An examination of the effects of the recent economic crisis on Major League Baseball attendance demand.

Hong, Sungil ; Mondello, Michael ; Coates, Dennis 等

An Examination of the Effects of the Recent Economic Crisis on Major League Baseball Attendance Demand

Major League Baseball (MLB) attendance in 2009 was about 6.77% below what it was in 2008 (30,338 per game for 2009 compared to 32,543 for 2008) (Brown, 2009), and 2008 was down slightly compared to 2007. The drop between 2008 and 2009 was the largest single-season loss in attendance since 1952, excluding years involving a work stoppage (Nightengale, 2009). More specifically, 22 of the 30 clubs experienced a decline in attendance, including four teams' attendance decreasing more than 20%. The Florida Marlins, Kansas City Royals, and Texas Rangers were the only teams able to boast 10% or greater increases in attendance (Brown, 2009).

The surprisingly large drop in attendance may be attributable to the recent economic crisis, the beginning of which was set at December 2007 by the National Bureau of Economic Research. In fact, MLB announced that "because of the economy, this year's (2009) total is 6.6 percent less than last year's total, but is actually only 5.2 percent lower when accounting for the reduced capacities of the two new ballparks in New York" (MLB.com, 2009, n. p.). It is certainly a possibility that the state of the economy in 2009 led to reduced attendance. Widespread unemployment and reduced incomes mean tighter budget constraints causing consumers to alter their spending patterns. Consumers may reduce the quantity of sporting events attended in response to the economic crisis both because their financial status is precarious and for psychological reasons. The psychological impact of the crisis may lead to a decrease in expenditures regardless of the consumers' actual financial status (Katona, 1974). This first contribution of this paper is that it empirically assesses the impact of the recent economic crisis on the largest single-season drop in attendance in MLB history between the 2008 and 2009 seasons. The key problem in this study is how best to represent the economic conditions in our analysis. The ideal economic indicators would vary daily and by city, but such data do not exist. The second contribution of this paper is to utilize "coincident indicators," produced by the Federal Reserve Bank of Philadelphia, reflecting the macroeconomic health of a given state in each month. These indicators reflect an improvement over the literature in which income per capita for a specific year is used to capture the role of consumer income in the determination of ticket demand.

The first section of the paper discusses what economic indicators should be used and why they are useful in measuring the effects of the recent economic crisis on MLB attendance demand. In the second section, the attendance demand model is presented along with descriptions of other explanatory variables. Finally, the last section discusses the empirical results, conclusions, and recommendations for future research.

Economic indicators

The Great Depression inspired economists to search for ways to detect or predict economic cycles (Conference Board, 2001). As an initial effort of gauging economic cycles, Arthur Burns and Wesley Mitchell of the National Bureau of Economic Research (NBER) developed a list of leading, coincident, and lagging indicators of economic activity in the United States as part of the NBER research program on business cycles (Stock & Watson, 1989).

Of the leading, lagging, or coincident indicators, the most useful for the study of baseball attendance is the index of coincident indicators. More specifically, the composite index of coincident indicators is designed to measure current economic conditions; that is, the coincident indicators provide a description of the actual circumstances that consumers face at a point in time. Of these three composite indexes of economic indicators, Crone (1994) claimed the composite index of coincident indicators is the most important index for dating business cycles. He stated that 18 of 22 business-cycle turning points in the U.S. economy over a period of 45 years closely corresponded with the index of coincident indicators, indicating an excellent benchmark for assessing the current pace of economic activity (Yamarone, 2004). Therefore, the composite index of coincident indicators is the best of the business cycle indicators to use as a determinant of attendance demand.

Composite index of coincident indicators

Two different coincident indexes have been released by the U.S. Department of Commerce and some regional U.S. Federal Reserve Banks. The index released by the Commerce Department (now published by the Conference Board) consists of four monthly data series--1) the number of jobs in nonagricultural establishments, 2) personal incomes, less transfer payments, adjusted for inflation, 3) the index of industrial production, and 4) manufacturing and trade sales adjusted for inflation (Crone, 1994). The Commerce Department's index is calculated based on month-to-month percent changes for each of these four series. The changes are standardized based on the long-run average absolute monthly change in the series; "this preliminary index is adjusted to grow over time at the same rate as real gross national product and is set to 100 in 1982" (Crone, 1994, p. 21). However, there is a primary issue about the methods used by the Commerce Department--the index is not derived from a formal mathematic or statistical model (Stock & Watson, 1989). More specifically, the same weight is assigned to each indicator in forming the composite index. It is unlikely that each indicator equally reflects the overall state of the economy (Crone, 1994).

In contrast with the traditional Commerce Department methodology, several procedures have been proposed to aid in the dating of recessions and expansions by using techniques based on econometric and time series analysis (Crone, 1994). Stock and Watson (1991) developed a probabilistic state space model based on time-series econometric techniques to estimate a latent process; this estimation is used as a coincident indicator of the economic activity. Their approach was based on the common movements across several economic data series best measuring the business cycles. Unlike the Commerce Department methodology, their methodology assigned the different weights determined by the degree of common movement in the indicators (Crone, 1994).

The Stock and Watson national index used the same data series as the Department of Commerce except for the number of nonagricultural jobs. Employee hours in nonagricultural jobs were used since economic output depends not only on how long they work but also on how many people are working (Crone, 1994). Their national index tracks the official business cycles closer than the method used by the Department of Commerce. However, the Stock and Watson index is not available at the state level and is, therefore, a poor means for measuring the regional economy (Crone & Clayton-Matthews, 2005). Accordingly, Crone and his colleagues of the Federal Reserve Bank of Philadelphia developed a new coincident index for each of the 50 states in order to describe recent economic trends at the state level.

