EX POST EVALUATION OF COHESION POLICY PROGRAMMES...

1

EX POST EVALUATION OF COHESION POLICY PROGRAMMES 2000-2006

WORK PACKAGE 10

‘Efficiency: Unit costs of major projects’

FINAL REPORT

25 October 2009

2

ACKNOWLEDGEMENTS This report has been written by RGL Forensics, Faber Maunsell/Aecom and Frontier Economics, following a call for tenders 2008.CE.16.0.AT.019. The authors are grateful for very helpful comments from the EC staff and particularly to Veronica Gaffey, Adam Abdulwahab, Kai Stryczynski and Francesco Maria Angelini (Evaluation Unit) and to the participants in the meetings of the Steering Committee. The team was also advised by a team of experts: Professor Bent Flyvbjerg, Mr Jacques Timmermans and Mr Nigel Grout. The authors are fully responsible for any remaining errors or omissions

DISCLAIMER

The European Commission and the authors accept no responsibility or liability whatsoever with regard to this text. This material is: - Information of general nature which is not intended to address the specific circumstances of any particular individual or entity. - Not necessarily comprehensive, accurate or up to date. - Not professional or legal advice. Reproduction or translation is permitted, provided that the source is duly acknowledged and no modifications to the text are made.

3

TABLE OF CONTENTS

TABLE OF FIGURES.................................................................................................................................5

TABLE OF TABLES ..................................................................................................................................7

0 EXECUTIVE SUMMARY...................................................................................................................8 0.1 Introduction..................................................................................................................................8 0.2 The high-level evaluation context before and after WP10 ......................................................8 0.3 Data gathering and limitations...................................................................................................9 0.4 Analysis of project delays ........................................................................................................10 0.5 The analysis of infrastructure unit costs................................................................................10 0.6 The analysis of productive investments.................................................................................13 0.7 The benchmarking database tool ............................................................................................14 0.8 Conclusions and recommendations .......................................................................................14

1 SECTOR SNAPSHOTS ..................................................................................................................36

2 INTRODUCTION .............................................................................................................................46 2.1 Background................................................................................................................................46 2.2 Purpose of this Report..............................................................................................................46 2.3 Structure of this Report ............................................................................................................47

3 IMPORTANT LESSONS FROM THE LITERATURE .....................................................................49 3.1 Introduction................................................................................................................................49 3.2 Unit cost definitions and measurement for infrastructure projects ....................................49 3.3 The determinants of infrastructure project costs ..................................................................55 3.4 Understanding time delays and cost overruns ......................................................................61 3.5 Defining and measuring ‘cost per job created’ ......................................................................66 3.6 Summary and conclusions.......................................................................................................70

4 INFORMATION GATHERING AND WP10 METHODOLOGY .......................................................71 4.1 Introduction................................................................................................................................71 4.2 The search for benchmark databases.....................................................................................71 4.3 Review of ex post evaluation activities at the Member State level ......................................74 4.4 Gathering data on the WP10 sample of major projects.........................................................77 4.5 Conclusions ...............................................................................................................................81

5 UNIT COST DEFINITION AND MEASUREMENT FOR WP10 INFRASTRUCTURE PROJECTS83 5.1 Introduction................................................................................................................................83 5.2 Categories of cost to be included............................................................................................84 5.3 Disaggregating the physical components of projects ..........................................................86 5.4 Defining unit cost indicators for Work Package 10 ...............................................................87 5.5 Information sought on the characteristics of projects..........................................................95 5.6 Measuring project completion times.......................................................................................96 5.7 Indentifying the causes of cost overruns and time delays...................................................97 5.8 Methods for achieving data comparability .............................................................................98

6 INFRASTRUCTURE: ESTIMATED AND ACTUAL UNIT COSTS AND COMPLETION TIMES.101 6.1 Introduction..............................................................................................................................101 6.3 Roads........................................................................................................................................110 6.4 Urban Transport ......................................................................................................................116 6.5 Water and Wastewater ............................................................................................................123 6.6 Energy ......................................................................................................................................129

4

6.7 Time delay and cost overrun analysis ..................................................................................134 6.8 The role of ex-ante risk assessment .....................................................................................139

7 PRODUCTIVE INVESTMENTS: COST PER JOB CREATED.....................................................140 7.1 Introduction..............................................................................................................................140 7.2 Methods used to estimate employment effects of productive investments .....................140 7.3 Total cost of productive investment projects in our sample..............................................142 7.4 Job creation analysis ..............................................................................................................150 7.5 Cost per job created................................................................................................................154 7.6 The role of delays in causing discrepancies........................................................................156 7.7 The type of funding that make up productive investments ................................................159 7.8 The amount of structural funds per job created ..................................................................161

8 DEVELOPMENT OF THE SPREADSHEET TOOLS....................................................................162 8.1 Introduction..............................................................................................................................162 8.2 Overall structure......................................................................................................................162 8.3 Input ..........................................................................................................................................163 8.4 Normalisation...........................................................................................................................164 8.5 Output.......................................................................................................................................164

9 SUMMARY AND CONCLUSIONS................................................................................................167 9.1 Review of the literature, previous evaluations and databases...........................................167 9.2 Definition and measurement of unit cost indicators for infrastructure.............................167 9.3 Methodology, data gathering and results.............................................................................168 9.4 Comparing estimated and actual unit infrastructure costs ................................................169 9.5 Analysis of cost overruns and time delays ..........................................................................170 9.6 Ex-ante risk assessment of major projects ..........................................................................173 9.7 ‘Cost per job created’ from productive investments ...........................................................174 9.8 Development of a spreadsheet tool.......................................................................................174

10 RECOMMENDATIONS.............................................................................................................176 10.1 Recommendations ..................................................................................................................176

ANNEX I – GRAPHICAL PRESENTATION OF OUR BENCHMARKING RESULTS..........................178

ANNEX II – DATA QUESTIONNAIRES ................................................................................................190

ANNEX III – TIPS FOR PROJECT MONITORING................................................................................205

