Impacts Of Entry In Airline Markets Effects Of Revenue Management On Traditional Measures Of Airline Performance

ARTICLE IN PRESS Journal of Air Transport Management 10 (2004) 259–270 Impacts of entry in airline markets: effects of revenue management on traditional measures of airline performance Thomas Gorin*, Peter Belobaba International Center for Air Transportation, Massachusetts Institute of Technology, Cambridge, MA 02139, USA Abstract Assessment of unfair competitive practices in airline markets is typically based on the analysis of changes in aggregate measures of airline performance, such as average fares, traffic and revenues Simulation results show that these measures can be greatly affected by the competitive revenue management situation For example, average fares on the incumbent carrier can either increase or decrease following entry by a new competitor, depending on whether one or both airlines perform revenue management Consequently, these measures on their own not constitute a reliable indication of the response of incumbent carriers, and provide even less information on their strategic intent, which is critical in identifying predation r 2004 Elsevier Ltd All rights reserved Keywords: Revenue management; Low-fare airline entry; Airline pricing Introduction The more profitable days of the airline industry in the late 1990s raised numerous questions regarding the potential for unfair competition and predatory pricing In the US, American Airlines was criticized by a number of its low-fare competitors (often referred to as low-cost carriers or LCCs), and sued by the US Department of Justice (United States of America vs American Airlines, 2001), while the Competition Tribunal in Canada attempted to determine whether Air Canada competed unfairly against CanJet (Competition Tribunal, 2000) and WestJet (Competition Tribunal, 2001) A large number of studies focused on market changes brought about by low-cost carriers and described their effect on fares and traffic In 1998, the United States Department of Transportation (US Department of Transportation, 1998) proposed a policy attempting to identify predatory practices based on high-level market measures Other studies of the potential for predatory pricing in airline markets include the work of Dodgson et al (1991) for the European Commission *Corresponding author Email-addresses: gorint9@mit.edu (T Gorin), belobaba@mit.edu (P Belobaba) 0969-6997/$ - see front matter r 2004 Elsevier Ltd All rights reserved doi:10.1016/j.jairtraman.2004.03.002 Here, the effect of airline revenue management on traditional measures of airline market performance and its importance in understanding the dynamics of airline markets, in particular with respect to competition issues is explored It is demonstrated through simulation of a single market scenario, that even when the incumbent carriers not respond to entry other than by matching the lowest available fare in the market, aggregate measures (such as average fares, revenues, loads and market share) change with the competitive revenue management situation and the entrant’s capacity As a result, such measures not constitute a reliable indication of the response of incumbent carriers, and provide even less information on the strategic intent of the incumbent carriers, which is critical in identifying predation Simulation confirms previous findings that under a realistic representation of competition, average market fares generally decrease following entry, while traffic increases For the nonstop incumbent carrier, however, we show that average fares, traffic and revenues behave very differently as a function of the competitive revenue management situation (that is, whether the incumbents and/or the low-fare new entrant perform revenue management), as well as new entrant capacity The results indicate that even airline-specific average fares, traffic and revenues provide a very incomplete picture of the effects of entry in a market, contrary to the suggestions of previous researchers ARTICLE IN PRESS 260 T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 Literature review In 1958, McGee (1958) argued that price predation was often an unprofitable business strategy and concluded that it would be unlikely to occur except under unusual market conditions, such as legal barriers to mergers and acquisitions Areeda and Turner (1975, 1976, 1978) claimed that pricing at or above average total cost could not be considered predatory and designed a test of predatory behavior, based on marginal cost In 1977, Williamson (1977) suggested a short-term output-maximizing rule as an alternative to Areeda and Turner’s marginal cost test Baumol (1979), Joskow and Klevorick (1979), and others also discussed predatory pricing in its more general economic setting and proposed tests or rules for evaluating whether a pricing strategy is predatory Most of the research on predation thus focuses on the comparison of revenues and costs, and the discussion among authors centers on the correct way to measure the appropriateness of a pricing strategy by an incumbent Research on entry into airline markets has focused mostly on the effects of entry on traffic and fares While many of these studies indicated a growing concern with respect to unfair competition and predatory pricing, few of these research efforts focused on identifying and understanding the dynamics of airline markets, and how they affect competition For instance, Bailey et al (1985), Morrison and Winston (1990), Windle and Dresner (1995), Perry (1995), and Oster and Strong (2001) all examined the impact of entry on average fares and traffic, distinguishing between entry by a low-fare carrier and a network carrier, and touch upon the issue of predation Perry, in particular, concludes that the typical impact of entry, at the local market level, is an increase in total traffic, a decrease in average market fare, and an increase in total market revenues For