The coincident indexes for the 50 states comprise four state-level indicators to summarize current economic conditions in a single statistic. The four state-level variables in each coincident index are 1) nonagricultural payroll employment, 2) unemployment rate, 3) average hours worked in manufacturing, and 4) real wage and salary disbursements (Crone & Clayton-Matthews, 2005). Nonagricultural payroll employment is intended to reflect actual changes in hiring and firing for the individual states by month. This is considered the most reliable employment series published for all the states.

The unemployment rate produced by the Bureau of Labor Statistics utilizes the current population survey, the payroll employment survey, state population estimates, and unemployment claims. While Stock and Watson's national index and the Conference Board's index include industrial production, the coincident index for the 50 states uses average hours worked in manufacturing in its place due to the unavailability of a comparable measure of industrial output at the state level (Crone & Clayton-Matthew, 2005). Finally, while the Bureau of Economic Analysis releases personal income and its components at the state level on a quarterly basis, the major component of personal income, real wage and salary disbursements, is included in the state index. The quarterly wage and salary disbursements produced by the BEA are adjusted by lagging the structure in the measurement equation for this variable to obtain the monthly data (Crone & Clayton-Matthew, 2005).

Usefulness of the composite index of coincident indicators for the 50 states

Consumer income is a determinant of demand and is commonly accounted for in the literature on attendance demand by income per capita measured at the metropolitan level. One problem with income per capita at the metropolitan level is that it is an annually reported variable so that it will not vary over the seasons. Moreover, as a measure of the financial situation of fans, income also may be lacking as it is only one of many alternative factors reflecting those circumstances. However, the use of a single economic indicator may lead to a different estimation of economic conditions because the indicators do not move together perfectly. Indeed, individual indicators may provide contradictory evidence on economic conditions (Crone, 1994).

In the literature, income is often found not to be a significant determinant of attendance, possibly because of the lack of variation over the league's season or because it is not a good measure of the purchasing power and economic circumstances of the fans (e.g., Alexander, 2001; Coates & Harrison, 2005; Coates & Humphreys, 2005, 2007; Coffin, 1996; Garcia & Rodriguez, 2002; Lemke, Leonard, & Tlhokwane, 2010; McEvoy, Nagel, DeSchriver, & Brown, 2005; Meehan, Nelson, & Richardson, 2007; Morse, Shapiro, McEvoy, & Rascher, 2008). As alternative measures, several attendance demand studies used consumption spending (Bird, 1982) and the unemployment rate (Dobson & Goddard, 1985; Jennett, 1984; Leadley & Zygmont, 2005), resulting in mixed results.

The coincident indexes have an advantage over income per capita and other alternative approaches because they vary by month within the season and they combine information from several indicators so they can better reflect the current economic conditions. One limitation of the coincident indexes is they cover an entire state rather than a metropolitan area. Consequently, if the economic circumstances of the city from which a team draws its fans do not correspond well to the circumstances of the entire state, then the coincident indicator may be a poor variable to capture the influence of income on demand.

In spite of the usefulness of the composite index of coincident indicators, little empirical research on consumer demand has utilized this index as an alternative measure of economic conditions. There are several tourism studies that have used leading indicators, which are widely used to judge or predict the future state of the economy, as explanatory variables in order to forecast tourism demand (e.g., Cho, 2001; Turner, Kulendran, & Fernado, 1997; Wander & Erden, 1979). However, we have found no demand studies that use coincident indicators for this purpose. Thus, we believe this study is a first attempt at adopting the coincident index for the investigation of consumer demand.

The Attendance Model

Attendance demand, like all demand, has well-known theoretical determinants. Among these determinants are income of the consumers, prices of tickets and of other goods and services, and preferences. The difficulty in estimating the demand equation is how to control for these theoretically relevant factors with the existing data. For example, it is difficult to control for preferences because there is no clearly defined or observable way to measure them. Additionally, in the case of sports attendance, ticket sales data are aggregated to the game or season level for a sport club, rather than linked to individual consumers. Consequently, aggregate rather than individual variables are used to proxy for correlates of individual characteristics that are themselves only proxies for tastes and preferences. In the context of individual game attendance, none of these aggregate variables will vary meaningfully and are not collected with sufficient frequency even if they did. Additionally, in the sports context, fan loyalty is important, and while it likely varies from season to season, its variability from game to game within the season is likely to be small. For these reasons, city or team fixed effects are used to capture the influence of these game-to-game time-invariant factors. Consequently, our regression model incorporates the current economic condition indicator as well as the most widely used demand determinants as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

where i indicates team, j indexes games, and t indexes the season, [beta]s are parameters to be estimated, [delta] represents the fixed effect parameter for each team, and [mu] is the error term. The variables used in the analysis are described below.

Data

Data were collected for every regular season game for the 29 MLB teams based in the United States. The composite index of coincident indicators is not available for the Toronto Blue Jays. The sample contains 4,696 games during the two seasons, but two of these are dropped because they were played in Japan. Multiple resources were used to collect the data such as Baseball-Reference.com, ESPN.com, and the Federal Reserve Bank of Philadelphia's website. Descriptive statistics for the variables used in the regression equation are presented in Table 1. Home attendance as a percentage of stadium capacity was 71.33% on average, with a standard deviation of 22.44. The mean of the coincident index for the states where the teams are located was 159.36 with a standard deviation of 17.99. The index declines over the sample period for all of the states, though to varying degrees. Figure 1 displays the index for each state for each month. Arizona, home to the Arizona Diamondbacks, had the highest of the composite index of economic indicators, while Michigan (home of the Detroit Tigers) had the lowest index. Michigan's index changed the most, New York's changed the least.

