Broadcasts of sporting events account for a large proportion of the revenue source for both networks and successful professional sport leagues and teams (Fort, 2006). The role of media in delivering sport content seems to be gaining further importance as professional sport leagues seek to expand their influence domestically and internationally. This importance has been reaffirmed since the outbreak of the COVID-19 pandemic, as the absence of stadium ticket sales has forced networks and professional leagues to rely on revenue sources beyond the stadium. Accordingly, the revenues generated from the sales of media rights and from media content distribution of live sporting events have become more important than ever (World Economic Forum, 2020). Finding ways to satisfy fans’ desires for media consumption of sport content will be pivotal to improve the future of the sport business.
Most sport demand studies are rooted in attendance research, which investigates the determinants of stadium attendance in various professional sports, including the European football leagues and North American college and professional sports (Borland & MacDonald, 2003; Szymanski, 2003; Villar & Guerrero, 2009). However, the attendance data present inherent problems with understanding the overall demand for professional sport, such as the possible bias season ticket holders cause, unobservable and excessive demands for sport that go beyond stadium capacity, and a generally high proportion of home supporters (Feehan et al., 2003; Forrest et al., 2005). Since previous works have found sport fans watching televised sporting events to have heterogenous preferences compared to those of fans attending games at a stadium, a separate investigation of viewership is needed to draw a more accurate implication (Mongeon & Winfree, 2012; Sung et al., 2019). Viewership studies have gained much attention in recent years to overcome such shortcomings, as well as to comprehend the ever-increasing size and significance of the sport media market (Ryu et al., 2019; Tainsky, 2010; Tainsky & Jasielec, 2014).
Most viewership studies use television ratings as a proxy for live sport demand (Alavy et al., 2010; Pérez et al., 2017; Tainsky & Jasielec, 2014; Wang et al., 2018). However, the immediate client for any professional sport is the broadcasting network, which acts as a patron, providing matches for its viewers (Evens et al., 2013). Only after the media company has purchased the media rights, and after the broadcaster has selected a match for televising, does a match become available to the audience for viewing. Therefore, it is of great importance to understand on what grounds broadcasters choose matches to televise, and whether broadcasters’ choices of matches are associated with the actual audience demand as reflected in the television ratings.
Unfortunately, there are only a few studies (Forrest et al., 2005, 2006) that have focused on broadcasters’ choices of sporting events. These studies, despite providing useful insights to researchers and practitioners, are rather outdated and are restricted to a single league in a specific period: the English Premier League (EPL) during its inception in the 1990s. Because the cultural background of professional sport fans and the socioeconomic environment surrounding each professional league vary from one country to another (Agah & Dixon, 2021; Jang, 2019; Ryu et al., 2019), the results of the aforementioned studies may limit the understanding of the latest trends in sport media consumption in other sport leagues. Specifically, each sport league within and across borders has different policies for broadcasting sporting events (Ma & Kurscheidt, 2020). European countries and the United States have varying policies regarding the accessibility of sporting events, degree of government intervention, and competition policies (Hoehn & Lancefield, 2003).
This study aims to fill the gap in the literature by investigating broadcasters’ choices of matches for televising and their relation to actual television viewership. First, we examined the determinants of television viewership and its relation to broadcasters’ choices of matches for live telecasts in order to understand whether the games to which broadcasters give a higher priority are indeed associated with higher television ratings. Sport broadcasters are rational decision makers and artistic producers of sport programs. Therefore, whether their choices of matches lead to higher audience demand is an open empirical question. Second, we examine the factors driving broadcasters’ choices of matches to be televised, in order to understand their general decision patterns. The former is investigated through the “viewership model” and the latter is investigated through the “broadcaster choice model.”
This study is conducted in the context of the Korea Baseball Organization (KBO), the highest-level professional baseball league, serving in Korea since 1982. The KBO provides a unique research setting not available in other professional leagues in that, with only a few exceptions, all matches in a season are televised nationally, and broadcasters are faced with a decision about selecting which matches to televise the next week and throughout the season. This distinctive structure allows broadcasters to choose which games to televise on a regular basis, making the KBO best suited to studying their decisions without having to study those of other competitors in the broadcasting rights market. The accumulated data based on the broadcasters’ weekly decisions during a KBO baseball season can provide practitioners and academicians with valuable insights to understand the general decision-making patterns. In addition, while the previous studies on networks’ selection of games to broadcast focus on North American and European sports broadcasting (Forrest et al., 2005), the current study contributes to the literature by extending this stream of research to Asian markets.
Literature Review
Determinants of Viewership for Sport
Most of the determinants of demand used to explain gate attendance in previous works could also explain a major part of viewership, while the effect of these determinants may show dissimilar trends (Tainsky, 2010). As one of the many determinants of the demand for sport, the uncertainty of outcome hypothesis presented by Rottenberg (1956) has driven a body of literature on understanding sport fans’ choices and preferences. Although Rottenberg’s original work was based on gate attendance demand, much of the viewership literature has also incorporated the topic to understand fans’ responses to outcome uncertainty. For instance, Buraimo and Simmons (2009) and Cox (2018) found that fans watching games on television preferred matches that were closely competitive or unpredictable, whereas this effect was unobservable in gate attendance.
More specific to the KBO viewership, Chung et al. (2016) examined the difference between pregame expectations and within-game expectations of outcome uncertainty. The authors found that fans were more responsive to predicted winning probability (i.e., the pregame expectation of outcome uncertainty) before a game started, whereas they were more responsive to real-time changes in winning probability (i.e., the within-game expectation of outcome uncertainty) after a game started. Similarly, Ryu et al. (2019) found that higher ratings of televised KBO matches were associated with greater outcome uncertainty. Additionally, Jung et al. (2020) found asymmetric responses from KBO attendance data: Home team fans showed less interest when the home team was more likely to win (i.e., less outcome uncertainty). However, there was no such effect when the visiting team was more likely to win. These findings align with those of Meehan et al. (2007) that fans tend to be attracted to a well-performing visiting team regardless of the outcome uncertainty. However, evidence of outcome uncertainty is still convoluted because many other studies have reported no evidence of uncertainty of outcome hypothesis (Buraimo & Simmons, 2015; Paul & Weinbach, 2015; Pérez et al., 2017; Sung et al., 2019). Overall, uncertainty of outcome hypothesis is a form of speculation of fan preferences, which are assumed to be inherently different from individual to individual. Thus, further investigation is needed to reveal fan preferences in various contexts, as Fort (2017) noted.
Common factors that could be considered with respect to the demand for sport include the effects of superstars (i.e., star athletes); proximity between competing teams (i.e., distance); rivalry; and duration of a sport team’s presence (i.e., team age). First, the superstar effect is characterized by a few individuals providing a large return based on their highly valued talent (Rosen, 1981). That is, in the sport context, a handful of athletes who show on-field prowess will earn higher salaries compared to other less talented athletes, making them a valuable product. Further, superstars generate externalities not only for their home teams but also for the overall league. Hausman and Leonard (1997) estimated the value of Michael Jordan for the National Basketball Association and found it to be more than 50 million dollars—more than any other player in the National Basketball Association. This superstar effect was also found in European soccer leagues (Buraimo & Simmons, 2015; Czarnitzki & Stadtmann, 2002), Major League Soccer (DeSchriver et al., 2016; Lawson et al., 2008; Sung & Mills, 2018), and Major League Baseball (Lewis & Yoon, 2018; Mullin & Dunn, 2002; Ormiston, 2014).
