I test whether predictable changes in the rate of information flow affect the market risk premium. Four periods of the year when the majority of firms are scheduled to announce earnings account for 70% of realized value-weighted excess returns – but only account for 50% of trading days - from 1926 to 2015. These earnings seasons have average returns of 5.3% and a Sharpe ratio of 0.61 while the remainder of the year has average returns of 2.4% and a Sharpe ratio of 0.25. Using regression-based tests to control for a variety of seasonal effects, I find the market risk premium is up to 17 basis points higher on days when large firms are scheduled to forecast earnings. I then run a series of tests related to (i) market, industry and firm volatility; (ii) cross-sectional measures of market risk; (iii) the retail industry, which delays reporting due to the holiday season; (iv) the relationship between earnings announcements, aggregate earnings and market returns; (v) the January and Wednesday effects; and (vi) economic indicators and macroeconomic cycles. The results support the information flow hypothesis linking earnings announcements to the market risk premium.