Football Fans’ Interest in and Willingness-To-Pay for Sustainable Merchandise Products

2.1 Value-Action Gap of Pro-Environmental Consumption

The theoretical framework of this study follows Blake’s (1999) approach of an environmental value-action gap. The environmental value-action gap illustrates individuals’ discrepancy between their concerns and knowledge about environmental problems and the subsequent lack of PEB or intentions of PEB. Several factors play a role here: On the positive side, environmental concerns and environmental knowledge are considered potential boosters for PEB intentions. On the negative side, the lack of PEB intentions is caused by three types of barriers: individuality, responsibility, and practicality.

2.2 Environmental Concerns and Environmental Knowledge as Contributors to PEC Intentions

Environmental concerns (EC) include individuals’ awareness of environmental problems, a supportive attitude towards PEB, and a willingness to solve environmental problems (Dunlap and Jones 2002). EC were found to affect PEB intentions significantly positively (Maichum, Parichatnon, and Peng 2016). Especially for the purchase of sustainably produced clothes, Rausch and Kopplin (2021) find a significant positive effect of EC on PEC intentions. Likewise, in another study of marathon runners, EC was positively associated with runners’ intentions to donate towards green initiatives (Triantafyllidis and Kaplanidou 2021). Among sport fans, women football fans were significantly more likely to eat vegetarian or vegan in everyday life than men fans. In the football stadium, women football fans were significantly more interested in the offer of vegan burgers or sausages than men irrespective of the EC level (Scharfenkamp and Wicker 2024). For international trail runners participating in a marathon in Greece, empirical evidence shows that environmental consciousness had a positive effect on WTP for voluntary carbon offsetting (Triantafyllidis and Kaplanidou 2018). Furthermore, the general attitude towards the environment and the recognition of environmental problems were positively associated with the WTP for offsetting carbon emissions (Lintumäki et al. 2023). Based on this evidence, we derive the first hypothesis:

Another precondition of PEC intentions is environmental knowledge (EK): Individuals are knowledgeable about the environment, if they are familiar with environmental concepts and understand the functioning of ecosystems as well as the interdependencies between their own behavior and the natural environment (Fryxell and Lo 2003). EK might lead to environmental attitudes (Ko and Jin 2017; Kollmuss and Agyeman 2002) and ultimately to PEB intentions (Kollmuss and Agyeman 2002). In the context of this study, football fans are assumed to have EK if they understand the consequences of their purchasing behavior of merchandise products on the ecosystem (Goh and Balaji 2016). Research examining the effects of EK on PEC intentions yielded mixed evidence: While some research indicated that EK neither influences the attitude towards sustainable products nor the intention to purchase such products (Jaiswal and Kant 2018), the majority of previous studies confirmed that EK is an essential precondition for PEB intentions (Chan 2001; Ko and Jin 2017; Kumar, Manrai, and Manrai 2017; Maichum, Parichatnon, and Peng 2016; Mostafa 2007; Wang, Qian, and Yu 2014; Yadav and Pathak 2016). Based on these findings, we derive the second hypothesis:

2.3 Barriers of PEC Intentions

Barriers like individuality, responsibility, or practicality might hinder individual intentions for PEC. The barrier of individuality considers cognitive structures or attitudes that hinder individuals to translate their positive environmental attitudes into PEC behavior or at least into PEC intentions (Blake 1999). Individuality in the sport fan context is reflected by past purchase behavior of merchandise products, awareness of sustainable brands, and interest in purchasing sustainable products.

