Abstract: Financial market participants’ knowledge about corporate events crucially depends on media generated information. The United States media landscape is characterized by two different types of media outlets, local and national. This study examines how media heterogeneity affects the information diversity in the dissemination and interpretation of corporate frauds. Using a local firm's AAER as a setting of fraud, I find that local reports are more original as measured by textual dissimilarity from exiting media content. In particular, textual analysis indicates that the local reports are more negative, emphasize litigation topics and de-emphasize executive-related topics. In addition, the information diversity provided by local reports is informative to investors, as local reports reduce information asymmetry and resolve market uncertainty more than national media reports do. This study builds on the recent literature that shows the effectiveness of local media but does not distinguish between the ex-ante deterrence effects from the ex-post interpretation effects of media coverage. My study explicitly identifies the ex-post channel and provides one of the first evidence comparing the relative informativeness of local versus national media.

Our paper documents a novel and robust empirical relation between analyst forecast errors and forecast revisions. We show that when analysts revise their forecasts positively (negatively), forecast errors will be significantly negatively (positively) correlated with revisions. This relation exists in both the US (on both the individual and the consensus levels) and foreign markets, and for both short-term and long-term forecasts. We propose an ambiguity-based model to explain this empirical finding, where analysts provide their forecasts strategically to help ambiguity-averse investors reduce ambiguity and form better beliefs. Our theory is supported by additional tests of model predictions in various settings.