Japanese professional soccer attendance and the effects of regions, competitive balance, and rival franchises.

Watanabe, Nicholas M.

Introduction

Baseball and the traditional sport of Sumo Wrestling have long been considered the two more popular sporting pastimes in Japan, with Nippon Professional Baseball (NPB) holding status as the premier professional sport league in the country (Whiting, 1990). While soccer leagues have existed domestically within Japan for over half a century, it was only with the formation of the J-League in 1992 that a rival professional team sport league emerged to challenge the dominance of the NPB. While the J-League has become a more established league in the two decades since its creation, the league still seems to be lurking in the shadow of the NPB. However, in recent years, the NPB has been plagued by its first ever strike and the forced merger of two teams to continue its existence, while the J-League has continued to expand and grow into an organization larger than the NPB.

In some sense, a battle between the NPB and J-League is raging, as both leagues compete against one another, trying to capture fan support and boost attendance numbers as much as possible. What is somewhat problematic in this competition is how professional sport franchises have been distributed in Japan. In North America there is often only one sport franchise per league in each market, or two franchises per league in the cases of larger markets such as New York and Chicago, while in Europe there are more markets with multiple franchises. In this sense, because of Japan's population density, there are often multiple franchises from a single league co-existing within the same city. In the case of the city of Tokyo, there are currently two J-League teams and two NPB franchises located within Tokyo (Honda, Kawai, & Hirata, 2009). Previously, many more franchises have existed within Tokyo, and in even greater numbers when looking at larger regional definitions around the Tokyo area.

In most situations professional sport leagues attempt to create monopoly power; however, this monopoly power may be lost in the case of professional sport leagues concentrating franchises within a single region such as found in Europe and Japan. Furthermore, professional sport franchises have often sought to reduce the number of teams within a single market, so as to have as few sport substitutes as possible (Nester, 1990). The unique distribution of the Japanese population, however, has forced teams into either sharing the large mega-markets with other rivals, or to try and stake a claim in a smaller market in hopes of building a loyal fan base through regional association. Cairns (1990) noted the issues in trying to measure different regions in a sophisticated manner within research. In this, there is difficulty in attempting to fully capture the "away" population which may attend a match or game, and measuring the true size of the market that attends sporting events (Cairns, 1990).

Japan presents an interesting environment in which to empirically examine and test different regional definitions of a market, and how they influence the attendance at sporting events. At the base level, the Japanese use three types of regional definitions: cities, prefectures, and regions (which are composed of multiple prefectures). To help better understand, take the case of the city of Yokohama (which neighbors Tokyo). Yokohama city is located within Kanagawa prefecture, a larger area that borders Tokyo as well. Furthermore, both Tokyo and Kanagawa prefecture are located within the "Kanto" region, which is composed of Tokyo, Kanagawa, and several other prefectures together. The three regional types constitute different sizes in regards to both area and population, and are thus of focus in three different models in this paper. These regions are also of interest because all three area definitions are widely used in Japan to this current day. This paper thus tries to help further the discussion and understanding of regions (and hence market size) in affecting attendance by considering all three regional levels, as well as the number of rival J-League and NPB franchises existing within the same regional areas.

Also considering the situation of professional sports in Japan, coupled with the country's relatively poor economic performance in recent years, there is a clear need for investigation of the professional sport leagues within the country. While there exists a number of studies focused on the NPB from an economic standpoint (Ohkusa, 1999; La Croix & Kawaura, 1999), there exist but a few studies on the J-League, mostly in the form of spectator preference studies (Nakazawa, Mahony, Funk, & Hirakawa, 1999; Mahony, Nakazawa, Funk, James, & Gladden, 2002), and one unpublished study which examines the effects that the NPB has on J-League attendance (Honda et al., 2009). Mahony et al. (2002) find a variety of factors that are important to the spectator's decision to attend matches, including the drama and level of uncertainty of a match. Professional sport in Japan is also of interest because of its use of separate structural models for the different sport leagues. That is, they use a North American style structure for professional baseball, and a European model for football (soccer), in what Humphreys and Watanabe (2011) describe as a "hybrid" structural model.

The one unpublished attendance study focused on the effects that distance between J-League and NPB franchises have on attendance (Honda et al., 2009) indicates the existence and proximity of a NPB team has a negative impact on J-League attendance. While this study shows the negative impacts the NPB has on J-League attendance, it does not fully consider the impacts J-League franchises may have on one another's attendance. Furthermore, the researchers use match-level variables to measure team quality, but ignore any attempt at considering the effects of uncertainty of outcome in matches, or competitive balance.

