Neighbors share a city, employers and news, so a similarity in their portfolios proves nothing at all about whether they ever spoke. Those shared factors were controlled for, and actual purchases compared. What was left after the controls?
Curiosity counts. Here is the answer.
Yes, and part of it really is conversation. The difficulty is that neighbors share a city, an employer, a local paper and exposure to the same news, so similar portfolios prove nothing about talking. The authors removed those factors and compared actual purchases, and neighbors’ purchases still predicted a small but measurable part of the choice. Two further checks support the conversation reading: the link is stronger in shares of local companies, and in states where people report more sociable relations with neighbors. The caveat: the conversations were never recorded or observed, only inferred indirectly.
A small, measurable share. A ten rise in neighbors’ purchases from a given industry was linked to a two point rise in the household’s own purchases from that industry. The link was stronger in and in more sociable states, and the authors attributed roughly a quarter to a half of to rather than to a shared environment.
The claim that a choice of share passes between neighbors sounds plausible, which is exactly why it is hard to prove. Neighbors share a city, employers, a local paper and an economic situation, so similar portfolios are expected even if they never spoke. The whole problem is separating a shared environment from a conversation.
The data allow an attempt. Household purchases at a broker during the 1990s were compared with purchases by neighbors in the same area, controlling for characteristics of the place and of the industry. The result: a ten percentage point rise in neighbors’ purchases from an industry was linked to a two point rise in the household’s purchases from it.
Two details strengthen the conversation reading. The link is stronger in shares of local companies, which are easy to talk about because everyone knows them, and it is stronger in states where people report more social interaction. Neither dependency is expected if shared news is all that is at work.
Their estimate is that roughly a quarter to a half of the correlation comes from word of mouth. That should be read as an estimate rather than a measurement: nobody recorded conversations between neighbors, and the communication was inferred indirectly from the patterns. What can be said is that a shared environment alone does not explain all of the similarity.
wat. When neighbors bought more from an industry, the household leaned that way too, especially in local shares and sociable areas.
wat wat
A financial idea can spread socially like a product or a rumor, even when nothing about the company’s value changes.
How do we know?
Household purchases at a broker between 1991 and 1996 were compared with neighbors’ purchases and with the characteristics of the place and the industry.
What is still unknown
Communication was inferred indirectly; the data contain no recording of conversations between neighbors.
Learn something?
Where to next
What happens to an important report when many companies publish on the same day?markets
How can easier trading in a share raise the danger of a fall?markets
Can the ability to bet against a share restrain financial reporting?markets