quarterly Journal of Economics 90: 475 — 498.

Anderson, S.P., de Palma, A., and Thisse, J .F. (1992). Discrete Choice Theory of Product
Differentiation, MIT Press, Cambridge.

Barnes, B. E. (1990). Electronic media audience behavior in the multichannel
environment: Patterns of demographic homogeneity and time spent viewing.
Unpublished doctoral dissertation, Northwestern University, Evanston, IL.

Beebe, J. (1977). Institutional structure and program choices in television markets.
Quarterly Journal of Economics 91: 15-3 7.

Berry, S., J. Levinsohn and A. Pakes (1995). Automobile Prices in Market Equilibrium,
Econometrica 63(4): 841-89.

Cabletelevision Advertising Bureau (1989). Cable Television Handbook 1989. New
York: Author.

Choi, E. K. (2005). Paper on the Regulation on the Broadcast Advertising Market [Bang-
Song-Gwang-Go Si-Jang Gyu-Jea-Won-Cheek-ae Gwan-Han So-Go],
Broadcasting Research, Special Issue 7 (27). downloaded from
http://www.kbc.go.kr/ on June 2, 2006.

Epple, D. (1987). Hedonic Prices and Implicit Markets: Estimating Demand and Supply
Functions for Differentiated Products, Journal of Political Economy, University
of Chicago Press, vol. 95(1), pages 59-80, February.

Freeman, A. (1993). The Measurement of Environmental Resource Values: Theory and
Methods. Washington: Resources for the Future: 516.

F eenstra, R. (1993). Measuring the Welfare Effect of Quality Change: Theory and
Application of Japanese Autos, NBER Working Papers 4401 , National Bureau of
Economics Research, Inc.

Fisher, F, McGowan, J ., & Evans, D. (1980). The audience-revenue relationship for local
television stations. Bell Journal of Economics 11: 694-708.

Garrod, G. (1994). Using the Hedonic Pricing Model to Value Landscape Features.
Landscape Research 19: 26-28.

Garrod, G. and Willis, K. (1992). Valuing the Goods Characteristics - An Application of

the Hedonic Price Method to Environmental Attributes, Journal of Environmental
Management 34 (1): 59-76.

119

Garson, 2007. Multiple Regression, Downloaded August 20, 2007, from
http://www2.chass.ncsu.edu/garson/PA765/regress.htm

Gensch, D. and Ranganathan, B. (1974), Evaluation of Television Program Content for
the Purpose of Promotional Segmentation, Journal of Marketing Research
11(Nov): 390-398.

Griliches, Z. (1971). Introduction: Hedonic Price Indexes Revisited in Price Indexes And
Quality Changes: Studies In New Methods Of Measurement (Ed) Z. Griliches.

Cambridge, pp. 3-15.

Guiltinan, J. (1987). The price bundling services: a normative framework. Journal of
Marketing 51(2): 74-85.

Fournier, G. & Martin, D. (1983). Does government-restricted entry produce market
power?: new evidence from the market for television advertising. The Bell
Journal of Economics 14 (1): 44-56.

J eon, B-S. (2004). Research on the Restructuring of the Broadcast Advertising System
[Bang-Song-Gwang-Go Je-Do Gae-Sun-Bang-An Yeon-Gu], A Forum for
Restructuring of the Broadcast Advertising System, hosted by Korean
Broadcasting Commission.

Jung, H-T. (2005). The Prospectus of Broadcast Advertising Market attributed to the
discontinuance of KOBACO monopoly system. presented at KAA Annual
Conference 2005. The summary cited from KAA Journal.

Jung, G-J. (2007). Urgency of Solving the Broadcast Advertising Policy Retrogressive to
Market Change [Hwan Kyoung Byun Hwa ae Yeok Hang Ha nun Gwang Go
Jung Cheak Hae Gyul Si Gup]. The Studies of Broadcasting Culture (January).

KBC Report 2005.

KF TC (2004). KOBACO’s Monopoly Advertising Sales is A Typical Unfair Trade
[KOBACO-cu Dock-Jum-Gwant-Go Pan—Mae-nun Dae-Pyo-Juck Bul-Gong-
Jung-Gu-Rae]. KAA.

Kim, K-W. (2004). Broadcast Advertising: Size is Established but System is Developing
[Bang Song Gwang Go: Gyu Mon nun Sun Jin Kook, Jae Do nun Hoo Jin Kook].
The Studies of Broadcasting Culture (July).

Kim, S-K., Kim, J-H., and Lee, S-S. (2003). The Policy for Competition in the Broadcast
Advertising Market [Bang Song Gwang Go Si Jang eu Gyung Jaeng Jung Chaek],
Seoul National University, Center for Corporate Competitiveness. Report 03-01.
Downloaded June 1, 2006, from http://ccc.snu.ac.kr/

120

Kim, J -H. (2006). The Review on the Broadcast Advertising System Fitting into the
Digital Era [Digital Si Dae ae Maat nun Bang Song Gwang Go Jae Do Gum To].
World Media News, April.

