Journal of Air Transport Management

Transcription

1 Journl of Air Trnsport Mngement 15 (9) 3 9 Contents lists vilble t ScienceDirect Journl of Air Trnsport Mngement journl homepge: Network competition in the open vition re Eric Pels VU University, De Boeleln 15, 1HV Amsterdm, The Netherlnds bstrct Keywords: Airline networks Competition Allinces Open vition re In the joint open vition re between the US nd EU will become relity. It is expected tht competition will increse. The rection of the irlines depends on the possibility to mke profits in new mrkets (mrkets tht cn now be entered). This, in turn, depends on network chrcteristics. In this pper we find tht full liberliztion of interntionl mrkets by mens of bilterl greement results in higher welfre thn the formtion of n llince. Crriers, however, will lso in fully deregulted vition mrkets most likely opt for n llince. This is result of built-in competitive effect of hub spoke networks. Only in mrkets where the reservtion price is very high (e.g. to London Hethrow), hub spoke irlines my enter competitive gme. Low-cost irlines, which do not operte extensive hub spoke networks, my find it profitble to enter new mrkets. Ó Elsevier Ltd. All rights reserved. 1. Introduction In the joint open vition re between the US nd EU will become relity. The greement between the US nd EU tht creted this open vition re ws endorsed by EU trnsport ministers on Mrch, 7. This drft US EU open skies greement deregultes the lrgest vition mrket in the world. According to US Trnsporttion Secretry Mry Peters, this greement will benefit both pssengers nd irlines: Tering down regultory brriers llows us to foster more ffordble nd convenient ir trvel nd gives our irline industry more opportunities to compete, innovte nd thrive. In the medi, there were lredy clims tht enhnced competition would lower trnstlntic fres with hundreds of Euros. The min competitive thret would come from low-cost crriers, s the removl of regultory brriers opens up trnstlntic mrkets for low-cost crriers. The mrket from London to the US ccounts for bout 1/3 of ll flights from Europe to the US, so one could expect tht entry will tke plce especilly in this lucrtive mrket. It is not the first time tht hopes were high when the deregultion of vition mrkets ws nnounced. When the US vition mrket ws deregulted in 197, the number of irlines incresed rpidly. But it decresed just s rpidly. Although fres decresed in rel terms, irlines lso succeeded in creting fortress hubs, protecting them from competition from other conventionl irlines, nd llowing them to chrge hub-premiums. Following the deregultion of the EU nd US mrkets, incresed competition E-mil ddress: pels@feweb.vu.nl usully cme from low-cost crriers on short-hul mrkets nd from other conventionl irlines offering indirect tickets on longhul mrkets. Competition between irlines on direct links between mjor hubs on the trnstlntic mrket ws usully limited to competition between US nd Europen irlines. More often thn not, US nd Europen irlines entered llince greements, to exploit network complementrities, nd to reduce competition. Although such llinces creted cler benefits for pssengers in terms of network connectivity, competition on routes between mjor hubs decresed. The expecttions from the joint vition re resemble the expecttions from the deregultion of the US nd EU vition mrkets: incresed competition, nd lower fres. But the experience of the deregultion shows tht we should be creful mking strong sttements bout n increse in the number of competitors. Wht will be the effects on irline networks of the new joint open vition re? In this pper we discuss the network chnges tht my occur s result of the joint open vition re. Specificlly, we discuss the chnges in the sector following the deregultion of the EU nd US mrkets, nd then see how this crries over to the trnstlntic mrket. The effects of the deregultion of the US nd EU mrkets re discussed in literture review in Section. Section 3 discusses model of network competition in the Open Avition Are, nd Section presents the theoreticl results. Section 5 concludes.. Effects of deregultion of vition mrkets in the US nd EU Prior to the deregultion in the US, Civil Aeronutics Authority, lter renmed s the Civil Aeronutics Bord (CAB), determined /$ see front mtter Ó Elsevier Ltd. All rights reserved. doi:.1/j.jirtrmn..9.1

2 E. Pels / Journl of Air Trnsport Mngement 15 (9) 3 9 routes nd regulted fres in the US to protect the crriers from destructive competition nd protect consumers, while llowing irlines to obtin resonble return on ticket sles. The CAB llowed fre increses to compenste for the higher irline operting costs following the oil crisis in the 197s, but it becme more nd more cler tht government regultions were too restrictive for the irline industry. In 1975 the CAB concluded tht the irline industry is nturlly competitive nd not monopolistic, nd in 197 the Airline Deregultion Act ws pssed. All restrictions on (domestic) routes, fres nd schedules were to be removed. Incresed irline operting efficiency nd competition were expected to benefit both irlines nd pssengers. Following deregultion of the US vition mrket, there