[FIGURE 1 OMITTED]

Dependent variable (AttenPct)

Because stadiums have widely varying seating capacities, we use each MLB team's game-by-game attendance as a percentage of stadium capacity as our dependent variable. It is common in the literature to use stadium capacity as an explanatory variable in the attendance equation, but our approach means we do not need to do that as a control for stadium size. Moreover, seating capacity is not rightly a determinant of demand, but rather a supply variable, so its inclusion in a demand equation is problematic (Demmert, 1973). Thus, several studies used attendance percentage as the dependent variable (e.g., Fizel & Bennett, 1989; Fort & Rosenman, 1999; Kaempfer & Pacey, 1986; Schollaert & Smith, 1987; Welki & Zlaptoper, 1999). Moreover, as reported in Table 2, the majority of the estimated coefficients when using attendance percentage and attendance level as dependent variables are similar in sign and level of statistical significance. Whether the dependent variable is actual attendance or the share of capacity, the value is limited on the up side. Attendance cannot exceed capacity, and the percentage of capacity cannot exceed 100%. In fact, during the 2008 and 2009 seasons, a total of 524 out of 4696 games were sold out, representing approximately 12.6% of our sample. Consequently, we estimate the model using a censored regression technique. The existence of censored observations constrained by stadium capacity results in parameter estimates that are biased and inconsistent (Meehan et al., 2007) leading us to use censored regression to estimate the model.

Explanatory variables

Economic conditions (Econcon). The coincident index of economic indicators for the 50 states released by Federal Reserve Bank of Philadelphia is used to reflect economic conditions during each month of the 2008 and 2009 seasons. There may be an issue for teams such as the Phillies, Mets, and Yankees, whose territories cover multiple states. Thus, we performed pairwise correlations for the coincident index for the state pairs, using the data from 2006 to the present. All of the correlations are positive and statistically significant and most are in the .9 and above range. Consequently, we feel comfortable linking a team to the coincident index of economic indicators for the state in which its stadium is located. Because of the unavailability of the index for Washington, D.C., the index for the state of Maryland was coded for the Washington Nationals.

Rottenberg (1956) hypothesized that attendance will be affected by the quality of the home team and the uncertainty of the outcome of the game. These variables can be measured in several ways. Our variables are described below:

Game uncertainty (HomeWin and OppWin). Winning percentages of the home and visiting teams prior to the game are used to capture game uncertainty of outcome in this study. Coates and Humphreys (2010) used these winning percentages as indicators of the game uncertainty of the outcome, along with betting line information in the National Football League (NFL). As Coates and Humphreys (2010) mentioned, these winning percentages also capture the quality of the teams.

Rottenberg (1956) suggested attendance would be greatest at games between evenly matched teams, all else constant. However, evidence from Forrest, et al. (2005), Buraimo and Simmons (2008), and Coates and Humphreys (2010) who use betting line information to measure the expected closeness of games, in Football Association games in England and the National Football League, respectively, found home attendance rises as the home team becomes a greater favorite.

Playoff uncertainty (monthsGameBehind). The measure of the playoff uncertainty assumes that when a game is significant in determining promotion or relegation, or for participation in the playoffs or a wildcard race, then fans are more attracted to the game, resulting in higher attendance (Borland & Macdonald, 2003). In the literature, the playoff uncertainty is widely measured by games behind the division leader in contention for playoff qualification (Baade & Tiehen, 1990; Butler, 2002; Canes, 1974; Demmert, 1973; Domazlicky & Kerr, 1990; Hill et al., 1982; Meehan et al., 2007; Noll, 1974; Schmidt & Berri, 2004; Simmons, 2006; Soebbing, 2008). However, one critical issue in this measure is the disability of identifying the time period defining contention for playoff qualifications. Thus, in this study, the month dummy variables are included and then the interaction of the month and the number of games behind the division leader was added. Without the month dummy variables, the interaction variable is likely picking up mostly the month effect, especially early in the season.

Team performance (PlayoffApp). It is reasonable that fans' expectations for the coming year and the decision to buy season tickets could depend on the last season's performance. To reflect these fans' expectations, a number of attendance studies used last season's appearance in the playoffs (Coates & Harrison, 2005; Coates & Humphreys, 2005; Meehan et al., 2007; Noll, 1974; Rivers & Deschriver, 2002). This variable is measured as a dummy variable for whether the home team appeared in the playoffs in the last season.

Interleague Play (Interleague). Since its inception in 1997, interleague play has contributed to an increase in the MLB attendance. According to Brown (2009), interleague play has drawn an average of 33,260 fans per game, compared to the intraleague average of 29,706 fans per game during the same span. This figure indicated that interleague play attracted 12.0 percent more fans than intraleague games. Therefore, a number of attendance demand studies have included the effect of interleague matchups (Boyd & Krebiel, 2006; Butler, 2002; Meehan et al., 2007). A dummy variable was used to measure the effect of interleague play on attendance by coding 1 if the home team played against another league's team and 0 otherwise.

