ArticlePDF Available

Abstract and Figures

The railway system is a fundamental component of the economy of most countries, since it has the capability to transport every day millions of passengers as well as millions of dollars' worth of goods from origin to destination. According to many empirical CO2 emissions, is one of the most environmentally friendly and safe transportation modes. Moreover, it is appreciated for its high energy efficiency. One of the most common and frequent issue about rail transport is the concept of arrival time, that is of punctuality and delay. The aim of the paper is twofold: it firstly provides a critical literature review on delay categories as a starting point for the development of a new, easy and complete classification responsibility. Secondly, applying this classification, a panel data analysis with fixed effects has been performed to understand motivation and responsibility of the delay on an important Italian railway line. Moreover, the application of survival analysis is used to understand the failure time probability of a train journey and to estimate the percentage of trains that arrive to a destination.
Content may be subject to copyright.
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
The importance of punctuality in rail transport
service: an empirical
investigation on the delay determinants
Daniele Grechi
, Elena Maggi
Università degli Studi dell’Insubria, Dipartimento di Economia
The railway system is a fundamental component of the economy of most countries, since it has the
capability to transport every day millions of passengers as well as millions of dollars’ worth of goods
from origin to destination. According to many empirical works and papers, rail, which produces very low
CO2 emissions, is one of the most environmentally friendly and safe transportation modes. Moreover, it
is appreciated for its high energy efficiency. One of the most common and frequent issue about rail
transport is the concept of arrival time, that is of punctuality and delay. The aim of the paper is twofold: it
firstly provides a critical literature review on delay categories as a starting point for the development of a
new, easy and complete classification of delay based on the link between motivations, causes and
responsibility. Secondly, applying this classification, a panel data analysis with fixed effects has been
performed to understand motivation and responsibility of the delay on an important Italian interregional
railway line. Moreover, the application of survival analysis is used to understand the failure time
probability of a train journey and to estimate the percentage of trains that arrive to a destination.
Keywords: Punctuality, rail transport service, railway delay model.
1. Introduction
An efficient rail system is an important element for the development of the economic
activities of a specific country or region. Economic exchanges, trade development, the
possibility of improving communications and travel are the basis of railway
development (Ponti and Beria, 2007; Beria and Grimaldi, 2017).
Over the time the expansion of rail track has favoured an increase in the productivity
of different industries and in the accessibility and competitiveness of various cities and
regions with the opportunity for a face-to-face communication process for knowledge
production (Kobayashi et al. 1997). As sustained by Romer (1986), knowledge is a non-
rival partially excludable good that can be available for firms or individuals through an
exchange process that happens crosswise the spatial networks (Batten et al. 1989).
However, the use of this transport mode is often curbed by the problem of low
punctuality of many trains. Thus, the concept of arrival time (Sahin 1999; D’Ariano et
Corresponding author: Daniele Grechi,
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
al. 2008), i.e. of punctuality and delay, is one of the key issues to be afforded. The
improvement of rail punctuality can help to promote the modal shift from the more
pollutant modes to the rail, which is one of the most environmentally friendly
transportation systems (Goverde 2005; Lackhove et al., 2011). Punctuality has been
defined in literature as: “The ability to achieve a safe arrival at a destination to an
advertised timetable” (Glyee 1994) or “a feature consisting in that a predefined vehicle
arrives, departs or passes at a predefined point at a predefined time” (Rudincki 1997).
As a consequence, the total delay is given by the difference between the scheduled time
and the effective time (Mattsson 2007) and it is a crucial topic in the daily operational
business of any transportation company (Huisman et al., 2005). For example, the shared
use of the same infrastructure by different railway services (high speed, freight transport
and local service), with different origins and destinations, speeds and halting patterns, is
probably the main reason of the propagation of delays throughout the network
(Vromans et al., 2006; Bergantino, 2015). Moreover unreliability, and the consequent
delay, happens when there are deviations from the official timetable, getting worse the
customer level of service and inducing a probable modal shift (Rietveld Bruinsma and
Van Vuuren, 2001; Olsson and Haugland, 2004; Freling et al., 2005).
The aims of the paper are: (i) to propose a new classification of delay based on the
link between motivation, causes and responsibility, on the basis of the literature review
results; (ii) applying this classification, to better understand motivation and
responsibility of the delay on a specific interregional Italian line. A panel data analysis
with fixed effects has been performed; the model variables represent technical elements
of the train (engine, weight, rank etc.) and other features of the journey (load factor,
direction, seasonality, number of stops, etc.). The main research question is which
factors positively or negatively affect the performance of a train journey, in terms of
punctuality (difference, in minutes, from the effective arrival time and the scheduled
Moreover, a survival analysis (Jardine et al., 1989; Andersson and Björklund 2011;
Andersson et al., 2012; Andersson et al., 2016) has been applied for the period 2013-
2016 on the same railway line, in order to understand the survival rate of the analysed
The paper is structured as follows. The next section presents the rail passenger
situation, the rate use and the different thresholds of delay used to identify the train
performance in Italy and in other European countries. Subsequently, a literature review
on punctuality and delay is provided and the major existing delay categories are
identified. In section 4 a new delay classification is proposed, that is then used in the
regression model described in paragraph 5. The following section describes the survival
analysis by a theoretical point of view and applies it to the same Italian railway line,
which has been considered in the regression model. The paper ends with some
concluding remarks.
2. Delay and use of rail in Italy: a European comparison
The Italian railway network is 17,000 kilometres long: the ratio between railway
network and motorway network is higher than in Spain and similar, but lower, than in
France (Policicchio 2007; 2017). Although there is a relevant number or railway
lines in Italy compared to other European countries, the Italians travel by train far less
than in Germany, France and the UK (Albalate et al., 2015).
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
According to the most recent and available data (2014, from RMMS and ERADIS, the average value of the indicator rail kilometres per
inhabitant is less than 1000 in UK, 1126 in Germany, 1359 in France and 804 in Italy
(Cartmell 2016). Even if in Austria, Sweden and Denmark people travels more by train
than in Italy, in other European countries trains are used less on average with respect to
other transport modes (in particular, car). In recent decades, Italy has hardly invested in
new railway lines to create a high-speed and/or high capacity network, with the purpose
to increase the quality and quantity of rail trips, making transport flows more
sustainable (Banister 2000; Giuntini 2006; Bergantino et al., 2013). This network has
the undoubted advantage to promote the shift of the demand for transport from road to
rail, and the introduction of new quality standards also thanks to the pressure of new
competitive operators (Curtis and Low 2013; Pendolaria 2016). Although the user of
short or medium distance rail transport services has different characteristics and needs
than the high-speed customer, there is the need to improve the quality also of these
services, in particular in terms of punctuality that is a key issue for commuting
passengers (De Luca and Pagliara 2007, Beria et al., 2016). Due to the historical and
geographical connotations of Italy, both high speed and “normal” railway services are
fundamental components of the national rail system that involves about 21.3 million
passengers every day (Relazione sulla qualità dei servizi, Trenitalia 2015) of which
more than 5 million of commuters (Pendolaria 2016). As regards delay, the latest Report
on the quality of Trenitalia's services (2016) indicates that the percentage of all
categories of trains that had more than one hour of delay was less than 1%. Moreover, in
2015 the 91.6% of medium and long-distance trains (so called Frecciarossa,
Frecciargento, Frecciabianca, InterCity and InterCity Night) and 97.9% of regional
trains were on time. According to the study on the prices and quality of rail passenger
services by Cartmell (2016), among the countries with large rail networks, in Germany
and Italy the delays in long-distance trains are higher than in other countries. In both
countries, less than 75% of the trains were punctual in 2014. However, the data on
punctuality of the European railways are not perfectly comparable, because the delay
threshold, in minutes, varies by the country according to the type of traffic and purchase
modality. The punctuality threshold in Germany is lower than in Italy, particularly on
long distance services, as shown in the table 1. In fact, the Italian railway company
(, 2017
) considers on time the passenger trains that arrive at destination with a
delay lower than 15 minutes for long distance services (including interregional ones)
and less than 5 minutes for short (regional) distance services. As regards freight trains,
the punctuality threshold is based on 30 minutes. These thresholds are frequently higher
than in other European countries.
Table 1: Delay and threshold in Europe
Regional Services
Distance Services
More than 5 minutes
More than 5 minutes
More than 2 minutes and 29 seconds
More than 4 minutes and 59 seconds
More than 5 minutes and 59 seconds
from 5 to 15 minutes due to the category of
RFI, that is the company of FS group which manage the railway infrastructure, calculates the
punctuality threshold for the traces purchased at least 5 working days in advance the date of utilization.
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
the journey
More than 5 minutes and 59 seconds
More than 5 minutes and 59 seconds
More than 5 minutes
More than 5 minutes
More than 3 minutes
More than 5 minutes
from 3 to 10 minutes due to the category of
the journey
from 5 to 10 minutes due to the category of
the journey
More than 5 minutes
More than 5 minutes
More than 5 minutes
More than 10 minutes
More than 5 minutes
More than 15 minutes
Sources: Personal Elaboration on various sources
3. Literature review: punctuality and delay in railway transportation
3.1 Punctuality and reliability definition
In the transport sector, focusing specifically on the rail sector, punctuality is an
important indicator to understand if the planned travel time is optimal. Dealing with
delays is a crucial issue in the daily operational business of any public and private
transportation company (Schöbel 2009). Some studies (Harris and Godward 1992; Bates
et al., 2001; Cavana et al., 2007) have shown that it is a critical element that companies
need to take into account for managing their service and it is a measure of the
operations’ reliability and performance (Veiseth et al., 2007). In fact, deviations from
scheduled time reduce the level of service (Dingler et al., 2010; Nagy and Csiszár 2015;
Olsson and Haugland 2004). Punctuality is a complex indicator and not only a simple
parameter to be taken into account. From the railway point of view (supply side), it is
useful to measure the service quality level and to understand if the infrastructure, even
in bad condition, is able to guarantee the connections. From the passenger demand side,
instead, punctuality is a fundamental element to plan a journey especially in the case of
interchange of different transport modes (Nagy and Csiszár 2015). Landex (2008)
argues that when a train is delayed, passengers are also late, and this can influence their
life quality and their future transport modal choice. Carey and Carville (2003) underline
that the structure and the physical organisation of a railway station and the number of
people waiting at a quay for a given train are factors that potentially affect the
timeliness of travel. In the literature many definitions of punctuality are available.
