Thredbo 9 Lisbon 59 september 2005 9 th

Thredbo 9, Lisbon, 59 september 2005 9 th International Conference on Competition and Ownership in Land Passenger Transport Ownership, Contractual Practices and Technical Efficiency: The Case of Urban Public Transport in France William ROY (LET, CNRS - Université Lumière Lyon 2) Anne YVRANDE-BILLON (ATOM, Université Paris 1 Panthéon-Sorbonne)

Introduction (1) • UPT in France : a dramatic evolution Ø An increasing operating deficit (€ 1. 6 billions in 2002). . . 1975 Receipts / Operating costs 1992 2002 80% 55% 34% Ø … that leads to an impressive amount of subsidy (€ 3, 2 billions in 2002, including investment expenditures) Thredbo 9, Lisbon, 5 -9 september 2005 2

Introduction (2) • Context : the European wind of reform concerning local public transport regulation • In France, networks are operated with : § Several governance structures and ownership regimes : Direct Administration, semi-public or a private company § Several risk sharing and regulatory contracts : Concessions, fixedprice, fixed-price on costs only, cost-plus • Organisational and regulatory frameworks have an effect on UPT performances ? § Ownership and competition influence performances ? § Contractual choices have an impact on performances ? Thredbo 9, Lisbon, 5 -9 september 2005 3

Choice of a regulatory scheme : make-or-buy ? Competition ? No Local Authority In principle Ex ante Incentives ? Direct public administration - Mixed company + Private company ++ Thredbo 9, Lisbon, 5 -9 september 2005 Management contract Gross cost contract Net cost contract 4

A typology of urban public transport delegation contracts (risk sharing) Cost risk borne by Local Authority Revenue risk borne by Operator Management Contract : = e s = se + (re-r)-(ce-c) Gross Cost Contract : = e - (c-ce) s = se + (re-r) Net Cost Contract : = e + (r-re)- (c-ce) s = se Operator “e” means expected, as opposed to realised “ ” is the operator’s profit ; “s” is the amount of subsidies “c” is the operating costs ; “r” is the commercial revenues Thredbo 9, Lisbon, 5 -9 september 2005 5

Local Authorities choices Modes of organisation en 2002 Modes of delegation en 2002 Direct public Administration 10% Delegation to a mixed company 21% Concession 2% Delegation to a private company 69% Net cost contracts 51% Management contract 20% Gross cost contract 27% (in % of the number of networks) Thredbo 9, Lisbon, 5 -9 september 2005 6

Whose performance is it ? Fares Operator’s performance LA’s performance (short term) Network’s performance Service specifications Environment LA’s performance (long term) Infrastructures Thredbo 9, Lisbon, 5 -9 september 2005 7

Efficiency and production process Inputs : Labour, capital, energy… Productive efficiency Efficiency Supply : Levels of services : quantity et quality Commercial efficiency Consumption : passenger trips Thredbo 9, Lisbon, 5 -9 september 2005 8

Frontier model : Battese et Coelli (1995) • Stochastic Frontier Analysis : a production function • Consider the production function • Where TE is the level of technical efficiency with • TE is a function of a set of explanatory variables W including the regulatory scheme y • A flexible functional form : translog y* • We estimate simultaneously : y. A Frontier : Ø Ø parameters b of the production function parameters d that explain TE Thredbo 9, Lisbon, 5 -9 september 2005 A x x* x. A 9

Hypothesis tested • Ownership structure and performance § Hypothesis 1 a: Private operators show higher technical efficiency than public ones § Hypothesis 1 b: Semi-public operators show higher technical efficiency than public ones, but lower technical efficiency than private operators • Contracts and performance § Hypothesis 2 a : Cost-plus contracts (management contracts in our typology) provide lower technical efficiency than fixed-price contracts (either gross contracts or net cost contracts) § Hypothesis 2 b : Net cost contracts and gross cost contracts are equivalent in terms of incentives to technical efficiency Thredbo 9, Lisbon, 5 -9 september 2005 10

Data and variables • 135 UPT networks between 1995 and 2002 Ø Networks without rail systems (only buses) Ø Networks serving more than 30 000 inhabitants • Inputs and control variables Ø Labour : driving and non-driving (including temporary and subcontracting) Ø Energy : consumption measured in equivalent diesel m 3 Ø Capital stock : number of vehicles Ø Length of network lines Ø Population • Organisational variables explaining inefficiency Ø Ownership : Direct administration / semi-public firm / private firm Ø Type de contract : C+ / GCC / NCC Thredbo 9, Lisbon, 5 -9 september 2005 11

Descriptive statistics Thredbo 9, Lisbon, 5 -9 september 2005 12

Models to be estimated • We estimate the following translogarithmic production frontier (we omit the cross-products between the inputs X and the control variables Z). • The technical inefficiency effect is defined by : § Ownership regimes model (1) § Regulatory contracts model (2) Thredbo 9, Lisbon, 5 -9 september 2005 13

Estimation results (Software FRONTIER 4. 1) • Model 1 : { PUBLIC ; MIXED ; (PRIVATE) } § PUBLIC & MIXED < PRIVATE (H 1 a) § MIXED < PUBLIC ( H 1 b) • Model 2 : { MANAG ; GROSS ; NET ; (PUBLIC & MIXED) } § § PUBLIC & MIXED < NET et GROSS (H 1 a) PUBLIC & MIXED <ns MANAG ( H 1 a) MANAG < NET & GROSS (H 2 a) NET < GROSS ( H 2 b) Thredbo 9, Lisbon, 5 -9 september 2005 14

Inefficiency scores Thredbo 9, Lisbon, 5 -9 september 2005 15

Conclusions • Our estimations support the conjecture that technical efficiency cannot be measured independently of the regulatory constraints. • Expected effects : § Private operators outperform public ones § Operators under cost-plus contracts exhibit a higher level of technical inefficiency than operators under a fixed-price contract • Unexpected effects : § A semi-public company is the worst organisational choice in terms of technical efficiency § Gross cost contracts are better than net cost contracts ? Thredbo 9, Lisbon, 5 -9 september 2005 16