Christopher Jeffords and Todd Potts presented a working paper entitled “Share the Love: Producing Wins in the NFL with a Cobb-Douglas Technology” at the annual IUP Sports Business Conference on April 12, 2014.
This paper assesses the degree to which average team salary cap shares devoted to offense, defense, special teams, and their component positions in the National Football League (NFL) are consistent with the optimal shares predicted by a simple constrained optimization model utilizing Cobb-Douglas production technology. Estimates are obtained using a strongly balanced panel with team and year fixed-effects for all 32 NFL teams over a 10-year time horizon, including 2002–2009 and 2012–2013.
The results indicate that the observed relative salary cap allocation between offense and defense is roughly consistent with that predicted by the model, but, relative to the model, teams devote too much salary cap to special teams and do not devote enough of the cap to players currently on the roster. Additionally, the model’s predicted optimal shares closely match the observed defensive and offensive positional cap allocations, but the predicted optimal shares are more accurate for the defensive allocations. The model over-predicts observed spending on all defensive and offensive positions, with the exception of under-predicting observed spending on the tight-end position.
Using a measure of absolute deviation from the aggregate predicted optimal shares, the results indicate that being consistent with the constrained optimization problem is correlated with higher winning percentages.
Department of Economics