Event Details

Dr Vikram Dokka will present a seminar to the Management Science Department

We consider pricing problem faced by a seller who sells a subset of products in a network of products. The prices of the products not sold by seller are random, with unknown independent distributions of which only partial information is available. Within the setting of ambiguity about price distributions, we first study adversarial buyer approach to pricing for a revenue maximizing seller. We develop a novel approach to pricing under ambiguity, referred to as selfish buyer approach, which can be seen as a relaxation of adversarial approach. We study efficiency of both adversarial and selfish models and show that selfish approach is less conservative against increasing valuation uncertainty. We develop the notion of semi-selfish model which unifies adversarial and selfish models into one framework. Using this, for general acyclic networks, we develop efficient algorithms to solve the resulting models. Using numerical experiments, we give insights into price performance of the models in path pricing setting where buyers choose paths after seller announces his/her link prices. Finally, we extend the adversarial and semi-selfish methodologies to pricing matchings with an application