Watch Out for the Overly Confident CEO

US News

When I was an officer in the Army, one of the tenets of leadership that I strove to emulate was to publicly give credit to my soldiers when something good happened and accept blame when something bad happened. It was a pretty simple approach to take. I did have great soldiers, so all I usually had to do was get out of the way, and magic happened. However, if I gave unclear orders or conflicting information, things were bound to get screwed up. Good: them. Bad: me.

According to research from Dr. Feng Li of the University of Michigan, in a paper titled Managers' Self-Serving Attribution Bias and Corporate Financial Policies, I'd have made a pretty decent CEO if I lived by the leadership tenets I had in the Army. Dr. Li discovered that there is a psychological phenomenon in the limbic systems of some CEOs which is a predictable indicator of poor corporate financial performance.

This psychological hiccup is called a self-attributing bias, a fancy psychological term for what happens when you think that you have the Midas touch and that when something goes wrong, it is obviously someone else's fault.

We see it all the time in our own daily lives. How many times have you thought to yourself "I'd have made a fortune in that investment if only [insert rational reason here] didn't happen!" That's this bias at work. You don't consider that perhaps it was your own shortcomings, such as following the herd or insufficient math skills that was the root cause of your poor investment outcomes. Instead, it's easier to blame something else and go about your merry way.

While small amounts of self-attributing bias probably won't really affect us in our personal lives, when a CEO has a significant amount of bias here, bad things tend to happen.

Dr. Li measured this bias in CEOs by evaluating the management discussion and analysis sections of10-K filings of publicly traded companies. To determine whether or not CEO had a higher amount of bias, Dr. Li looked for a disproportionate amount of sentences that contained a structure similar to: "[Good thing] [was caused by] [me/we]." or "[Bad thing][was caused by] [external factors]."

Dr. Li found that CEOs who had higher instances of bias demonstrated in their filing language also exhibited the following behaviors:

--Issuing overly optimistic earnings forecasts. Since these CEOs were overconfident about their cash flows, they did not account for possible variance. Their earnings forecasts were more likely to lead to reporting misses than the average earnings forecast.

--Higher use of leverage. Since they thought that their strategies were going to work, these CEOs weren't afraid to take on debt for their projects.

--Increased spending in property, plant and equipment spending. Much of the debt that these CEOs took on went into costly investments in internal projects and infrastructure.

--Decreased dividends. Many times, companies issue dividends when they do see better opportunities to invest in internally to increase earnings for the shareholders. Since the biased CEOs had stronger beliefs in their own capabilities, they were willing to lower the weighted average cost of capital of internal projects, modeling out a lower desirability of issuing dividends.

--Poor market reaction to acquisitions. These CEOs believed that if the Midas touch worked in their own companies, it would work when they applied that same touch to acquisition targets. The markets often begged to differ.

If you are going to invest in individual stocks, you should spend a few extra minutes poring over recent 10-K statements to see if the CEO demonstrates self-attribution bias. While it's not a guarantee of future underperformance, Dr. Li's research shows that it's indicative of forthcoming financial woe. Better yet, stick to index funds. Don't you have better things to do with your time?

Jason Hull is a candidate for the CFP(R) Board's certification, is a Series 65 securities license holder, and owns Hull Financial Planning.



More From US News & World Report
View Comments (0)