New stadium (Novelty). A large body of attendance research has demonstrated a positive effect of a new stadium on attendance (e.g., Borland & McDonald, 2003; Coates & Humphreys, 2005; McEvoy et al., 2005; Noll, 1974; Zygmont & Leadley, 2006). During the 2008 and 2009 seasons, the Washington Nationals played at a new stadium both years, and the New York Yankees and Mets played at a new stadium in 2009. To capture the novelty effect, the analysis included a dummy variable that takes value 1 when playing at the new stadium and 0 otherwise.

Star player (HomeStar and OppStar). Team composition plays a fundamental role in facilitating fan support (Brandes, Frank, & Nuesch, 2008). Recent studies on attendance demand clearly indicated that star players contribute to driving attendance demand (Berri & Schmdit, 2006). To assess the impact of popularity of star players on attendance, the number of previous season All-Stars until mid-season of the current year, and current season All-Stars for the second half of the season were coded for both home and visiting teams.

Schedule (Weekend, Opening and Holidays). To control for when the game is played, we included weekend, opening day, and holiday (Memorial Day, Independence Day, and Labor Day) game dummy variables. These dummies were used to control for possible unusually high attendance driven by special circumstances.

Average ticket price (AvgTicket). Ticket (or admission) price has been used in almost all attendance demand models in MLB (Coates & Humphreys, 2007). Consistent with demand theory, it is assumed that as price increases game attendance should decrease. Thus, it is expected that the coefficient on the ticket price variable will be negative and significant. Additionally, there has been an argument whether attendance demand is price elastic or inelastic in the literature. Most recent research has confirmed that teams set ticket prices in the inelastic portion of attendance demand, possibly due to other revenue sources such as concessions, local broadcasting, and parking (Coates & Humphreys, 2007; Fort, 2004, Marburger, 1997; Krautmann & Berri, 2007; Zimbalist, 1992). Thus, it is expected that the absolute value of ticket price elasticity is less than 1.

Limitations of the analysis

The data we have used in the analysis is not ideal. Many variables are problematic. Here we describe three limitations of the analysis owing to data issues that have not previously been mentioned.

Ticket prices in attendance demand studies are problematic. The best measure would link the ticket price paid by the marginal ticket buyer. Instead, available ticket price variables are the average ticket price calculated either by dividing ticket revenue by tickets sold or by constructing a weighted average of ticket prices by the proportion of seats available at that price. Variable ticket pricing by day of the week or opponent has become more common and is prevalent in our data. We are not aware of any game-by-game ticket price data for all the clubs in Major League Baseball, and even if that data existed it would suffer from being an average. Consequently, while our ticket price variable is not ideal, we believe it is a reasonable approximation of the relevant price.

A second limitation is related to playoff contention. We include games behind the division leader and that variable interacted with the month dummy variables. But games behind the division leader may be less important in our time period than it was for studies conducted before the advent of the wildcard in baseball playoffs. We have considered games behind in the wild card race and believe that games behind the division leader addresses this issue adequately. In both the 2008 and 2009 seasons, and in both the American and the National Leagues, the wild card winner had a better record at the end of the season than at least one of the division winners. Moreover, in three of four cases, the team in second in the wild card race was farther behind the wild card leader than behind their division leader at the end of the season.

The third limitation is the lack of weather data in the model. Past attendance models have measured the effect of weather on daily attendance, but we have not included that in the current analysis. Two approaches are common in the attendance demand studies. The first approach is concerned with the temperature for the day of the game. For example, as continuous variables, the average of the daily low and high temperature (Meehan et al., 2007) and the temperature reported during the game (Bruggink & Eaton, 1996; Paul, Paul, & Yelencsics, 2008) were used. Meanwhile, Butler (2002) used two dummy variables; one is a dummy variable equal to 1 if game temperature was less than 55[degrees] F; the other is a dummy variable equal to 1 if game temperature was greater than 94[degrees]F. The second approach concerns rain. Meehan et al. (2007) used the number of inches of rain for a given day, while DeSchriver (2007) and Butler (2002) included a dummy variable to indicate whether the game was played during rainy weather conditions. These different approaches to measurement of weather conditions have generated contradictory results. Researchers should therefore address more accurate methods that can be employed to assess the impact of weather conditions. However, omitting weather should not alter our results regarding the coincident indicator, because it is highly unlikely that daily weather and monthly coincident indicator are highly correlated.

Empirical Results and Discussion

Table 2 reports regression results from the censored regression analysis with the team fixed effects. The censored regression model was found to be significant with a log likelihood statistic of -16851.31.

The coefficient on the coincident indicator variable Econ is significant and positive, as expected. This clearly indicates that healthier economic conditions are associated with higher attendance. Thus, the decrease in the composite index between 2008 and 2009, representing the recent economic downturn throughout the whole economy, influenced a decrease in attendance between the 2008 and the 2009 seasons. The estimated regression coefficient of 0.38 appears small. However, at the mean values of attendance percentage and Econ, the coefficient estimate implies an elasticity of 0.85. In other words, a one percent increase in Econ, the coincident indicator, implies a 0.85 percent increase in attendance. For the period of our data, the average decline in Econ is over 7.7%, suggesting a rough impact of the economic decline on attendance from the start of 2008 until the end of 2009 of about 6.5%, very nearly the 6.77% actual decline mentioned in the introduction.

