Revenue Management in the Airline Industry - Thesis Example

?Revenue Management in the Airline Industry I. Introduction It has always been business instinct to increase profitability with the utmost efficiencylike in the way seasonal products are sold at higher prices when demand is at peak and at lower prices when there appears to be no demand for it at all. The basic idea underpinning revenue management therefore is not entirely new in the practice when it was coined three decades ago, strictly speaking. In the field of operations research, RM only evolved from a theory. It is itself a practice, a systems- and methods-based approach that optimizes revenues and attempts to reduce cost. American Airlines stated that revenue management aims to “maximize passenger revenue by selling the right seats to the right customers at the right time” (qtd. in Yu 69). However, not only did the RM concept spread like wildfire in the airline industry, it also garnered adequate reputation in other industries. Revenue management as it is applied can entirely aid in pricing decisions and optimize revenues. II. Historical Brief The law of supply and demand has always been institutional in the business literature and practice. Even as commerce started to flourish, merchants already had to make demand-management decisions specifically in terms of structure, pricing, and quantity in the hope of maximizing profit and avoiding loss (Talluri and van Ryzin 4). However the old idea of RM as businesspeople perceive it three decades ago until now is different in the sense that revenue management focuses on the way decisions are made through a technology-based system (not theoretical therefore) which should be more responsive to the uncontrollable and hardly predictable variables and constraints in a certain industry (Talluri and van Ryzin 4). The airline services sector was the first to employ the principle of revenue management. The efficiency of reservation control systems was based on quantitative researches which centered on “controlled overbooking” (McGill and van Ryzin 233). Overbooking depended on the probability of the number of passengers who shows up during boarding time (McGill and van Ryzin 233); and which is technically necessary in effort to replenish the could-be lost in revenues in case of cancellations or no-shows among passengers (Belobaba et al. 93). In an industry with low marginal costs, fixed capacity, perishable product, irregular demand, and varied market segments such as the aviation industry, excess inventory may be minimized by forecasting through historical data in order to maximize revenues (Dunne and Lusch 42). Airlines during the 1970s started offering restricted discount rates where passengers in the same aircraft compartments have actually paid different prices (McGill and van Ryzin 234). Prices were offered at a different range at predetermined periods to different market segments without having to compromise the level of travelling experience. Therein, comes the groundwork for yield management which was later called, revenue management. This principle was first grasped by Kenneth Littlewood of British Overseas Airways Corporation (BOAC, known today as the British Airways) in his mathematical proposition that in adopting discount schemes, the value they yield should exceed the expected return of future full fare bookings (qtd. in McGill and van Ryzin 233). (BOAC was offering discount rates for customers who reserved for seats twenty-one days before the actual time of flight). In the United States, the American Airlines adopted the same scheme through its Super Saver Fares in 1977 which would later be encapsulated in the RM framework especially after the passage of the Airline Deregulation Act of 1978 that paved the way for the wide practice of such principle in the modern business context (Hall 600; McGill and van Ryzin 234). Before the deregulation, US airlines were controlled by the Civil Aeronautics Board (CAB). The pricing schemes underwent dramatic change after the deregulation where new low-cost carriers threatened major carriers. The deregulation allowed airlines to alter their price offerings and schedules, and expand their services without legal restrictions. On the other hand since the legislation, competition also stiffened in the airline industry. PeopleExpress, a new low-cost carrier, operated with minimal costs and offered fares which were 50-70% less than the major competing airlines’ (Hall 600). Migration of “price-sensitive discretionary travellers” from high-cost to low-cost carriers accounted for the cumulative losses obtained by major airlines despite the advantages as established carriers (i.e. wide offers of services and coverage) (Hall 600). American solved the problem through imposing purchase restrictions and capacity-controlled fares. This prevented business travellers to purchase new low fares in a way that they had to purchase nonrefundable discounts 30 days before the planned departure; and a seven-day minimum stay was likewise required to avail of the promotion (Hall 600). Data-processing tools served to control seat inventory manually (Smith et al. 24). However, flights