the incumbent carriers, this translates into a decrease in average fare, a decrease in its revenues (as the LCC increases its own revenues) and an increase in its local traffic Dodgson et al (1991) provide a definition of predatory practices in the airline industry and concepts of relevance in identifying these practices In addition, they highlight the irrelevance of cost-based tests of predation in airline markets Baumol (1982), Baumol et al (1983), Bailey and Baumol (1984), Hurdle et al (1989), Whinston and Collins (1992) focus on the contestability of airline markets and how it might affect competition in the airline industry, but conclude that the hypothesis of contestability is inappropriate Thus, past efforts to investigate competitive behavior in airline markets have involved almost exclusively the analysis of aggregate market measures of average fares and traffic to evaluate the response and intent of incumbent carriers Overall, none of these studies have provided a satisfactory method to evaluate the possibility of predation, given the dynamics of airline networks and revenue management More importantly, none of the previous research has attempted to estimate the impact of airline revenue management on incumbent performance after entry Approach The approach used to examine the effects of revenue management and new entrant capacity on airline markets is based on simulation Simulation presents the advantage of allowing for the representation of a dynamic competitive airline market where passenger choice and revenue management controls applied by the airlines are represented Analytical models tend to be limited to static observations that overlook the complexity of airline pricing, scheduling and revenue management processes, not to mention the intricacies of modeling demand, booking behaviors, forecasting, and competitive airline interactions For example, in the case of n flight legs with k fare classes per flight leg, with a booking period of 360 days, revenue management optimization involves in the order of nk optimizations per time period within the booking horizon This does not account for multiple frequencies within a market In short, even in a single market case, the number of steps required in an analytical model quickly makes it intractable In addition, the simplifications required to manage such analytical representations inevitably lead to inconclusive findings Rather than oversimplifying, the Passenger Origin Destination Simulator (PODS), a simulator of a competitive airline network, is used.1 To illustrate the impacts of entry, a single market network is simulated, with a set of competing airlines offering service in this market Initially there are two incumbent carriers, Airline offering nonstop service, and Airline offering connecting service The new entrant carrier (Airline 3) then comes in with a schedule identical to that of the nonstop incumbent carrier and with different aircraft capacity levels In the simulations, it is assumed that the market does not structurally change after entry For example, we assume that conditional passenger preference towards any particular airline remains unchanged by entry: Given that the passenger does not choose to travel on Airline 3, his/her preference between airlines and is Abundant literature is available on PODS, including a detailed description of the underlying algorithms (Hopperstad, 2000), general discussions of the structure of PODS by Belobaba and Wilson (1997) and Lee (1998), an explanation of the forecasting models used in PODS by Zickus (1998) and Skwarek (1996), and a validation of the passenger choice model by Carrier (2003) In all these references, various revenue management methods used by airlines are also described ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 the same as his/her preference when there are only airlines and operating in the market Similarly, we assume that total potential demand remains a function of price, as governed by the existing price–demand curve in the market, irrespective of the number of competitors in the market While these assumptions are not overly restrictive, it may be argued that low-fare entry has a structural effect on the market For tractability reasons, and since there is little evidence of this in the literature, structural change is not modeled In these simulations, airlines operate in a short-haul market (283 miles) with a total maximum potential demand of 165 one-way passengers per day at the current lowest available fare level of $63 Of this passenger demand, 35% is business oriented and the remaining 65% is leisure demand The difference between leisure and business passengers resides in business passengers’ willingness to pay a higher fare, greater sensitivity to fare restrictions, and later booking behavior Fig shows the price–demand curves of leisure and business passengers in the market, and the potential for leisure demand stimulation at fares below the current lowest fare of $63 Two competitive scenarios are simulated to allow comparisons ‘before’ and ‘after’ new entry into an airline market In the base case, two incumbent airlines compete, one of which offers only nonstop service in the market while its competitor offers only connecting service In the second scenario, we add a third carrier—the new entrant—which then also offers nonstop service in this market, and competes with both incumbents but more directly with the nonstop incumbent carrier The purpose of Airline 2—the connecting incumbent carrier—is to act as a ‘relief valve’ for the excess market demand and to allow passengers to have an alternative to the nonstop carrier Airline thus represents all the connecting alternatives available to passengers in a more 200 180 160 Business Leisure Demand 140 120 100 80 60 40 20 $50 $100 $150 $200 Fare $250 $300 Fig Price–demand curves in PODS for the simulated market 261 realistic market As a result, we assume that Airline offers a large capacity