According to Gylee (1994), punctuality is: “the ability to achieve a safe arrival at a
destination to an advertised timetable”. Otherwise Rudnicki (1997) defines punctuality
as a measured value that is able to indicate if a given known vehicle arrives or departs at
a specific point in a previously set time. Subsequently, Hansen (2001) defines the same
concept as a percentage of railway journeys that arrive or depart in a specified station of
a railway network no later than a specified time in minutes. Moreover, Olsson and
Haugland (2004) describe train non-punctuality as a deviation, usually negative, from
the defined timetable. Veiseth et al., (2007) give a definition of punctuality similar to
Hansen (2001), as the percentage of trains that arrive on time at their final destinations.
However, this percentage is considered as a reductive indicator by Olsson and Haugland
(2004) and by Bititci and Veiseth (2005) because some other useful data are hidden
(such as the delay and the recovery time for an intermediate stop). Mattsson (2007) use
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
a mathematical equation to define the concept of total delay based on the difference
between the effective time and the minimum time scheduled. Noland and Polak (2002)
focus their attention on travel time variability that is a measurement of the uncertainty
of trip journey times in transportation, and introduce in this concept also delays, early
arrivals and cancellations. According to Nystrom (2005), punctuality is an agreement
between passengers and the company, one of the most important components of the
measured quality of the service. Passengers put high expectations on the reliability of
train schedule, which strongly influences the positive perception of the travel (Salkonen
and Paavilainen 2010). Harris and Godward (1992) shows that the reliability of the
arrival time is often more important than the train rapidity. The shared use of the same
infrastructure by different railway services (high speed, regional, interregional and local
service, long haul and freight at the same moment), with different origins and
destinations, different speeds, and different halting patterns, is probably the main reason
for the propagation of delays throughout the network (Vromans et al., 2006). However,
the specific delay volume relationship is dependent on the traffic mix on a route
(Dingler et al., 2009; Krueger 1999; Bronzini and Clarke 1985). Different train types
have different operating characteristics influencing the total delay that a train
experience. Heterogeneity in these train characteristics causes additional conflicts,
increasing delays (Dingler et al., 2009; Pachl 2002; Leaflet 2005; Abril et al., 2008).
3.2 Categorization of the delay
In literature there are some authors sustaining that unreliability, and the consequent
delay, happens when there are deviations from the official timetable (Bruinsma Rietveld
and Van Vuuren 1999; Rietveld Bruinsma and Van Vuuren 2001). Unscheduled delays
can be caused by numerous events including: mechanical failures, malfunctioning
infrastructure, weather conditions, excessive boarding times of passengers, accidents at
highway-railroad grade crossings, etc. (Vromans et al., 2006; Carey 1999). Delays may
be divided into different categories, but the terminology differs between different
authors. Regarding the size of delays, Gylee (1994) defines primary delays as the delays
with the greatest impact, while secondary delays as delays that are a consequence of the
primary ones. In this last case, the delay of a train spread to the others that are
following, causing a phenomenon that is called “cascading effect” by Dingler et al.,
(2010). The terms primary and secondary delays are used differently in Norway:
according to Veiseth et al. (2007), secondary delays indicate the delays caused by other
delayed trains, while primary delays are caused directly by the train, not considering the
influence of the other ones. Gibson et al., (2002) instead call exogenous delays the
primary delays of Glyee (1994) and reactionary delays the secondary delays defined by
Veiseth et al. (2007), but with more emphasis on the interaction between different train
operators. Carey (1999) underlines that there is a difference between exogenous delays
and knock-on delays. The first ones are due to events such as failure of equipment or
infrastructure, delays in passengers boarding or alighting and they are equivalent to the
concept of secondary delays developed by Glyee (1994). The second ones are directly
related to a failure of the train.
Higgins et al. (1995) classify the delay, combining different causes at the same
moment. They identify three categories of delay:
Track related delay: it occurs when a train have a slowdown caused by track
problems or a sudden and unexpected stop (e.g. infrastructural problems).
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Train dependent delay: caused when a train is forced to slowdown in a line
section for reasons other than track problems (e.g. locomotive failure).
Terminal/scheduled stop delay: delay that happen in a scheduled stop and is
related to loading/unloading, train connections, fuelling and crew problems.
Müller-Hannemann and Schnee (2009) focus their attention on the importance for
passengers, but also for railway companies, to have a real-time information system that
can up-to-date train status information and provide to a user valid timetable information
in the presence of disturbances. They decide to classify the delay, according to the
different possible motivations: disruptions in the operational flows, accidents,
malfunctioning or damaged equipment, construction work, repair work, and extreme
weather conditions like snow and ice, floods, and landslides. In their analysis, they
focus on the concept of real time information. The usefulness of immediate information
is crucial for the passenger who is able to find alternatives to reach the destination. For
example, in Germany, an online system manages every day 6 million of forecast
messages about timetable changes and also the latest prediction of the current situation.
Another classification is provided by Nelson and O’Neill (2000). By analysing the
U.S. railway lines in the period 1998-2000, they categorize the reasons of delay linked
to its nature, identifying (i) engineering causes (referred to tracks, structures, stations,
signal and communication instruments), (ii) mechanical causes about the rolling stock
and (iii) transportation causes regarding decisions of the railway manager, dispatching
procedure. In addition, they define other specific factors related to delay that are
construction work, problems related to passengers, extraordinary circumstances and
cascade delays deriving from the circulation of freight trains. Mechanical delay is a
component that is common for any transport operators and is representative of a failure
of a train component. Nelson and O’Neill (2000) found that the major causes in this
case are an engine failure, braking system and coach components problems. Moreover,
they highlight that 13% of the total delay is due to extraordinary events, such as weather
conditions, vandalisms problems with vehicles at crossing lines and police
interventions. The authors argue that passengers are not directly responsible for most of
the delays, but they are a contributing factor. For example, a train could be delayed by
the presence of an incremental extraordinary number of passengers deriving from a train
suppression or by the waiting for delayed passengers by the train crew. The influence of
the train stops in a railway station on the delay was also studied in deep by Harris
(2015), Harris Mjøsund and Haugland (2013) and Harris and Andersson (2007).
Analysing the dwell time, that is the whole process of train stop in station, they have
made some measurement about the duration of delays in station stops that concern the
entire process of boarding and alighting passengers.
A summary of the different types of delay classification provided by the literature can
be found in Table 2.
Table 2: Summary of the literature review on delay classification
Delay Classification
Scientific papers
Primary and secondary delay
Glyee 1994
Cascade delay
Dingler et al., 2010
Exogenous and reactionary delays
Gibson et al., 2002
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Exogenous delays and knock
on delays
Carey 1999
Track related, dependent and terminal stop delays
Higgins et al., 1995
Motivations of delay
Hannemann and Schnee 2009
Engineering, mechanical, transportation and other
extraordinary causes
Nelson and O’Neil 2000
Dwell time Harris 2015; Harris et al., 2013; Harris and
Andersson 2007
3.3 Other literature findings related to delay and punctuality
Olsson and Haugland (2004), using data on Norwegian railway and the Pearson
indicator, found a negative correlation between punctuality and the load factor of local
trains. In fact, when the load factor is high in peak hours or days there is less
punctuality, while in the non-working days and in non-commuting hours the punctuality
is better. They analyse also the relation between punctuality and cancelled trains for the
Oslo area, finding a positive correlation. In fact, given a certain railway infrastructure
capacity, possible traffic problems are due to broken trains on the line (Burdett and
Kozan 2006). Infrastructure capacity, in terms of number of trains on a specific line in a
predetermined time, is one of the elements that can influence a journey and its possible
delay. According to some authors (Dingler et al., 2009; Pachl 2002; Leaflet 2005; Abril
et al., 2005), the relationship between performance and infrastructure capacity is
negative: as the number of trains increases, the average delay rises, worsening the
performance. This relationship is clearly affected by the number of the tracks available
on a specific line. Moreover, it is possible to have adjunctive delays in crossing times in
a railway station with interchange binary located along a single-track line; in that case,
the delay regards not only the train itself but also all the trains traveling along the line in
a specific moment. A possible solution to compensate (small) delays is represented by
the recovery time. It is a procedure that add supplementary minutes to the total running.
