The monetary value of competitive balance for sport consumers: a stated preference approach to European professional football.

Pawlowski, Tim ; Budzinski, Oliver

Introduction

Ever since the pioneering work of Rottenberg (1956) and Neale (1964), the uncertainty of outcome hypothesis (UOH) has played a major role in the economic analysis of professional sport leagues. The UOH suggests that increasingly imbalanced leagues potentially influence fan interest in a negative way and, consequently, stadium attendance and TV viewership will decrease.

Against this background, the increasing imbalance of the top five national football leagues in Europe as well as of the UEFA Champions League may reflect a worrying development (Pawlowski, Breuer, & Hovemann, 2010). Moreover, other European football leagues also increasingly become less balanced (e.g., Denmark) or have displayed a comparatively and considerably low degree of competitive balance (CB) over years (e.g., the Netherlands) and might face the risk of losing fan interest. Figure 1 presents the competitive balance ratio (a measure developed by Humphreys, 2002) for Germany, Denmark, and the Netherlands for the last two decades illustrating this decline in CB.

However, decades of empirical research within sports economics have not been successful in establishing clear evidence for the importance of the uncertainty of match or seasonal outcomes for attendance or TV viewers in European professional football. (1)

Thus, a puzzling gap between major and well-established sports economic insights and the apparently actual behavior of sports fans--the consumers of the product-- surfaces and calls for analyzing possible reasons for this gap. One avenue of research that could contribute to closing this gap tackles possible mismatches in the conceptualization of CB in empirical research, on the one hand, and in the eyes of the fans, on the other hand. In other words, fans may perceive CB in a different way than it is measured in science. As a consequence, the lack of evidence for CB influencing fans' behavior might be caused by attributing CB's influence on fans' behavior to measures of CB that do not reflect how CB actually influences fans' consumption decisions.

The resulting question is whether it is really appropriate to investigate the effects of CB based on aggregate (past) attendance figures. This requires an unambiguous, stable, and continuous relationship between variations of CB and consumption reactions of fans. However, if this relationship is characterized by discontinuities and threshold effects, then the traditional approach may lose some of its explanatory value. For instance, if the (statistically measured) observed variations in CB have not been large enough to bother fans, they may not affect their consumption patterns and, consequently, demand will not be influenced. It might well be that fans' decision to consume does not react continuously to variations in CB. As long as a sufficient level of CB is not undercut, consumption patterns remain stable or even constant. However, if CB falls below that critical level, then a massive drop in demand occurs. If this is the case, another question arises against the background of the decreasing CB in many European football leagues: are we in fact close to some tipping point beyond which poor CB would deter fans?

In order to approach such kinds of phenomena, it is helpful to distinguish between objective competitive balance (OCB) and perceived competitive balance (PCB). OCB refers to the statistically measured competitive balance, whereas PCB denounces the competitive balance as perceived by the fans. Traditional approaches in sports economics assume an identity between OCB and PCB (i.e., OCB = PCB). Only if this holds true, the statistically measured OCB is a perfect proxy for PCB and qualifies as a good measure to test theories about how CB influences fans' consumption behavior. The hypothesis that we formulate in the preceding paragraph, however, states a case where OCB and PCB are not identical. In order to empirically address PCB in a direct way, we need to ask the fans, i.e., to make use of a stated preferences approach (see more details in the following section). In doing so, we find support for the hypothesis that PCB can considerably differ from OCB, so that OCB is not necessarily a good proxy for PCB.

While a previous paper (Pawlowski, 2013b) is focused on the fans' intention to consume in the German league only, this paper employs a different stated preference approach (i.e., the fans' willingness-to-pay [WTP]), and includes data on three major leagues from Germany, Denmark, and the Netherlands. (2) In doing so, it provides a robustness check on the previous research. However, the main point of this paper is to provide a unique contribution by analyzing possible (systematic) differences between PCB and OCB.

Despite the necessity of further research to corroborate our findings, the paper provides evidence for the relevance of discontinuities in the relationship of CB and demand reactions. The results indicate that there is indeed a tipping point (threshold) above which the consumption reaction is rather inelastic to variations in CB. However, crossing the threshold leads to a massive demand reaction. This insight obviously has important implications both for sports economics research and the management of sports leagues.