ws lrge-scle entry of new crriers, followed by the rpid deprture of lmost ll of them. Immeditely fter the deregultion, there were bout mjor crriers, while some 15 yers lter there were six or seven. It thus ppers tht competition did not increse following the deregultion, lbeit tht fres decresed in rel terms since the deregultion. The decline in fres from 197 to 195 represented svings of $11 billion US to pssengers in 19 (Khn, 19). The disciplining effect of competition ws, however, geogrphiclly unevenly distributed. Airlines were free to operte their most efficient networks, nd most irlines decided to operte the hub-nd-spoke network, which llows for the exploittion of density economies. The number of competitors my hve ctully decresed on routes strting or terminting t hub. On routes between hubs nd on long hul, connecting flights, there my, however, be fierce competition. Hub irlines offer their locl pssengers direct flights, with greter frequency, better irport fcilities, nd more destintions thn its competitors. Becuse of this service qulity, irlines re ble to chrge hub premium to its locl pssengers. The empiricl evidence hs widely vrying estimtes of the hub premium, but 15% is probbly resonble point estimte. But lso the exploittion of mrket power cn led to higher fres; lck of competition llows irlines to crete so-clled fortress hubs nd rise fres without the thret of entry. Zhng (199) shows tht irlines using hub spoke networks my not hve n incentive to invde ech other s network, becuse this my lower profits in the originl network. Zhng uses the network depicted in Fig. 1 to mke this point, where irline 1 uses H s hub, serves AH nd BH directly, nd AB indirectly, while irline uses K s hub, serves AK nd BK directly, nd serves AB indirectly. This network is not relistic since the mrket between hubs is missing, but similr results re obtined when this mrket is included. When irline 1 invdes mrkets AK nd BK, the price decreses becuse of incresed competition. Airline cn behve by behving ggressively in the AB-mrket: incresed output in AB mrket lowers verge costs on the AK nd BK links becuse of density economies. Becuse irline behves ggressively in ABmrket, irline 1 looses output in AB mrket (irline cptures prt of the AB mrket of irline 1). As result, mrginl costs on the AH nd BH links increse. Averge costs thus increse. Becuse Fig. 1. Network configurtion. verge cost increse on AH nd AB links, the number of pssengers in the AH nd BH mrkets decreses (flights re more expensive). Becuse output is decresed in the originl network (HAB), the dditionl profits of the new AK nd BK mrkets hve to be blnced ginst losses in the other mrkets. When density economies re strong (effects mentioned bove re strong) nd willingness-to-py is high, ttcking the network of irline decreses profits for irline 1. If irline lso responds by invding the AH nd BH mrkets, the effects resemble effects of invsion of AK nd BK mrkets by irline 1. Therefore, entry in competitor s network my led to lower overll profits. Insted, more often thn not irlines choose to enter llince greements rther thn to enter competitive gme. The concentrtion on the US vition mrket resulted from number of bnkruptcies nd vrious forms of coopertion between crriers (vrying from loose greements to full mergers). Allinces (ntionl or intercontinentl) llow irlines to reduce competition (in the cse of prllel llinces), but lso to dd destintions to the network without invding new mrkets, so tht density economies cn be exploited (in the cse of complementry llinces). The process of llince formtion therefore cn be seen s continution of the network formtion process tht strted with the formtion of hub spoke networks. The policy implictions of the bove mentioned findings re interesting. In deregulted vition networks, where irlines cn enter nd leve mrkets t their own discretion, irlines using hub spoke networks, most likely do not hve n incentive to enter ech other s networks. Insted, they opt for the formtion of fortress hubs, where one irline (the mjor hubbing irline t the irport) domintes the irport in the number of flights, nd the formtion of llinces. Low-cost crriers, which do not operte hub spoke networks nd therefore do not fce the network effects resulting from indirect pssengers, do not suffer from the network effects described bove when they enter new mrkets. In deregulted environment, low-cost crriers re therefore likely to enter new mrkets s soon s they see potentil profits, while coopertion with other irlines my be difficult due to lrge differences in operting procedures. The deregultion of the EU vition mrket ws fr more grdul compred to the US cse. But the outcomes re, to certin extent, similr. Mny Europen irlines were stte compnies with rdil networks. In the deregulted environment, more nd more irlines will be privtized, while the shift of the rdil network to the hub spoke network relted to the timing of the flights, to llow for more convenient trnsfers. Airlines with hub spoke irlines did not invde ech other s networks, nd lso in the EU there ws concentrtion: some irlines went bnkrupt (Swissir, Sben), while other irlines entered llince greements (the Air