mystic - final report · project st-97-sc.2101 author: pdc, peter davidson consultancy (gb) ... ses...

84
Final Report for Publication MYSTIC Towards Origin – Destination Matrices for Europe Project ST-97-SC.2101 Author: PDC, Peter Davidson Consultancy (GB) Project Coordinator: PDC, Peter Davidson Consultancy (GB) Partners: Sub Consultants PDC, Peter Davidson Consultancy (GB) Co-ordinators AGDER, Research Foundation (NO) BEL, Baxter Eadie Limited (GB) CBS, Statistics Netherlands (NL) CNS, Community Network Services Limited (GB) INRETS, Institute National de Recherche sur les Transports et leur Securite (FR) IVT, IVT Heilbronn (DE) IVV, Ingenieurgruppe IVV-Aachen (DE) NEA, Transport Research and Training Foundation (NL) NTUA, National Technical University of Athens (GR) SES, Ministry of Transport and Tourism (FR) DETR, Department of the Environment, Transport and the Regions (GB) TTR, Transport and Travel Research (GB) Project Duration: January 1998 – September 1999 Date: August 2000 Project Funded by the European Commission Under the Transport RTD Programme of the 4 th Framework Programme

Upload: doancong

Post on 13-Feb-2019

213 views

Category:

Documents


0 download

TRANSCRIPT

Final Report for Publication

MYSTIC Towards Origin – Destination Matrices for Europe

Project ST-97-SC.2101

Author: PDC, Peter Davidson Consultancy (GB)

Project Coordinator: PDC, Peter Davidson Consultancy (GB)

Partners:

Sub Consultants

PDC, Peter Davidson Consultancy (GB) Co-ordinators

AGDER, Research Foundation (NO)

BEL, Baxter Eadie Limited (GB)

CBS, Statistics Netherlands (NL)

CNS, Community Network Services Limited (GB)

INRETS, Institute National de Recherche sur les Transports et leur Securite (FR)

IVT, IVT Heilbronn (DE)

IVV, Ingenieurgruppe IVV-Aachen (DE)

NEA, Transport Research and Training Foundation (NL)

NTUA, National Technical University of Athens (GR)

SES, Ministry of Transport and Tourism (FR)

DETR, Department of the Environment, Transport and the Regions (GB)

TTR, Transport and Travel Research (GB)

Project Duration: January 1998 – September 1999

Date: August 2000

Project Funded by the European Commission Under the Transport RTD Programme of the 4th Framework Programme

- 2 -

Contents 1. THE MYSTIC PARTNERS ....................................................................................................................4

2. EXECUTIVE SUMMARY.......................................................................................................................7

2.1 INTRODUCTION ........................................................................................................................................7

2.2 THE DEVELOPMENT OF THE PASSENGER O - D MATRICES ..........................................................7

2.3 THE INVESTIGATION OF FREIGHT SURVEY STRATEGY....................................................................9

2.4 RECOMMENDATIONS ON THE WAY FORWARD................................................................................11

3. OBJECTIVES OF THE PROJECT..........................................................................................................13

3.1 PASSENGER OBJECTIVES......................................................................................................................13

3.2 FREIGHT EDI S URVEY OBJECTIVES ...................................................................................................13

3.3 FREIGHT SHIPPERS S URVEY OBJECTIVES ........................................................................................14

4. ACHIEVEMENT OF THE OBJECTIVES ......................................................................................16

4.1 INTRODUCTION ......................................................................................................................................16

4.2 BACKGROUND ........................................................................................................................................16

4.3 AIM OF THE MYSTIC P ROJECT .........................................................................................................17

4.4 THE FRAMEWORK .................................................................................................................................18

4.5 KEY ISSUES ............................................................................................................................................20

4.5.1 Data Owners Confidentiality ..................................................................................................20

4.5.2 The Treatment of Air Passengers........................................................................................21

4.6 PASSENGER AND FREIGHT O -D MATRICES......................................................................................22

4.6.1 The Use of EDI to Gather Freight Data .............................................................................23

4.6.2 Producing an OD Transport Chain Freight Matrix.........................................................23

5. TECHNICAL DESCRIPTION OF THE PROJECT.....................................................................25

5.1 INTRODUCTION ......................................................................................................................................25

- 3 -

5.2 INCEPTION REPORT AND MODELLING FRAMEWORK (WP2) ........................................................25

5.3 PASSENGER ACTIVITIES (WP3 AND WP4) .....................................................................................26

5.3.1 Results Achieved .......................................................................................................................35

5.3.2 Presentation Software .............................................................................................................43

5.4 ELECTRONIC DATA PROCESSING (WP5) ........................................................................................43

5.5 SHIPPERS SURVEY (WP6)..................................................................................................................50

6. CONCLUSIONS AND RECO MMENDATIONS ...........................................................................64

6.1 OVERALL CONCLUSIONS : FREIGHT.................................................................................................64

6.2 OVERALL CONCLUSIONS: PASSENGERS..........................................................................................67

6.3 PROPOSED METHODOLOGY FOR BUILDING THE PAN-EUROPEAN PASSENGER MATRIX .......68

6.4 THE WAY FORWARD .............................................................................................................................70

6.5 WHAT WILL THIS PROVIDE?...............................................................................................................71

7 REFERENCES ................................................................................................................................................73

ANNEX 1: PUBLICATIONS, CONFER ENCES AND PRESENTATIONS .................................75

ANNEX 2: DESIRE LINES ..........................................................................................................................77

- 4 -

1. The MYSTIC Partners

Peter Davidson Consultancy – Partner & Project Co-ordinator

PDC is a private sector consultancy with particular expertise in the areas of transportation

planning, software development and market research, including the use of stated preference

analysis techniques. The consultancy has undertaken work at all spatial levels: urban;

interurban; regional; national and trans-national. Its clients include infrastructure authorities,

government agencies and major financial institutions.

AGDER - Partner

As a regional independent and non-profit organisation, AGDER is one of the largest research

institutions in Norway and also comprises a Transport Studies Group. In the transport arena,

AGDER’s main activities comprise of analysis of transport demand and supply, forecasting

scenarios of transport activities and transport policy assessment and analysis.

AGDER has 10 years experience in local, national and Europe-wide projects including freight

demand for the Nordic link corridor, establishment of OD matrices in Southwest Norway as

well as IN FOSTAT and MESUDEMO for the EU.

Baxter Eadie – Partner

Baxter Eadie is a transport consultancy with specialist expertise in Ports & Shipping,

Insurance and Finance, Economics and Statistics, Intermodal and Landside Transport.

Particular interests of the consultancy include shipping, containerisation, marine insurance,

development banking and road/rail transport. Additional specialists from related disciplines

are also associated with the company, providing a wide-ranging area of expertise. Recent

European work has included STEMM and INSPIRE.

CBS – Partner

As the Central Bureau of Statistic of the Netherlands, CBS is tasked to provide official

statistical information and economic statistics. CBS provides all the official statistics of the

Netherlands, and its work programme is comprehensive. Statistics available include all

economic sectors, national accounts, consumer and producer price indices and

environmental statistics.

Community Network Services Ltd – Partner

CNS provides customised, managed solutions to serve the information needs of communities

with a common requirement for the electronic exchange of information. CNS creates and

develop advanced port community systems for simplified customs clearance throughout the

- 5 -

UK. CNS is experienced in web portals, such as the Southampton Port Information Network

(SPIN) and Port Automated Cargo Environment (PACE). Other services include value added

managed network service and website design/hosting. CNS has also been partners in the

3Snet, POSEIDON and FINE European projects.

Institute National de Recherche sur les Transports et leur Securite – Partner

INRETS is a French public research institute specialising in the field of transportation, under

the patronage of the Ministries of Transport, Industry and Research. It’s activities cover

technological and socio-economical research, and INRETS has recently developed a strong

involvement in European fourth and fifth framework projects.

IVT Helibronn – Partner

IVT is a small private research institute based in Germany. The main areas of work for IVT

are transport statistics, statistical methodology for transport studies, design and analysis of

transport and travel surveys, empirical mobility and transport studies, travel forecasting,

quality management in both public and freight transport and data analysis.

Ingenieurgruppe IVV Aachen – Partner

A private consultancy based in Germany, IVV is a highly experienced European project

agency. IVV specialises in transport research and analysis, urban and regional planning,

network studies and software development.

IVV has been responsible for projects on all spatial levels from local to European wide

including CAMPARI, INFOSTAT, Federal highway master plans, modal split surveys and park

& ride Studies.

National Technical University of Athens – Partner

The main activity of the National Technical University of Athens (NTUA) is obviously one of

education, but the Telecommunications Laboratory, a division of NTUA has a wide range of

experience in computer networks, mobile telecommunications, network management and

conformance testing.

The laboratory has a history of working in European transport projects including

EUROSCOPE, THERMIS and INFOSTAT where it provided support relating to advanced

information management systems, decision support systems and applications of second

generation GIS.

NEA, Transport Research and Training Foundation – Partner

- 6 -

NEA is an independent organisation specialising in research, consultancy and training

services in the field of transport and logistics. NEA’s employees include economists,

econometrists, engineers, EDP personnel and educationalists.

NEA’s activities cover a broad range of transport related topics. These include modelling,

forecasting and database construction, market evaluation, policy support, cost/benefit and

feasibility studies. NEA’s econometric analyses are supported by databases of transport

flows and infrastructure.

Previously, NEA has worked on a number of European Commission projects including

INFOSTAT, the SPREAD working group and has constructed European regional trade

models.

Service Economique et Statistique – Partner

SES is part of the Directorate for Economique and International Affairs (DAEI) within the

French Ministry of Equipment, Transport and Housing. It is respons ible for public statistics

and economic studies for the Ministry. SES operates as a branch of INSEE, the National

Statistical Institute, and is also the French correspondent for EUROSTAT.

DETR – Sub Consultant

The Department of Environment, Transport and the Regions is a national government

department in the United Kingdom. Its role in the MYSTIC project was as a sub-consultant

and provided expert technical assistance in the matrix development process. The DETR

were co-funders of the United Kingdom’s contribution to the MYSTIC project.

Transport and Travel Research – Sub Consultant

Based in the United Kingdom, TTR is a transport consultancy offering specialist research and

consulting services to both public and private sector organisations throughout Europe.

TTR’s main activities include transport policy, energy and the environment studies, transport

telematics, public transport, market and social research and project management.

TTR offer advice and services from initial concept through to implementation and appraisal.

This covers qualitative surveys, proposal preparation, system specification, evaluation of

systems and schemes, project and programme planning, policy analysis and public

consultation.

- 7 -

2. Executive Summary

2.1 Introduction

The MYSTIC project set out to address the issue of how to build origin-destination passenger

and freight matrices for Europe from currently available data and to chart a path which could

lead to a continuous process of matrix development. Such matrices and the information they

contain are essential for the development of the Trans European Transport Network (TEN),

as well as the linkages that feed into it. They are essential for the development and

justification of the Common Transport Policy (CTP) and vital for effective decision-making.

They underpin the actions of Government and are needed to maintain the infrastructure and

to regulate the transport system equitably. Transport operators and infrastructure providers

are desperately in need of a European origin-destination matrix – with a detailed knowledge

of where people and freight come from and go to, they can optimise their service provision so

as to make their services more attractive and capture new markets (e.g. from road) with the

confidence in knowing that the business investment decisions they are making, will give them

the return on investment that they have planned for.

The project has been carried out through a total of seven work packages which have broadly

covered co-ordination, preparation of an Inception Report, the development of new

methodologies for data analyses, the development of the passenger matrices, the

development of procedures for the collection of freight data through electronic methods, the

development of freight shippers surveys and preparation of progress, work package and final

reports.

This report details what has been achieved in relation to the objectives that were set.

2.2 The Development of the Passenger O - D Matrices

The objective of the passenger component of MYSTIC was to develop a methodology for

building a pan-European origin – destination matrix and to test it using two case studies, as

follows:

1. The European case study merged the existing origin-destination data sets for seven

countries together to produce a European origin-destination matrix covering road and

rail. The objective, to develop the tools and techniques for merging different datasets

and to show how they could be used to produce a European trip matrix, was

successfully achieved and produced a useful matrix.

- 8 -

2. The UK case study assembled the major detailed datasets in the UK and showed how

different detailed datasets could be held in a detailed National database suitable for

building detailed trip matrices for infrastructure design, operation and assessment. This

database is currently in use by the UK Government for its programme of major multi -

modal studies covering eleven key locations throughout the UK.

The case studies demonstrated that the overall approach outlined here can be used to build

the trip matrix and that existing data can make a valuable contribution to producing the pan

European passenger origin-destination trip matrix which has the following components:

A major new pan-European passenger screen line origin-destination survey: (see

below)

Existing origin-destination datasets: e.g. household o-d interviews, on-system o-d

interviews (i.e. at roadside, on-train and at-airport), already-constructed o-d matrices: The

study showed that these existing data sources could make a valuable contribution to the

matrix.

Existing non origin-destination data: e.g. count data, population and demographic

data: This data would be used in conjunction with the matrix improving methodology to

improve the resulting matrix.

Existing ticket sales data: The methodology developed for the UK case study can be

used to derive air and rail origin-destination passenger trip matrices from ticket sales data.

Connect to Transport Operators Databases: Transport operators collect origin-

destination data and it is important to elicit the support of all data owners to contribute data

and make use of the matrix. This would involve tackling the important issue of data

confidentiality for which MYSTIC shows the way forward.

Research into Methodology: The methodological advances made by MYSTIC can be

used and built upon to improve the matrix. The methodology itself can be improved with

further research so as to incorporate a wider range of data types with modelling.

The study showed the importance of establishing an organisational structure, which secures

sensitive data and provides as free a flow of information as possible within the constraints that

data owners place upon the use of their data by others. The database approach provides a

framework within which this can be organised optimally, providing the continuity of keeping

the pan-European passenger origin-destination trip matrix as a working tool. In the database

approach, the data is held in a database, which can be added to as data becomes available

and matrices can be built from it – matrices which can be used at both a detailed scheme

appraisal level and at a pan-European level for policy analysis.

