Abstract:It is well-established that the home advantage (HA), the phenomenon that on average the local team performs better than the visiting team, exists in many sports. In response to the COVID-19 outbreak, spectators were banned from football stadiums, which we leverage as a natural experiment to examine the impact of stadium spectators on HA. Using data from the first division of the German Bundesliga for seasons 2016/17 to 2023/24, we are the first to focus on a longer time horizon and consider not only the first but all three seasons subject to spectator regulations as well as two subsequent seasons without. We confirm previous studies regarding the disappearance of the HA in the last nine matches of season 2019/20. This drop materialised almost entirely through a reduction of home goals. The HA in season 2020/21 (with spectator ban during most matches) was very close to the pre-COVID-19 season 2018/19, indicating that teams became accustomed to the absence of spectators. For season 2021/22, with varying spectator regulations, we detect a U-shaped relationship between HA and the stadium utilisation rate, where HA increases considerably for matches with medium stadium utilisation which is associated with a larger difference in running distance between the home and away teams.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to explore the impact of the absence of spectators during the COVID - 19 pandemic on the home advantage (HA) in the German Bundesliga. Specifically, using data from the 2016/17 to 2023/24 seasons, the study focuses on the following aspects:
1. **Disappearance of home advantage**: During the last nine rounds of the 2019/20 season, due to the pandemic - induced ban on spectators, the study found a significant decline in home advantage, mainly manifested as a decrease in the number of home goals.
2. **Long - term impact**: The study not only examined the impact during the first spectator ban but also analyzed the situation in the subsequent three seasons (2020/21, 2021/22, and 2022/22) to determine whether this impact was temporary or long - term.
3. **Impact of the partial return of spectators**: In the 2021/22 season, the number of spectators gradually recovered but not completely. The study found a U - shaped relationship between home advantage and stadium utilization, that is, the home advantage increased significantly at medium utilization.
4. **Mechanism exploration**: The study further explored possible mechanisms, such as team adaptation to spectator - free games, changes in player behavior (e.g., differences in running distances), etc.
### Main conclusions
- **End of the 2019/20 season**: The home advantage almost completely disappeared, mainly because the number of home goals decreased significantly.
- **2020/21 season**: Although most games were still without spectators, the home advantage recovered to a level close to that before the pandemic, indicating that the teams gradually adapted to the spectator - free game environment.
- **2021/22 season**: As the spectators gradually returned, a U - shaped relationship emerged between home advantage and stadium utilization, and the home advantage increased significantly at medium utilization.
- **2022/23 and subsequent seasons**: The home advantage has basically recovered to the pre - pandemic level.
### Research methods
The study used a multiple regression model, controlling for multiple fixed effects (such as the home team, the away team, the match day, the ranking difference, etc.), and considering other potential confounding factors (such as participation in European competitions, new coach appointments, weather conditions, etc.). Through these methods, the study ensured the robustness of the results and provided an in - depth understanding of the mechanisms of changes in home advantage.
### Formula representation
To describe the model more precisely, the following is the Markdown - format representation of the regression equation:
\[
Y_{is} = \beta_0+\sum_{k \neq 2018/19} \beta_{1,k} S_k(is)+\beta_2 S_{2019/20}(is) \times M_{26 - 34}(is)+\gamma_1 S_{2021/22}(is) \times SU(0,0.25]+\gamma_2 S_{2021/22}(is) \times SU(0.25,0.75]+\delta MD_j(is)+\alpha_{1,H(i)}+\alpha_{2,A(i)}+\delta Ranking(H - A)_{is}+\epsilon_{is}
\]
where:
- \( Y_{is} \) represents the match result variable (such as the difference between home and away goals, a dummy variable for home or away game victory, etc.)
- \( S_k(is) \) represents the season dummy variable
- \( S_{2019/20}(is) \times M_{26 - 34}(is) \) represents the interaction term for rounds 26 - 34 of the 2019/20 season
- \( SU(0,0.25] \) and \( SU(0.25,0.75] \) are dummy variables for low and medium utilization respectively in the 2021/22 season
- \( MD_j(is) \) represents the match - day fixed effect
- \( \alpha_{1,H(i)} \) and \( \alpha_{2,A(i)} \) represent the fixed effects of the home team and the away team respectively
- \( Ranking(H - A)_{is} \) represents the fixed effect of the ranking difference between the two teams
- \( \epsilon_{is} \) represents the error term