We focus on CEO specifications because information about a firm’s CEO is usually reported in ExecuComp, whereas information about its CFO is
sometimes missing (our sample size drops from 5,768 to 4,709 observations when we adopt CFO specifications instead). In addition, ExecuComp provides
the exact date on which a manager becomes CEO, but only gives the appointment year for the CFO, which introduces noise to the data. Finally, we
examine the data visually and note that, in many cases, CEO and CFO tenure overlap, which renders it difficult to distinguish between the CEO’s and CFO’s
effects on forecasting behavior empirically.
our sample size drops from 5,768 to 4,709 observations when we adopt CFO specifications instead