- 9 -

The case studies also showed the use and limitations of existing data and highlighted the

need to collect a consistent set of new data with which to both harmonise existing data and to

add valuable additional information. This led to the need for and specification of, a new pan-

European passenger screen line origin-destination survey so as to complete the survey

picture for matrix building. This new survey would mean identifying a set of screen lines

throughout Europe which intercepted the major international flows and surveying them with

roadside, on-train and at airport personalised origin-destination interviews on a rolling

programme and using the data for enriching and infilling the matrix with better and better data

over time. For road this could mean undertaking a series of roadside interview sites along the

major screen lines. For rail this would mean interviewing on-train on routes that cross the

screen line. For air this would mean interviewing at airports for flights that cross the screen

line. The survey design would need to give careful attention to sampling ensuring that the

sample can be expanded to represent all travel that crossed the screen line. The data needs

to be collected to a structured and consistent way particularly with regard to the origin and

destination addresses, which need to be collected with sufficient precision at a European level

and so that they can be used at a detailed level for scheme appraisal.

The new pan-European passenger screen line origin-destination survey would provide a

robust and consistent data source to help provide information missing in the current data sets.

This would be best undertaken in conjunction with the initiatives in member countries as we

are aware that at least one country is planning (and maybe already undertaking) a roadside

interview survey for national matrix building. It is envisaged that the first phase would require

about 250 sites to be surveyed, costing about 3m euro for fieldwork address coding and data

processing.

2.3 The Investigation of Freight Survey Strategy

Until recently customs records have provided a data source suitable for compiling origin-

destination flows of freight commodities between member countries which with the help of

additional surveys, could provide enough information from which to compile the statistics

needed by member countries including detailed matrices of freight flows (e.g. tonnes lifted,

tonne kilometres and freight commodity value). This valuable data source disappeared with

the disappearance of customs duty between EU member countries. The MYSTIC project

therefore investigated freight data collection strategy with two possible candidate

methodologies for collecting freight origin-destination data. They were as follows:

1. The freight shippers survey sampled shipments of freight and traced the shipment

through each organisation that handled the goods with telephone interviews until it was

traced to its final destination. This method involved a series of telephone calls to each

company involved in the shipment chain asking detailed information including

behavioural information about the alternatives that the company could have used.

- 10 -

2. The freight EDI survey sought to investigate the possibility of intercepting the Electronic

Data Interchange (EDI) messages, which pass over the Internet from one company to

the next in the freight commodity chain, in advance of the goods reaching them.

The EDI survey showed that the information required (i.e. origin, destination and other data

items) did indeed exist in the computer systems of the originator of the consignment although

the data items required usually spanned more than one organisation and spanned more than

one computer system which made automating the process difficult to achieve in the short

term.

However the opportunity should be taken to introduce a EU wide standardised data collection

format now - because the timing is optimum. The industry (particularly that associated with

the logistics operator and with just-in-time manufacture) is undergoing rapid change and in

response software suppliers are redesigning their software. The opportunity now arises, if an

EU-wide standard is introduced, that EDI software suppliers would design-in the new

standard so that future releases would have the new standard embedded. If the standard was

so designed this could provide the platform needed for the full-scale automation of freight

origin-destination data collection with EDI.

The shippers survey successfully found a methodology for tracing freight shipments from

origin to destination through each link of the transport chain so that a matrix could be built for

each commodity flow. This is a significant finding because such a methodology had hitherto

eluded researchers.

The ‘shippers’ survey methodology can also be used to elicit a wealth of other detail including:

the origin-destination characteristics of the chain, sequence of modes, intermodal platform,

physical dimension of the commodity flow, behavioural data, costs and tariffs, as well as an

in-depth investigation of the logistic organisation of the shipper, thereby considerably

improving our ability to understand and forecast the movements of goods and their modal

split. Eliciting this full set of information can be costly and if the objective is the origin-

destination matrix then slimming the interview down to the essential data items can further

reduce costs. For behavioural and other data an additional questionnaire can be administered

to a sub sample.

The survey methodology could be scaled-up so as to elicit enough information from which to

build a freight origin-destination transport chain matrix for Europe. Additional benefit would be

gained by undertaking the survey at the European scale for several reasons. For example as

a transport chain may have links in several member countries, shippers could be interviewed

by their own Nationals – increasing the success rate and reducing the cost per interview.

It is therefore recommended that the Commission undertake a pan-European freight

‘shippers’ origin-destination survey so as to build the European freight transport chain origin-

- 11 -

destination matrix. The cost of the complete survey is likely to be high and would need to be

set up as a rolling programme with Eurostat taking the lead - although due to the initial

research nature of the survey it would be best to instigate the survey as part of the European

research programme. The fieldwork should be phased with an initial one-third coverage that

could cost in the region of 3.5m euro including fieldwork, survey design and analysis costs.

The first phase should be combined with the recommended passenger research project (see

below). The rolling programme could cover the other two thirds over the rest of a ten-year

period under the auspices of Eurostat. The survey should be sponsored by member

governments who should provide visible backing for the survey – shippers are reluctant to

respond to non-governmental initiatives.

The data would be put into the origin-destination database and the organisational framework

put in place to provide information for government and maintain, market and offer users the

service of supplying extracts from the data, matrices and tools. The organisational framework

needs to be set up with great care so as to safeguard the data and maintain the high quality

standard of the data required by member governments.

2.4 Recommendations on the Way Forward

We recommend the Commission organise a research project to design, set up and initiate the

first four-year phase of a rolling programme of passenger screen line origin-destination

surveys and freight ‘shippers’ origin-destination surveys, put this data into a European origin-

destination database and build the passenger and freight matrix. The surveys could

subsequently be taken over by Eurostat and undertaken as a ten year rolling programme so

as to keep the matrices reasonably up to date. The two new surveys are pan-European and

are as follows:

A new passenger screen line origin-destination survey: so as to provide the core of the

passenger matrix. It would cover major international and European passenger flows by

interviewing passengers at the roadside, on train and at airports on transport links that cross

optimally located screen lines across Europe.

A new freight ‘shippers’ origin-destination survey so as to provide the core data for the

freight matrix. The survey methodology would draw upon the results of MYSTIC suitably

adapted for the different member countries with a questionnaire that just elicited the

information needed for matrix building. The survey would cover all countries with about a one

third coverage

The research project could also set up the combined passengers and freight European origin-

destination database together with the service to supply information to government (and

others), set up the database maintenance, marketing and support as well as providing a

general service of supplying matrices and tools to users. The origin-destination database

- 12 -

could comprise the new survey origin-destination data together with pan-European household

travel diary origin-destination survey data, rail and air ticket sales data, counts and other non

origin-destination data. The project could use the origin-destination database to build the pan-

European passenger and freight origin-destination matrix as described above. Operating the

database could become a near self-financing service, run on a five year rolling contract from

the Commission with the organisation undertaking analysis for the Commission and acting as

a focus for data supply to others.

The research project could encourage and help member governments build their national

origin-destination database. It could develop computer linkages with other databases (e.g.

transport operators) providing exchange of data with them and so that the data can be used

to improve the matrix. It could provide a focus of expertise for origin-destination data and its

uses and support the activities of researchers, analysts and modellers. It could research

methodologies for matrix merging, combining and improving and develop the tools that could

be made available to member governments and data contributing organisations. It could also

cover collation of other data and the passenger survey design and analysis.

The new passenger screen line origin-destination survey is likely to cost 3m euro including

fieldwork, address coding and data processing (but excluding the survey design and analysis

costs which would be part of the main research project). The new freight ‘shippers’ origin-

destination survey could cost 3.5m euro including fieldwork, survey design and analysis costs.

The research project itself is likely to cost in excess of 1.5m euro, which with the surveys

could give a total project cost of 8m euro. The research project should deliver a pan-

European passenger and freight origin-destination matrix using the above methodology. It

should instigate a rolling programme of passenger and freight origin-destination surveys

covering a ten-year period, which should be repeated so that the matrix is never more than

ten years out of date.

- 13 -

3. Objectives of the Project

3.1 Passenger Objectives

The purpose of the passenger part of the MYSTIC project was to develop a methodology for

building a pan-European passenger o-d matrix. The detailed objectives to be achieved were

as follows:

To bring together heterogeneous data sets from a number of European countries and

merge them into a pan-European o-d matrix covering road and rail

To develop a detailed UK based o-d matrix to show how datasets could be held in a

database for use in infrastructure scheme design and development

To develop mathematical and statistical tools and techniques that would be used by the

case studies,

To develop software tools that could be used for both analytical and presentational

purposes.

The methodological developments, the two case studies, mathematics and software

development when taken together succeeded in developing a methodology for building the

pan-European passenger o-d matrices which could be used both at a strategic pan- European

level for policy and strategy hypothesis testing and at a very detailed level for scheme

assessment and design – the same database technology and tools being used at a range of

levels of spatial detail and extent of the study areas.

The project demonstrated the efficacy of the recommended methodology with two case

studies: one aimed at the pan-European policy level and another at the detailed transport

planning and infrastructure appraisal level (the UK case study). The pan-European Merged o-

d matrix this study produced may be useful for research purposes but has weaknesses that

make it unsuitable for use in policy analysis. If it is used then the user should take full account

of the construction of the matrix, because, for example, certain types of trip may be omitted

(e.g. short distance trips or trips from non-Nationals of the country concerned) in addition to

the weaknesses inherent in the datasets used to com prise the matrix cells concerned

3.2 Freight EDI Survey Objectives

Background Information about origins and destinations of transport flows is an essential

requirement for assessing needs for transport infrastructure provision, for analysis of internal

and external costs and benefits of such transport activity, and for development of policy

- 14 -

initiatives to influence and provide facilities for these activities. Statistics describing patterns

of origin – destination distributions are of value to operators and economic actors within the

transport industry, for purposes of marketing, development of facilities, and planning of

operations.

Sample surveys to collect such statistics have been carried out on a comprehensive national

basis since 1974 in the U.S.A, since 1978 in the United Kingdom, and during the 1980s and

1990s in France, Germany and the Netherlands, for international trade flows. Because of the

expense involved in the large surveys required, these studies have been generally been

carried out at intervals of five years or longer. Efficient sample design is essential to

minimise costs, and in turn requires a comprehensive sample frame of consignments. In the

United Kingdom Customs provided a complete frame of consignments until 1993. Since the

introduction of the Single Market this frame has not been available for intra-European trade.

An alternative methodology with ship manifests as a sample frame was used to survey United

Kingdom – Continent transport flows during 1996, within the STEMM project, but was only

partially successful because of incompleteness of the sample frame.

The questions that have been addressed by this study are “Can complete census information

be collected easily from transport operators with electronic control systems?” and “How does

information from these operators relate to overall transport patterns?”

The Work Plan objectives of Work Package 5, the EDI study, were therefore to:

Assess the extent of computer based control systems among transport operators;

Find out the content of these systems;

Develop and evaluate statistical procedures to compile statistical descriptions of all

transport chains in the overall origin-destination flows.

3.3 Freight Shippers Survey Objectives

The specific problem, which Work Package 6 of MYSTIC addressed, was to establish policy-

oriented information on the flows of goods in Europe. To do that the project developed an

approach based on a survey of shippers’ international related activities. The survey of

shippers was seen to be the basis for understanding and measuring the relationship between

organisational and economic factors of the firm on the one hand, and the modal choice

behaviour on the other hand. This information is essential for transport policy.

The objective of this aspect of the MYSTIC project was to show the technical possibilities for

applying the transport chain concept developed by the INFOSTAT project. In doing so it

- 15 -

contributed to a better understanding of the organisation of the transport chain for

international trade and particularly a better understanding of the modal organisation.

In addition, Work Package 6 had an objective to develop methodologies for building up

matrices of freight flows along the transport chains.

The research tasks attributed to the Shippers surveys part of MYSTIC had two objectives:

To develop a behavioural analysis of the modal choice for international transport chains

To produce a methodology for estimating Origin/Destination international transport chain

freight matrices.

The research task of the behavioural survey can be stated as follows: Develop a behavioural

analysis of the organisation of a transport chain and subsequently develop new statistical

methods aiming at building up transport chain matrices of good flows.

Parallel to the behavioural shipper survey, the O/D shipper survey was carried out in order to

estimate the level of trade by activity and on a region-to-region basis.

By undertaking in-depth surveys on selected corridors and segments, the MYSTIC freight

study has aimed at the possibility of developing an innovative approach of complex problems

that could be applied to the EU as a whole.

- 16 -

4. Achievement of the Objectives

4.1 Introduction

The MYSTIC project was launched at a time when

transport modelling was gaining rapidly in sophistication,

integrated and comprehensive transport models were used for national planning in many

European countries based on detailed national zoning systems, but normally with few or no

foreign zones added,

several international transport corridor studies had been carried out based on statistics

compiled, processed and adapted specially for those projects,

the European Union (EU) had in the Framework Programs financed the development of

integrated and comprehensive transport models experiencing problems of data shortage,

available algorithms, iteration procedures and computer capacity were rapidly getting

more powerful,

European transport policy was putting high priority on the Pan-European dimension,

exemplified by Trans -European Transport Network (TEN-T) by the EU,

generally available transport databases to supply the necessary statistical information for

modelling, planning and policy purposes were shrinking in scale and scope.

4.2 Background

The conceptual framework for Pan-European transport planning has been produced in the

Fourth Framework RTD (Research and Technical Development) project INFOSTAT and has

been termed ETIS (European Transport Policy Information System). In INFOSTAT (1998)

Final Report for Publication indicators and variables are presented in a systematic way

relevant for transport modelling, planning and policy issues. Under the heading of "Passenger

transport demand indicators" is found "Total annual interzonal passenger transport flow by trip

purpose, mode (or combination of modes) and type of trip chain." Similarly under the heading

of "Freight transport demand indicators" is found "Total annual interzonal freight transport flow

by commodity group, mode (or combination of modes) and type of transport chain." Both

indicator groups are to be specified on a relevant origin/destination (o-d) base and have been

given the highest priority in the report (fundamental in the terminology of the report).

- 17 -

In integrated and comprehensive transport modelling, such as the four-stage transport model,

the two first stages deal with transport generation/attraction and the distribution between

zones before transport mode distribution and transport route assignment are solved.

Transport between zones or as it is most frequently denoted origin/destination (o-d) flows are

basic to integrated and comprehensive transport planning.

