Presentation - Competition and Regulation in Network Industries

Intermodal competition: studying the
pricing behavior of the French Rail
Patricia Perennes, RFF Centre d’économie de la Sorbonne
Introduction 1/2
In France, the railway national passengers’ transport is still characterized by a
monopoly. However, is SNCF’s pricing behavior the one of a monopoly?
 For certain service there is a strong intermodal competition of air and road
 Train tickets’ prices are not totally freely set by SNCF (price regulation)
The fact this price cap based regulation exists is also an opportunity for
an economist to analyze how a transportation company facing intermodal
competition sets its prices.
Usually, such an analysis is hard to conduct since transportation price are set
following yield management principles (Antes et al. (2004), Bergantino
This article also gives some insight on the type of competition that would
better suit passengers’ rail transportation
 The European commission wants to liberalize this industry
 Which kind of competition is better suited for this industry (open access vs.
Introduction 2/2
This article analyses SNCF’s pricing strategy on most of the O&D it
operates from/to Paris with HS trains, taking into account the limited
leeway that the company enjoys to set its prices because of price
Two analyses
 Empirical analysis of SNCF’s pricing behavior on most of the
origin/destination pairs (O&D) it operates from/to Paris with highspeed (HS) trains
 Qualitative analysis based on prices data for 19 selected routes
and for 2 time periods
This article relies on two unique data sets entirely collected for the
present study
Intermodal competition
 Rail network:
 HS trains covering most of mainland
France (even if HS tracks network is
much more smaller)
 Pricing behavior: for HS trains yield
management to a certain extent,
price based on the number of
kilometers for “regular” trains
 Airlines:
 Air France has quite a dense
network, but the number of routes it
offers has decreased since the 80s
because of the competition of the HS
 Pricing behavior: Yield management
 Road network:
 A well developed motorway network, with relatively expensive tolls
(fixed prices monitored by the state)
 A secondary network free of charge
Intermodal competition
Regulation of train tickets’ prices
Regular trains:
 “Kilometric reference”
 BF_ICi= A*kmi + B
 kmi is the number of kilometers for the O&D i
 A and B are numeric constants set each year by SNCF and approved
by the French Secretary of Transportation (SoT)
 Reduction coefficient may be applied
HS trains:
 System is more complex: state monitoring and yield management
 State monitoring: price cap
 Yield management: price can be freely set above this price cap
 The price caps are choosen by SNCF (and approved by SoT) for
each O&D
 This price cap cannot deviate too much from the “kilometric
reference” (40% leeway)
 (1-40%)( A*kmi + B)  BF_TGVi  (1+40%)( A*kmi + B)
Empirical analysis
We use the ratio between actual basic fare and kilometric basic fare
This ratio should be included between 0.6 and 1.4 (in reality between 0.9 and 1.39)
If this ratio is below 1, that means the tickets is relatively cheap compared to other
 Strong competition ?
If this ratio is above 1, that means the tickets is relatively expensive compared to other
 Low competition ?
To conduct this empirical analysis, we use a data set entirely collected for the present
 To calculate the endogenous variable R, we collect the 172 basic fares and the yearly
constants A and B, the number of kilometers (tariff kilometers)
 For the exogenous variables we collect driving times, flying times, train riding times,
driving costs (gas and tolls), driving times to the closest airport, numbers of tracks
kilometers, track access charges, numbers of psgers in the destination station, etc.
Panel data model:
 Data are available for 6 years (2007-2012)
(i: city pairs, t: years)
 Explanatory variables (intramodal competiton):
 Relative driving time (driving time/train riding time)
 Existence of a plane alternative (base on the relative total length of the travel
wen flying)
 Driving costs (tool fees and gas expense)
 Existence of a LCC service
 Control variables:
 Number of yearly passengers in the destination stations
 Track access charges
 “Region” (all destination cities were grouped in 9 “regions”)
Results 1/2
Explanatory variables’ coefficients are significant
 The existence of an air alternative to train impacts train ticket price
 Decrease of €6 to €8 for a €100 ticket if there is an airline alternative
 The existence of a low cost carrier service decreases train ticket price
 Tickets are less expensive if the relative driving time is small (i.e. close to 1)
 For a €100 ticket for a service where driving or taking a train have similar
duration, the price would have been €9 to €12 more expensive if the train
was two time faster.
 The most expensive the driving cost per kilometer, the higher the ratio
 For a €100 ticket an increase of €0.05 of the driving cost per kilometer
(the average driving cost per kilometer is €0.16) would lead to an increase
of €3 to €5 euros.
“Region” is not significant
The coefficient associated to the annual numbers of passengers is negative.
 Passengers getting off in important stations pay relatively less than
passengers getting off in smaller one.
Access charge, seems counter intuitive
 Endogeneity?
Results 2/2
Plane alternative
Relative car duration
LCC alternative
Cost per km by car
Price per km access charge
Ln Passengers
r2 within
r2 between
r2 overall
OLS (1)
More than one third of the variation is explained with R2 around 0.35-0.48
Qualitative analysis 1/2
Given the nature of the data at disposal, the analysis conducted in this
part is more qualitative than quantitative
 The goal of this analysis is to corroborate the previous empirical analysis
Based on another data set, also entirely collected for this study:
 Prices were collected during three months on SNCF’s, LCC’s and Air France’s
 Simple comparison of prices evolution for 19 O&D (Air France, LCC and
This qualitative analysis boils down to 3 “rules”
1. SNCF’s and Easy Jet’s prices seem correlated
2. Air France’s prices are usually much higher than Easy Jet’s/SNCF’s prices.
However a strong increase in Air France’s prices is usually followed by a small
increase in SNCF’s prices
3. A few days before the train/plane departure SNCF’s prices hit the maximum
price set by regulation, therefore taking the train is a much cheaper option
than flying
Qualitative analysis 2/2
Main result:
 The regulation effectively impact SNCF’s pricing beavior:
 On the 151 TGVs in our data set, 133 TGVs hit the cap set by the
 A few days before departure, taking the train is therefore a much
cheaper option than flying (for 100% of the destinations)
Regarding intramodal competition:
 LCC:
 Price difference between LLC and train tickets are low on average
€20.40 (regular train ticket are between €79 and €122)
 Air France:
 Less conclusive. A graphical comparison of prices series also
indicates that a strong increase in Air France’s prices is usually
followed by a small increase in SNCF’s prices. However, this effect is
hard to summarize in a synthetic index.
Is SNCF’s pricing behavior the one of a monopoly ?
 SNCF adapts its price depending on the potential intermodal
competition it faces
 Prices regulation effectively restricts SNCF’s ability to set its prices
Food for thought in the context of the rail industry liberalization:
 Intermodal competition dampers the monopolistic behavior of the
railroad incumbent. Need for an intramodal competition for all the
Regarding French price regulation:
 It definitively has an impact on SNCF’s price behavior
 But what is exactly the goal of this legislation? Public service? Antitrust?