Distance between competing teams also has an impact on demand. According to Hotelling’s (1929) location model, demand for a certain firm is dependent on the distance between consumer and product. That is, all else being equal, products that are farther away from a consumer become less ideal as the associated costs may increase. Hence, a negative linear relationship between consumer demand and distance exists. Empirically, previous studies have shown that an increase in distance is negatively associated with attendance demand, and this effect becomes more pronounced as the season progresses (Forrest & Simmons, 2002; Lemke et al., 2010). This effect could also be related to substitution behavior, according to which sport fans switch their favorite teams based on the distance of their residence to the sport team. Winfree et al. (2004) found that introducing a new team in a region with an existing one can lead to a decrease in fans for the existing team because the new team may absorb this loss. In terms of viewership markets, Tainsky and McEvoy (2012) found that distance had a negative impact on demand; specifically, when no home team match was televised, out-of-market teams that were located closest to fans were preferred.
Relevant to distance, literature suggests that rivalry can boost fan interest in matches (Beckman et al., 2012; Buraimo & Simmons, 2008; Lemke et al., 2010; Paul, 2003). However, the definition of “rival” is different across leagues. There are various factors that influence the formation of rivalry, such as history, geographical proximity, frequency of meeting, and similarity (Kilduff et al., 2010; Tyler & Cobbs, 2015). Sung et al. (2017) urged that the investigation of sport rivalry in demand studies should take intensity, direction, and duration into account. In other words, how strongly a fan feels about the rivalry, whether both teams’ fans feel the same way or not, and how long the rivalry has been in existence are important factors to be taken into consideration when examining the influences of rivalry.
Finally, sport teams with a longer history and existence influence fan loyalty and, thus, viewership (Coates & Humphreys, 2005). Over time, a team with a longer presence in a particular region is more likely to form a loyal fan base (Borland & MacDonald, 2003). Because higher fan loyalty implies a greater demand for the team, previous demand studies have widely used team age as a proxy to measure the effects of fan loyalty on demand (Coates & Humphreys, 2005; Sung & Mills, 2018; Tainsky & Jasielec, 2014; Wooten, 2018).
Determinants of Broadcasters’ Demands and Choices
For the last several decades, the exposure of televised sport content has increased, and, as a result, so have the media rights fees because of the uptick in demand from more broadcasters (Noll, 2007). In particular, Noll speculated that the growth in commercial broadcasters has increased the demand for broadcasting rights. Hence, the fees for the right to air a program and airing hours have significantly increased. However, the objectives of public and commercial broadcasters differ from each other. Public service broadcasters, established to support the public welfare, are more concerned with program diversity and providing programs of merit goods characteristic than commercial broadcasters, which are primarily interested in profit maximization (Solberg, 2007). For this, many public service broadcasters receive license fees (e.g., BBC in the United Kingdom), whereas commercial broadcasters seek to maximize television ratings to pursue the highest possible revenues from advertisements and subscriptions (Gratton & Solberg, 2007).
Unfortunately, research on the determinants that influence broadcasters’ choices of matches to televise is limited. Among the dearth of literature, Forrest et al. (2005) examined the choices of British broadcasters for EPL games. They found that broadcasters preferred matches involving contenders for the EPL championship or European Championship League qualification, matches between local rivals, and matches held on weekends. Also, matches with greater unpredictability, and a larger sum but a smaller gap between the competing teams’ wages, were selected by broadcasters with higher priority. Forrest et al. (2006) examined broadcasters’ choices for EPL matches based on a five-season data set and reported a similar pattern of results.
Broadcasting of the KBO League and Research Questions
Established in 1982, the professional KBO league has a long history in Korea, and it is the most popular professional sport league in Korea (KBO, 2020). The KBO league was founded with six teams and in 2015 it expanded to 10 teams. Unlike Major League Baseball in North America, comprised of three divisions (i.e., West, Central, and East Divisions), the KBO league comprises a single division of 10 teams. There is also a minor league called the KBO Futures League, which is similar to Minor League Baseball in the United States.
Until 1994, only national terrestrial broadcasting networks were able to televise KBO matches (Ha, 2014). Subscription-based cable and satellite channels started broadcasting KBO games in 1995 and 1997, respectively. In the early 2000s, sport channels were established in the pay TV market, expediting the expansion of the sport broadcasting market.
The uniqueness of the KBO television market is that, unlike other professional sport leagues in the United States and Europe, all games, with only a few exceptions, have been televised nationally via pay TV sport channels since 2008. This is because the KBO requires sport channels to air at least 95% of total matches. The three terrestrial channels in Korea have to air a minimum of 10 matches. Both sport channels and the three terrestrial channels provide national coverage, so all KBO matches are televised nationally. Because of pay TV’s reasonable subscription prices, the penetration rate of pay TV services in the Korean broadcast market has reached almost 100%; the basic program package includes all five sport channels responsible for KBO broadcasts at a cost of approximately $20 USD per month (KISDI Report, 2019).
KBO games are scheduled from Tuesday through Sunday. Five games are played each day, and five sport channels are available to televise the matches. To distribute the games without dispute, a system was implemented to determine the order of pick for match selection. At the beginning of each season, the order of pick (e.g., 1–5) is decided randomly through a number draw and assigned to the five channels. The channel with the first pick receives priority in selection, and according to the established order, each channel chooses which game to televise the following week from the available matchups. The order of selection rotates sequentially every week. Thus, if one channel chooses first in one round, it chooses second in the next round and so on. Under these circumstances, a unique situation is created where broadcasters face decisions every week about the most preferred games to televise. The selection process results in a lineup of matches each week that are ranked according to the broadcaster’s choice order, thus providing a unique opportunity to examine the determinants that influence broadcasters’ choices.
RQ1: Does the broadcasters’ choice order for the televised games affect the television ratings of the game?
RQ2: Which factors, besides the broadcasters’ choice order, affect the television ratings of the game?
RQ3: What are the determinants of the broadcasters’ choice order for the games to be televised?
Methods
Data Structure
A panel data set, that is, a data set containing cross-sectional time-series data, was collected from the game record data of the 2018 KBO league pennant races held from March to October 2018. There are 10 teams in the league, and each team plays every other team 16 times, resulting in 144 games for each team. This results in 720 games total. A total of 718 games were televised; the two games not televised by any channel were excluded from the analysis. Further, three terrestrial TV channels in Korea chose 10 games to televise, and the sport channels televised the remaining games. These 10 games were excluded from the analysis because the sport channels’ choices of games to be televised was the key variable of interest. As a result, a total of 708 games were analyzed in the panel data set. The panel variable was the matchup (i = 1, . . . 45), and the time variable was the round of the game (t = 1, . . . 16).
Model and Analyses
Two regression models were examined. The first, the viewership model, examined which factors, including the broadcasters’ choice order, affected a game’s television ratings (i.e., DV = log-transformed TV ratings). The second model, the broadcaster choice model, examined the factors affecting the broadcasters’ choice order for the games to be televised (i.e., DV = broadcasters’ choice order). Broadcasters’ choices for games were decided a week in advance; thus, game choice decisions were made on a weekly basis for the following week’s games. Each of the five available sport channels took turns choosing first out of the five available games each week. In practice, broadcasters may have their own agendas within the market, so the demand for a certain type of content, such as baseball games, may vary from one broadcaster to another. Therefore, an analysis of the data set based on the game choices of all available sports channels and the respective television ratings of the televised games can reveal valuable insights to understand the broadcasters’ decision-making patterns and their relation to the actual audience demand for the live sport broadcast.
