‘Good questions outrank easy answers.’ –Paul Samuelson
This paper documents a close connection between the timing of corporate news disclosures and CEOs’ absences from headquarters. I identify CEO absences by merging publicly available flight histories of corporate jets with real estate records of CEOs’ property ownership near leisure destinations. I find that CEOs go to their vacation homes just after companies report favorable news, and CEOs return to headquarters right before subsequent news is released. More good news is released when CEOs are back at work, and CEOs appear not to leave headquarters at all if a firm has adverse news to disclose. When CEOs are away from the office, stock prices behave quietly with sharply lower volatility. Volatility increases immediately when CEOs return to work. Mandatory Form 8-K disclosures of material company news are more likely to be filed late if news occurs while CEOs are at their vacation homes. […]
The paper’s results seem consistent with an agency cost hypothesis, under which CEOs might slow down their firms’ news disclosures for personal convenience on the days that they requisition company aircraft for golf or ski trips. However, the observed associations between news disclosures and vacation schedules may well be endogenous, if CEOs plan to be away from the office when the company expects to have little news to announce. To understand more clearly the direction of causation between disclosures and CEO absences, I conduct a variety of tests, examining how company news announcements change when CEOs return to headquarters at unexpected times. I also estimate a bivariate probit model of news days and vacation days, in which weather at the CEO’s vacation site is used as an instrumental variable that should be associated with trips to the vacation home but not be connected to company news developments. Much of the analysis from these tests supports the agency interpretation, with news releases appearing to occurr less frequently simply because the CEO is absent from the office. However, it is difficult to test causation in the other direction, which would require an instrumental variable associated with news at headquarters but uncorrelated with the CEO’s decision to take time off.