The recovery time is decided by the rail companies and differs according to the
geographical location or country. Pachl (2002) distinguishes between regular and
special recovery time: the first one indicates the supplementary time usually added to
running time (as a percentage), while the second one is introduced when there are speed
restrictions due to maintenance work (on track, line, powerline, signal, informatics
components) or track malfunctions or problematic weather conditions. According to
Beyene (2012) and Kroon et al. (2014), since a temporary speed reduction can cause
delay that the train is not able to make up during its journey, there is the need to
reschedule the timetable on the lines that are involved in this reduction; otherwise a
delay is caused. Moreover, the eventual deceleration zones that can be required by a
speed reduction could cause a supplementary delay. Both speed limitations and the
unavailability of a sufficient number of platforms for all the trains are conditions that
can influence the dwell time and can raise the time for boarding and alighting
passengers. In the case of unavailable platforms or maintenance work, the role of the
circulation manager is very important, in reprogramming the traffic, using the computer
systems. He has to apply operational priority rules, taking two important decisions of
delay management: the “wait-depart decisions” and the “priority decisions”. The first
one is about the choice to maintain or not a connection in case of delay, while the
second one concerns the order in which a certain train is allowed to pass on a specific
track (Dollevoet et. Al., 2014). In other words, the normal scheduling timetable should
be modified giving priority to the most important trains, according to the commercial
agreements. These delay management decisions should be taken also in the case of
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
limited capacity of the tracks, as studied by Ginkel and Schöbel(2007) and
Schachtebeck and Schöbel (2010). In fact, if two or more trains use the same piece of
infrastructure (single track or double track), a priority rule should be given to one of
4. New classification proposal
As underlined by the literature review, there is not a common view in defining and
classifying the different types and determinants of the train delay. For this motivation
the following paragraph is dedicate to a systematic organization of the concept of arrival
delay with a classification that takes some insights from some of the works mentioned
above but focuses on the delay causes and its responsibility. This new classification will
be used into a regression model presented in the next section.
Five macro causes of the delay have been identified, as represented in Table 3.
As regards the first one, the concept of delay due to circulation causes is directly
linked to the concept of secondary delay expressed by Olsson and Haugland (2004) and
the concept of exogenous delay described by Gibson in 2003. In this case the delay of
the train is due to a delay of a preceding train. For example, the train B which follows
the train A is forced to stand still outside the station, because the train A uses more time
than planned for the boarding and alighting operations.
Table 3: Classification of different delay causes.
Causes of delay
Circulation problems
Train Failure
Infrastructure Failure
Preparation Delay
External Delay
Source: Personal elaboration
The second cause of delay is the train failure: as for other types of vehicles it is
possible that a train has a failure and is unable to resume his march (or it takes some
time to be repaired on site). This type of failure can occur in the station of departure or
during the journey. The possible causes may be represented by a failure of the
locomotive, problems with a door of a wagon or a malfunction of some train
As regards the third cause, it is possible that the railway infrastructure has mechanical
breakdowns (switches, tracks, power lines). This type of failure affects indirectly a train,
that has to wait for a despatch order to continue the planned journey. The preparation
delay occurs when the trains (engine and/or coaches) are not ready at the starting
station. The motivation is related to problems about the availability of the effective
material due to failure or absence of the corresponding train, if the material does not
arrive from the depot. It is possible that the train is delayed due to failure of the track or
the electric line, but in this case is classified as a failure of the infrastructure. The last
delay category refers to the external causes, as explained by Nelson and O’Neill in
2000. For this kind of situations (e.g.: intervention of law enforcement, strikes, seismic
and weather events, accidents not attributable to railway operators) the role of the
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
railway operator is secondary, but it is important to classify and to analyze this typology
to understand its incidence on the performance of a railway journey.
Each of these macro causes is linked to the identification of the responsibility source
(excluding the external delay). The model that will be briefly presented in the next
section relates to Italian reality, where there are two organisations responsible for the
RFI, that was established in July 2001 as the 'Infrastructure Company' of the
State Railways Group in response to the Community Directive transposed by
the Italian Government on the separation between the network operator and
the transport services provider; it is responsible for delays due to circulation
problems and infrastructure failure.
Trenitalia S.p.A., that is a 100% owned subsidiary of State Railways, and is
the leading Italian company for the management of both rail and passenger
freight; it has responsibility for train failure and departure delay.
On the lines where other rail companies compete with Trenitalia, such as Italo-NTV
firm, they can be also responsible for the second and fourth types of delay.
Moreover, in addition to above presented macro causes, in the regression model the
concept of physiological delay has been included. This is a variable that has been
introduced to check when there is a delay but there is not a precise motivation and refers
to all the mini-causes that may occur during a trip, such as a temporary failure at a door
or a slowdown due to previous trains, that are resolved quickly. This last classification
is relevant for the model because can allow us to classify also minor delays that are not
included in the above presented classification. In our case the physiological delay
counts all the delays between 5 to 9 minutes, but it is possible to apply with different
limits to another model. It is important to remark that in this model the delay is
associated to a specific cause only when a railway journey has more than 9’ of delay at
the arrival point. This is due to the availability of our data; the rail operator has not
provided the motivation for delay lower than 10 minutes. This model is adaptable to
other realities with a variation of the range of the performance (related to the concept of
delay of the rail operator of a specific country).
5 Regression model
5.1 Description of the model
The aim of the model is the validation of the proposed classification of delay and
subsequently the analysis of the value and weight of the different delay determinants
from a statistical point, using panel data. The regression model takes inspiration from
the study provided by Harris and Godward (1992), who applies a similar kind of
analysis to verify which factors would affect the delay of a generic train journey using
UK data of the late 80’. They found that distance covered, and train length were
statistically significant in determining punctuality. For them it would be realistic to
expect the increase in delay to be proportionate to the route length.
The model here presented is applied to a well-defined sample of railway journeys in
working days commuting hours (from 6.00 am to 9.00 am and from 16.00 pm to 20.00
pm) in the period 2013-2016. The data, which have been collected from the official web
site, in addition to the information provided by Trenitalia Long-
Haul refer to 16 trains along the Milan-Genoa line (a total of 15,600 observations). The
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
data concern the characteristics of the line, the trip and the train, considering also its
performance in terms of time.
The normality of the performance data is confirmed and sustained by the analysis
made in previous papers (Goverde et al., 2001; Murali et al., 2010).
The model that is used in this paper is a panel data with fixed effect: a statistical
model where the parameters are fixed or non-random quantities, and the single
observations about n entities or individuals, or cross-sectional observations are
described for two or more moments over time (day, months, years) (Hsiao, 2014).
In panel data, in which longitudinal observations for the same elements, for a fixed
effect model exist, the dependent variable should be measured for each individual on at
least two occasions and those measurements could be comparable. The term fixed
effects estimator refers to an estimator for the coefficients in the regression model
including those fixed effects (Allison, 2005; 2009).
Table 4 presents all the variables that are included in the complete model, specifying
the typology and the related literature.
Table 4: Regression variables.
Variables References Format
Performance This variable represents the final arrival
time of the journey measured in minutes.
Harris and
Goodward, 1992;
Harris, 2007;
Harris et al., 2013;
Harris et al., 2015
Numerical, logarithm
of the performance
Causes of delay
Thess variable are related to the
motivation of the delay. There is a
variables per categories
Abril et al., 2005;
Burdett and
2006; Landex,
2008; Gibson et
al., 2002; Olsson
and Haugland,
binary-5 categories
ID_Rail Numerical
Model of locomotive
From official Trenitalia Data it was
possible to derive the real engine for each
The 3 locomotives are: e464, e444, e402
Harris and
Goodward, 1992;
Trenitalia Libro
servizi universali
binary-3 categories
Weight with
This Variable represent the weight of the
Harris and
Godward, 1992;
Trenitalia Libro
servizi universali
Features of the
journey (numerical
and binary)
Load Factor
This variable represents the estimated load
factor for each journey. The estimation is
from Trenitalia
Olsson and
Haugland, 2004;
Alwadood et al.
Costs This Variable is related to the cost of a
single trip in standard second Class.
Bergantino et al.,
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Number of_Stops Number of stops per journey per train
Vromans et al.,
2006; Harris 2007;
Harris et al., 2013;
Harris et al., 2015
Travel Time
Expressed in minutes it is related to the
planned travel time (official Trenitalia
Harris and
Godward, 1992;
Carey, 1999;
Bergantino et al.,
This variable is about the effective
relation of the journey (Genoa-Milan or
Harris and
Goodward, 1992;
Olsson and
Haugland 2004
Morning_Evening This variable is about the hour of the trip
(Morning or Evening)
Olsson and
Haugland 2004;
Skjæret 2002
This variable represents the ownership of
the train (Thello, Trenitalia, Trenitalia
Bentivogli and
Panicara, 2011 binary-3 categories
Season The season of the year related to the
journey considered
Dobney et al.,
2009; Olsson and
binary-4 categories
Day The day related to the journey (From
Monday to Friday)
Dobney et al.,
2009; Olsson and
Haugland, 2004
binary-5 categories
year-Date Numerical
Speed Restriction
The variable assumes value 1 if there is a
speed restriction on the track. This
variable is unique for all the track, so
assume the same value with 1 or more
speed restriction
Beyene, 2012;
Landex, 2008;
Olsson and
Haugland, 2004
Binary- 2 categories
Weather Conditions
The variable assumes value 1 if there is a
weather alert.
Dobney et al.,
Huisman and
Boucherie, 2001;
Mattsson, 2007
Binary- 2 categories
Source: Personal elaboration
5.2 Regression results
The results of the regression (see Table 5) confirm that all the causes of delay, that are
directly related to the logarithm of the performance, are statistically significant in
relation with the train performance (with a p value <0.001). According to the
coefficients, external causes (such as floods, suicides, fires, accidents at level) are the
typology of delay that affect more the performance of the train, followed by train
failure, and, with a similar coefficient departure delay and infrastructure failure (electric
line, rails, level passes). However, since the coefficients of these variables vary in a
small range (1.20-1.15), the impact of these types of motivation on delay is similar. The
circulation problems influence on delay, indeed, is smaller. In fact, even if the
circulation conflict occurs more frequently than the other causes, it causes minor
disadvantages in terms of minutes of delay. The last cause that is represented by
physiological delay has, logically, the smaller coefficient of the classification. This is
due to the fact that, as explained above, in this category there are only delays from 5 to
9 minutes.
