Linguistic Cues of CEO Personality and its Effect on Performance

Abstract

Chief Executive Officer (CEO) personality belongs to the most important constructs to understand strategic change and performance of companies but its measurement remains a major obstacle. Therefore, I study whether freely available, large-scale corporate documents reveal linguistic hints about CEOs' personality dimensions (Big-5) by applying a validated, closed based approach developed by Yarkoni (Journal of Research in Personality, 44, 2010) to a sample of FORTUNE 500 companies. I compare the results with validated, machine-learning trained open-based results from a specifically developed tool of 3000 CEOs of S&P 1500 firms obtained from conference calls (Harrison, Thurgood, Boivie, & Pfarrer, Strategic Management Journal, 2019). In order to provide a basis of comparison, I supplement these linguistic personality profiles with a sample of convicted criminal CEOs that provide an archetype of linguistic extreme cases. In addition, these linguistic profiles are complemented with a subsample of actual personality orientation, using the “positive” leadership attribute of CEO humility identified via a video metrics approach. Finally, after controlling for commonly used content analytical constructs (e.g., entrepreneurial orientation, market orientation) to capture CEOs' cognitive state, I link these cross-sectional linguistic profiles to longitudinal firm- level performance data. The analysis indicates that linguistic manifestations of CEO personality explains in conjunction with observable individual, industry and firm-level data a great share of firm performance. While there appear complementarities between open- based and closed based language approaches, the analysis also indicates that the open-based linguistic trait of neuroticism is negatively related to various measures of firm performance. In particular, these measures appear to be more robust against genre effects and language that is unique among CEOs (self-selection effects). The video 2 metric approach confirms the importance of certain linguistic personality markers, in particular neuroticism. Hence, the approach sheds light on the important role of validation procedures in the realm of computer-aided-content-analysis (CATA) as well as the conducive and detrimental effects of CEO personality cues on firm performance. The paper discusses practical implications for share and shareholders wishing to infer personality cues of key decision makers from large-scale corporate documents as well theoretical implications for the advancement of the upper- echelon research.