Some remarks on the production of o-d matrices synthetically might be appropriate at this

stage. Transport models can create such matrices based on the particular model structure,

the model input and the model maximisation attributes. If there is no external database to

confront the model output with, there is no way of knowing how reliable the o-d matrices might

be, and models cannot be calibrated to achieve better fit to reality.

So there are good reasons why o-d transport matrices should be produced empirically for

some base years. From these base years the matrices can be updated by various techniques

to any recent year relevant for model input and output confrontation. In this manner there

could be a fruitful interplay between relevant transport data and transport models.

Today, however, no comprehensive and generally available set of Pan-European o-d

matrices exists for any base year. This applies both for

transport modes,

commodities (freight),

trip purposes (passengers),

transport chains,

countries and international regions.

4.3 Aim of the MYSTIC Project

The overall aim of the MYSTIC project, as indicated above, has been to develop methodology

and corresponding software to produce (estimate) origin/destination (o-d) matrices for both

passenger and freight transport at the European level. Although the overall aim is the Pan-

European country-to-country flows, a regional specification (NUTS 2 or more detailed

specification) at both ends of the flow is desirable as well as some commodity specification for

freight transport and travel purpose for passenger transport. The o-d matrices should be

produced for those single transport modes being in the most keen competition with each other

and for the most important transport mode combinations (transport chains) as has been

explicated in chapters 2 -4 above.

- 18 -

The methodologies for o-d matrix estimation to be developed in the MYSTIC project as has

already been underlined above, have to be suitable for application to typical Pan-European

transport modelling, planning and policy problems. In addition the o-d matrix estimation

methodologies developed must cover both current conditions when ETIS is not yet available

as a total database and a future situation when ETIS may be established.

The current data situation is even so problematic that available past data for the base

(reference) years are so old that o-d matrices have to be estimated (calibrated) for more

recent years to be really useful for policy and analytic purposes. Although the MYSTIC project

deals with o-d matrix estimation for the past, the project is closely related to forecasting

issues as mentioned above. Models to be used for o-d flow forecasting at the European level

are significantly influenced by the kind of data that are available. On the path to the future

ETIS the MYSTIC project has, therefore, analysed and proposed several methodologies,

procedures and techniques and corresponding software. This chapter presents the framework

for these proposals.

4.4 The Framework

Relevant data that might be exploited for the production of Pan-European o-d matrices are

regularly compiled by The European Conference of Ministers of Transport (ECMT), the United

Nations Economic Commission for Europe (ECE) and EUROSTAT of the EU both in

conjunction with each other and separately. EUROSTAT compiles both foreign trade statistics

and transport statistics as a consequence of several directives. Many European countries

produce trade and transport sta tistics and national travel and freight surveys more or less

regularly. Roadside traffic counts are numerous and the various transport modes (companies

and national as well as international organisations) might have data sufficient for detailed o-d

matrices for some past years. So the situation is not completely bleak, but offer a great

challenge for the maximum use of existing data and knowledge.

One disturbing problem is the increasing tendency for transport companies and organisations

to keep transport data confidential even when such data are only intended for statistical

purposes. In the MYSTIC project an appeal was made with the help of Directorate General

(DG) VII, to a number of countries to make existing data on passenger transport available for

the project. This has on the one hand created value added by exploiting and combining

existing data from responding countries into the beginning of Pan-European o-d passenger

transport matrices for road and rail transport, and on the other hand paved the way for a

continuing improvement, updating and extending of the matrices. Further success will depend

upon co-ordinated co-operation from many countries, and therefore it is important to build

relations with the data holders.

- 19 -

Within the given cost and time budget constraints the MYSTIC project has worked under the

aim to come up with cost-effective procedures for statistical sound data compilation and

processing in relation to Pan-European o-d matrices. The constraint necessitates, however, a

focus on those procedures that promise an answer within reasonable time and costs.

The MYSTIC project seeks to derive in a sense a best-estimate o-d pattern with as recent

data sets as possible, to specify the need for collecting new data sets so as to complete the

o-d pattern together with developing compatible methodologies for modelling the transport

mode choice.

The project deals with new statistical and computing methodologies, which are available and

beginning to be applied to transport. In general terms the project addresses issues such as:

merging different data sets, with different levels of precision and different levels of

confidence and still providing some best-estimate base year o-d matrices;

improving existing and developing new survey methods for collecting better data for o-d

matrix estimation.

The MYSTIC project has produced a framework within which the different passenger and

freight o-d transport chain matrix building procedures have been fitted. It comprises Pan-

European survey methodology, framework for the analysis of such data and storage within

the ETIS, and framework for connecting together existing data sets from European countries

so as to build matrices comprising the whole of Europe. This demanding challenge the

MYSTIC project has answered somewhat differently for passenger and freight transport due

to a different database situation.

For passengers the work in the MYSTIC project has shown due to information gathered from

several countries, that it might be possible to produce matrices approximating First

Generation Pan-European o-d Passenger Matrices.

To arrive at the conclusion relating to passenger transport the MYSTIC project has studied

and exploited transport models, algorithms and available data sets from several countries

both for their usage in producing o-d matrices and for proposing methodologies for future

usage.

Within the MYSTIC project work has been carried out to find methods and models usable for

data analysis and data production relevant for the project and exploit some of them. With in

the framework context of this chapter can be mentioned efforts to:

identify and use methods to harmonise and merge o-d matrices from different sources,

- 20 -

identify and use methods to transform limited information such as link counts into relevant

o-d information,

study scope and limitation of transport models to generate o -d relevant information,

exploit transport models and case studies for their ability to create o-d relevant

information,

assess statistical precision in the o-d data and data production procedures.

The classical transport planning model has been extensively used as tool in creating master

plans for urban areas. Efforts have also been put into using the model for regional planning,

but only more recently has the model been applied to nations. Few attempts have been made

to use it on a Pan-European scale. Thus there is a very limited volume of data available in

Europe on border crossing travel, and in the MYSTIC project the consortium have had to

make maximum use of national data, lacking data on travel between countries.

Most efforts put into the MYSTIC project on the passenger side was related to the task of

developing methodology for building Pan-European passenger o-d observed matrices from

different data sets and sources. The study has developed passenger o-d matrices for Europe

in the most complete form possible recognising the current data and modelling constraints.

In relation to o-d passenger database generation the MYSTIC project has been able to get

and process information from eight countries: Denmark, Finland, France, Germany, Great

Britain, Netherlands, Norway and Sweden. So far no Central and Eastern European countries

(CEEC) are included. Focus has been on producing Pan-European o-d matrices for the rail

and road transport modes.

4.5 Key Issues

4.5.1 Data Owners Confidentiality

We found that some data owners were willing to make their data available to us in its entirety.

However there was also the opposite - a distinct reluctance by data owners to give their data.

This is not a new problem and we had to address it in detail – and solve it - if we were to

develop a methodology for building the pan European o-d matrix using existing data. The

database approach offered us a way to ameliorate the impact of at least some, of the

concerns of data owners.

In some cases data is increasingly being seen as a commodity of value and some data

owners are reluctant to ‘give’ their data to others – especially if they gain nothing in return.

- 21 -

The database approach allows them to retain control of their data. In this way data does not

need to be given to the database system and the data is clearly under the control of the data

owner. There would need to be some contractual arrangement between the data owner and

the operator of the database system that could be a normal commercial arrangement (or can

be free). This project sought to use the case study to assess these aspects for suitability to

see whether they could contribute to the o-d database part of the ETIS.

We found that some data owners were willing to make their data available to us in its entirety

but this could well become increasingly unlikely especially for the more powerful datasets. It is

also increasingly important for data to be maintained – preferable at its source. The idea of

the database approach was that data owners, who generally had a vested interest in making

sure that the data was correct and well maintained, should hold the source data. In the

database approach the database system would connect to the source database and draw an

extract from it, which it would use for matrix building. As more source data is added to the

source database the data owner would undertake the necessary processing tasks and

maintain the integrity of the source data and keep it up-to-date. This is not a task that the

database system would wish to undertake and is best undertaken by the data-owning

organisation itself anyway. In the database approach the important point is that the

connection or linkage to the source data should be maintained as part of the database system

so that as further data is added to the source data future extracts are easy. It could be that

the source data is corrected or updated or data is added in some way, in which case a better

version of the source data can be accessed and a n extract taken for matrix building.

The database approach to matrix building could provide an effective solution to this problem if

the case studies were successful. This view is endorsed by the success of the DETR

research project, which successfully negotiated the use of highly confidential rail ticket sales

data for use by MYSTIC and others.

4.5.2 The Treatment of Air Passengers

The derivation of the pan-European o-d matrix for air passengers was thought to be important

and to be a question that MYSTIC should address. It was considered that analysing the road

and rail o-d matrices from the case studies and using the results to draw relevant conclusions

could address the problems of deriving an air passenger o-d matrix. One of the richest data

sources for air passenger o-d movements is that obtained from accumulating the individual air

ticket sales. These data are collected and could possibly be used for o-d matrix building

although there are some organisational problems (e.g. several organisations are involved in

the data) and logistical problems (e.g. that they span several computer systems) to be solved.

At the commencement of the project the European Commission were investigating the

possibility of assembling the air ticket sales data into their own database and it was felt to not

be part of this project to repeat this exercise (or even part of it).

- 22 -

However the issue of using air ticket sales data centres on the problem of relating the airport-

to-airport nature of a ticket to the ultimate origin and destination of the person using the ticket.

Consider the case of a single stage trip from origin zone to origin airport to destination airport

to destination zone. The use of airport-to-airport ticket data translates into finding the

travellers origin zone when the boarding airport is known and finding the destination zone

when the alighting airport is known. This problem is more complex when two or more aircraft

are to be used depending upon whether there is one (or more) ticket(s) for whole trip (i.e.

from the first airport visited to the last airport visited). The problems associated with using

ticket sales data for air are similar to those using rail where the ticket sale relates to the

boarding station to alighting station and it is required to infer the origin zone to destination

zone. They also exist for other public transport modes. The use of ticket sales data could be

analysed using rail and during the course of this project we (and others) were asked by DETR

to develop origin-destination matrices for rail passengers using ticketing data as part of

another exercise. MYSTIC used the UK rail passenger o-d matrices developed under this

exercise and we were able to analyse the rail methodology and understand how it could be

applied to air. This could provide a suitable methodology for developing pan European air

passenger matrices.

4.6 Passenger and Freight o-d Matrices

In the MYSTIC project dealing with passengers, 1997 was chosen as base year.

The main problems encountered, to turn the limited information available into approximations

to Pan-European rail and road o -d matrices for the selected countries can be summarised as:

turning information from years other than 1997, frequently as far back in time as

1992/1993 into 1997 data,

turning national travel surveys and other information giving little or no data on foreign

travels into international specification on countries and international zones in the NUTS

administrative system,

turning national roadside link counts on average annual daily traffic (aadt) and other

partial information into relevant Pan-European o-d information,

releasing and transforming railway information into relevant Pan-European o-d

information,

transforming national travel surveys covering resident population between certain age

groups into travels by all people irrespective of age groups and nationality,

- 23 -

distinguishing between more infrequent long-distance Pan-European travels and the

more frequent short-distance border-crossing shopping and leisure trips,

getting estimates on the use of transport chains by travellers between zones.

For freight the project is based on alternative approaches.

4.6.1 The Use of EDI to Gather Freight Data

At the more micro level the MYSTIC project have tried out a few approaches to o-d

estimation. One such approach has been the possible use of stored data in the computer

system of operators, another a shippers survey and developing methodology for estimating o-

d international transport chain freight matrices and for understanding behaviour in modal

choice.

In the first of the above-mentioned approaches the MYSTIC project wanted to:

assess the extent of computer based control systems among transport operators;

find out the content of these systems,

develop and evaluate statistical procedures in order to compile statistical descriptions of

all transport chains in the overall o-d flows.

The study has been based on interviews with operators in two countries. If the aims of the

study were fulfilled, this approach could be used to get information on o-d freight flows from

operators. To find out the potential of this source of information, the MYSTIC project had

divided the study into some distinct tasks, as follows:

data description of the elements of interest,

enumeration and classification of a frame of transport operators,

interview programme,

pilot electronic data interchange (EDI) collection,

methodology description and report preparation.

4.6.2 Producing an OD Transport Chain Freight Matrix

The other freight transport study mentioned above was focused on transport chains and dealt

with developing:

- 24 -

new statistical methods of registering flows aiming at building up o-d transport chain

matrices of freight flows,

a behavioural model of the organisation of a transport chain.

The objective of the work has been to give a basis for the necessary future data collection

with focus on getting a method for producing an o-d transport chain matrix on a regular basis.

The sub-task dealing with behavioural analysis was considered important because it would

provide insight in the significant factors that play a role in the decision process of modal

choice.

The shippers' survey have been based on interviews with shippers in France and the

Netherlands and renders regionalized o-d information specified on commodity, type of

transport chain and mode combination to the extent it is possible to gather information from

shippers in such a way that it represents all flows. Estimating o-d single mode matrices is also

possible by decomposing transport chains. By this it may be possible to study the probability

of using unimodal transport or a multimodal transport chain, including applying mathematical

models in order to predict the choice of the transport chain.

The MYSTIC project has shown the technical possibility to apply the transport chain concept

developed by the INFOSTAT project. As such the behavioural analysis has contributed to

identify and to measure some of those explanatory variables that are required in the ETIS

system as instrument of policy decisions.

- 25 -

5. Technical Description of the Project

5.1 Introduction

As indicated in section 2 the overall aim of the MYSTIC project was to develop a methodology

or set of methodologies for building origin-destination matrices of passengers and freight

movements and to build a pan European matrix or matrices of passenger movements based

on the available data.

This latter objective was introduced to the project in late April 1998, some three months after

the project had begun and it involved a change in the direction of the project that meant that

efforts in the first stage of the project were misdirected.

The project was divided neatly into two major streams of work - passengers and freight, and

within the freight section there was a further subdivision into firstly an examination of the use

of Electronic Data Processing for collecting data on the movement of freight around Europe,

and secondly on the use of a shippers’ survey to identify the movement of freight.

The Project was divided into 7 work packages - WP1 and WP7 were for coordination and

reporting respectively and WP2 was concerned with the production of an Inception Report

and the development of an overall modelling framework. The passenger work was divided

between WP3 and WP4, the EDP part of the project was addressed in WP5 and the shippers

Survey in WP6. The progress of these work packages and the results achieved are

described in more detail in the following paragraphs.

