Rail access charges and internal competition in High Speed Trains Óscar Álvarez San-Jaime Pedro Cantos Sánchez Rafael Moner-Colonques José J. Sempere-Monerris University of Valencia 2nd MEETING ON TRANSPORT ECONOMICS AND INFRASTRUCTURE IEB Barcelona January 21 2016 1

Motivation  In July 2012 the Spanish government announced a plan with different

measures to liberalize the rail industry:  To separate the main operator (RENFE) in four differentiated companies:

passenger, freight, wear and tear and rolling-stock.  ADIF remains as the public entity that owns and manages the infrastructure.  To favor the entry of private companies:  In freight sector (already possible but with poor results).  In long-distance services (HSR lines and other commercial corridors), although

this plan has been stalled.  Subsidised services (local and regional services) will be redefined and

maintained (perhaps through franchising systems). Also this measure has been dismissed.

2

Motivation  Apart from this plan, there is an important debate on the level of rail

track access charges:  ADIF has historically complained about the low charges, and got significant

increases in 2012 and 2013. Even so, the losses for ADIF in 2014 were around €230 million.  RENFE has objected that this increase is inefficient, because the new charges are above the marginal infrastructure costs.  Potential entrants argue that high charges will make entry difficult to new operators.  Public administrations are forcing to introduce “break-even” constraints in most of the public entities.

3

4

Overview of the presentation 1. Introduction:  Related literature

 Objectives and contribution.

2. The theoretical model 3. Calibration of the model  Description of the calibration process  Results

4. Conclusions and policy recommendations

5

Introduction  Competition “for the market” (by franchising systems) has been the

traditional way to introduce competition in the passenger rail industry.  Germany, Sweden and the Netherlands introduced franchising systems for

many regional and subsidised services.  UK extended the franchising system to all the passenger network.  Competition “in the market” has been occasionally promoted :  Some corridors in UK (London-Birmingham, Peterborough and

Cambridge, or London-Hull).  In Sweden (Stockholm–Gothenburg–Malmö) and Germany (Hamburg– Cologne), but entrants supply low price/low quality services.  In other European corridors rail entrants offer similar high quality services to the incumbent (Prague-Ostrava, Vienna-Salzburg or Milan-Naples). In the latter, Trenitalia competes with NTV: number of services , prices  and global rail traffic  (around 15%). 6

Related Literature  Preston et al (JTEP, 1999) use the PRAISE software designed to predict

the effect of competition between operators, by simulating the decisions on a sample of individuals:  They consider different competition scenarios between the two operators.  Only scenarios of “cream skimming” and “fare reductions” are generally feasible

with competition.

 A paper by Johnson and Nash (JRTP&M, 2012) uses an improved

version of the PRAISE software to model open access competition on a HSR international route:  ”On-track” competition has benefits to users in terms of fares and services.  There is an important loss of profitability for the incumbent.  Only entry is feasible if it leads to a notable cost reduction and additional traffic

is generated.  Then, wouldn’t it be better to franchise? 7

Related Literature  Adler et al (TR-B, 2010) show that strategic interaction between operators is key

to assess transport investments.  Ivaldi and Vibes (JTEP, 2008) analyze inter- and intra-modal competition in the transport industry, using a game theory approach.  Consumers choose a transport mode and an operator to travel with; operators

strategically decide on prices for the services.  They conclude that the entry of low cost operators can notably increase the levels of Consumer Surplus (CS).  Cantos et al (ET, 2015) define a model where there is competition between air

travel and HSR in frequencies and prices. The model is simulated for the routes Madrid-Sevilla and Madrid-Valencia:  Entry provokes increases in CS, but an important loss in rail profitability.  Entry is found to be welfare enhancing only when it generates large increases in

traffic (around 25-30%).

8

Introduction: objectives and contribution  Our paper develops a theoretical model where we introduce as novel

features:

 Rail operators that compete in prices and number of services, but now access fees

for the use of the rail infrastructure are endogenous.  The model is solved for different rail structures: vertical integration versus separation between rail infrastructure and operations.

 The main goal of the paper is:  To study the interplay between the access fee and competition variables chosen by operators  To analyze the effects of entry of a new train operator in an HSR corridor both on competition and access fees.  And specially, to evaluate the impacts on industry profitability, consumer surplus and social welfare under different market structures and for different Spanish corridors. 9

The theoretical model  We consider the standard deterministic quadratic and separable utility

function for a trip: 1 U  y   rQr  acQc  (brQr2  bcQc2  2dQrQc ) 2

 where we assume there is one train operator company (TOC) and one

outside option (private car).  And r = ar + vr nr, where vr stands for the per unit utility of one more service, and nr is the number of rail services.  Maximization of U subject to the corresponding budget constraint yields

a system of inverse demand functions. By inverting the system, the direct demand equations (for car and train) are:

10

Qr 

ar bc  ac d bc vr nr bc d   p  pc r 2 2 2 2 br bc  d br bc  d br bc  d br bc  d

Qc 

acbr  ar d dvr nr br d   p  pr c br bc  d 2 br bc  d 2 br bc  d 2 br bc  d 2

The theoretical model  In the case of entry a new rail operator the three demand functions (for

car and the two TOCs) is obtained from a new utility function as: 1 U  y   r1Qr1   r 2Qr 2  acQc  (brQr21  brQr22  bcQc2 )  d (Qr1Qc  Qr 2Qc  Qr1Qr 2 ) 2

 We solve a three-stage game:  In the first-stage the infrastructure operator selects the access per service fee, fr , to maximize its profit: i  ( f r  tr )nr  Fi  Then, the profit maximizing train operator chooses:  Number of services in the second stage,  And prices are set in the third stage.