Seminal works in sport demand studies, such as Rottenberg (1956), Borland and MacDonald (2003), and Fort (2006), identified factors determining stadium attendance, such as price, income, population, stadium capacity, viewing quality at stadium, game characteristics, substitutes, and consumer preferences. Some of these factors, such as price, stadium capacity, and viewing quality, are only relevant to stadium spectatorship and are not applicable to this study focusing on TV viewership. Additionally, income and population do not apply to the TV audience market of the KBO league because there are no local broadcasts like in the United States (i.e., all games are televised nationally), leaving the income and population factors constant in the TV audience market. Considering the nature of the KBO broadcasting environment, we focused on game characteristics and audience preferences. An overview of the dependent and independent variables is presented in Table 1.
Descriptions of Variables
| Categories | Variables | Descriptions | Type |
|---|---|---|---|
| TV ratings | Ratingsa | Mean TV ratings per minute of the live game | Continuous |
| LnRatings | Log-transformed Ratings | Continuous | |
| Choice order | ChoiceOrder | Broadcasters’ choice order (First choice through fifth choice) | Ordinal |
| Past game attributes | LnLagRatings | Log-transformed mean ratings of the same matchup’s most recent game series | Continuous |
| LagScoreDiff | Mean score difference of the same matchup’s most recent game series | Continuous | |
| HomeRank | Current standing of home team up to most recent game | Continuous | |
| AwayRank | Current standing of away team up to most recent game | Continuous | |
| Current game attributes | SumScore | Sum of both team’s score | Continuous |
| ExtraInnings | Whether the game went into overtime (extra innings = 1 and no extra innings = 0) | Dummy | |
| ScoreDiff | Final score difference | Continuous | |
| Predicted game attributes | Theil | Measure of uncertainty of outcome by Theil (1967) times 100 for interpretation | Continuous |
| BetReverse | Reverse bet outcome defined as any outcome that has turned out to be opposite of a priori information | Dummy | |
| Team attributes | SumTeamAge | Sum of competing teams’ age | Continuous |
| Rival | Whether the matchup is between two rival teams or not (rival = 1 and nonrival = 0) | Dummy | |
| LagPostSeason | Whether the matchup involves a team which appeared in at least one game from the previous season’s postseason games (yes = 1 and no = 0) | Dummy | |
| LnTotalSalary | Log-transformed sum of total salary of both home and away team | Continuous | |
| Allstar | Number of players designated for Allstar game in 2018 | Continuous | |
| External factors | Dista | Distance between the two teams’ home stadiums in kilometers | Continuous |
| LnDist | Log-transformed Dist | Continuous | |
| Weekend | Whether the game is part of the weekday series (weekend series = 1 and weekday series = 0) | Dummy | |
| NightTime | Whether the game was held during the night time (night game = 1 and daytime game = 0) | Dummy | |
| Temp | The highest temperature of the game day in Celsius | Continuous | |
| Weather | Whether the day was sunny (sunny = 1 and cloudy/rainy = 0) | Dummy | |
| Month | Month (March/April, May, June, July, August, September, and October) | Categorical | |
| Network | Network channels (KBS N Sports, MBC Sports+, SBS Sports, SPOTV, SPOTV2, and Multi) | Categorical |
aRaw data before log transformation.
For the viewership model, the dependent variable was log-transformed TV ratings (LnRatings), and the primary independent variable was the broadcasters’ choice order (ChoiceOrder). Other predictors of LnRatings included the following five categories of factors: (a) past records of the same matchup (i.e., past game attributes); (b) the attributes of the game being watched (i.e., current game attributes); (c) predicted on-field performances of each team and the associated match outcomes (i.e., predicted game attributes); (d) attractiveness of the matchup (i.e., team attributes); and (e) other external factors.
The explanatory variables related to past game attributes included LnLagRatings, LagScoreDiff, HomeRank, and AwayRank, and LnLagRatings was the log-transformed mean rating of the same matchup’s most recent game series. The most recent game series referred to the two or three consecutive games between the same matchup teams. KBO usually schedules three games between the same matchup teams in a row, but occasionally organizes only two consecutive games. LagScoreDiff was the mean score difference of the same matchup’s most recent game series. HomeRank and AwayRank were the home and away teams’ standings within the league up to the most recent game.
The explanatory variables related to current game attributes included SumScore, ScoreDiff, and ExtraInnings. SumScore represented the sum of both teams’ scores, and ScoreDiff was the final score difference. ExtraInnings was a dummy variable indicating whether the game went into overtime (extra innings = 1 and no extra innings = 0).
The explanatory variables related to the team attributes included SumTeamAge, Rival, LagPostSeason, LnTotalSalary, and Allstar. SumTeamAge was the sum of competing teams’ ages. Rival was a dummy variable indicating whether the matchup was between two rival teams or not. There were four historical rival matchups in KBO, which were coded as Rival (rival = 1 and nonrival = 0). The rivalry in the KBO did not convert mechanically to the matches between the two geographically adjacent teams; instead, the historical rivalry naturally originated from teams sharing the same hometown or region. LagPostSeason was a dummy variable indicating whether the matchup involved a team that appeared in at least one game from the previous post season (yes = 1 and no = 0). LnTotalSalary measured the log-transformed sum of the home and away teams’ wages of players as a proxy of team quality. Allstar was used as an indicator of the total number of players in a match who were designated as All-Star players in the 2018 season, to account for the superstar effect of high-profile players in the current year.
The variables related to the external influencers included LnDist, Weekend, NightTime, Temp, Weather, Month, and Network. LnDist was the log-transformed distance between the two teams’ home stadiums in kilometers. Weekend was a dummy variable indicating whether the game was a part of the weekday series (weekend series = 1 and weekday series = 0). Weekday series included games held on Tuesday through Thursday, and the weekend series included games held on Friday through Sunday. NightTime was a dummy variable indicating whether the game was held at night (night game = 1 and daytime game = 0). Day games were those that started at 2:00 p.m. and night games started at 6:30 p.m. Temp was the highest temperature of the game day in Celsius. Weather was a dummy variable indicating whether the day was sunny (sunny = 1 and cloudy/rainy = 0). Month represented a categorical variable indicating the month when the game was held (March/April through September). Because there were only a few games held in March, March and April were combined into one category, March/April. Finally, Network was a categorical variable indicating the broadcast channel. The games were televised via one of the five channels or sometimes by multiple channels simultaneously; thus, the total number of categories of the Network was six (five channels plus simultaneous broadcasts by multiple channels). Quite reasonably, broadcasters avoided choosing the game already selected by another broadcaster, so that each channel could televise different games. But there were some exceptional cases when a broadcaster selected a game that was already selected by another broadcaster in an earlier order, resulting in the same game being televised by multiple channels simultaneously; these cases occurred when part of the daily game lineup was canceled due to bad weather conditions. Although only 24 games (3.3% of the 708 games) were televised by multiple channels, it was necessary to control for the effects of simultaneous broadcasting in the regression model by having the variable Network include a category representing the case when multiple channels broadcast the same game.
Regarding the broadcasters’ choice model, it was reasoned that broadcasters would likely select the game matchup which is expected to produce the highest ratings when making a choice about the coming week’s games for telecast, based on either experience or hard data from a similar set of variables as those of the viewership model. Current game attributes are not available to broadcasters who need to make decisions at least a week ahead of the actual games. Also, while a 10-day forecast was available in Korea, the forecast data of the past dates were not available for data collection. Therefore, the three current game attributes (SumScore, ExtraInnings, and ScoreDiff) and the two game-day factors (i.e., Temp, Weather) were excluded from the broadcaster choice model. All other explanatory variables were identical to those in the viewership model.