5.2 Inception Report and Modelling Framework (WP2)

As the first deliverable from the MYSTIC project, an Inception Report was submitted to the

European Commission on 12th May 1998. This report contained an outline description of the

work to be undertaken in the project and described the links between the different subtasks

within work packages and between the work packages.

The Inception Report contained modifications to the distribution of individual partners’ work

allocations between the different work packages as more was learned about the way forward

for the project. It also contained detailed specifications of the work plans and sub-task

descriptions for the three major streams of work. The production of the report was carried out

principally by AGDER, though the project coordinator PDC undertook the final editing.

- 26 -

The Inception Report constituted the first Deliverable (D1) of the project, outlining the work to

be undertaken and describing the links between the different sub-tasks within each work

package, the links between the work packages, and the links between MYSTIC and other

related projects in more detail than was possible in the Technical Annex. The Inception

Report also serves as a document of reference for the partners throughout the lifespan of the

project.

It outlined in detail the processes and deliverables of the project. The Technical Annex

constituted the scientific aspect, and is what the end product will be evaluated against. The

primary purpose of the Inception Report was to refine the link between the Technical Annex

and the research to be undertaken, by developing and establishing an ex ante overall

methodological framework for the project. Secondly, the document contained outlines of the

work to be undertaken in the individual Work Packages, the so-called Work Programmes that

are important guiding instruments for the accomplishment of the research. This was seen as a

continuous process starting with the initial thinking around the proposal and leading to

recommendations of the project

5.3 Passenger Activities (WP3 and WP4)

These two work packages were closely interlinked and are described together. The general

objective of WP3 was to develop new methodologies for data analyses to support, in

particular WP4, and WP6 also.

The following flowchart describes the MYSTIC methodology. MYSTIC was intended to deliver

a ‘process methodology’ - that can be then be used in the future.

- 27 -

MYSTIC Passenger O-D database: Architecture

Comprises data and processes

Raw O-D dataN, S, DK, NE, FR, DE, UK

Synthetic matrices e.g. STREAMS

Harmonised O-D database by country

Available data O-D matrices

HARMONY.DB

HARMONY.DB

Matrix buildingprocedure

BUILD.PR

Validation & sensitivityanalysis

RAW.DB

BUILD.PR

Harmonisation procedures

The sub tasks for WP3 were focussed on the following subjects

ST1: development of a methodology for describing national O - D raw data sets in a

standardised way

ST2 target definition of the pan - European O - D matrix to be built

- 28 -

ST3 methodology for harmonising national O - D data sets

ST4 methodology of merging and combining O - D matrices

The results of these research activities serve as tools for the WP4 European

Passenger Matrix

ST5 raw data description guidelines

ST6 O - D data harmonisation guidelines

ST7 merging procedures for O - D matrices

ST8 combination procedures for O - D matrices

From a statistical viewpoint both the merging and combining of O-D matrices required

complex mathematical procedures. Only the ‘classical’ merging problem where all O - D

matrices refer to the same zoning system and where variances are known can use traditional

methods. In the MYSTIC project however O - D matrices from different countries use different

zoning systems (grouping of cells in external areas) that gave rise to the ‘combination’ rather

than ‘merging’ problem. Moreover, since national O - D matrices are partly synthetic

variances of O - D matrix elements are frequently unknown. As a consequence demanding

mathematical work had to be done in WP3 but this resulted in the development of new

methodologies for matrix merging and combining.

The Passenger Working Group in its initial meetings formulated a set of sub-tasks for WP4

that were incorporated into the Inception Report. Further meetings following the submission of

the Inception Report expanded the list of sub-tasks to the following:

ST 1 Definition of scope and detail

ST 2 Collect raw O-D datasets

ST 3 Produce harmonised O -D datasets

ST 4 Develop and build matrices

ST 5 Merge Europe – wide road and rail O – D matrices

ST 6 Enhance matrices with synthetic data

ST7 Validate assembled matrices

ST8 Report and recommendations

- 29 -

Tasks 1 and 2 - Definition of Scope and Collection of Raw Data

The initial element of WP4 was to identify those countries and data sources that could

contribute significantly to the required data sets. A core of countries was identified from which

data would be central to the project and other countries from which data would be sought to

support the dataset coverage. The core countries were:

UK Sweden

NL Denmark

German Finland

French Norway

‘Supporting countries’ comprise the following:

Austria Italy

Belgium Luxembourg

Denmark Portugal

Greece Spain

Ireland Switzerland

Data was received from all the core countries listed above. In the case of the UK the data set

was particularly rich and considerable effort was spent on the merging and harmonisation

processes.

- 30 -

Table 5—1 Summary of Data Sources and Types

Country Mode Base

year Source

Number

of

zones

Geographical

area

National

or Inter-

national

Institution

data holder

Rail 1992 Ticket sales Road

Administration

Road 1992

Roadside

interviews, car

ownership

register

1000

Parishes and

groups of

parishes

Inter-

national Rail

Administration

Denmark

Air 1997

Airports Internation

al

Copenhagen

airport

authorities

Rail 1992 NTS

450 Municipalities national Road

Administration

Road 1992 NTS

450 Municipalities national Road

Administration

Finland

Air 1992 NTS

450 Municipalities national Road

Administration

Road

Rail

France

A7ir

1993

Enquete

Transport

Departements

NUTS3/NUTS

2

National

and Inter-

national

Service

Economique et

Statistique

Road 1991 RSI, modelling 453 NUTS 3

Rail 1991 Tickets,

Modelling 453

NUTS 3

Germany

Air 1991 Counts,

Modelling 194

NUTS 2

Inter-

national

Ministry of

Transport

- 31 -

Country Mode Base

year Source

Number

of

zones

Geographical

Area

National

or Inter-

national

Institution

Data Holder

Road 1981-

1991

RSI, IPS, 3000

Study specific

zones

Rail 1997 Tickets,

modelling 11,000

Wards

(NUTS5/6)

Grt Britain

Air 1995 CAA

Inter-

national

DETR

Rail 1995 Ticket sales,

counts 435

municipalities TOI, State

Railways

Rail 1997 Ticket sales 19 Counties State Railways

Road 1995 Car ownership

register, counts 435

municipalities

Air 1995 Travel survey,

airport statistics

National

TOI

Norway

Air 1997

National/in

ternational

Directorate for

Aviation

Rail

Road

Air

1993

NTS 668

sams

National

Rail

Administration

Sweden

Rail 1992

Tourist data-

base, counts,

hotel statistics

300

Eurostat,

Regions 1989

Inter-

national

Transek AB/

Swedish

Railways

Task 3. Produce harmonised o – d datasets

In Task 3 there was a requirement for software to be available to translate and transform the

different datasets into the harmonised form. Harmonisation is concerned with ensuring that

the data from the different countries is to a consistent specification. This included, for example

- 32 -

transforming the geo-referencing details on the data to a common zonal system (NUTS),

transforming modes and purposes to a common definition and adjusting the data to a

common base year. The following diagram illustrates the harmonisation process

Task 4. Develop and build matrices

Task 4 also required software to merge the different datasets and to ensure their statistical

specification. In the United Kingdom, the DETR commissioned a program called ERICA for

their regional models and PDC worked on developing that software to bring it up to date with

modern computing standards (e.g. 32 bit processing rather than 16 bit) and have been testing

it on the UK data. The data generally came with expansion factors so as to expand the data

up to represent a full days (or month’s or year’s) worth of travel. The data were then built into

o-d matrices by accumulating the expanded numbers of trips recorded between each pair of

- 33 -

zones. Generally the zone system to which each trip origin and destination was recorded as

going to (or coming from) was generally particular to that particular survey. Within the country

of origin, the trips were generally recorded to a zone system that accorded with the NUTS

system. However outside the country, trip origins and destinations were generally aggregated

sometimes to a zone system that bore little resemblance to NUTS. Differing external zone

systems were accounted for in the merging process whereby a zone was split up into the

NUTS system (generally at NUTS2) using demographic data.

A similar situation occurred with the UK datasets where the external zones were rather too

coarse than the consistent NUTS3 level required for the matrix building, and the zone splitting

was undertaken for the UK datasets too.

Many of the datasets were supplied to us as o-d matrices and in this case they by-passed the

matrix-building task. In some cases the harmonisation factors were applied after building the

matrices and in some cases it was best undertaken beforehand.

Task 5. Merge Europe – wide road and rail O – d matrices

The matrices built from each dataset were merged together so as to produce one Merged

matrix covering those countries for which we had o-d data. Country to country movements,

where we had only one observation from one dataset, were put directly into the Merged

matrix. Country to country movements where we had more than one observation from

different datasets were merged according to the merge methodology which usually meant

taking the average. Where variances had been calculated for both datasets then the

averaging could use variance weighting. In certain circumstances some datasets did not

contain all the trips (this was generally either due to the long distance cut-off or because the

dataset was from a household survey and therefore omitted trips from foreigners) and in this

case if there was an estimate with all the trips it was taken in preference to the former.

Splitting the external zones for a dataset was undertaken on the basis of population data for

the zones concerned. It was applied before the merging process.

When merging the matrices from the individual datasets, the merge conditions were set

according to the table given below. The merging process was undertaken in pairs of matrices

and matrices were merged by concatenating one dataset matrix into the result from the

previous merge. The two matrices being merged are given in the left hand columns and

“Result” means the result of the previous merges. Internal refers to a zone within the country

whose matrix is being merged and External refers to a zone in neither country.

- 34 -

From Internal A Internal B External

Matrix A Matrix B To Int A Int B External Int A Int B External Int A Int B External

Variance

Germany Denmark TA AV TA AV TB TB TA TB Max Default

Result Holland TA AV TA AV TB TB TA TB Max Default

Result UK IPS TA AV TA AV TB TB TA TB Max Default

Result Rest of

UK

TA AV TA AV TB TB TA TB Max Record

Variance

Result Norway TA TA TA TA TB TB TA TB Max Default

Result Sweden TA TA TA TA TB TB TA TB Max Default

Result Denmark TA TA TA TA TB TB TA TB Max Default

Result France TA TA TA TA TB TB TA TB Max Default

Key

TA Take Matrix A Value

TB Take Matrix B Value

AV Take average of the two values

Max Take the larger of the two values

Task 6. Enhance matrices with synthetic data

The merged matrix only contained an estimate of the origin-destination movements for those

zone pairs for which we had data (i.e. was included in one of the datasets listed above).

Those countries where we did not have any data, we took synthesised matrix cell estimates

from the STREAMS EC project. This included such countries as Spain, Portugal, Belgium,

Austria, Italy, Greece, and Ireland. The origin to destination matrix cells concerned were

simply replaced with the synthesised matrix cells.

Task 7. Validate assembled matrices

Details of the results achieved with the validation procedure are given in section 5.3.1 below.

- 35 -

Task 8. Report and recommendations

We have recommended the Commission organise a research project to design, set up and

initiate the first phase of a rolling programme of passenger screen line o-d surveys. The

project would also set up the passenger (and freight) o-d database together with the service

to supply information to government (and others), set up the database maintenance,

marketing and support as well as providing a general service of supplying matrices and tools

to users. Operating the database could become a near self-financing service, run on a five

year rolling contract from the commission with the organisation undertaking analysis for the

Commission and acting as a focus for data supply to others. The project should deliver a pan

European passenger and freight o-d matrix using the methodology developed in the tasks set

out above. It was further recommended that the Commission should instigate a rolling

programme of passenger and freight o-d surveys covering a ten-year period, which should be

repeated so that the matrix is never more than ten years out of date.

5.3.1 Results Achieved

In terms of the UK case study the results achieved show the merging of many different data

sets has been accomplished. There is a reasonably equal spread of trips over the areas from

which the data has been taken and that areas where there is most likely to be more trips, i.e.

London and Manchester, do attract more trips. It must be noted that this matrix is only

complete for trips within and to/from the four areas where the survey data has been used, i.e.

South East, Manchester/Yorks, Birmingham and Scotland. Parts of the Matrix starting and

finishing in places other than these are partial and do not contain all the trips.

The case study has shown a methodology that successfully merges data from different

roadside and other interview surveys. Many completely different datasets have been

combined to produce a single matrix that represents trip movements within and between the

locations where the surveys were undertaken. The success of this study stems from the

methodology and merging software being flexible enough to adapt to very different datasets.

The complete UK matrix is given as Annex 2.

On the European level the results of the study have also been successful in showing the

feasibility of producing a harmonised and merged matrix from different data sources. The

following tables illustrate the type of data the study has been able to produce.

The first table is a summary of some of the key o-d flows from five EU countries, measured in

000’s of passengers per annum for the year 1997 by road and rail combined, for trips over a

threshold distance of 50km to 100 km depending upon the dataset. They show that of these

countries, the highest flow is between France and Germany with around 57 million two-way

passenger trips per year; followed closely by trips between the Netherlands and Germany

with just over 48 million two-way passenger trips per year. After this come the flows between

- 36 -

Denmark and Germany (just over 32 million two-way passenger trips per year) and between

France and the UK (just over 26 million two-way passenger trips per year). There is a clear

gap then to the next highest flows between the Netherlands and the UK (6 million) and

between Denmark and France (4 million). The flows between Denmark and the Netherlands

and between France and the Netherlands were about 1 million two-way passenger trips per

year and between Denmark and the UK they were about a half a million two-way passenger

trips per year.

Destinations

Denmark France Netherlands UK Germany

Denmark 0 2208 639 233 16353

France 2043 0 638 13381 28880

Netherlands 561 636 0 2820 23982

UK 240 13381 2820 0 3149

Origins

Germany 16454 29567 23024 2996 0

Source: MYSTIC. These figures were obtained from merging different data sets that had different definitions e.g. some data sets

excluded trips less than 100km., some 80km., some 50km., and some were for country residents only. See Chapter 4, section 4.3

Harmonisation

Turning to the proportion of trips by mode, car or rail, for the above five countries, car carries

the most trips, with in all 93% of the market share (172 million two-way passenger trips per

year between the countries) with rail carrying the remaining 7% (13 million two-way

passenger trips per year between the countries). The following table show the market share

by country measured in terms of 000s of two-way trips in 1997.

