War of Ideas
The CNBC effect: Are narcissistic CEOs more likely to be ousted?
Self-serving attribution bias is the tendency people have to attribut success to their own abilities, while attributing failure to outside factors. Readers will probably not be shocked to hear that many CEOs of large corporations exhibit this particular characteristic, but research by finance professor Andy Kim of the Nanyang Business School in Singapore suggests that ...
Self-serving attribution bias is the tendency people have to attribut success to their own abilities, while attributing failure to outside factors. Readers will probably not be shocked to hear that many CEOs of large corporations exhibit this particular characteristic, but research by finance professor Andy Kim of the Nanyang Business School in Singapore suggests that this kind of thinking is ultimately self-defeating for business leaders. What caught my eye about the paper was the dataset Kim used to measure this type of bias:
We measure the self-referencing behavior and the SAB of CEOs using transcripts of 6,931 CEO interviews on CNBC from 1997 to 2006 as in Kim and Meschke (2011). CEO interviews on financial media are close to the ideal setting for this research: the journalist (show host) frequently asks about the firm’s past performance and the CEO answers by attributing the firm’s success to specific causes. Although many CEOs must go through interview preparation supported by the companies’ Public Relations Division, the unobserved effect of PR training would only add noise to our empirical study and bias against finding the results. Moreover, analyzing the linguistic content of the spoken language of the CEO enables us to estimate cross-sectional variation in the psychological bias of the CEOs, namely the SAB.
We use the computational linguistic technique in Li (2010), where he measures the SAB using the management discussion and analyses (MD&A) sections of annual reports of public US public companies. We also use the Linguistic Inquiry and Word Count (LIWC) dictionary. LIWC is one of the most widely used text analysis software developed by James W. Pennebaker, Roger J. Booth, and Martha E. Francis. We first identify the “causal” sentences from the CEOs’ words in the transcripts. We capture whether the singular and plural first-person pronouns (the 22 words in LWIC dictionary under “I” and “We” categories, such as I, me, my, we, our, and us) are used in the sentences, and count it as “causal_we.” Equivalently, we capture whether second- or third-person references (the 47 words in LWIC dictionary under “You,” “SheHe,” and “They” categories, such as you, your, they, and their), including “competitor(s),” “industry,” or “economy,” are used in the causal sentences, and count them as “causal_other.” For each interview, “causal_we” minus “causal_other” divided by the total word count of the CEO is defined as the measure of self-referencing.
Here’s what he found:
With the novel dataset, we find that stock market investors do not welcome the CEOs who excessively attribute the outperformance of the company to their own ability and the underperformance of the company to bad luck or the economy. In addition, our hand-collected CEO turnover data based on Execucomp enables us to find consistent results with the prediction by Gervais, Heaton, and Odean (2011): CEOs with SAB are more likely to be fired sensitively to performance, especially if the governance is stronger. CEO dismissal event study reveals that the stock market response to the announcement of forced turnover of CEOs with SAB are significantly more positive by up to 9.7% over the event window of [-1,1] days. We also find inverse U-shaped relation between the stock market response to acquisition announcements and the self-attribution bias of the CEO in CNBC interviews.
Of course, the people with the kind of self-confidence that comes with this sort of bias may also be the most likely to get into these positions in the first place.