Justin Ehrlich: 0000-0001-5729-6461

Document Type



Winter 1-29-2021


home field advantage, COVID-19, gridiron, football, attendance




Econometrics | Sports Studies


While economists have previously noted that home field advantage is affected by crowd density, isolating this effect is difficult since crowd size is likely endogenous with team ability and game matchup. The COVID-19 pandemic has presented a unique natural experiment since local governments have introduced safety protocols that varies widely across the United States. These safety protocols have limited attendance in varying ways for live sporting events, including National Football League games. Given the differential (and exogenous) attendance restrictions, we were able to isolate three broad categories of attendance: 1) games without attendance restrictions (seasons 2016-2019), 2) games with limited in-person attendance, and 3) games without fans. We developed a model to predict the home team point differential, which allowed us to estimate home field advantage of these broad categories. We found that playing in partially full stadiums does not impact relative home team performance (compared to 2016-19 seasons). Conversely, we found that playing with no fans completely eliminates home field advantage. We also tested if fan density mattered in partial attendance game and found that attendance percentage does not statistically affect the home point differential. Hence, our results reveal that the presence of fans matters but the density of fans does not. Furthermore, we were able to compare these results with the ‘predictions’ offered by the betting market. We found that the betting market correctly predicted that games with partial attendance (due to social distancing) did not impact home field advantage in a statistically significant way and that home team performance would suffer in games played without attendance. Finally, we find evidence that the betting market exhibited a ‘learning effect’ since the market predicted an increasingly strong ‘no fan’ effect as the season progressed.

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.