Frnce- KLM merger being the most fr reching). Low-cost irlines ppered lso in Europe fter the deregultion of the EU vition mrkets. Low-cost irlines opened new mrkets (e.g. Rynir London Crcssonne), but lso entered locl mrkets of estblished hub-crriers (e.g. esyjet on Pris Toulouse). Another importnt impliction of deregultion nd bilterl greements between the US nd Europen countries is tht US irlines nd Europen irlines cn form llinces, thus expnding the number of intercontinentl destintions (nd benefit from economies of density) without extending the network size. To summrize, the deregultion of vition mrkets led to the doption of hub spoke networks by full-service crriers. Such networks llow full-service irlines to form fortress hubs, nd prevent full-service irlines from entering ech other s locl mrkets. Low-cost irlines re not limited by the network effects, nd cn enter mrkets whenever they see profits. In the next sections, we see how these findings crry over to trnstlntic mrkets.

3 E. Pels / Journl of Air Trnsport Mngement 15 (9) Model setup In the pst decde, interntionl irline llinces rpidly gined importnce. Bilterl greements between the US government nd Europen countries gve US irlines ccess to irports within the Europen country in question, nd vice vers. The Open Avition Are extends the bilterl greements between two individul countries to n greement between the US nd EU. For resons explined bove, it seems likely tht the process of llince formtion nd concentrtion will continue. The hub spoke networks used by full-service crriers rely on trnsfer pssengers to keep the cost per set low. As result, smll irliners re reltively expensive to operte, so tht consolidtion leds to lower operting costs. In this section we present theoreticl nlysis of interntionl irline llinces, for exmple on the trnstlntic mrket. The nlysis concerns three scenrios. The first scenrio is the bse scenrio. In this scenrio the irlines (or llinces) re only llowed to operte service to one foreign destintion ( gtewy ), nd no coopertion between irlines or llinces is llowed. Following the deregultion of the vition mrket, irlines or llinces re llowed to operte flights to other foreign destintions. There re two options. In scenrio II, the irlines (or llinces) invde the competitor s network; irlines choose not to cooperte. In scenrio III, the irlines or llinces decide to cooperte. The networks re depicted in Fig.. Assume tht we hve two irlines which operte networks s in Fig.. In the bse scenrio, which we will cll scenrio I, irline opertes hub-nd-spoke network with A s the hub. One of the destintions in s network, B, lies in foreign country. Airline b lso opertes hub-nd-spoke network nd hs B s its hub. A is irline b s foreign destintion. In this network, there is only (direct) competition on link AB. We ssume the cost per link is qudrtic function of the number of pssengers using the link: C Q þ (q/ )Q þ f, where Q is demnd nd f is the fixed cost per link. The inverse demnd function in ll mrkets is P Q/, where Q is demnd in the mrket under considertion 1, so tht we hve for exmple in mrket AB: P AB ðqab þ Q AB b Þ=, where the superscript denotes the irline nd the subscript denotes the mrket. In this setting irline s profits re: ( p I P AB $QAB þ P AC$Q AC þ P BC $Q BC þ P $Q QAB þ Q BC þ 1 $Q q $ QAB þ Q BC þ 1 ) ( $Q Q AC þ Q BC þ 1 $Q q $ Q AC þ Q BC þ 1 ) $Q $f As result of, for exmple, n open skies greement between the two regions, the two crriers re llowed to serve the other foreign destintion from their own hub. If the irlines decide to do so noncoopertively, they re in effect invding ech other s mrkets; this is scenrio II. For exmple, irline b opertes n indirect route between A nd C nd competes with irline, which opertes direct route. Both irlines lso operte two new indirect mrkets next to the new direct mrket. Corresponding profits re: p II PAB $Q AB þ P AC $Q AC þ P AD $Q AD þ P BC $Q BC þ P BD $Q BD Q þ P $Q AB þ QBC þ Q BD q $ QAB þ Q BC þ QBD Q AC þ Q BC þ Q q $ QAC þ Q BC þ Q Q AD þ Q BD þ Q q $ QAD þ Q BD þ Q 3$f Rther thn invding ech other s mrkets, the irlines cn lso choose to cooperte; this is scenrio III. The crriers jointly operte the AB-route. Pssengers trveling on the BC- nd AD-routes use one or both of the cooperting crriers (depending on which irline offers the service on the AB-route), pssengers trveling on the route use both crriers. Following Brueckner (1997) we mke the following ssumptions concerning the llince. Revenues from the mrkets where pssengers my hve to trvel on both irlines re shred evenly between the crriers. The fixed costs of operting the AB-route re lso shred evenly between the crriers. Crrier s profit function in this scenrio is: Fig.. Network configurtions. 1 This specifiction is quite common in the vition economics literture; see e.g. Brueckner (1997), Brueckner nd Spiller (1991), Prk (1997) nd Zhng (199). In effect the irlines re restricted to operte hub-nd-spoke network, while link between, for exmple, B nd D is lso n (theoreticl) option. Optimlity of the hub-nd-spoke network is studied by Pels (). Here we ssume the irline chooses to operte hub-nd-spoke network; we return to this lter.