In empirically examining J-League attendance, several factors are considered within this study. In addition to regional definitions, competitive balance and the concept of parity are considered. It is believed under the Uncertainty of Outcome Hypothesis (UOH) (Rottenberg, 1956; Neale, 1964) that when there is less certainty of the outcome of a match, attendance will increase. Within the North American literature, findings have indicated the importance competitive balance and UOH has on attendance (Soebbing, 2008), while European research has tended to display less support for the UOH (Buraimo & Simmons, 2008). The Noll-Scully competitive balance metric is included within the models within this research to test the importance of competitive balance and UOH in Japanese professional football.

This research, thus, considers whether the UOH is of significance within an Asian sport league. Therefore, this study will seek to make to major contributions in being one of the first empirical study to attempt to understand the determinants of attendance at J-League matches from an economic standpoint, as well as being one of a handful of studies (Fort & Maxcy, 2003; Soebbing, 2008) which considers the role of competitive balance in the empirical examination of live attendance at sporting events. Additionally, the consideration of market size and regions within this study helps to further the understanding of how different market sizes can affect attendance. In this, different regional definitions and population sizes are employed to try and give a more developed approach to measuring different regional and market sizes for sport franchises.

J-League

The J-League was formed in 1993 as part of Japan's bid to host the 1994 Federation Internationale de Football Association (FIFA) World Cup Finals, which ultimately failed, with the tournament awarded to the United States. The J-League began with a simple system with only one division with 10 teams. Prior to 1998, teams were promoted from the Japanese Football League (JFL); however, this changed with the addition of a second division called J2 in 1999. This created a promotion and relegation system by combining teams from the JFL and J-League so that J1 would have 16 clubs and J2 would have 10 clubs.

The restructuring of the JFL allowed for a multi-tiered system in which teams could move from between the JFL, J2, and J1 via promotion and relegation. The J-League has continued to evolve this system, by adding more teams to each of the leagues in the past 10 years, with J1 now consisting of 18 clubs, J2 of 15 clubs, and the JFL of 18 clubs. Thus, in a period of 20 years the J-League managed to grow from a 10-club league, with the JFL as a separate entity, to having a three-league tier system with full promotion and relegation just like that used in soccer leagues throughout Europe and South America.

In its current format, the J-League season runs from the early spring through to December, with the champion being determined by the team with the most points gained from wins and ties at the end of the season. This format closely mirrors the structural format used in European football leagues. For the period of 1993 through 2008, which is the focus of this study, there have been fluctuations in the number of teams in the league. The largest number of franchises has always been focused around the Tokyo regional area known as "Kanto", where up to 9 to 10 teams, or about half the league, have been based at any one time. As one moves away from the Kanto region to the north or west, the number of franchises tend to decrease, especially when moving to the exterior islands of Kyushu and Hokkaido, which have had at most one or two franchises at any one time in the top flight of Japanese football.

Demand for Sport

Following the seminal works of Rottenberg (1956) and Neale (1964), who both proposed an Uncertainty of Outcome Hypothesis (UOH), in more recent years there has been a boom in both theoretical and empirical literature attempting to understand what factors truly are influential in determining the demand for sport. Probably one of the more insightful studies is an elegant review of the literature (Borland & Macdonald, 2003), which draws conclusions from the body of sport demand literature and theorizes that demand for sport can be broken down into five categories: consumer preferences, economic, quality of viewing, characteristics of the sporting contest, and supply capacity. It is with this and other theoretical foundations which this study will attempt to understand attendance at J-League matches.

In considering this taxonomy of determinants of demand, there are clearly a variety of factors which need to be considered in attempting to understand the demand for sport. Probably one of the more important lines of sport research focuses on the effect price has on attendance, as not only is price considered to be important through neoclassical economic theory in determining demand, but also because price data is often very difficult to obtain for data sets in sport (Borland & Macdonald, 2003). Empirical investigation has found that price has little (Whitney, 1988), or no affect on the demand for attendance at sporting events (Burdekin & Idson, 1991; Fort, 2005). Further research has shown tickets to sporting events are priced in the inelastic range (Scully, 1989; Fort & Quirk, 1996), indicating the prices of tickets may not be significant determinants of demand as predicted by theory. Rather, other variables may have larger effects on consumer choice of attending sporting events, or other game-day revenue sources such as concessions and parking help teams set prices in the inelastic range (Coates & Humphreys, 2007). Due to unavailability of price data, as well as the empirical findings, this paper takes the position that the omission of price from the specified model as being unfortunate, but a necessary consequence which should not have major effects of the results presented.