5

TABLE OF FIGURES

Figure 1: Actual unit cost data for road projects.......................................................................................11 Figure 2: Indicative shares of 5 cost categories for infrastructure types.................................................51 Figure 3: DG XVI’s key determinants of infrastructure project costs .......................................................56 Figure 4: Key determinants of infrastructure project delays and cost overruns .......................................62 Figure 5: Common causes of cost overruns and project time delays ......................................................62 Figure 6: Reasoning for measuring cost per job created .........................................................................68 Figure 7: Cost categories provided in ERDF application forms ...............................................................84 Figure 8: Level 1 Rail: ‘All in’ estimated unit costs for all projects..........................................................102 Figure 9: Level 1 Rail: ‘All in’ actual unit costs for all projects................................................................103 Figure 10: Level 2 Rail: Estimated and actual unit ‘build’ cost of trackwork ..........................................104 Figure 11: Type 2 Rail: Estimated and actual unit ‘build’ cost of stations ..............................................105 Figure 12: Level 2 Rail: Estimated and actual unit ‘build’ cost of bridges ..............................................105 Figure 13: Level 2 Rail: Estimated and actual unit ‘build’ cost of tunnels ..............................................106 Figure 14: Type 1 Rail: Average cost overrun scores ............................................................................107 Figure 15: Type 1 Rail: Average time delay scores ...............................................................................108 Figure 16: Type 2: Average score of cost overrun and time delay categories.......................................109 Figure 17: Level 1 Road: ‘All in’ estimated unit costs for all projects .....................................................110 Figure 18: Level 1 Road: ‘All in’ actual unit costs for all projects ...........................................................111 Figure 19: Level 2 Road: Estimated and actual unit ‘build’ cost of pavement........................................112 Figure 20: Level 2 Road: Estimated and actual unit ‘build’ cost of bridges............................................112 Figure 21: Level 2 Road: Estimated and actual unit ‘build’ cost of tunnels............................................113 Figure 22: Type 1 Road: Average cost overrun scores..........................................................................113 Figure 23: Type 1 Road: Average time delay scores .............................................................................115 Figure 24: Type 2: Average score of cost overruns and time delays.....................................................115 Figure 25: Level 1 Urban Transport: ‘All in’ estimated unit costs for all projects and metro and tram benchmarks ............................................................................................................................................116 Figure 26: Level 1 Urban Transport: ‘All in’ actual unit costs for all projects and metro and tram benchmarks ............................................................................................................................................117 Figure 27: Level 2 Urban Transport: Estimated and actual unit ‘build’ cost of trackwork ......................118 Figure 28: Level 2 Urban Transport. Estimated and actual unit ‘build’ cost of stations .........................119 Figure 29: Level 2 Urban Transport. Estimated and actual unit ‘build’ cost of stations .........................119 Figure 30: Level 2 Urban Transport. Estimated and actual unit ‘build’ cost of bridges..........................120 Figure 31: Type 1 Urban Transport: Average cost overrun scores........................................................120 Figure 32: Type 1 Urban Transport: Average time delay scores ...........................................................122 Figure 33: Type 2 Urban Transport. Average score of each overrun and delay category, descending by overrun score..........................................................................................................................................122 Figure 34: Level 1 Water and Wastewater: ‘All in’ estimated unit costs for all projects.........................123 Figure 35: Level 1 Water and Wastewater: ‘All in’ actual unit costs for all projects ...............................124 Figure 36: Level 2 Water and Wastewater: Estimated and actual unit ‘build’ cost of land ....................125 Figure 37: Level 2 Water and Wastewater: Estimated and actual unit ‘build’ cost of water supply .......125 Figure 38: Type 1 Water and Wastewater: Average cost overrun scores .............................................126 Figure 39: Type 1 Water and Wastewater: Average time delay scores.................................................127 Figure 40: Type 2 Water and Wastewater. Average score of each overrun and delay category, descending by overrun score .................................................................................................................128 Figure 41: Level 1 Energy: ‘All in’ estimated unit costs for all projects excluding pipelines...................129 Figure 42: Level 1 Energy: ‘All in’ actual unit costs for all projects excluding pipelines.........................130 Figure 43: Level 2 Energy: Estimated and actual unit ‘build’ cost of turbines........................................131 Figure 44: Level 2 Energy: Estimated and actual unit cost of land ........................................................131 Figure 45: Type 1 Energy: Average cost overrun scores.......................................................................132 Figure 46: Type 1 Energy: Average time delay scores ..........................................................................133 Figure 47: Type 2 Energy: Average score of each overrun and delay category, descending by overrun score.......................................................................................................................................................134 Figure 48: Cost overrun per km against absolute delay per km.............................................................137 Figure 49: Total cost estimates and cost breakdowns for ‘new build’ projects ......................................143 Figure 50: Cost variance for new-build projects (actual minus forecast costs)......................................144 Figure 51 Cost variances for individual cost components of new-build projects....................................145 Figure 52 Actual cost divided by estimated cost of new build, shown as a percentage ........................146

6

Figure 53: Total cost estimates and breakdowns for expansion projects ..............................................147 Figure 54 Actual cost divided by estimated cost of expansion projects, shown as a percentage .........148 Figure 55 Cost variance and cost breakdowns for expansion projects..................................................149 Figure 56: UK Merseyside Special Investment Fund Expenditure.........................................................150 Figure 57: Estimates of full-time equivalent (FTE) jobs created by ‘new build’ projects........................151 Figure 58 Variance in number of jobs created by ‘new build’ projects...................................................152 Figure 59: Estimates of FTE jobs created for expansion projects..........................................................153 Figure 60 Actual minus forecast number of jobs created by expansion projects...................................154 Figure 61: Estimated versus actual cost per job created by new build projects ....................................155 Figure 62: Estimated versus actual cost per job created by expansion projects ...................................156 Figure 63: Actual and projected start dates of ‘new build’ projects........................................................157 Figure 64: ‘New build’ project durations (actual minus projected project durations)..............................157 Figure 65: Actual and projected start dates of expansion projects ........................................................158 Figure 66: Expansion project durations (actual minus projected start dates) ........................................159 Figure 67: Proportions of different types of funding used for productive investments ...........................160 Figure 68: ERDF funding per estimated job created..............................................................................161 Figure 69: Spreadsheet tool main menu ................................................................................................162 Figure 70: Input sheet for Rail sector projects........................................................................................164 Figure 71: Road sector single project output sheet................................................................................165 Figure 72: Urban Transport sector summary output sheet ....................................................................166 Figure 73: Actual unit cost data for road projects...................................................................................170 Figure 74: Actual unit cost data for road projects...................................................................................171 Figure 75: Level 1 Road Costs...............................................................................................................179 Figure 76: Two-lane carriageway level 1 construction costs..................................................................180 Figure 77: Road widening level 1 construction costs .............................................................................180 Figure 78: Level 2 construction costs for carriageways .........................................................................181 Figure 79 Level 2 construction costs for tunnels....................................................................................182 Figure 80 Level 2 construction costs for bridges....................................................................................182 Figure 81 Metro construction costs ........................................................................................................183 Figure 82: Light Rail and Tram construction costs.................................................................................184 Figure 83 Metro services cost of construction from Prof. Bent Flyvbjerg’s research.............................185 Figure 84 Level 1 Rail Costs ..................................................................................................................186 Figure 85 Level 2 Rail Costs – Tracks ...................................................................................................187 Figure 86 Wind farm construction costs .................................................................................................188 Figure 87 Water treatment facilities’ construction costs.........................................................................189 Figure 88 Water supply facilities’ construction costs..............................................................................189