Findings and Discussion
Descriptive Statistics
Table 2 shows the descriptive statistics of the key variables. The highest television rating for the KBO games was 4.13%, whereas the mean rating was 1%. That is, approximately half a million fans watched KBO games on average. The highest viewership level was recorded at approximately 2.15 million.1 Regarding the game attribute dummy variables, a total of 7% of the games went into extra innings, whereas 9% of games were between rivals. Additionally, about 43% of the total games resulted in reverse outcomes, based on betting predictions. For other external factors, 48% of the total games were played on the weekend, and as many as 86% of the games were played after 6:30 p.m. Games played in sunny weather accounted for almost half the games, whereas the rest were played under cloudy or rainy weather conditions.
Descriptive Statistics of the Key Variables
| Categories | Variables | Mean (frequency)a | SD (%)a | Minimum | Maximum | N | Type |
|---|---|---|---|---|---|---|---|
| TV ratings | Ratingsb | 1.00 | 0.56 | 0.10 | 4.13 | 708 | Continuous |
| LnRatings | −0.17 | 0.61 | −2.26 | 1.42 | 708 | Continuous | |
| Choice order | ChoiceOrder | 2.95 | 1.41 | 1.00 | 5.00 | 708 | Ordinal |
| Past game attributes | LnLagRatings | −0.17 | 0.58 | −1.88 | 1.01 | 592 | Continuous |
| HomeRank | 5.40 | 2.93 | 1.00 | 10.00 | 708 | Continuous | |
| AwayRank | 5.38 | 2.84 | 1.00 | 10.00 | 708 | Continuous | |
| LagScoreDiff | 4.04 | 1.97 | 0.50 | 13.50 | 592 | Continuous | |
| Current game attributes | SumScore | 11.14 | 5.17 | 1.00 | 30.00 | 708 | Continuous |
| ExtraInnings | 0.07 | — | 0.00 | 1.00 | 708 | Dummy | |
| ScoreDiff | 4.11 | 3.11 | 0.00 | 18.00 | 708 | Continuous | |
| Predicted game attributesc | Theil | 29.39 | 0.92 | 15.67 | 30.10 | 708 | Continuous |
| BetReverse | 0.43 | — | 0.00 | 1.00 | 708 | Dummy | |
| Team attributes | SumTeamAge | 45.06 | 15.81 | 12.00 | 72.00 | 708 | Continuous |
| Rival | 0.09 | — | 0.00 | 1.00 | 708 | Dummy | |
| LagPostSeason | 0.22 | — | 0.00 | 1.00 | 708 | Dummy | |
| LnTotalSalary | 23.45 | 0.16 | 23.10 | 23.76 | 708 | Continuous | |
| Allstar | 4.78 | 3.80 | 0 | 15 | 708 | Continuous | |
| External factors | Distb | 210.22 | 126.79 | <0.01d | 405.35 | 708 | Continuous |
| LnDist | 4.82 | 1.94 | −6.91 | 6.00 | 708 | Continuous | |
| Weekend | 0.48 | — | 0.00 | 1.00 | 708 | Dummy | |
| NightTime | 0.86 | — | 0.00 | 1.00 | 708 | Dummy | |
| Temp | 20.95 | 5.89 | 3.90 | 33.40 | 708 | Continuous | |
| Weather | 0.50 | — | 0.00 | 1.00 | 708 | Dummy | |
| Month | — | — | — | — | 708 | Categorical | |
| March/April | 146 | 20.62 | — | — | |||
| May | 119 | 16.81 | — | — | |||
| June | 120 | 16.95 | — | — | |||
| July | 109 | 15.40 | — | — | |||
| August | 63 | 8.90 | — | — | |||
| September | 117 | 16.53 | — | — | |||
| October | 34 | 4.80 | — | — | |||
| Network | — | — | — | — | 708 | Categorical | |
| KBS N Sports | 138 | 19.49 | — | — | |||
| MBC Sports+ | 134 | 18.93 | — | — | |||
| SBS Sports | 140 | 19.77 | — | — | |||
| SPOTV | 135 | 19.07 | — | — | |||
| SPOTV 2 | 137 | 19.35 | — | — | |||
| Multie | 24 | 3.39 | — | — |
aFrequency and percentage for dummy and categorical variables. bRaw data before log transformation. cPredicted game attributes. dTwo teams share the same stadium, and the distance between these two teams was coded as 0.001 km. eMulti indicates the games were televised simultaneously by more than two channels.
Factors Affecting Television Viewership (RQ1–2)
To examine RQ1 and RQ2, the viewership model was analyzed. The log-transformed TV ratings (LnRatings) were regressed on the broadcasters’ choice order (ChoiceOrder), and a predetermined set of independent variables, including five time-invariant variables (SumTeamAge, Rival, LagPostSeason, LnTotalSalary, and Allstar). The number of cases used in the analyses was 592. A total of 116 cases were dropped from the analyses because two variables related to the past game attributes—LnLagRatings and LnScoreDiff—were not available for the season’s opening game series.
The random-effects and FGLS models were used to analyze the viewership model. Because the effect of the time-invariant variable was unidentifiable in the fixed-effects model, the random-effects model was first employed to analyze the data. The results of Breusch–Pagan’s Lagrangian Multiplier test for the random-effects model were statistically significant, χ2(1) = 64.39 (p < .01). Rejection of the null hypothesis of the Breusch–Pagan’s Lagrangian Multiplier test (i.e., the variance of the unique errors [ui] of the panels does not differ from zero) indicated the existence of random effects; therefore, the unique panel characteristics needed to be accounted for in the regression model, which the pooled Ordinary Least Squares was unable to do. In addition, the result of the likelihood ratio test for the heteroscedasticity of error terms was statistically significant, indicating the presence of heteroscedasticity, χ2(44) = 114.77 (p < .01); thus, clustered robust SEs were used for the random-effects regression. Meanwhile, there was no evidence of significant multicollinearity issues, as indicated by the fact that all variance inflation factor values were below four. The random-effects model was significant, χ2(31) = 2516.01 (p < .01); the overall R2 was 84.13%, and the ρ, or the fraction of variance due to the panel variable, was 17.22%.
The random-effects model, however, imposed a strong assumption that the unique errors (ui) of each panel were uncorrelated with the predictors [i.e., cov(Xit, ui) = 0]. The Hausman test is typically used to examine this assumption; the test compares the differences in the coefficients estimated from the random-effects model to those of the fixed-effects model, which does not require the assumption of independent covariance. However, the viewership model in the current study includes several time-invariant variables. The specifications of the random- and fixed-effects models cannot be the same, because the former includes time-invariant variables, whereas the latter does not; for this reason, the Hausman test was not performed.
Instead, we used the FGLS model, which does not require the assumption that the random-effects model imposes, that is, that the unobserved heterogeneity is uncorrelated with the predictors. Further, the FGLS generates more efficient estimators than the Ordinary Least Squares in the presence of heteroscedasticity and autocorrelation in the panel data set (Bai et al., 2021). The Woodridge test for autocorrelation was also performed, and the result was significant, F(1, 44) = 32.89 (p < .01), indicating the existence of a first-order correlation within panels. Therefore, the FGLS model was used while correcting for heteroscedasticity and autocorrelation. The FGLS model was significant, χ2(31) = 2,246.38 (p < .01); heteroscedasticity was allowed, and a common first-order autocorrelation of .359 was used in the FGLS model.
