The idea of providing differentiated network services to customers with different usage profiles is gaining strong commercial support. Many companies are providing solutions to policy-based management for desired Quality of Service (QoS). These network-based services range from application level control to bandwidth management. However, the onus of deciding or designing these policies rests on network managers. The view taken by most solution providers is that of a static organization where requirements can be assessed in advance. Consequently, at the user end overprovision of services has become a norm. This proposal investigates the economics of differentiated network service arrangements in business environments and the design and development of realistic network environments to conduct experiments and evaluate performance of a wide class of pricing approaches. The class of approaches we propose to investigate range from purely deterministically guaranteed QoS through contracts to real-time pricing that provides consumers with option to buy a QoS for each application at the time of consumption. To test the various pricing schemes, the token-bucket paradigm will be used. It will simulate a given price/contract arrangement. The framework for economic experiments developed by this project will provide a benchmark tool to assess network efficiency under different contractual arrangements.