The following variables were used to capture the importance of competitive balance and the uncertainty of outcome hypotheses. For game uncertainty, the coefficients on the home (HomeWin) and visiting team's (OppWin) current winning percentage to date variables are statistically significant and positive. This result supports the idea that fans are eager to see good teams play, regardless of whether it is the visiting team (Coates & Humphreys, 2010). Thus, fans consider the quality of both home and visiting teams for attendance. For playoff uncertainty, the coefficients on the playoff uncertainty variables before the month of July (MarchGameBehind, AprGameBehind, MayGameBehind, JunGameBehind) were found to be insignificant while these variables for the months of July, August, September and October (JulyGameBehind, AugGameBehind, SepGameBehind, OctGameBehind) were significant and negative. This has two important implications. One defines when fans start to recognize playoff contention and the other is when the effect of the division leader on attendance is more prominent than other months. Indeed, the results are inconsistent with Noll's (1974) judgment on when the contention, or race, for playoff spots begins. Specifically, he judged the playoff contention as if the second-place team averaged five games or fewer behind the leader between August 1 and the end of the season. However, it should be noted that he did not provide statistical justification why the playoff contention starts from the month of August, rather it was based on his personal impressions.

The significant negative coefficients on the month dummy variables indicated average attendance is lower in other months relative to average attendance at games in March, the omitted month in the regression. There were very few games played in March, 15 out of 4,696, and two of those were played in Japan. These March games are, of course, "Opening Day" games for the home teams and, therefore, have unusually high attendance. To control this abnormal attendance, the opening day dummy was included, resulting in the very strong impact on attendance. It is, therefore, not surprising that average attendance in other months is lower than in March. More importantly, consider the pattern in the coefficients on the month dummy variables. As the weather heats up through spring and into summer, and kids get out of school, average attendance is rising--the month coefficients increase (-2.10 in April, 2.46 in May, 3.94 in June, 10.70 in July)--until July. Average attendance starts to decline again through August and September, until it jumps up substantially in October. Not much should be made of this increase in October, however, as there are only 56 games played in October in the data compared to between 700 and 825 games each month from April through September.

Surprisingly, the PlayoffApp had a negative impact on attendance. The negative impact of the PlayoffApp may be explained with two features of the teams that appeared in the 2007 and 2008 playoffs. One feature is that several teams that appeared in previous playoffs showed poor performance in that year. For example, the Cleveland Indians and the Colorado Rockies appeared in the 2007 playoffs but performed poorly in the 2008 season, resulting in a decrease in attendance. The Chicago White Sox and the Milwaukee Brewers appeared in the 2008 playoffs and also performed poorly in the 2009 season, resulting in a decrease in attendance. If one includes interactions between the playoff appearance dummy and the month of the season dummies (excluding an October interaction), those interactions are each individually statistically significant, and negative, while the playoff appearance dummy is positive and statistically significant. (These results are available upon request.) This relates to a second possible explanation for the negative coefficient on the playoff appearance variable; several teams' continued dominance may result in fans losing interest, particularly early in the season. Eckard (2001) found a significant decrease in attendance when a team is on a run of dominance. Thus, the appearances of several dominant teams in the 2007 and/or 2008 playoffs, such as the New York Yankees, the Los Angeles Angels, the Chicago Cubs, the Boston Red Sox, and the Los Angeles Dodgers may generate the possible loss of fan interest, or not produce additional interest in the team, resulting in lower early season attendance. The estimated coefficients on the playoff appearance month interactions suggest that interest, in the form of attendance, rises slowly throughout the season for the teams that repeatedly contend for the playoffs.

The estimated coefficients on the all-star variables for home (HomeStar) and visiting teams (OppStar) were significant and positive. What this result reveals is that home spectators are interested in seeing both the home teams' and visiting teams' star players. Visiting teams' star players' scarcity value (produced by less opportunity of seeing them) is one possible explanation for the positive association with attendance. Berri and Schmidt (2006) document the impact of star players from the visiting team on attendance in the NBA.

The novelty effect was positive and statistically significant despite only three teams playing at a new stadium during the time span used. The coefficients on the weekend, the holidays and the interleague variables were also significant and strongly positive, as expected. These results are consistent with findings of previous attendance studies.

Finally, the AvgTicket was found to be negative and statistically significant. The negative sign on this variable indicated that higher ticket prices were associated with lower attendance, as implied by demand theory. Of the 29 MLB teams used in the analysis, eight teams (the Arizona Diamondbacks, Baltimore Orioles, Cincinnati Reds, Cleveland Indians, Los Angeles Dodgers, Oakland Athletics, Pittsburgh Pirates, and San Diego Padres) lowered seasonal average ticket prices in the 2009 season. However, the lowered ticket prices in the 2009 season did not seem to help boost attendance among the eight teams, as reflected by each team's decline in attendance (i.e., the Arizona Diamondbacks: -9.6%; Baltimore Orioles: -3%; Cincinnati Reds: -9.1%; Cleveland Indians: -10.7%; Los Angeles Dodgers: 0.7%; Oakland Athletics: -8.5%; Pittsburgh Pirates: -1.6%, and San Diego Padres: -14.7%).

The estimated absolute value of ticket price elasticity (.04 < 1) supports previous findings of ticket pricing in the inelastic portion of the demand (Coates & Humphreys, 2007). In other words, MLB teams could change ticket prices in response to a change in demand reflected in attendance (Narayan & Smyth, 2003). As Coates and Humphreys (2007) mentioned, this is due to "the inter-related pricing decision on tickets, concessions, and other related goods made by revenue maximizing monopolists in the NBA and MLB" (p. 169).

Conclusion

Using 4,696 games during the 2008 and 2009 seasons, the main focus of the proposed study was to investigate the effect of the recent economic crisis on attendance in MLB. Additionally, we included a variety of standard explanatory variables based on attendance demand theory to isolate the influence of the general economic circumstances, represented by the coincident indicator, on game-by-game attendance. In this section, the findings and the contributions of this paper are reviewed.