Because both heteroscedasticity and autocorrelations were corrected for in the FGLS model, and the random-effects model requires a strong assumption that the unobserved heterogeneity is uncorrelated with the independent variables, the FGLS model was deemed more appropriate for the given data—even though the analyses of the two models resulted in a similar pattern of results. Table 3 shows the results of both the random-effects and FGLS models, but the following interpretations focus on the results of FGLS model.
Regression Results of Viewership Model
| Variables | Random effects | Robust SE | FGLS | SE | |
|---|---|---|---|---|---|
| Choice order | ChoiceOrder | −0.131*** | 0.016 | −0.155*** | 0.014 |
| Past game attribute | LnLagRatings | 0.131*** | 0.040 | 0.171*** | 0.033 |
| LagScoreDiff | −0.008 | 0.005 | −0.011** | 0.005 | |
| HomeRank | −0.005 | 0.004 | −0.009*** | 0.003 | |
| AwayRank | 0.001 | 0.004 | 0.002 | 0.003 | |
| Current game attribute | SumScore | 0.006*** | 0.002 | 0.005*** | 0.002 |
| ExtraInnings | 0.149*** | 0.036 | 0.135*** | 0.032 | |
| ScoreDiff | −0.031*** | 0.004 | −0.032*** | 0.003 | |
| Predicted game attribute | Theil | −0.001 | 0.006 | 0.009 | 0.009 |
| BetReverse | 0.034* | 0.019 | 0.020 | 0.015 | |
| Team attribute | SumTeamAge | 0.004* | 0.002 | 0.002* | 0.001 |
| Rival | 0.002 | 0.053 | −0.009 | 0.046 | |
| LagPostSeason | -0.064 | 0.053 | −0.068** | 0.031 | |
| LnTotalSalary | 1.580*** | 0.260 | 1.322*** | 0.152 | |
| Allstar | −0.004 | 0.007 | 0.002 | 0.004 | |
| External factors | LnDist | 0.011 | 0.011 | 0.014 | 0.010 |
| Weekend | −0.044* | 0.023 | −0.046** | 0.018 | |
| NightTime | 0.096*** | 0.031 | 0.095*** | 0.024 | |
| Temp | −0.004 | 0.006 | −0.005 | 0.004 | |
| Weather | −0.012 | 0.022 | −0.029* | 0.016 | |
| Montha | — | — | — | — | |
| May | 0.204*** | 0.056 | 0.175*** | 0.046 | |
| June | 0.156** | 0.064 | 0.154*** | 0.051 | |
| July | 0.094 | 0.073 | 0.073 | 0.063 | |
| August | 0.150* | 0.088 | 0.199*** | 0.065 | |
| September | 0.017 | 0.059 | −0.002 | 0.050 | |
| October | 0.018 | 0.082 | −0.004 | 0.064 | |
| Networkb | — | — | — | — | |
| MBC Sports+ | 0.116*** | 0.038 | 0.129*** | 0.030 | |
| SBS Sports | 0.135*** | 0.033 | 0.137*** | 0.031 | |
| SPOTV | -0.002 | 0.033 | 0.001 | 0.031 | |
| SPOTV 2 | −0.276*** | 0.043 | −0.278*** | 0.030 | |
| Multi | 0.204** | 0.085 | 0.107* | 0.060 |
Note. FGLS = feasible generalized least squares.
aReference month: March/April. bReference channel: KBS N Sports; DV = LnRatings; N = 592.
*p < .10. **p < .05. ***p < .01.
With respect to RQ1, ChoiceOrder had a significant and positive impact on LnRatings (βFGLS = −0.155, p < .01). The significant coefficient for broadcasters’ choices of matches (ChoiceOrder) indicates that the broadcasters’ selection priorities were well aligned with the actual television ratings. This result, however, should be interpreted with caution because it does not indicate that the rank of the games decided by the broadcasters, which is not revealed to consumers, changes consumer preference for the games. Rather, the result indicates that broadcasters’ choices of games were well reflected in the future audience demand for the sport telecasts.
Regarding RQ2, it appears that all of the current game attributes were significant in predicting television ratings. However, not all variables related to the past game attributes, the team attributes, and the external influencers had significant impacts on the audience viewership, while none of the predicted game attributes were statistically significant.
Specifically, three of the four past game attributes were statistically significant. LnLagRatings had a positive influence on TV ratings (βFGLS = 0.171, p < .01), indicating the existence of an inertia of interest from the previous matchups. LagScoreDiff (βFGLS = −0.011, p < .01) was also statistically significant. The results indicated that a one-point increase in the mean score difference of the same matchup’s most recent game decreased ratings by 1.1%. That is, KBO fans preferred games where the previous result of the same matchup had a smaller final score difference. Finally, between the home and away teams’ standings in the league, only HomeRank was found to negatively influence ratings (βFGLS = 0.009, p < .01); this result indicates that the matchups involving a home team with higher league standing are associated with greater television viewership.
The three current game attributes, SumScore (βFGLS = 0.005, p < .01;
None of the two predicted game attributes, Theil (the outcome uncertainty) and BetReverse, significantly influenced TV ratings. This contradicts the previous literature, which found fans’ preference for outcome uncertainty in the KBO context (Chung et al., 2016; Ryu et al., 2019).
For the five team attribute variables, it was most apparent that LnTotalSalary (βFGLS = 1.322, p < .01;
With respect to the external influencers, Weekend, NightTime, and Weather appeared to be significant predictors of television ratings. Nighttime games (NightTime, βFGLS = 0.095, p < .01) were associated with higher television ratings. The television ratings for weekend series games were lower than those for weekday series games (Weekend, βFGLS = −.046, p < .05). Further, sunny days were associated with lower television ratings when compared to cloudy or rainy days (βFGLS = −0.029, p < .10), but only at the 90% level. These results suggested the possible substitute behavior of KBO fans; that is, during the daytime, on weekends, and on sunny days, other available substitutes to watching televised baseball games may exist, including physically attending the game or engaging in other non-baseball-related activities.
Factors Affecting Broadcasters’ Choices of Games (RQ3)
To examine RQ3, broadcasters’ choice order (ChoiceOrder) was regressed on the predetermined set of independent variables using ordered logit and ordered probit models. The number of cases for both models was 592. To test for the heteroscedasticity of error terms, the likelihood ratio test was employed to compare the log likelihood between the restricted model assuming homoscedastic error structure and the unrestricted model lifting the homoscedasticity assumption, thus allowing heteroscedasticity with no cross-sectional correlation. The result was statistically significant, indicating the presence of heteroscedasticity in the error structure, χ2(44) = 151.01 (p < .01). With the presence of heteroscedasticity, clustered robust SEs were used for the ordered logit and ordered probit analyses, which allowed for the intrapanel correlation of the observations. Both the ordered logit model, χ2(25) = 135.44 (p < .01), and the ordered probit model were significant, χ2(25) =151.56 (p < .01).
The results presented in Table 4 show that both the ordered logit and the ordered probit models produced a similar pattern of results.