Uncertainty of Outcome

Uncertainty of outcome can be traced to the prior mentioned UOH (Rottenberg, 1956; Neale, 1964), which theorizes that fan attendance will increase as game outcomes become more uncertain. From this hypothesis, uncertainty of outcome can be considered to be an important determinant of demand, and is, thus, widely employed within demand studies. Forrest and Simmons (2002) offer a definition of uncertainty of outcome as: "a situation where a given contest within a league structure has a degree of unpredictability about the result and, by extension, that competition as a whole does not have a predetermined winner at the outset of the competition" (p. 229). Within the uncertainty of outcome literature, there exist three basic variations in how researchers have attempted to consider this unpredictability of sporting contests: at the match-level, season-level, and long-run (Borland & Macdonald, 2003). Not only is the uncertainty of outcome considered within these three different levels within demand studies, but the metrics used to measure the concept of uncertainty also vary within each of these levels.

First, examining the literature focused on match-level attendance, research has focused into how the demand for attendance changes from one match to the next depending on the level of uncertainty in each match. The first study that attempted to employ uncertainty of outcome at the match level was Hill, Madura, and Zuber's (1982) study of Major League Baseball (MLB) attendance. In this research, uncertainty was measured through the use of variables for both the home and away team, which measured how many games behind the leader each team was. It is found that the number of games a home team is behind a leader has a stronger effect than the number of games the away team is, which may be indication that team strength may have a stronger effect than uncertainty in determining the demand for sport (Borland & Macdonald, 2003).

Following the work of Hill et al. (1982), a number of researchers proposed new and innovative means through which to measure match-level uncertainty (Whitney, 1988; Carmichael et al., 1999; Borland & Lye, 1989). The results from these studies make it hard to draw any conclusions as to the nature of the role uncertainty of outcome in determining attendance (Borland & Macdonald, 2003). Additionally, the possibility that uncertainty does not manifest itself well at the match-level is considered by researchers, warranting investigation of the phenomena over longer time periods.

The focus of seasonal-level uncertainty studies is on how uncertainty of outcome affects attendance at matches from one season to the next. These studies are traced back to the work of Noll (1974), who used the average number of games behind a leader a team was, as a measure of uncertainty. Like the match-level studies, there are also a number of metrics in which season-level uncertainty of outcome can be measured, and has been employed in attendance studies. Metrics included to measure seasonal uncertainty include direct measures of uncertainty (Noll, 1974; Borland & Lye, 1989; Borland & Macdonald, 2003), as well as competitive balance metrics (Soebbing, 2008). Results for season-level uncertainty of outcome studies find stronger evidence of a relationship between attendance and uncertainty than is evidenced in match-level studies, suggesting the effect of uncertainty on attendance manifesting in longer time periods.

Finally, considering the long-run uncertainty of outcome, there is but a handful of studies which have examined how uncertainty of outcome has affected the demand for sport over longer periods of time (Borland, 1987; Schmidt & Berri, 2001, Humphreys, 2002). While Borland's (1987) study found little or no connection between long-run uncertainty and attendance, the works of Schmidt and Berri (2001) and Humphreys (2002) find a strong relationship between uncertainty of outcome and attendance at matches. The two later works are also notable in that they (Schmidt & Berri, 2001; Humphreys, 2002) employ competitive balance variables as measures of uncertainty of outcome, showing a strong link between uncertainty of outcome and attendance, and likewise, competitive balance and attendance.

In reviewing the literature on the three levels of uncertainty of outcome studies, it is clear that studies focused on longer periods of time find stronger evidence of uncertainty of outcome being related to attendance. Furthermore, it is evidenced that further investigation on the role of uncertainty of outcome and attendance is necessary. Additionally, it is shown within this literature that competitive balance is a legitimate incarnation of uncertainty, yet there remains an absence of competitive balance metrics within the UOH/attendance literature. From this, it is necessary to further consider the relation between competitive balance, how it is measured, and its role within the current demand for sport literature.

