Bargaining power and value sharing in distribution network
In 2014, Roson and Hubert published an interesting paper on Networks and Spatial Economic journal titled “Bargaining power and value sharing in distribution network: a cooperative game theory approach” applies the basic ideas of trading in a network, bargaining power of individual nodes, and game theory as discussed in the class in a sophisticated way. This paper uses networks to represent markets, where nodes are agents with different bargaining power who are distributed economic surplus obtained in the market. Agents can also cooperate, grouped into coalitions. For each coalition and its associated networks, a network market equilibrium (NME) is found by means of an optimization problem, taking the link flows into consideration, with a goal of maximizing the total net welfare obtained from a network, subject to an equilibrium situation that flows are equal in both directions of transaction. For the surplus allocation, the authors use the cooperative game theory, where the equilibrium is reached at the Shapley value assigned to each agent a payoff which is proportional to her “contribution” in all possible forming coalitions, thus the agents’ bargaining power can be obtained based on the Shapley value.
The authors then give an example of a fictitious network consisting of five agents. Except for 1) the Shapley value is calculated in a seemingly complicated way (with equation 3 in the article), and 2) a supply-demand curve is embedded to each agent, where for an individual agent (which could also be a single market) it has a demand for a good, and the supply is provided by other agents in the same network and the supply curve is drawn in a stepwise manner, the method of total welfare distribution, which depends on the relative bargaining power, is the same as what has been discussed in the in the class.
This paper presents to me a perfectly relevant example of model construction on welfare distribution according to bargaining power on a market network, and broadens my horizon in the application and extension of the ideas discussed in the class. However, as the authors themselves state, their model assumes a static network, the flexible and modifiable network structure is still to be explored. After all, this is also a complex topic left undiscussed in the class.
Link: http://link.springer.com/article/10.1007/s11067-014-9270-6