Market Share by Country (000s of trips per year)

Denmark France Netherlands UK Germany All

Road 18.795 40.074 26.820 19.225 66.629 171.543

Rail 0.638 4.868 1.179 0.365 5.412 12.462

19.433 44.942 27.999 19.590 72.041 184.005

- 37 -

Source: MYSTIC. These figures were obtained from merging different data sets that had different definitions e.g. some data sets

excluded trips less than 100km., some 80km., some 50km., and some were for country residents only. See Chapter 4, section 4.3

Harmonisation

It is also important to consider the extent to which the outputs can be considered ‘validated’

The aim of a validation procedure is to ensure the data delivered in the merged / combined

matrices are on the one hand complete and consistent with respect to the input data and zone

representation (technical validation) and on the other hand to check the plausibility of the data

("quality" validation). In theory, there are many techniques conceivable to perform a matrix

validation however the procedures and the results of the validation process in MYSTIC can be

summarised as follows:

The technical validation was mainly performed by analysing the desire-lines and the data of

the intermediate results of the merging and combination process. Here, the technical

problems caused e.g. by different formats of the input data especially with respect to the

handling of border crossing trips have been revealed and have been solved. The technical

validation process illustrated how “good” the matrix was.

The following tables present some results of the quality comparison of the MYSTIC case

study data. As already described in the previous chapters, the emphasis of the MYSTIC -

project lay on the development of methodologies to merge and combine existing data. Thus

the project had to rely on the data available and did not have the possibility to influence the

quality of the input data. Therefore, a big variety of data had to be taken into account in the

case study that lead to divergent results if one compares the data country by country. The

table below shows the average number of trips per inhabitant and year for the different

countries in the MYSTIC Case Study Road Matrix.

In general it becomes obvious, that a border-crossing trip is a very infrequent event (in

average only a few trips per inhabitant and year!). Therefore, all methods relaying on

interviews of people face the problem of a scattered / diffuse data set needing very big

expansion factors to reveal total trips.

- 38 -

Trips per Inhabitant Country

Border crossing Internal

1 Austria * 4,80 0,00

2 Belgium * 3,14 0,00

3 Germany * 2,25 20,00

4 Denmark * 5,38 0,00

5 Spain * 0,05 0,00

6 Finland * 0,03 77,12

7 France * 0,80 3,85

8 Greece * 0,10 0,00

9 Ireland * 0,36 0,00

10 Italy * 0,22 0,00

11 Luxembourg * 13,45 0,00

12 The Netherlands 2,68 143,19

13 Portugal * 0,06 0,00

14 Sweden * 0,73 10,13

15 Great Britain * 0,46 60,49

* Denotes partial or inconsistent data

Trips per inhabitant per country in the MYSTIC Case Study matrix

The following table shows the average zone size per country for the NUTS2 and the NUTS3

levels. The table reveals considerable differences even for those countries with similar

regional structure. This means that due to the big differences of zone sizes the share of

intrazonal trips – trips that have origin and destination in the same zone – differs very much

between countries. The problem of intrazonal trips cannot be neglected, especially in the case

when no distance thresholds are considered. Thus, as intrazonal trips are not represented in

a trip-matrix, it cannot be recommended to use different zone sizes for presenting matrices

with no distance threshold! This is a fundamental problem (which occurs not in this extent on

a national or regional scale) and should be taken into account when a future ETIS is to be

made operational.

- 39 -

Consequently, the different size of NUTS zones is a fundamental topic when investigating

comparable data sets from different countries. Taking for example the trips within the

Netherlands and Finland. Although, both data sets have no distance threshold, the figures

'trips/inhabitant' are much higher in the Netherlands than in Finland not least due to the big

differences in the average zone size.

Area NUTS 2 NUTS 3

[km2] (number) [km 2/zone] (number) [km 2/zone]

Austria 83.858,3 9 9.317,6 35 2.396,0

Belgium 30.518,1 11 2.774,4 43 709,7

Germany 357.020,8 38 9.395,3 445 802,3

Denmark 43.094,4 1 43.094,4 15 2.873,0

Spain 504.790,0 18 28.043,9 52 9.707,5

Finland 338.147,0 6 56.357,8 19 17.797,2

France 543.964,6 26 20.921,7 100 5.439,6

Greece 131.625,0 13 10.125,0 51 2.580,9

Ireland 70.273,1 1 70.273,1 8 8.784,1

Italy 301.316,0 20 15.065,8 103 2.925,4

Luxembourg 2.586,4 1 2.586,4 1 2.586,4

The Netherlands 41.028,5 12 3.419,0 40 1.025,7

Portugal 91.905,0 7 13.129,3 30 3.063,5

Sweden 410.934,2 8 51.366,8 24 17.122,3

Great Britain 241.751,0 35 6.907,2 65 3.719,2

Total 3.192.812,4 206 15.499,1 1.031 3.096,8

Average zone size per country

Moreover, different zone sizes must also be taken into account when analysing pictures

representing desire-lines between zones. Naturally, the flow to one big zone is bigger than

each single sub-flow to three zones having together the same size as the one big zone. This

occurs for example for Denmark, which "unfortunately" represents on NUTS2-level only one

zone.

- 40 -

From the above tables it is clear that,

the distance threshold plays an important role, e.g. comparing the internal data for Germany (only trips longer 100 km) with the data from the Netherlands (all trips),

the zone size plays a role as well,

however, people in very small countries like Luxembourg make more border crossing 'trips/inhabitant' because the country is so small that more trips are likely to go across the border,

there are also other reasons for an unbalance of national data: Taken as an example France it is obvious that here is another fact responsible for the considerable low demand data (e.g. in this case the fact that the household data consider only French residents),

the countries with zero figures in internal data did not contribute to the MYSTIC case study and received the border crossing trips from other data sets.

Altogether, the validation of the MYSTIC European case study data showed that from the

technical point of view the methods applied worked and produced solid trip-matrices.

However, due to the disperse situation of the national data feeding the process (data with and

without distance thresholds, data generated from household and roadside interviews or

modelled data, consideration of residents only or of all travellers, etc.) it cannot be

recommended to use the Case Study Matrix for practical planning processes on the European

level. This is also true for the combined matrices containing the MYSTIC case study and the

STREAMS data. Here the same shortfalls apply.

However, the national data as such can be used separately, especially due to the fact that

they are tailored to the NUTS3 level. The following maps show for exa mple the results of the

validation of the German Data by performing assignments. The maps present the link loads

on the German road network for a total matrix including all trips between NUTS3 zones. The

tables above show the results of an assignment of the German input data for MYSTIC

(harmonised to 1997, 100 km distance threshold applied). Although there are differences in

the total level of trips it can be seen that the MYSTIC data represent the flows on the German

autobahn-network (often TERN elements) in a proper way. Only the trips in conurbations or

big cities show big differences due to the short distances applied.

- 41 -

Map1: Trips on the German Road Network – NUTS3 – no distance threshold

- 42 -

Map 2: Trips on the German Road Network – NUTS3 – distance threshold >100km

- 43 -

5.3.2 Presentation Software

An important part of the passenger side of the MYSTIC project was the also development of

the presentation software that was seen as an essential part of the dissemination and use of

the MYSTIC outputs.

The objective of the presentation software was to present the output of the passenger part of

the MYSTIC project in such a way that would be user friendly and easy to understand. It was

designed to allow the user to access the MYSTIC European matrix either in a non-graphical

way, by calculating number of trips between selected zones or in a graphical way with the

help of a GIS system. Moreover, this software was aimed at providing the documentation

corresponding to OD information on the data sources, the data sets and the methodology

used for the computation of the flows.

The presentation software was designed to serve two kinds of users. These who want to have

a friendly presentation of selected data e.g. collapsed tables, desire lines and those who want

to use the OD information (the tables themselves) e.g. transport experts. Both groups of users

need to have clear documentation on the data sources, data sets, and access to the MYSTIC

methodologies and this will be completed in the next period as part of the Final Report. A

CD-ROM will be produced for disseminating the software.

5.4 Electronic Data Processing (WP5)

The collection of complete chain origin and destination data by conventional questionnaire-

based means is now proving very difficult because of the loss of the detailed Customs

database from which to select a sample. The object of this work package is therefore to

ascertain whether statistics of complete origin and destination chains could alternatively be

obtained from the computer records of transport operators.

This Work Package was subdivided into 5 tasks as follows:

ST 1 Data description of the elements of interest

ST 2 Enumeration and classification of a frame of transport operators

ST 3 Interview programme

ST 4 Pilot EDI collection

ST 5 Methodology description and report preparation

The following paragraphs describe this Work Package in more detail.

- 44 -

Task 1 – Data Description of the Elements of Interest

The work package partners created a detailed list of relevant elements with definitions. This

was intended as a basis on which to describe freight movements, and forms the main input to

the Interview Questionnaire as required for Task 3. Task 1 was completed in March 1998

although the list and description of elements was not set in stone. It was left open to make

adjustments as may be required when other tasks in this work package progressed.

The conclusions derived from this activity were that the origin and destination of transport

chain, associated with description of the consignment being moved, are core elements of any

data collection to describe transport chains. This obvious statement is reflected in the

frequent separation of these data from other chain descriptors within and between logistics

operators’ electronic systems. Development of electronic systems for logistics (i.e. door-to-

door) operations is at an early stage, and expanding rapidly. To ensure availability of data

elements for statistical purposes, it is essential that official statisticians specify what

information is going to be required, and discuss with industry trade associations and software

suppliers how best to integrate these requirements into new systems, before these systems

are designed and implemented.

Task 2 – Enumeration and Classification of a Frame of Transport Operators.

The aims of the work package were:

to find out if transport operators’ information systems contain data describing transport

chains from origin to destination.

to evaluate if statistics describing the transport chain patterns can be compiled from these

systems using electronic means, and with a small or negligible burden on the operators.

to estimate what proportion of operators has such a systems.

to estimate what proportion of total traffic is covered by operators with such systems.

to estimate if the transport chain patterns of the subgroup from whom statistics are

collected is representative of the transport chain patterns of the whole population.

to estimate the quality of coverage of total transport obtainable with statistics compiled

from these systems.

to specify a methodology for compiling these statistics.

to specify methodology for improving the overall quality of statistics produced.

- 45 -

Therefore interviews were conducted with a sample of companies, from which o-d database

systems can provide information to compile statistics describing transport trips. The following

questions needed to be answered:

how representative are the interviewees of the group from which they are selected?

what proportion of the group would be able to provide information for statistical

compilation?

what will the sampling errors are in using results from those able to provide data as

representative of the whole group?

is the transport activity of the group from which interviewees are selected typical of the

transport activity of the whole population?

if not, for what sub-group is the sample representative?

how will the collected statistics of o-d patterns be scaled up to the group total?

what proportion of total traffic does this group cover?

what possible methods might be used to obtain statistics for the remaining parts of the

total population?

In order to answer these questions, a clear definition of the frame of organisations with which

we are concerned was enumerated. It was intended that groups of organisations in scope for

this study were:

Road transport operators.

Marine transport operators

Freight forwarders

Logistics operators

Inter-modal operators

Industrial and commercial companies carrying out their own logistics operations.

- 46 -

The first five categories were straightforward, and lists were compiled reasonably

satisfactorily. The last group presents more difficulties of identification. Initial suggestions

were that retail supermarkets should be included, since it is known that they do their own

account logistics organisation. But such an ad hoc definition provides no information about

the total extent of their own-account operations.

To avoid overlap and to enable this task to yield some estimate of the o-d activity as a whole,

based on a surveyed sample, it was felt that the highest level of operator would be the most

appropriate target. The focus would therefore be on (4) and (6) above. To that end, a

Logistics Operator for the purposes of this study was defined as follows:

A Logistics Operator/Manager is concerned with the provision of complete supply chain

solutions. As such, his activities will include the management or direct supply of the following

services: warehousing, consolidation, groupings, packing, transporting, unpacking, storage,

distribution and local delivery. It may also include ancillary value-added services such as

stock control, just-in-time management, pricing and repackaging for customer presentation.

Task 3 – The Interview Programme

On launch of this task within Work Package 5, it was decided at an early stage that the

questionnaire would consist of two sections: an initial brief section, known as the "pilot

interview", which will be precise but simple, and a second section which will be in more detail

and which will determine the data availability as required for this task.

The "pilot interview" will be with the managing director or the relevant department head of the

target Logistics Operator (LO). This interview is to determine general background data of th is

operator, nature, extent and names of sub-contractors, total activity in throughput and

estimated market share. The "main" interview will be conducted with the person responsible

for data processing (this could be the same person for a smaller operator). This will gather

information on data systems, nature of inputs, and aggregation of information and possible

outputs as required for this work package. The potential output data will be compared with

the parameters enumerated in Task 1 to determine if such data could indeed describe a

complete supply chain for the purposes of the MYSTIC project.

The questionnaire will therefore be along the lines noted below. It is structured on the basis

that the only interest is in those operators who have an electronic database. However, not all

questions are always relevant for all operators. Once the details of the questionnaire were

- 47 -

agreed, this was formatted into a numerical, logical-flow structure with blank boxes on the

right hand side to allow for varying lengths of answers.

It was agreed by work package partners that each interview would be initiated with a letter

explaining the purpose and objectives of the interview. For the United Kingdom, this letter

came from the DETR and was addressed directly to the managing director or department

head of the targeted company. The letters were sent out in three tranches and a period of up

to two weeks elapsed before follow-up telephone calls were made. Of the original 60-

targeted companies in the six sectors, 30 were finally interviewed.

Interviewing was carried out in the period from January 1999 to June 1999. Considerable

initial effort was needed to arrange the interviews. Having decided which companies to

interview it was necessary to telephone them to establish the name of the most suitable

contact. In some cases this took little time, but in others it took several calls to establish the

correct person. Two people both arranged and carried out the interviews. Arranging

interviews was also time consuming because it could sometime take several calls before the

potential interviewee was there to receive the call. Most companies were willing to take part,

although there were some refusals. On average interviews took about an hour, but when

travelling was taken into account the whole process could take several hours. It was often

difficult to get all the information required from one person. Sometimes they could give partial

or estimated answers. At other times it was necessary to speak to somebody else, or

because of the expense of carrying out a second interview to leave the question unanswered.