Past purchase behavior is known to be connected with individual attitudes towards PEB (e.g. Thormann and Wicker 2021, 2024). While Kim and James (2016) find that past purchase behavior of merchandise products was significantly and positively associated with future intentions to consume merchandise products, there is no evidence on this link for sustainable clothes, yet. Similar findings are evident for sustainable hotels (Han and Kim 2010). Drawing from these findings, we suggest the following third hypothesis:

A lacking awareness of the sustainable impact of clothing and sustainable labels might be another individual barrier hindering individuals to pursue PEC intentions. Previous research indicated that individuals in general had a limited awareness of the environmental impact of clothing (Goworek et al. 2012) and corresponding labels. Harris, Roby, and Dibb (2016) identified three reasons for this low level of awareness: First, clothing is not an altruistic purchase, suggesting that sustainability of clothes was evaluated very low in consumers’ purchasing decision criteria. Second, the sustainability of clothes is too complex, meaning that consumers lack knowledge about mechanisms in the textile industry and e.g. the link between prices and the payment of cotton farmers. The third reason is the high diversity of ethical concerns among consumers, meaning that while some consumers might be concerned about animal welfare, some prioritize that their clothes were produced child-labor free. Due to the generally low level of awareness about the sustainable impact of clothing, the fashion industry has developed specific labels, like e.g. the green button, that reflect the extent to which items were produced in an environmentally and socially sustainable manner. Therefore, we assume that the awareness of sustainable labels might reduce the lack of knowledge about at least some mechanisms in the textile industry towards more sustainability, thus supporting PEC intentions. The resulting hypothesis is:

A particular interest in sustainably produced clothes is a necessary precondition for the purchase of these goods. Specifically, green attitudes were significantly positively related to purchase intentions of sustainable products (Jaiswal and Kant 2018). The interest in sustainable products goes beyond that of awareness of sustainable labels/products, as it includes that consumers do not only have environmental concerns, but also consider the particular ecological consequences of their purchase behavior (Jaiswal and Kant 2018). A number of studies have shown a positive link between the consumer’s interest and willingness to buy sustainable products (Akehurst, Afonso, and Gonçalves 2012; Chan 2001; Dagher and Itani 2014). Therefore, interest in sustainable products might consequently boost individuals’ interest in and WTP for sustainable merchandise clothes:

Responsibility is a second barrier that hinders individuals with environmental concerns or knowledge to pursue their pro-environmental intentions, because “they do not feel that they (as individuals) should take the responsibility for helping to solve environmental problems” (Blake 1999, 266). In the context of sport fans, the barrier of responsibility can be captured by the fans’ lacking identification or psychological link with the football team (Greenwood, Kanters, and Casper 2006). Individual commitment and a certain emotional attachment to the football club are indicative of fan identification (Branscombe and Wann 1991). In the strongest form, fans perceive themselves to be part of the football team (Wann and Branscombe 1992). It is possible that the football club’s aim to comply with sustainability initiatives might be considered a mutual goal by football fans. Therefore, a high team identification of football fans might be leveraged by the club to reach this goal (McCullough and Cunningham 2011; McCullough and Kellison 2016) as individuals categorize themselves to be part of specific groups as part of their social identity (Tajfel and Turner 1986). Regarding PEC intentions, Kwon, Galen, and Jeffrey (2007) outline the relevance of team identification for sport fans as it explains 42.6 % of the variance in purchase intentions of merchandise products. In this study, a lacking identification with the football team (Brown 1998) is assumed to be related to a low feeling of belonging and no/low perceived need to contribute to the club’s common goals (Wann and Grieve 2005). The corresponding hypothesis is:

A third barrier is practicality, implying that sport fans lack sufficient income to translate their EC and EK into PEC intentions. Based on the low-cost hypothesis of Diekmann and Preisendörfer (2003), individuals only formulate behavioral intentions in situations where perceived costs in terms of financial costs, time, convenience, and effort are low (Wicker 2018). An example for such a successful translation is the reduction of one’s carbon footprint with a simultaneous financial cost reduction by using public transport instead of using the private and more convenient car (Becken 2004; Wicker 2019). The purchase of sustainably produced merchandise products is likely associated with higher prices because of higher costs for raw material and labor. Conversely to previous literature claiming that consumers are willing to pay a price premium for sustainable clothes (Hamilton and Zilberman 2006), Tey, Brindal, and Dibba (2018) showed that both an additional value and an appealing esthetic is necessary for sustainable clothes being purchased. Therefore, football fans with lower income are less likely to purchase sustainably produced merchandise clothes, as reflected in the last hypothesis:

3.1 Data Collection

Quantitative data were collected using an online survey of football fans and spectators of Arminia Bielefeld from April to June 2022. The link to the survey was provided on the club’s social media platforms, in newsletters, and on the website of the collaborating university. Since the survey was anonymous, it is not possible to report the channel through which respondents were recruited. Overall, 1,021 respondents completed the survey and 1,019 answers could be used for the empirical analysis. The collected respondents’ characteristics and structure were discussed with club officials of Arminia Bielefeld to ensure that the sample is comparable to the perceived fan base of the club.