 The train operator cost structure is as follows: TCr  cr n 2r  f r nr  Fr 11

The theoretical model  We solve the model for different scenarios before and after the entry of

a new rail operator. Before entry:  The first one considers a vertically integrated monopoly (VIM).  The second one presents the case of a vertically separated monopoly (VSM).

 And the scenarios after entry:  A vertically integrated dupoly (VID): there is VI and entry of a new rail

operator  A vertically separated duopoly (VSD): there is VS and entry of a new rail operator  A last scenario (VRD) is defined assuming that access charge is equal to the rail infrastructure cost, tr.

12

Calibration of the model  Values for elasticities are taken from González-Savignat (2008) and Cantos et al

(2009):

Own –price elast.

Cross-price elast.

Own frequency elast.

Train

-0.75

0.12

0.15

Car

-0.30

0.12

-

 Data for current traffic, price and frequency for each mode, per day and

direction in 2011 in Madrid-Valencia and Madrid-Sevilla corridor are: MAD-BCN Rail traffic point to point per year Rest of internal traffic per year Average rail price per passenger Train services per day and direction Total car passengers per year Car price trip

13

2,428,118 2,738,092 €105 27 4,728,500 €80

MAD-VLC 1,925,000 448,000 €75 15 3,062,000 €57

MAD-SVL 2,140,942 656,298 €88 18 3,100,000 €65

Calibration of the model  With these data a system of four equations is defined: 

bc pr  0.75; br bc  d 2 qr

d pc  0.120; br bc  d 2 qr



bc vr nr br pc  0.150;   0 . 30 ; br bc  d 2 qr br bc  d 2 qc

 And values for br , bc , d and vr can be recovered: MAD-BCN MAD-VLC MAD-SVL

br 0.021 0.033 0.031

bc 0.042 0.048 0.056

vr 3.472 4.750 4.514

d 0.009 0.010 0.011

 Then, knowing the current traffic levels and the number of rail services,

the values of ar ,aa and cr are finally calibrated: ar

14

ac

cr

MAD-BCN

207.68

406.48

362.65

MAD-VLC

126.78

285.14

339.35

MAD-SVL

145.58

329.74

285.14

Calibration of the model  And the cost for the rail infrastructure owner: Ci  tr nr  Fi  To separate variable infrastructure from fixed costs we use different

information directly from ADIF and literature (Wheat et al, 2009).  From the current figures, we can approximate tr = 1,400 and and and Fi per day and direction will be as shown below.  Then in the cost function for the rail operators: TCrj  (cr nrj  f r ) nrj  Fr  Again we need a calibration for Fr. Using data provided from RENFE and a

breakdown of the rail operating costs provided by Crozet and Chassagne (2013), fixed costs per day, direction and corridor are: MAD-BCN Fixed inf. costs (Fi) Fixed oper. costs (Fr)

15

MAD-VLC

MAD-SVL

€66,143

€39,244

€47,675

€132,208

€38,177

€54,845

MAD-BCN Train price inc.

Results

VIM 115

VSM 92

VID 89

VSD 66

VRD 80

VSD pessim 56

VRD pessim 67

1

0.80

0.77

0.57

0.70

0.49

0.58

53

66

80

56

67

# trains inc.

23.38

10.30

20.58

8.48

17

# trains entr.

4.23

10.30

20.58

8.48

17

Train price entr.

# total trains

27

13.5

27.61

20.60

41.16

17

34

1

0.50

1.02

0.76

1.52

0.63

1.26

# pass. inc.

5492

4044

4883

3408

4101

# pass. Entr

3251

4044

4883

3408

4101

Total rail pass. Road traffic Access fee Incumb. Profits

6044

4812

8743

8088

9766

6816

8202

1

0.80

1.45

1.34

1.62

1.13

1.36

6477

6742

5896

6037

5675

6311

6013

1

1.04

0.91

0.93

0.88

0.97

0.93

1400

6916

7076

6055

1400

5238

1400

1

4.94

5.05

4.33

1.00

3.74

1.00

260658

148961

-4280

33637

74178

-13464

13711

-6847

33637

74178

-13464

13711

Entrant’s profits ADIF’s profits Consum Surplus Social Welfare

16

-66143

8323

90571

29750

-66143

-897

-66143

1636935

1509729

1805475

1718285

1914955

1594068

1730476

1

0.92

1.10

1.05

1.17

0.97

1.06

1831450

1667013

1884920

1815309

1997168

1566243

1691755

1

0.91

1.03

0.99

1.09

0.86

0.92

MAD-BCN

VIM

VIS

Entrant’s profits

VID VSD Fr is 25% lower for the entrant 107230 26205

VRD

VSD pessim

VRD pessim

66689

19588

46763

Total oper. prof.

