Ram Chellappa
Associate Professor, Information Systems & Operations Management
Emory University
Friday, October 31, 2014
10:00am – 11:30am Speakman Hall 318
Seminar Title: On the Temporal Nature of Sales to Rank Relationships of Digital Music
Abstract
A significant amount of work in IS, economics and marketing has used the relationship suggested by Chevalier and Goolsbee (2003) to impute demand from the relative sales-rank of a product in its category. However many industries, in particular the music industry are subject to temporal changes suggesting that the sales to rank relationship may not be fixed. Analyses of weekly album sales data not only reveals statistically different sales distribution week-to-week, but also significant differences in this relationship. To account for this difference, our research incorporates two temporal factors, namely size of the competition and market. We further plan to contrast physical and digital ranks and sales of albums to examine any distinct differences in consumption patterns.
Prior research suggests that the relationship between sales of a book and its category rank is a power-law distribution, more specifically a Pareto distribution. A number of papers in IS have used this relationship for the online book industry (specifically Amazon.com) to be of the form: R =α Sθ . While this relationship may very well hold for the book industry, the music industry is undergoing great changes with digitization. A number of elements are unique to this industry: First, weekly distribution of sales is quite different from each other and over the years increasingly fewer titles are accounting for a greater percentage of sales. Second, album sales largely exhibit a distinct exponential decay sales pattern beginning with their week of release and thirdly, there are significant seasonality issues wherein competition (as measured by number of new titles) is most severe in the 4th quarter. Our research is focused on specifically examining the temporal variation in the shape parameter θ .