Decomposing the Congestion Effect and the Cross-Platform Effect inTwo-Sided Networks: A Field Experiment

This paper highlights how the provision of information about userparticipation can serve as a strategic marketing tool for firms seekingto grow two-sided exchange networks. A two-sided exchange network is abusiness model (such as Ebay or Craiglist) where revenue is generatedfrom persuading people to buy and sell items through that particularexchange. It is not immediately clear whether broadcasting informationabout the number of sellers will grow further seller participation. Onthe one hand, a strong rival presence may dissipate payoff (a'congestion effect'). On the other hand, a large number of rivals maysignal high buyer demand (a 'cross-platform effect'). We use fieldexperiment data from a B2B web site that brings together buyers andsellers of used equipment and real estate. Before each seller made aposting request, the web site randomized whether to disclose the numberof buyers and/or sellers, and the exact number to disclose. We find thatwhen presented together with the number of buyers, a larger number ofsellers makes sellers less likely to list their products, indicating anegative congestion effect. However, when the number of sellers ispresented in isolation, its negative impact on entry is significantlyreduced, indicating a positive cross-platform effect. Higher buyersearch intensity amplifies the moderating role of demand uncertainty.The results suggest that information on the number of users can be aneffective tool to grow two-sided networks but should be usedstrategically. A network can attract more users by advertising densecompetition when demand is not transparent, especially insearch-intensive markets.