In a business that lives by the rule of “What have you done for me lately?” standing by while a new coach – or a CEO – piles up losses in the short term may be necessary to produce winning ways in the long run. However, as these authors point out, it’s not so much if you should make a change as when you should make it.

Whether individual leaders are directly responsible for organizational performance is a question that dominates strategy research. Franco Bernabe, the CEO of ENI, a large, Italian energy company, once said that “…I realized that leaders could make a difference. They could transform situations that seemed impossible.” (Linda Hill & Suzy Wetlaufer, “Leadership When There Is No One To Ask: An Interview with ENI’s Franco Bernabe,” Harvard Business Review, July-August, 1998). Bernabe, for one, believes that leaders can affect organizational outcomes. Strategy researchers such as Sayan Chatterjee, Michael Lubatkin and William Schulze support Bernabe. They argue that: “Our field’s [strategic management] theory, research and pedagogy are based on the intuition that management matters: Firms, through calculated actions, can protect their earnings from market forces in ways that are valuable to investors.” (“Toward a strategic theory of risk pr emium: Moving beyond CAPM, ” Ac ademy of Management Review, Pp. 556-568, 24(3), 1999). On the other hand, research on leadership succession suggests the opposite. Studies from Major League Baseball, the National Basketball Association and the National Football League suggest that organizational leaders (the field manager in baseball, the head coaches in football and basketball) have little, if any, impact on a team’s performance. These conclusions are drawn from studies that found that a change in leaders between seasons had no impact on team performance. However, they are inconsistent with the widely held belief that individual leaders do, in fact, affect organizational performance.

Interestingly, studies from English professional soccer have found that, on average, teams that went through a mid-season management change did not perform as well as teams that did not change managers during the season. It would seem that a new manager has little time to have an impact on a poorly performing, demoralized team in the middle of a season. Between-season changes, on the other hand, give new managers (such as coaches in hockey) an opportunity to use training camp, pre-season conditioning, player selection and repositioning to influence the team’s structure and level of play. Yet, as already mentioned, the average performance of teams with between-season changes is no different than the average performance of teams without them.

So, why don’t teams realize the expected benefits of between-season succession? One explanation, we suggest, is that between-season appointees have neither the opportunity to observe and assess the team’s strengths and weaknesses under competitive conditions, nor the information for building on the resources at hand. On the other hand, coaches appointed the previous season have had the opportunity to assess the talent of each player, and can then work more effectively with the general manager between seasons to make the appropriate draft choices and trades, and to improve the integration of the new players. As a result, teams that changed coaches during the previous season should perform better, on average, in subsequent seasons than teams that did not make changes. It is a theory, we believe, that can be applied to the world of business and organizational performance. To test this theory, we used a model in which we predicted the positive, neutral and negative effects on National Hockey League (NHL) teams with previous-season, between-season and within-season successions of general managers and coaches.

The data

There are several reasons that we chose to use a professional sports organization, the NHL, to study the effect of leaders on organizational performance. First, the performance metric is very clear: points obtained divided by points available. Second, sports teams are organizations worthy of study in their own right. Recent articles in Forbes magazine reported that the value of NHL franchises after the 2001/2002 season ranged from $286 million (U.S.) for the Detroit Red Wings to $86 million (U.S.) for the Edmonton Oilers (see Table 1). Finally, up to this point, only a very few researchers have used the NHL to study leadership succession.

Our main source of NHL data was Total Hockey: The Official Encyclopedia of the National Hockey League (2nd Edition, 2000), and the annual NHL Official Guide & Record Books, (both published by Dan Diamond and Associates, Inc.) Our data covered the years 1942-43 to 2001-02. We started with the ’42-’43 season because it is considered the beginning of the modern era in hockey. This is the year that the NHL was reduced to “The Original Six.” Prior to this season the NHL had grown from a three-team league (1917-1918) to a 10-team league (1926-27), before settling into a six-team league when the Brooklyn Americans folded after the 1941-42 season. We included all teams that played in the NHL during that 60-year period, with the exception of the Oakland/California Seals, which became the Cleveland Barons after the 1975-76 season, and merged with the Minnesota North Stars after the 1977-78 season. This is the only team that was dropped from the NHL during the 60-year period.

In order to assess the impact of succession on team performance, we eliminated alternative explanations for team performance by incorporating several variables for each of the 30 teams in our analysis. These variables included the previous season’s performance, level of rivalry in the division within which each team played, number of all-stars on each team as chosen by the Professional Hockey Writers Association at the end of each season, teamwork (number of assists divided by number of goals), and player turnover from the previous season. In addition, we included control variables for both the coach’s ability (measured as the win/loss/tie record throughout the coach’s career) and coach tenure (number of games coached with his current team) at the beginning of each season, as well as any league reorganizations prior to the current season. For each team for each season, we also noted whether the coach and general manager were, or were not, the same person. Furthermore, we eliminated the impact of time because it can skew results. We included our final control variable-the impact of the team itself-to address the argument that certain teams, because of their legacy, could affect the results. We wanted to make sure that someone could not say, for example: “but the Montreal Canadiens are the Montreal Canadiens.” Finally, we ascertained whether there had been a coaching or general-manager change in the previous season, between seasons and/or during the current season (see Table 2).

By including all 30 teams in the NHL, we obtained 843 team-season observations. During the 60 years covered in our study, there were a total of 238 coaches and 120 general managers. This suggests that the average tenure for coaches was 3.54 seasons, while the average tenure for general managers was 7.0 seasons. The latter ranged from a partial season for Roger Crozier with the Washington Capitals to 28 seasons for Harry Sinden with the Boston Bruins.