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Table 5: Results of regression analysis with all the data (model 1)
Variable Coefficient
P-value Literature
Constant -12.6416
Circulation 0.82912 0.00 *** Verified
Train_failure 1.16354 0.00 *** Verified
1.15508 0.00 *** Verified
Preparation delay 1.15063 0.00 *** Verified
External causes 1.20505 0.00 *** Verified
Physiological delay
0.50107 0.00 *** Verified
Weight_Brake -1.12 0.01 *** Not in literature
Load_Factor 0.49 0.00 *** Verified
Costs 0.02 0.12 Not significant in
this model
Travel_time -0.03 0.00 *** Verified
Morning or Evening
0.05 0.17 Not significant
Winter -0.06 0.00 *** In contrast with
Summer -0.06 0.00 *** Verified
Monday 0.03 0.02 ** Verified
Tuesday 0.03 0.16 Not Significant
Wednesday 0.02 0.19 Not significant
Friday 0.04 0.006 *** Verified
Speed Restriction 0.077 0.09 ** Verified
Weather conditions
0.071 0.07 ** Verified
Additional information
Mean Performance 0.45
RMS Performance 0.60
R^2 LSDV 0.73
R^2 within 0.67
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Durbin-Watson 1.84
Source: Personal elaboration using Gretl ( Note: the variables not statistically
significant and not relevant for the analysis are not in the table
As regards the other determinants, confirming the findings of Olsson and Haugland
(2004), Alwadood et al. (2012) and Harris (2007), high load factor increases delay,
negatively influencing the train performance; in fact, the coefficient is statistically
significant. There is an inverse relationship between total travel time, which is
dependent on the trip distance, and delay. The data show that the regional trains have a
lower average delay compared to long-hauls trains with longer scheduled travel time.
The costs and morning-evening variables (MOR-EV, that indicates if the journey is
done in the morning or in the evening) are not significant in terms of p-value. As
regards winter season variable, the model results are in contrast with the findings of
Olsson and Haugland (2004), who sustain that the delay increases in winter. Probably
this is due to the characteristics of the line analysed and the weather conditions of the
regions. In summer season indeed the delay decreases, although this is not supported
by the literature. As claimed by Dobney et al. (2009) and Olsson and Haugland (2004),
in the initial and the ending working days of a week, Monday and Friday, the delay
results greater. Finally, the regression shows that both speed restrictions and weather
conditions negatively affect train performance, confirming the findings of Beyene
(2012) for the first variable and of Dobney et al. (2009), Huisman and Boucherie
(2001), Mattsson (2007) for the second one. In addition, other restricted models were
developed, by dividing the data into different categories, according to the direction of
the train (from North to South or viceversa) and to the departure time (morning or
afternoon). The following table shows the resulting coefficients.
Table 6: Results of restricted models and comparison with model 1 (coefficients value)
Variable Model 1: all
Model 2:
Model 3:
South- North
Model 4:
time in
Model 5:
time in
const 0.11 (***) 0.16 (***) 0.070 (***) 0.20 (***) 0.03 (***)
Circulation 0.83 (***) 0.78 (***) 0.88 (***) 0.78 (***) 0.73 (***)
Train_failure 1.16 (***) 1.10 (***) 1.22 (***) 1.11 (***) 0.82 (***)
Infrastructure_failure 1.15 (***) 1.14 (***) 1.17 (***) 1.06 (***) 0.91 (***)
Preparation delay 1.15 (***) 1.14 (***) 1.15(***) 1.13 (***) 0.89 (***)
External causes 1.21 (***) 1.17 (***) 1.24 (***) 1.13 (***) 0.87 (***)
Physiological delay 0.50 (***) 0.45 (***) 0.55 (***) 0.42 (***) 0.83 (***)
R^2 0.73 0.67 0.67 0.67 0.88
Source: Personal elaboration using Gretl (
It is possible to observe that there are very small differences between the general
model and the models considering only one direction. The results of model 4 and 5 are
much more interesting: in the morning the variables infrastructure failure and departure
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
delay are more relevant than the external causes variable and the other causes. The
external causes are more important in afternoon trips. The weight of the physiological
delay is higher in the morning than in the other cases. The R square is similar for the
first three sub-models while it is much higher for the morning trips model.
6 Survival Analysis
6.1 Definition of survival analysis and censoring
The survival analysis is a statistical method to analyse the expected duration of time
until a specific event happens. It may be applied to different issues, for example in the
epidemiological area, the event of interest may be the death of a patient or the relapse
following a disease or the response of a patient to a specific treatment. In general, in the
survival analysis literature, death or failure, is considered the "event of interest". It is
also called reliability analysis in engineering, duration analysis in economics, and event
history analysis in sociology (Miller. 2011; Kleinbaum and Klein, 2010; Cleves, 2008).
The first step in a survival analysis is the calculation of "survival time", as the
difference between the time to event occurred and the time of entry into the study of a
statistical unit and it is typically a positive number (Despa, 2010).
According to the general theory and concepts of survival analysis and model
estimation (Kiefer 1988; Lancaster 1990; Klein and Moeschberger, 2005), it is possible
to underline that:
Survival functions generate a hazard function for a consumer i, that describes
the probability of defeat at time t, that is indicated as hi(t).
The hazard function can be transformed into a survival function, which
represents the probability Si (t) that a consumer survives at time t conditioned
to the fact that it is "alive" at t-1 time, that is Si (t) = (Si (t-1) x 1-hi (t)), with
Si (1) =1
S(t) is constant in the time interval between two events. S(t) is a step function
that changes its value only if the event happens.
Time to event: The time between the subject's entry into the study until a
particular "outcome".
In this technique, some units of analysis are censured, i.e. removed from the
observation before failure, if for a certain period are no information, or when they leave
the study, or if the study ends before the outcome of interest is revealed. They are
counted as "alive" for the time they were followed in the study. Furthermore, it is
important to remember that dropouts are related to outcomes and treatment and they can
distort the results.
There are two different types of censoring: left and right. Left censoring occurs when
an observation is below a certain value, while right censoring when it is above a certain
value, but in both cases the exact value in unknown.
Examples of censored observations are (Klein and Moeschberger, 2005):
end of study
inability to follow the subject
the minimum time t until a subject "survived".
There are several fields in which it is possible to use and analyze data with this
technique (health, mechanical, railway). It is important to underline that the data
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
distribution is not normal but exponential, Weibull and log normal distribution, as
described in Table 6.
Table 7: Survival Analysis Distribution.
S(t) S
urvival function
Exponential distribution t
Weibull distribution
Log normal distribution
Sources: Miller, 2011; Carlin and Louis, 1997; Cox and Oakes, 1984. Note: ϕ is the cumulative function
of the normal
6.2 Applications of the survival analysis to the railway sector
To the best of our knowledge, in literature only few applications of the survival
analysis to the railway field can be found. The aims of these applications are very
different than those of our analysis, because they do not concern the rail delay issue.
Jardine et al., (1989) used survival analysis to determine the risk of failure of diesel
locomotives in Canada within the maintenance-related repair cost process. In their
paper, they have decided to concentrate their attention on checking the Weibull form of
the hazard function. They apply the failure time concept to the components of train
equipment which have a well-defined point of failure after a length of time. Moreover,
they consider some censored data that are determined when the engine has a motor
change for another reason (e.g. a scheduled overhaul), or when “failure” has still not
occurred before the end of the observation period. Grube-Carvers and Patterson (2015)
use survival analysis to test the relationship between urban rapid rail transit and the
beginning of gentrification in the three largest cities of Canada. More recently,
Andersson et al., (2016) used survival analysis to estimate the cost of renewal of railway
tracks. They used a sample of censored data containing nearly 1,300 observations on the
Swedish main railway lines. They use a Weibull distribution to understand the failure
time and develop a regression models to estimate the deterioration elasticities for total
tonnage as well as for passenger and freight tonnages separately.
6.3 Application of survival analysis to Milan-Genoa railway line
The survival analysis has been applied to 15,684 train data relating to the Milano-
Genoa line (equally divided in both directions) in the period 2013-2016 only in working
days. The failure event is represented by the suppression of the train, that consequently
does not arrive at the destination. The data censored to the right is represented by trains
that did not have the cancellation, because they have not final destination in Genoa; in
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
fact, there is a number of journeys that finish in other destinations such as
Ventimiglia/Nice Ville (West) or La Spezia/Livorno (East). It is important to underline
that the right censoring is not present in the South-North direction since all the trains
have their final destination in central Milan. The objective of this analysis is to test the
data and verified the results with the declared percentage of the quality reports produced
in the same years by Trenitalia (“Relazioni per la qualità del servizio di Trenitalia”,
2013, 2014, 2015, 2016). In these documents there are the data of trains that arrive to a
destination and the percentage of cancellations. The spatial survival analysis (Ibrahim et
al., 2005; Grube-Cavers and Patterson, 2015) can be very useful to check if there are
any points (e.g. climbs, mountains) that can cause train failure. This analysis is useful
also from a forecast point of view as it is possible, through estimates, to hypothesize
what percentage of trains will arrive at their destination. Due to the data it is possible to
understand the exact position of the train when there is the event (suppression) from a
temporal and geographical point of view. Although this sample is not representative of
the entire national railway system, the results of the analysis is even confirmed by the
data provided by Trenitalia in its quality reports. In fact, as shown by Figure 1, in all the
models the survival rate is close to 99% and Trenitalia declare that more than the 98%
of the analysed trains arrive at destination. It should be pointed out therefore that the
arrival is not related to the average train delay. Stata software was used to perform this
analysis with the STS function (Cleves, 2008; Lambert and Royston, 2009).
