The power of fear: Facial emotion analysis of CEOs to forecast firm performance

We present a framework to predict the firm performance based on CEO's facial emotion dynamics during video interviews. This is a challenging pattern recognition problem that involves robust detection of various emotional expressions in the uncontrolled environment and classification of spontaneous spatio-temporal data with different personal and biological signature traits. We assess the reliability of outcomes that result from the classification through noise suppression and information maximization with correlation analysis. The experimental results show the strong and positive reaction of firm performance merely to CEO's Fear emotion. We analyze the determinants that affect Fear emotion and find that Fear is more in newer CEO, and is more when CEO has less control over the board of the firm. These findings may contribute to prediction of the market value of firms and actions taken on effective investment.