Competitive Balance

Competitive balance has become a popular word of use when considering the relative strengths between teams competing within the same league. While the phrase hasn't quite reached buzz-word status, the phrase is often misused by media and fans alike. Thus, the question arises: what is competitive balance? One of the more cited definitions of competitive balance comes from Forrest and Simmons (2002), who define it as: "a league structure which has relatively equal playing strength between league members." [italics added] (p. 229). Fort and Maxcy (2003) additionally comment on competitive balance, arguing that competitive balance research is necessary from both a UOH perspective, as well as an analysis of competitive balance (ACB) perspective. From this, a need for further investigation of competitive balance, both as a measure of uncertainty of outcome in attendance studies (Fort & Maxcy, 2003), as well as in examining the changes and evolution of competitive balance within leagues is noted

Early empirical examination of competitive balance (Scully, 1989; Fort & Quirk, 1995) focused on the metric known as the standard deviation of win percentage (SDWPCT). This competitive balance metric has two different variations, the season-specific, and the team-specific. Season-specific SDWPCT takes all the teams in a league over a specific number of years, and calculates how far teams in a league deviate from the mean win percentage. Team-specific SDWPCT, on the other hand, calculates how far a single team deviates from the mean win percentage over many seasons of play. The SDWPCT has been widely employed within the literature focused on the ACB line of competitive balance research, but has almost never been employed in UOH studies. Soebbing (2008) uses a season-specific SDWPCT to examine seasonal attendance at MLB games, and find the metric to be negative and significant in regards to attendance, as hypothesized. What is problematic about the season-specific SDWPCT is there will only be one measurement for all teams in a league, and likewise, the team-specific SDWPCT would allow a team only one measurement of competitive balance over a several-year period. Thus a team could theoretically have the same competitive balance metric over a 100-year dataset, while actually experiencing much larger fluctuations and winning trends during this time (Eckard, 1998; Humphreys, 2002).

Further innovations in competitive balance metrics have sought a variety of manners in which to measure the concept, including distribution of championships (Eckard, 1998) through the Hirfindahl-Hirschman Index's (HHI), reordering of teams from one season to the next (Maxcy, 2002) in the Spearmen Rank Correlation Coefficient, a variance decomposition of win percentage (Eckard, 2001), gini-coefficients to measure disparity (Schmidt & Berri, 2001), and the competitive balance ratio (CBR), which is a ratio of the averages of the team and season-specific SDWPCT's (Humphreys, 2002).

In regards to demand studies focused on competitive balance, the research findings indicate that competitive balance does have some affect on the demand for attendance at sporting events (Humphreys, 2002; Soebbing, 2008). Within these demand studies, the metrics of the HHI (Humphreys, 2002), CBR (Humphreys, 2002), and SDWPCT (Soebbing, 2008) have been employed, with only the CBR and SDWPCT evidencing a strong relationship between competitive balance and attendance. As this study is focused on seasonal attendance, it is only logical that this study will employ the Noll-Scully ratio of SDWPCT to idealized SDWPCT as Soebbing (2008) did to test the UOH, as part of a demand model to estimate attendance.

Research Questions

This paper will attempt to answer a series of research questions focused on fan attendance of live sporting events. First, this work will attempt to estimate the affects of competitive balance on attendance, and in doing this, test the UOH. Second, considering the clustered nature of Japanese sport, this paper will attempt to examine the differences which come about from using different regional definitions to consider the number of competing J-League and NPB franchises in the each region type. It is hypothesized that the competitive balance measure will have an inverse relationship to attendance, that is, the better competitive balance, the more attendance will increase. Finally, it is believed there will be some fluctuation in the regional J-League and NPB franchise count variable, but in general, there will be a significantly negative relationship between these variables and attendance.

Data

Data collection for this project was conducted through the use of a variety of sources, mainly in Japanese, and focuses on the 1993 through 2008 J-League seasons. Market population data was collected through the Japanese Ministry of Internal Affairs and Communications' Statistics Bureau Home Page. Seasonal average attendance, team records, championships, and stadium capacities were all collected through the official J-League website (http://www.j-league.or.jp/). The Noll-Scully competitive balance metric was calculated by employing the standard deviation of mean win percentage for all J-League teams in each season. In this, because ties are worth only one-third of a win, a correction was made for this in calculating win percent, and the Noll-Scully. Variables for number of rival J-League franchise and NPB franchise variable for a city, prefecture, and region were gathered by finding all stadium locations of each team in each season. Summary statistics of the data are presented in Table 1.