Research Design and Data Analysis

This section provides details on the sample selection procedure, the PCB and WTP measures, as well as the estimator employed in this paper. (3)

To analyze the PCB and WTP by the fans, a written survey among football fans in Europe was conducted. (4) In contrast to previous research on football fans' perceptions in Europe (inter alia, Konigstorfer, Groeppel-Klein, & Kunkel, 2010), three countries (Germany, Denmark, and the Netherlands) with different levels of 'quality' of their major leagues were selected. Quality, here, refers to international competitiveness as measured by the UEFA ranking where (at the time of investigation) Germany was third, the Netherlands were ninth, and Denmark was 12th. Since it might well be that fans in the stadium differ from fans watching football on TV with regard to their perception of CB, both "types" of fans have been included in the sample. Therefore, the survey took place in the stadium as well as in bars where football matches are broadcasted live. (5)

Furthermore, to control for possible heterogeneity between fans of different teams, cities were chosen with different types of first division teams performing either "constantly good" (CGP), "constantly bad" (CBP), or "volatile" (VP) during the last 10 years (see Table 1).

The survey was conducted in German, Dutch, and Danish language. The overall degree of PCB in the German Bundesliga was assessed via the following question: "Thinking back to previous seasons, how would you rate the level of suspense of the Bundesliga on a scale of 0-10 (0=not at all suspenseful...10=very suspenseful)?"

"Suspense" is written in italics since the native terms that we used (for instance, the German term "Spannung") can be misunderstood when translated into English without the context of the overall sentence. For instance, a dictionary translation of "Spannung" into English may also yield "excitement" next to "suspense," "tension," or "tenseness" and, obviously, excitement can include many other dimensions than close competition between the playing teams and uncertainty of outcome.

However, we are very confident that in all the three countries, the native terms in the context of the wording of the questions were understood to target suspense in the sense of close competition and outcomes remaining uncertain for a long time. To test this assumption, 11 (in addition to the overall PCB measure) inquired items reflecting the short-, mid-, and long-term UO were first reduced by applying a principal component analysis and then tested as correlates of the overall degree of PCB by applying ordered probit and logit models with robust standard errors as well as clustered standard errors clustered by favorite teams. As shown in Pawlowski (2013a) for all three countries and in Pawlowski (2013b) for the German subsample only, the overall degree of PCB is (partly) explained by these reduced items.

Three questions related to the overall degree of PCB--"At which level of overall suspense (on a scale of 0-10) would you (1) ...start to lose interest in the league; (2) ...not watch a match in the stadium; (3).. .not watch a match on television"--were aimed at the fans' intention to consume.

Furthermore, two distinct scenarios were tested to investigate the WTP of fans for CB: "Imagine you could increase the level of suspense in the LEAGUE by making a financial contribution!" as well as "Imagine you could make sure that the level of suspense in the LEAGUE does not decrease in the future by making a financial contribution!" (6)

The WTP-data is used to estimate two PCB-dependant demand functions. These so called Kaplan-Meier survival functions are simply derived by arranging the sample's WTP values in ascending order and calculating the proportion of the sample that have a WTP greater than each value (Bateman et al., 2002):

S(WTP) = [n.sub.j]/N j = 0 to J [n.sub.j] = [J.summation over (k = j+1)][f.sub.k] (1)

with [f.sub.k] = fan k in the sample; N = total number of fans in the sample and [n.sub.J] = total number of fans in the sample with a WTP that is greater than WT[P.sub.j]. The mean WTP is the area bounded by the Kaplan-Meier survival function (Bateman et al., 2002):

[bar.WTP] = [J.summation over (j=0)] S(WT[P.sub.j])(WT[P.sub.j+a] - WT[P.sub.j]) (2)

In addition, the median value is displayed at the point at which the function reaches a probability of 0.5 (Bateman et al., 2002).

Results

Overall, the inquiries took place before/during 14 matches in the first divisions of the respective leagues and the complete data base contains n=1,689 observations (with n =1,203 for Germany; n =267 for Denmark, and n =219 for Netherlands).

Table 2 summarizes the distribution of variables reflecting interest and consumption patterns as well as the socio-demographic and economic characteristics of respondents in the German, Danish, and Dutch samples.