involved different sets of variables which eventually lead to the development of the Dynamic Inventory Allocation and Maintenance Optimizer (DINAMO), a decision support tool that produced benefits on overbooking, productivity, revenue mix, and pricing flexibility (Smith et al. 24). PeopleExpress lost its head in the game as American Airlines recovered through DINAMO. PeopleExpress announced bankruptcy from $160 million in losses from 1984 to 1986 (Talluri and van Ryzin 9). The company’s CEO admitted that losses were due to the lack of a prompt development of an automated revenue management which American Airlines developed and used effectively in the latter’s markets (Talluri and van Ryzin 9). III. Characteristics of the Airline Industry: RM Applicability The concept of revenue management was eventually translated by other sectors in the travel and tourism industry. Dunne and Lusch (42) describe five major characteristics where the principle of revenue management can be applicable. Different Market Segments: In all types of industries, market segmentation is common since customers diverge in wants and needs. In the airline industry, there are a number of variables such as the origin and culture of the passenger, and purpose, and length of the travel (Shaw 24). An effective market segmentation through revenue management prevents cross-selling among the segments (Dunne and Lusch 42). Since there is no significant difference in the quality of services offered, price discrimination is thus elaborate. Although this type of discrimination does not come single-handedly from consumer demographics but rather through product differentiation (business and leisure). Therefore in terms of pricing, customers will have to pay varying prices depending on their corresponding willingness to pay and need (Belobaba et al. 79). Fixed Capacity: The number of passengers that an aircraft can carry is highly dependent on the number of seats available like hotels are in their number of rooms as to the maximum quantity of guests they could lodge in. In industries alike, this typically accounts for the high proportion of fixed costs. in the travel and tourism industry. A fixed capacity compels airlines to manage their capacity in relation to the demand to prevent losing when revenues could be maximized (Dunne and Lusch 42). Fluctuating Demand: Demand in service industries vary irregularly either for a short-term or long-term basis or a combination of both. Demand in the airline industry can be determined in the number of passengers that appeared for boarding and the number of walk-in passengers who cannot be boarded due to the lack of capacity (in a technical term called spill or rejected demand) (Belobaba et al. 48). Aside from seasonal factors, the fixed capacity of airlines itself affect pricing decisions and demand (Dunne and Lusch 42). Because airlines could not sacrifice customer service such as in cancelling flights in case of many unsold seats, demand and supply will have to be forecasted effectively otherwise, it would result in overcapacity or undercapacity (Ben-Yosef 119). Low Marginal Cost: Airlines usually incur high fixed costs and low variable costs. Where a schedule is already fixed, selling an empty seat entails a low marginal cost since, after all, the costs of fuelling the engine or cleaning the aircraft relative to carrying another passenger load is almost non-existent (Holloway 165; Dune and Lusch 42). In this regard, the production of one more unit does not significantly affect the aggregate costs. Airlines would only have to maximize revenue therefore, revenue management is an essential tool (Dune and Lusch 42). Perishable Inventory: If a seat remains unsold, the possible revenue it could have generated for that specific schedule has already perished. It cannot be transferred to another flight schedule. In determining the maximum potential, there is a need to refer to the historic statistical data for an accurate forecast. Potential losses from unused inventory will easily be minimized through revenue management schemes (Dunne and Lusch 42). IV. Revenue Management Techniques/ Methods a. Overbooking Overbooking is the method to offset potential revenues lost due to cancellations and no-shows. Overbooking is more of a need in airlines than it is tactical. Apparently, it should have a standard limit, an optimum, since extensive overbooking would significantly compromise the reputation of the airline (Vasigh et al. 303; Wang and Bowie n. pag.). RM will determine this limit that would require historical patterns regarding cancellations, go-shows, and no-shows. b. Discount allocation Discount allocation is an RM method that solves the problem of the number of discounted fares that should be offered in order to invite demand for a specific flight; and protect for full fares (Smith et al. 9). Discount allocation assumes that it would be unlikely to reduce the number of unoccupied seats by offering full fares alone. c. Origin-destination control/ Traffic management OD control is a reservations control process based on the origin and destination of the travel for a product mix that provides optimum revenues. In essence, this