260658

148961

21925

100326

181408

6124

60474

ADIF’s profits

-66143

8323

90571

29750

-66143

-897

-66143

Total rail profits

194515

157284

112497

130076

115265

5227

-5669

Consum. Surplus

1636935

1509729

1805475

1718285

1914955

1594068

1730476

1

0.92

1.10

1.05

1.17

0.97

1.06

1831450

1667013

1917972

1848361

2030220

1599295

1724807

1

0.91

1.05

1.01

1.11

0.87

0.94

99741

52640

79815

Social Welfare

Fr is 50% lower for the entrant 140282 59257

Entrant’s profits Total oper. Prof.

260658

148961

54977

133378

214460

39176

93526

ADIF’s profits

-66143

8323

90571

29750

-66143

-897

-66143

Total rail profits

194515

157284

145549

163128

148317

38279

27383

Consum. Surplus

1636935

1509729

1805475

1718285

1914955

1594068

1730476

1

0.92

1.10

1.05

1.17

0.97

1.06

1831450

1667013

1951024

1881413

2063272

1632347

1757859

1

0.91

1.07

1.03

1.13

0.89

0.96

Social Welfare

17

MAD-VLC Train price inc.

VIM 75

VSM 57

VID 63

VSD 45

VRD 57

VSD pessim 38

VRD pessim 47

1

0.76

0.84

0.60

0.76

0.51

0.63

38

45

57

38

47

# trains inc.

12.92

5.78

11.56

4.53

9.05

# trains entr.

3.49

5.78

11.56

4.53

9.05

Train price entr.

# total trains

15

7.5

16.41

11.56

23.12

9.06

18.10

1

0.50

1.09

0.77

1.54

0.60

1.21

# pass. inc.

2178

1574

1981

1298

1617

# pass. entr.

1314

1574

1981

1298

1617

Total rail passengers Road traffic Access fee Incumb. profits

2438

1859

3492

3148

3962

2596

3234

1

0.76

1.43

1.29

1.63

1.06

1.33

4194

4316

4025

4045

3874

4161

4027

1

1.03

0.96

0.96

0.92

0.99

0.96

1400

3740

3769

3424

1400

2985

1400

1

2.67

2.69

2.45

1.00

2.13

1.00

47320

20996

-6568

1694

12739

-10372

-3781

-5579

1694

12739

-10372

-3781

Entrant’s profits ADIF’s profits

-39244

-21694

-369

-15847

-39244

-24884

-39244

Consumer Surplus

628439

590168

664809

633190

689340

602412

638474

1

0.94

1.06

1.01

1.10

0.96

1.02

636515

589470

652294

620732

675574

556784

591668

1

0.93

1.02

0.98

1.06

0.87

0.93

Social Welfare

18

MAD-SVL Train price inc.

VIM 88

VSM 68

VID 72

VSD 52

VRD 65

VSD pessim 43

VRD pessim 54

1

0.77

0.82

0.59

0.74

0.49

0.61

41

52

65

43

54

# trains inc.

15.84

6.87

13.74

5.52

11.04

# trains entr

3.61

6.87

13.74

5.52

11.04

Train price entr.

# total trains

18

9

19.45

13.74

27.48

11.04

22.08

1

0.50

1.08

0.76

1.53

0.61

1.23

# pass. inc.

2754

1947

2426

1627

2013

# pass. entr

1556

1947

2426

1627

2013

Total rail passeng.

Road traffic Access fee Incumb. profits

3010

2315

4310

3894

4852

3254

4026

1

0.77

1.43

1.29

1.61

1.08

1.34

4110

4243

3881

3940

3756

4063

3915

1

1.03

0.94

0.96

0.91

0.99

0.95

1400

4358

4411

3934

1400

3437

1400

1

3.11

3.15

2.81

1.00

2.46

1.00

75164

35238

-11663

2981

18697

-13703

-3784

-11916

2981

18697

-13703

-3783

Entrant’s profits ADIF’s profits

-47675

-21053

10889

-12858

-47675

-25187

-47675

Consumer Surplus

752938

698850

807880

764454

844190

721079

774435

1

0.93

1.07

1.02

1.12

0.96

1.03

780426

713034

795190

757558

833909

668486

719193

1

0.91

1.02

0.97

1.07

0.86

0.92

Social Welfare

19

Conclusions and policy recomendations  Separation without entry is not a good policy, because it leads to a

reduction in prices, frequencies and lower industry profitability and consumer surplus.  Entry of a new TOC has the following effects,  On the profitability of the industry:  Profits for the entrant are very sensible to the access fee. BCN-MAD is the most

profitable corridor for the entrant.  Incumbent losses profits with the entry  On CS and welfare:  Entry is generally benefitial for the consumers.  Final effect on welfare depends on the scenario considered and the increase in

the rail demand provoked by entry.  Scenarios with marginal infrastructure cost pricing maximize welfare, but losses increase for the infrastructure owner ADIF. 20

 Thank you very much for your attention!!!

21