3.2 Questionnaire and Variables

The survey included a number of questions measuring past purchase of merchandise products, environmental concern (Kilbourne and Pickett 2008; Paul, Modi, and Patel 2016), environmental knowledge (Ko and Jin 2017), awareness of sustainable labels (Henninger 2015), team identification (Wann and Branscombe 1993), and respondents’ sociodemographic characteristics (gender, age, education, income). The employed scales have acceptable reliability, with Cronbach’s alphas of 0.893 (environmental concern), 0.773 (environmental knowledge), and 0.931 (team identification). Mean indexes were computed for these three constructs (Table 2).

WTP was assessed using the contingent valuation method (Orlowski and Wicker 2019a), which is an established method and was used in previous studies assessing football fans’ WTP (Frick and Wicker 2018; Wicker et al. 2016). At the heart of the survey was a CVM scenario, with respondents randomly receiving either the version with a t-shirt or a hoodie. The scenario started with informing respondents about societal expectations towards sport organizations to become more socially and environmentally sustainable and the implementation of sustainability aspects in the DFL’s licensing criteria from the 2023/24 season onwards, forcing clubs’ business processes to become more sustainable. Then it was assumed that given these developments, Arminia Bielefeld wants to become more socially and environmentally sustainable and that sustainably produced products will be offered in the fan store in addition to conventionally produced ones. It was supposed that respondents want to purchase a shirt/hoodie (current price €25/€60), with a socially and environmentally sustainably produced shirt/hoodie being offered in addition to the conventional product. The sustainably produced shirt/hoodie would look similar and be of the same quality as the regular shirt/hoodie. Then respondents were asked for their level of interest in this sustainably produced shirt/hoodie on a five-point scale. This variable represents the first outcome of interest, which is PEC intentions.

The second and third measure of PEC intentions stem from the assessment of WTP. After the interest question, respondents were asked for their likelihood of purchasing the shirt/hoodie at different prices. Thus, WTP was assessed using a payment card format (Orlowski and Wicker 2019a), which was selected to reduce hypothetical bias (Whitehead and Wicker 2019). The answers were converted into a continuous variable, with respondents clicking on 5 (very likely) receiving the respective Euro value and respondents not clicking on five for any price being assigned a WTP = €0. Based on respondents’ answers, two WTP variables were calculated, a WTP dummy and a logged WTP variable reflecting the amount of WTP of those respondents having a positive WTP (i.e. WTP > 0).

3.3 Empirical Analysis Strategy

The empirical analysis starts by providing descriptive statistics to give an overview of the sample structure and respondents’ WTP. In a second step, statistical significant differences based on chi²- or t-tests between the mean values for the subsamples of respondents for either the sustainably produced t-shirt or hoodie are reported. The same comparison is conducted for the subsamples of respondents with a WTP = 0 and a WTP > 0.

Third, two sets of regression models for the subsamples of the t-shirt versus hoodie scenario were estimated. The first estimation is a linear regression model explaining the interest in a sustainably produced merchandise product. The second is a logistic regression that estimates the likelihood of respondents stating a positive WTP. The third is a log-linear regression model estimating the correlates of the logged amount of WTP of those respondents with a positive WTP. This approach to first examine the extensive margin with a logistic regression to find out which respondents have a positive WTP for sustainable clothes, followed by an analysis of the intensive margin (here: amount of WTP for those with a positive WTP, hence excluding zeros) is a common procedure for datasets including a high percentage of zeros (Humphreys, Lee, and Soebbing 2010, 2011). Therefore, this two-step approach has been frequently applied in research examining the determinants of football fans’ WTP (Frick and Wicker 2018; Wicker et al. 2016).