relative to demand in this market (identical to that of the nonstop incumbent carrier), even though its connecting flight options (paths) are far less desirable than those of Airline The loads, revenues and overall performance of Airline are therefore not of particular interest in this discussion From here on, we thus refer to the nonstop incumbent simply as the incumbent carrier 3.1 Base case: no entrant competition Without new entrant competition, the market is served by two competing incumbent carriers, each offering three daily departures Airline offers three daily nonstop flights while its competitor offers three connecting flights, each with 30 seats on each flight, for a total of 90 seats per day in the market for each carrier Table summarizes the frequency, capacity and baseline pricing of the incumbent carriers All other characteristics are exactly the same for both airlines There is no passenger preference for either airline, other than the preference induced by path quality (nonstop vs connecting paths) With identical fare levels and restrictions for each fare product, the only difference between the two competitors is therefore the fact that one carrier offers nonstop service while its competitor offers connecting service, as shown in Fig As a result of the connecting service, total travel times (origin to destination, including connecting time) are greater on Airline 2, which affects passenger choice in the simulation The baseline prices for each fare class are set as shown in Table 2, along with the restrictions associated with each individual fare class in this baseline scenario Y class is the unrestricted fare class in the market; while B, M and Q classes are increasingly restricted The more restrictive the fare class in terms of advance purchase requirements and restrictions (roundtrip, Saturday night stay, and nonrefundability requirements), the cheaper the associated fare We refer to this fare structure as the standard fare structure As described in the literature on PODS, these fare settings lead to a higher relative utility of higher fare classes (Y and B) for business passengers, and conversely, a greater relative utility of lower fare classes (M and Q) for leisure passengers Finally, since the purpose is in part to examine the impact of revenue management on ‘traditional’ measures of incumbent performance, we allow the incumbent carriers to either accept requests for seats on a firstcome, first-serve basis (FCFS), or to use Fare Class Revenue Management (FCRM) In the case of FCFS seat request acceptance, passengers book seats in a FCFS manner, and the only controls that enable airlines to differentiate between fare products are advance ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 262 Table Capacity, frequency and pricing overview without entrant competition Carrier Capacity Frequency Pricing Airline Airline 90 seats (3  30) 90 seats (3  30) Three daily flights Three daily flights Four fare classes with four different fare levels Y, B, M and Q (see Table 2) Table Two-tier fare structure details (new entrant carrier) Airline Origin Destination Fare class Fare Airline Airline Y M Hub H2 $135 $53 Restrictions Roundtrip requirement Saturday night stay Nonrefundable Advance purchase No Yes No Yes No No No 14 days Fig Single market network with two competing carriers Table Fare classes, associated fares and restrictions for the standard fare structure in the base case scenario Fare class Y B M Q Fare $261 $135 $92 $63 Restrictions Roundtrip requirement Saturday night stay Nonrefundable Advance purchase No Yes Yes Yes No No Yes Yes No No No Yes No days 14 days 21 days purchase requirements that effectively close down a fare class beyond a given deadline, or restrictions that have an impact on the passengers’ buying decision In the case of FCRM, a combination of Booking Curve detruncation, Pick-up forecasting, and Expected Marginal Seat Revenue algorithm (Belobaba, 1987, 1992), is used (see Gorin, 2000) Under Fare Class Revenue Management, advance purchase requirements and restrictions still apply, and are reinforced by revenue management controls to protect seats for later-booking high-fare passengers, in turn limiting seats made available to early-booking low-fare passengers In the remainder of the paper, Fare Class Revenue Management is referred to simply as Revenue Management (RM), as opposed to FCFS acceptance of seat requests 3.2 New entrant scenario Upon entry, the new entrant carrier offers three daily nonstop flights scheduled at the exact same times as the nonstop incumbent carrier’s flights (Airline 1) The nonstop incumbent’s schedule is mirrored to eliminate the effect of schedule preference on passenger choice In this scenario, passengers now have the option of flying on the nonstop incumbent carrier, its nonstop new entrant competitor, or the connecting incumbent carrier The new entrant offers a two-tier fare structure as follows (Table 3): Fully unrestricted Y class fare set at $135 (the same fare as the B class fare on the incumbent carrier in the base case), approximately 48% lower than the previous Y fare Restricted M class fare (roundtrip and Saturday night stay requirements with 14 days advance purchase) priced $10 below the base case Q fare on the incumbent, at $53 This two-tier fare structure is based on the observation that low-fare new entrants typically offer a simplified fare structure compared to incumbents The notion of simplification does not necessarily involve the removal of all restrictions and advance purchase requirements, but rather a decrease in the number of fare classes offered, and consequently in the complexity of the fare structure In addition, low-fare new entrants typically offer substantially lower fares relative to the incumbents’ standard fare structure To test the effect of the entrant’s capacity on market performance, various capacity levels offered by the new entrant on its three daily flights were simulated