7

TABLE OF TABLES

Table 1: Summary of the data gathering results ........................................................................................9 Table 2: Percentage delays by project phase and sector ........................................................................10 Table 3: Cost overrun summary by country and sector – average percentage differences between estimated and actual costs .......................................................................................................................12 Table 4: Time delay summary by country and sector - average percentage differences between estimated and actual completion times ....................................................................................................13 Table 5: How this report addresses the tasks (as per the Tender Specifications for WP10)...................47 Table 6: Cost per job created found by CSES (2006)..............................................................................69 Table 7: Summary of our data gathering results ......................................................................................79 Table 8: Structure of the intended WP10 sample of 155 major projects..................................................81 Table 9: Actual structure of the WP10 sample of 96 major projects ........................................................81 Table 10: Extract from questionnaire for road projects sent to Member States.......................................86 Table 11: Project disaggregation and corresponding Levels 1-3 unit cost indicators..............................87 Table 12: Unit cost indicators for rail projects ..........................................................................................88 Table 13: Unit cost indicators for road projects ........................................................................................89 Table 14: List of unit cost for Urban Transport projects ...........................................................................91 Table 15: List of unit cost for Water projects ............................................................................................93 Table 16: List of unit cost for Energy projects ..........................................................................................94 Table 17: General attribute information sought on infrastructure projects ...............................................95 Table 18: Sector-specific project attributes ..............................................................................................95 Table 19: Project completion times template ...........................................................................................96 Table 20: Type 1 and Type 2 cost overrun and time delay categories ....................................................97 Table 21: Price indices used for inflation adjustment...............................................................................98 Table 22: Exchange rates for non-Euro Member States..........................................................................99 Table 23: Rail: Estimated completion times, actual completion times, absolute delay and percentage delay .......................................................................................................................................................108 Table 24: Road: Estimated completion times, actual completion times, absolute delay, percentage delay................................................................................................................................................................114 Table 25: Urban transport estimated completion times, actual completion times, absolute delay, percentage delay ....................................................................................................................................121 Table 26: Water and Wastewater: Estimated completion times, actual completion times, absolute delay, percentage delay ....................................................................................................................................127 Table 27: Energy: Estimated completion times, actual completion times, absolute delay, percentage delay .......................................................................................................................................................133 Table 28: Level 1 Cost overruns: Average score (0-3) by sector...........................................................135 Table 29: Percentage delays by project phase and sector ....................................................................135 Table 30: Level 1 time delays: Average score (0-3) by sector ...............................................................136 Table 31: Cost overrun summary by country and sector – average percentage differences between estimated and actual costs .....................................................................................................................138 Table 32: Time delay summary by country and sector - average percentage differences between estimated and actual completion times ..................................................................................................138 Table 33: Employment effects projected for the Doncaster-Sheffield Airport productive investment ....141 Table 34: Employment effects projected for an Italian productive investment.......................................142 Table 35: Summary of the data gathering results ..................................................................................169 Table 36: Cost overrun summary by country and sector – average percentage differences between estimated and actual costs .....................................................................................................................172 Table 37: Time delay summary by country and sector - average percentage differences between estimated and actual completion times ..................................................................................................173

8

0 EXECUTIVE SUMMARY

0.1 Introduction

0.1.1 This is the Final Report on Work Package 10 (WP10) of the European Commission, Directorate-General for Regional Policy’s ex post evaluation of Cohesion Policy programmes 2000-2006 financed by the European Regional Development Fund (ERDF) in Objective 1 and 2 regions. The title of Work Package 10 is ‘Efficiency – Unit costs of major projects’.

0.1.2 The main objectives of the WP10 Study were (1) to undertake an ex-post evaluation of the performance (cost overruns and time delays) of major infrastructure projects and productive investments co-financed by ERDF based on a representative sample; (2) to develop a database of unit cost benchmarks and project characteristics to assist the Commission in the appraisal of future project financing requests; and (3) to develop an Excel-based tool that includes the project data collected through the WP10 exercise (with the potential to be populated with data from other infrastructure projects funded by the EC) to facilitate project appraisal and evaluation in the future.

0.1.3 What follows is a short summary of the findings and the key messages and recommendations that have emerged from our WP10 Study, including our assessment of the sample of 155 ERDF co-financed major projects.1

0.2 The high-level evaluation context before and after WP10

0.2.1 Our review of existing relevant academic literature and professional studies revealed an extensive body of research into the causes of infrastructure cost overruns and project delays and of their tendency to differ in magnitude and scale across countries and sectors. Specifically, ‘optimism bias’ in project forecasting, and other difficulties in the estimation of project costs, are widely acknowledged. Likewise, the lack of good quality project data is widespread, which makes it difficult to use benchmark data as a basis for forecasting.

0.2.2 For these reasons, robust cost estimation, project appraisal and evaluation has and will remain a critical area of responsibility for public agencies that are recipients of EU funds. In that respect, the availability of reliable and consistent benchmark data on project unit costs could provide a useful tool to improve the appraisal and evaluation process. However, bar a couple of exceptions, there are no relevant up-to-date databases of infrastructure costs.2 Some high level total cost benchmark data was available but there were problems with the definition of unit costs and with data comparability in general. (The benchmarks we did find are presented in Annex I of this report.)

0.2.3 One of DG Regio’s objectives for WP10 was to set up an infrastructure project cost database, which might provide reliable and robust cost benchmark data for the appraisal and evaluation of major projects in the future. We hope, therefore, that the database we have constructed and the analysis we have conducted are useful first steps in establishing such a database and, perhaps, a useful foundation for future work in this area.

0.2.4 DG Regio might, however, also wish to consider co-operation with European financial institutions, for example, the European Investment Bank (EIB) and the European Bank

1 Please note that the work undertaken to produce this and earlier WP10 reports have benefitted from the regular advice of three external experts. Those experts were Prof. Bent Flyvbjerg, Mr Jacques Timmermanns and Mr Nigel Grout. Any errors are those of the project team and not of our external experts. 2 The most notable exception is the World Bank ROCKS database for large highway projects.

9

for Reconstruction and Development (EBRD) to improve the scope and quality of the database and, in turn, the quality of project appraisal and evaluation for all parties, including DG Regio.

0.2.5 Likewise, the development of a ‘generally accepted’ set of (sector-specific) unit cost definitions for major infrastructure projects could contribute greatly to the availability of comparable benchmark data. This could be done by DG Regio in conjunction with these other European project finance institutions. Moreover, the general adoption by government ministries, planning agencies and project implementation bodies across EU Member States could, in turn, lead to improvements in project planning, appraisal, monitoring and evaluation (at the Member State as well as at the EU level).

0.2.6 We developed, for the purposes of WP10, a three-tier (Level 1 to 3) approach to the definition of unit cost indicators to reflect various degrees of cost disaggregation. This was to acknowledge that unit cost benchmarks are useful only if they accurately reflect the average cost of sufficiently disaggregated project components. If DG Regio pursues the development of a common set of unit cost indicators, they could be defined according to the three-tier approach developed in this Report.

0.3 Data gathering and limitations

0.3.1 We were required, as previously noted, to incorporate in our Study, an assessment of 155 major projects (115 infrastructure and 40 productive investments). The information required for these assessments was, however, more difficult to acquire from the Member States than had been initially expected. This, however, was due, in large part, to the fact that the information required to put values to our unit cost indicators is not information that the Member States are required to submit to the European Commission in order to receive funds or to facilitate ex post evaluation.3

0.3.2 The results of our data gathering efforts are summarised in Table 1 below, and set out in greater detail in section 4 of this report.

