The Identification of Social Learning in Online Markets
27 Pages Posted: 20 Sep 2013 Last revised: 12 Jun 2015
Date Written: June 3, 2014
I investigate the identification of social learning in online markets. In these markets, there is often a two-stage decision making process where a consumer first chooses whether or not to search a product and then chooses whether or not to buy it. Additionally, the social learning signals are regularly observable before the search decision. I posit that the ideal test for learning involves estimating the effect of a signal on the search probability, but one can also identify learning by finding the signal's effect on the purchase probability. However, using the purchase decision results in an increased likelihood that learning is incorrectly rejected. To demonstrate this, I perform three empirical exercises. The first uses simulation, while the others use data from an experimental and a real-world online market for music. The advantage of the latter two data sets is that I observe both a search decision (listening to the sample of a song) and a purchase decision (the downloading of a song). The results of the paper suggest a possible explanation for the difference in the amount of empirical evidence supporting word-of-mouth versus observational learning. Further, it provides researchers and managers a road map for identifying learning in line markets.
Keywords: Observational Learning, Word-of-Mouth, Search Good, On-Line Markets, Consider-Then-Choose
JEL Classification: L15, L82, M31, D83
Suggested Citation: Suggested Citation