As shown in Table 2, the majority of respondents is highly interested in football and regularly consumed first division football in the previous season by watching either football matches live in the stadium and/or on TV. While the distribution with regard to the general interest in football is fairly similar in the three countries, differences occur with respect to the underlying consumption patterns, i.e., the respondents attend fewer matches in the stadium and watch fewer matches on TV in the Danish and Dutch samples compared to the German sample.

Furthermore, the majority of respondents are male and young in all countries. However, while the majority of respondents are unmarried in the German and Danish samples, the majority of respondents in the Dutch sample are married. Also, some country specific differences occur with regard to the distributions of household size, monthly household net income, and work status.

In general, some country specific differences might be the result of different habits and preferences in the three countries analyzed. However, due to the rather small sample sizes we cannot rule out the possibility of specific selection biases. Since this is the first study to focus on the specific population of football fans and the distribution of characteristics across the total population of football fans in the three countries is unknown, it is difficult to judge the representative nature of the three subsamples. However, a rough comparison between the German subsample in this study and a representative study conducted by SPORTFIVE (2009) does not suggest the existence of significant bias in the German subsample (Pawlowski, 2013a, b).

The following figure provides an overview on the distribution of the PCB measure in the overall sample as well as the subsamples. In general, the distributions are skewed to the left indicating a rather high degree of PCB in the analyzed leagues. However, country specific differences in the degree of perceived CB are obvious: first and foremost, Danish fans perceive the Danish first division to be less balanced (mean: 6.62) than the German (mean: 8.11) and Dutch (mean: 7.75) fans perceive the degree of balance in their first divisions.

[FIGURE 2 OMITTED]

To test the relevance of PCB for football fans, WTP-questions have been formulated as discussed above. Figures 3 and 4 display the estimated Kaplan-Meier survival functions for both, the WTP to ensure the current level of PCB as well as the WTP to increase the current level of PCB in the respective league. Furthermore, median and mean WTP values are displayed with the corresponding standard errors for the latter derived by bootstrapping with r=999.

More than 50% of the respondents are (in general) willing to pay for either improving or maintaining the current degree of PCB in the corresponding national leagues. The average value is around 3 Euros per stadium ticket per game. However, it turns out that the Danish fans are willing to pay even more than 5 Euros per stadium ticket per game to increase the current level of PCB in the Danish league.

Discussion

The results highlight the country specific (monetary) values of PCB by the fans and offer important insights into the role of CB policy in professional sports. First, in most cases only marginal differences exist between the three countries. However, Danish fans are willing to pay more than 160% of the value that fans in Germany and the Netherlands are willing to pay (five Euros compared to three Euros) to increase the current level of CB and, furthermore, are considerably more sensitive to changes in PCB. Consequently, in the eyes of the fans competitive imbalance appears to be a more serious issue in the Danish league compared to the other two.

[FIGURE 3 OMITTED]

At first sight, this represents an astonishing result since the statistically measured levels of CB (= OCB-levels) display a comparatively well-balanced Danish league (again compared to the other two).

So, while

OC[B.sub.Denmark] > OC[B.sub.Germany] > OC[B.sub.Netherlands] (3)

follows from the statistical competitive balance ratio (see Figure 1), the perception of the fans in the three countries lead to a

PC[B.sub.Denmark] <<< PC[B.sub.Netherlands] < PC[B.sub.Germany] (4)

ranking (see Figure 2).

The first conclusion that can be drawn from this result is that PCB actually matters. It makes a difference whether we look at statistical measures for CB (OCB) or whether we look at the perception of the fans (PCB). This insight alone offers important implications both for science and management. Understanding the effects of CB on economic success, attendance figures, profits, etc. of leagues requires considering that fans perceive CB in a different manner than statistics measure it. And it is certainly not a far-stretched conclusion that fans' perception drives fans' behavior more than statistical measures. Obviously, this has important implications also for the management of leagues.