Table 1: Summary of the data gathering results

Numbers of projects Total Infrastructure Productive investments

Target sample size 155 115 40

Projects on which data sought 173 128 45

Unfinished projects 26 16 10

Data returns received 96 66 30

0.3.3 The direct implications of the relatively poor response rate are (i) a reduced sample

size and (ii) an infrastructure sample that is biased in favour of road and rail projects. An indirect implication is the limiting effect on the scope and depth of the statistical analysis that was possible for the WP10 exercise. However, we believe that the statistical analysis has been more severely affected by the scope and quality of the information that was made available than by any sector bias.

3 In this respect, WP10 represents a first step of this new aspect of monitoring of major projects.

10

0.3.4 Under the 2000-2006 Objective 1 and Objective 2 ERDF programmes, 271 major projects were adopted for the 11 Member States that were the subject of the WP10 Study. The total estimates of expenditure involved in these projects were circa €33bn, of which approximately €15bn was to be the contribution of the ERDF. Therefore, the sample of 96 major projects analysed in this report represents about 35 per cent of the total of 271 projects financed with EU structural funds in the period of study.

0.4 Analysis of project delays

0.4.1 Table 2 below sets out the average delay in project implementation by project phase and sector (the percentage represents the delays expressed as a proportion of the time estimated for each phase of the project).

Table 2: Percentage delays by project phase and sector

Project phase Rail

(%)

Road

(%)

Urban

(%)

Water

(%)

Energy

(%)

Planning 36.4 19.2 37.7 37.0 14.2

Funding 115.4 0.0 60.1 73.9 71.8

Permissions 31.8 3.0 7.7 33.1 22.5

Site preparation 47.3 27.7 18.4 153.2 21.2

Construction 51.6 22.0 13.4 37.9 11.0

0.4.2 Table 2 indicates significant delays can occur at each phase of project development, with delays in funding contributing to significant delays in most sectors other than road projects for which the site preparation and construction phases were the most likely to cause delay.

0.5 The analysis of infrastructure unit costs

0.5.1 Our analysis of infrastructure unit costs was constrained by the limited sample sizes and level of detail available for each project, as summarised above and discussed in Sections 5 and 6 of this Report. Notwithstanding these limitations, we expect our database to provide benchmark cost data that will be useful in the appraisal of future ERDF-project financing requests.

0.5.2 The database also contains information on individual project characteristics which, for example, identifies the urban or rural location of the project, the geographic terrain, and the project complexity. We did not, however, succeed in gathering sufficient data on project characteristics to enable us to carry out a statistical analysis of the possible relationship between certain project characteristics and costs (for example, to determine with any rigour the expected impact of the urban/rural split of a project on overall cost).

0.5.3 The level of data we have collected has enabled us to produce useful Level 1 (and some Level 2) unit cost benchmarks. For example, Figure 1 below shows our benchmark unit cost data for road projects, which may be useful for future project appraisal, at least at the initial appraisal stage.

11

Figure 1: Actual unit cost data for road projects

0

10

20

30

40

50

60

70

80

DE02 - P

rojec

t A

DE02 - P

rojec

t B

DE03 - P

rojec

t A

ES04 - P

rojec

t A

GR03 - E

gnatia

GR05 - K

ifisso

s

GR03 - P

rojec

t A

GR04 - A

ttiki

GR05 - P

rojec

t A

GR05 - C

hrisoupoli

GR04 - P

rojec

t A

GR05 - S

iatist

a

IE02 - H

urlers

IE02 - W

aterg

rassh

ill

IE02 - A

shford

IE05 - B

allys

hannon

IE02 - B

allinco

llig

IT06 - O

riental

e

IT03 - C

arlo Feli

ce

PL05 - P

ulawy

PL05 - E

lbag

Uni

t cos

t (€m

/km

)LandTaxesContingencySoftBuild

0.5.4 If, as recommended below, the database is adequately maintained and updated over time, it will provide an increasingly useful information resource on major projects. It will also extend the boundaries of the statistical analysis that can be undertaken to enable, for example, the analysis of the relationships between different project characteristics and cost overruns and/or overall time delays. It might also enable the use of the database in a ‘reference class forecasting’ approach to benchmarking and cost estimation.4

0.5.5 Our analysis of cost overruns and time delays was influenced by the same data constraints that limited our analysis of unit costs. However, we did collate information where it was available and this has allowed us to undertake a descriptive analysis. Table 3 and Table 4 provide an illustrative summary of this analysis. Specifically, Table 3 shows a summary of average percentage cost overruns for all observations in our sample, broken down by sector and by country. A positive value indicates a cost overrun, while a negative value indicates a cost saving.

4 Reference class forecasting is a method that predicts the outcome of a project based on actual outcomes in a reference class of similar projects to that being forecast. It has been applied in the context of major infrastructure projects by Prof. Bent Flyvbjerg; see, for example, Flyvbjerg (2007), ‘Eliminating bias through reference class forecasting and good governance’.

12

Table 3: Cost overrun summary by country and sector – average percentage differences between estimated and actual costs

(%)/ (number of projects) Rail Road Urban

Transport Water Energy Weighted average

by sector

Germany -4.3% (6) -10.0% (3) -6.2%

Spain 12.8% (6) 30.7% (1) 17.4% (2) 15.8%

France 32.9% (1) 32.9%

Great Britain 110.7% (1) 110.7%

Greece 74.3% (2) 19.7% (8) 20.1% (2) 0.0% (1) 26.6%

Ireland 2.1% (5) 74.1% (1) 14.1%

Italy 62.4% (5) -5.0% (2) -0.9% (1) 37.6%

Poland 19.7% (2) 80.9% (2) 50.3%

Portugal 9.0% (1) 3.3% (4) 4.4%

Weighted average by Member State

26.9% 9.4% 45.4% 11.3% 20.7% 21.2%

0.5.6 Table 4 provides an analogous summary for percentage delays (calculated as the ratio of actual completion time to estimated completion time). Again, positive values correspond to actual delays, while negative values indicate that the actual average completion time was shorter than expected.

13

Table 4: Time delay summary by country and sector - average percentage differences between estimated and actual completion times

(%)/ (number of projects)

Rail Road Urban Transport Water Energy

Weighted average

by sector

Germany 40.2% (6) 4.7% (3) 28.4%

Spain 15.3% (6) 27.3% (1) 55.9% (2) 25.7%

France 4.9% (1) 4.9%

Great Britain 0.0% (1) 0.0%

Greece 24.4% (2) 17.8% (7) 13.2% (2) 12.6% (1) 17.7%

Ireland 9.0% (5) 52.2% (1) 16.2%

Italy 88.4% (1) 88.4%

Poland 5.9% (1) 2.7% (2) 3.8%

Portugal 258.3% (1) 41.5% (4) 84.9%


25.8% 13.2% 49.6% 66.7% 29.8% 26.2%

0.5.7 The values presented in tables 3 and 4 are averages and, as such, they mask the high variance of results which we obtained for each project. However, we note that most of the projects in our database were not completed on time and without any cost overruns.