In most cases however enough basic information could be obtained in order to understand

the practicalities of getting supply chain information from company computer systems.

The broad conclusions from UK survey were that the results demonstrated that the data

required were available but, particularly for international shipments, not necessarily all in one

computer system. To monitor each complete supply chain it would ofte n therefore be

necessary to extract data from more than one system, especially when subcontracting was

involved. Considerable resources would therefore be needed to set up a data extraction and

analysis system.

The UK survey mainly covered the larger companies within the sectors covered. One way

forward might be to set up an electronic collections system for these larger companies, whilst

still using conventional paper-based methods for others. The burden on companies providing

- 48 -

data could however be significant at the set up stage and they would need to be sure that

they were going to see some major benefit.

At the UK interviews companies gave estimates of market share and through put in a variety

of units e.g. revenue, units of production, pallets, journeys. It would be necessary to establish

that a standard unit of measure existed that could be used for grossing to population totals.

Conclusions from the Netherlands surveys were that to undertake a survey in order to collect

information about the level of use of electronically means was very useful for Statistics

Netherlands. The outcome was more positive than thought beforehand.

In case a similar regular survey should be held, the frame should be defined properly. This

should be done for weighting purposes afterwards. It should be possible to produce results

for all Dutch enterprises.

It should be studied how this survey could be combined with existing obliged surveys.

Information about the level of electronically data exchange could lead to more integrated data

collection procedure. In this case more information could be collected at one time. This

would very much decrease the response burden for the enterprises and also speed up the

production of figures. It was also concluded that is certainly worth to study a pan-European

enlargement of the survey.

Task 4 - Pilot EDI Data Collection

The questionnaire was designed as follow up to the initial sample in particular to test some of

the initial responses against the reactions of the respondents' IT department and to explore

details of the data capture potential.

Although much of the core data is available in company systems it is not structured in the

manner envisaged by the initial questionnaire. For example for statistical purposes some

goods movements can be viewed as a chain with each multi-modal element being regarded

as a link, while commercial records only show it as a rope only recording the total movement.

Equally statistically a set of deliveries from a depot is seen as a circle statistically (with a+b+c

- 49 -

travelling from the depot to A, b+c from A to B etc) while commercially the movements are

recorded as a star pattern (i.e. Depot to A, depot to B etc.)

Information is scattered across systems, companies, countries and even continents. Fo r

example one organisation was only responsible for exports, others in the group were

responsible for internal distribution, imports and supplies of raw materials. Equally it was

common for data to be split across a manufacturer’s systems and those of its haulers and

between operational and accounting databases - in some cases in different IT centres. In

some cases a fuller high-level picture may be available from International Head Offices, but

the difficulty of obtaining this sort of information from for example American or Japanese – let

alone “flag of Convenience” operators is well known.

There were no common messages structures in use that would allow extraction of the core

data. In many cases even where "standard" EDIFACT messages were in use the comment

was that every customer used a slightly different version so in effect each message was

bespoke.

The conclusions from this task were that there is a willingness on the part of United Kingdom

industry to participate; however the data is not easily ava ilable from the systems sampled.

Once the information is received in a suitable format there is no difficulty in creating a suitable

database and then designing enquiries to answer statistical questions. The major problem is

collecting the data in an appropriate single structure since commercial organisations,

understandably, hold the information the information in a manner appropriate to their

operational and accounting requirements rather than for the benefit of European Statistics.

The only organisation that may have know of the complete movement from origin to final

destination appears to be the Originator (the Exporter or Manufacturer), however unless they

are also the Transport Operator they almost certainly will not have full details of the multi-

modal movement (For example they may well not know details of ferries, depot reorganisation

of loads and will certainly not be aware of co-loaders when the movement is a part load).

It appears that there may be a need to collect partial information about the Transport Chain

from several organisations and for the Collecting Authority to collate this with all the

complication of achieving Unique Consignment References and eliminating double reporting.

Existing systems have difficulty in supplying the information in the structure required. It may

well be that the best solution is for the European Statistical authorities to define the minimum

data set needed for planning, and then to discuss the requirement with European and

National Trade bodies, with a view to obtaining Memoranda of agreement that individual

organisations will include the requirements as they redesign and replace their internal

- 50 -

systems over the next decade or so. However, this does not solve the immediate statistical

gathering problem.

Task 5 - Methodology Description

The methodology proposed for the collection of complete transport chain data is set out in the

following paragraphs.

The sample frame should be a list of industrial companies generating consignments. This

frame should be stratified by sector, since the characteristics and complexity of transport

chains used is significantly different between sectors. The frame can be compiled from

business registers classified to industrial nomenclature, supplemented by information from

trade associations. Standard register data needs to be expanded by information about use

of electronic systems for control of transport and logistics operations. Such additional

information will require a preliminary survey enquiry to identify those companies from whom

data can potentially be collected electronically. This preliminary simple survey will identify

those using electronic systems to control logistics, and those using third party logistics

companies. At this stage, it may be necessary to obtain information about the volume of

transport carried out for each company, to be used as part of estimation of the total transport

characteristics.

Collection of data electronically from company systems implies expenditure in setting up the

collection process and links. If this expenditure of resources and/or money is by the

collection agency, then it may be possible to select a fractional sample of the population from

which the total picture can be estimated. However, the concept is that data on all transport

chains used by each company will be collected. Even if the costs are met by the collecting

agency, some companies are likely to be reluctant that only their data be used to represent

the whole sector. If set-up and collection costs are to be met by the companies, this

reluctance will be reinforced.

A separate set-up is likely to be required for each company, taking account of the technical

difficulties. Once in place, electronic data collection will then be straightforward. Validation

and correction procedures will be required, although the setting up process should enable

standardisation of data collected. The reason the process is expensive is because the basic

information is not in a standard form.

5.5 Shippers Survey (WP6)

As indicated in Section 2 of this report the research tasks attributed to the Shippers surveys

part of MYSTIC had two objectives:

- 51 -

To develop a behavioural analysis of the modal choice for international transport chains

To produce a methodology for estimating Origin/Destination international transport chain

freight matrices.

To fulfil these objectives, two surveys were carried out: an in-depth survey of shippers in

France and in the Netherlands in relation with their exporting activities, another one devoted

to the volume of international trade. Most resources wee spent on the first survey in order to

fix the behavioural organisation of shippers.

In many aspects the shippers surveys in MYSTIC were very innovative. Not only did they

provide a deep and new source of information on the transport chain organisation of

international flows of these two countries, but also they constituted a basis for a wide

approach of freight transport of other countries. The question arises at the end how the

experience built up to develop the following orientations in the shippers surveys in MYSTIC,

could serve to expand the approach on a European scale:

Testing the feasibility of tracing international flows and capturing intermodal chains

through dedicated shippers’ surveys

Understanding the influence of the organisation of the transport chain on the physical

flows

Identifying the logistic determinants of the freight transport

Experimenting with methods to combine shippers’ surveys quantitative results with

traditional customs and modal data to produce the real o-d matrices region to region

The behavioural research task was therefore to develop an analysis of the organisation of a

transport chain and subsequently develop new statistical methods aiming at building up

transport chain matrices of fl ows. Parallel to the behavioural shipper survey, the o-d shipper

survey was held in order to estimate the level of trade by activity and on a region-to-region

basis. The focus lies on the France – The Netherlands flows and shippers were surveyed in

the Netherlands. The objective was to fix whether and to what extent a limited survey among

firms might be an instrument of estimation of total international demand for transport on a

regional level. If successful it would be a promising method, helping to fill current and growing

statistical gaps. Combined with the behavioural model and possibly other sources and

methods, it might offer a tool for building up a multimodal European database of freight flows.

As detailed in the Inception Report, the Shipper Survey has been divided into 10 sub tasks as

follows

ST 1 Lessons of the past and effects for MYSTIC

- 52 -

ST 2 Lessons of other RTD projects

ST 3 Preparation of the sample

ST 4 Selection of survey method

ST 5 Methodological approach of the behavioural problem

ST 6 Establishment questionnaire and test

ST 7 Actual surveying

ST 8 Analysis of results

ST 9 Development of an OD method

ST 10 Development of a behavioural model

A summary of those sub tasks are described in the Inception Report:

The following paragraphs describe progress with the Work Package in more detail

Tasks 1 and 2 Lessons of the past and effects for MYSTIC and lessons of other RTD

projects

Task 1 was completed with the production of a report by INRETS on “Lessons of the past -

the French experience” which drew upon past shippers surveys carried out in France to

identify where changes should be made and where ideas could be carried forward. A draft of

a report for Task 2 (“Lessons of other RTD projects”) was produced by AGDER. Two

shippers' surveys were described in detail for Task 1

Large-scale Dutch shippers survey 1987

French shippers survey 1988

Also three other surveys were described in less detail:

Shippers survey port choice model 1989

Shippers survey mode choice 1990

NIPO survey 1996 (NIPO is a Dutch market research institute)

From the literature study it appeared that the Dutch surveys were seldom used (for various

reasons) for modelling or a more in-depth analysis. On the other hand, the French survey was

- 53 -

used in a large number of projects. The success of the 1988 French survey in the past was

one of the reasons for using this approach in the MYSTIC survey.

Tasks 3 and 4 Preparation of the Sample and Selection of Survey Method

In order to assist the design of the MYSTIC shippers survey for tasks 3 and 4, the following

were obtained:

translation into English of the questionnaire from the Nord Pas de Calais shippers survey which is a project which has begun with shippers in that region who have international chains

the production of an economic typology and a transport typology of shippers in the Netherlands by NIPO;

the production of international road transport flows by region in the Netherlands and France by CBS;

a synergy typology analysis NIPO / CBS: common definition of sectors of activities and CBS tables of number of firms per SBI sector

a comparison of O - D analysis issued from the NEAC system with French custom information by INRETS;

analysis of Kompass business data by INRETS;

an approach to the INSEE economic database by INRETS.

For task 4 much work was done on the typology of the shippers in relation to the constitution

of the matrices on a regional level. It was concluded that a two-step approach would be taken,

separating the problem of total transport demand (economy related) from the question of the

type of transport chain (transport related). This would lead to an O - D matrix of the total

volume of exchanges between regions by sector of activity / type of goods and a separate

register of the organisation of the transport of the exchange of these goods.

Task 5. Methodological Approach of the Behavioural Problem

The structure of the survey allows three main levels of investigation, each one including one

or several types of questionnaire forms:

The establishment level with the pre-interview and the establishment forms,

The shipment level with 3 selected shipments by establishment,

- 54 -

The transport chain level by questioning all the operators who are involved in the

transport process of the shipm ent (operator questionnaire) and describing the provided

transport services (journey leg questionnaire).

Structure of the survey

Task 6. Establishment Questionnaire and Actual Survey

These two tasks were the major part of WP6 and covered the survey process. This covered a

number of discrete steps as detailed below:

In the case of the pre-questionnaires and establishment questionnaires the first step

concerned questions about the economic and logistical characteristics of the firm. This part

was filled in with the director or the representative of the logistics and transport department to

whom a pre-questionnaire has been previously sent by mail.

The surveys carried out in the past did not include such pre-questionnaires. This preliminary

step was introduced in order to collect into the same form the main requested figures

concerning the firms activity. The idea was to let the respondent know about the approach

followed in the survey and to gather the required figures prior to the interview. It was also to

reduce the total time of the face-to-face interview. The questions covered

The yearly volumes the establishment has shipped

The shipments distribution by weight

The shipments geographic breakdown by large destination zones

- 55 -

The organisation mode of the shipments and the relation between the used transport

means and the large destination regions

The availability of container handling equipment and vehicles or containers on own

account.

In the face-to-face questionnaire, the previous questions were completed by information on:

Firms activity and size

Regularity of the sales along the year, and months of full activity for sales

Storage conditions

Trading partners with number of customers accounting for 80% at least of the turn-over

Accessibility and distance to different transport means

In relation with the WP5 aims, use of computerised information system recording the

transport movements

The second level of investigation referred to the shipments. The process was to select at

random three shipments among a list of ten.

The common questionnaire for all shipments asked the following types of questions:

The departure time of the shipment

The destination location and the nature of the customer

The yearly shipped volumes to this customer, in tons and in number of shipments

The modalities of the order

The product constituting the main shipment's weight and the eventual special transport

features as hazardous goods, controlled temperature, fragile products

The value of the shipment

The weight of the shipment

The conditions of departure

The mode of appearance of the shipment or load

- 56 -

The transport « exchange terms » determining who was contractually in charge of the

transport operation.

The procedure of identification of the different operators involved in the transport of the

shipment is one of the key-dimensions of the survey. The procedure (the same as used with

success in the last 1988 French survey) consisted of identifying the operators according to

the logic of contacting-contacted. Each interviewed person is asked to give the address of

those that he or she has contacted to organise the transport of the shipment, from one

operator to another, up to the complete reconstitution of the chain.

The final component is the reconstitution (operator questionnaire) and physical chain

(journey leg questionnaire). The survey was carried out by phone, among all the identified

operators. The questionnaires were related to the different profiles of operators: customer,

carrier or transport auxiliary, other shipper establishment or specific cases as the carrier of

the empty combined unit.

The first aspect of the operator questionnaire, the customer or consignee questionnaire, the

first question allows checking whether the customer has ordered for its own establishment or

for another one. In the last case, when the customer appears as a central buying office, only

the information related to his part in the transport was collected. This concerns the gathering

of the name and address of possibly contracted carriers and of the establishment for which

the order has been given and which will be surveyed next as the final consignee.

For other customers or final consignees, the collected information related to:

The main activity of the firm, its nature (production unit, warehouse, wholes ale or retail

shop) and its size (number of employees of the establishment and of the company it belongs

to),

The conditions of access to the main transport facilities,

The arrival time of the shipment, the agreed schedule upon this arrival time and the

possible shipment delay.

The particular case when the consignee is not a customer but a logistical operator was also

taken into account, including the identification of the principal (i.e. the agent who has given

the order of transport, generally either the shipper or his customer). Furthermore the type of

services which are provided are described: operations on goods as packing, labelling, or

products fitting, warehousing, dispatching on other destinations at local, regional, national or

international levels.