Based upon our study in this paper, however, we can offer a second insight. In the case of the Danish league, the decrease in OCB seems to influence perceptions in a stronger way than the level of OCB. While the Danish league indeed is on average characterized by a (slightly) better OCB than the other two in both periods (see Figure 1), the decrease of CB is much more dramatic in Denmark than in the Netherlands and in Germany. The Competitive Balance Ratio indicator for Denmark has decreased by around 36% (from 0.88 to 0.56) between the two periods, whereas for Germany it has decreased by around 20% (from 0.66 to 0.53) and the Dutch one remained almost unchanged (increase by around 5% from 0.41 to 0.43). Thus, the difference between OCB and PCB (levels) can be explained by changes of OCB being a stronger influence on fans' perception than OCB levels. If this hypothesis can be corroborated by further research, it offers additional important insight of how to analyze the CB/economic effects interface.

[FIGURE 4 OMITTED]

Finally, as a third conclusion from our analysis, we find supportive evidence for our hypothesis that the relation between CB and fans' consumption includes a discontinuity in terms of a tipping point above which changes in CB are not very relevant for fans whereas fans consumption behavior does change significantly once CB falls below that crucial threshold. In other words, changes in consumption behavior seem to be triggered by PCB falling below a crucial threshold, where the WTP for CB improvements "jumps" to a higher level.

Summing up, by employing a stated preferences approach, the paper provides empirical evidence for our suggestion, that (systematic) differences between PCB and OCB might serve as a possible explanation for the gap between the UOH and the (lack of) its empirical validation with regard to European professional football. Nevertheless, stated preference methods are based on what people say rather than what people do as pointed out by Zou and Hobbs (2006). Furthermore, the question arises whether the simplistic WTP-scenarios employed are appropriate in this research context. Therefore, the estimated absolute WTP-values in this study should be treated with caution. However, even if (due to the above mentioned methodological restrictions) the derived absolute WTP-values are biased, this bias at least should be the same bias for all countries, i.e., the detected country specific differences (relative WTP-values) should remain rather robust. Therefore, despite these shortcomings, the paper presents a promising avenue for future research into the development and application of other methods to test the UOH.

References

Bateman et al. (2002). Economic valuation with stated preference techniques: A manual. Cheltenham, UK: Edward Elgar.

Buraimo, B., & Simmons, R. (2008). Do sports fans really value uncertainty of outcome? Evidence from the English Premier League. International Journal of Sport Finance, 3(3), 146-155.

Czarnitzki, D., & Stadtmann, G. (2002). Uncertainty of outcome versus reputation: Empirical evidence for the First German Football division. Empirical Economics, 27, 101-112.

Forrest, D., & Simmons, R. (2002). Outcome uncertainty and attendance demand: The case of English soccer. Journal of the Royal Statistical Society, Series D (The Statistician), 51(2), 229241.

Forrest, D., & Simmons, R. (2006). New issues in attendance demand: The case of the English

Football League. Journal of Sports Economics, 7(3), 247-266.

Hart, R. A., Hutton, J., & Sharot, T. (1975). A statistical analysis of association football attendances. Applied Statistics, 24(1), 17-27.

Humphreys, B. R. (2002). Alternative measures of competitive balance in sports leagues. Journal of Sports Economics, 3(2), 133-148.

Koenigstorfer, J., Groeppel-Klein, A., & Kunkel, T. (2010). The attractiveness of national and international football leagues: Perspectives of fans of "star clubs" and "underdogs". European Sport Management Quarterly, 10, 127-163.

Neale, W. C. (1964). The peculiar economics of professional sports: A contribution to the theory of the firm in sporting competition and in market competition. Quarterly Journal of Economics,78(1), 1-14.

Pawlowski, T. (2013a). Wettbewerbsintensitat im ProfifuBball--Eine empirische Untersuchung zur Bedeutung fur die Zuschauer. Wiesbaden: Springer-Gabler Research (ISBN: 978-3-65800210-7).

Pawlowski, T. (2013b). Testing the uncertainty of outcome hypothesis in European professional Football--A stated preference approach (unpublished working paper).

Pawlowski, T., & Anders, C. (2012). Stadium attendance in German professional football - The (un)importance of uncertainty of outcome reconsidered. Applied Economics Letters, 19, 1553-1556.

Pawlowski, T., Breuer, C., & Hovemann, A. (2010). Top club's performance and the competitive situation in European domestic football competitions. Journal of Sports Economics, 11, 186202.