is the way to manage the capacities of a set of connecting flights where demand is under consideration (Smith et al. 10). V. Mathematical Models The mathematical models for capacity allocation to optimize revenues can be categorized into two: the static and dynamic models. Static models presume that demands for different fare classes come in non-overlapping periods whereas dynamic models surmise that demands arrive at no particular order (Barz 63). In this section, only the static models of which EMSR algorithm, the most notable mathematical model commonly used in airlines is one, are discussed. a. Littlewood’s Rule: Optimization In order to discuss EMSR, there is a need to discuss Kenneth Littlewood’s rule. Littlewood’s rule is only applicable to a single flight leg with two classes of fare products such that P1>P2. With capacity c given there are no cancellations and overbooking, the probability is denoted in P(.). Low-paying passengers book their flights before high-paying passengers would so that the Littlewood’s rule relies heavily on the demand distribution of P1, D1. The problem is in determining the acceptable demand for class 2 before determining the demand for class 1, X1 (Talluri and van Ryzin 35). An acceptable demand for P2, wherein the number of seats protected for high-paying passengers is kept at an optimum level x, should satisfy P2 ? P1Prob(X1 ? x) If in case class 2 offers are all consumed and a demand for the remaining n units of capacity appears from a class 2 customer, satisfying this demand would equate to revenues; rejecting it would entail selling up n units at class 1 value if and only if the demand for class 1 is greater than or equal to n units (X1 ? n). The expected marginal revenue, that is, the potential of reserving yth unit for class 1 is P1 Prob(X1 ? n), hence it is only reasonable to accept X1 if and only if P2 ? P1 Prob(X1 ? n). The idea here is that as long as the potential revenues generated from class 2 is greater than or equal to the potential of class 1, accepting the demand for class 2 results in optimal value (Talluri and van Ryzin 35). Furthermore, class 2 demand should be accepted if the remaining capacity is more than the optimal protection level and otherwise, if equal to or less. The optimal protection level y should satisfy: P2 ? P1 Prob(X1 ? y) and P2 ? P1 Prob(X1 ? y + 1). Where a continuous demand distribution of F1(x) is used, y can be translated in Littlewood’s rule: P2 = P1 Prob(X1 > y) or y = F1 -1 (1 - P2 /P1). Therefore, booking limit for class 2 would be the difference between the actual capacity and the optimal protection level b = c - y, where bid prices can be determined at (x) = P1 Prob(X1 >n) (Talluri and van Ryzin 36). b. Expected Marginal Seat Revenue: Heuristics EMSR or the expected marginal seat revenue is the widely used mathematical model in airline revenue management systems. It resides on the concept of the Littlewood’s rule that is applied to successive pairings as an approach to a multi-class capacity allocation problem (Phillips 161). In a multi-class case, EMSR will determine the booking limits of each fare class relative to another that is if, they have the same inventory of seats (Belobaba et al. 98); and as such it can be simply interpreted as the “expected gain from adding another seat” (Vasigh et al. 196). The expected marginal revenue of protecting Sth seat can be determined in: EMSR1 = f1 P(S1) where f1 is the average revenue for a specific class i, and P(S1) is the probability that there will more demand for fare class i than what its protection level, S1 imposes. Throughout 1987 to 1992, Belobaba introduced EMSR-a and EMSR-b. It is said that EMSR-b produces results that enable airlines to obtain better optimal revenues than can EMSR-a, although EMSR-a is more well-documented than the other (Talluri and van Ryzin 45). However, there are several conflicting accounts to this end and some reports that not any of the either outperforms the other (Talluri and van Ryzin 48). b.i EMSR-a EMSR-a aggregates the protection levels of a specific class against other classes. Using Littlewood’s rule, the protection level y for a higher class P3 against a specific class P1 is determined in (Phillips 161): y31 = F1 -1 [(P1 - P3 ) /P1] while determining the protection level for class P3 against lower class P2 is: y32 = F2 -1 [(P2 - P3 ) /P2] The overall protection level of class P3 is simply the sum of the two protection levels (Phillips 161). Where number of fare classes j ? 