Do an carriers dominate their connecting markets? A methodology for the analysis of market concentration on transfer routes * & Jaap de Wit *Dept of Planning Faculty of Geosciences Utrecht University, the Netherlands g.burghouwt @geog.uu.nl Transfer dominance in Conclusions and research agenda Hub and spoke networks United States Transfer dominance in Conclusions and research agenda. Spatial concentration of airline networks around a limited number of hubs 2. Temporal concentration: adoption of wave-system structures 8 Temporal concentration Competition Frequency 6 4 2 6 77 94 28 45 62 79 96 23 685 855 25 95 365 535 75 875 245 Time of the day Delta-Atlanta 999 Mixed market concentration result: Fortress hubs, local route concentration, hub premia (e.g. Borenstein, 989) Hub-competition and decreasing transfer dominance (e.g. DoT, 99)
,2,8,6,4,2 99 992 994 996 998 Y e a r very small airports small airports medium airports large hubs primary hubs Herfindahl index US market Direct routes 984 987 99.59.62.632 All routes.53.52.56 Hub and spoke networks in Spatial concentration: stable due to bilateral regulation of ica-services but about to change Temporal concentration: adoption and intensification of wave-system structures Source: Borenstein, 992 3 Wave-system structure Air France Paris CDG 3 Temporal concentration Number of airports with wave-system structure 2 2 6 6 78 96 4 32 5 68 86 24 69 87 5 23 4 59 77 95 23 Departures 6 78 96 4 32 5 68 86 24 69 87 5 23 4 59 77 95 23 Departures Number of airports 5 4 3 2 99 999 absent very poor poor limited good very good 99 999 Competition Market concentration result: Fortress hubs, local concentration and hub premia (e.g. Lijesen et al., 2) Increased hub- competition and decreasing transfer dominance? Concentration levels of an flights on an airports Transfer dominance in Entropy-coeffici Source: Frenken et al.
Hub-competition JFK-BCN AMS LHR BRU CDG JFK ZRH MXP BCN MAD FCO FRA LIS Research questions How to measure market concentration in transfer markets without the availability of transfer passenger data? To what extent do an carriers dominate their transfer markets? What are the most important determinants of transfer dominance? Transfer dominance in Weighted indirect connectivity (WNX) Total number of indirect connections (per indirect route, geographical market, airport or airline) meeting criteria of minimum and maximum connecting time, weighted by: Transfer time Routing factor Indirect connectivity measure based on insights from the connectivity measures of Veldhuis (997) (CNU) and Bootsma (997) (CNX) weighted indirect 2.4* TI + RI WI = 3.4 TI: Transfer index RI: Routing index Tj: Maximum connecting time connection j Th: Transfer time TI Th T j connection: WI Transfer index.9.8.7.6.5.4.3.2. EUR- EUR = WNX= SUM WI EUR- ICA 2 3 4 5 6 7 8 9 2 3 Transfer time in hours ICA- ICA Herfindahl index indirect H connectivity = ( x / ) 2 i i xi X i is the weighted indirect connectivity of an airline in a certain transfer market (e.g. JFK-BCN)
JFK-BCN market 999 (weekday) Airline Hub WNX Market share SN BRU.65.6 TP LIS.57.2 SR ZRH.99.7 IB MAD.25.35 KL AMS 2.9. BA LHR 2.56.9 LH FRA 2.44.7 AF CDG 3.6.2 AZ FCO/MXP.9.4 TOTAL 29.564. Source: OAG Herfindnahl=.8 Moderately concentrated AMS LHR BRU CDG JFK ZRH MXP MAD FCO FRA LIS BCN Transfer dominance in Herfindahl values at indirect routes (one-stop, EU airline hubs only) Avg. Herfindahl index number of routes 99 999 99 999 intra-eu.88.86 329 5764 EU-ICA (outbound).86.82 2363 726 ICA-EU (inbound).83.8 276 7 ICA-ICA (directional).89.8 889 383 Total.86.83 8997 2363 herfindahl index Distribution Herfindahl index,9,8,7,6,5,4,3,2, EU to ICA market, 999 EU other EUR North-America Latin-America Asia Pacific Middle East Africa Latin-Am. 99 2 4 6 8 2 4 6 8 2 22 24 26 28 3 Percentile Asia-Pac 99 32 34 36 38 H=,5: TRB concentrated city-pair limit 4 42 44 46 48 5 Development market concentration transfer routes (one-stop, EU airlines) Overall decrease of market concentration on transfer routes EU airlines still dominate their transfer markets to a large extent. In 999, 7% of indirect routes monopolized, % not concentrated according to TRB standards North- America and Asia-Pacific markets most competitive Transfer dominance in