New entrant capacity ranges between 15 and 50 seats per flight, with intermediate capacity settings of 25 and 30 seats Finally, the new entrant carrier either accepts seat requests on a FCFS basis, or use RM In most of the simulations presented here, we assumed that all competitors use the same revenue management system (or lack thereof) Upon entry, we assume that the incumbents not fully match the entrant’s fare structure Nonetheless, it would be unrealistic to presume that the incumbent ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 carriers would let the entrant offer a lower fare than their lowest available fare We therefore model an observed limited incumbent fare response in which the incumbents match the lowest fare of the new entrant in their most restrictive fare class As a result, the incumbent carriers are offering a fare of $53 in their Q class, which is more restrictive than the M class fare offered on the new entrant carrier at the same price Table summarizes the type of service, frequency, capacity, fares and revenue management approach of each carrier in the competitive case Results 4.1 Impact of entry on market-level measures Table shows the effect of increasing new entrant capacity on traffic, revenues and average fares for the total market It also shows the effect of revenue management on each of the above-mentioned measures As new entrant capacity increases, total traffic increases in all cases, total market revenues increase (even though they initially decrease slightly following entry in the case of RM), and average fares decrease The effects of increasing entrant capacity on these traditional measures of airline performance are thus relatively straightforward, and in line with previous studies It also appears that the initial effect of entry is 263 far greater than the effect of increasing new entrant capacity Competitive revenue management settings not appear to significantly affect the relative impact of entry on total traffic, as shown in Table Total market revenues and average fares, however, are affected quite differently by the competitive revenue management situation In the case of entry without revenue management (by all carriers), total market revenues increase by as much as 13% at 150 seats on the new entrant, and increase with new entrant capacity Comparatively, when all carriers use RM, Table shows that total market revenues initially decrease slightly with entry, but increase slowly as new entrant capacity increases In the case of RM, total market revenues remain below baseline revenues (without a new entrant) In addition, while pre-entry revenues were higher in the case of RM than with FCFS, post-entry revenues are actually lower in the case of RM at all entrant capacity levels tested We explain these results in Section 5, and show that they are highly dependent on the assumption that the incumbent carriers not match the new entrant’s fare structure While average fares decrease in both cases (whether airlines are using FCFS or RM), the magnitude of the decrease is greater in the case of RM In addition, postentry average fares are generally lower when all carriers use RM than when they accept seat requests on a FCFS basis Table Competitive case summary Competitive case Airline Airline Airline (New entrant) Service Nonstop Connecting Nonstop Frequency & capacity  30  30  15–25–30 or 50 Fares by fare class Revenue management Y B M Q $261 $261 $135 $135 $135 n/a $92 $92 $53 $53 $53 n/a FCFS or FCRM FCFS or FCRM FCFS or FCRM Table Absolute and relative impact of entry on average market fare, revenues and traffic, as a function of entrant capacity and competitive revenue management settings Total market Traffic Revenues Average fare FCFS FCRM FCFS FCRM FCFS FCRM Absolute No entrant  15  25  30  50 126 172 179 182 191 122 163 178 182 190 $15,205 $15,699 $16,639 $16,919 $17,232 $15,914 $15,237 $15,247 $15,420 $15,775 $121.15 $91.37 $92.88 $93.01 $90.32 $130.04 $93.52 $85.85 $84.93 $82.85 Relative to no entrant  15  25  30  50 37% 43% 45% 52% 33% 45% 48% 56% 3% 9% 11% 13% À4% À4% À3% À1% À25% À23% À23% À25% À28% À34% À35% À36% ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 264 These results can be explained by looking more closely at the impacts of entry on each carrier The greater relative (and absolute) decrease in total market revenues and average fares in the case where all carriers use RM is a direct consequence of the combination of all carriers using RM with the incumbents not matching the entrant’s fare structure In particular, the new entrant is now able to forecast, and thus protect seats for, latebooking high-fare demand Combined with the fact that the new entrant offers lower fares than the incumbent in equivalently restricted fare classes, this leads to substantial revenue dilution at the market level (from the incumbent carriers), causing the observed decrease in market revenues and average fare, as shown in Table Fig shows the impact of revenue management on the mix of passengers on the incumbent and new entrant carrier, and illustrates that revenue management allows Fig Passenger distribution by fare class on the incumbent and new entrant, as a function of the competitive revenue management situation, and at  30 seats on the new entrant the new entrant to increase loads in its higher fare class (Y class) It also shows that revenue management leads to an increase in unrestricted fare traffic (combined Y and B class loads) at the total market level Because the higher fare class passengers on the new entrant actually pay lower fares than they did on the nonstop incumbent, overall market revenues and average fares decrease 4.2 Impact of revenue management and new entrant capacity on traditional measures of incumbent performance Table summarizes the average fare, traffic, revenues, and market and revenue share on the incumbent carrier as a function of entrant capacity and competitive revenue management settings It shows