Abstract
My dissertation contains three essays in financial economics. The first essay shows that productivity shocks to the top 100 U.S. companies (as identified in Gabaix (2011)) contain systematic information. Specifically, I show that shocks to the top 100 firms predict future shocks to geographically close firms. Intra-sector trade links are an important economic channel for the cascade effect. However, these geographic spillovers are not only restricted to firms' explicit interactions. State income tax payments is another dominant channel through which the shocks propagate. Moreover, the results indicate that market participants, including equity analysts, do not fully incorporate the geographic information contained in shocks to the top 100 firms. Consequently, a trading strategy that exploits the slow diffusion of information generates an annual risk-adjusted return of 7.5%. The second essay explores the economic impact of locally-dominant firms. In particular, we apply the Gabaix (2011) method to the U.S. states and identify firms that influence state-level business cycle (i.e., locally-dominant firms). We find that idiosyncratic shocks to locally-dominant firms, which are not among the largest 100 U.S. firms (i.e., nationally-dominant firms), explain a significant portion of local business cycle. Shocks to locally-dominant firms also impact the U.S. aggregate fluctuations. Specifically, shocks to locally-dominant companies explain over 50% of U.S. GDP growth, which is larger than the impact of nationally-dominant firms. These findings show that shocks to locally-dominant firms can have large aggregate impact on the U.S. business cycle. The third essay investigates whether equity analysts are subject to in-group favoritism when forecasting earnings of firms. Specifically, we argue that equity analysts may have less favorable views about firms that are not headed by CEOs of their own “group." We define groups based on gender, ethnicity, and political attitudes. Examining analysts' earnings forecasts we find that, compared to female analysts, male analysts have lower assessments of firms headed by female CEOs than of firms headed by male CEOs. As a result, earnings surprises of firms headed by female CEOs are systematically upward biased. Results are very similar if we define in-groups based on ethnicity or political attitudes. Analysts' “buy" and “sell" recommendations are also biased towards their own in-group. Examining cumulative abnormal returns (CARs) surrounding earnings announcements, we do not find that the market undoes this bias.