1. The Impact of Rail Fares Complexity on Demand
Paul Metcalfe (PJM Economics)
Chris Heywood and Rob Sheldon (Accent)
Tony Magee (ATOC)
Paul Metcalfe
Tel +44 (0)

2. Rail fares complexity: pros and cons
A complex fare structure benefits customers as it offers the opportunity to find the preferred combination of cost, speed, flexibility, and comfort.
But complexity may also become a barrier to ticket purchase for customers who are not willing or able to apply the necessary time and effort to evaluate all the options.
The current fare structure in the UK is perceived as complex.
Which? (2011): Only 1% of rail passengers interviewed correctly identified all the various restrictions of Advance, Off-Peak, and Anytime tickets
SPAFT (2011): 15% stated they had previously purchased a ticket that was not valid on the train they were taking and therefore had to purchase another ticket
MW - Change to 1st bullet

3. Relevant previous work
Accent 2013 sought to determine the optimal number of fares and conditions for a flow in terms of maximising demand.
Found that consumers can ‘buy into’ increased complexity if they can see that this provides cheap deals and this is the price area that they focus on.
Relevant but excluded Advance fares.
SPAFT 2011 did incorporate Advance fares and found a broad lack of understanding or awareness of Advance tickets and restrictions.
But did not quantify the elasticities of demand or revenue with respect to simplification or appraise the desirability of alternative fares regimes.
MW - Changed to clarify the position in relation to guidance

4. Purpose of this study Overall objective Specific objectives
To understand the impacts on passenger demand of simplification of the fares structure, with a particular focus on Advance fares
Specific objectives
To estimate whether the typical Advance fares structure coexisting within the overall fares regime has any adverse impact on demand
To determine if a more optimal Advance fares structure may exist that differs from the typical Advance fares structures offered currently
The overall objective of the study was to look at the potential traffic impacts of the proposed Sizewell C construction on the wellbeing of local communities in an innovative and comprehensive manner which complements traditional modes of transport assessment. The study results are intended to inform Suffolk County Council in their discussions with EDF Energy on means of addressing the impacts.
More specifically, the research aimed to achieve the following objectives:
To review approaches to assessing the social and community impacts of changes in traffic flows (especially HGVs) on people within directly affected communities, identifying best practice and also any correlations between pre-construction perceptions and actual experiences during construction.
To research, through appropriate engagement activities with individuals in communities living (relatively) nearby the route, the perceived effects of an increase in traffic flows on the B1122 and the A12 through Yoxford, having regard to any previous comparable experiences.
To analyse the type, scale and range of impacts that are envisaged to arise, to examine the consistency and consensus in such views and to identify any correlations between those views and respondent characteristics such as their location and socio-economic profile.

5. Survey sample Pilot study 1 119 observations Pilot study 2
Main stage
1041 observations
Web survey
Route: London-Leeds
93% bought a train ticket in the last month
22% of these considered making the trip by car

6. Survey design: matching real world complexity
Ticket type choice
Do not travel by train
Train service choice

7. Alternatives OR Advance Anytime 58 different ticket types Offpeak
Super-offpeak
25 possible train services
Preferred departure/arrival time + 12 earlier and later services (increments equal to the base frequency for the route in question)
Respondents who chose an off-peak or super-offpeak tickets, or a ticket that is specific to a company do not see all the options
Single/return, first class/standard, valid on any company/valid on specific train-operating companies OR
“Not train” alternative

8. Complexity levels
The alternatives shown depend on a ‘complexity’ attribute, with four levels 1
As now: The price of advance tickets varies from service to service. 2
The price of advance tickets is the same whichever train service is booked. 3
Advance tickets will no longer be offered for sale and so passengers would have to choose one of the other ticket types, or not to travel by train at all. 4
A more expensive Advance is added to the offer that is valid in a time band rather than just on a specific service. Current-style Advance tickets would also be available in this scenario.

9. Other attributes Relative to the base fare Fare
9 levels, from -20% to +20%, in 5% increments
Duration
Relative to the base duration

10. Analysis: Nested mixed logit model
Lower level
train service choice
One separate choice for each leg (outward and return)
Variables: fare, duration, deviation from ideal departure/arrival time
Log sum term
The expected utility from the choice of train service if a ticket type is chosen, depending on the restrictions regarding the train operating company they can be used on and on travel on off-peak or super off-peak times
Upper level
ticket type choice
One choice for each trip
Variables: fare, ticket can be used on any service (for each leg), log sum term (each leg), “not train” alternative, “not train” alternative interacted with complexity levels 2, 3, and 4

11. Results: train service choice
Variable
Coef.
Std. error
P>|z|
95% Conf. Interval
Fare
-0.114
0.003
0.000
-0.120
-0.108
Duration (hours)
-0.428
0.063
-0.551
-0.304
number of hours after ideal departure/arrival time
-0.898
0.050
-0.996
-0.801
number of hours in excess of 2 hours after ideal departure/arrival time
-2.349
0.100
-2.545
-2.153
number of hours before ideal departure/arrival time
-1.926
0.069
-2.063
-1.790
number of hours in excess of 2 hours before ideal departure/arrival time
-1.496
0.128
-1.746
-1.245

12. Results: ticket type choice
Variable
Coef.
Std. error
P>|z|
95% Conf. Interval
fare
-0.035
0.001
0.000
-0.036
-0.034
ticket can be used on any service in the outward leg
-1.653
0.072
-1.794
-1.512
ticket can be used on any service in the return leg
-2.181
0.098
-2.374
-1.988
logsum (outward leg)
0.275
0.019
0.239
0.312
logsum (return leg)
0.146
0.017
0.113
0.180
Not train
-9.605
0.468
-8.688
Not train * Complexity level 2 (all advance tickets same price)
1.056
0.216
0.632
1.480
Not train * Complexity level 3 (no advance tickets)
4.799
0.401
4.013
5.585
Not train * Complexity level 4 (flexible-advance tickets available)
-0.086
0.222
0.698
-0.522
0.349

13. Results: share of preference Origin: Leeds, Destination: London
Ticket type
Complexity levels 1
(As now) 2
(All advance tickets same price) 3
(No advance tickets) 4
(flexible-advance tickets available)
Advance
Single
43.27%
45.60% 0%
18.92%
1st class
30.23%
22.93%
12.63%
Advance Flexible
42.63%
16.00%
Anytime
4.15%
4.53%
8.46%
1.48%
1.19%
1.29%
2.38%
0.42%
Return
4.61%
7.48%
1.47%
1.78%
1.79%
3.18%
0.62%
Off-Peak
7.11%
7.88%
13.38%
2.50%
1.86%
1.61%
0.53%
Super Off-Peak
1.41%
2.16%
0.47%
5.33%
6.77%
7.49%
2.18%
Not train
0.43%
1.35%
53.86%
0.16%

14. Preliminary conclusions
If all advance tickets had the same price, the impact on overall rail demand would be negative, but likely to be small. The impact on the share of the different ticket types would also be small
If advance tickets were not available, the demand for rail travel from Leeds to London would fall more than 50%
New “flexible advance” tickets could take more than 50% of the market

15. Next steps
Segment results (users vs. non-users, frequent vs. non-frequent users, journey purpose)
Calibration of results to reflect real-world market shares of train vs. other modes and of the different train ticket types
Design tool to simulate the impacts on rail demand and revenue for hypothetical changes in the availability of ticket types and fares