We consider a dynamic pricing setting where a firm sells a perishable product over a finite selling horizon. Different from the standard setting, the firm faces multiunit demand and can separately quote a price for every batch size. Customers differ in their attraction to the product and their preference regarding the batch size. These two attributes are depicted by a random variable each and are the basis for the calculation of customers’ willingness-to-pay. The resulting customer choice model is very challenging to work with and we put some effort into reducing its complexity. We develop optimality conditions for the stage-wise optimization problem. As finding the optimal solution in every state is non-trivial, we resort to formulating a fluid approximation model. With a simplifying assumption, we can solve this approximation and subsequently verify that this assumption indeed holds for the optimal solution. The resulting static pricing policy is approximately optimal in our dynamic setting. However, instead of applying this static policy, we use it to ensure approximate optimality of the three novel heuristics we developed in this paper. We test all heuristics in a simulation study against an upper bound and analyze patterns in the corresponding policies to gain managerial insights. For example, we find that a piecewise linear pricing structure performs very well and might be an easy-to-communicate alternative to full nonlinearity.
 

Dynamic Pricing, enabled by technological developments, is gaining more importance in fields beyond the airline industry, including retail, where neglecting multiunit demand leads to suboptimal prices and lost revenues. In these fields, nonlinear pricing is a common static pricing strategy that explicitly takes multiunit demand into account but lacks the possibility to dynamically adapt prices. In this paper, we bring the strengths of both pricing strategies together by combining them to multiunit dynamic pricing. We formulate the corresponding stage-wise optimization problem. To account for customers' preferences regarding batch size, we adapt an adequate customer choice model based on (random) willingness-to-pay. The willingness-to-pay is defined by a combination of customer’s attraction to and consumption of the product. These two aspects of customers’ preferences are private information, but the firm may have (partial) access to the information of the current customer. The firm is monopolistic and can price-discriminate between different order sizes by quoting nonlinear batch prices. This work investigates three cases of what information is observable: attraction to the product, consumption of the product, or both. We solve the resulting optimization model analytically and derive closed-form expressions of the optimal solution in two of the cases. Moreover, we proof the desirable monotonicity in time and capacity is still intact. Building on this monotonicity, we show dynamics of the optimal pricing policy. Finally, we examine the value of information in a numerical study to gain managerial insights regarding the importance of knowing customers’ preferences.