The Performance Effects of Wise and Unwise Managerial Dismissals

Abstract

The decision to dismiss a coach is challenging because poor performance tends to coincide with both bad luck and low coaching ability. We differentiate between dismissals following actual poor performance on the pitch (wise dismissals) and dismissals following seemingly poor performance due to bad luck (unwise dismissals). To categorize dismissals, we use "expected goals," which are less vulnerable to random variation in match outcomes. Using data from European football, we find that wise dismissals increase subsequent performance compared to a control group of nondismissals with similarly poor performance on the pitch. However, unwise dismissals do not improve subsequent performance compared to a control group with similar strings of bad luck. (JEL D81, J44, L83) I. INTRODUCTION Understanding how managerial change affects the performance of organizations is of critical importance. One main area of previous research is the sports industry, where the focus is on the impact of head coach changes on team performance (Giambatista, Rowe, and Riaz 2005). Coaches are ultimately responsible for team performance and thus hold official authority (Grusky 1963). Based on the considerable influence of coaches on teams through strategic, operational, and motivational channels, research suggests that coaches play crucial roles in determining team performance (Kattuman, Loch, and Kurchian 2019; Muehlheusser et al. 2018; Rowe et al. 2005). However, most empirical studies across different sports conclude that there is no positive effect of involuntary coach changes on team performance (Audas, Goddard, and Rowe 2006; Gamson and Scotch 1964). Most prominently, studies using data from European football show that while performance improves after a within-season coach dismissal, performance also improves for a control group of counterfactual dismissals that did not occur (Paola and Scoppa,

Underestimating randomness: Outcome bias in betting exchange markets

“…Many laboratory studies have found an outcome bias in different settings, e.g., legal decisions (Alicke, Davis & Pezzo, 1994), medical decisions (Baron & Hershey, 1988), investment decisions (König-Kersting, Pollmann, Potters, & Trautmann, 2021;Ratner & Herbst, 2005) and ethical judgments (Gino, Moore & Bazerman, 2009;Gino, Shu & Bazerman, 2010). Consistent with the literature of psychology and economics, recent research in sports economics has also shown that decision-makers underestimate the role of randomness in match outcomes and assign too much weight to the observed outcomes when they evaluate performance (Flepp & Franck, 2021;Gauriot & Page, 2019;Kausel, Ventura & Rodríguez, 2019;Lefgren, Platt & Price, 2015).…”

“…Consistent with insights from the laboratory, a few studies have provided the first evidence of the outcome bias in the field (Emerson et al, 2010;Tinsley, Dillon & Cronin, 2012). Recently, research in sports economics has shown that decision-makers underestimate the role of randomness in match outcomes and assign too much weight to the observed outcomes when they evaluate performance (Flepp & Franck, 2021;Gauriot & Page, 2019;Kausel et al, 2019;Lefgren et al, 2015). Using a formal model based on Bayesian updating, Lefgren et al (2015) found that coaches in the NBA change their strategy more often after losing a game than after winning a game, even when comparing narrow wins and losses.…”

“…These findings indicate that luck is overly rewarded in such performance evaluations and consequently might lead to inappropriate future strategy adjustments. For instance, Flepp and Franck (2021) showed that coach dismissals in soccer lead to a boost in subsequent team performance only when previous match outcomes were the result of actual bad performance rather than bad luck. This finding emphasizes the importance of unbiased performance and decision-making evaluations, as coach dismissals are very costly at the top level of soccer.…”

“…Following Brechot and Flepp (2020) and Flepp and Franck (2021), we plot the OLT rank against the xGT rank to show the distribution of the rank combinations for all teams in all but the first match weeks of the seasons (see Figure 1). 1 For all rank combinations above the black line in Figure 1, the LTD variable is negative, and for all rank combinations below the black line, the LTD variable is positive.…”

“…Instead of calculating the differences in expected goals and then continuously adding these differences to create a table ranking, we formed thresholds for expected goal differences to determine whether the match outcome was considered a win, draw or loss. Following Flepp and Franck (2021), we considered an expected goal difference greater than 0.5 to be a win, an expected goal difference less than 0.5 to be a loss and everything in between a draw. We then allocated points as in the OLT, i.e., 3 points for a win, 1 point for a draw and 0 points for a loss.…”