How much do CEOs really matter? Reaffirming that the CEO effect is mostly due to chance

• Author: Markus A. Fitza
• DOI: http://doi.org/10.1002/smj.2597
• Date: 01 March 2017
• Published date: 01 March 2017

Strategic Management Journal

Strat. Mgmt. J.,38: 802–811 (2017)

Published online EarlyView 2 December 2016 in WileyOnline Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2597

Received 27 July 2016;Final revisionreceived 13 September 2016

COMMENTARY

HOW MUCH DO CEOS REALLY MATTER?

REAFFIRMING THAT THE CEO EFFECT IS MOSTLY

DUE TO CHANCE

MARKUS A. FITZA1,2*†

1Department of Management, Mays Business School, Texas A&M University,

College Station, Texas, U.S.A.

2Frankfurt School of Finance & Management, Frankfurt, Germany

How much of the variance in rm performance can be attributed to CEOs? This question has

been the focus of a long debate in management research. In a recent study, I showed that a large

portion of the performance differences that are often attributed to CEOs might in fact be due to

chance. In a recent article, Quigleyand Grafn argue that my conclusions can be avoided if more

advanced methodological approaches are applied. Here I show that this is not the case, in fact

if more realistic assumptions of how chance can affect rm performance are made, the effect of

CEO leadership is almost indistinguishable fromthe effect of chance, independent of the estimation

methodology. Copyright © 2016 John Wiley& Sons, Ltd.

INTRODUCTION

The question of how important CEOs are for

explaining performance differences between rms

has been the focus of a long debate in strategic

management (e.g., Lieberson and O’Connor, 1972;

Mackey, 2008; Quigley and Hambrick, 2015;

Thomas, 1988). In my 2014 article “The use of

variance decomposition in the investigation of the

CEO effect: How large must the CEO effect be to

rule out chance?” I contributed to this debate by

showing that a large portion of the so-called “CEO

effect” (the amount of variance in rm performance

that can be attributed to CEOs), as it was measured

Keywords: variance decomposition; CEO effect; chance;

sources of performance heterogeneity; CEOs

*Correspondence to: Markus A. Fitza, Department of Man-

agement, Mays Business School, Texas A&M University, 420

Wehner Building, College Station, TX 77843-4221, U.S.A.

E-mail: markustza@tamu.edu

†The author is also a permanent visiting Professor at the Faculty

of Business and Law, Universityof Newcastle, Australia.

Copyright © 2016 John Wiley & Sons, Ltd.

by previous variance decomposition studies, was

in fact caused by chance effects. I argued that after

industry, year, and rm effects are controlled for,

a rm’s performance has at least two remaining

components, (1) performance due to the CEO’s

abilities and (2) performance caused by chance

events, encompassing luck and random inuences.

A proper method to determine how much of the

performance differences between rms are due to

the inuence of CEOs should not attribute such

chance effects to the CEO, since by denition such

effects are completely out of the control of the

CEO and thus should not be used to determine how

much CEOs matter.

To showhow past variance decomposition analy-

ses attributed chance effects to the CEO I created a

dataset of rm performance in which I left the struc-

ture of the data intact (rms, industries, etc.) but

replaced real rm performance with random noise

(a set of Gaussian distributed random numbers that

are independent of each other) which represented

the chance element in a rm’s performance. Using