4 E. Pels / Journl of Air Trnsport Mngement 15 (9) 3 9 ( QAB p III P AB$Q AB þ P AC $Q AC þ P BC $Q BC þ P $Q þ Q BC þ Q q $ QAB þ Q BC þ Q ) ( Q AC þ Q BC þ Q q $ Q AC þ Q BC þ Q ) 3 $f In scenrios I nd II irline mximizes its own profits with respect to q i, where subscript i denotes the mrkets served by. In scenrio III the crriers mximize their joint profits in the AB nd mrkets. In the other mrkets the crriers mximize their own profits scenrio II more profitble upper bound fesible prmeter spce π II - π I = scenrio I more profitble lower bound fesible prmeter spce Network competition in the open vition re: theoreticl perspective In this section we compre the three different scenrios. The purpose of the nlysis is to see if nd how irline (nd b due to symmetry) will chnge its strtegy compred to the bse scenrio (I) if it is llowed to enter the BD mrket (nd through tht mrket the indirect AD nd mrkets). We do this by compring the mximum profits in the different scenrios. The prmeter spce in which we cn compre profits is bounded by the second order for profit mximiztion nd the requirements of nonnegtive quntities nd mrginl costs, see e.g. Pels () for detils. From the optiml prices nd quntities reported in Appendix A, it is cler tht nlyticl comprison of profits in the three scenrios is for ll prcticl purposes meningless. Insted we mke grphicl comprison of scenrios by plotting p x p y, where x nd y represent different scenrios..1. Airline strtegies In Fig. 3 we compre scenrios I nd II for f. 5 We see tht irline will not invde irline b s mrkets if both nd q re reltively high; in tht cse profits in scenrio I re lwys higher. If f increses, the re in which scenrio I is more profitble increses; see Appendix B, Fig. B1. In scenrio II, irline opertes one extr direct link. Revenues hve to be high to compenste: Q nd/or P must be reltively high. When q is low, mrginl costs (MC 1 qq), nd with it mrginl revenues, re reltively high. But when q is high mrginl costs nd prices re low: irline my not generte sufficient revenues under scenrio II, nd scenrio I will be preferred. Even though it extrcts revenues from three extr mrkets (AD, BD nd ), the high vlue of q combined with the downwrd pressure of competition on prices leves irline with loss. Fig. 3. Comprison of profits in scenrios I nd II, f. This result is similr to the result obtined by Zhng (199), who finds tht n invding irline s profits my decrese s result of invsion of the competitor s locl mrkets. It is lwys more profitble for irline to enter n llince. This is no surprise, s competition is eliminted from the AB mrket. Moreover, under this scenrio, irline is ble to sell tickets in the AC nd mrkets, nd shres the fixed costs of the AB-route with irline b. Finlly, network effects llow irline to obtin higher profits due to economies of density. In Fig. we compre profits in scenrio II nd III, gin for f. We see tht only if nd q re low, the fully competitive scenrio (II) will be preferred by the irline. In scenrio III there is no competition, nd s result prices re higher nd, ceteris pribus, demnd will be lower compred to scenrio II. In scenrio III, less direct links re operted, llowing the irlines to profit from economies of density. But when nd q re low nd demnd is low, costs re reltively high becuse there is little opportunity to exploit density effects. In tht cse, the positive effect of higher prices on profits my not blnce the negtive effects of reltively high costs nd lower demnd in scenrio III. However, if f is sufficiently high it is lwys optiml to enter n llince greement (under which fixed costs re lower); see Appendix B, Fig. B, where the indifference curve lies outside (below) the fesible prmeter spce. Above the curve, scenrio III is more profitble. Hence inside the fesible prmeter spce, it is lwys more profitble to enter n llince greement. From the nlysis in this section we conclude tht n irline will expnd its horizon if it gets the opportunity to enter new 1 3 Crrier my use crrier b s cpcity in the BC mrket. Likewise, crrier b cn use s cpcity in the AD mrket. These curves lso exist outside the fesible prmeter spce, nd re bckwrd bending (the curves used to compre consumer surplus in the following section re lso exist outside the fesible re nd re lso bckwrd bending). Only those prts of the curves tht re of economic interest (i.e. re locted inside or prtly inside the prmeter spce) re plotted. 