First, our findings with respect to the control variables are consistent with theoretical expectations. Ticket price exerts a negative and statistically significant effect on game attendance. Better teams, based on winning percentage to date, draw better, whether they are home or visitors. Teams with more star players draw better than teams with few stars. Interleague games draw better than intraleague games, and Opening Day, holidays, and weekend games draw better than weekday games.

Second, as problematic as it may be, the games behind variables provide an interesting implication pertaining to the time period defining contention for playoff qualification. Indeed, previous research only demonstrated fans lose their interests as a team is getting farther behind the division leader. However, the negative and significant playoff uncertainty variables for July, August, September and October seem to indicate that fans start to recognize playoff contention in July. It is possible that the present findings are partly attributable to the increased media coverage of MLB playoff races that begins after the All-Star Game.

Finally, this study suggests a previously unused measure of the economic conditions, the composite index of coincident indicators, to control for the influence of income on ticket demand. Indeed, most research on attendance demand has used annual per capita income despite its lack of variation over the season, with mixed results. To overcome the limitations of per capita income, this study first applies the composite index of coincident indicators as a measure of buying power and economic circumstances. The empirical evidence indicates the recent economic crisis contributed to a decline in MLB attendance over the period 2008 through 2009; the implication of our results is that deteriorating economic circumstances explain a decline of about 6.5% compared to the reported decline of 6.77%. The success of the composite index of coincident indicators in explaining the impact of the recent economic crisis on attendance in MLB suggests the indicator is a viable proxy for income in game attendance demand studies. As a practical matter, the findings here provide information that can help team marketers schedule events and promotions and set prices throughout the season.

References

Alexander, D. L. (2001). Major League Baseball monopoly pricing and profit-maximizing behavior. Journal of Sports Economics, 2(4), 341-355.

Baade, R. A., &Tiehen, L. J. (1990). An analysis of Major League Baseball attendance, 1969-1987. Journal of Sport and Social Issues, 14(1), 14-32.

Berri, D. J., & Schmidt, M. B. (2006). On the road with the national basketball association's superstar externality. Journal of Sports Economics, 7(4), 347-358.

Bird, P. J. (1982). The demand for league football. Applied Economics, 14(6), 637-649.

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

Boyd, C. T., & Krehbiel, C. T. (2006). An analysis of the effects of specific promotion types on attendance at Major League Baseball games. Mid-American Journal of Business, 21(2), 21-32.

Brandes, L., Franck, E., & Nuesch, S. (2008). Local heroes and superstars: An empirical analysis of star attraction in German soccer. Journal of Sports Economics, 9(3), 266-286.

Brown, M. (2009). Inside the numbers: Final 2009 MLB regular season attendance. Retrieved from http://www.bizofbaseball.com/index.php?option=com_content&view=article&id= 3592:inside-the-numbers-final-2009-mlb-regular-season-attendance&catid=56:ticketwatch&Itemid=136

Bruggink, T. H., & Eaton, J. W. (1996). Rebuilding attendance in Major League Baseball: The demand for individual games. In J. Fizel, E. Gustafson & L. Hadley (Eds.), Baseball economics: Current research (pp. 9-31). Westport, CT: Praeger.

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, 146155.

Butler, M. R. (2002). Interleague play and baseball attendance. Journal of Sports Economics, 3(4), 320-334.

Canes, M. E. (1974). The social benefits of restrictions on team quality. In R. G. Noll (Ed.), Government and the sports business (pp. 81-113). Washington, DC: Brookings Institution.

Cho, V. (2001).Tourism forecasting and its relationship with leading economic indicators. Journal of Hospitality & Tourism Research, 25, 399-420.

Coates, D., & Harrison, T. (2005).Baseball strikes and the demand for attendance. Journal of Sports Economics, 6(3), 282-302.

Coates, D., & Humphreys, B. R. (2005). Novelty effects of new facilities on attendance at professional sporting events. Contemporary Economic Policy, 23(3), 436-455.

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

Coates, D., & Humphreys, B. R. (2010). Week to week attendance and competitive balance in the National Football League. International Journal of Sport Finance, 5(4), 239-252.

Coffin, D. A. (1996). If you build it, will they come? Attendance and new stadium construction. In J. L. Fizel, E. Gustafson, & L. Hadley (Eds.), Baseball economics: Current research (pp. 3346). Westport, CT: Praeger.

Conference Board. (2001). Business cycle indicators handbook. New York, NY: The Conference Board.

Crone, T. M. (1994). New indexes track the state of the states. Business Review, January/February, 19-31.

Crone, T. M., & Clayton-Matthews, A. (2005). Consistent economic indexes for the 50 states. Review of Economics and Statistics, 87(4), 593-603.

Demmert, H. G. (1973). The economics of professional team sports. Lexington, MA: Lexington Books.

DeSchriver, T. D. (2007). Much Adieu about Freddy: Freddy Adu and attendance in Major League Soccer. Journal of Sport Management, 21(3), 438-451.

Dobson, S. M., & Goddard, J. A. (1995). The demand for professional league football in England and Wales, 1925-92. The Statistician, 259-277.

Domazlicky, B., R., & Kerr, P., M. (1990). Baseball attendance and the designated hitter. American Economic Review, 34, 62-68.

Eckard, E. W. (2001). Baseball's Blue Ribbon economic report: Solutions in search of a problem. Journal of Sports Economics, 2(3), 213-227.

Fizel, J. L., & Bennett, R. W. (1989). The impact of college football telecasts on college football attendance. Social Science Quarterly,70(4), 980-988.