that now, all three aggregate measures of incumbent performance (traffic, revenues and average fare) are affected very differently by entry as a function of the competitive revenue management situation Comparatively, the entrant’s capacity has a smaller impact The entrant’s capacity does nonetheless have some impact on incumbent performance When all carriers accept requests for seats in a FCFS manner, the effects on traffic, revenues and average fares are consistent with usual expectations in the case of a limited response to entry: Incumbent traffic and revenues decrease, while its average fare increases (after initially decreasing due to the lower Q fare after entry), with increasing new entrant capacity In the case of RM, the effect is far less intuitive: Incumbent traffic decreases, but the relative decrease is lower at intermediate new entrant capacity Similarly, the average fare also decreases, but to a greater extent at intermediate new entrant capacity Revenues, on the other hand, behave more intuitively and decrease with increasing new entrant capacity These intuitive—and less intuitive—results are actually a consequence of the competitive revenue management Table Absolute and relative impact of entry on incumbent average fare, revenues and traffic, as a function of entrant capacity and competitive revenue management settings Airline Traffic Revenues Average fare Market share Revenue share FCFS FCRM FCFS FCRM FCFS FCRM FCFS (%) FCRM (%) FCFS (%) FCRM (%) 67 47 40 37 22 61 34 36 34 22 56 44 42 41 34 75 41 31 30 21 Absolute No entrant  15  25  30  50 84 80 72 67 43 75 56 64 61 42 $8,490 $6,923 $6,982 $6,925 $5,877 $12,003 $6,197 $4,794 $4,559 $3,370 $101 $86 $97 $103 $138 $160 $112 $75 $75 $80 Relative to no entrant  15  25  30  50 À4% À14% À20% À49% À26% À15% À19% À43% À18% À18% À18% À31% À48% À60% À62% À72% À15% À4% 2% 36% À30% À53% À53% À50% ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 Fig Incumbent revenues and average fares as a function of competitive revenue management settings and new entrant capacity setting (combined with the limited fare match on the part of the incumbent carriers) Fig illustrates the differences in revenues and average fare on the incumbent carrier as a function of the competitive revenue management situation and the new entrant’s capacity When airlines accept passenger requests on a FCFS basis, incumbent traffic decreases with increasing new entrant capacity, by up to 49% at high new entrant capacity (150 seats) By adding capacity in the market at fares that are relatively more attractive, the new entrant is able to divert much of the incumbent’s former traffic Capacity constraints combined with demand stimulation limit the incumbent’s loss of passengers at low entrant capacity levels As capacity increases on the new entrant, diversion also increases, hence the increasing losses in traffic on the incumbent carrier The fact that none of the carriers practice revenue management then affects the average fare on the incumbent carrier: Since leisure passengers book first and airlines accept passenger requests on a FCFS basis, it is consequently not surprising that the first passengers to be diverted from the incumbent carrier are leisure passengers The new entrant carrier accepts bookings from the bottom up, and thus starts with low-fare traffic, which frees up capacity on the incumbent carrier The joint effect of demand stimulation (achieved through the lower Q fare) and diversion of traffic to the new entrant leads to a decrease in high-fare class (Y, B and M) loads, but a slight increase in Q class loads on the incumbent carrier when the new entrant comes in at low capacity, as shown in Fig This leads to a decrease in the average fare on the incumbent carrier As new entrant capacity increases, more of the low-fare traffic is able to book the less restricted (but equally cheap) fare on the new entrant, thus freeing capacity on the incumbent This leads to an increase in Y class bookings on the incumbent (since seats are more likely to remain 265 Fig Incumbent loads by fare class as a function of entrant capacity and with FCFS on all carriers available on the incumbent in this FCFS acceptance of seat requests scheme), and accordingly an increase in the average fare on the incumbent carrier The ensuing effect on revenues is initially a moderate decrease (À18%) followed by a greater decrease as the new entrant diverts more and more traffic from the incumbent carrier, mostly from lower fares, to an extent that cannot be compensated by the increase in Y class bookings Note that there is little competition between the new entrant’s unrestricted fare class and that of the incumbent carriers Indeed, given the more attractive fare structure on the new entrant, and the fact that seat requests are accommodated on a FCFS basis, early booking, low-fare passengers will overwhelmingly choose to travel on the new entrant, and thus fill up its capacity This leaves the incumbents to share the remainder of the demand, namely, the late-booking, high-fare passengers With even larger (greater than total leisure demand) new entrant capacity, however, we would have started to observe diversion from the incumbent’s higher fare classes to the new entrant When all carriers use RM, the incumbent carrier loses relatively more traffic at extreme levels of entrant capacity (3  15 and  50) and relatively less traffic at intermediate levels of entrant capacity (3  25 or  30) The traffic recovery at intermediate levels of entrant capacity is a direct consequence of the incumbent’s use of revenue management As shown in Fig 6, the incumbent carrier initially loses traffic in all fare classes except the lowest class (Q), in which its loads increase because of low-fare demand stimulation, making up for some of the loss of higher-class revenues When new entrant capacity further increases to  25 and  30, the incumbent recovers some traffic, but in the lowest fare class only