5 The ssumption f is common in the literture. Pels () shows tht fully connected network is most likely the optiml network rther thn hub-nd-spoke network for f. The ssumption f my thus my led to sitution where optiml strtegies in sub-optiml networks re nlyzed. If we ssume tht due to legl or other constrints irline cn only serve destintion D from its hub in scenrio II, we cn mke this ssumption. Moreover, if f increses, hub-nd-spoke is likely to become more profitble (Pels, ), nd the point we try to mke in this pper does not chnge (see Appendix B). Remember tht in scenrio II ll mrkets re competitive. 1 scenrio III more profitble scenrio II more profitble π III - π II = Fig.. Comprison of profits in scenrios II nd III, f.

5 E. Pels / Journl of Air Trnsport Mngement 15 (9) (foreign) mrkets. Assuming tht the prmeters f, nd q re sufficiently lrge, the irlines will enter n llince greement. The literture shows tht density economies re strong in the vition industry (see e.g. Cves et l., 19), while conventionl irlines hve reltively high costs compred to low-cost irlines. Low-cost irlines re ble to operte with reltively low fixed costs due to extensive outsourcing. We see tht conventionl US irlines will not invde Europen networks if the level of demnd nd economies of density re reltively high. The US irlines operte more links, so fixed costs re higher nd revenues hve to be higher to compenste. But when density economies re high, mrginl costs nd mrginl revenues re low. We lso see tht joining n llince greement with Europen crrier is in most cses more profitble thn invding the Europen crrier s network. Only when the level of demnd nd density economies re low, there re not enough opportunities to exploit density economies. It is then lwys more profitble to enter n greement compred to the bse scenrio... Consumer effects We now compre consumer surplus in the different scenrios. Consumer surplus in scenrio II is lwys higher thn consumer surplus in scenrio I inside the fesible prmeter spce. This is strightforwrd, becuse prices re likely to be lower under scenrio II becuse of incresed competition nd density effects. To compre scenrios I nd III nd II nd III, respectively, we plot CS x CS y inside the fesible prmeter spce, where CS stnds for consumer surplus nd x nd y re different strtegies. In Fig. 5 we see tht t low levels of nd q, consumer surplus is lower under scenrio III thn it is under scenrio I. Elsewhere inside the fesible prmeter spce, consumers prefer scenrio III (i.e. n llince). In scenrio III, competition is reduced compred to scenrio I (which leds to higher prices). On the other hnd, by linking the networks, new mrkets re opened, nd there re opportunities to exploit economies of density, which cn led to lower prices. The ltter effect only outweighs the former if (mximum reservtion price) nd q re sufficiently high. Brueckner (1997) finds tht codeshring increses welfre for most prmeter combintions, but the networks nlyzed re different form the networks nlyzed here. Brueckner compres scenrios I nd III for the cse where both irlines operte hub-ndspoke networks with three spoke mrkets, one of which is the intercontinentl mrket on which both irlines compete. In our model, we hve one less spoke mrket but by entering n llince 1 1 > CS I < CS I upper bound fesible prmeter spce - CS I = lower bound fesible prmeter spce Fig. 5. Comprison of consumer surplus I nd III, f. 1 1 < CS II greement, n irline cn open up new mrkets. This is not possible in Brueckner s model. In Fig. we see tht consumer surplus in scenrio II is higher for lmost ll prmeter combintions. Only if q is close to its mximum vlue, there is (smll) region in which consumers would prefer scenrio III. Although irline (nd lso irline