Author(s): ARRIGO, M., GOMEZ-GONZALEZ, C., LANG, M., Institution: UNIVERSITY OF LAUSANNE, Country: SWITZERLAND, Abstract-ID: 1664

Despite growing awareness, racism in sport remains an issue. While overt discrimination is visible, identifying subtler forms—such as distinguishing taste-based from statistical discrimination (Becker, 1971), especially when biases are implicit—remains challenging. To address this, prior research has studied fan behavior using game-based markets such as baseball cards and, more recently, Fantasy Football (e.g., Bryson & Chevalier, 2015).
Fantasy Football is an online competitive game in which participants have a virtual budget to assemble a squad of players and earn points based on those players’ real-life performance. This setting allows us to observe hiring decisions where past performance information is fully transparent. We used a unique dataset of over 3 million observations from the Italian Fantasy Football (Fantacalcio) across three seasons. First, we estimated Ordinary Least Squares (OLS) models to test whether Black players were underpaid relative to non-Black players with comparable performance. Second, we exploited a distinctive feature of Fantacalcio: unlike English Fantasy Football (Bryson & Chevalier, 2015), where player prices are platform-set, teams are formed through auctions within private leagues of 6–12 participants. We reshuffled players’ skin-tone labels within each position and league and ran Monte Carlo simulations to generate the expected dispersion in the share of Black players across teams under random allocation. If participants were indifferent to skin tone, observed dispersion would match this benchmark. Instead, higher dispersion would indicate that some teams systematically selected more Black players and others fewer, revealing heterogeneity in preferences and suggesting that skin tone influenced performance evaluations.
We found no difference in auction prices between Black and non-Black players conditional on performance. However, allocation across teams showed substantial excess dispersion relative to random assignment. On average, observed dispersion exceeded its benchmark by about 19%, consistently across seasons. Placebo tests using irrelevant attributes (e.g., odd vs. even day of birth) confirmed that the pattern was not mechanically driven.
These results suggest that competition may correct price discrimination when sufficient non-discriminatory participants are present (Becker, 1971). Nevertheless, allocation patterns revealed persistent racial preferences among a subset of fans, even in highly competitive and transparent environments. The findings provide insights into individuals’ racial biases and performance evaluation in sport and beyond.

Becker, G. S. (1971). The Economics of Discrimination (2nd ed.). University of Chicago Press.
Bryson, A., & Chevalier, A. (2015). Is there a taste for racial discrimination amongst employers? Labour Economics, 34, 51–63. https://doi.org/10.1016/j.labeco.2015.03.002