What did we find?

As we expected, most of the control variables contributed to some aspect of team performance. The team variable was a factor because clubs such as Anaheim, Chicago, Columbus, Los Angeles, Pittsburgh, Tampa Bay and Vancouver performed significantly worse than other teams. Previous season’s performance positively and significantly affected current season’s performance in that better teams in the previous season did well in the current season, and poorly performing teams in the previous season performed poorly in the current season. In addition, league reorganization, number of all-stars per team and a coach’s ability had a positive effect on performance; player turnover had a negative effect. Four variables had no impact on performance: whether or not the general manager and coach were the same person, divisional rivalry, teamwork and coach tenure.

Control variables aside, it is interesting to note that in the case of both coaches and general managers, a team’s performance was related to the particular pattern of succession, as we had predicted. First, teams that changed coaches or general-managers during the previous season performed significantly better than teams that did not. Second, teams that changed coaches or general-managers between seasons performed the same as teams that did not. Finally, teams that changed coaches or general-managers during the current season performed significantly worse than teams that did not.

What are the implications for organizations?

Overall, our findings support the contention that succession does affect performance. However, the pattern of results from our study indicates that there may be a more important question: When is the best time for the succession to occur? We discovered that only changes made during previous seasons had a positive effect on performance; changes made between seasons had no effect. Within-season changes had a negative effect.

Our findings have several implications for succession planning in sports organizations. While teams may find it appealing to make managerial changes between seasons, our results imply that such a strategy diminishes performance in the next season. Furthermore, a successor who is brought in during the current season and does not immediately achieve positive results should not be released prior to the next season. The successor needs to become familiar with the team’s strengths and weaknesses under competitive conditions, and the opportunity to use this knowledge to improve the team during the between-season period.

Another implication of our findings concerns the internal barriers that may discourage or prevent midseason managerial changes, particularly because successions are associated with declining performance. For example, might the interests of some individuals’ conflict with his obligation to make such a change? For example, consider the impact that a general manager’s own contract may have on his decision to make a coaching change. General managers with at least one season left in their contracts may be more willing to make a mid-season coaching change that results in short-term loss for long-term gain, thereby increasing their bargaining power. However, given the pressures to produce positive results in the short term, general managers whose contracts expire at the end of the current season may not risk worsening their record just when they are looking for a new contract.

Though our study only addressed succession in the context of NHL teams, there are clear implications for other business organizations. Although most businesses do not have a “between-season,” we believe they can still maximize their long-term benefits by taking into account the timing of successions. More specifically, they can realize long-term gains if they give new managers the time to assess the organization’s resources firsthand, and to apply that organization-specific knowledge when making appropriate and effective changes. On September 27, 1989, General Dynamics announced that William Anders would succeed Stanley Pace (who hand-picked Anders for the job) as chair and CEO on Jan. 1, 1991. He would join the company as vice-chair on Jan. 1, 1990. As Kevin Murphy and Jay Dial described in their Harvard Business School case “General Dynamics: Compensation and Strategy” (Dec. 3, 1997): “Anders spent his first year with GD undertaking a comprehensive assessment of the company’s strategy, its operations and markets, and its financial structure. He discovered that it was no longer ‘business as usual,’ and quickly concluded that GD was heading towards serious financial trouble.”

Anders concluded that GD was too diversified, and over the next two-and-a-half years, he divested all non-defence divisions and the defence divisions responsible for military aircraft, space systems, missiles and electronics. GD was left with nuclear submarines and tanks. Its share price, which had dipped to a low of $25.25 (U.S.) in December 1990, from a high of $79 in February, 1987, rebounded to $92 by December, 1993. Its market cap, which had dropped from $2.04 billion in 1987 to $1.05 billion in 1990, increased to $2.9 billion by 1993.

Both the GD case study and our NHL research suggest that long-term performance should improve when new managers have the opportunity to experience how things work in the organization before making changes. Otherwise, a successor may make changes that do not use the organization’s resources effectively. At the same time, boards of directors and owners may have to accept short-term losses in order to benefit in the long term.

Our results also imply that the timing of succession changes is important. If an organization waits until it is in a steep decline before making a managerial change, there may not be sufficient time for it to benefit from the succession, especially since performance will worsen at first. Furthermore, delaying a managerial change for a less disruptive time may not benefit the firm in the long run. In cyclical businesses, for example, a succession during the quieter part of the operating cycle may not provide the new manager with the opportunity to assess how the organization responds when resources are being taxed. Consequently, it may take longer for the potential long-term gains from a succession to be realized, or worse, they may not be realized at all.

Lastly, our findings have implications for managerial changes at lower levels, particularly in departments where employees work on teams that perform interdependent tasks. After a departmental managerial change, senior managers may have to tolerate short-term results while the new manager gains firsthand experience of how the team functions under pressure. Senior managers should then facilitate the opportunity for a “between-season” period in which the successor can work through the appropriate changes with the departmental team. In addition, senior managers should consider whether the contracts of lower-level managers correspond to long-term performance gains and not simply to short-term results.

So, does changing leaders affect the organization’s performance? We believe our results suggest that it does. However, our results also suggest that organizations need to consider two important points if they are to maximize long-term gains: First, question and assess the timing of the change. Second, as the old adage says, understand that there can be no long-term gain without short-term pain.