0.5.8 These initial results appear to confirm the usefulness of this type of analysis. If the database is maintained and updated with more data on project performance, it has the potential to be useful in the risk assessment which is required as part of major project appraisal.

0.6 The analysis of productive investments

0.6.1 WP10 is also concerned with the ‘result’ efficiency of productive investment projects (i.e., projects that involve direct support to enterprises), where ‘result’ efficiency is measured according to the ‘cost per job created’ as a result of these investments.

0.6.2 Apart from the difficulties in accurately measuring Structural Fund employment effects (specifically job creation and maintenance), the different types of productive activity, as reflected in the different capital / labour ratios of different industries, casts doubt over the appropriateness of ‘job creation’ and, hence, over the appropriateness of ‘cost per job created’, as both absolute and relative indicators of the performance of productive investments.

0.6.3 Decisions on the granting of funds for productive investments are, moreover, based on a wide cost-benefit analysis, rather than on the basis of the cost-effectiveness of job creation only (unlike direct employment measures in general). While comparisons between estimates of job creation and ‘cost per job created’ with the outturn values for these indicators might be useful on a project-by-project basis, comparisons between

14

the values of these indicators across projects are unlikely to be meaningful without appropriate adjustments for different industry circumstances.

0.6.4 In our view, it would be more appropriate to evaluate these projects using a wider cost benefit approach and then compare comparative efficiency using benefit-cost ratios.

0.7 The benchmarking database tool

0.7.1 As part of this Report, we designed and built spreadsheet-based database tools of project unit costs to provide a repository for the information we have collected during our analysis, thereby facilitating easy access to data relating to specific projects. It will also be useful as a repository for new project data as these become available. At the start of our study, in line with the Terms of Reference, we set out to develop database tools for both infrastructure projects and productive investments. However, our analysis has shown that the benchmarking of productive investment on the basis of the ‘cost per job created’ is unlikely to be meaningful without appropriate adjustments for different industry circumstances. For this reason, only the infrastructure projects database tool was eventually developed for general use.

0.7.2 The spreadsheet tool is structured as a set of input and output sheets. The input sheets allow the Commission services to enter raw project data. The data is then put through a set of normalisation sheets, which convert the raw data into comparable unit costs, adjusting for exchange rates and inflation. The output sheets will allow users (such as project appraisers) to search for projects of a similar class and to consult summary sheets for an overview of all projects in a particular sector.

0.8 Conclusions and recommendations

0.8.1 Despite the difficulties encountered in our data gathering exercise and insufficient information to draw statistically robust conclusions, we have endeavoured to provide a comprehensive overview of the available data for ERDF co-financed large infrastructure projects and productive investments, including:

• estimated and actual total and unit costs per ERDF infrastructure project;

• available unit cost benchmarks from existing databases and previous evaluations; and

• Jobs created per ERDF productive investment, total cost per job created and the amount of ERDF funding per job created.

0.8.2 The database of infrastructure unit cost benchmarks and project characteristics and the Excel-based tool can facilitate project appraisals undertaken by DG Regio in the future. However, a lack of data availability has been a limiting factor in our attempt to (i) undertake the ex-post evaluation of the performance (cost overruns and time delays) in the delivery of major infrastructure projects in the sample, the other main objective of WP10 and (ii) to provide robust unit cost benchmarks.

0.8.3 The benchmarking and data gathering exercise undertaken in WP10 should, however, be viewed as a first step in the development of a comprehensive benchmarking tool for future evaluations of projects and investments financed with European Commission funds (ERDF, Cohesion, ISPA etc), and be able to apply both established and new methodologies (e.g., in the application of the reference class forecasting approach).

0.8.4 Going forward, and in order to maintain and grow the database, it will be necessary to collect new project data and information in a consistent and regular manner, which has

15

not been the case in the past. It will also be necessary to establish channels for the regular provision of quality and consistent data on major projects by the Member States and to put in place robust systems for data verification.

0.8.5 The desire for improved major infrastructure project appraisal, design and evaluation is not new. In particular, in some European countries such as Denmark, Netherlands, Norway and the UK, public sector agencies, industry and academia have, over recent years, become increasingly aware of the need to develop new methods to improve cost estimation. This reflects the fact that major projects have historically had a poor record in terms of cost overruns and delays in implementation.

0.8.6 The work done in this WP10 study appears to follow these more recent developments in some European countries, which may offer opportunities for collaboration between DG Regio and the EU Member States in this area in the future.

0.8.7 Our recommendations are set out below.

Development of an EU-wide database of major infrastructure project costs

0.8.8 The Excel spreadsheet tool developed as part of the WP10 exercise provides a useful starting point in the development of such a database for EU-funded projects. The database can be built up over time to provide a valuable on-going source of benchmark cost data.

0.8.9 We recommend, therefore, that the Commission considers how best to ensure that the database is maintained on an ongoing basis and that the quality and quantity of project data improves over time. Three further recommendations may contribute to the usefulness of the database, namely:

• the adoption of common sets of unit cost definitions at national and EU level;

• the reform, at the Member State level, of major project monitoring and reporting and of the depth of project appraisal carried out when requesting EU funds; and

• The participation of European financial institutions, such as EIB and EBRD that are also involved in major project co-financing.

0.8.10 The collaborative approach suggested above can have a number of potential benefits, including the increased number of data points and the potential for sharing the costs of maintaining the database. It might also facilitate the multilateral development and consequent greater use of common sets of cost definitions. These could, as noted already, be defined according to the three-tier Level 1 to 3 approach that we have developed for WP10 (see section 5.4 of this Report).

Standardize and improve the ex ante risk assessments in funding applications

0.8.11 We also recommend that the application of the methods for the ex ante assessment of project risks that are set out in the Commission’s most recent guide to cost-benefit analysis would represent a significant improvement in the quality of such assessments (which are legally required for all major projects). 5 , 6 Our recommendation is based on a review of several of the relatively few project dossiers on which detailed reports were available in support of the ERDF application forms (which generally provided only the conclusions of such reviews).

5 European Commission, Directorate General Regional Policy, Guide to Cost-Benefit Analysis of investment projects Structural Funds, Cohesion Fund and Instrument for Pre-Accession Final Report 16/06/2008 6 A useful manual from the US Government Accountability Office is ‘Cost estimating and assessment guide: Best practices for developing and managing capital program costs’, GAO Applied Research and Methods (2009).

16

Standardize and improve project monitoring and reporting for major projects

0.8.12 Based on the responses we received to our enquiries and questionnaires, it appears that relatively few project completion or progress reports on major ERDF-funded projects are produced by or on behalf of the Member States. Furthermore, the usefulness of the few reports we reviewed for project evaluation purposes was limited.

