Most service departments have occasionally idle capacity and occasionally saturated capacity. The reason for these phenomena is that a service department faces a random demand for the service it renders. How shall its transfer pricing be influenced by this randomness? The paper discusses this question by analyzing an abstract model of a service department. The model depicts a service department as a M/D/l queieng system with an upper limit UL on the time in the process. All requests for service take the same amount of time. The net benefit (defined as the benefit minus any direct costs) of satisfying a request is uniformly distributed and is known at the time the request materializes. If the net benefit is larger than the net transfer price (defined as the transfer price minus any direct costs) and the delivery time does not exceed UL then the request is submitted to the service department. The paper develops two algorithms to find the optimal transfer pricing and the best simple transfer price, respectively. Numerical computations indicate that the expected net benefits earned per unit time by employing the solutions found by the two algorithms differ by at most 3%. They also investigate the relationship between the best simple net transfer price and the cost of service capacity.
