Revenue Management at American Airlines
Airlines face highly cyclical demand; American reported profitability in the strong expansion of 2006–2007 but massive losses in the severe recession of 2008–2009. Demand also fluctuates day to day. One of the ways American copes with random demand is through marginal analysis using revenue management techniques. Revenue or “yield” management (RM) is an integrated demand-management, order-booking, and capacity-planning process.
To win orders in a service industry without slashing prices requires that companies create perceived value for segmented classes of customers. Business travelers on airlines, for example, will pay substantial premiums for last-minute responsiveness to their flight change requests. Other business travelers demand exceptional delivery reliability and on- time performance. In contrast, most vacation excursion travelers want commodity-like service at rock-bottom prices. Although only 15 to 20 percent of most airlines’ seats are in the business segment, 65 to 75 percent of the profit contribution on a typical flight comes from this group.
The management problem is that airline capacity must be planned and allocated well in advance of customer arrivals, often before demand is fully known, yet unsold inventory perishes at the moment of departure. This same issue faces hospitals, consulting firms, TV stations, and printing businesses, all of whom must acquire and schedule capacity before the demands for elective surgeries, a crisis management team, TV ads, or the next week’s press run are fully known.
One approach to minimizing unsold inventory and yet capturing all last-minute highprofit business is to auction off capacity to the highest bidder. The auction for freewheeling electricity works just that way: power companies bid at quarter ‘til the hour for excess supplies that other utilities agree to deliver on the hour. However, in airlines, prices cannot be adjusted quickly as the moment of departure approaches. Instead, revenue managers employ large historical databases to predict segmented customer demand in light of current arrivals on the reservation system. They then analyze the expected marginal profit from holding in reserve another seat in business class in anticipation of additional “last-minute” demand and compare that seat by seat to the alternative expected marginal profit from accepting one more advance reservation request from a discount traveler.
Suppose on the 9:00 a.m. Dallas to Chicago flight next Monday, 63 of American’s 170 seats have been “protected” for first class, business class, and full coach fares but only 50 have been sold; the remaining 107 seats have been authorized for sale at a discount. Three days before departure, another advance reservation request arrives in the discount class, which is presently full. Should American reallocate capacity and take on the new discount passenger? The answer depends on the marginal profit from each class and the predicted probability of excess demand (beyond 63 seats) next Monday in the business classes.
If the $721 full coach fare has a $500 marginal profit and the $155 discount fare has a $100 marginal profit, the seat in question should not be reallocated from business to discount customers unless the probability of “stocking out” in business is less than 0.20 (accounting for the likely incidence of cancellations and no-shows). Therefore, if the probability of stocking out is 0.25, the expected marginal profit from holding an empty seat for another potential business customer is $125, whereas the marginal profit from selling that seat to the discount customer is only $100 with certainty. Even a pay-in-advance no-refund seat request from the discount class should be refused. Every company has some viable orders that should be refused because additional capacity held in reserve for the anticipated arrival of higher profit customers is not “idle capacity” but rather a predictable revenue opportunity waiting to happen.