Dynamic Pricing Part 2: Price Discrimination and Revenue Management

As we noted in our first dynamic pricing entry, the basic tools for implementing revenue management originated in the airline industry.  In today’s installment, the goal is to provide some insights into how revenue management works in the airline industry and to begin to consider how these tools apply (or don’t apply) to ticket pricing.

As a starting point, it is necessary to specify the revenue management goal.  In the airline industry the most basic goal is to maximize the revenue collected from a given flight (the network structure of airlines can often motivate more complicated goals).  In the case of sports, or performing arts, the corresponding goal is to maximize the revenue produced by a given event.  More generally, we can think of the goal of revenue management is to maximize the value of a firm’s inventory.  Often a distinction is made that revenue management is especially useful when inventory is perishable.  The key point is that if an airliner takes off with an open seat or a stadium has empty seats at game time, that the firm has forever lost that unit of inventory.

The two key ingredients needed to implement revenue management are a system for segmenting customers and forecasts of segment level demand.  In the airline industry, a very basic system of segmentation might be to group customers into a business and leisure segments.  These segments are thought to differ in terms of traits such as price sensitivity and flexibility.  Specifically, business travelers are thought to be willing to pay higher prices and to have more restrictive schedules.  The second necessary input is segment level demand forecasts.  There are many ways of forecasting demand ranging from complex statistical models to simple heuristics but the salient point is that the revenue manager needs to be able to accurately forecast how many customers from each segment will want to travel on a given flight or attend a specific game.

A core concept for implementing revenue management is the idea of “expected marginal revenue.”  For example, let us assume that we are trying to manage the revenue produced by a flight with 100 seats. Also we have a business segment willing to pay a high price that tends to book close to the departure time, and a price sensitive leisure segment that books long in advance.  The revenue manager’s job is to decide how many low priced seats to sell to the leisure segment or, differently stated, to decide how many seats should be reserved for the late arriving business segment.  One way to do this is to reserve just enough seats such that the expected marginal revenue from saving a seat for a business customer is equal to the expected marginal revenue from allowing the seat to be purchased by a leisure customer.

Let us further assume that the business customer pays $1,000 per ticket while the leisure traveler pays $100.  The revenue manager’s decision rule would be to reserve sufficient tickets for the business segment such that the expected marginal revenue produced by the last seat allocated to the business customers is the same as if the seat were sold to the leisure customers.  Returning to our example, if we can sell unlimited $100 seats the revenue manager would reserve enough seats so that there is at least a 10% chance to sell a $1,000 business ticket.

In many ways we could make the preceding discussion sports specific simply by changing “flight” to “game” and by using a segmentation system that is more sports specific (casual versus hardcore fans?).  But as you might expect, when we get to actual implementation it often becomes difficult to directly transfer travel industry yield management techniques to a sports context.  In our next entry we will discuss several of these challenges.

Mike Lewis and Manish Tripathi, Emory University 2013.