Model Specification

The regression equation employed to estimate results is:

Average Franchise [Attendance.sub.it] = 0 + [sub.1]Competitive Balance [Measure.sub.t] + [sub.2]Win [Percent.sub.it] + [sub.3]First Stage [Championship.sub.it] + [sub.4]Second Stage [Championship.sub.it] + [sub.5]Overall [Champion.sub.it] + [sub.6][Population.sub.it] + [sub.7]Stadium [Capacity.sub.it] + [sub.8]New Stadium [Dummy.sub.it] + [sub.9]World Cup [Dummy.sub.t] + [sub.10]Rival J-League Franchise [Variable.sub.it] + [sub.11]Rival NPB Franchise [Variable.sub.it] + [sub.12]Time [Trend.sub.i] + [sub.13]Time Trend [Squared.sub.i] + [sub.13]Franchise [Dummy.sub.t] [[micro].sub.it] where t index's seasons, and i index's J-League franchises.

Within the model, seasonal average attendance is used as the dependent variable. Independent variables focused on team performance include: the Noll-Scully competitive balance metric, and team win percent to test the effects of team strength. The Noll-Scully is fairly easy to understand, in that the lower the number, the better competitive balance is. The first stage and second stage championship variables represent the division of the season into two halves, as was done in the earlier years of the league. In this system, the two winners of both halves would then meet in a final match at the end of the season to denote the overall league champion. The Overall Champion dummy thus indicates the winner of this match from the earlier years, and the leader of the table in more recent years.

Stadium variables include capacity, to consider supply capacity of sport, which brings up the issue of using a Tobit model because of sell-outs at sporting contests (Feehan, 2006) as well as the supply capacity of the sporting product (Borland & Macdonald, 2003). However in the case of the J-League, there are few sell-outs, and thus as indicated in research, a Tobit model is not necessary (Forrest & Simmons, 2002). The World Cup dummy variable indicates those years in which a FIFA World Cup is held, as the mega-event should generate a heightened interest in soccer. The population variable simply measures the population of the prefecture for the city and prefecture level models, and for the entire regions (composed of multiple prefectures tabulated together) population for the region level model. Time trend and time trend squared variables are included within the model to capture the effects of time on attendance in the new league. It is possible that with a new league, there may be a novelty effect in the early seasons, so it is important to try and capture any such effect. Additionally, franchise dummies are also placed within the model to try and capture any franchise-specific effects on attendance.

Finally, there are the J-League franchise and NPB franchise regional variables, which measure the number of other franchises from these leagues in the same regional area. Because Japan often uses three different definitions of regions, all three types were encoded, and included in separate models. Therefore, there are J-League franchise variables for city, prefecture, and regional levels, and similarly, NPB franchise variables for city, prefectural, and region. Thus, the first model is run using the J-League and NPB city variables, the second one using prefectural variables, and the third using region variables.

A Hausman test was run on the results of a fixed-effects and random-effects model run with the panel data. Results of the Hausman test were insignificant, indicating the use of a Generalized Least Squares (GLS) random-effects model would be of best choice (Gujarati, 2003). This formal test was employed on all models run for the three types of regional definition, and found that the GLS should be employed for each. Results can be found in Table 2.

[FIGURE 1 OMITTED]

Results

The observed variation in attendance explained the three models explained 80% (the overall r-square) in all three models. There are notable possible errors and limitations of the data set and model specification which may have contributed to not having a higher r-squared. The Noll-Scully was negative and significant in all three models at the 1% level, indicating that competitive balance is a strong indicator of attendance in J-League matches, and confirming the UOH.

Team strength and stadium capacity, and the World Cup dummy variables were all positive and significant in all three regression models, indicating their importance in determining J-League attendance. Curiously, the population variable was insignificant in the city level model, but was positive and significant in the prefecture and regional level models. The J-League franchise variable is positive and significant at the city and prefectural level, and negative and significant at the regional level. The NPB franchise variable was insignificant at the city and prefecture level, and negative and significant at the regional level. The time variable was negative and significant, indicating that J-League attendance has decreased over time; the square term of time was positive and significant, indicating that the decrease is increasing over time.

Discussion, Limitations, and Future Directions

As the results indicate, competitive balance as measured by the Noll-Scully is found to be strongly related to attendance in the J-League. Not only are these results similar to the findings of Soebbing (2008), but they also help to build a stronger case for the importance of competitive balance and the UOH within the sport attendance literature. Furthermore, the results in this paper help to expand the current theoretical and empirical understanding of the UOH line of competitive balance research. That is, fans do seem to respond to certain levels of parity/disparity within a sport league when making the decision to attend matches. This will hopefully translate into the continued and further use of competitive balance and testing of the UOH with attendance as well as demand for sport literature.

In all three models, the results for performance variables showed that win percent was significant in fan decision to attend matches, but none of the championship variables had any effect in any of the models. A separate model was run with the championship variables being lagged; however, these did not change the significance of these variables. These results indicate that fans want to watch strong teams, but that winning championships may not be of consequence for fans who attend Japanese soccer matches.

