How much do leaders matter?

How much do leaders matter? Coaches, CEOs, political leaders… it differs.

How much do CEOs matter? According to a paper co-authored by HBS’ former dean Nithin Nohria[1] in 2010 estimates that CEOs explain 14% of a company’s profitability. According to another study of Stanford’s Rock Center for Corporate Governance[2], CEOs attribute 30% of a firm’s performance to themselves, while directors are even more generous, and attribute them 40% of the results. These results certainly explain why CEOs are paid generously but are they that relevant?

In a paper released in 2019 by Christopher Berry and Anthony Fowler[3], they explore the answer to how much do coaches matter. They use a novel technique called RIFLE –randomization inference for leader effects—to calculate the effectiveness of leaders in general. In simple terms, RIFLE is a regression with a dummy for each leader –coaches in their paper. To control for serial correlation and real coach effects, they randomly permuted each coach’s tenure as a block. Then they artificially created a dataset where the data has the same features as the real one, except that the coaches do not align with the outcomes that they actually got. Then if the real dataset consistently shows a larger coach effect than the artificial one, they can conclude that there is a positive or negative coach effect.

Berry and Fowler found that coaches explain 20-30% of the team’s performance –14 wins per season in the NBA if you go from a coach that is two standard deviations below the mean to one that is two standard deviations above the mean.

When using the same approach to understand the impact that CEOs have, the answer that they found is… almost no effect. The same applies to political leaders in democracies, whereas not so much in autocracies. What does this mean? probably that the more complex the environment and the larger and more diverse the stakeholder base is, the less power the leader has. The “game” for these leaders is certainly more complex as well, making it a balancing act.

We are left with a wide range of potential leader impact on their organization’s performance: from 0% to 40%. What are the lessons that we can derive from this?

First, that it is still important to have well-prepared business and political leaders, but it is humbling –and common-sense– to acknowledge that one leader cannot radically change the direction and performance of a complex organization or territory. And second, to bring it back home with people analytics, it is critical to understand that people are much more powerful than what a model can predict. We, humans, are capable to achieve the most impressive feats, ones that have never done or even imagined before. Algorithms might underestimate individuals, but we need to remain human-centric rather than machine-centric and use data and analytics to complement our skills and ability to create and make things happen.


[1] Wasserman, Anand, Nohria (2010)

[2] CEOs and Directors on Pay, 2016 Survey on CEO Compensation

[3] Berry, C.R., & Fowler, A. (2019). How Much Do Coaches Matter?


HiQ Labs, Inc. v. LinkedIn Corp. – reflections from the 9th Circuit


Making Sense of the Virtual Environment: The Power of People Analytics for Reflecting on the Pandemic

Student comments on How much do leaders matter?

  1. Interesting post, Manuel! It is fascinating that such analyses were conducted, and even more surprising that the results were that CEOs… may not matter. I agree that this conclusion cannot be treated as a fact because of the reasons you mention, but assuming this is true, I think it might be due to a lot of different factors:
    Is it because individual CEOs have little power compared to the board? Or maybe because CEOs are not very diverse in practice (they tend to come from similar backgrounds, have similar experiences…)? Finally, how much time does it take for a CEO to see the results of his/her actions (we might risk attributing the merits of one to their successor)?

  2. Super interesting. This is a topic I think about a lot given the importance of leadership HBS constantly emphasizes, which I worry will perpetuate this aura of grandiose a lot of modern corporate leaders seem to ascribe to themselves (which of course, manifests in the increasing CEO-worker pay gap).

    That 0-40% range reminded me of the theory of leadership Gen. (Ret.) Stan McChrystal explored in his book “Leaders: Myth and Reality”. He argued that the impact leaders have are largely contextual, and outcomes will depend on the fit between a leader’s character and the situation he faces (social, organizational, etc.), hence rather than sitting at the top of a pyramid, leaders should reposition themselves as a node as part of a network.

    This finding isn’t new – basically the concept of servant leadership – but when comp plans don’t reflect that leaders are “there to help you be successful” and “part of the same team”, platitudes like those just rings hollow to me.

Leave a comment