“The social learning channel is observational,” explains Ederer. “When I hear that my friend wants to purchase a certain asset it makes me think that the asset is of high quality because I trust my friend’s judgment, and therefore I buy the asset.”
The social utility channel, on the other hand, “is what we often refer to as ‘keeping up with the Joneses.’ When my friend buys an asset, it doesn’t necessarily make me change my beliefs about the quality of the asset, but his possession of it makes me worry about falling behind him in terms of wealth and consumption. If the asset’s value shoots up, he’ll be rich and I’ll be poor, so that makes me buy the asset.”
Both social learning and social utility have been widely studied theoretically, but have been difficult to study in practice. Typically, an investor making the decision to purchase an asset implies that he also purchased that asset, making it difficult to separate the effects of learning that a friend thinks highly of an asset, inherent in social learning, from the effects of learning that a friend actually has the asset, as occurs in social utility. To disentangle the two channels, Ederer and his co-authors worked with a large financial brokerage firm in Brazil to conduct a high-stakes field experiment. The brokerage firm offered a new financial asset to pairs of clients who shared a social relationship. The minimum investment was R$2,000, the equivalent of more than $1,000 U.S. dollars at the time of the study, which was approximately 50% of the average investor’s monthly income in the sample.
In the experiment, a broker called the first investor in the social pair and offered him the new asset, explaining that was in limited supply and that even if he wanted to buy it, he might not be able to. When the first investor indicated that he wanted to buy the asset, the broker then called the second investor in the social pair. The second investor was randomly assigned to one of three study treatments. In the first, the control group, he heard the same sales pitch as the first investor but was told nothing about his peer’s desire to purchase the asset. In the second, he was truthfully told that his peer wanted to buy the asset but was not allowed to, which, the authors hypothesized, should turn on the social learning effect. In the third treatment group, he was told that his peer wanted to purchase the asset and did purchase, which should activate social utility as well as social learning.
The authors compared how many of the second investors in each group purchased the asset and found that both social learning and social utility significantly influenced purchase decisions. Those in the control group had a purchase rate of 42%, compared to 71% of those in the group who heard that their peer wanted the asset but didn’t get it, and almost 93% of those in the group who heard that their peer wanted the asset and was able to buy it.
“When you hear that your friend wanted to buy the asset and also got the asset you’re even more likely to buy,” says Ederer.
The authors also found that the more sophisticated the first investor was, the stronger the social learning effect for their peers. “Social learning from peers matters for financial decisions, especially among unsophisticated investors,” says Ederer. “There’s just more to learn from a sophisticated investor. If I hear that one of my finance colleagues wants to buy an asset, there’s a lot of information contained in that decision. She’s a finance professor; she’s very sophisticated; I trust her judgment; there’s a reason why she wants to buy the asset. In contrast, if I hear that my grandfather, who is not a sophisticated investor, wants to buy an asset, I learn relatively little.”
Ederer and his co-authors also examined the purchase decisions of the second investors whose peers did not want to purchase the asset and found that their purchase rate dropped to about 35%. “People tend to associate with people who are like themselves, so we would expect a different result from investors who are connected with people who wanted to invest compared to investors who are connected with people who didn’t want to invest.”
The study has implications for understanding financial bubbles, says Ederer. “We now have some evidence that financial bubbles may form because of informational cascades based on social learning or because of herd behavior that derives from the preferences of investors to behave like their peers.”
The study applies outside of the context of assets, Ederer explains, to social marketing and health interventions such as curbing risky adolescent behaviors. He points, as an example, to the marketing of the Nissan Cube in Canada. Nissan took applications from people who wanted to prove that they were cool enough to drive the Cube. Those who were selected were given a Cube to drive with the hope that they would influence others to buy the car. “It’s like harnessing the ‘keeping up with the Joneses’ effect,’” Ederer says. “If I see cool people driving this car, then maybe I should also drive this car. You also see this on Facebook, where they suggest pages that you should like because your friends like them.”
Ederer adds, “Our methodology applies in any setting where you have interactions with ‘keeping up with the Joneses’ effects and learning effects. This can range from how CEOs make acquisition decisions to whether teenagers decide to smoke or not.”
“Understanding Mechanisms Underlying Peer Effects: Evidence from a Field Experiment on Financial Decisions,” is published in Econometrica by Leonardo Bursztyn (UCLA Anderson School of Management), Florian Ederer (Yale School of Management), Bruno Ferman (FGV Escola de Economia de São Paulo), and Noam Yuchtman (UC-Berkeley Haas School of Business).