Kim, J -K. (1999). Suggestion: Radio Audience Ratings and the Meaning of The research
of Radio Audience. [J ae-eun: Radio Chung-Chuy-Yul-gwa Chung-Chuy-Za Cho-
Sa-eu Eu-Mee]. Producer‘Joumal [Bang-Song-Si-Dea], 16. Retrieved February 25,
2007, from
http://wvvw.pdnet.or.kr/dataroom/dataroom3_2.asp‘?volidx=l 5&idx=l 22&pagenu
m=1.

Kim, S-H. (2006). Prospering the Media Industry and Advertising Policy in the
Terrestrial Broadcasting Market. [Media San Up Hwal Sung Hwa wa Jee Sang Pa
Gwang Go Jung Cheak], Studies of Broadcasting Culture (October).

Kim, Y-J. and Lee, W-J. (2004). Media Competition Surrounding the Advertising
Market: The Current States and Trends under the Advertising Price Regulation
[Gwang-Go-Si-Jang-eul Dul-Ru-San Mae-Che-Gan Gyung-Jang: Gwang-Go-
Ryo-Yul Gyu-Jae-System Ha-ae-seo-eu Gwang-Go-Si-Jang Hyun-Hwang-Gwa
Byun-Dong-Choo-Yi-rul Jung-Sim-eu—ro], presented at the Transitional Korean
Journalism Seminar. Korean Society for Journalism & Communication Studies
2004.

KOBACO (2006). Introduction to Venture and Inno-biz Broadcasting Advertising
[Venture Inno-biz Gwang-Go An-Nae]. Downloaded from
http://www.kobaco.co.kr/kor/broadcasting/market/market_venture.asp on July 17,
2006.

KOBACO (2007). The Types of Broadcast Advertising and Sales [Bang Song Gwang Go
Yoo Hyung Mit Pan Mae Bang Sik]. Downloaded from
http://www.kobaco.co.kr/businessintro/businessfbrod_ad_sale.asp on June 9, 2007.

Koschat, M. and Putsis Jr., W. (2002). Audience Characteristics and Bundling: A hedonic
analysis of magazine advertising rates. Journal of Marketing Research 39(2), pp.
262-273.

Lancaster, K. (1966). A New Approach to Consumer Theory," Journal of Political
Economy, 74(2), pp. 132-157.

Lancaster, K. (1971). Consumer Demand: A new approach. New York: Columbia
University Press.

Lee, G-W. (1995)*, The Broadcast Advertising System in the Netherlands [Netherlands-
eu Bang-Song Gwang-Go], KOBA CO Broadcast Advertising Research Series (6):
190.

121

Lee, G-W. (2000). Research on the Advertising price restructuring Considering the
changes of media influence [Bang-Song Mae-Che-Ryuk Byun-Hwa-ea Dda-run
Gwang-Go Yo-Geum Jo-Jung-Bang-An Yeon-Gu]. KOBACO, Research Project:
28.

Lee, H. K (1995). Study on the Efficiency of Advertising Price of Television Program
[Television Program Gwang-Go-Yo-Gum Hyo-Yul-Sung-ae Gwan-Han Yeon-
Gu], The Korean Journal of Advertising 6 (1): 225-249.

Lee, H. K. (1997). Study Comparing the Advertising Price of 4 Major Media [4-Dae
Mae-Che Gwang-Go-Yo-Gum Hyo-Yul-Sung Bee-Gyo Yeon-Gu]. The Korean
Journal of Advertising 8 (1): 171-194.

Lee, H. K. (2004). Study on the Pricing Structure of Television Advertising [Television
Gwang-Go Yo-Gum-Gu-Jo-ae Gwan-Han Yeon-Gu: Shi-Gup-eul Jung-Shim—eu-
ro]. The Korean Journal of Advertising 15(1): 13 1-151.

Levin, H. (1980). Fact and fancy in television regulation: An economic study of policy
alternatives. New York, NY: Russell Sage Foundation.

McAfee, R., McMillan, J. and Whinston, M. (1989). Multiproduct monopoly, commodity
bundling, and correlation of values. Quarterly Journal of Economics 104(2): 371-
383.

Min, K-S. (1995), Broadcast Advertising in England [Young-Kook-eu Bang-Song
Gwang-Go]. KOBA C 0 Broadcast Advertising Research Series: 151.

Monroe, KB. (1990). Pricing: Marketing Profitable Decisions. McGraw Hill, New York.
Recited from J edidi et al. (2003): Measuring Heterogeneous Reservation Prices
for Product Bundles, Marketing Science 22 (1): 107-130.

Napoli, RM. (2003). Audience Economics: Media Institutions and the Audience
Marketplace, Columbia University Press.