0.8.13 We recommend that the Commission takes the necessary steps to improve the project reporting process for major projects with the objectives of improving the project monitoring process and the availability of project data for planning, project appraisal and evaluation.

0.8.14 In particular, the following improvements should be considered:

• The use of standard definitions for project costs to be required in project application forms, appraisal and monitoring.

• The preparation of regular project progress reports to be a condition for funding.

• The submission of regular project progress reports (to a specified format) to be a condition for continuing drawdown of funds.

• The submission of a project completion report (to a specified format) to be a condition of funding.

17

0 RÉSUMÉ

0.1 Introduction

0.1.1 Le présent document est le Rapport final sur le Volet 10 (WP10) de l’Évaluation ex post des programmes de la politique de cohésion 2000-2006 financés par le Fonds européen de développement régional (FEDER) dans les régions éligibles aux objectifs 1 et 2 de la Commission européenne, Direction générale de la politique régionale. L’intitulé du Volet 10 est « Rentabilité : coûts unitaires de grands projets ».

0.1.2 Les principaux objectifs de l’Étude du Volet 10 étaient (1) de procéder à une évaluation ex post de la performance (dépassements de coûts et retards) des grands projets infrastructurels et des investissements productifs cofinancés par le FEDER à partir d’un échantillon représentatif ; (2) de préparer une base de données de coûts unitaires de référence et de caractéristiques de projet afin d’aider la Commission à apprécier les futures demandes de financement de projet ; et (3) de mettre au point un outil basé sur Excel reprenant les données de projet recueillies pendant l’étude du Volet 10 (pouvant être renseigné avec les données d’autres projets infrastructurels financés par la Commission européenne) afin de faciliter l’appréciation et l’évaluation des projets futurs.

0.1.3 Suit un bref résumé des conclusions ainsi que les principaux messages et recommandations issus de notre Étude du Volet 10, dont notamment notre évaluation de l’échantillon de 155 grands projets financés par le FEDER.7

0.2 Le contexte de l’évaluation de haut niveau avant et après le Volet 10

0.2.1 Notre analyse de la littérature universitaire et des études professionnelles pertinentes existantes a révélé un important corpus de recherches sur les causes des dépassements de coûts et retards et sur leur tendance à varier d’un pays et d’un secteur à l’autre en termes de grandeur et d’échelle. En particulier, un « biais optimiste » dans les prévisions de projet, et d’autres difficultés concernant l’estimation des coûts de projet, sont largement reconnus. De la même manière, le manque de données de projet de bonne qualité est généralisé, ce qui rend difficile l’utilisation de données de référence comme fondement des prévisions.

0.2.2 Pour ces raisons, procéder à une estimation des coûts, une appréciation de projet et une évaluation consistantes est et restera un domaine de responsabilité critique pour les agences publiques qui reçoivent des financements de l’Union européenne. À cet égard, la disponibilité de données de référence fiables et cohérentes sur les coûts unitaires de projet pourrait constituer un outil utile en vue d’améliorer le processus d’appréciation et d’évaluation. Toutefois, à quelques exceptions près, il n’existe pas de base de données des coûts infrastructurels à jour et pertinente.8 Quelques données de référence de haut niveau sur le coût total étaient disponibles, mais il y avait des problèmes concernant la définition des coûts unitaires et la comparabilité des données en général. (Les références que nous avons trouvées sont présentées dans l’Annexe I de ce rapport.)

0.2.3 L’un des objectifs de la DG Régio concernant le Volet 10 était de créer une base de données des coûts de projets infrastructurels, qui pourrait fournir des données de

7 Noter que les travaux entrepris pour produire ce rapport ainsi que les précédents rapports du Volet 10 ont bénéficié des conseils réguliers de trois experts externes. Ces experts étaient le professeur Bent Flyvbjerg, M. Jacques Timmermanns et M. Nigel Grout. Les éventuelles erreurs ont été commises par l’équipe de projet et non par nos experts externes. 8 L’exception la plus remarquable concerne la base de données ROCKS de la Banque Mondiale pour les grands projets autoroutiers.

18

référence fiables ou consistantes sur les coûts en vue de l’appréciation et l’évaluation des grands projets futurs. Nous espérons donc que la base de données que nous avons construite et l’analyse que nous avons réalisée constituent des premières étapes utiles dans l’établissement de cette base de données et, peut-être, une base utile pour les travaux futurs dans ce domaine.

0.2.4 Néanmoins, la DG Régio pourrait également envisager une coopération avec les institutions financières européennes, comme la Banque européenne d’investissement (BEI) et la Banque européenne pour la reconstruction et le développement (BERD), afin d’améliorer l’étendue et la qualité de la base de données et, à son tour, la qualité de l’appréciation et de l’évaluation des projets pour toutes les parties, y compris la DG Régio.

0.2.5 De la même manière, l’élaboration d’une série de définitions (sectorielles) « généralement admises » des coûts unitaires pour les grands projets infrastructurels pourrait beaucoup contribuer à la disponibilité de données de référence comparables. La DG Régio pourrait s’en charger en conjonction avec ces autres institutions européennes de financement de projet. De plus, l’adoption générale par les ministères nationaux, les agences de planification et les organismes d’exécution de projet des États membres de l’Union européenne pourrait, à son tour, conduire à des améliorations de la planification, de l’appréciation, du suivi et de l’évaluation des projets (aussi bien au niveau des États membres que de l’UE).

0.2.6 Pour les besoins du Volet 10, nous avons mis au point une approche à trois niveaux (Niveau 1 à 3) pour la définition des indicateurs de coûts unitaires pour refléter divers degrés de ventilation des coûts. Le but était ici de reconnaître que les coûts unitaires de référence ne sont utiles que s’ils reflètent avec précision le coût moyen de composantes de projet suffisamment ventilées. Si la DG Régio poursuit l’élaboration d’un ensemble commun d’indicateurs de coûts unitaires, ceux-ci pourraient être définis conformément à l’approche à trois niveaux mise au point dans ce Rapport.

0.3 Collecte de données et limites

0.3.1 Comme indiqué précédemment, on nous avait demandé d’incorporer dans notre Étude une évaluation de 155 grands projets (115 projets infrastructurels et 40 investissements productifs). Les informations nécessaires pour ces évaluations étaient, cependant, plus difficiles à acquérir auprès des États membres que nous l’avions anticipé à l’origine. Toutefois, ceci était dû en grande partie au fait que les informations nécessaires pour donner des valeurs à nos indicateurs de coûts unitaires ne sont pas des informations que les États membres sont tenus de soumettre à la Commission européenne pour recevoir des fonds ou pour faciliter l’évaluation ex post.9

0.3.2 Les résultats de nos travaux de collecte de données sont présentés de manière sommaire dans le Table 1 ci-dessous, et sont repris de manière plus détaillée à la section 4 de ce Rapport.

9 À cet égard, le Volet 10 représente une première étape de ce nouvel aspect du suivi des grands projets.

