Social Interaction and Investor Behavior in Stock Markets

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

This project investigates whether investor behavior in stock markets is influenced by social interaction. One channel through which social interaction might influence investors is that they spread information and ideas about stocks to one another directly, through word-of-mouth communication. However, in spite of their familiarity, such hypotheses about social interaction have received little direct support in stock-market data. In an effort to bring large-sample evidence to bear on these questions, this project conducts two studies of social interaction and investor behavior in stock markets. The first study investigates the idea that stock-market participation by households is influenced by social interaction. A simple model is built in which any given "social" investor finds it more attractive to invest in the market when the participation rate among his peers is higher. The model predicts higher participation rates among social investors than among "non-socials". The theory is tested using data from the Health and Retirement Study. Social households are defined as those who interact with their neighbors, or who attend church. The investigators test whether the social investors are indeed substantially more likely to invest in the stock market than non-social households, controlling for other factors like wealth, race, education and risk tolerance. The investigators also test an auxiliary prediction of their model that the impact of sociability is stronger in states where stock-market participation rates are higher. A second study investigates word-of-mouth effects on the holdings and trades of money managers. The idea is a very simple one, premised on the assumption that fund managers who work in the same city are more likely to come into direct contact with one another, and hence to exchange ideas by word-of-mouth. Consider for example a fund manager working for the Fidelity fund family, which is located in Boston, and her decision of whether or not to buy, say, shares of Intel in a given quarter. Our basic prediction is that this decision will be more heavily influenced by the decisions of fund managers working for the Putnam family, which is also located in Boston, than by the decisions of fund managers located in other cities. Using a large database of mutual fund holdings, the investigators look to see whether a mutual-fund manager is more likely to hold (or buy, or sell) a particular stock in any quarter if other managers from different fund families located in the same city are holding (or buying, or selling) that same stock. In these regressions, we are able to control for the distance between the fund manager and the stock in question and for heterogeneity in investment styles. Finally, in a model in which investors spread information and ideas about stocks directly to one another by word of mouth, it takes time for ideas about stocks to travel over large distances, much as with a contagious disease. The investigators can test for such trading dynamics by checking whether a given fund manager's trades responds to the trades of other fund managers with more of a lag when these other managers are located in more distant cities.