National Football League head coach race, performance, retention, and dismissal

Highlights

•	Test for racial discrimination in the retention and dismissal of National Football League head coaches.
•	Statistically significant evidence that Non-White head coaches have longer employment spells in the Rooney Rule era.
•	Head coach employment tenure and dismissal are largely driven by raw performance and to a lesser degree, relative performance.
•	No statistically significant evidence that dismissal, raw performance, or relative performance differs by race.

Abstract

The authors tested for evidence of racial discrimination in the employment retention of National Football League head coaches. A robust data set spanning the modern history of the sport (1985–2018) was generated to examine managerial employment tenure, dismissal, and subsequent organizational performance. After controlling for performance differences and heterogeneity between head coaches, the authors uncover statistically significant evidence that Non-White head coaches experience longer employment spells relative to White head coaches in the Rooney Rule era. No statistically significant evidence of racial differences in the rate at which head coaches are fired is found. Both employment tenure and dismissal are largely driven by raw performance, and to a lesser degree, relative performance. Finally, the relationship between head coach race and organizational performance is examined, but no statistically significant differences by race are uncovered.

Keywords:

• Race
• Performance
• Employment retention
• Dismissal

Notes

1 These are exclusively short spells (usually one or two games) at the end of a season following an in-season HC firing.

2 We note that we use the terms “Non-White” and “minority” interchangeably moving forward.

3 Win%Tenure and ATS%Tenure are positively correlated at 0.6691. This value is at an acceptable level with respect to collinearity (Kutner, Nachtsheim, & Neter, 2004).

4 We assess the sensitivity of this approach by alternatively operationalizing functional experience using a) the number of seasons of NFL and NCAA coaching experience at hire and b) the combined number of seasons of coaching experience at both levels at hire. There are no substantive changes in the results when using these alternative measures.

5 We do not account for NCAA playing experience as only two HCs in the sample did not play at the college level.

6 Staw and Hoang (1995) demonstrated that draft position is related to player employment duration in the NBA. We attempted to include an indicator denoting whether the HC was selected in the NFL Draft as a player. However, it is collinear with YearsNFLPlayer (r=0.7420). We tested this indicator in place of YearsNFLPlayer, and the results are not substantially different.

7 These include: quarterback (QB), running back (RB), wide receiver (WR), tight end (TE), offensive line (OL), defensive line (DL), linebacker (LB), defensive back (DB) and those not playing at the college or NFL level (NoPosition).

8 We also test a ten-year measure of team success and find no substantial differences in results.

9 We also divide the sample based on the median value of spell winning percentage (.463) and the results do not change in a substantive manner.