Figure 1: Survival Analysis 2013-2016.
Source: Personal Elaboration using Stata and Excel on official data from
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
7 Conclusions
One of the most challenging goal of nowadays is to make more sustainable the
growing passengers flows, rebalancing the modal shift in favour of train, that is less
pollutant and energy consuming than car. Since the value of travel time in transport
modal choice is often more important than the price, the issue of train punctuality is of
key importance. Understanding the most frequent delay causes and their relative weight
compared to the others can help the rail companies in identifying the most effective
strategies to improve punctuality, positively affecting the choices of the travellers.
The present paper gives a contribution to the existing literature in different ways.
First, since there is not a common view in defining and classifying the different types
and determinants of the train delay, a systemic classification of delay causes, and its
responsibility is proposed and developed on the basis of the results of the literature
review. Second, it is the first analysis on the delay determinants performed in Italy,
where there is a specific railway system, with different characteristics than other
European countries, in the use, investments and also in the definition of the thresholds
considered to identify if a train is on time or not. This analysis has been developed by
using panel data regression models, focusing on an important railway line of the
Northern Italy, connecting the most important Italian economic city (Milan) with
another urban area (Genoa) in which a key port is located. The regression analysis
includes more variables than the majority of the other works on this issue, indicating
which are the main causes of train delays, generally confirming the results of previous
works but also determining the importance of new factors and giving a weight to each
determinant for the considered line. Third, the paper applies – to the best of our
knowledge – for the first time the technique of survival analysis to determine the
probability of train arrival at the destination on the same line. The resulting survival rate
is confirmed by the empirical observations made by the railway company, Trenitalia, in
its quality reports, suggesting the statistically goodness of the analysed sample.
In the future it would be interesting to extend both the econometric and survival
analysis to other lines or, if the railway operator will be available to provide information
(since today this was not possible), to the whole Italian railway system, making
comparison between different lines, regions and periods of time. A possible other
application can concern the railway systems of other countries. Finally, another possible
step would be the consideration of the concept of seasonality in the survival analysis,
checking if the number of trains, that arrive to a destination, differs according to the
travel season (Dobney et al., 2009; Olsson and Haugland, 2004).
Abril, M., Barber, F., Ingolotti, L., Salido, M. A., Tormos, P., Lova, A. (2008) "An
assessment of railway capacity", Transportation Research Part E: Logistics and
Transportation Review
(5), pp. 774
Albalate, D., Bel, G., Fageda, X. (2015) "Competition and cooperation between high-
speed rail and air transportation services in Europe", Journal of Transport
, pp. 166
Allison, P. D. (2005) Fixed effects regression methods for longitudinal data using SAS,
Institute, Cary.
Allison, P. D. (2009) Fixed effects regression models (Vol. 160), SAGE publications,
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Thousand Oaks.
Alwadood, Z., Shuib, A., Hamid, N. A. (2012, April) "Rail passenger service delays:
An overview", Business Engineering and Industrial Applications Colloquium
2012 IEEE, Kuala Lumpur.
Andersson, M., Björklund, G. (2011) "Marginal railway track renewal costs: a survival
data approach", Working Paper 2012:29, Centre for Transport Studies,Stockholm, SE.
Andersson, M., Björklund, G., Haraldsson, M. (2016) "Marginal railway track renewal
costs: A survival data approach", Transportation Research Part A: Policy and
87, pp.
Andersson, M., Smith, A., Wikberg, Å., Wheat, P. (2012) "Estimating the marginal
cost of railway track renewals using corner solution models", Transportation Research
Part A: Policy and Practice
(6), pp.
Banister, D. (2000) "Sustainable urban development and transport-a Eurovision for
Transport Reviews
(1), pp. 113
Bates, J., Polak, J., Jones, P., Cook, A. (2001) "The valuation of reliability for personal
travel", Transportation Research Part E: Logistics and Transportation Review, 37(2-
3), pp. 191
Batten, D. F., Kobayashi, K., Andersson, Å. E. (1989) "Knowledge, nodes and
networks: an analytical perspective", In Knowledge and industrial organization,
, Berlin, Heidelberg.
Bentivogli, C., Panicara, E. (2011) "Regolazione decentrata e servizio concentrato: le
ferrovie regionali viaggiano su un binario stretto?", Banca d’Italia, Le infrastrutture in
Italia: dotazione, programmazione, realizzazione,
Bergantino, A. S. (2015), “Incumbents and new entrants", In: Rail economics, policy
and regulation in Europe, Edward Elgar Publishing Limited, Cheltenham.
Bergantino, A. S., Capozza, C., Capurso, M. (2013) "L’effetto della liberalizzazione
ferroviaria sulle politiche di prezzo delle compagnie aeree e ferroviarie. Evidenze
preliminari sui principali collegamenti ad Alta Velocità in Italia", Rivista di Economia
e Politica dei Trasporti, 3
(6), pp. 1
Beria, P., Grimaldi, R. (2017) "Reality and opportunities for on-track competition in
HSR", In: Evaluating High-speed Rail. Interdisciplinary Perspectives, Routledge,
Abingdon on Thames.
Beria, P., Redondi, R., Malighetti, P. (2016) "The effect of open access competition on
average rail prices. The case ofMilan–Ancona”, Journal of Rail Transport Planning &
(3), pp. 271
Beyene, M. (2012). Effect of speed reductions for train punctuality, Master's thesis,
Institutt for bygg,
anlegg og transport, Trondheim, NO
Bronzini, M. S., Clarke, D. B. (1985) "Estimating rail line capacity and delay by
computer simulation", Tribune des Transports, 2(1), pp. 5-11.
Bruinsma, F. R., Rietveld, P., Van Vuuren, D. J. (1999) "Unreliability in public
transport chains", Working Paper 98-130/3, Tinbergen Institute, Amsterdam, NE.
Burdett, R. L., Kozan, E. (2006) "Techniques for absolute capacity determination in
railways” Transportation Research Part B: Methodological, 40(8), pp. 616-632.
Carey, M. (1999) "Ex ante heuristic measures of schedule reliability" Transportation
Research Part B: Methodological, 33(7), pp. 473-494.
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Carey, M., Carville, S. (2003) "Scheduling and platforming trains at busy complex
stations", Transportation Research Part A: Policy and Practice, 37(3), pp. 195-224.
Carlin, B. P., Louis, T. A. (1997) "Bayes and Empirical Bayes Methods for Data
Statistics and Computing
(2), pp.153
Cartmell, J. (2016). Study on the Prices and Quality of Rail Passenger Services, Steer
Davies Gleave
, Brussels.
Cavana, R. Y., Corbett, L. M., Lo, Y. L. (2007) "Developing zones of tolerance for
managing passenger rail service quality", International Journal of Quality & Reliability
(1), pp. 7
Cleves, M., Gould, W., Gould, W. W., Gutierrez, R., Marchenko, Y. (2008). An
introduction to survival analysis using Stata, Stata press, College Station.
Cox, D. R., Oakes, D. (1984) Analysis of Survival Data (Vol. 21), CRC Press, Boca
Curtis, C., Low, N. (2016). Institutional barriers to sustainable transport, Routledge,
D’Ariano, A., Pacciarelli, D., Pranzo, M. (2008) "Assessment of flexible timetables in
real-time traffic management of a railway bottleneck", Transportation Research Part
C: Emerging Technologies
(2), pp. 232
De Luca, M., Pagliara, F. (2007). "La ferrovia nelle aree metropolitane italiane", Atti
del XIV Convegno nazionale SIDT,
Despa, S. (2010). "What is survival analysis", StatNews #78, Cornell University
Statistical Consulting Unit, Ithaca, NY.
Dingler, M., Koenig, A., Sogin, S., & Barkan, C. P. (2010, August) "Determining the
causes of train delay", AREMA Annual Conference Proceedings, Orlando.
Dingler, M., Lai, Y. C., Barkan, C. (2009) "Impact of train type heterogeneity on
single-track railway capacity", Transportation Research Record: Journal of the
Transportation Research Board
, (2117), pp. 41
Dobney, K., Baker, C. J., Quinn, A. D., Chapman, L. (2009) "Quantifying the effects of
high summer temperatures due to climate change on buckling and rail related delays in
east United Kingdom", Meteorological Applications: A journal of forecasting,
practical applications, training techniques and modelling
16(2), pp. 245
Dollevoet, T., Huisman, D., Kroon, L., Schmidt, M., Schöbel, A. (2014) "Delay
management including capacities of stations", Transportation Science, 49(2), pp. 185-
ERADIS European Railway Agency Database of Interoperability and Safety, (accessed 24 july 2018)
Freling, R., Lentink, R. M., Kroon, L. G., & Huisman, D. (2005) "Shunting of
passenger train units in a railway station", Transportation Science, 39(2), pp. 261-272.
Gibson, S. (2003) "Allocation of capacity in the rail industry", Utilities Policy, 11(1),
pp. 39
Gibson, S., Cooper, G., Ball, B. (2002) "Developments in transport policy: the
evolution of capacity charges on the UK rail network", Journal of Transport
Economics and Policy (JTEP)
(2), pp. 341
Ginkel, A., Schöbel, A. (2007) "To wait or not to wait? The bicriteria delay
management problem in public transportation", Transportation Science, 41(4), pp. 527-
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Giuntini, A. (2006) "Una storia che pendola. Successi e sconfitte dell'Alta velocità
ferroviaria in Italia", Memoria e ricerca, (23), pp. 1000-1029.