Regression Results of Broadcaster Choice Model
| Variables | Ordered logit | Robust SE | Ordered probit | Robust SE | |
|---|---|---|---|---|---|
| Past game attributes | LnLagRatings | −1.323* | 0.740 | −0.822** | 0.396 |
| LagScoreDiff | −0.181 | 0.121 | −0.087 | 0.058 | |
| HomeRank | 0.068 | 0.048 | 0.044* | 0.025 | |
| AwayRank | 0.102** | 0.051 | 0.056** | 0.025 | |
| Predicted game attributes | Theil | 0.338** | 0.165 | 0.182** | 0.091 |
| BetReverse | 0.141 | 0.206 | 0.081 | 0.109 | |
| Team attributes | SumTeamAge | 0.012 | 0.042 | 0.008 | 0.023 |
| Rival | −1.326 | 1.742 | −0.754 | 0.987 | |
| LagPostSeason | 1.940* | 1.134 | 1.138* | 0.638 | |
| LnTotalSalary | −27.849*** | 4.607 | −15.474*** | 2.456 | |
| Allstar | 0.044 | 0.161 | 0.016 | 0.090 | |
| External factors | LnDist | −0.334 | 0.232 | −0.197 | 0.130 |
| Weekend | 0.115 | 0.287 | 0.025 | 0.144 | |
| NightTime | 0.164 | 0.410 | 0.066 | 0.220 | |
| Montha | — | — | — | — | |
| May | 1.227 | 0.892 | 0.654 | 0.499 | |
| June | 1.013 | 0.760 | 0.520 | 0.428 | |
| July | 0.696 | 0.964 | 0.366 | 0.532 | |
| August | 1.338 | 0.976 | 0.725 | 0.536 | |
| September | 0.571 | 0.935 | 0.312 | 0.515 | |
| October | −0.536 | 1.170 | −0.319 | 0.644 | |
| Networkb | — | — | — | — | |
| MBC Sports+ | −0.357 | 0.587 | −0.282 | 0.319 | |
| SBS Sports | 0.152 | 0.750 | 0.022 | 0.384 | |
| SPOTV | 0.625 | 0.525 | 0.299 | 0.293 | |
| SPOTV 2 | −0.483 | 0.599 | −0.326 | 0.324 | |
| Multi | −1.982** | 0.982 | −1.388** | 0.550 |
aReference month: March/April. bReference channel: KBS N Sports; DV = ChoiceOrder; N = 592.
*p < .10. **p < .05. ***p < .01.
It appears that three of the four past game attributes played significant roles when broadcasters made decisions about which games to televise in the coming week. LnLagRatings (βologit = −1.323, p < .10; βoprobit = −0.822, p < .05) was a significant and negative predictor of ChoiceOrder. This indicates that the higher the TV ratings of the most recent game series between the same matchup teams, the more the broadcasters prioritized the game. AwayRank (βologit = 0.102, p < .05; βoprobit = 0.056, p < .05) was a significant and positive predictor of choice order; likewise, HomeRank (βoprobit = 0.044, p < .10) was a significant and positive predictor of the broadcasters’ choice order, but only for the ordered probit model at the 90% significance level. These results indicate that broadcasters preferred games between teams at higher league standings up to the most recent game, and that away teams’ league standings (vs. home teams’) were a more stable influencer of broadcasters’ game choices. However, LagScoreDiff did not influence the broadcasters’ choice order.
Of the two predicted game attribute factors, Theil (βologit = 0.338, p < .05; βoprobit = 0.182, p < .05) had a positive and significant impact on ChoiceOrder; the positive relationship indicated that when a higher uncertainty of the outcome of a game is anticipated, broadcasters are less likely to choose the game for the next week’s telecast. This indicates broadcasters’ tendency to select games with more certain than uncertain outcomes. BetReverse did not influence the broadcasters’ choice order.
Of the five team attribute factors, the sum of both teams’ total salaries, LnTotalSalary (βologit = −27.849, p < .01; βoprobit = −15.474, p < .01), was a significant and negative predictor of broadcasters’ choice order, suggesting that broadcasters preferred to select games between teams with a higher total sum of wages. LagPostSeason (βologit = 1.940, p < .10; βoprobit = 1.138, p < .10) had a significant and positive impact on ChoiceOrder, albeit at a weak level of significance. This implies that broadcasters did not prioritize teams that advanced into the prior postseason. One possible interpretation is that broadcasters took into account the inverse relationship between playing in the postseason in the previous season and TV ratings, as evidenced in our viewership model. The remaining three team attribute factors—Allstar, Rival, and SumTeamAge—did not show any statistical significance.
Finally, none of the external influencers, including the physical distance between teams (i.e., LnDist) and the day and time of the games (i.e., Weekend and NightTime), had a significant impact on broadcasters’ choice order.
General Discussion
There is a dearth of literature on the topic of understanding the choices of broadcasters in the context of live sport programs and TV audiences’ demands. The empirical results of the current study provide both broadcasters and academia with valuable insights to understand the sport broadcast market dynamics. The key findings of the current study were twofold. First, our analyses revealed that the broadcasters’ choices of matches for live telecasts well reflected the actual TV audience demand. Second, different patterns of demand were observed for the broadcaster choice and the viewership models. Most notably, the predicted game attributes, such as outcome uncertainty, appeared to be significant predictors of the broadcasters’ choice of games for live telecasts but not for the viewership; rather, viewership was determined by the past and current game attributes as well as the team attributes.
These findings serve as a useful avenue for discussion. The fact that the games selected with higher priority were indeed associated with greater audience viewership indicates that the KBO broadcasters, although they do not go through systematic analyses before selecting games for live telecasts, have a reasonably good sense of which games will result in high TV ratings. Thus, it is reasonable to assume that broadcasters are rational decision makers whose objective is to maximize ratings (Noll, 2007).
Nonetheless, there are some notable discrepancies in the results from the viewership model and the broadcaster choice model, indicating that broadcasters have less than perfect understanding of the factors driving TV audience demand, or there are some factors broadcasters are unable to consider at the time of game selection. First, smaller score difference in the same matchup in the most recent game series led to higher viewership, but broadcasters did not take into consideration the past games’ score differences, despite the fact that such information is readily available to broadcasters when they select games. Second, broadcasters used prediction of possible outcome while viewers used real-time game information to make decisions. Specifically, unlike Forrest et al. (2005), broadcasters tended to choose games with lower outcome uncertainty while viewers showed little interest in the unpredictability of games. This might be due to the fact that the current ongoing game attributes are unavailable to broadcasters at the time of game selection but can be useful information for TV viewers as viewers can switch channels at any time. Third, broadcasters preferred games involving teams close to the top of the league standings, whereas viewership decreased for home team matches with higher league standings. A possible reason for these results might be that while broadcasters believed high-ranking teams or contenders for playoff spots could lure bigger TV audiences (Forrest et al., 2005), home fans might have chosen to physically attend at stadiums rather than watching the televised matches when their teams performed well. It should be noted, however, that such substitution effects were not directly examined in the current study. Future research should examine the possibility that home fans prefer to attend stadiums than watch games on TV when their teams are ranked high in the league standings.
There were also some similar patterns from the viewership and broadcaster choice models. Both TV viewers and broadcasters preferred games with higher TV ratings in the previous game series, showing evidence of the habitual nature of viewership demand for sport (Tainsky & Jasielec, 2014). This indicates that both viewers and broadcasters tend to maintain their interest levels in the same matchups. Alternatively, knowing that audience interest in each matchup has inertia to some degree, broadcasters could make risk-averse decisions by prioritizing matchups that previously generated high TV ratings. As Noll (2007) speculated, commercial broadcasters of the KBO ought to focus on the popularity of the game because of its potential for revenue from additional subscriptions and advertisement fees. Also, both TV viewers and broadcasters preferred games with greater sums of total salary. This result is consistent with prior research demonstrating a positive relationship between total salary and audience demand (Forrest et al., 2005). Interestingly, both TV viewers and broadcasters were less likely to prefer games that had appeared in the previous season’s postseason. This result might be a reflection of the fluctuating pattern of the league standings of the KBO teams from 1 year to another; that is, one team may participate in a postseason series 1 year but not in the following year. However, there is no empirical data to support this argument except for the low rank–order correlation (ρ = .20, ns) between the 2017 and 2018 league standings of the 10 KBO teams.
