Television audience demand for football: disaggregation by gender, age and socio-economic status

The first journal paper to report a modelling exercise on the size of television audiences for football (Forrest et al., 2005) stimulated a large successor literature: 29 such papers on football (as well as several more on other sports) were identified in a review by Van Reeth (2020), who noted that most had been published only since 2015. Typically they were motivated in part by the increasing dependence of elite leagues on revenue derived from the sale of television rights. In the latest year for which data were available at the time of writing, the proportion of clubs' total revenue accounted for by broadcasting in Europe's 'big five' leagues ranged from 39 per cent (Germany) to 59 per cent (England) (Deloitte, 2019). When these figures are compared with the proportions earned from matchday income (11-17 per cent across the 'big five'), the shift of emphasis in the academic literature from studying stadium demand to studying television demand is clearly validated: it is of obvious importance to the sport to understand what appeals to the audience in the market that supplies the bulk of revenue. While two studies (Alavy et al., 2010; Buraimo et al., 2020) have broadened the scope of the relevant literature by attempting to model how audience size varies within a match as events on the field unfold, most contributions follow the precedent of Forrest et al. (2005) by regressing the TV audience figure for the whole match on covariates describing game characteristics, typically including measures of the quality of the two teams, ex ante outcome uncertainty and match significance, as well as controls for time of day, week and season, and, sometimes, club fixed effects.