Determinants of market concentration on transfer routes Potential market size/ gravity (Borenstein, 992; DOT, 99, Goetz, 992) Distance between endpoints (e.g. Borenstein, 992; Dot 99; Goetz, 22) Regulatory regime (e.g. DoT, 999) Airport concentration at endpoints of the transfer route (e.g. DoT, 99) Presence of direct service Network strategy of carriers serving endpoints: HS versus PP (e.g. DoT, 99; DoT, 999) OLS regression: market concentration on intra- an transfer routes (EU airlines only) Coefficient R 2 change Gravity (log),297 ***,2 Distance (sqrt),24 ***,66 airport concentration endpoints Herfindahl dep airport (log) Herfindahl arrival airport (log) -,75 *** -,64 **, Direct connection -,4 *, available Dummy other,5, *á=.; ** á =.5; *** á =. dependent: Herfindahl (transformed inverse), h<, R 2 =,8 OLS regression market concentration on EU<>ICA transfer routes (EU airlines and hubs only) Coefficient R 2 change Gravity (log),424 ***,24 Distance (sqrt),53 **, airport concentration endpoints,2 Herfindahl dep airport (log) -,273 *** Herfindahl arrival airport (log) -,247 *** Direct service available,77 ***,5 Market region (dummy) North-America (North- America=ref.) Latin-America -,3 ** Asia-Pacific,2 Middle-East -,3 Africa,9 *** *á=.; ** á =.5; *** á =. dependent: Herfindahl (transformed inverse), h<, R 2 =,33,9 Transfer dominance in Conclusions Use of WNX measure of indirect connectivity allows for the analysis of market concentration on transfer routes without availability of transfer passenger data Various studies show an increase in hubdominance at local hub-routes in and the US Market concentration at indirect routes via EU airline hubs decreased between 99 and 999 but is still high Market potential, distance and airport concentration at endpoints most important determinants for transfer dominance Research agenda Work in progress Vailidity of regression-analysis: endogeneous versus exogeneous variables herfindahl index and normality herfindahl= model How to include network strategies of carriers and regulatory regimes? Which variables are missing? Include 2-stop transfer routes and non-eu airlines/ non-eu airline hubs Include global alliances and recent data
.25 3.25.75 6.25 4.75 9.25 7.75 2.25.75 5.25 3.75 8.25 6.75 2.25 9.75 22.75.25 3.25.75 6.25 4.75 9.25 7.75 2.25 5.25 8.25 2.25.75 3.75 6.75 9.75 22.75 Number of weighted indirect connections per airport per day Spatial concentration WNX 9858 8727 5 45 4 35 3 25 2 5 5 999 99 LH AF BA KL SR SK SN IB AZ BA LH AZ SK AY IB SK OS TP AF EI BU VM AF LH AY BA JK LH FI IW BA Network concentration.9.8.7.6.5.4.3.2.. 99 99 992 993 994 995 996 997 998 999 low-cost national regional extra-eu NC Wave-system structure British Airways London Gatwick BA at London Gatwick (red) SAS at Copenhagen (green) Air France at Marseille(red) Régional Airlines at Clermont- Ferrand (green) 3 3 3 3 2 6 78 96 4 32 5 68 86 24 69 87 5 23 4 59 77 95 23 Departures 2 6 78 96 4 32 5 68 86 24 69 87 5 23 4 59 77 95 23 Departure Cumulative sum of indirect connections 2 BA LGW SK CPH Cumulative sum of indirect connections 2 AF MRS VM CFE Time of the day Time of the day 99 999 999 999 >> Average dominance (%) at indirect route markets EU<>ICA, 999.9.8.7 Average transfer dominance.6.5.4.3.2 hub AMS CDG FRA LHR EU North-America Asia Pacific Africa other EUR Latin-America Middle East market region