This study has managerial and theoretical implications. First, managerially, the findings of this study can serve as guidelines for broadcasters when selecting games most likely to maximize viewership. Second, although the current research context is unique in that five rights-holding channels make weekly decisions to choose which games to televise throughout a season, the findings can be applied to other professional leagues and sports. This is because in most cases rights-holding broadcasters do not televise all available matches included in the media rights package, but select a subset of matches for live telecasts. Therefore, broadcasters are faced with decision tasks to select games for live telecasts, although the number of game selections may vary from one league to another. For example, Olympic rights-holding broadcasters do not televise all matches staged at the Olympic Games, just the ones they select. Theoretically, this study replicates the stream of research that investigates audience demand for media sport (Buraimo & Simmons, 2015; Chung et al., 2016; Jung et al., 2020; Paul & Weinbach, 2013; Sung et al., 2019; Tainsky, 2010) and extends it by incorporating broadcasters’ demand for sport games, similar to Forrest et al. (2005).
Despite the contributions that our work makes to the current stream of literature, there are a few limitations that future research could address. First, our data cover a single year of observation, which can lead to additional time-variant determinants being omitted. A longitudinal study covering data from multiple years could reveal any shifts in preference for both broadcasters and viewers and help achieve a deeper understanding of the decision-making process. Second, because we used average ratings at the national level, we were unable to examine the fluctuations in ratings throughout each game. Nor was it possible to understand the specific media consumption patterns of local TV audiences. The use of minute-by-minute data, as well as data separated by local media markets, could help researchers uncover changing patterns of viewership throughout the game and audience behaviors in relation to their geographical location.
Note
The population of Korea was approximately 52 million in 2019 (http://www.kostat.go.kr).
References
Agah, N., & Dixon, J.C. (2021). The uniqueness of spectator sport from a strategic management perspective: The case for spectatoritis. Journal of Global Sport Management, 6(1), 7–29. https://doi.org/10.1080/24704067.2019.1641830
Alavy, K., Gaskell, A., Leach, S., & Szymanski, S. (2010). On the edge of your seat: Demand for football on television and the uncertainty of outcome hypothesis. International Journal of Sport Finance, 5(2), 75–95.
Bai, J., Choi, S.H., & Liao, Y. (2021). Feasible generalized least squares for panel data with cross-sectional and serial correlations. Empirical Economics, 60(1), 309–326. https://doi.org/10.1007/s00181-020-01977-2
Beckman, E.M., Cai, W., Esrock, R.M., & Lemke, R.J. (2012). Explaining game-to-game ticket sales for Major League Baseball games over time. Journal of Sport Economics, 13(5), 536–553. https://doi.org/10.1177/1527002511410980
Borland, J., & MacDonald, R. (2003). Demand for sport. Oxford Review of Economic Policy, 19(4), 478–502. https://doi.org/10.1093/oxrep/19.4.478
Buraimo, B., & Simmons, R. (2008). Do sports fans really value uncertainty of outcome? Evidence from the English Premier League. International Journal of Sport Finance, 3(3), 146–155.
Buraimo, B., & Simmons, R. (2009). A tale of two audiences: Spectators, television viewers and outcome uncertainty in Spanish football. Journal of Economics and Business, 61(4), 326–338. https://doi.org/10.1016/j.jeconbus.2008.10.002
Buraimo, B., & Simmons, R. (2015). Uncertainty of outcome or star quality? Television audience demand for English Premier League football. International Journal of the Economics of Business, 22(3), 449–469. https://doi.org/10.1080/13571516.2015.1010282
Chung, J., Lee, Y.H., & Kang, J.H. (2016). Ex ante and ex post expectations of outcome uncertainty and baseball television viewership. Journal of Sports Economics, 17(8), 790–812. https://doi.org/10.1177/1527002514551002
Coates, D., & Humphreys, B.R. (2005). Novelty effects of new facilities on attendance at professional sporting events. Contemporary Economic Policy, 23(3), 436–455. https://doi.org/10.1093/cep/byi033
Cox, A. (2018). Spectator demand, uncertainty of results, and public interest: Evidence from the English Premier League. Journal of Sport Economics, 19(1), 3–30. https://doi.org/10.1177/1527002515619655
Czarnitzki, D., & Stadtmann, G. (2002). Uncertainty of outcome versus reputation: Empirical evidence for the first German football division. Empirical Economics, 27(1), 101–112. https://doi.org/10.1007/s181-002-8361-7
DeSchriver, T.D., Rascher, D.A., & Shapiro, S.L. (2016). If we build it, will they come? Examining the effect of expansion teams and soccer-specific stadiums on Major League Soccer attendance. Sport, Business and Management, 6(2), 205–227. https://doi.org/10.1108/SBM-05-2014-0025
Evens, T., Iosifidis, P., & Smith, P. (2013). The political economy of television sport rights. Palgrave Macmillan.
Feehan, P., Forrest, D., & Simmons, R. (2003). Premier league soccer: Normal or inferior good? European Sport Management Quarterly, 3(1), 31–45. https://doi.org/10.1080/16184740308721938
Forrest, D., & Simmons, R. (2002). Outcome uncertainty and attendance demand in sport: The case of English soccer. Journal of the Royal Statistical Society. Series D, 51(2), 229–241. https://doi.org/10.1111/1467-9884.00314
Forrest, D., Simmons, R., & Buraimo, B. (2005). Outcome uncertainty and the couch potato audience. Scottish Journal of Political Economy, 52(4), 641–661. https://doi.org/10.1111/j.1467-9485.2005.00360.x
Forrest, D., Simmons, R., & Buraimo, B. (2006). Chapter 5: Broadcaster and audience demand for Premiere League football. In C. Jeanrenaud & S. Kesenne (Eds.,) The economics of sport and the media. (pp. 93–105). Edward Elgar.
Fort, R. (2006). Sport economics (2nd ed.). Pearson Prentice Hall.
Fort, R. (2017). Pro sports league antitrust ‘Beliefs’: Applied theory and the Rule of Reason. Managerial and Decision Economics, 38(5), 655–663.
Gratton, C., & Solbert, H.A. (2007). The economics of sports broadcasting. Routledge.
Ha, S. (2014). A study on changes in producing components of Korean TV sports relay broadcasting: Focused on professional baseball relay broadcasting [Unpublished doctoral dissertation, Konkuk University].
Hausman, J.A., & Leonard, G.K. (1997). Superstars in the National Basketball Association: Economic value and policy. Journal of Labor Economics, 15(4), 586–624. https://doi.org/10.1086/209839
Hoehn, T., & Lancefield, D. (2003). Broadcasting and sport. Oxford Review of Economic Policy, 19(4), 552–568. https://doi.org/10.1093/oxrep/19.4.552
Hotelling, H. (1929). Stability in competition. Economic Journal, 39(153), 41–57. https://doi.org/10.2307/2224214
Jang, H. (2019). Salary distribution and team outcome: The comparison of MLB and KBO. Journal of Global Sport Management, 4(2), 149–163. https://doi.org/10.1080/24704067.2018.1537678
Jung, H., Moon, C.G., & Sung, Y.T. (2020). Outcome uncertainty and demand for baseball: A single-game approach in the Korean professional baseball League. International Journal of Sport Finance, 15(1), 39–52.