Goverde, R. M. (2005) Punctuality of Railway Operations and Timetable Stability
PhD. Thesis
, TUDELFT, Delft, NE
Goverde, R. M., Hansen, I. A., Hooghiemstra, G., Lopuhaa, H. P. (2001) "Delay
distributions in railway stations", In 9th World Conference on Transport Research,
Grube-Cavers, A., & Patterson, Z. (2015) "Urban rapid rail transit and gentrification in
Canadian urban centres: A survival analysis approach", Urban Studies, 52(1), pp. 178-
Gylee, M. (1994, September) "Punctuality Analysis-A Basis for Monitoring and
Investment in a Liberalized Railway System”, Proceedings of seminar held at the 22nd
PTRC european transport forum, Warwick.
Hansen, I. A. (2001, August). Improving railway punctuality by automatic piloting.
In Proceedings IEEE Intelligent Transportation Systems (pp. 792-797).
Harris, N. G. (2015) "A European Comparison of Station Stop Delays",
International Congress on,
Harris, N. G., Godward, E. W. (Eds.). (1992) Planning passenger railways: A
handbook, Transport Publishing Company Limited, London.
Harris, N. G., & Anderson, R. J. (2007). An international comparison of urban rail
boarding and alighting rates. Proceedings of the Institution of Mechanical Engineers,
Part F: Journal of
Rail and Rapid Transit
(4), 521
Harris, N. G., Mjøsund, C. S., Haugland, H. (2013) "Improving railway performance in
Norway. Journal of Rail Transport Planning & Management", 3(4), pp. 172-180.
Higgins, A., Ferreira, L., & Kozan, E. (1995) "Modelling single line train
operations", Transportation Research Record, Journal of the Transportation Research
Board, Railroad Transportation Research
, pp. 9
Hsiao, C. (2014) Analysis of panel data (No. 54), Cambridge university press,
Huisman, D., Kroon, L. G., Lentink, R. M., Vromans, M. J. (2005) "Operations
research in passenger railway transportation". Statistica Neerlandica, 59(4), pp. 467-
Huisman, T., Boucherie, R. J. (2001) "Running times on railway sections with
heterogeneous train traffic", Transportation Research Part B: Methodological, 35(3),
pp. 271
Ibrahim, J. G., Chen, M. H., Sinha, D. (2005) Bayesian Survival Analysis, Springer,
Jardine, A. K. S., Ralston, P., Reid, N., & Stafford, J. (1989) "Proportional hazards
analysis of diesel engine failure data". Quality and Reliability Engineering
(3), pp. 207
Kiefer, N. M. (1988) "Economic duration data and hazard functions", Journal of
economic literature
(2), pp.646
Klein, J. P., & Moeschberger, M. L. (2006). Survival analysis: techniques for censored
and truncated data. Springer Science & Business Media.
Kleinbaum, D. G., Klein, M. (2010) Survival analysis (Vol. 3), Springer, New York.
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
Kobayashi, K., Okumura, M. (1997) "The growth of city systems with high-speed
railway systems", The annals of regional science, 31(1),pp. 39-56.
Kroon, L., Maróti, G., Nielsen, L. (2014) "Rescheduling of railway rolling stock with
dynamic passenger flows", Transportation Science, 49(2), pp. 165-184.
Krueger, H. (1999) "Parametric modeling in rail capacity planning", In Simulation
Conference Proceedings ,
Lackhove, C., Brinkmann, F., Scheier, B., Mbakwe, I., & Böhm, T. (2011). Advancing
life-cycle-management for railway signalling and control systems. Towards Life Cycle
Sustainability Management
Lambert, P. C., Royston, P. (2009) "Further development of flexible parametric models
for survival analysis", Stata Journal, 9(2), pp. 265.
Lancaster, K. (1990) "The economics of product variety: A survey", Marketing
(3), pp. 189
Landex, A., Kaas, A. H., Nielsen, O. A. (2008) Methods to estimate railway capacity
and passenger delays, PhD Thesis, Technical University of Denmark (DTU), Lyngby,
Mattsson, L. G. (2007) "Railway capacity and train delay relationships", In: Critical
Springer, Berlin, Heidelberg.
Miller Jr, R. G. (2011) Survival analysis (Vol. 66), John Wiley & Sons, Hoboken.
Müller-Hannemann, M., & Schnee, M. (2009) "Efficient timetable information in the
presence of delays", In: Robust and Online Large-Scale Optimization, Springer, Berlin,
Murali, P., Dessouky, M., Ordóñez, F., Palmer, K. (2010) “A delay estimation
technique for single and double-track railroads", Transportation Research Part E:
Logistics and Transportation Review
(4), pp. 483
Nagy, E., Csiszár, C. (2015) "Analysis of delay causes in railway passenger
transportation", Periodica Polytechnica Transportation Engineering, 43(2), pp. 73-80.
Nelson, D., O’Neil, K. (2000) "Commuter rail service reliability: On-time performance
and causes for delays", Transportation Research Record: Journal of the Transportation
Research Board
, (1704), pp. 42
Noland, R. B., Polak, J. W. (2002) "Travel time variability: a review of theoretical and
empirical issues", Transport reviews, 22(1), pp. 39-54.
Nyström, B. (2005) Punctuality and railway maintenance, PhD Thesis, Luleå tekniska
universitet, Luleå, SE
Olsson, N. O., Haugland, H. (2004) "Influencing factors on train punctuality—results
from some Norwegian studies", Transport policy, 11(4), pp. 387-397.
Pendolaria Report (2016) "Situazione e scenari del trasporto ferroviario pendolare in
Rapporto Pendolaria
, Lega
mbiente, Roma.
Policicchio, F. (2007) Lineamenti di infrastrutture ferroviarie (Vol. 2), Firenze
University Press, Firenze.
Ponti, M., Beria, P. (2007) "La rotaia arrugginita e il vagone del futuro", il
(6), pp. 1028
Rete Ferroviaria Italiana RFI,, (accessed 24 july 2018)
Rietveld, P., Bruinsma, F. R., Van Vuuren, D. J. (2001) "Coping with unreliability in
public transport chains: A case study for Netherlands", Transportation Research Part
A: Policy and Practice
(6), pp. 539
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
RMMS Rail Market Monitoring, , (accessed 24
july 2018)
Romer, P. M. (1986) "Increasing returns and long-run growth", Journal of political
(5), pp. 1002
Rudnicki, A. (1997) "Measures of regularity and punctuality in public transport
IFAC Proceedings Volumes
, Nantes.
Şahin, İ. (1999) "Railway traffic control and train scheduling based on inter-train
conflict management", Transportation Research Part B: Methodological, 33(7), pp.
Salkonen, R., & Paavilainen, J. (2010, July) "Measuring railway traffic punctuality
from the passenger’s perspective", General proceedings of the 12th World Conference
on Transport Research (WCTR),
Schachtebeck, M., Schöbel, A. (2010) "To wait or not to wait—and who goes first?
Delay management with priority decisions", Transportation Science, 44(3), pp. 307-
Schöbel, A. (2009) "Capacity constraints in delay management", Public
(2), pp. 135
Skjaeret, J. (2002) "Kapasitetsberegninger", Working Paper, NSB Bane
Trenitalia, (2013) Relazione sulla qualità dei Servizi;
file/allegati/trenitalia/area_clienti/Relazionesullaqualitadelservizio.pdf (accessed 24
july 2018)
Trenitalia, (2015) Relazione sulla qualità dei Servizi;
2015.pdf (accessed 24 july 2018)
Trenitalia, (2016) Relazione sulla qualità dei Servizi;
(accessed 24 july 2018)
Trenitalia, Libro composizioni servizi universali (2013);
2014.pdf (accessed 24 july 2018)
Trenitalia, Libro composizioni servizi universali (2014);
2015.pdf (accessed 24 july 2018)
Trenitalia, Libro composizioni servizi universali (2015);
2016.pdf (accessed 24 july 2018)
Trenitalia, Libro composizioni servizi universali (2016);
2017.pdf (accessed 24 july 2018)
Veiseth, M., & Bititci, U. S. (2005, July). Performance measurement in railway
operations–improvement of punctuality and reliability. In Proc. of the 5th Int. Conf. on
Performance measurement and management
(pp. 801
Veiseth, M., Olsson, N., & Saetermo, I. A. F. (2007), "Infrastructure’s influence on rail
punctuality", WIT Transactions on The Built Environment, 96, pp.481-491.
Vromans, M. J., Dekker, R., Kroon, L. G. (2006) "Reliability and heterogeneity of
railway services", European Journal of Operational Research, 172(2), pp. 647-665.
European Transport \ Trasporti Europei (Year) Issue 70, Paper n° 2, ISSN 1825-3997
This work would not have been possible without the data collection process made by
the "Comitato Pendolari Genova Milano".
We are grateful to Trenitalia Divisione Passeggeri Long Haul and Trenitalia DPR
Liguria for their assistance with precious additional information.
Finally, we would also like to show our gratitude to the responsible of the “Comitato
Pendolari”, Dr. Enrico Pallavicini, for his commitment during these years for this cause.
... From the microeconomic/performance accounting viewpoint, the definition of expected quality of service in public contracts can take into account "punctuality indicators" [4,5], typically defined in terms of the percentage of trains arrived "on time", i.e., within a given delay, at stations, which can mean either at their final destination or also at intermediate "significant" stops. The delay threshold varies from country to country [6] (e.g., 3 min in the Netherlands [7], 5-10 min in the UK [6], 6-10 min in Belgium [8], 5 min in Germany [6], and 5-15 min in Italy for commuter and long-distance trains, respectively [5,[9][10][11][12]; for a full list, readers are referred to [13,14]). Measures can also be based on the "average delay per train" [4,15], on the standard deviation of the travel time [4], and on behavior of a single train as experienced by its passengers, with a single number with the physical dimension of time, which can then be used for subsequent elaborations. ...