Forrest, D., Simmons, R., & Buraimo, B. (2005). Outcome uncertainty and the couch potato audience. Scottish Journal of Political Economy, 52(4), 641-661.

Fort, R. (2004). Inelastic sports pricing. Managerial and Decision Economics, 25(2), 87-94.

Fort, R., & Rosenman, R. (1999). Sports economics: Current research. In E. G. J. Fizel, & L. Hadley (Ed.), Streak management (pp. 119-134). Westport, CT: Praeger Publishers.

Garcia, J., & Rodriguez, P. (2002). The determinants of football match attendance revisited. Journal of Sports Economics, 3(1), 18-38.

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

Jennett, N. (1984). Attendances, uncertainty of outcome and policy in Scottish League Football. Scottish Journal of Political Economy, 31(2), 176-198.

Kaempfer, W. H., & Pacey, P. L. (1986).Televising college football: The complementarity of attendance and viewing. Social Science Quarterly, 67(1), 176-185.

Katona, G. (1974). Psychology and consumer economics. The Journal of Consumer Research, 1(1), 1-8.

Krautmann, A., & Berri, D. (2007).Can we find it at the concessions? Understanding price elasticity in professional sports. Journal of Sports Economics, 8, 183-191.

Leadley, J. C., & Zygmont, Z. X. (2005). When is the honeymoon over? National Basketball Association attendance 1971-2000. Journal of Sports Economics, 6(2), 203-221.

Lemke, R. J., Leonard, M., & Tlhokwane, K. (2010). Estimating attendance at Major League Baseball games for the 2007 season. Journal of Sports Economics, 11(3), 316-348.

Marburger, D. R. (1997). Optimal ticket pricing for performance goods. Managerial and Decision Economics, 18(5), 375-381.

McEvoy, C. D., Nagel, M. S., DeSchriver, T. D., & Brown, M. T. (2005). Facility age and attendance in Major League Baseball. Sport Management Review, 8(1), 19-41.

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.

MLB.com. (2009, n. p.). MLB finishes 2009 season with fifth highest attendance ever. Retrieved from http://mlb.mlb.com/news/press_releases/pressrelease.jsp?ymd=20091006&content_id =7371722&vkey=pr_mlb&fext=.jsp&c_id

Morse, A. L., Shapiro, S. L., McEvoy, C. D., & Rascher, D. A. (2008). The effects of roster turnover on demand in the National Basketball Association. International Journal of Sport Finance, 3, 8-18.

Nightengale, B. (2009, September 30). Decline at gate a challenge for undeterred commissioner Selig. USA Today. Retrieved from http://www.usatoday.com/sports/baseball/2009-09-29mlb-attendance_N.htm

Noll, R. (1974). Attendance and price setting. In R. Noll (Ed.), Government and the sports business (pp. 115-157). Washington, DC: Brookings Institute.

Paul, R. J., Paul, K. K., & Yelencsics, K. (2008). Good day, sunshine: Attendance for the Florida state league. Southern Business & Economic Journal, 31(1/2), 63-74.

Rivers, D. H., & Deschriver, T. D. (2002). Star players, payroll distribution, and Major League Baseball attendance. Sport Marketing Quarterly, 11, 164-173.

Rottenberg, S. (1956). The baseball players' labor market. The Journal of Political Economy, 64(3), 242-258.

Schmidt, M. B., & Berri, D. J. (2004). The impact of labor strikes on consumer demand: An application to professional sports. American Economic Review, 94(1), 344-357.

Schollaert, P. T., & Smith, D. H. (1987). Team racial composition and sports attendance. The Sociological Quarterly, 28(1), 71-87.

Simmons, R. (2006). The demand for spectator sports. In W. Andreff & S. Szymanski (Eds.), Handbook on the economics of sport (pp. 77-89). Cheltenham, UK: Edward Elgar Publishing Limited.

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, 119-216.

Stock, J. H., & Watson, M. W. (1989). New indexes of coincident and leading economic indicators. NBER macroeconomics annual, 4, 351-394.

Stock, J. H., & Watson, M.W. (1991). A probability model of the coincident economic indicators. In K. Lahiri & G. H. Moore (Eds.), Leading economic indicators: New approaches and forecasting records (pp. 63-85). New York, NY: Cambridge University Press.

Turner, L. W., Kulendran, N., & Fernando, H. (1997). Univariate modelling using periodic and non-periodic analysis: inbound tourism to Japan, Australia and New Zealand compared. Tourism Economics, 3(1), 39-56.

Welki, A. M., & Zlatoper, T. J. (1999). U.S. professional football game-day attendance. Atlantic Economic Journal, 27(3), 285-298.

Yamarone, R. (2004). The trader's guide to key economic indicators. Princeton, NJ: Bloomberg Press.

Zimbalist, A. (1992). Salaries and performance: Beyond the Scully model. In P. Sommers (Ed.), Diamonds are forever: The business of baseball (pp. 109-133). Washington, DC: Brookings Institution.

Zygmont, Z. X., & Leadley, J. C. (2006). When is the honeymoon over? National Hockey League attendance, 1970-2003. Canadian Public Policy, 32(2), 213-232.

Sungil Hong [1], Michael Mondello [2], and Dennis Coates [3]

[1] Hong Kong Baptist University

[2] University of South Florida

[3] University of Maryland, Baltimore County

Sungil Hong is an assistant professor in the Department of Physical Education. His current research focuses on attendance demand and innovation diffusion in professional sports.