The stadium capacity variable was positive and significant at all levels, meaning the more seats a stadium has, the more fans can attend, which indicates supply capacity is an important determinant of attendance. The dummy variable for new stadiums was insignificant, which meant fans were not responsive to new stadiums, possibly indicating the lack of a novelty effect for Japanese soccer fans. The population variable was insignificant in the city model, and was positive and significant at the prefectural and regional level. This may be indication that fan bases for J-League franchises are not well measured at the city level, as well as the use of prefectural population data for a city may not be the best fit. Issues do arise in the lack of proper city population data for the time set used within this study, and future research will hopefully turn up reliable city level data for all Japanese cities for further investigation into this matter.

Finally, considering the J-League and NPB franchise variable, the results paint a very intriguing picture. Considering the results from the three models using different regional variables, it would seem to be an indication that at the city and prefectural level, it is actually beneficial in regards to attendance to have multiple J-League franchises. Furthermore, at the city and prefectural level, NPB franchises do not seem to have any influence on J-League attendance. This picture changes as one moves to a regional level, where having several J-League and NPB franchises in the same region hinders attendance. In this sense, it may indicate that local rivalries may be healthy, but over-saturation of a several prefectural area may be a negative for J-League franchises. In regard to the franchise specific dummy variables, the variables in general reported values that were negative and significant, hinting that many franchises actually had negative effects on attendance; the fixed effects are not presented in the tables because of space limitations.

There are the possibilities of a few limitations which may hamper this study, and thus need to be discussed and acknowledged. One possible limitation of this study is omitted variable bias, especially in regards to no variables used for price or distance within this study. In regards to price, this variable was omitted because of a lack of data. The distance variable was not employed because the distance between clubs often overlapped over city and prefectural boundaries, and because of movement of clubs as well as promotion and relegation, rival clubs and derby matches are not necessarily always captured by distance. Where Cairns (1990) noted the issues which went along with calculating market effects for areas having multiple franchises, the case of Japanese regions may be even more problematic.

Two issues, which are related to one another, that may have also affected this study are those of the data and number of observations. This research only examined the J-League from 1993 to 2008 due to the fact that the Japanese government has only posted population numbers through to 2008, which limits the potential number of observations in this study. Furthermore, there may have been some errors or inaccuracies in the data, such as misrepresented attendance numbers. However, because of the general reliability of the Japanese government databases and J-League website, it is believed that the data employed within this study is quite reliable.

As displayed in the results of this paper, as well as prior research (Humphreys, 2002; Soebbing, 2008) there is a need for the continued examination of the relationship which exists between competitive balance and attendance and sport demand. To date most studies have focused primarily on the SDWPCT and the Noll-Scully, and thus there is a need to consider other measures of competitive balance within demand models. Future studies into attendance should also strive to reach outside of the more popular and "major" leagues, and attempt to also understand why consumers attend sporting events for least popular leagues, such as Major League Soccer (MLS), the now defunct Arena Football League (AFL), and so forth. Finally, it would seem prudent for further investigation of the issues of various regional definitions, and attempt to better understand the effects distance, substitutes, and other factors within a region may have on game attendance.

Conclusion

Prior literature into competitive balance and attendance has found there is evidence of a relationship between these two factors (Schmidt & Berri, 2001; Humphreys, 2002; Soebbing, 2008). The results of this study are important, as they help to further the argument and present evidence which backs the findings of prior research, that competitive balance does indeed have a significant effect on attendance. While the previous studies did so using primarily large professional sport leagues from western nations, this research used a more novel context in considering the effect of competitive balance on attendance at a new and emerging league in a non-western nation. The results indicate that even in other cultural contexts, in this case Japan, competitive balance remains an important determinant of attendance, highlighting the need for further study of the concept in conjunction within and outside of research focused on the attendance for sport.

This model also considers other important issues related to the demand for sport attendance, especially the difficulty of considering different regional definitions in regards to the "market" for a sport franchise. The findings of this study suggest that rival franchises both within and outside of a league may have either positive or negative effects based on attendance, with the negative effects coming when one moves to larger regional definitions. Additionally, this model is novel not only in using a rather unexamined league to conduct an attendance study, but also in considering the complex nature of regions and travel within Japan. Further research is needed to better understand the effects of distance and regions for the demand for sport both within Japan and other countries. In conclusion, this study can be considered to be the first stepping stone which Soebbing (2008) called for, in considering the UOH and competitive balance in other leagues outside the MLB, as well as helping to push forward the consideration of various regional definitions in demand for sport research.