... Indeed, one of the basic measures used to classify a train as "on time" or "late" has long been the delay at the final destination. On top of this basic value, punctuality indicators are usually defined in terms of the percentage of trains arriving within a given delay, which is typically lower for short-haul and commuter services (e.g., 3-5-6 min, depending on the country) and larger for long-haul services (e.g., 10-15 min) [13,14]. This approach is both easy to understand and somewhat natural from a train operator's viewpoint, as the goal is to guarantee the overall stability of the timetable from the operational perspective, which clearly depends on rolling stock being available for the next train at the expected time in the expected location to ensure that the delay of one train is not transferred to the next. ...
... All the relevant technical aspects of train operation are considered, from delay distribution models [28], line planning and timetabling [22], timetable stability analysis and optimisation [23], delay management [24], simulation [23], and performance evaluation in terms of reliability and availability of service [4,6,13]. In [14], a systemic classification of delay causes was proposed based on a comprehensive literature review which includes the main definitions of punctuality and reliability, together with delay thresholds in many European countries. In [35], the focus is on punctuality reporting systems, aimed at performing an in-depth analysis of the delay causes and of the train run. ...
Full-text available
Indicators of expected quality of service in public contracts are often based on some kind of “punctuality”, usually defined in terms of the percentage of trains arriving at the final destination (and/or at intermediate significant stops) within a given delay. Passengers, however, tend to use the word “punctuality” with a more general meaning, mostly as a synonym for expected delay at their own destination, and especially in case of commuters are much less tolerant of even smaller delays than train operators would normally allow. In particular, measuring the delay only at the final destination is perceived as largely inadequate, leading to underestimation of the actual percentage of late trains, and in turn undermining passengers’ trust in official performance statistics. In this paper, we take the passengers’ perspective, introducing a family of delay indices called D-indices aimed at capturing the overall performance of a train “as a whole”, taking into account both the delays at the sampling points and the mutual location and order of such sampling points. In this paper, all indicators have the physical dimension of time in order to be easily replaceable to other delay measures. We first present typical approaches and definitions of punctuality in the literature, then introduce D-indices while exploring their features, pros and cons, and relevant properties. We validate and discuss our approach by comparing this model with existing approaches both theoretically and by comparison with selected datasets consisting of about one hundred trains transcribed over the last three years.
... Technically, the Italian railway company RFI considers on time the passenger trains that arrive at the destination with a delay lower than 15 min for long haul services and less than 5 min for (regional) distance services. Considering these values, it is possible to affirm that in Europe these thresholds are smaller than in Italy (for a more specific classification see Cartmell (2016) and Grechi and Maggi (2018)). ...
... As claimed by numerous authors, such as Alfieri, Groot, Kroon, and Schrijver (2006), Peeters and Kroon (2008), Florian, Bushell, Ferland, Guerin, and Nastansky (1976), and Grechi and Maggi (2018) the number of trains circulating on a given railway network is a key element that can potentially influence the punctuality or delay of a predetermined journey. This relationship is negative as the increasing number of trains passing in a day/h on a railway section can determine deterioration of the single and potentially overall line performance. ...
Full-text available
Covid-19 has strongly influenced the mobility of people in several ways and the limitations of mobility in Europe and Italy have conditioned the flows of workers and travelers. Railways, which are one key transport sector, had a clear decrease in passengers all over the world. The analysis, based on six Lombard/Piedmont lines (North-West of Italy), was carried out in the periods immediately preceding and following the lockdown, in which it was possible to move without draconian restrictions, and with mostly regular railway traffic, but at the same time, the number of commuters and occasional travelers was reduced by 40–60%. Using official data from Trenitalia and Trenord, the role of load factor in the railway performance is analyzed, to verify the impact of a substantial reduction in passenger numbers on the overall performance of six railway lines with a special focus on peak hours. The results obtained are not univocal and provide interesting indications, in fact, for some lines, there is an improvement in performance while for others there are no statistically significant differences with the previous year that means a similar performance, which is a starting point for future developments of the work at the national or regional level. (Proof File)
... В задачу данного исследования не входило проведение сравнительного анализа используемых на различных железнодорожных сетях параметров и методов, однако нельзя не отметить многоплановость ведущегося научного поиска . Так, наряду с комплексным рассмотрением влияния [1][2][3][4], детальному анализу подвергаются и отдельные проблемы, например, влияния многоплатформенных станций [5], содержания инфраструктуры [6][7][8] . Нельзя не отметить подходы к изучению вопросов общего расстояния маршрута на выполнение расписания (например, автор [4, c . ...
... • уровень выполнения расписаний пассажирских поездов по станциям посад- 3 • ки (высадки) пассажиров в пути следования -пасс пв γ ; • уровень выполнения расписаний по прибытии пригородных поездов на промежуточные станции -приг пв γ . Эти, как и другие показатели, в настоящее время нормируются исходя из принципа «от достигнутого уровня», что не позволяет при нормировании заданий на предстоящие периоды учитывать объективные условия эксплуатационной работы на железных дорогах . ...
One of the main tasks of railway employees is to ensure 100 % punctuality of passenger and suburban trains. However, this is impossible due to the action of various reasons, comprising actual reliability of technical equipment and vehicles, natural and other factors. Various companies have different standards and practices of setting and monitoring relevant indicators. The objective of the study was to find out the degree of influence of the e factor formulated as «the number of station stops on the route» on the rate of punctuality of passenger and suburban trains. Calculations and approbation of the suggested model were performed using the example of the JSC Russian Railways. Russian Railways standardise punctuality indicators based on the past performance principle. This does not guarantee that the objective conditions for organising operational work on various railways are fully considered. Besides, it is suggested to consider as main conditions: the level of the use of transit capacity, the technical condition of the infrastructure and rolling stock, etc. However, the factor of the number of station stops of passenger or suburban trains en route envisaged by the traffic schedule is not considered. The greater is the number of station stops, the greater impact, in the absence of a possibility to recover the delay and catch up the traffic schedule, this factor has on the level of traffic punctuality. In turn, the chance to get back on traffic schedule in passenger long distance traffic is higher than in suburban traffic with short routes. The number of station stops varies significantly across the railways, which indicates the unequal conditions of their operation according to this factor. The numerical value of the e factor, as well as of the values of the share of delayed trains (calculated separately for passenger and suburban trains) were determined: by delay in departure – á dep ; by delay in arrival at intermediate points of the route – á int ; by delay in arrival at the destination – á ar . Based on these data, parameters have been established that have allowed to determine the relationship between the number of station stops (e) and the change in the share of delayed trains. Using the methods of mathematical statistics, the insignificant influence of the e parameter on the values of á dep , á int and á ar has been established. It is shown that the punctuality of passenger and suburban trains is significantly influenced by the traffic conditions after departure from the initial station and especially by the possibility to come back on the traffic schedule after possible delays along the route. In this case, one should consider the combined organisation of passenger and freight traffic on most lines of the considered network. It is proposed to optimise the number of standardised indicators in view of their reduction.
... It is very important especially for passengers as a criterion of public transport quality. Passengers should consider interchanges as an integral part of transport process [1,2]. ...
... i.e. to identify the survival duration of dissonants/consonants in TODs. Previously several studies in transportation have applied a continuous time survival analysis method(Anastasopoulos et al., 2012; Bergman et al., 2018;Grechi and Maggi, 2018;Louie et al., 2017;Rahimi et al., 2019;Zheng et al., 2019) ...
Full-text available
Residential dissonants, residents who are not satisfied with land use patterns in their neighbourhood, are a threat to transit-oriented development (TOD) policy because of their unsustainable transport choices. However, it is not known if their level of dissatisfaction is reduced in TODs, and if so, the time duration it takes. This study tracks dissonance status of 98 TOD residents using five waves of panel data spanning over nine years from Brisbane, Australia. The residents were classified into TOD dissonant and TOD consonant (opposite of dissonants) groups and a discrete time survival analysis technique was applied to identify time-to-event for these groups. An event was recorded if a dissonant became a consonant, or vice versa. Two discrete time hazard models were estimated using binary logistic regression analysis (one for each transition) to identify socio-demographic and built environment characteristics associated with the occurrence of an event. Results showed that about 46% of the TOD residents were dissonants at baseline. The survival functions were significantly different between dissonant and consonant classes. About half of the dissonants took-on the characteristics of consonants in just four years. In contrast, TOD consonants remained consonants relatively longer (median survival duration is 9 years). Groups that were likely to become dissonants were those with low educational status, and people born overseas. The findings suggest that TODs have an autonomous effect on changing attitudes over time, which verifies the 'reverse causality' hypothesis, and therefore, TODs are likely to be dissonant free naturally presumably as residents experience the benefits of TOD living. The process could be sped up with targeted policy interventions (e. g., concessionary travel card, rent relief to bear the high cost of living in TODs) for those being as, or likely to be susceptible to become dissonant.
... The function minimizes the sum of the squares of the distances between the observed data and those of the curve that represents the function itself (Paruolo, 1999). According to DeFries and Fulker (1985) and Gunst (2018) this methodology is applied in numerous fields, such as business studies (Gazzola et al., 2019b) transportation studies (Grechi & Maggi, 2018) and cognitive fields (Folgieri et al., 2014). We have elaborated the regression models using Gretl ( ...