Peel, D. A., & Thomas, D. A. (1992). The demand for football: Some evidence on outcome uncertainty. Empirical Economics, 17(2), 323-331. Rottenberg, S. (1956). The baseball player's labour market. Journal of Political Economy, 64(3), 242-258.

SPORTFIVE (2009). European Football--England, France, Germany, Italy and Spain. Hamburg, Germany: SPORTFIVE.

Zou, N. N., & Hobbs, J. E. (2006). Modelling functional food choice and health care impacts: A literature review. University of Saskatchewan, Department of Agricultural Economics (Research Project Number CMD-06-02).

Authors' Note

The authors gratefully acknowledge that this research benefited from a UEFA Research Grant. The authors would also like to thank Dennis Coates, David Forrest, Arne Feddersen, and the participants of the second Conference on Football and Finance in Munster, as well as the participants of the 4th ESEA/XIV IASE Conference in London for valuable comments on earlier drafts of this paper. Additionally, the authors would like to thank Christopher Dick, Paul Downward, Mikkel Draebye, Thomas Junod, Stefan Kesenne, Stephanie Leach, Christian Muller, Katalin Monostori, and Gerco van Dalfsen for their help in conducting the research project as well as Nadine Neute for editorial assistance.

Endnotes

(1) Starting with the seminal work by Hart, Hutton, and Sharot (1975), most of the studies analyze the (potential) impact of short-term UO on stadium attendance in European professional football. Those studies predominantly found either no significant effect (e.g., Hart et al., 1975; Forrest & Simmons, 2002, 2006) or an effect not supporting the UOH (e.g., Buraimo & Simmons, 2008; Czarnitzki & Stadtmann, 2002; Peel & Thomas, 1992). For an overview on previous research please refer to Pawlowski (2013a, 2013b).

(2) In contrast, Pawlowski (2013a) is about the fans' intention to consume in the three mentioned countries.

(3) Please note, that some parts in this method section are similar to Pawlowski (2013b). However, for a comprehensive understanding of the analysis in this paper it appears to be necessary to describe these methodological issues here as well.

(4) A football fan is characterized to have a major interest in professional football competitions and consumes the product, i.e., either attends a game live in the stadium and/or watches a game live on TV.

(5) As ongoing data analysis shows, no obvious differences between both fan "types" exist with regard to the PCB measures and intention to consume (Pawlowski, 2013b). However, this has to be taken with caution since it is not entirely clear that fans watching football matches on TV live in bars represent a good proxy for real "couch potato" fans watching football at home. The latter are not directly considered in our database.

(6) Response categories are: "0[euro]", "1[euro] to 2[euro]", "3[euro] to 5[euro]", "6[euro] to 10[euro]", "11[euro] to 15[euro]", "16[euro] to 20[euro]", "21[euro]" or more per stadium ticket per game.

Tim Pawlowski [1] and Oliver Budzinski [2]

[1] University of Tuebingen

[2] Ilmenau University of Technology

Tim Pawlowski is a professor of sport economics, sport management and sport media research in the Faculty of Economics and Social Science. His research interests include the analysis of sports demand, the financing of sport systems, and the economics of league competitions.

Oliver Budzinski is a professor of economic theory in the Institute of Economics. His research interests include competition economics, media economics, and sport economics.

Table 1

Country   Date         City          Club

G         11.09.2011   Cologne       1. FC Koln
G         17.09.2011   Hamburg       Hamburger Sport-Verein (HSV)
G         17.09.2011   Leverkusen    Bayer 04 Leverkusen
G         01.10.2011   Dortmund      Borussia Dortmund (BVB)
G         16.10.2011   Cologne       1. FC Koln
G         23.10.2011   Leverkusen    Bayer 04 Leverkusen
NL        22.10.2011   Utrecht       FC Utrecht
NL        20.11.2011   Groningen     FC Groningen
NL        27.11.2011   Enschede      Twente Enschede
NL        03.12.2011   Groningen     FC Groningen
DK        06.11.2011   Copenhagen    FC Kobenhavn
DK        07.11.2011   Sonderjysk    Sonderjysk E
DK        20.11.2011   Midtjylland   FC Midtjylland
DK        27.11.2011   Sonderjysk    Sonderjysk E