b) will opt for n llince if f is sufficiently high, consumers prefer the fully competitive scenrio II. Compring scenrios II nd III, there re two counterviling effects. First, in scenrio II ll mrkets re competitive. In scenrio III there is no competition; irlines collude or hve locl monopolies. Therefore prices cn be expected to be higher in scenrio III. On the other hnd, in scenrio III irline opertes one less direct link (AD). Pssengers in the AD nd mrkets therefore hve to trvel one extr link. This increses the number of pssengers on the AB, AC nd BD links, nd lowers the prices in ll mrkets in which these links re used. It turns out the ltter effect is not strong enough to blnce the former effect unless q is very high. Thus pssengers re most likely better of in scenrio II. The uthorities my hve the intention to protect consumers in either the AB mrket or, in nive pproch following the conjecture tht llinces re lwys hrmful becuse they eliminte competition, in ll mrkets. If they do so by simply not llowing irlines to enter llinces, to encourge the irlines to dopt the fully competitive strtegy, we see tht t high levels of nd q the irlines do not choose to invde their competitor s network; i.e. the irlines stick to the strtegy of scenrio I. However, t these levels of nd q it is likely tht consumers will prefer the codeshring scenrio over the bse scenrio. Although the uthorities hve the intention to protect consumers, in this cse the consumers would ctully be better off if the uthorities would not intervene. Whether or not the government should intervene (i.e. forbid n llince) then depends on the prmeter vlues. If levels of demnd nd density economies re high the government should not intervene. To conclude, it is most likely dominnt strtegy for n irline to enter n llince. Consumer surplus is then most likely lower thn under scenrio II. Authorities my forbid llince in fvor of scenrio II, in which consumer surplus is higher thn in I. But then it my be dominnt strtegy for irlines to keep their old network. 5. Conclusion upper bound fesible prmeter spce lower bound fesible prmeter spce - CS II = > CS II Fig.. Comprison of profits in scenrios II nd III, f. In this pper we rgue tht the effects of the Open Atlntic Avition Are my resemble the effects of the erlier deregultion

6 E. Pels / Journl of Air Trnsport Mngement 15 (9) 3 9 of the US nd EU mrkets. Following the deregultion of these mrkets, irlines rerrnged their networks in process of hubnd-spoke network formtion nd consolidtion. Rther thn to compete, the irlines using hub spoke networks stick to their fortress hubs, nd enter llince greements. The lck of competition on low-density routes my be intrinsic to the vition sector; demnd my be too low to llow more thn just few (or even one) crrier to obtin profitble mrket shre. This is result of network chrcteristics. Routes between hubs nd e.g. trnstlntic routes (i.e. routes with reltively high reservtion price, such s routes to London Hethrow) re more competitive. In the ltter cse, the number of direct competitors my be low, but there re number of indirect lterntives vilble. The theoreticl nlysis in Sections 3 nd confirms tht when density effects re importnt, irline consolidtion will be the most dominnt strtegy. When the uthorities intervene by forbidding coopertion to stimulte competition, irlines will stick to their originl networks. Two remrks re in order. First, the uthorities my decide to protect consumers in the AB mrket. This my then hppen t the expense of the consumers in the other mrkets; the remrks bove bout the undesired policy effects re bsed on consumer surplus. Second, in this pper, consumers re better off if the prices re lower. Network effects hve influence (due to economies of density), but frequency plys no prt. Aggregte frequencies my be higher if irlines enter n llince greement, to the benefit of pssengers. Scenrio III might become more ttrctive to pssengers if frequency is lso considered. Including frequency in the nlysis is therefore n importnt point on the reserch gend. Low-cost crriers will