Date of Degree


Document Type


Degree Name





Lin Peng

Committee Members

Karl Lang

Xi Dong

Dexin Zhou

Subject Categories

Finance and Financial Management


Social Networks, Social Connectedness Index, Institutional Investors, Portfolio Returns, Market Return Predictability, Flu


This dissertation consists of three chapters that encompass social network, institutional investor behavior, financial market efficiency, and market return predictability.

Chapter 1: We use social network data from Facebook to show that institutional investors are more likely to invest in firms that are located in areas the investor is more socially connected to. This effect is particularly large for investments in small and informationally opaque firms. After controlling for social connectedness, investments do not vary with the geographic distance between investors and firms. We find no evidence that investors achieve higher returns when investing in firms located in areas that the investor is socially connected to. In contrast, we show that firms in locations that are socially proximate to capital have higher valuations and higher liquidity, again with the largest effects for smaller firms with lower analyst coverage. In addition to these cross-sectional results, we document that, during Hurricane Sandy, there was a disproportionately adverse effect on the liquidity of firms that were socially connected to regions affected by Sandy. Our results suggest that the geographic structure of social networks affects firms’ access to capital and can contribute to geographic differences in economic outcomes.

Chapter 2: We provide the first empirical evidence on the link between long-short anomaly portfolio returns---the bedrock of the cross-sectional literature---and the time-series predictability of the market excess return. Using 100 long-short anomaly portfolio returns and a variety of forecasting methods, and with machine learning tools that extract signals efficiently, we find that the anomaly portfolio returns evince economically and statistically significant predictive ability for the market excess returns. Our results can be explained by asymmetric arbitrage costs and overpricing dominance in the equity market. Consistent with this explanation, we find that out-of-sample return predictability is significantly related to various market frictions associated with arbitrage costs.

Chapter 3: A strand of the economics literature finds that the health conditions will affect the productivity of workers and have an impact on the real economy. This chapter explores its implications for the financial market. We hypothesize that the health condition of investors affects information processing and diffusion in stock markets. To test this hypothesis, we examine the effect of Influenza on the price reactions to earnings news. We find that the immediate return response to an earnings surprise is weaker, and the post-announcement drift is stronger during times of high flu intensity. In addition, we find that high flu intensity is associated with a decrease in both institutional and retail investor attention. The distraction effect for institutional attention is weaker on earnings announcement days but still significant. Our evidence is consistent with a causal relationship between investor inattention and market underreaction.