The rst line of work described above focuses on modeling trade and its associated
frictions, assuming simple trading networks (often with a single middleman). On the other
hand, the second line of work focuses on the impact of network structure but does not
account for trade frictions. The interaction of both these eects is not well understood. In
a more complex network, the search problem facing an agent will depend on her location in
the network, and the presence of such frictions will naturally given rise to dierent equilibria
than in models such as Blume et al. (2009).
This paper provides a starting point to bridge this gap: as in Blume et al. (2009) we
consider a trading network connecting consumers to producers, but as in Rubinstein and
Wolinsky (1987), agents randomly meet over time and engage in non-cooperative bargaining
protocols. Thus, our paper provides a general framework that incorporates three important
features of markets. First is the underlying network structure: not all pairs of agents can
interact in the market. The second is the non-cooperative bargaining setting: no agents have
the power to set prices, the prices are formed through a negotiation process. Finally, the
third is the search cost: agents discount their payo if they do not nd a proper trading
partner or fail to negotiate. The possibility of not nding a proper trading partner is an
important additional search cost in our model.