TALKS

October 1, Stanford GSB Operations, Information and Technology Seminar, Stanford, CA

October 12-15, INFORMS Annual Meeting, Washington, DC

January 8-9, Fifth bi-annual Conference on The Economics of the Software and Internet Industries, Toulouse, France

WORKING PAPERS

“Word of Mouth and Recommender Systems: A Theory of the Long Tail”

I present a model to assess the extent to which recommender systems can account for the "long tail", an increase in the tail of the sales distribution. Consumers face a search problem within a pool of horizontally differentiated products supplied by a monopolist. They are endowed with a taste profile that determines their probability of matching with any given product, but arrive to the market uninformed and cannot identify which products are more likely to yield a match. Consumers may search for a match by drawing products from the assortment or by seeking word of mouth recommendations from other consumers. Product evaluations prior to purchase and the exchange of recommendations are both shown to arise endogenously, increasing firm profits and the concentration of sales. Introducing a recommender system to act as an intermediary in the recommendations exchange further increases firm profits and affects sales concentration. Insights are derived on the mechanisms driving concentration in artistic markets and their implications for the long tail debate. The model is suited for experience good markets such as music, cinema, literature and video game entertainment.

SSRN link

“Peer-to-Peer File Sharing and the Market for Digital Information Goods”
Under revision for the Journal of Economics and Management Strategy (JEMS)
Joint with Ramon Casadesus-Masanell

We study competitive interaction between two alternative models of digital content distribution over the Internet: peer-to-peer (p2p) file sharing and centralized client-server distribution. We present microfoundations for a stylized model of p2p file sharing where all peers are endowed with standard preferences and show that the endogenous structure of the network is conducive to sharing by a significant number of peers, even if sharing is costlier than freeriding. We build on this model of p2p to analyze the optimal strategy of a profit-maximizing firm, such as Apple, that offers content available at positive prices. We characterize the size of the p2p network as a function of the firm's pricing strategy, and show that the firm may be better off setting high prices, allowing the network to survive, and that the p2p network may work more efficiently in the presence of the firm than in its absence.

SSRN link

Also check out our HBS Case and our Q&A for HBS Working Knowledge.

“Bandwidth Allocation in Peer-to-Peer File Sharing Networks”
Computer Communications 31 (2008) 257–265
Joint with Albert Creus-Mir and Ramon Casadesus-Masanell

We present a model of bandwidth allocation in a stylized peer-to-peer file sharing network with S peers (sharers) who share files and download from each other and F peers (freeriders) who download from sharers but do not contribute files. Assuming that upload bandwidth is scarcer than download bandwidth and efficient allocation, we compute the expected bandwidth obtained by each peer. We show that (i) while the exact formula is complex, S/(S+F) is a good approximation and (ii) sharers (freeriders) obtain bandwidth larger (smaller) than S/(S+F). The paper constitutes a first step towards a general analytical foundation for scarce resource allocation in peer-to-peer file sharing networks.

SSRN link

Watch the presentation at the NET Institute Conference 2007 (streaming video part 3)

“Non-Cost-Raising Discrimination: A Rationale for Functional Separation in Broadband Wholesale”

We present a vertical differentiation model to analyze the quality-wise strategy of an incumbent telecommunications operator under open access regulation. The incumbent serves a wholesale market subject to a price cap and there is free entry by third-party operators. We show that it is always profitable to foreclose the market by degrading the wholesale quality supplied to competitors in a non-recoverable fashion. Unlike the cost-raising discrimination literature, our result is robust to the number of competitors and the price cap level. We show that functional separation, a structural remedy aimed to separate the incumbent's wholesale and retail operations, better aligns supply-side incentives with those of consumers. We characterize market coverage and social welfare implications, examine the robustness of our results and discuss the incentives for discrimination that may arise with the deployment of next-generation networks. Our analysis and the empirical evidence from the UK market suggest that structural remedies exhibit good properties for open access regimes.

Direct link