References

Borland, J., & Lye, J. (1989). Attendance at Australian Rules Football: A panel study. Applied Economics, 24, 1053-1058.

Borland, J., & Macdonald, R. (2003) Demand for sport. Oxford Review of Economic Policy, 19(4), 478-502.

Burdekin, R., & Idson, T. (1991, January). Consumer preference, attendance and the racial structure of professional basketball teams. Applied Economics, 23, 179-186.

Buraimo, B., & Simmons, R. (2008). Competitive balance and attendance in Major League Baseball: An empirical test of the uncertainty of outcome Hypothesis. International Journal of Sport Finance, 3(3), 146-155.

Cairns, J. (1990). The demand for professional team sports. British Review of Economic Issues, 12(28), 1-20.

Carmichael, F., Millington, J., & Simmons, R. (1999). Elasticity of demand for Rugby League attendance and the impact of BskyB. Applied Economics Letters, 6(12), 797-800.

Coates, D., & Humphreys, B. R. (2007). Ticket prices, concessions and attendance at professional sporting events. International Journal of Sport Finance, 2, 161-170.

Eckard, E. W. (1998). The NCAA cartel and competitive balance in college football. Review of Industrial Organization, 13, 347-369.

Eckard, E. W. (2001). Free agency, competitive balance, and diminishing returns to pennant contention. Economic Inquiry, 39(3), 430-443.

El-Hodiri, M., & Quirk, J. (1971). An economic model of a professional sports league. Journal of Political Economy, 79, 1302-1319.

Feehan, P. D. (2006). Attendance at sports events. In W. Andreff & S. Szymanski (Eds.), Handbook on the Economics of Sport (pp. 90-99). Cheltenham, UK: Edward Elgar Publishing.

Forrest, D, & Simmons, R. (2002). Outcome uncertainty and attendance demand in sport: The case of English Soccer. The Statistician, 51, 229-241.

Fort, R. (2005). The golden anniversary of 'the baseball players' labor market'. Journal of Sports Economics, 6(4), 347-358.

Fort, R., & Maxcy, J. (2003). Comment on: "Competitive balance in sports leagues: An introduction. Journal of Sports Economics, 4(2), 154-160.

Fort, R., & Quirk, J. (1995). Cross-subsidization, incentives, and outcomes in professional team sports leagues. Journal of Economic Literature, 33(3), 1265-1299.

Fort, R., & Quirk, J. (1996). Over-stated exploitation: Monoposony versus revenue sharing in sports leagues. In J. Fizel, E. Gustafson, & L. Hadley (Eds.), Baseball economics: Current research. Westport, CT: Praeger Publishers.

Gujarati, D. N. (2003). Basic econometrics (4th ed.). New York, NY: McGraw-Hill/Irwin.

Hill, J. R., Madura, J., & Zuber, R. A. (1982). The short run demand for Major League Baseball. Atlantic Economic Journal, 10(2), 31-35.

Honda, D., Kawai, S., & Hirata, T. (2009). The impact of Japanese Professional Baseball for attendance demand for the J.League. Western Economics Association International Meetings, March 24-27, Japan.

Humphreys, B. R. (2002). Alternative measures of competitive balance in sports leagues. Journal of Sports Economics, 3(2), 133-48.

Humphreys, B. R., & Watanabe, N. M. (2011). Business and finance of international sport leagues. In M. Li, E. Macintosh, & G. Bravo (Eds.), International Sport Management. Champaign, IL: Human Kinetics.

Japanese Ministry of Internal Affairs and Communication. (n.d.). Statistics Bureau homepage. Retrieved from http://www.stat.go.jp

J-League. (n.d.). Retrieved from http://www.j-league.or.jp/

La Croix, S. J., & Kawaura, A. (1999). Rule changes and competitive balance in Japanese Professional Baseball. Economic Inquiry, 37(2), 353-368.

Mahony, D. F., Nakazawa, M., Funk, D. C., James, J. D., & Gladden, J. M. (2002). Motivational factors influencing the behaviour of J. League spectators. Sport Management Review, 5, 1-24.

Maxcy, J. (2002). Rethinking restrictions on player mobility in Major League Baseball. Contemporary Economic Policy, 20(2), 145-159.

Nakazawa, M., Mahony, D. F., Funk, D. C., & Hirakawa, S. (1999). Segmenting J. League spectators based on length of time as a fan. Sport Marketing Quarterly, 8(4), 55-65.