Country   Performance   Game

G         CBP           1. FC Koln--1. FC Nurnberg
G         CGP / VP      HSV--Borussia Monchengladbach
G         CGP           Bayer--1. FC Koln
G         VP            BVB--Augsburg
G         CBP           1. FC Koln--Hannover 96
G         CGP           Bayer--FC Schalke 04
NL        CBP           FC Utrecht--SC Heerenveen
NL        VP            FC Groningen--VVV-Venlo
NL        CGP           Twente--Vitesse Arnheim
NL        VP            FC Groningen--NEC Nijmegen
DK        CGP           FC Kopenhagen--Lyngby BK
DK        CBP           Sonderjysk E--Aarhus
DK        VP            FC Midtjylland--Aalborg
DK        CBP           Sonderjysk E--Lyngby BK

Notes: Country: G = Germany, NL = Netherlands, DK = Denmark;
Performance: CBP constantly bad performance throughout the
seasons 2002/03-2011/12, CGP constantly good performance
throughout the seasons 2002/03-2011/12, VP volatile
performance throughout the seasons 2002/03-2011/12.

Table 2: Sample characteristics, with distributions in %. All cells
in brackets sum to 100 if missing or "other" categories, which are
omitted here, are included.

                                  German sample

Interest and consumption patterns

interest in football (high;       (88; 10; 2)
moderate; low)

matches attended in person
in the last season
(0; 1-5; 6-10; 11-21; >21)        (8; 41; 17; 18; 16)

matches watched on television
in the last season
(0; 1-5; 6-10; 11-21; >21)        (4; 12; 15; 27; 41)

Socio-demographic and economic background

gender (male; female)             (73; 23)

age (< 29; 30-49; [greater        (51; 34; 10)
than or equal to] 50 years)

family status (single; married)   (43; 27)

household size (1; 2; 3; 4;       (17; 27; 21; 15; 9)
[greater than or equal to]
5 person) monthly household
net income

(< 1,500; 1,501-2,500; >          (31; 24; 24)
2,500 Euro) work status

(employed; apprentice/student;    (58; 21; 3; 2)
pensioner; unemployed)

total number of observations      1,203

                                  Danish sample

Interest and consumption patterns

interest in football (high;       (79; 17; 4)
moderate; low)

matches attended in person
in the last season
(0; 1-5; 6-10; 11-21; >21)        (6; 23; 18; 30; 22)

matches watched on television
in the last season
(0; 1-5; 6-10; 11-21; >21)        (2; 17; 17; 23; 40)

Socio-demographic and economic background

gender (male; female)             (85; 12)

age (< 29; 30-49; [greater        (39; 43; 13)
than or equal to] 50 years)

family status (single; married)   (45; 30)

household size (1; 2; 3; 4;       (22; 23; 14; 21; 14)
[greater than or equal to]
5 person) monthly household
net income

(< 1,500; 1,501-2,500; >          (30; 20; 40)
2,500 Euro) work status

(employed; apprentice/student;    (70; 12; 4; 6)
pensioner; unemployed)

total number of observations      267

                                  Dutch Sample

Interest and consumption patterns

interest in football (high;       (82; 15; 2)
moderate; low)

matches attended in person
in the last season
(0; 1-5; 6-10; 11-21; >21)        (7; 34; 16; 24; 17)

matches watched on television
in the last season
(0; 1-5; 6-10; 11-21; >21)        (10; 26; 17; 19; 26)

Socio-demographic and economic background

gender (male; female)             (82; 16)

age (< 29; 30-49; [greater        (28; 46; 21)
than or equal to] 50 years)

family status (single; married)   (30; 45)

household size (1; 2; 3; 4;       (15; 25; 13; 22; 13)
[greater than or equal to]
5 person) monthly household
net income

(< 1,500; 1,501-2,500; >          (23; 23; 35)
2,500 Euro) work status

(employed; apprentice/student;    (66; 12; 6; 3)
pensioner; unemployed)

total number of observations      219

Figure 1: Competitive Balance Ratio of the top five clubs in the
Danish, German, and Dutch professional football leagues before
and after the turn of the millennium

Competitive Balance Ratio

         1991/92-2000/01   2001/02-2010/11

Danish   0,88              0,56
German   0,66              0,53
Dutch    0,41              0,43

Source: Pawlowski, 2013a, p. 64

Note: Table made from bar graph.