In both cases, the revenue management system evaluates the trade-off between carrying more passengers in lower fare classes at the ARTICLE IN PRESS 266 T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 expense of high-fare traffic, which has become less likely to materialize on the incumbent, and thus opens the availability of lower fare classes This leads to an increase in loads, but a decrease in overall revenues on the incumbent As new entrant capacity keeps increasing, diversion from the incumbent to the new entrant increases further, and loads decrease on the incumbent in all fare classes (including the lowest fare class) In addition, the fact that the new entrant also practices revenue management has an impact on the mix of traffic losses on the incumbent: The new entrant is now able to forecast latebooking high-fare passengers, and thus protect seats for these passengers At the time of booking, these passengers are now faced with a choice between airlines and 3, and are more likely to choose Airline 3’s lower fares (with comparable restrictions and advance purchase requirements) over those of Airline This has the Fig Incumbent loads by fare class as a function of entrant capacity and with FCRM on all carriers effect of diverting passengers from all fare classes on the incumbent carrier—a substantial difference from the FCFS case where the new entrant only diverted early booking (thus low-fare) passengers until it ran out of seats to sell This behavior has a direct negative effect on the incumbent’s average fare: It decreases sharply following entry, by as much as 53%, as its traffic shifts towards lower fare classes The decrease in Q class loads on the incumbent carrier also has the effect of leading to a slight recovery in the incumbent’s average fare as the new entrant’s capacity becomes very large (150 seats) More passengers get diverted from the incumbent, and in particular from Q class, hence the increase in average fare Entry also leads to a decrease in incumbent revenues ranging from 48% of pre-entry revenues to as much as 72% of pre-entry revenues at high entrant capacity In addition, the contribution of Y class to total incumbent revenues shifts from over 66% of revenues without a new entrant to about 38% when the new entrant offers 150 seats in the market Fig illustrates the effect of entry on incumbent traffic and the diversion from the incumbent to the new entrant carrier, which is responsible for the variations in revenues and average fares described previously In particular, Fig shows how pre-entry traffic on the incumbent gets re-distributed when passengers are offered additional competitive service by the new entrant Tables and also show the effect of revenue management on average fares, traffic, revenues, and revenue and market share as new entrant capacity varies Such a comparison shows that revenue management has a substantial impact on relative changes in revenues, fares, traffic and revenue shares upon entry In particular, it appears that when all carriers use RM, relative decreases in total average market fares, and increases in total traffic, are greater than without Fig Effect of entry on incumbent loads and diversion from incumbent to entrant at  30 capacity on entrant ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 revenue management, while in fact the actual response of the incumbent carriers was identical in both cases (Table 5) Similar conclusions can be drawn at the incumbent carrier level Table shows, for example, that when the entrant comes in with 90 seats per day, the relative decrease in revenues on the incumbent is 18% in the case of FCFS, or 62% in the case of RM The incumbent’s average fare in the case of FCFS increases by 2% while it decreases by 53% in the case of RM A naive comparison of the case of entry when all carriers allocate seats on a FCFS basis with the case where all carriers use revenue management could lead to the conclusion that the response of the incumbent carrier was far more aggressive in the latter case However, the response was identical—the incumbent effectively did not match the new entrant’s fare structure Performance of RM relative to FCFS, on the incumbent carrier As discussed in Section 4.2, and as shown in Table 6, the absolute and relative impacts of entry on the incumbent carriers are much more dramatic in the case of RM, relative to that of FCFS acceptance of seat requests It is important to stress here that these results by no means imply that RM negatively impacts the revenues of the airline using it Rather, we show in the following paragraphs that the incumbent carrier does in fact gain from using revenue management—compared to where it would accommodate requests for seats on a FCFS basis The apparent greater impact of entry on incumbent revenues is a combination of the effect of all carriers using RM and lower fares on the new entrant carrier diverting high-fare traffic from the incumbent carrier 5.1 Incumbent revenue gains from fare class revenue management Table shows Airline 1’s traffic, revenues, average fare and market and revenue share, as a function of new 267 entrant capacity, in the case where only the new entrant uses revenue management while the incumbent carriers use FCFS (labeled FCFS vs FCRM), and where all carriers use RM (labeled FCRM All) Airline benefits from using RM as opposed to FCFS, given that the new entrant carrier uses RM The relative increase in revenues from using RM decreases with increasing new entrant capacity, as the excess capacity on the new entrant still leads to revenue dilution Nonetheless, relative revenue gains attributable to RM over FCFS for the incumbent vary from 27% at low entrant capacity (3  15), to 6% when the new entrant has a very high capacity By comparison, loads are lower on the nonstop incumbent carrier in the case where all carriers use RM than when the incumbents are using FCFS Once again, this is a consequence of the fact that