19

Tableau 1: Résumé des résultats de la collecte de données

Nombre de projets Total Projets infrastructurels

Investissements productifs

Taille de l’échantillon cible 155 115 40

Projets sur lesquels des données ont été demandées 173 128 45

Projets non terminés 26 16 10

Retours de données reçus 96 66 30

0.3.3 Les implications directes de ce taux de réponse relativement mauvais sont (i) une taille

d’échantillon réduite et (ii) un échantillon de projets infrastructurels dans lequel les projets routiers et ferroviaires sont surreprésentés. Cette situation a également une implication indirecte : l’effet limiteur sur l’étendue et la profondeur de l’analyse statistique possible pour le Volet 10. Nous pensons cependant que l’analyse statistique a été davantage compromise par l’étendue et la qualité des informations mises à disposition que par le biais sectoriel.

0.3.4 Dans le cadre des programmes 2000-2006 du FEDER dans les régions éligibles aux objectifs 1 et 2, 271 grands projets ont été adoptés pour les 11 États membres qui faisaient l’objet de l’Étude du Volet 10. Les estimations de dépenses impliquées dans ces projets tournaient autour de 33 milliards d’euros au total, dont 15 milliards d’euros environ devait être la contribution du FEDER. L’échantillon de 96 grands projets analysé dans ce rapport représente donc environ 35 % du total de 271 projets financés par des fonds structurels européens pendant la période de l’étude.

0.4 Analyse des retards prévus

0.4.1 Le Table 2 ci-dessous présente le retard moyen de mise en œuvre des projets par phase de projet et par secteur (le pourcentage représente les retards exprimés en proportion du temps estimé pour chaque phase du projet).

Tableau 2 : Retards en pourcentage par phase de projet et par secteur

Phase de projet Rail

(%)

Route

(%)

Transports urbains

(%)

Eau

(%)

Énergie

(%)

Planification 36,4 19,2 37,7 37,0 14,2

Financement 115,4 0,0 60,1 73,9 71,8

Autorisations 31,8 3,0 7,7 33,1 22,5

Préparation du site 47,3 27,7 18,4 153,2 21,2

Construction 51,6 22,0 13,4 37,9 11,0

0.4.2 Le Table 2 indique que des retards significatifs peuvent intervenir à chaque phase du développement de projet, les retards de financement provoquant des retards

20

significatifs dans la plupart des secteurs autres que les projets routiers pour lesquels les phases de préparation du site et de construction étaient les plus susceptibles d’entraîner des retards.

0.5 L’analyse des coûts unitaires des projets infrastructurels

0.5.1 Notre analyse des coûts unitaires des projets infrastructurels était limitée par la taille des échantillons et par le faible niveau de détail disponible pour chaque projet, comme il est indiqué brièvement ci-dessus et expliqué dans les Sections 5 and 6 de ce Rapport. Nonobstant ces limites, nous espérons que notre base de données fournira des données de référence sur les coûts qui seront utiles pour l’appréciation des futures demandes de financement de projet par le FEDER.

0.5.2 La base de données contient également des informations sur les caractéristiques des différents projets qui, par exemple, identifient la situation urbaine ou rurale du projet, le terrain géographique et la complexité du projet. Nous n’avons pas réussi, cependant, à recueillir suffisamment de données sur les caractéristiques des projets pour nous permettre d’exécuter une analyse statistique du rapport possible entre certaines caractéristiques et certains coûts de projet (par exemple, pour déterminer avec une certaine rigueur l’impact attendu de la ventilation urbaine/rurale d’un projet sur le coût global).

0.5.3 Le niveau de données que nous avons recueilli nous a permis de produire des coûts unitaires de référence utiles de Niveau 1 (et quelques-uns de Niveau 2). Par exemple, la Figure 1 ci-dessous montre nos données sur les coûts unitaires de référence pour les projets routiers, qui peuvent être utiles pour l’appréciation de futurs projets – du moins lors de la phase d’appréciation initiale.

21

Figure 1 : Données des coûts unitaires réels pour les projets routiers

0.5.4 Si, comme il est recommandé ci-dessous, la base de données est maintenue et actualisée comme il convient au fil du temps, elle constituera une source d’informations de plus en plus utiles sur les grands projets. Elle repoussera également les limites de l’analyse statistique pouvant être réalisée pour permettre, par exemple, l’analyse des rapports entre les différentes caractéristiques de projet et les dépassements de coûts et/ou retards globaux. Elle pourrait également être utilisée dans une méthode d’analyse comparative et d’estimation des coûts de type « reference class forecasting ».10

0.5.5 Notre analyse des dépassements de coûts et des retards a été influencée par les contraintes liées aux données, qui ont limité notre analyse des coûts unitaires. Nous avons cependant regroupé les informations lorsqu’elles étaient disponibles, ce qui nous a permis de réaliser une analyse descriptive. Le Table 3 et le Table 4 présentent une synthèse illustrative de cette analyse. En particulier, le Table 3 montre une synthèse des dépassements de coûts moyens en pourcentage pour toutes les observations de notre échantillon, ventilés par secteur et par pays. Une valeur positive indique un dépassement de coût, tandis qu’une valeur négative indique une économie.

10 La méthode de prévision de type « reference class forecasting » prédit le résultat d’un projet sur la base des résultats réels d’une catégorie de référence de projets comparables au projet pour lequel les prévisions sont établies. Elle a été appliquée dans le contexte des grands projets infrastructurels par le professeur Bent Flyvbjerg ; voir, par exemple, Flyvbjerg (2007), « Eliminating bias through reference class forecasting and good governance ».

0

10

20

30

40

50

60

70

80

DE02 - Pro

ject A

DE02 - Pro

ject B

DE03 - Pro

ject A

ES04 - Pro

ject A

GR03 - Egnati

a

GR05 - Kifis

sos

GR03 - Pro

ject A

GR04 - Attik

i

GR05 - Pro

ject A

GR05 - Chris

oupoli

GR04 - Pro

ject A

GR05 - Siat

ista

IE02 - Hurle

rs

IE02 - Wate

rgras

shill

IE02 - Ash

ford

IE05 - Ball

yshan

non

IE02 - Ball

incollig

IT06 - Orie

ntale

IT03 - Carl

o Felice

PL05 - Pulaw

y

PL05 - Elbag

Coû

t uni

taire

(€m

/km

)

TerrainImpôtsImprévusCoûts accessoiresConstruction

0

10

20

30

40

50

60

70

80

DE02 - Pro

ject A

DE02 - Pro

ject B

DE03 - Pro

ject A

ES04 - Pro

ject A

GR03 - Egnati

a

GR05 - Kifis

sos

GR03 - Pro

ject A

GR04 - Attik

i

GR05 - Pro

ject A

GR05 - Chris

oupoli

GR04 - Pro

ject A

GR05 - Siat

ista

IE02 - Hurle

rs

IE02 - Wate

rgras

shill

IE02 - Ash

ford

IE05 - Ball

yshan

non

IE02 - Ball

incollig

IT06 - Orie

ntale

IT03 - Carl

o Felice

PL05 - Pulaw

y

PL05 - Elbag

Coû

t uni

taire

(€m

/km

)