Michael Mondello teaches finance and analytics the MBA sport and entertainment management program at the University of South Florida. His research interest focuses on financial/economic issues related to sports organizations including competitive balance, economic impact analysis, contingent valuation, ticket pricing, and stadium financing.

Dennis Coates is a professor of economics at UMBC. His published work has ranged from topics relating to voting behavior of legislators to the economic effects of stadiums and professional sports franchises.

Table 1. Descriptive statistics for the variables to predict MLB
attendance demand during the 2008 and 2009 seasons (N = 4694)

Variable           Mean     SD      Minimum   Maximum

AttendancePct      71.33    22.44   21.00     100.00
Econcon            159.36   18.00   112.86    226.08
Interleague        0.10     0.30    0.00      1.00
HomeWin            49.98    9.79    0.00      100.00
OppWin             50.11    10.01   0.00      100.00
PlayoffApp         0.29     0.46    0.00      1.00
HomeAllstar        2.25     1.47    1.00      8.00
OppAllstar         2.18     1.43    1.00      8.00
Weekend            0.48     0.50    0.00      1.00
Holidays           0.02     0.13    0.00      1.00
Opening            0.01     0.11    0.00      1.00
AvgTicket          26.10    10.20   14.31     72.97
Novelty            0.05     0.22    0.00      1.00
MarchGameBehind    0.00     0.04    -0.50     1.50
April              0.15     0.36    0.00      1.00
AprGameBehind      0.29     1.17    -6.50     9.50
May                0.18     0.38    0.00      1.00
MayGamebehind      0.59     2.14    -8.50     14.00
June               0.16     0.37    0.00      1.00
JunGamebehind      0.81     2.86    -9.50     18.50
July               0.16     0.37    0.00      1.00
JulyGamebehind     0.90     3.45    -10.00    27.00
August             0.17     0.38    0.00      1.00
AugGamebehind      1.42     5.12    -18.00    31.50
September          0.16     0.37    0.00      1.00
SepGamebehind      1.75     6.21    -21.00    41.00
October            0.01     0.11    0.00      1.00
OctGamebehind      0.11     1.64    -8.50     40.00

Table 2. Summary of Regression Analysis for
Variables Predicting MLB Attendance

Variables      Attendance Percentage (DV)          Attendance (DV)

               [beta]   Stnd.   t-stat   P-value   [beta]     Stnd.
                        Error                                 Error

Econcon        0.38     0.04    10.56    0.00      193.79     15.67
Interleague    8.06     0.84    9.57     0.00      3619.01    369.39
HomeWin        0.13     0.04    3.56     0.00      57.69      16.17
OppWin         0.13     0.02    5.79     0.00      57.67      9.98
PlayoffApp     -3.62    0.75    -4.83    0.00      -1492.43   328.01
HomeAllstar    0.68     0.22    3.14     0.00      288.82     95.22
OppAllstar     2.02     0.14    14.14    0.00      911.20     62.46
Weekend        13.88    0.40    34.92    0.00      6124.33    174.34
Holidays       6.72     1.57    4.28     0.00      3258.88    688.00
Opening        44.91    2.37    18.99    0.00      19942.94   1033.45
AvgTicket      -0.12    0.05    -2.29    0.02      -42.44     23.75
Novelty        6.54     1.33    4.90     0.00      -3712.05   585.93
MarchGame      6.93     6.99    0.99     0.32      1751.13    3097.72
Behind
April          -2.10    5.56    -0.38    0.71      -2075.44   2427.84
AprGame        0.23     0.26    0.90     0.37      67.23      112.34
Behind
May            2.46     5.62    0.44     0.66      -32.52     2455.83
MayGame        -0.08    0.13    -0.58    0.56      -57.36     58.45
behind
June           3.94     5.65    0.70     0.49      568.11     2470.33
JunGame        -0.05    0.10    -0.45    0.65      -40.30     45.20
behind
July           10.70    5.64    1.90     0.06      3833.84    2463.29
JulyGame       -0.18    0.08    -2.15    0.03      -126.80    35.80
behind
August         10.27    5.63    1.82     0.07      3617.53    2461.05
AugGame        -0.22    0.06    -3.85    0.00      -126.37    24.90
behind
September      7.23     5.64    1.28     0.20      2390.24    2463.64
SepGame        -0.33    0.05    -6.86    0.00      -171.53    21.02
behind
October        16.26    6.17    2.63     0.01      7222.95    2703.40
OctGame        -0.88    0.16    -5.41    0.00      -457.46    71.95
behind

Variables      Attendance (DV)

               t-stat   P-value

Econcon        12.37    0.00
Interleague    9.80     0.00
HomeWin        3.57     0.00
OppWin         5.78     0.00
PlayoffApp     -4.55    0.00
HomeAllstar    3.03     0.00
OppAllstar     14.59    0.00
Weekend        35.13    0.00
Holidays       4.74     0.00
Opening        19.30    0.00
AvgTicket      -1.79    0.07
Novelty        -6.34    0.00
MarchGame      0.57     0.57
Behind
April          -0.85    0.39
AprGame        0.60     0.55
Behind
May            -0.01    0.99
MayGame        -0.98    0.33
behind
June           0.23     0.82
JunGame        -0.89    0.37
behind
July           1.56     0.12
JulyGame       -3.54    0.00
behind
August         1.47     0.14
AugGame        -5.07    0.00
behind
September      0.97     0.33
SepGame        -8.16    0.00
behind
October        2.67     0.01
OctGame        -6.36    0.00
behind

The model also includes home team dummy variables.