KBO. (2020). Korean Baseball Organization. https://www.koreabaseball.com/
Kilduff, G.J., Elfenbein, H.A., & Staw, B.M. (2010). The psychology of rivalry: A relationally dependent analysis of competition. Academy of Management Journal, 53(5), 943–969. https://doi.org/10.5465/amj.2010.54533171
KISDI Report. (2019). 2019 survey of broadcasting industry. Korea Information Society Development Institute.
Lawson, R.A., Sheehan, K., & Stephenson, E.F. (2008). Vend it like Beckham: David Beckham’s effect on MLS ticket sales. International Journal of Sport Finance, 3(4), 189–195.
Lemke, R.J., Leonard, M., & Tlhokwane, K. (2010). Estimating attendance at Major League Baseball games for the 2007 season. Journal of Sport Economics, 11(3), 316–348. https://doi.org/10.1177/1527002509337212
Lewis, M., & Yoon, Y. (2018). An empirical examination of the development and impact of star power in major league baseball. Journal of Sport Economics, 19(2), 155–187. https://doi.org/10.1177/1527002515626220
Ma, Y., & Kurscheidt, M. (2020). Doing it the Chinese way: The politically-led commercialization of professional football in China. Journal of Global Sport Management. Advance online publication. https://doi.org/10.1080/24704067.2020.1753332
Meehan, J.W., Jr., Nelson, R.A., & Richardson, T.V. (2007). Competitive balance and game attendance in Major League Baseball. Journal of Sports Economics, 8(6), 563–580. https://doi.org/10.1177/1527002506296545
Mongeon, K., & Winfree, J. (2012). Comparison of television and gate demand in the National Basketball Association. Sport Management Review, 15(1), 72–79. https://doi.org/10.1016/j.smr.2011.09.001
Mullin, C.J., & Dunn, L.F. (2002). Using baseball card prices to measure star quality and monopsony. Economic Inquiry, 40(4), 620–632. https://doi.org/10.1093/ei/40.4.620
Noll, R.G. (2007). Broadcasting and team sports. Scottish Journal of Political Economy, 54(3), 400–421. https://doi.org/10.1111/j.1467-9485.2007.00422.x
Ormiston, R. (2014). Attendance effects of star pitchers in major league baseball. Journal of Sport Economics, 15(4), 338–364. https://doi.org/10.1177/1527002512461155
Paul, R.J. (2003). Variations in NHL attendance: The impact of violence, scoring, and regional rivalries. American Journal of Economics and Sociology, 62(2), 345–364. https://doi.org/10.1111/1536-7150.t01-2-00003
Paul, R.J., & Weinbach, A.P. (2013). Uncertainty of outcome television ratings for the NHL and MLS. Journal of Prediction Markets, 7(1), 53–65. https://doi.org/10.5750/jpm.v7i1.587
Paul, R.J., & Weinbach, A.P. (2015). The betting market as a forecast of television ratings for primetime NFL football. International Journal of Sport Finance, 10(3), 284–296.
Pawlowski, T., & Anders, C. (2012). Stadium attendance in German professional football—The (un)importance of uncertainty of outcome reconsidered. Applied Economics Letters, 19(16), 1553–1556. https://doi.org/10.1080/13504851.2011.639725
Pérez, L., Puente, V., & Rodríguez, P. (2017). Factors determining TV soccer viewing: Does uncertainty of outcome really matter? International Journal of Sport Finance, 12(2), 124–139.
Rosen, S. (1981). The economics of superstars. The American Economic Review, 71(5), 845–858.
Rottenberg, S. (1956). The baseball players’ labor market. Journal of Political Economy, 64(3), 242–258. https://doi.org/10.1086/257790
Ryu, Y., Kim, K., Paik, J.W., & Cheong, Y. (2019). Determinants of audience demand for the televised professional baseball matches in Korea: An analysis of the post-season matches from 2008 to 2016. International Journal of Sport Marketing and Sponsorship, 20(1), 184–202. https://doi.org/10.1108/IJSMS-12-2017-0127
Schreyer, D., Schmidt, S.L., & Torgler, B. (2018). Game outcome uncertainty and television audience demand: New evidence from German football. German Economic Review, 19(2), 140–161. https://doi.org/10.1111/geer.12120
Solberg, H.A. (2007). Sports broadcasting: Is it a job for public service broadcasters? A welfare economic perspective. Journal of Media Economics, 20(4), 289–309. https://doi.org/10.1080/08997760701668169
Sung, H., & Mills, B.M. (2018). Estimation of game-level attendance in major league soccer: Outcome uncertainty and absolute quality considerations. Sport Management Review, 21(5), 519–532. https://doi.org/10.1016/j.smr.2017.12.002
Sung, H., Mills, B.M., & Mondello, M. (2019). Local broadcast viewership in Major League Soccer. Journal of Sport Management, 33(2), 106–118. https://doi.org/10.1123/jsm.2018-0022
Sung, H., Mills, B.M., & Tainsky, S. (2017). From schadenfreude to mitfreude? Estimating viewership loss and rivalrous relationships in otherwise neutral markets. Sport Management Review, 20(2), 159–169.
Szymanski, S. (2003). The economic design of sporting contests. Journal of economic literature, 41(4), 1137–1187. https://doi.org/10.1257/jel.41.4.1137
Tainsky, S. (2010). Television broadcast demand for National Football League contests. Journal of Sport Economics, 11(6), 629–640. https://doi.org/10.1177/1527002509355636
Tainsky, S., & Jasielec, M. (2014). Television viewership of out-of-market games in league markets: Traditional demand shifters and local team influence. Journal of Sport Management, 28(1), 94–108.
Tainsky, S., & McEvoy, C.D. (2012). Television broadcast demand in markets without local teams. Journal of Sport Economics, 13(3), 250–265. https://doi.org/10.1177/1527002511406129
Theil, H. (1967). Economics and information theory. McGraw-Hill.
Tyler, B.D., & Cobbs, J.B. (2015). Rival conceptions of rivalry: Why some competitions mean more than others. European Sport Management Quarterly, 15(2), 227–248. https://doi.org/10.1080/16184742.2015.1010558
Villar, J.G., & Guerrero, P.R. (2009). Sport attendance: A survey of the literature 1973–2007. Rivista di Diritto e di Economia dello Sport, 5(2), 112–151.
Wang, C., Goossens, D., & Vandebroek, M. (2018). The Impact of the soccer schedule on TV viewership and stadium attendance: Evidence from the Belgian Pro League. Journal of Sport Economics, 19(1), 82–112. https://doi.org/10.1177/1527002515612875
Winfree, J.A., McCluskey, J.J., Mittelhammer, R.C., & Fort, R. (2004). Location and attendance in major league baseball. Applied Economics, 36(19), 2117–2124. https://doi.org/10.1080/0003684042000287664
Wooten, J.J. (2018). Rivalries in US Professional Soccer. International Journal of Sport Finance, 13(2), 105–118.
World Economic Forum. (2020). This is how COVID-19 is affecting the world of sport. https://www.weforum.org/agenda/2020/04/sports-covid19-coronavirus-excersise-specators-media-coverage/
