Are CEOs really as important as we’re led to believe?
A book of randomness and chance I’m reading suggests not.
Leonard Mlodinow’s ‘The Drunkard’s Walk: How randomness rules our lives’ has a beautiful example that illustrates the role of chance in performance and that shows how assumptions are made when using past performance to predict future results.
He takes the CEOs of the top 500 (Fortune 500) companies and assumes each of them has a certain probability of success each year, defined by their own company’s measure. And also assumes that these successful years occur at a frequency of 60%. He asks whether this means in a 5-year period each CEO will have a 60% success rate (3 successful years).
And he shows that the chances that in a given 5-year period any single CEO’s success rate will match the underlying rate is 1 in 3. In terms of the Fortune 500 companies this means that over the past 5 years (not an insignificant amount of time) around 333 CEOs would demonstrate results that did not reflect their true ability!! And by chance alone around 10% of the CEOs will have a 5 year winning or losing streak.
I have always argued that CEOs (and football managers, etc) often seem to be given credit and apportioned blame in amounts disproportionate to their real observable effect on a company. If only because they haven’t been at the company long enough for the results to shake off the randomness that can crop up anywhere.
Mladinow’s book has more insights on related topics such as
• The Law of Large (and small) Numbers (small sample sizes not reflecting true results, like for example “8/10 women prefer this shampoo to others” when only 100 women have been asked..)
• Human preferences for spotting patterns where there is true randomness, and vice-versa.
• Determinism where actors (in the social science sense) believe and act as though they have a greater say in an outcome than is true.
Mladinow also highlights throughout the book areas such as medical trials, court trials and betting shops, as well as day-to-day life, where knowledge of randomness and chance is absolutely critical.