2, the protection level yj for a specific class Pj is yj = Fi -1 [(Pi - Pj) /Pi] In a normal distribution, the booking limit for class j is defined as: yj = + ? -1 [(Pi - Pj) /Pi] EMSR-a however disregards the pooling effect from summing up the demand across classes thereby producing unreliable results, in fact far from what is optimal (Talluri and van Ryzin 46). This makes EMSR-a faulty in a very important way however intuitive it may appear. b.ii. EMSR-b The inventory structure of available seats for future flights is important on the way booking limits should be calculated using the EMSR-b model since it was developed for nested inventory control systems in the first place (Belobaba et al. 98). EMSR-b allows airlines to “serially nest their booking classes” with underlying premises that 1.) there is a separate and independent demand for each fare class; each fare class has a stochastic demand that usually exemplify a normal probability distribution; lower class seats are usually booked first than the higher class seats; and/or etc. (Belobaba et al. 99). Capacity allocation does not concern itself with allocating seats to different fare classes but rather in determining the protection level for higher fare classes and nested booking limits for lower fare classes (Belobaba et al. 98). What EMSR-a lacks is what EMSR-b possesses. EMSR-b does not ignore the pooling effect or the statistical averaging effect in aggregating. EMSR-a determines the protection level whereas EMSR-b focuses on the demand from future fare classes which will be regarded as one ‘artificial class’ where it is assumed that demands from this class are equal to the aggregate future demands while the expected gain is equal to the weighted-average revenue (Talluri and van Ryzin 47; Barz 77). P(Sj > yj) = (Pj + 1) / (Pj) where in determining the protection level yj in stage j + 1, the collective future demand for artificial class j, j-1,...,1 is Sj = Dk while the weighted-average revenue is j = {Pk E[Dk]} / {E[Dk]}. With the assumption that the demand for each class j is normally distributed and independent with mean j and variance j2 (indicating uncertainty), the protection level is defined as yj = + z? where the mean is = k while the variance of the total demand is 2 = k2 and the inverse of the normal distribution is z? = -1 (1- pj + 1/ j ) EMSR-b may not be entirely comprehensive however with the fact that EMSR-b assumes the equivalence between the expected revenue of a future accepted request and the expected revenue of all future requests. Protection levels are also deemed necessary in what comprises an optimum and the weighted-average may create an impact to these outcomes (Talluri and van Ryzin 48). VI. The Impacts of Revenue Management The success of revenue management as an approach is already evident in the way it gained popularity from the aviation industry to other segments in the travel industry, and finally to other types of industries such as the manufacturing, hospitality, and tourism industries that are characterized by five attributes presented in Section III. Although, its infallible flexibility to other industries is still yet to be quantified and qualified. In the airline industry, there can be no doubt as to its effectiveness as an appropriate measure where an optimum product mix depends on reliable and appropriate forecasting systems and methods, but for the sake of substantiating the claim, statistical evidences will be presented. a. Economic Impacts The case for American Airlines draws a concrete example as to the actual impacts of RM (as presented in Table 1). After the innovating and installing RM, American Airlines accounted $1.4 billion in quantifiable benefits within the late 1980s and early 1990s; and expected $500 million in revenues annually for the succeeding years (With the use of the revenue opportunity model, Table 1 presents no significant pattern changes in the revenue opportunity earned by overbooking.) (Smith et al. 27). Delta Airlines employed the same practice on revenue management and accounted $300 million change in revenues annually. Wilson studies the economic impacts of RM. He studies the revenue changes as a result of the application of revenue management in two carriers (an innovator and a laggard) serving on a first-come, first-served basis (91). The innovator gains basic knowledge about a leg-based EMSR control while the competing laggard follows. Table 2 shows the percentage change in revenues using a six method combinations of the control options of FCFS EMSR-a, and EMSR-b at DF (demand factor, which is the fraction of demand to capacity) values of 0.9 and 1.2 (Wilson 91). The six method combinations approach relates back to the development of RM as an essential operational strategy (Wilson 91). Figures show how the innovator benefits from applying its basic knowledge on RM and improves more upon the application of both versions of EMSR whereas the laggard obtains negative impacts but recovers somehow in the process. Mak performs a simulation study and uses the American Airlines’ revenue opportunity model in studying the different systems figures such as number of classes, capacity constraints in order to determine the revenue impacts (115). EMSR-b works best when demand factors are higher and lesser capacities such that it creates a positive impacts on revenues as the number of fare classes increases (121). At the same time, he concludes that EMSR-a and EMSR-b almost perform equally with regard to achieving revenue opportunity (117). To better understand the advantages of RM application, Talluri and van Ryzin present numerical examples (48). b. Competitive Advantage Since the inception of revenue management as a technological tool in competitive airline markets, airlines experienced increases in revenues while the opposite is experienced by laggards (Belobaba and Wilson 3). The authors assume that new entrants do not significantly change the market structures. They have concluded that the airline that had used RM before its competitors enjoys a competitive advantage in different markets (Belobaba and Wilson 3). This is a replication of the Wilson’s findings in that the laggard experiences negative effects on its revenues. c. Overall Advantages Aside from achieving competitive edge, revenue management had most importantly improved the management efficiency (i.e. in terms pricing schemes and inventory control) in the airline industry in real-time, and integrating decisive measures (which can be tactical and strategical) for sustainability (Phillips 123). More importantly, RM has provided airlines access to different markets in which demand is likewise easily satisfied from processes called market segmentation, product differentiation, determining booking limits, and estimating spills (Phillips 123). VII. Conclusion Revenue management should result in an improved forecasting accuracy, optimum overbooking limit, and efficient traffic management. Achieving this promises better revenue performance. Furthermore, the application of revenue management should not only meet short-term objectives but also most especially the long-term such as in achieving competitive edge and equally, profitability. However the success of the RM application also depends on identifying the different variables under the most intelligent assumptions. On the other hand, although RM has proven its capacity in operations management, its effects in the more intangible aspects of business and which are technically more crucial in the long-term objective have received inadequate attention. Wang and Bowie (n. pag.), for instance, conclude that while revenue management induces operational efficiency, it creates long term negative effects on intangible affects on business-to-business relationships in terms of trust or commitment. Spill can definitely affect the reputation of the airline in the process especially if it goes untreated appropriately in spite of the fact that RM optimizes overbooking limits. In this light, further studies must be performed to determine such effects and mitigate them should they be harmful to the industry. VIII. Appendix Table 1: Economic Impacts of Revenue Management at American Airlines Year Revenue Opportunity Earned (%) Revenue Earned (US $ million) 1988 92 210 1989 93 235 1990 90 225 Note: The revenue opportunity is calculated by taking the difference between the revenues from using EMSR-a and EMSR-b analytical methods and No Control method (Mak 116). Table 2 Symmetric Scenario Revenue Effects (Wilson 91) DF = 0.9 DF = 1.2 Scenario Carrier 1 Carrier 2 Total Carrier 1 Carrier 2 Total EMSRa/FCFS +8.2 -1.4 +3.4 +68.7 -3.1 +32.8 EMSRb/FCFS +10.4 -1.4 +4.5 +68.7 -3.1 +32.8 EMSRa/EMSRa +8.2 +8.4 +8.3 +43.1 +43.2 +43.2 EMSRb/EMSRb +10.3 +11.0 +10.6 +46.5 +47.0 +46.8 EMSRa/EMSRb +7.5 +7.4 +7.4 +42.3 +44.8 +43.6 Works Cited Barz, Christiane. Risk-averse Capacity Control in Revenue Management. New York: Springer Berlin Heidelberg, 2007. Belobaba, Peter, Amedeo Odoni, and Cynthia Barnhart. The Global Airline Industry. United Kingdom: John Wiley and Sons, 2009. Belobaba, Peter, and John Wilson. “Impacts of Yield Management in Competitive Airline Markets.” Journal of Air Transport Management 3.1 (1999): 3-9. Ben-Yosef, Eldad. The Evolution of the US Airline Industry: Theory, Strategy and Policy. Netherlands: Springer, 2005. Dunne, Patrick, and Robert Lusch. Retailing. China: Thomas South-Western, 2008. Hall, Randolph. Handbook of Transportation Science. Kluwer Academic Publishers, 2003. Holloway, Stephen. Straight and Level: Practical Airline Economics. England: Ashgate Publishing Ltd., 2008. McGill, Jeffrey and Garrett van Ryzin. "Revenue Management: Research Overview and Prospects." Transportation Science 33.2 (1999): 233-256. Web. 10 Jan. 2011. . Phillips, Robert. Pricing and Revenue Optimization. USA: Stanford University Press, 2005. Shaw, Stephen. Airline Marketing and Management. England: Ashgate Publishing Ltd., 2007. Smith, Barry, John Leimkuhler, and Ross Darrow. “Yield Management at American Airlines.” The Institute of Management Sciences 22.1 (1992): 8-31. Web. 11 Jan. 2011. . Talluri, Kalyan and Garrett van Ryzin. The Theory and Practice of Revenue Management. USA: Springer Science Business Media, 2004. Talluri, Kalyan, Garrett van Ryzin, Itir Karaesmen, and Gustavo J. Vulcan. “Revenue Management: Models and Methods.” Informs. Informs, 2008. Web. 11 Jan. 2011. . Vasigh, Bijan, Tom Tacker, and Ken Fleming. Introduction to Air Transport Economics: From Theory to Applications. England: Ashgate Publishing Ltd., 2008. Wang, Xuan Lorna, and David Bowie. “Revenue management: the impact on business-to-business relationships.” Journal of Services Marketing 23.1 (2009): 31-41. Web. 14 Jan. 2011. . Wilson, John. The Value of Revenue Management Innovation in a Competitive Airline Industry. MS Thesis. Massachusetts Institute of Technology, Massachusetts, 1995. Print. Yu, Gang. Operation Research in the Airline Industry. USA: Kluwer Academic Publishers, 1998. Read More