The performance of football club managers: skill or luck?

Abstract

This paper develops a performance management tool and considers its application to the football industry. Specifically, the resulting model evaluates the extent to which the performance of English Premier League football club managers can be attributed to skill or luck when measured separately from the characteristics of the team. We first use a specification that models managerial skill as a fixed effect and we then implement a bootstrapping approach to generate a simulated distribution of average points that could have taken place after the impact of the manager has been removed. The findings suggest that there are a considerable number of highly skilled managers but also several who perform below expectations. The paper proceeds to illustrate how the approach adopted could be used to determine the optimal time for a club to part company with its manager.

Acknowledgements

The authors are very grateful to Simon Burke who provided extremely useful suggestions on the design of the methodology used in this paper. We would also like to thank Graham Mackrell from the League Managers Association (LMA) for providing the data pertaining to managerial sackings/resignations and David Matthews for his assistance with data collection.

Notes

¹Association football is commonly referred to as ‘soccer’ in the USA in order to distinguish the game from American football. Throughout the paper, we use the term ‘football’ rather than ‘soccer’.

⁵However, since it would be very difficult to construct a plausible cumulative measure of net transfer spending, we do not consider this issue further.

⁶An increase in transfer expenditure of a similar figure leads to only a 0.1 fall (0.07 rise) in the expected number of points, although as we note, these parameter estimates are not statistically significant.

⁸Three of the best performing managers were from Chelsea, the club with the highest budget in the Premier League over the sample period.

⁹We should again note here that Bobby Robson managed only four games within our sample period and therefore, as we argue below, this is not really a sufficient number to effectively evaluate his performance.