mrkets whenever they see profits. Although low-cost crriers will operte in thin mrkets between secondry irports, they only do so if they re ble to mke money. In the US, nd in the EU, low-cost crriers engge in cherry-picking: select only the most profitble mrkets. At this moment low-cost crriers re not llowed to fly on trnstlntic routes. But s soon s these mrkets re opened up for low-cost crriers, they will lso follow the cherry-picking strtegy in the Open Atlntic Avition Are. This mens tht they will fly on thick routes offering competition to full-service crriers. Also in the Open Atlntic Avition Are it is to be expected tht competition will come minly from low-cost crriers. The importnt question then is whether low-cost irlines will enter such mrkets. In such mrkets, they my not be ble to chieve the quick turn-round times tht re so importnt for their strtegies. Also, they my not be ble to generte enough revenues on long-hul flights if they stick to the singe clss ircrft configurtion. But the most importnt brrier to entry my be the vilbility of slots t the irports on routes tht hve enough potentil demnd to be interesting for low-cost irlines. Slot-lloction therefore should lso be high on the gend of policy mkers. Appendix A. Optiml prices nd quntities Scenrio I Q AB ðq 1Þ½ðq 1Þ þ 1Š P AB ½ð5q 7Þq þ 1Š ðq þ 1Þ ðq 3Þ½ðq 1Þ þ 1Š Q AC q 3 þ 5q Q BC P AC 1 P BC 1 ½ð1q 17Þq þ 3Š þ 3 q ½ðq 11Þq þ 3Š þ 5q Scenrio II Q AB 9 Q AC 9 ½ðq 3Þq þ 3Š 3 þ q ½ðq 9Þq þ 3Š þ q QAD ½ðq 9Þq þ 3Š þ q 9 q 3 1q þ 9 QBC 9 q 3 1q þ 9 QBD 9 h q þ 3 q 3i 1q þ Q 9 P AB 1 9 P AC 1 9 P AD 1 9 P BC 1 9 P BD 1 9 P 1 9 ½ðq 3Þq þ 3Š q þ ½ð1q 9Þq þ 3Š q þ 9 ½ð1q 9Þq þ 3Š q þ 9 ½ð1q 9Þq þ 3Š q þ 9 ½ð1q 9Þq þ 3Š q þ 9 ½ðq 1Þq þ 3Š 3q þ 1

7 E. Pels / Journl of Air Trnsport Mngement 15 (9) Scenrio III Q AB ½ð3q 5Þq þ Š þ ðq 1Þ P AB ½ðq 13Þq þ Š q þ Q AC ½ð3q Þq þ Š þ q P AC 1 ½ð19q Þq þ Š q þ Q AD Q BC Q q þ 3q q þ 3q q þ 3q P AD P BC P ½ð1q 1Þq þ Š q þ 3 ½ð1q 1Þq þ Š q þ 3 ½ð1q 1Þq þ Š q þ 3 Appendix B. The effect of incresed fixed costs In Fig. B1, where f 1, we see tht the re in which profits in scenrio I exceed profits in scenrio II is lrger compred to Fig. 3, where f. This is not surprising, becuse in scenrio II one more link is operted. If the fixed costs of this dditionl link re (very) high, the extr revenues resulting from need to be high to compenste. These revenues re, however, independent of the level of the fixed costs. Therefore, if f increses, scenrio II becomes more nd more unttrctive to the irline. In Fig. B we see tht, for similr resons, scenrio II is no longer n lterntive for scenrio III if f increses; if f is too high, the indifference curve flls outside the fesible re nd scenrio III is lwys preferred. Hence, if f is high, the uthorities need to be extr creful implementing nive policy of simply not llowing n llince to protect consumers; irlines re more likely to stick to scenrio I. 1 1 scenrio III more profitble π III - π II = Fig. B. Comprison of profits in scenrios II nd III, f scenrio II more profitble scenrio I more profitble π II - π I = Fig. B1. Comprison of consumer surplus I nd III, f 1. References Brueckner, J.K., The Economics of Interntionl Codeshring: An Anlysis of Airline Allinces, Office of Reserch Working Pper University of Illinois t Urbn Chmpign. Brueckner, J.K., Spiller, P.T., Competition nd mergers in irline hub-nd-spoke networks. Interntionl Journl of Industril Orgniztion 9, Cves, D.W., Christensen, L.R., Trethewy, M., 19. Economies of density versus economies of scle: why trunk nd locl service irline costs differ. Rnd Journl of Economics 15, Khn, A.E., 19. Surprises of irline deregultion. Americn Economics Review: Ppers nd Proceedings 7, Prk, J.-H., The effects of irline llinces on mrkets nd economic welfre. Trnsporttion Reserch E 33, Pels, E.,. Airport Economics nd Policy: Efficiency, Competition, nd Interction with Airlines. Ph.D. thesis, Vrije Universiteit/Tinbergen Institute, Thel Thesis, Amsterdm. Zhng, A., 199. An nlysis of fortress hubs in irline networks. Journl of Trnsport Economics nd Policy 3,