1. AN ASSESSMENT OF THE DESIRABILITY OF ON-TRACK COMPETITION: THE IPSWICH-LONDON ROUTE First Conference on Railroad Industry Structure, Competition and Investment, Toulouse, November 7-8, 2003 ANTHONY GLASS e-mail:aglass@its.leeds.ac.uk

2. 1. INTRODUCTION A higher or lower service frequency Desirability depends on the accompanying fare changes

3. 2. BARRIERS TO SERVICE COMPETITION Mechanism for the moderation of competition Approach to track access charging Content of a franchise agreement Capacity constraints

4. 3. ENTRY STRATEGIES Vertical product differentiation Horizontal product differentiation Niche market entry Mixed strategies

5. 4. GAME THEORY-SEQUENTIAL EQUILIBRIUM When a strategy is optimal for the remainderof the game based on a rational interpretation of previous play and a rational expectation of future play

6. 5. GAME THEORY-REPEATED GAME ACROSS SUBSETS First version of the chain-store game M+1 players; 1,…,M; 1,…,P; M  [1, P]; M<P; and m 1 enters p 1

7. 6. GAME THEORY-REPEATED GAME WITHIN A SUBSET Second version of the chain-store game V  [1,  ) and V+1 players T=max[S+(p-1)]

8. 7. GAME THEORY-REPUTATION ACQUISITION Ad hoc assumptions Fight is defined as a fare cut and accommodate is defined as no fare change or a fare increase ; Bayes theorem; weighted by elasticity and size of cut

9. 8. TWO PLAYER SERVICE FREQUENCY GAMES 1. Entrant Higher/Lower No Move 2. Incumbent Fight Accommodate

10. 9. PAY-OFFS (Entrant, Incumbent)Pay-offs (No Move, No Move)(0, 0) (Higher, Fight) ( , -  +  ) (Higher, Accommodate) ( , -  ) (Entrant, Incumbent)Pay-offs (No Move, No Move)(0, 0) (Lower, Fight) ( ,  +  ) (Lower, Accommodate) ( ,  )

11. 10. SOLVING THE TWO PLAYER GAMES A weighted or unweighted average  t is calculated  t <  t * , ,  and  e.g.  0 such that  +  > 

12. 11. CASE STUDY Buffer Station B (Brighton) Buffer Station A (Norwich) London Ipswich

13. 12. DATA Network information- timetable, fares, ticket shares by operator and base demand for each OD pair DETR (2000)1997-1999 modal shares Tastes and preferences- generalised journey time elasticities of demand and rail service valuations Variable cost per train km and variable cost per train hour

14. 13. EMPIRICAL METHODOLOGY A sample of 100 individuals Monopoly scenarios and two timetable changes 10% and 20% fare increase and decrease, and no fare change Profit maximising fare combination Nash-Bertrand equilibrium and sequential equilibrium

15. 14. SOCIAL WELFARE EquilibriumFares  CS  PS  SW Monopoly 20%FC ,20%FF , 20%RF ,20%SF  11.2%42.4%53.6% NB (L) FGE:No ,AR:10%  -17.5%-6.8%-24.3% NB (H) FGE:20% ,AR:20%  27.5%5.1%32.6% SE (L)  tj FGE:10% ,AR:10%  -17.2%-7.5%-24.7% 1-  tj FGE:No ,AR:10%  -17.5%-6.8%-24.3% SE (H)  tj FGE:10% ,AR:20%  26.5%6.4%32.9% 1-  tj FGE:No ,AR:20%  26.0%7.1%33.1% SE (LA)  tj FGE:10% ,AR:No  -15.4%-7.7%-23.1% 1-  tj FGE:No ,AR:No  -15.7%-7.0%-22.7%

16. 15. OPERATOR PROFIT Equilibrium  FGE  AR  SC  PS NB (L)-4.49%-1.84%-0.44%-6.8% NB (H)4.16%0.02%0.95%5.1% SE (L)  tj -4.50%-2.58%-0.44%-7.5% 1-  tj -4.50%-1.80%-0.44%-6.8% SE (H)  tj 4.14%1.30%0.95%6.4% 1-  tj 3.90%2.27%0.95%7.1% SE (LA)  tj -4.67%-2.60%-0.44%-7.7% 1-  tj -4.66%-1.91%-0.44%-7.0%

17. 16. REPUTATION ACQUISTION AR Fare Change No  10%  20%  FGE Fare Change 10%  -0.82%-5.42%-9.83% No  -3.90%-13.30%-32.30% SE(H) represents a costless investment and the biggest possible contribution to reputation acquisition Reputation acquisition costs so AR will probably play no change

18. 17. CONCLUSIONS More head-on competition can result in a rise in social welfare More uncertainty about the revenue stream of a TOC Limited spare capacity