Neale, W. C. (1964). The peculiar economics of professional sports. The Quarterly Journal of Economics, 78(1), 1-14.

Nester, D. C. (1990). Labor exemption to antitrust scrutiny in professional sports. Southern Illinois University Law Review, 15, 123-144.

Noll, R. (1974). Attendance and price setting. In R. Noll (Ed.), Government and the Sports Business. Washington, DC: Brookings Institute.

Quirk, J., & Fort, R. D. (1992). Pay dirt: The business of professional team sports. Princeton, NJ: University of Princeton Press.

Rottenberg, S. (1956). The baseball players labor market. Journal of Political Economy, 64, 242-258.

Schmidt, M. B, & Berri, D.J. (2001). Competitive balance and attendance: The case of Major League Baseball. Journal of Sports Economics, 2(2), 145-167.

Scully, G. (1989). The business of Major League Baseball. Chicago, IL: University of Chicago Press.

Soebbing, B. P. (2008). Competitive balance and attendance in Major League Baseball: An empirical test of the uncertainty of outcome hypothesis. International Journal of Sport Finance, 3(2), 119-126.

Whiting, R. (1990). You gotta have wa. New York, NY: Vintage Books.

Whitney, J. (1988). Winning games versus winning championships: The economic of fan interest and team performance. Economic Inquiry, 26, 703-724.

Nicholas M. Watanabe

University of Missouri

Nicholas M. Watanabe, PhD, is an assistant teaching professor in the Department of Parks, Recreation and Tourism's sport management emphasis. His research interests include sport management and sports economics.

Table 1: Summary Statistics

Variable Mean Standard Dev Min Max

Average Attendance 16001 7089 5356 47609
Noll-Scully 1.855 0.2860 1.423 2.327
Win Percent 0.4827 0.1402 0.1444 0.8778
1st Stage 0.0431 0.2036 0 1
2nd Stage 0.0431 0.2036 0 1
Overall Champ 0.0627 0.2430 0 1
Pre Population (1000) 6172 2776 877 12838
Region Population (1000) 29704 11267 5535 41977
Capacity 32157 16285 9000 72327
New Stadium 0.0510 0.2204 0 1
World Cup Year 0.2392 0.4274 0 1
J-League City 0.2314 0.4225 0 1
J-League Prefect 0.7529 0.8681 0 3
J-League Region 4.125 2.649 0 8
NPB City 0.5451 0.6909 0 3
NPB Prefect 0.8392 0.6711 0 3
NPB Region 3.545 2.198 06 6
Time 8.976 4.498 1 16
Time Squared 100.7 81.42 1 256

n=255

Table 2: Regression Results - Random Effects GLS Regression

 Dependent Variable is Average Attendance

Variable City Prefecture

 Coef. Std Err. Coef. Std Err.

Noll-Scully -3142 1031 *** -3158 1030 ***
Win Percent 7533 2160 *** 7278 2171 ***
1st Stage -1599 1279 -1664 1277
2nd Stage 1425 1242 1190 1412
Overall Champ 1319 1231 1659 1221
Population (1000) 0.2200 0.8227 1.2289 0.7288 *
Capacity 0.1964 0.0263 *** 0.2035 0.0268 ***
New Stadium 439 1094 435 1079
World Cup Year 1206 567 ** 1124 566 **
J-League City 2059 931 ** -- --
NPB City -393 1169 -- --
J-League Prefec -- -- 1373 617 **
NPB Prefec -- -- -757 1454
J-League Region -- -- -- --
NPB Region -- -- -- --
Time -1774 239*** -1770 239 ***
Time Squared 96.65 12.70*** 95.14 12.74 ***
Constant 36442 3356*** 34094 3319 ***
[R.sup.2] 0.7480 0.8019

 Dependent Variable is
 Average Attendance

Variable Region

 Coef. Std Err.

Noll-Scully -3278 1012 ***
Win Percent 7735 2122 ***
1st Stage -1757 1250
2nd Stage 1074 1388
Overall Champ 1220 1205
Population (1000) 1.278 0.7253 *
Capacity 0.1739 0.0266 ***
New Stadium 98.20 1064
World Cup Year 996 561 *
J-League City -- --
NPB City -- --
J-League Prefec -- --
NPB Prefec -- --
J-League Region -724 324 **
NPB Region -1754 984 *
Time -1686 264 ***
Time Squared 84.16 14.26 ***
Constant 13315 16703
[R.sup.2] 0.8091

* p<0.10, ** p<0.05, *** p<0.01