all carriers protect seats for late-booking passengers, and thus spill lowfare, early-booking passengers when they use RM Average fares, however, are much higher (up to +90% on the incumbent carrier at a capacity of  15 on the new entrant carrier) when all carriers use RM, as the mix of passengers includes more high-fare customers Therefore, RM brings revenue gains to the incumbent carrier, as would be expected In the previous simulation results, it is the fact that the new entrant is also using revenue management in conjunction with a two-tier fare structure (not matched by the incumbents) that leads to lower incumbent revenues than in the case where none of the carriers apply revenue management These results show that upon entry, it is not revenue management that is responsible for the decrease in revenues on the incumbent carrier, but rather the combination of revenue management, differentiated products, and lower fares on the new entrant that lead to a change in traffic mix by fare class, and a decrease in Airline 1’s revenues When the incumbent carriers switch to revenue management, their revenues increase compared to the case where only the new entrant carrier uses revenue management Finally, the new entrant’s revenues increase as it begins to use RM as opposed to FCFS (against FCFS on the incumbents), as shown in Table Table Airline traffic, revenues, average fares and market and revenue share as a function of the revenue management situation Airline No entrant  15  25  30  50 Traffic Revenues Average fare Market share Revenue share FCFS vs FCRM FCRM All FCFS vs FCRM FCRM All FCFS vs FCRM FCRM All FCFS vs FCRM FCRM All FCFS vs FCRM FCRM All 84 82 77 72 47 75 56 64 61 42 $8,490 $4,862 $4,421 $4,250 $3,173 $12,003 $6,197 $4,794 $4,559 $3,370 $101.11 $58.94 $57.28 $58.62 $68.11 $160.07 $111.62 $74.94 $74.68 $79.54 67% 48% 43% 40% 24% 61% 34% 36% 34% 22% 56% 31% 27% 25% 18% 75% 41% 31% 30% 21% ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 268 Table Airline traffic, average fares and revenues Airline 3  15  25  30  50 Traffic Average fare Revenues FCFS All FCFS vs FCRM FCFS All FCFS vs FCRM FCFS All FCFS vs FCRM 45 74 88 139 41 66 80 133 $66.53 $67.23 $67.55 $69.67 $132.78 $115.12 $104.43 $85.39 $2,969 $4,949 $5,925 $9,674 $5,412 $7,606 $8,395 $11,372 Table Incumbent traffic, revenues and average fares under revenue management as a function of whether the incumbents match the new entrant’s two-tier fare structure Airline Traffic Absolute No Entrant  15  25  30  50 Rel to limited match  15  25  30  50 Revenues Average fare Limited match Full match Limited match Full match Limited match Full match 75 56 64 61 42 75 79 78 78 76 $12,003 $6,197 $4,794 $4,559 $3,370 $12,003 $6,707 $6,388 $6,300 $6,149 $160.07 $111.62 $74.94 $74.68 $79.54 $160.07 $84.39 $81.63 $81.18 $80.51 43% 22% 27% 80% Revenue management therefore increases airline revenues, and it is the new entrant’s pricing structure which is primarily responsible for the decrease in revenues on Airline 5.2 Incumbent performance under revenue management when it matches the new entrant’s two-tier fare structure Table shows the nonstop incumbent’s traffic, revenues and average fares when incumbents match the new entrant’s two-tier fare structure, under the assumption that all carriers use revenue management The incumbent’s revenues increase by as much as 82% over the case where it does not match the new entrant’s fare structure (at  50 seats on the new entrant carrier) The reason for this dramatic increase in revenues lies in the impact of matching the new entrant’s two-tier fare structure on incumbent loads, and the distribution of this traffic between fare classes Table shows that the average fare on the incumbent carrier actually increases when it matches the new entrant’s lower fares The incumbent’s passengers, on average, now generate more revenues than when the incumbents maintained their traditional fare structure and did not fully match the new entrant The only exception arises at low entrant capacity, where in this case the nonstop incumbent’s average fare decreases, but revenue increases nonetheless In this case, the large increase in 8% 33% 38% 82% À24% 9% 9% 1% loads offsets the decrease in average fare, leading to an increase in revenues Table shows that incumbent performance, when it uses revenue management, is highly dependent on the incumbent’s fare structure relative to that of the new entrant As mentioned, the apparently poorer revenue performance of the incumbent carrier in the case where it uses revenue management but does not match the new entrant’s fare structure, is mostly caused by the difference in fare structure between carriers Conclusions The simulations show that even with a minimal response by incumbents to new entry—consisting of a match of only the lowest new entrant fare, with more restrictions placed upon it—traditional measures of the performance of an incumbent carrier are affected dramatically based solely on whether the airlines in a market practice revenue management In particular, in the case where the new entrant comes in with the same capacity as the incumbent (90 seats per day), the simulated relative effect on the nonstop incumbent’s revenues and average fares is À18% and +2%, respectively, when all carriers accept passenger requests on a FCFS basis, as compared to À62% and À53%, respectively, when all competitors use RM ARTICLE IN PRESS T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 Simulation results also show that average fares on the nonstop incumbent carrier can either increase or decrease following entry, depending on whether the airlines perform revenue management, again with no difference in the incumbents’ pricing response For example, when all carriers accept seat requests on a FCFS basis, the nonstop incumbent’s average fare increases with new entrant capacity, as primarily lowfare passengers are diverted from the incumbent carriers to the new entrant In