Information Diffusion Effects in Individual Investors' Common Stock Purchases Covet Thy Neighbors' Investment Choices
Working Paper 13201
---- Abstract -----
We study the relation between households' stock purchases and stock purchases made by their neighbors. A ten percentage point increase in neighbors' purchases of stocks from an industry is associated with a two percentage point increase in households' own purchases of stocks from that industry. The effect is considerably larger for local stocks and among households in more social states. Controlling for area sociability, households' and neighbors' investment style preferences, and the industry composition of local firms, we attribute approximately one-quarter to one-half of the correlation between households' stock purchases and stock purchases made by their neighbors to word-of-mouth communication.
-----Part-----
Despite the fact that individuals collectively hold about one-half of the U.S. stock market, information diffusion effects among individual investors—the relation between the investment choices made by an individual investor’s neighborhood and the investor’s own investment choices—have received relatively little attention in the academic literature, probably because of the lack of detailed data. If present, such effects undoubtedly can affect individual investors’ asset allocation decisions. Moreover, trades based on information diffusion might be sufficiently correlated and condensed in time to affect stock prices.
In the domain of institutional investors, Hong, Kubik, and Stein (2005) study word-of-mouth effects among mutual fund managers and find that “…a manager is more likely to hold (or buy, or sell) a particular stock in any quarter if other managers in the same city are holding (or buying, or selling) that same stock.” This study complements their work by ascertaining whether such trading patterns are a broader phenomenon. For example, individual investors may seek to reduce search costs and circumvent their lack of expertise by relying on word-of-mouth communication with those around them. Indeed, Hong, Kubik, and Stein (2004) present a model in which stock market participation may be influenced by social interaction. Such social interaction can serve as a mechanism for information exchange via “word-of-mouth” and/or “observational learning” (Banerjee (1992), Ellison and Fudenberg (1993, 1995)). Duflo and Saez (2002, 2003) present evidence of peer effects in the context of retirement plans. They find that an employee’s participation in retirement plans and choices within those plans are affected by participation decisions and choices made by other employees in the same department.
In the international arena, Feng and Seasholes (2004) present evidence of herding effects among individual investors who hold individual brokerage accounts in the People’s Republic of China. A unique feature of their data (investors seeking to place trades in person can` do so only in the brokerage house in which they opened their accounts) enables Feng and Seasholes to disentangle word-of-mouth effects from common reaction to releases of public information. They find that common reaction to public information (trades placed across branches in the same region, local to the company), rather than word-of-mouth effects (trades placed in the same branch), seems to be a primary determinant of herding in that context.
Grinblatt and Keloharju (2001) find that proximity to corporate headquarters, the language of communication with investors, and the company’s CEO’s cultural origin are important determinants of Finnish households’ stock investments. Whereas these findings could be consistent with word-of-mouth effects influencing portfolio choice, they could also reflect households’ tastes for familiarity—preference to invest in companies that disseminate annual financial reports in their native tongues or feature a CEO with the same origin.
We study information diffusion effects among U.S. individual investors by using a detailed data set of common-stock investments 35,673 U.S. households made through a large discount brokerage in the period from 1991 to 1996. Throughout the paper, we loosely refer to the correlation between households’ investments and their neighbors’ investments as “information diffusion.” This term is intended to encapsulate several potential reasons why such correlation exists—word-of-mouth effects, similarity in preferences, as well as common local reaction to news. To further characterize information diffusion and word-of-mouth effects, we consider state-level measures of sociability and find that the level of sociability prevailing in the state to which the household belongs (likely a strong correlate of the presence of word-of-mouth effects) can explain a significant portion of the overall diffusion effect. Moreover, we disentangle the diffusion into the influences of common preferences, structure of the local industry, and word-of-mouth effects.

Putting our results in perspective and comparing them with the findings from Feng and Seasholes (2004) delivers a new, richer understanding of the different mechanisms that govern individuals’ investment decisions across various societies. Indeed, whereas Feng and Seasholes (2004) report that individual investors’ correlated investment decisions are driven by common reaction to locally-available news, with no evidence of word-of-mouth effects among Chinese investors, our estimates suggest that word-of-mouth effects among U.S. investors are strong, particularly in more social areas. This discrepancy is consistent with the differences in the fundamental characteristics of the two societies. Freedom House, which has been producing annual ratings of political and civil rights for more than 200 countries for the past three decades (Freedom House (2004)), has ranked the U.S. among the highest and the People’s Republic of China among the lowest along the dimension of civil liberties. An essential ingredient of the civil liberties score is prevalence of open and free discussion (or absence thereof). Coupled with the fact that many, if not most companies in the People’s Republic of China are at least partly government-owned, it is very plausible that exchanging investment-relevant information in a society deprived of open and free discussion and many other civil liberties is rare and modest.
Even within the U.S., there is variation in sociability (e.g., membership in clubs, trust in other people). If word-of-mouth is an important contributor to households’ stock purchases, the observed correlation in a household’s portfolio allocation and that of its neighbors should be higher in the more social areas. Other explanations for information diffusion effects, such as correlated preferences and common local reaction to news, should not vary with the sociability of the community. Using state-level variation in sociability measures enables us to differentiate among the competing hypotheses that can explain trading patterns of U.S. investors.

Overall, we find a strong information diffusion effect (“neighborhood effect”): a ten percentage point increase in purchases of stocks from an industry made by a household’s neighbors is associated with an increase of two percentage points in the household’s own purchases of stocks from that industry. We pay particular attention to the differentiation between information diffusion effects related to local stocks (defined as companies headquartered within 50 miles from the household) and the effects related to non-local stocks. Whereas the key neighborhood effects—similarity in preferences, the impact of the structure of the local industry, and word-of-mouth—can prevail among the investments both local and non-local to the household, most of those effects will likely be far more pronounced among local investments because, as demonstrated for both professional money managers (Coval and Moskowitz (2001)) and individual investors (Ivković and Weisbenner (2005)), the flow of value-relevant information regarding local companies appears to be higher and of better quality than the comparable flow regarding remote, non-local companies.