Full-text available
By adopting mainly the "principal-agent theory", the study analyses how non-profit organizations (NPOs) corporate governance structure could increase the trust of the donors and therefore affect their ability to receive donations from taxpayers. Starting from a literature review we concentrated our attention on the non-profit sector where NGOs represent the largest category. In Italy, starting from 2006, all NPOs could receive funding deriving from taxes paid by citizens when making tax returns with the so called '5 per thousand' of the personal income tax. We analyzed the corporate governance disclosure practices of the first Italian 100 NPOs that received the highest donations from 5 per thousand. In particular, we elaborated a CGI index that includes governance and informativeness. This paper shows how an efficient and transparent corporate governance structure motivates the donors to donate 5 per thousand to NPOs that demonstrate good corporate governance. The findings suggest that taxpayers are inclined to allocate 5 per thousand to organizations where the information level, from a governance point of view, is high, easily available and clear about the purpose in the specified field of research.
... A railway vehicle derails when its wheels run off the rails and, in effect, the rails no longer provide the guidance of the vehicle in the lateral direction. Since the risk of derailment is of the ultimate safety concern in the operation of railway vehicles, it has been a subject of many studies in the past (for example, [12]) as well as the topic of ongoing researches [6,10,15,16,19,20,24]. While various scenarios of derailment are possible and the respective criteria have been proposed and are in use [18], one of the earliest relevant works was published by Joseph Nadal in 1896 [17]. ...
Full-text available
This paper investigates the variation of track geometrical parameters that lead to a local increase of specific dynamical quantities of a railway vehicle, possibly beyond their acceptable values. In particular, the changes of track geometry are investigated near track points where the running safety or ride comfort are significantly decreased during the vehicle motion due to track irregularities. The investigated dynamical quantities include the lateral and vertical forces at the wheel-rail contact as well as the acceleration of the vehicle body. The vehicle motion has been simulated using a non-linear model of a passenger car moving along a nominally tangent stiff track with random geometrical irregularities. The relationship between the local track condition and the maxima of the dynamical quantities was investigated with the statistical method proposed by the author. The performed analysis clearly identifies the characteristic variation of track irregularities that leads to a large increase of the investigated dynamical quantities at some track points.
Full-text available
The Yinchuan–Chongqing high-speed railway (HSR) is one of the “ten vertical and ten horizontal” comprehensive transportation channels in the National 13th Five-Year Plan for Mid- and Long-Term Railway Network. However, the choice of node stations on this line is controversial. In this paper, the problem of high-speed railway station selection is transformed into a classification problem under the edge graph structure in complex networks, and a Scatter-GNN model is proposed to predict stations. The article first uses the Node2vec algorithm to perform a biased random walk on the railway network to generate the vector representation of each station. Secondly, an adaptive method is proposed, which derives the critical value of edge stations through the pinching rule, and then effectively identifies the edge stations in the high-speed railway network. Next, the calculation method of Hadamard product is used to represent the potential neighbors of edge sites, and then the attention mechanism is used to predict the link between all potential neighbors and their corresponding edge sites. After the link prediction, the final high-speed railway network is obtained, and it is input into the GNN classifier together with the line label to complete the station prediction. Experiments show that: Baoji and Hanzhong are more likely to become node stations in this north–south railway trunk line. The Scatter-GNN classifier optimizes the site selection strategy by calculating the connection probabilities between two or more candidate routes and comparing their results. This may reduce manual selection costs and ease geographic evaluation burdens. The model proposed in this paper can be used as an auxiliary strategy for the traditional route planning scheme, which may become a new way of thinking to study such problems in the future.
Full-text available
A transport business that has reached financial sustainability is one that is providing a service at a price that not only covers its costs but also creates a profit for upcoming contingencies. A focus on rail infrastructure financial sustainability is of paramount importance to guarantee the availability of punctual rail transport to remote potential users. To evaluate the sustainability of mass rapid transit on the relation among hypothetical key aspects of sustainability—perception of property; willingness to pay for maintenance, repair, and operations; confidence in the Roads and Transport Authority; and citizen participation in the rail project—and railway service punctuality, the most important result variable, was studied according to the specialized literature on rail transport sustainability. Leading information was collected by means of personal questionnaires of more than 1000 railway users according to the Krejcie Morgan formula for the calculation of the sample size knowing the population size. Qualitative plus quantitative information was gathered from different ways (technical test of the rail system, discussions with users, focus-group discussions, and interviews with key informers).The outputs by means of the statistical analysis allowed understanding two key perceptions. Firstly, beyond a half decade after construction, during a system intervention, a smaller perception of public property for the railway system was related to better service punctuality. This idea contrasts with the vast majority of the publications, which highlight a regular, direct relationship between perception of property and sustainability of railway systems. Secondly, in spite of three-quarters of users accepting that they would contribute monetarily for maintenance, repair, and operations service, such payments were not imminent because of the lack of confidence in the Roads and Transport Authority. In this situation, more than one-third of the metro stations were identified as non-punctual, beyond a half decade after construction.
This study analyzes the perceptions of individuals on retrospective rail punctuality indicators to determine the most useful indicator according to socio-demographic characteristics, regular trip behavior variables, and railways transportation habits variables. In choice situations, individuals must choose between four punctuality indicators and an out option. Common punctuality indicators have been selected among those proposed by the authority for quality of service in transport, as well as a new punctuality indicator from the financial literature: Delay-at-Risk. Thus, via an online survey and econometric modeling, we show that respondents appreciate the usefulness of punctuality indicators for planning their long-distance rail trips. The usefulness is reinforced by the fact that respondents employ several modes for regular trips and frequent train users. Moreover, they have already experienced missed appointments or connections. The risk attitude and prudence of respondents also play an important role but not totally in the expected direction. Lastly, Delay-at-Risk, although unknown and more complex in its formulation, exhibits some characteristics that are appreciated by users.
Conference Paper
Full-text available
Train service operation and modelling requires a good understanding of the delays which may disrupt performance at stations. However, few of these smaller delays are measured properly, let alone understood. This paper reports ongoing research on European railways into the magnitude and distribution of minor delays at stations, the reasons for their propagation, and possible management strategies to mitigate them. Results will be important not only for train service modelling, but also for their direct application to immediate service improvement.
Full-text available
One of the most important quality indicators of public transportation is punctuality. Deviations from schedule reduce the level of service. Analyzing historical data, exploring and categorizing the causes of delays correlations can be determined. Based on them, the schedule deviations are predictable. In our research the schedule deviations on railway stations have been investigated based on the manually registered information of the Hungarian Railways. The study mainly focused on the delay causes that were generated by random external factors. Particularly, effects of the certain weather conditions have been highlighted. The analysis has been conducted on railway lines with different infrastructure. Contexts based on the results of the research can be built into traffic prediction models.
Italy is among the few countries where open-access rail competition exists, and the entry of the newcomer NTV during 2012 radically changed the market conditions. The aim of this paper is to shed some light on the effect of competition on price strategies. To do that, we focused on the period before and after the opening of a new HS direct service between Milano and Ancona. Through a web crawler, we collected available fares, from 30 to 1 days in advance of departure and for 15 months, between September 2013 and December 2014. We statistically analysed the sample i) to estimate the price reactions of the incumbent after the new entry, and ii) to compare the price strategies of the two operators. Results show that the incumbent reduced economy class prices by about 15%. The newcomer is slightly cheaper for bookings well in advance of the departure date, but raised prices in the very last days before departure. Their fare strategy also changed after three months of operations, when they applied lower prices. Finally, we see that Trenitalia does not respond to NTV’s prices, whereas the opposite is true, demonstrating that the newcomer is the price-taker in the short-run.
Economic theory advocates marginal cost pricing for efficient utilisation of transport infrastructure. A growing body of literature has emerged on the issue of rail marginal infrastructure wear and tear costs, but the majority of the work is focused on costs for infrastructure maintenance. Railway track renewals are a substantial part of an infrastructure manager’s budget, but in disaggregated statistical analyses they cause problems for traditional regression models since there is a piling up of values of the dependent variable at zero. Previous econometric work has sought to circumvent the problem by aggregation in some way. In this paper we instead apply corner solution models to disaggregate (track-section) data, including the zero observations. We derive track renewal cost elasticities with respect to traffic volumes and in turn marginal renewal costs using Swedish railway renewal data over the period 1999–2009. This paper is the first attempt in the literature to apply corner solution models, and in particular the two-part model, to disaggregate renewal cost data in railways. It is also the first paper that we are aware of to report usage elasticities specifically for renewal costs and therefore adds important new evidence to the previous literature where there is a paucity of studies on renewals and considerable uncertainty over the effects of rail traffic on renewal costs. In the Swedish context, we find that the inclusion of marginal track renewal costs in the track access pricing regime, which currently only reflects marginal maintenance costs, would add substantially to the existing track access charge. EU legislation requires that access charges reflect the ‘costs directly incurred as a result of operating the train service’, which should include a marginal renewal cost component. This change would also increase the cost recovery ratio of the Swedish infrastructure manager, thus meeting a policy objective of the national government.
In this paper, renewal costs for railway tracks are investigated using survival analysis. The purpose is to derive the effect from increased traffic volumes on rail renewal cycle lengths and to calculate associated marginal costs. A flow sample of censored data containing almost 1300 observations on the Swedish main railway network is used. We specify Weibull regression models, and estimate deterioration elasticities for total tonnage as well as for passenger and freight tonnages separately. Marginal costs are calculated as a change in present values of renewal costs from premature renewal following increased traffic volumes. The marginal cost for total tonnage is estimated to approximately SEK 0.002 per gross ton kilometre.
This paper treats knowledge stocks as endogenous public goods. Rather than being used up in the process of production, knowledge is expanded and enhanced by way of exchange processes on a network consisting of nodes and links in geographical space. The nodes take the form of human settlements such as villages, towns or metropolitan regions, and the links between nodes consist of transportation routes and communication channels which facilitate knowledge acquisition and knowledge expansion.