Bremen Computational Logistics Group Working Papers 3
A coalition of freight carriers is considered which has to decide how to allocate a pool of transport requests among its members. The literature is aware of a number of solution approaches which usually assume truthful behavior of the freight carriers. However, the used negotiation protocols are mostly not proven to enforce truthful behavior. This paper gives some insights into the impact of non-truthful behavior via computational experiments. We solve the collaborative problem via a genetic algorithm (GA) which is operated by an auctioneer. The GA’s individu- als are allocations of requests to carriers. To calculate the fitness of an individual, the carriers bid on the allocations. Bidding below a carrier’s true valuation could ceteris paribus increase its profits. However, understated valuations can influence the search process negatively, in particular when a favoured allocation is dismissed wrongly. It is shown via computational experiments that for six tested instances, bidding non-truthfully is individually, but not collectively, rational and results in a kind of prisoner’s dilemma.