The dynamic informativeness of scheduled news

We propose a method to identify the informativeness of a future scheduled announcement at the daily level, exploiting the discontinuity it creates in the term structure of option volatility. We implement the strategy in a panel data model to estimate the relation between prior signals and the future announcement. This method allows to separate substitutes from complements, can isolate multiple signals within the same quarter and can condition on the timing and signal characteristics. We find that analyst forecasts substitute earnings announcement information and recommendation provide extra information on top of forecasts. Moreover, our evidence suggests that insiders sell to avoid uncertainty when the announcement is far away but pull forward earnings information when they trade one month before.