Do Bubbles Have a Birthdate? The Role of College Interaction in Portfolio Choice

Abstract

We study the link between social interaction and stock market bubbles. We argue that an increase in social interaction may facilitate the birth of a cascade-type pattern and indirectly of a bubble. We concentrate on a form of interaction that is rooted back in the past: college-based interaction - defined as the one that relates the portfolio choice of an investor to that of the other investors who went to the same college. We explain it in terms of a common cultural imprinting and the development of long-term friendship and alumni networks and we directly quantify this bonding effect. We study how it affects bubble-related portfolio decisions: the choice to focus in growth stocks, the decision to invest in a particular stock, the choice to herd and the decision to concentrate the portfolio in few stocks.

We use a new dataset with information on portfolio choice - broken down at the stock level - wealth, income and demographic characteristics of a big panel of investors as well as information on the college they attended and their family situation at the time. We show that the impact of college-based interaction is statistically and economically significant. Investors invest in the same stocks in which their former classmates do and skew their portfolios towards growth stocks if their former classmates do the same. Moreover, investors are more likely to herd with the other investors who went to the same college than with the rest of the population. College-based interaction also affects investors' decision to concentrate their portfolios in few stocks.

College-based interaction is stronger than the other sources of interaction (professional and geographical) and ranks third as the single most important factor affecting portfolio choice, with an explanatory power higher than that of all standard determinants of portfolio choice, such as hedging non-financial income risk, information and familiarity and so on. This holds even after controlling for all the standard motivations brought forward in portfolio theory, such as hedging of non-financial income risk, familiarity and information effects, wealth and income effect, a host of demographic, geographic and professional dummies, trend-chasing and momentum behavior.

Keywords: Portfolio choice, education, individual investors, herding

JEL Classification: G11, G12, I21, G14