Competitive Dynamics in Forecasting: The Interaction of Skill and Uncertainty

Authors


  • The authors' names are listed alphabetically.

Correspondence to: Natalia Karelaia, INSEAD, Boulevard de Constance, 77305 Fontainebleau, France. E-mail: natalia.karelaia@insead.edu

ABSTRACT

The outcomes in many competitive tasks depend upon both skill and luck. Behavioral theories on risk taking in tournaments indicate that low-skilled individuals may have incentives to take more risks than high-skilled ones. We build on these theories and suggest, in addition, that when luck is more important in determining outcomes, the increase in risk taking is larger for low-skilled than high-skilled individuals. We test this hypothesis by analyzing stock analysts' forecasts of companies' earnings per share under market conditions that vary in volatility and thus imply different levels of luck in outcomes. Specifically, noting that forecasts that deviate widely from the consensus—which is observable by the analyst—potentially carry career-related rewards but also reputational risks, we examine the degree of deviation from consensus exhibited by analysts of different skill levels (measured by both past forecasting accuracy and education) in different market conditions. We find that average deviations from consensus increase as markets become more volatile. At the same time, under conditions of high volatility, low-skilled analysts exhibit larger increases in deviations from consensus than high-skilled analysts. These field data results support our hypothesis based on of risk taking in tournaments. We discuss alternative interpretations such as, for example, self-serving attributions and indicate directions for future research. Copyright © 2012 John Wiley & Sons, Ltd.

Ancillary