- 57 -

In the remaining part of the questionnaire the customer was asked the same questions as the

shipper on the division of the transport operation. If the customer had no part in this transport,

the questionnaire ended. In other cases he was questioned about his role in the shipment:

Type of transport (on own account or hired transport),

Carriers and third parties contacted,

Criteria for choosing the transport,

Transport price paid by the customer or consignee.

In the case of carrier or transport questionnaire these operators are questioned about the

characteristics of their firm with information notably on their nationality, the number of

employees of the establishment and their belonging to a multi-establishment company or to

an international group or network. The rest of the questionnaire allowed identification of the

place of the operator in the whole organisational and decisional chain with:

The identification of the principal:

The nature of the economic links with this principal;

The intervention of the carrier in the choice of the transport and his choice criteria;

And, as for all operators, the appeal to third parties with the identification of the operators

to which they have subcontracted the totality or a part of the operations they were in charge

of.

This information was completed by the already obtained information concerning the charge

split between the shipper and the customer and their role in the transport organisation. It then

allowed building chain variables that described:

Who is at the origin of the chain;

To how many carriers the shipment has been entrusted (operators of first level);

The appeal to subcontracting relations;

The type of this subcontracting:

- Simple subcontracting when the operators subcontract to only another one,

- Multiple subcontracting when they subcontract to several others,

- Subcontracting in line when there are several successive subcontracts.

- 58 -

The last part of the questionnaire concerns the price that the carriers or transport auxiliaries

invoiced for their services. With regard to the breakdown of this price for maritime shipments

between the different inland, port and maritime parts and additional data on the type of

maritime transport, and the criteria of choice of the loading port.

In the case of the shipper company questionnaire this questionnaire identifies the cases of

intervention of operators who are not transport professionals: other establishments of the

shipper's company which are involved in the organisation of the shipment or its carriage on

own account, or other shippers supplying transport on own account ("disguised" hired

transport).

Finally, in the case of the empty combined unit questionnaire the questionnaire concerns the

combined shipments departing from the establishment and for which the carrier who has

brought the empty unit is also surveyed. This questionnaire is very short and deals only with

the nature of the principal who ordered the transport for this unit and the place where the unit

was warehoused. The carrier is also asked whether he has made the carriage of the shipment

and in this case the interview goes on with the usual carrier questionnaire.

As referred to in above there is also a journey leg questionnaire. Each operator who has been

involved in a transport is questioned about the number of journey-legs he has undertaken,

counting as many legs as needed stops either to change vehicle or transport mode or to

operate on the goods for consolidation, stuffing, packing, etc.

The collected information allows describing:

The origin and destination of each journey- leg, the nature of these intermediate platforms

and their access to transport facilities

The time of departure and arrival which enables to know both the transport time and the

immobilisation time for logistics operations

The provided services

The transport mode used and the type of vehicle,

The characteristic of the line

The total weight of the load and the corresponding number of customers

It is then possible to describe the physical chains: the combination of modes or vehicles, the

number of legs, and all other characteristics as consolidation, containerisation or various

logistical services provided.

- 59 -

Task 8 Analysis of results

The procedures described above result in two matrices for the sum of all transport modes. A

separate harmonising procedure can adjust the two sets by some averaging method. The

overall fit between the two matrices was good and in view of the exploratory nature of the

MYSTIC project, no harmonising of the two sets of matrices was carried out.

To highlight implications of the calculations, tonnes were summed to four levels in the macro

presentation below:

Intra-EEA, CH area (European Economic Area and Switzerland),

Intra-CEEC (Central and Eastern European countries),

Deliveries from EEA, CH to CEEC,

Deliveries from CEEC to EEA, CH.

All Intra-European goods transport (1000 tonnes) 1997

GDP and

foreign trade

The sum of

transport

modes

Difference in

per cent

1 Intra-EEA, CH 1 431 881 1 432 465 -0.0

2 Intra-CEEC 56 944 58 678 -3.0

3 1 – 2 72 391 73 314 -1.3

4 2 – 1 122 561 121 610 +0.8

5 TOTAL 1 683 777 1 686 067 -0.1

The close results generated by the two approaches do not mean that the actual observable

result in the end might not be different from the projected one in the MYSTIC project. When

the ECMT has produced statistics for 1995 and 1996 by the end of 1999, a follow -up

procedure might adjust factors to this new information that probably will affect the coefficients

used and the contents of the matrices.

- 60 -

The following survey shows the overall Pan-European freight transport development

calculated in the MYSTIC project and combined with matrices produced in the INFOSTAT

project.

All Intra-European goods transport (1000 tonnes)

1989 1990 1991 1992 1994

1 Intra-EEA,

CH

1 094

229

1 117

962

1 140

306

1 178

292

1 256

552

2 Intra-CEEC 29 513 19 343 17 780 25 088 46 716

3 1 - 2 31 294 33 730 36 736 45 748 57 561

4 2 - 1 70 950 63 301 76 175 86 462 100 217

5 TOTAL 1 225

986

1 234

336

1 270

997

1 335

590

1 461

046

The two surveys presented above show the dominating role of transport between countries in

the EEA, CH area. Gross domestic product (GDP) in this area grew at a decreasing rate

between 1989 and 1992 while the level of tonnes transported between the countries was

maintained and even grew much more than GDP between 1991 and 1992. Average elasticity

between tonnes and GDP was about 1.5 in these years.

Between 1992 and 1994 growth in tonnes was above 3 per cent per year and growth in GDP

just above 1 per cent per year. Even though growth in tonnes might have been somewhat

overvalued in 1994 due to the change of statistical regime with the introduction of the Single

Market in 1993, growth in tonnes has to be down by 1.5 per cent if the implicit elasticity

should be as low as the one observed for 1989-1992. The most likely conclusion seems to be

that still overall growth in goods transport between countries in the EEA, CH area was higher

than growth in the overall GDP.

Between 1994 and 1997 transport volumes in trade between countries in the EEA, CH area

accelerated compared to 1992-1994. However, the growth rate of GDP also accelerated and

was on average 2 per cent per year between 1994 and 1997. The implied elasticity between

tonnes and GDP was 2.25 in the period.

- 61 -

Between 1994 and 1997 transport volum es in trade between countries in the EEA, CH area

accelerated compared to 1992-1994. However, the growth rate of GDP also accelerated and

was on average 2 per cent per year between 1994 and 1997. The implied elasticity between

tonnes and GDP was 2.25 in the period.

Trade between CEEC countries decreased for some years after 1989, but seems now to be

increasing. The same tendency has applied to exports from the CEEC countries to countries

in the EEA, CH area. Exports from the EEA, CH area to the CEEC countries may have grown

all the time, but the level of exports is less than the level of imports from the CEEC countries.

Task 9 Development of an OD method

The goal of the origin-destination (o-d) survey is to test whether it is possible to collect a

reliable o-d matrix of the total volume of exchanges between regions (by commodity group

and by mode of transport) on international relationships, using a short questionnaire.

The reason such an approach is tested is that it becomes more difficult to retrieve reliable o-d

data caused by the implementation of the internal free market within the European Union.

Information that used to be known is missing because the registration of some variables has

been shifted from the member-state border to the community border. In general no

international o-d information on regional level is available. According to the new directive on

road transport, interregional transport information will be collected, however not on the initial

origin and final destination of the good transported if different from the points of loading and

unloading, respectively. Therefore, alternative methods have to be found to fill these gaps in

information.

The shipper survey is an opportunity to test the approach. Appointments have been made

with shippers and they had to answer a (large) number of questions. Therefore, it requires a

relatively small effort to introduce a few extra questions. No specific shippers have been

selected for the o-d survey, all shippers exporting or importing to or from France that were

questioned in the large shipper survey were also questioned for the o-d survey.

The o-d survey is carried out amongst Dutch shippers in the shipper survey. This has been

done because in the Netherlands information is available for the total population of exporting

and importing Dutch firms in relation with France. This information is not available for French

firms. The availability of this information is crucial in order to raise the sample to a total for the

population on the o-d-relation.

The following information is gathered for the o-d relations:

Origin region (place of the establishment; from the shipper survey)

- 62 -

Destination region

Commodity group

Mode of transport

Total yearly weight (in tons)

Total yearly value

Activity of the Dutch establishment (from the shipper survey)

Number of employees from the Dutch establishment (from the shipper survey)

Task 10 Development of a behavioural model

Not only the MYSTIC project can show the technical possibility to apply the transport chain

concept developed by INFOSTAT, but it will also contribute to a better understanding of the

organisation of this transport chain inside international trade and particularly of the modal

choice. As such the behavioural analysis will contribute to identify and to measure those

explanatory variables that are required in the future European Transport policy Information

System (ETIS) as instruments of policy decisions.

The period January to June 1999 was used to hold the surveys in France and in the

Netherlands. A maximum of 300 exporting firms and 3 shipments by firms have been

surveyed in France and in The Netherlands, in a number of regions and activities. Destination

of the shipment might be all over in the world, depending upon the type of transport chain,

which is used.

The principle of the survey was to trace the logistic/transport organisation from the production

place to the destination, surveying also the transport legs that are used. The questionnaire

included 5 parts:

F1 pre-interview

F2 plant questionnaire

F3 shipment questionnaire

F4 operator questionnaire

F5 transport leg

- 63 -

F1 to F3 were face to face, whilst F4 and F5 were telephone interviews. Attention was paid to

the complexity of transport chains in international transport (compared to domestic flows).

Such surveys as just achieved by MYSTIC are quite innovative and complex. For this reason

in the previous period a test phase of the questionnaire and the clustering of the firms in

France and the Netherlands took place.

.

- 64 -

6. Conclusions and Recommendations

6.1 Overall Conclusions: Freight

The shippers survey successfully found a methodology for tracing freight shipments from

origin to destination through each link of the transport chain so that a matrix could be built for

each commodity flow. It could also be used for collecting behavioural data about freight. This

is a significant finding because such a methodology had hitherto elided researchers.

Suitably amended, the survey methodology could be scaled-up so as to elicit enough

information from which to build a freight origin-destination transport chain matrix for Europe.

The amendments include reducing the length and scope of the questionnaire so as to only

collect those items of information needed for the matrix. This would reduce the cost per

interview and improve the success rate. In view of the differences in response rate found in

France to that in the Netherlands the survey methodology would need to be adapted to suit

individual countries. It was found to be absolutely essential that the member government

should sponsor the survey. They should provide visible backing for the survey, as shippers

are reluctant to respond to non-governmental initiatives. If behavioural data were also

required then a sub sample could be given an additional questionnaire. Additional benefit

would be gained by undertaking the survey at the European scale for several reasons. For

example as a transport chain may have links in several member countries, shippers could be

interviewed by their own Nationals – increasing the success rate and reducing the cost per

interview.

For a pan European freight survey, some details would need to be worked out during the

piloting stages especially to adapt the questionnaire to local circumstances. The cost of the

complete survey is likely to be high and would need to be set up as a rolling programme with

Eurostat taking the lead - although due to the initial research nature of the survey it would be

best to instigate the survey as part of the European research programme. Based on the

French experience the fieldwork cost per country could be in the region of 0.5m euro for more

or less complete coverage and for 15 countries with initially one-third coverage this could cost

in the region of 2.5m euro. To this should be added the survey design and analysis costs that,

if combined with the recommended passenger research project (see below), could cost about

1m euro bringing the Freight costs to 3.5m euro for the freight component. The rolling

programme could cover the other two thirds over the rest of a ten-year period under the

auspices of Eurostat.

The data would be put into the o-d database and the organisational framework put in place to

provide information for government and maintain, market and offer users the service of

supplying extracts from the data, matrices and tools. The organisational framework needs to

- 65 -

be set up with great care so as to safeguard the data and maintain the high quality standard

of the data required by member governments. This could be initiated with a research project,

which could subsequently be taken over by Eurostat as a service for member governments,

transport operators and private clients.

The EDI survey showed that the information required (i.e. origin, destination and other data

items) did indeed exist in the computer systems of the originator of the consignment although

the data items required usually spanned more than one organisation and spanned more than

one computer system which made automating the process difficult to achieve in the short

term. The availability of the data would lend itself to adopting paper-based interview methods

to elicit o -d freight flow information.

When taken together the shippers survey and the EDI results show that paper based

interview methods with a short interview, referencing the company’s computer systems (if

they are exist), could be an effective basis for a methodology for estimating the pan European

freight commodity o-d matrix.

However the opportunity should be taken to introduce a EU wide standardised freight data

collection format now - because the timing is optimum. The industry (particularly that

associated with the logistics operator and with just-in-time manufacture) is undergoing rapid

change and in response software suppliers are redesigning their software. The opportunity

now arises, if an EU-wide standard is introduced, that EDI software suppliers would design-in

the new standard so that future releases would have the new standard embedded. If the

standard was so designed this could provide the platform needed for EDI to be successful.

It is possible to set out specific recommendations for a common Europe-wide electronic data

exchange standard that could then be taken up by the IT industry, along the lines of the

essential data inputs set out in Deliverable 5. These can be summarised as follows:

What? (describes the object being transported)

Commodities as defined, and by:

Weight

Volume

(consider floor space)

Type (see below)

Hazardous/Non-hazardous

Temperature Control (if applicable)

- 66 -

Value

Tariff (cost of transport)

Fragility

How? (identify all links and modes of the total chain of transport)

For each link:

Mode of Transport (Road, Rail, Sea, Air), and by:

Origin

Type of Packaging

Unit

Carrier (i.e. means of transport)

Shipping Document

Liability Convention

Terms of Sale

Start Time

Planned Finish Time

Actual Finish Time

Transhipment (if relevant), including:

Point of transfer

Environment

Method

Intermediate Destination (Repeat as necessary ...)

For Whom?

Shipper/Consignee

Name

- 67 -

Company structure

Domicile

Final Destination

By Whom? (identify the nature of the carrier(s) for each link defined under “How”)

Company

Name

Company structure

Domicile

Means of Communication

Appendix 1 of Deliverable 5 provides more detail.

6.2 Overall Conclusions: Passengers

The study was successful and our ideas about the methodology required to produce a pan

European passenger o-d trip matrix were honed-up into a workable methodology that is

outlined below. The case studies demonstrated that the overall approach outlined here can be

used to build a the trip matrix and that existing data can make a valuable contribution to

producing the pan European passenger o-d trip matrix. The study showed the importance of

establishing an organisational structure, which secures sensitive data and provides as free a

flow of information as possible within the constraints that data owners place upon the use of

their data by others. The database approach provides a framework within which this can be

organised optimally.