The independent variables include environmental concern, environmental knowledge, team identification, and respondents’ sociodemographic characteristics. In the last two model specifications, the interest in a sustainable merchandise product is added as an explanatory variable. To avoid distortions due to multicollinearity, a correlation analysis for all independent variables has been conducted. As all bivariate correlations were below the critical threshold of 0.8 (Hair et al. 2013), no distortions are expected. Nevertheless, for all sets of estimations, we run separate models using environmental concerns, environmental knowledge, or awareness of sustainable labels, recognizing that these variables are not statistically highly correlated, but still conceptually related. Since heteroscedasticity is likely in large cross-sectional data, heteroscedasticity-consistent standard errors were included in each regression model. The predictive capacity of each regression model is either provided by the Pseudo-R ² for logistic regressions or the R ² for (log-)linear regressions.

With 43 % of respondents stating a zero WTP, the present sample is characterized by a high share of zeros, which might represent an issue in the empirical analysis. Specifically, these zeros might not only be composed of genuine zeros, but also include protest zeros (Frick and Wicker 2018). To address this issue, we provide supplementary analyses on the full sample including the zero responses and estimate zero-inflated Poisson regressions (Eisenberg et al. 2015; Lambert 1992; Muñiz, Rodríguez, and Suárez 2014; Otto, Pawlowski, and Utz 2021) – similar to previous WTP research in sport (Lintumäki et al. 2023). The predictive capacity of these models is provided by the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC).

4.1 Descriptive Statistics

Table 1 shows the descriptive statistics of all variables. The detailed items of mean indexes like EC, EK, sustainable labels, and team identification are displayed separately in Table 2. Table 3 reports the mean values for the subsamples of respondents for either the sustainably produced t-shirt or hoodie, including the corresponding chi²- or t-tests. On average, respondents were interested in purchasing a sustainably produced merchandise product (4.07) (Table 1). Interest in the sustainable shirt is significantly higher for t-shirts with 4.16 on average compared to 3.97 for the hoodie (Table 3). The WTP for a sustainably produced t-shirt is significantly lower (€19.42) than for a hoodie (€32.28) (Table 3). Overall, 57.1 % of respondents stated a positive WTP. Average WTP of those respondents with a positive WTP is €29.86 for a sustainable shirt and significantly lower than for a sustainable hoodie (€65.50) (Table 3).

With respect to the explanatory variables, respondents stated to be more environmentally concerned (4.254) than to have environmental knowledge (3.574) and to be partly interested in sustainable labels (3.485). The identification with Arminia Bielefeld among respondents is rather high with a mean of 4.123. The majority of respondents (81.9 %) purchased merchandise products during the last 2 years (Table 1). No significant differences between respondents of the t-shirt or hoodie scenario are evident (Table 3).

Turning to sociodemographic characteristics, nearly one third of respondents are female (29.5 %). The average age is 29.2 years and average income is €1,664. Concerning educational level, 26.7 % of respondents have lower education (below A-levels), 42.5 % have A-levels, and 30.8 % a university degree. The mean values for environmental concern are 4.25, 3.57 for environmental knowledge, and 4.12 for team identification. According to the on average high team identification, 88.42 % of respondents stated to be a fan of Arminia Bielefeld. Most respondents (81.9 %) have purchased merchandise products of the club in the last 2 years. Approximately 92 % of respondents answered the survey in season until May 14th, the residual questionnaires were answered offseason until June 2022. Half of the sample received the shirt and the hoodie scenario (Table 1). No significant differences in sociodemographics between respondents of the t-shirt or hoodie scenario are evident (Table 3).

Table 4 compares the two subsamples of respondents with a WTP = 0 and a WTP > 0. The average WTP among those who stated a positive WTP for sustainable clothing is €45.28. Respondents answering that they have no WTP for a sustainable hoodie or t-shirt have a significantly lower interest in sustainable merchandising clothes, lower EC, lower EK, higher past consumption, and a higher probability for a lower education compared to respondents with a positive WTP for sustainable merchandise clothing (Table 4). Based on these findings, it is possible that protest statements are included among those respondents with zero WTP. The finding regarding education is in line with Lübke (2022) documenting that climate change deniers are less educated.