contrast, the incumbent’s average fare decreases with increasing new entrant capacity when all carriers use revenue management, as more high-fare passengers shift to the new entrant, which is now protecting seats for them Focusing on average fares in the evaluation of airline performance before and after entry could lead to the incorrect conclusion that there is a difference in response from the incumbent carrier between the two cases, when in fact there is none The effect of entry on the incumbent is far greater (and worse) when both the incumbents and the new entrant use revenue management Given the new entrant’s more attractive fares and use of revenue management, it diverts mostly high-fare passengers from the nonstop incumbent, and thus substantially hurts incumbent revenues This result illustrates that revenue management can be as important for new entrant carriers as for incumbents The simulations also show that under revenue management on all carriers, matching the new entrant’s fare structure substantially increases incumbent revenues by up to 82% (compared to limited match by the incumbents) This emphasizes the importance of the competitive pricing situation on individual airlines’ revenue performance, even in a simple single market environment Acknowledgements This research would not have been possible without the programming talent of Craig Hopperstad, who wrote the Passenger Origin Destination Simulator, and implemented the revenue management methods in the simulation environment We also thank the Sloan Foundation for its generous support of the MIT Global Airline Industry Program References Areeda, P.P., Turner, D.F., 1975 Predatory pricing and related practices under section of the Sherman Act Harvard Law Review 88, 697–733 Areeda, P.P., Turner, D.F., 1976 Scherer on predatory pricing: a reply Harvard Law Review 89, 891–900 Areeda, P.P., Turner, D.F., 1978 Williamson on predatory pricing Yale Law Journal 87, 1337–1352 269 Bailey, E.E., Baumol, W.J., 1984 Deregulation and the theory of contestable markets Yale Journal on Regulation 1, 111–137 Bailey, E.E., Graham, D.R., Kaplan, D.P., 1985 Deregulating the airlines MIT Press Series on the Regulation of Economic Activity MIT Press, Cambridge, MA Baumol, W.J., 1979 Quasi-permanence of price reductions: a policy for prevention of predatory pricing Yale Law Journal 89, 1–26 Baumol, W.J., 1982 Contestable markets: an uprising in the theory of industry structure American Economic Review 72, 1–15 Baumol, W.J., Panzar, J.C., Willig, R.D., 1983 Contestable markets: an uprising in the theory of industry structure: reply The American Economic Review 73, 491–496 Belobaba, P.P., 1987 Air Travel Demand and Airline Seat Inventory Management M.I.T Doctoral thesis, Cambridge, MA Belobaba, P.P., 1992 Optimal vs Heuristic Methods for Nested Seat Allocation, AGIFORS Reservation Control Study Group Presentation, Brussels Belobaba, P.P., Wilson, J.L., 1997 Impacts of yield management in competitive airline markets Journal of Air Transport Management 3, 3–10 Carrier, E.J., 2003 Modeling airline passenger choice: Passenger preference for schedule in the passenger origin destination simulator M.I.T Master’s thesis, Cambridge, MA Competition Tribunal, 2000 Commissioner v Air Canada/CanJet (CT-2000/004), Ottawa Competition Tribunal, 2001 Commissioner v Air Canada (CT-2001/ 002), Ottawa Dodgson, J., Katsoulacos, Y., Pryke, R., 1991 Predatory behaviour in aviation, a report to the competition directorate of the European commission Commission of the European Community Official Publications, Luxemburg Gorin, T.O., 2000 Airline revenue management: sell-up and forecasting algorithms M.I.T Master’s thesis, Cambridge, MA Hurdle, G.J., Johnson, R.L., Josokow, A.S., Werden, G.J., Williams, M.A., 1989 Concentration, potential entry, and performance in the airline industry The Journal of Industrial Economics 38, 119–139 Hopperstad, C.H., 2000 Passenger origin destination simulator technical specifications (Revision 1) November Unpublished working paper Seattle, WA Joskow, P.L., Klevorick, A.K., 1979 A framework for analyzing predatory pricing policy Yale Law Journal 89, 213–270 Lee, A.Y., 1998 Investigation of competitive impacts of origin– destination control using PODS M.I.T Master’s thesis, Cambridge, MA McGee, J.S., 1958 Predatory price cutting: the standard oil (N.J.) case Journal of Law and Economics 1, 137–169 Morrison, S.A., Winston, C., 1990 The dynamics of airline pricing and competition The American Economic Review 80, 389–393 Oster, C.V., Strong, J.S., 2001 Predatory Practices in the US Airline Industry US Department of Transportation, Office of the Assistant Secretary for Aviation and International Affairs, Washington Perry, L.J., 1995 The response of major airlines to low-cost airlines In: Jenkins, D., Preble Ray, C (Eds.), Handbook of Airline Economics First Edition McGraw-Hill, New York Skwarek, D.K., 1996 Competitive Impacts of yield management systems components: forecasting and sell-up models, M.I.T Master’s thesis, Cambridge, MA United States of America vs American Airlines, 2001 Report No 991180-JTM, Washington US Department of Transportation, 1998 Proposed statement of enforcement policy on unfair exclusionary conduct by airlines Office of the assistant secretary for public affairs, Washington ARTICLE IN PRESS 270 T Gorin, P Belobaba / Journal of Air Transport Management 10 (2004) 259–270 Whinston, M.D., Collins, S.C., 1992 Entry and competitive structure in deregulated airline markets: an even study analysis of people express RAND Journal of Economics 23, 445–462 Williamson, O.E., 1977 Predatory pricing: a strategic and welfare analysis Yale Law Journal 87, 284–340 Windle, R.J., Dresner, M.E., 1995 The short and long run effects of entry on US domestic air routes Transportation Journal 35, 14–25 Zickus, J.S., 1998 Forecasting for Airline network revenue management; revenue and competitive impacts M.I.T Master’s thesis Cambridge, MA