Owen, B., and Wildman, S. (1992). Video economics. Cambridge, MA: Harvard
University Press.

Park C-P. (2004). The System of Broadcast Advertising Sales. KOBACO.

Park, J-R. (2005). The Paradigm Shift in Advertising Media and Media Research - What
could be a reasonable way to go? [Kwang—Go Media eu Paradigm Byun-Hwa wa
Media Research - Ba Ram Jick Han Byun-Hwa nun Moo But 11 Ka?], KAA Focus
2005, Retrieved May 19, 2007 from
http://64.23 3.167.104/search‘?q=cache:R6Uesz gL4J :mvwkaaorkr/k/mag/Z 00

122

5/0l_02/20050102KAA__06.pdf+TNS,+%EB%AF%B8%EB%94%94%EC%96%
B4%EB%A6%AC%EC%84%9C%EC%B9%98&hl=en&ct=clnk&cd=2&gl=us.

Park, S-K. (1996). Determinants of renewal and adjustment of license fees of network
prime time programs. The journal of media economics 9 (3), pp. 1-19.

Park, W-K. and Lee, G-W. (2007). Study on the Broadcast Ad Pricing and Sales System:
Comparative study on Korea and Other Countries. [Bang Song Gwang Go Yo
Geum gwa Pan Mae Bang Sik ae Gwan Han Yeon Gu: Gook Nae Wae Bee Gyo
rul Jung Sim eu ro], Spring Symposium for Advertising Studies, held on April 6th,
2007.

Pakes, A. (2003). A Reconsideration of Hedonic Price Indices with an Application to
PC’s, American Economic Review 93 (5): 1578-1596.

Palrnquist, R. (1991). Hedonic Methods, in Braden, J. B. and Kolstad C. D. (Ed.), 1991,
Measuring The Demand For Environmental Quality, North-Holland, The
Netherlands.

Poltrack, D. (1983). Television Marketing: Network, Local and Cable. New York, NY:
McGraw-Hill.

Ratchford, B. (1 97 5). The new economic theory of consumer behavior: an interpretive
essay. The Journal of Consumer Research 2(2): 65-75.

Rosen, S. (1974). Hedonic prices and implicit markets: Product differentiation in pure
competition, Journal of Political Economy 82: 34-55.

Schmalensee, R. 1984. Gaussian demand and commodity bundling. Journal of Business
57: $211-$230.

Spence, M. and Owen, B. (1977). Television Programming, Monopolistic Competition,
and Welfare. Quarterly Journal of Economics 91(1):]03-126.

Steiner, W. (1952). Program patterns and preferences and the workability of competition
in radio broadcasting. Quarterly Journal of Economics 66: 194-233.

Tabachnick G. and F idell, L. (1996). Using Multivariate Statistics, third ed. New York:
HarperCollins.

Takada, H. and Henry, W. (1993). Analysis of Network TV Commercial Time Pricing for
Top-Rated Prime Time Programs. Journal of Current Issues and Research in
Advertising 15(2): 59-70.

The Korea Broadcast Law (2006).

123

The Order of Korea Broadcast Law (2006).

Vanslembrouck, 1., Van Huylenbrouck, G. and Verbeke, W. (2002). Determinants of the
Willingness of Belgian farmers to Participate in Agri-environmental Measures,
Journal of Agricultural Economics 53: 489 —51 1.

Venkatesh, R. and Kamakura, W. (2003). Optimal Bundling and Pricing under a
Monopoly: Contrasting Complements and Substitutes from Independently Valued
Products. Journal of Business 76(2): 21 1-231.

Waterman, D. and Yan, M. (1999). Cable Advertising and the Future of Basic Cable
Networking. Journal of Broadcasting & Electronic Media 43 (4): 645.

Webster, J. and Lichty, L. (1991). Ratings analysis: Theory and practice. Hillsdale, NY:
Lawrence Erlbaum Associates.

Wildman, 8. (Spring 2003). Modeling the Ad Revenue Potential of Media Audiences: An
Underdeveloped Side of Media Economics, Journal of Media Economics &
Culture 1(2): 7-37.

Wirth, M. and Bloch, H. (1985). The broadcasters: The future role of local stations and
the three networks. In E. Noam (Ed.), Video media competition: Regulation,
economics and technology (pp. 121-137). New York, NY: Columbia University
Press.

Yang, B-H. (2007). It is not an Option but a Must [Bang Song Gwang Go Jae Do Gae
Sun eun Sun Tack a nin Sang Zone]. The Studies of Broadcasting Culture
(March).

Yoo, H-S. (2005). Let the Market determine Broadcast Advertising Time Sales [Bang-

Song-Gwang-Go Pan-Mae-nun Shi-Jang-Gyung-Jaeng-Won-Ry-ae Mat-Gye-Ya-
Han-Da]. Downloaded from wwwadvertisingco.kr on June 2, 2006.

124

111111111111111)111111