4.2 Interest in and WTP for Sustainably Produced Merchandise Products

Table 5 displays the results of linear regressions. Regarding the interest in a sustainably produced merchandise products, the results reveal that a higher EC, higher EK, and a higher awareness of sustainable labels are significantly associated with a higher interest in a sustainable t-shirt or hoodie (models 1–6). Therefore, hypotheses H1 and H2 are supported. While a higher team identification is only weakly significant and positively associated with a higher interest in a sustainable t-shirt (model 1), there is no significant effect on the interest in a sustainable hoodie (models 2–5). However, past consumption of the club’s merchandise products is significantly positively associated with the interest in both sustainable products (models 1–6). Therefore, hypothesis H3 is supported.

The results of logistic regressions explaining the probability of respondents to have a WTP higher than zero are presented in Table 6. The results reveal that EC only have a significant positive effect on the likelihood that respondents have a WTP higher than zero for a sustainable t-shirt (model 1). Thus, hypothesis H1 is only partly supported. The probability for a WTP higher than zero for a sustainable hoodie is only significantly positively affected by EK (model 5). Consequently, hypothesis H2 is only partly supported. Past consumption of the club’s merchandise products has a significant positive impact on the probability of a positive WTP for a sustainable t-shirt (models 1–3). The interest in sustainably produced merchandise products significantly positively affects the likelihood of a positive WTP in both subsamples (models 1–6). Hence, hypothesis H5 is supported. For sustainable hoodies, being a university graduate and income are significantly positively related to the probability of a positive WTP. Consequently, hypothesis H7 stating a positive link between income and PEC intentions, is partly supported. Conversely, age is significantly negatively related to the probability of a positive WTP in the hoodie subsample (models 4–6).

The results of log-linear models for the subsample of respondents with a positive WTP are presented in Table 7. We find that EC are significantly positively associated with the WTP for a sustainable hoodie or t-shirt (models 1–6), so that hypothesis H1 is supported. Conversely, EK is only significantly positively related to the WTP for a sustainable t-shirt, thus partly supporting hypothesis H2 (model 2). A higher awareness of sustainable labels is significantly positively related to the amount of WTP for a sustainable t-shirt (model 3). Referring to our hypotheses, we learn that H4 is supported. A higher interest in sustainable products is significantly positively associated with the amount of stated WTP for a sustainable hoodie or t-shirt (models 1–6). Hypothesis H5 is supported. However, team identification is significantly negatively associated with the amount of WTP for a sustainable t-shirt in this subsample (model 3), so that hypothesis H6 is not supported. As income has no significant effect on neither the interest, nor the probability for a positive WTP, nor the exact WTP, hypothesis H7 is not supported.

The results of the supplementary zero-inflated Poisson regressions are presented in Table 8 for sustainable t-shirts and sustainable hoodies. Like in the main estimations, the results reveal that higher EC are significantly positively associated with the WTP of those respondents who potentially have a WTP larger than zero (models 1 & 4), thus supporting hypothesis H1. While a higher EK is only significantly related to the WTP for a sustainable hoodie in the main estimations, a higher EK is also significantly related to the potential WTP for a sustainable t-shirt (models 2 & 5), thus supporting hypothesis H2. Looking at hypothesis H3 and a potential positive link between past purchase behavior and PEC intentions, the estimations provide no supporting evidence. With respect to the awareness of sustainable labels, a significant positive correlation to the potential WTP of a sustainable t-shirt partly supports hypothesis H4 (model 3). Moreover, strong supportive evidence is found for hypothesis H5 that the interest in purchasing a merchandise t-shirt or hoodie is significantly associated with PEC intentions in all models. Like in the main estimations, there is no supportive evidence for hypothesis H6 as team identification is significantly negatively associated with potential WTP of a sustainable t-shirt. Finally, there is no robust evidence on a significant association between respondents’ income and their PEC intentions, so that hypothesis H7 is not supported either.