The continuity of keeping the pan European passenger o-d trip matrix as a working tool was

successfully addressed with the database approach whereby the data is held in a database

which can be added to as data becomes available and matrices can be built from it – matrices

which can be used at both a detailed scheme appraisal level and at a pan European level for

policy analysis.

The case studies highlighted additional problems and showed the way they could be

overcome. They showed the use and limitations of using existing data and highlighted the

need to collect a consistent set of new data with which to both harmonise existing data and to

add valuable additional information (e.g. about trip lengths, frequency, market segmentation,

spatial detail etc).

- 68 -

In summary, therefore, the passenger component of the MYSTIC project has provided a

comprehensive ‘toolbox’ consisting of a methodology for merging heterogeneous datasets

plus the software for actually carrying out the processes. The project has achieved its

purpose in developing a methodology for creating a pan-European passenger o-d matrix. It

has also successfully achieved its detailed objectives that were as follows;

To bring together heterogeneous data sets from a number of European countries and

merge them into a pan-European o-d matrix covering road and rail

To develop mathematical and statistical tools and techniques that would be used by the

case studies,

To develop a detailed UK based o-d matrix to show how datasets could be held in a

database for use in infrastructure scheme design and development

To develop a strategic pan-European o-d matrix using existing data to determine the

contribution existing data can make to the overall pan-European matrix

6.3 Proposed Methodology for building the Pan-European Passenger Matrix

The success of the UK case study in scaling-up the regional on-system o-d database and

matrix building methodology to a national scale gave confidence that it could also be applied

to help produce the pan-European passenger o-d matrix. The overall components of the

methodology for producing this trip matrix are set out in the following paragraphs

Existing o-d datasets: e.g. household o-d interviews, on-system o-d interviews (i.e. at

roadside, on-train and at-airport), already-constructed o-d matrices: The study showed that

these existing data sources could make a valuable contribution to the matrix. Other data is

needed to harmonise individual data sets into a common specification.

A new Pan European household travel diary survey: This is a survey of all long-distance

travel that a sample of households make in a predefined period. It should be sampled so as to

provide a complete representative cross section of travel. The data provides valuable

information that is consistent across all modes. It could be used to provide matrix cells, to

harmonise other data, to help with synthesising and to provide a consistent market

segmentation schema.

A new Pan European passenger roadside and on-system interview o-d survey: This

would be a major new survey undertaken in association with member governments, instigated

with an EC research project which would design and undertake the first phase of a rolling

programme of surveys which could be handed over to Eurostat to continue as a rolling

programme. The on-system interviews would comprise roadside, on-train and at-airport

interviews at locations where major international transport links cross a particular screen line

- 69 -

e.g. at a natural barrier. We recommend undertaking an initial 250 interview sites on important

road, rail and air links at major European screen lines. Subsequent phases in the rolling

programme would address other screen lines over a ten-year period. Typical sample fractions

would be 10% of all travellers crossing the screen line. The data from this would provide

robust estimates of all cells that traverse the screen line and as such builds -in to the matrix a

relatively high level of reliability.

Encourage National Governments to build national o-d databases: A Government

collects on-system o-d data for planning new transport infrastructure and services. This data

is often not held centrally and can be much more useful assembled into a National database

for matrix building. The MYSTIC toolbox can help member governments in this process and

can help in connecting National o-d databases together.

Assemble other o-d data: Transport operators and others collect o-d data for planning new

transportation and operating their services. This data can be used to build the matrix and it is

important to elicit the support of all data owners to contribute data and make use of the

matrix. Existing ticket sales data would also be included. This would help derive rail and air

o-d matrices. The air ticket sales data can be treated with the methodology developed for the

UK case study to derive origin-destination passenger trip matrices in conjunction with the

European household travel diary survey.

Merging and combining the above different types of o-d data: The methodological

advances made by MYSTIC can be used and built upon to improve the matrix. The

methodology itself can be improved with further research

The data would be put into an o-d database and the organisational framework put in place to

maintain, market and offer users the service of supplying extracts from the data, matrices and

tools. The organisational framework needs to be set up with great care so as to safeguard the

data. This could be initiated with a research project, which could subsequently be taken over

by Eurostat as a service for member governments, transport operators and private clients.

The service is unlikely to cost a lot to operate (provided the new data collection costs are

excluded) and if successful could be self financing.

It may be that certain member governments want to set up and operate their own o-d

database that should be welcomed. In these cases Eurostat would need to set up data

linkages with the organisational framework to support it so that data could be swapped with

ease.

- 70 -

6.4 The Way Forward

We recommend the Commission organise a research project to design, set up and initiate the

first phase of a rolling programme of passenger screen line o-d surveys and freight ‘shippers’

o-d surveys. The project would also set up the combined passengers and freight o-d

database together with the service to supply information to government (and others), set up

the database maintenance, marketing and support as well as providing a general service of

supplying matrices and tools to users. Operating the database could become a near self-

financing service, run on a five year rolling contract from the commission with the organisation

undertaking analysis for the Commission and acting as a focus for data supply to others.

The o-d database could comprise the new survey o-d data together with the pan European

household travel diary o-d survey data, rail and air ticket sales data, counts and other non o-d

data. The project could use the o-d database to build the pan European passenger and freight

o-d matrix as described above. To support this activity the project would need to research

methodologies for matrix merging, combining and improving and develop the tools which

could be made available to member governments and data contributing organisations. The

project could develop computer linkages with other databases proving exchange of data with

them and so that the data can be used to improve the matrix. The project should provide a

focus of expertise for o-d data and its uses and support the activities of researchers, analysts

and modellers.

Against a background of increasing competition and privatisation, it can take considerable

time to gain the support and confidence of some organisations. In order to provide optimal

chances for success it would be best if the timescale for the project were of sufficient length to

help achieve its objectives. The surveys are likely to take three years to plan, execute and

integrate with the other data to produce matrices. It could be preferable to extend this to four

years so as to gain the support of as many organisations as possible. The project could set

up the system in such a way as to effect a smooth handover to Eurostat after the four years.

The main part of the research project would need to cover the other components listed above:

collation of other data (existing o-d datasets, ticket sales data, household travel diary data,

non o-d data), methodological research (into matrix building and improving) as well as

building the database (passengers as well as freight) and matrix (including harmonisation,

building, combining, merging and improving). It would also need to cover the passenger

survey design and analysis. This is likely to cost in excess of 1.5m euro.

The new passenger screen line o-d survey is likely to cost 3m euro including fieldwork,

address coding and data processing (but excluding the survey design and analysis costs

which would be part of the main research project). The new freight ‘shippers’ o-d survey could

cost 3.5m euro including fieldwork, survey design and analysis costs. The main research

- 71 -

project is likely to cost 1.5m euro that with the surveys could give a total project cost of 8m

euro. The project should deliver a pan European passenger and freight o-d matrix using the

above methodology. It should instigate a rolling programme of passenger and freight o-d

surveys covering a ten-year period, which should be repeated so that the matrix is never

more than ten years out of date.

6.5 What Will This Provide?

The pan European passenger and freight o-d matrix would be invaluable in a whole raft of

ways – ways vital to the success of Europe plc, and European commerce and welfare as well

as to the European transport system. It could provide valuable base statistics so as to help

governments plan both their economies and their transport policy. It could help governments

improve the quality of their trade statistics and help plan the freight distribution systems so as

to provide the best balance between being both effective and environmentally friendly. It could

provide information about the demand for travel and allow inter country comparisons of

mobility, trip patterns, trip length, trips per citizen etc so that policies can be targeted at

locations where they are most needed. It could support modelling, which could open up the

whole field of forecasting and allow analysis of what-if scenarios, alternative -pricing

mechanisms etc to show their impacts on the transport networks.

The o-d matrices at member government level would fit together at country borders so that

countries themselves can plan their transport system knowing that the data fits together with

that of their neighbours. This is particularly important for those countries that have problems

with transit traffic.

Infrastructure providers could use the o-d matrices to help plan their schemes, services and

operations. This is as important for appraisal of the TEN’s as it is for the other parts of the

transport network.

European rail transport is key to maintaining mobility in Europe in the long term. However

train operators are faced with the need to capture new markets from road when at present

they have no idea how big this market is, where it wishes to travel to, its detailed market

segmentation and how to influence it. The o-d matrix together with its supporting database

could help understand these markets for them and play a key role in capturing road travel,

passengers and freight.

The o-d database could provide a consistent set of data, matrices and tools for European

policy analysis, member Government policy analysis, infrastructure appraisal, transport

planning and network operational optimisation. The consistency of the database will help

- 72 -

ensure that differences of view are not based on differences in the information used to

support them.

O-d data is very expensive to collect and is a valuable resource. In many cases this valuable

resource is simply thrown away. It should be assembled, stored in a database as a resource

for the future - and used. The recommended approach of assembling a pan European o-d

database could provide a fundamental and valuable resource for the future – for everyone

- 73 -

7 References

Beardwood, J.E. (1980) Sub-sample and Jack-knife: A general technique for the estimation

of sampling errors, with applications and examples in the field of transport planning.

Transportation Research, 24A, pp 211-215.

Cascetta, E. (1984) Estimation of trip matrices from traffic counts and survey data: a

generalised least squares estimator. Transportation Research, 18B, pp289-299.

Cascetta, E. and Nguyen, S. (1988) A unified framework for estimating or updating

origin/destination matrices from traffic counts. Transportation Research, 22B, pp437-455.

Clavering, J., Watson, S., Collop, M. and Hyman, G. (1997) Removing double counting

from RSI databases. Proc. 25th European Transport Forum, Seminar F. PTRC, London.

Department of Transport. (1981) Traffic Appraisal Manual. Department of Transport,

London.

Department of Transport (1982) Trip Matrix Building and Validation Program Suite, User

Documentation. HECB/R/27. Department of Transport, London.

DETR. (1997) Removing double counting from RSI databases, Final Report. Department of

the Environment, Transport and the Regions, London.

Gaudry, M. (1999) SPQR – The Four Approaches to Origin – Destination Matrix Estimation:

Some considerations for the MYSTIC Research Consortium. Montreal: University of Montreal

Gunn, H.F. and Whittaker, J.C. (1981) Estimation errors for well-fitting gravity models. ITS

Working Paper 149. Institute for Transport Studies, University of Leeds, Leeds.

Gunn, Mijer, Lindvelt and Hofman (1997) Estimating base matrices: the combined

calibration method. Proc. 25th European Transport Forum, Seminar E. PTRC, London.

Hajek (1977) Optimal sample size of roadside interview origin-destination surveys. Report

RR208. Ministry of Transport and Communications, Ontario.

Kirby, H.R. and Murchland, J.D. (1981) Gravity model fitting with both origin-destination

data and modelled trip-end estimates. Proc. 8th Int. Symp. on Transportation and Traffic

Theory, pp 348-375.

Kuwahara, M. and Sullivan, E.C. (1987) Estimating origin-destination matrices from

roadside survey data. Transportation Research, 21B, pp233-248.

- 74 -

Logie, M. and Hynd, A. (1990) MVESTM matrix estimation. Traffic Engineering and Control,

31, pp 454-459, 534-537 and 541-544.

Maher, M.J. (1983) Inferences of trip matrices from observations on link volumes: a Bayesian

statistical approach. Transportation Research, 17B, pp 435-447.

MVA Consultancy (1993) Deriving best estimates of trips from the RSI database. LATS

Report XV4.01. The MVA Consultancy, London

Spiess, H. (1987) A maximum likelihood for estimating origin-destination matrices.

Transportation Research, 21B, pp 395-412.

- 75 -

ANNEX 1: PUBLICATIONS, CONFERENCES AND

PRESENTATIONS

The following list covers the key events:

Publications:

Gaudry. M. SPQR – The Four Approaches to Origin – Destination Matrix Estimation: Some

Considerations for the MYSTIC Research Consortium. 23 pgs. September 1999, publication

CRT-99-35, Centre de Recherche sur les Transports, University of Montreal

Conferences

November 1999, European Transport Research Conference, Lille. The MYSTIC project was

demonstrated with the aid of an aid of a laptop slide presentation and brochures at a stand in

the main exhibition hall.

September 2000, Association of European Transport, Cambridge. Proposed presentation of

MYSTIC

Presentations

September 9th 1999. Presentation of the MYSTIC results and developments in methodology

to the UIC conference, Paris

September 20 th 1999. Presentation of the MYSTIC results and developments in methodology

to the EUROSIL multimodal workshop, Athens

September 25 th 1999. Presentation of the MYSTIC results and developments in methodology

to the MESUDEMO conference

September 28th 1999. Workshop at EUROSTAT, Luxembourg. This was a formal

presentation by each of the MYSTIC partners about the overall results and conclusions of the

project.

The results and conclusions from the EDI survey have been separately presented at the

following meetings:

February 2000, Automated Customs and International Trade association, Annual Meeting,

London

March 2000. Transport Statistics User Group, London

- 76 -

June 2000. European Commission Concerted Action Meeting on Intermodal Transport,

Brussels (at this meeting the results and conclusions of the shippers survey were also

presented)

- 77 -

ANNEX 2: THE DESIRE LINES

Figure 4.3: Mystic Case Study – Road – NUTS0 – no presentation threshold

- 78 -

Figure 4.4: Mystic Case Study – Road – NUTS2 – no threshold

- 79 -

Figure 4.5: Mystic Case Study – Road – NUTS2 – presentation threshold: >100000 Trips per

year

- 80 -

Figure 4.6: Mystic Case Study – Road – NUTS2 – presentation threshold: >200000 Trips per

year

- 81 -

Figure 4.7: Mystic Case Study combined with STREAMS – Road – NUTS0 – no presentation

threshold

- 82 -

Figure 4.8: Mystic Case Study combined with STREAMS – Road – NUTS2 – presentation

threshold: >200000 Trips per year

- 83 -

Figure 4.9: German data (trips > 100 km) – Road – NUTS0 – no presentation threshold

- 84 -

Figure 4.10: German data (trips > 100 km) – Road – NUTS2 – presentation threshold:

>200000 Trips per year