First-year student Rohan Rajiv is blogging once a week about important lessons he is learning at Kellogg. Read more of his posts here.
A Google search for “CEO Marathon study” reveals a slew of articles talking about how marathon running CEO’s lead successful companies.
The study has been quoted in numerous blogs and articles since it was published. Now, let’s take a closer look at the abstract of the study (the first line will do).
This study finds a positive relation between CEO fitness and firm value. For each of the years 2001 to 2011, we define CEOs of S&P 1500 companies as being fit if they finish a marathon. The literature suggests that fitness moderates stress and positively affects cognitive functions and performance. Accordingly, we find the strongest effects on firm value in subsamples where fitness is most important, i.e., for CEOs with high workload, above median age, and above median tenure. Fit CEOs are further associated with significantly higher abnormal announcement returns in M&A bids for large, public, and cross-border targets, concomitant with high stress. Our findings can explain the importance of CEO fitness in the managerial labor market and the trend among CEOs to stay fit.
This is a classic example of the identification error – mixing correlation with causality. Here is an example of the kind of issues that this study might have:
1. What if CEO fitness does indeed have a positive relation with firm value, but given the absence of other control variables, we don’t realize that it is actually healthy eating that really contributes to firm success? This is a great example of omitted variable bias.
2. There is just a positive relation between CEO fitness and firm value. What if it is high firm value that causes CEOs to get fit and not the other way around?
3. Are marathons the only way CEOs mitigate high stress environments? Could yoga or video games be an option?
I could go on.
I have been guilty of not being critical about the identification error multiple times. And, a discussion of this study in our statistics class was a nice wake up call to be more critical about results from studies. Studies that might be published in obscure publications might be obscure for a reason. And, we should worry when we see studies that just find a positive relation between two variables.
To illustrate, some researchers in the 1980s concluded that the key to success was to boost self esteem. So, an entire generation of parents and educators focused on boosting self-esteem of kids by giving prizes to everyone and ensuring they always felt encouraged. It was only in the late 90s when a panel of psychologists who reviewed these studies concluded that the researchers had done it wrong. A trait that causes success is self control / high willpower. Kids with high self control inevitably do well and have high self esteem. So, while self esteem might be correlated to success, it is self control that has a causal relationship. There are many such examples.
Pay attention. This stuff matters.
Rohan Rajiv is a first-year student in Kellogg’s Full-Time Two-Year Program. Prior to Kellogg he worked at a-connect serving clients on consulting projects across 14 countries in Europe, Asia, Australia and South America. He blogs a learning every day, including his MBA Learnings series, on www.ALearningaDay.com.