TerrainImpôtsImprévusCoûts accessoiresConstruction

22

Tableau 3: Synthèse des dépassements de coûts par pays et par secteur – différences moyennes en pourcentage entre les coûts estimés et les coûts réels

(%)/ (nombre de projets) Rail Route Transports

urbains Eau Énergie Moyenne pondérée

par secteur

Allemagne -4,3 % (6) -10,0 % (3) -6,2 %

Espagne 12,8 % (6) 30,7 % (1) 17,4 % (2) 15,8 %

France 32,9 % (1) 32,9 %

Grande-Bretagne 110,7 %

(1) 110,7 %

Grèce 74,3 % (2) 19,7 % (8) 20,1 % (2) 0,0 % (1) 26,6 %

Irlande 2,1 % (5) 74,1 % (1) 14,1 %

Italie 62,4 % (5) -5,0 % (2) -0,9 % (1) 37,6 %

Pologne 19,7 % (2) 80,9 % (2) 50,3 %

Portugal 9,0 % (1) 3,3 % (4) 4,4 %

Moyenne pondérée par État membre

26,9 % 9,4 % 45,4 % 11,3 % 20,7 % 21,2 %

0.5.6 Le Table 4 présente une synthèse analogue pour les retards en pourcentage (calculés comme le rapport du délai de réalisation réel sur le délai de réalisation estimé). Là aussi, les valeurs positives correspondent à des retards réels, tandis que les valeurs négatives indiquent que le délai de réalisation moyen a été plus court que prévu.

23

Tableau 4: Synthèse des retards par pays et par secteur – écarts moyens en pourcentage entre les délais de réalisation estimés et réels

(%)/ (nombre

de projets) Rail Route Transports

urbains Eau Énergie Moyenne pondérée

par secteur

Allemagne 40,2 % (6) 4,7 % (3) 28,4 %

Espagne 15,3 % (6) 27,3 % (1) 55,9 % (2) 25,7 %

France 4,9 % (1) 4,9 %

Grande-Bretagne 0,0 % (1) 0,0 %

Grèce 24,4 % (2) 17,8 % (7) 13,2 % (2) 12,6 % (1) 17,7 %

Irlande 9,0 % (5) 52,2 % (1) 16,2 %

Italie 88,4 % (1) 88,4 %

Pologne 5,9 % (1) 2,7 % (2) 3,8 %

Portugal 258,3 % (1) 41,5 % (4) 84,9 %

Moyenne pondérée par État-membre

25,8 % 13,2 % 49,6 % 66,7 % 29,8 % 26,2 %

0.5.7 Les valeurs présentées dans les Tableaux 3 et 4 sont des moyennes et, de ce fait, elles masquent la grande disparité des résultats que nous avons obtenus pour chaque projet. Nous notons cependant que la plupart des projets de notre base de données n’ont pas été réalisés dans les délais et sans dépassements de coûts.

0.5.8 Ces premiers résultats paraissent confirmer l’utilité de ce type d’analyse. Si la base de données est maintenue et actualisée avec d’autres données sur la performance des projets, elle pourrait être utile pour l’évaluation des risques exigée dans le cadre de l’appréciation des grands projets.

0.6 L’analyse des investissements productifs

0.6.1 Le Volet 10 s’intéresse également à l’efficacité « de résultat » des projets d’investissement productif (c’est-à-dire les projets qui comportent un soutien direct aux entreprises), sachant que l’efficacité « de résultat » est mesurée en fonction du « coût par emploi créé » en conséquence de ces investissements.

0.6.2 En dehors des difficultés rencontrées pour mesurer de manière exacte les effets sur l’emploi des Fonds structurels (en particulier la création et le maintien des emplois), les différents types d’activité productive, tels que reflétés dans les différents ratios capital / main-d’œuvre des différents secteurs d’activité, remettent en cause le bien-fondé de la « création d’emploi » et donc le bien-fondé du « coût par emploi créé » comme indicateurs absolus et relatifs de la performance des investissements productifs.

0.6.3 Les décisions relatives à l’octroi de fonds en vue d’investissements productifs reposent, de plus, sur une analyse coûts/avantages plus générale, et non sur la seule efficacité-coût de la création d’emploi (contrairement aux mesures directes en faveur

24

de l’emploi en général). Bien que les comparaisons entre les estimations de création d’emploi et le « coût par emploi créé » d’une part et les valeurs des résultats réels de ces indicateurs d’autre part puissent être utile au cas par cas pour chaque projet, il est peu probable que les comparaisons entre les valeurs de ces indicateurs pour plusieurs projets présentent un intérêt en l’absence d’ajustements appropriés pour tenir compte des situations différentes des divers secteurs d’activité.

0.6.4 À notre avis, il serait plus intéressant d’évaluer ces projets en utilisant une méthode coût/avantage plus générale, puis de comparer leur efficacité relative au moyen de ratios coûts/avantages.

0.7 Les outils de base de données de référence

0.7.1 Dans le cadre de ce Rapport, nous avons conçu et construit des outils de base de données de coûts unitaires de projet basés sur un tableur pour stocker la somme des informations que nous avons recueillies au cours de notre analyse, facilitant ainsi l’accès aux données relatives à des projets particuliers. Ces outils seront également utiles en tant que lieu de stockage des nouvelles données de projets à mesure que celles-ci deviennent disponibles. Au début de notre étude, conformément aux Termes de référence, nous nous sommes mis à développer des outils de base de données pour les projets infrastructurels et pour les investissements productifs. Cependant, notre analyse a montré que l’analyse comparative des investissements productifs sur le fondement du « coût par emploi créé » serait sans doute sans intérêt en l’absence d’ajustements appropriés pour tenir compte des situations différentes des divers secteurs d’activité. Pour cette raison, seule la base de données des projets infrastructurels a été finalement mise au point pour un usage général.

0.7.2 Le tableur est structuré comme une série de feuilles de saisie et de résultat. Les feuilles de saisie permettent aux services de la Commission d’enregistrer les données brutes sur les projets. Les données passent ensuite par plusieurs feuilles de normalisation, qui convertissent les données brutes en coûts unitaires comparables, en tenant compte des taux de change et de l’inflation. Les feuilles de résultat permettront aux utilisateurs (comme les appréciateurs de projet) de rechercher des projets de catégorie similaire et de consulter les feuilles de résumé pour obtenir une vue d’ensemble de tous les projets d’un secteur particulier.

0.8 Conclusions et recommandations

0.8.1 Malgré les difficultés rencontrées dans nos travaux de collecte des données et malgré les informations insuffisantes pour tirer des conclusions statistiquement robustes, nous nous sommes efforcés de produire une vue d’ensemble complète des données disponibles pour les grands projets infrastructurels et les investissements productifs cofinancés par le FEDER, dont notamment :

• les coûts totaux et unitaires estimés et réels par projet infrastructurel du FEDER ;

• les coûts unitaires de référence disponibles à partir des bases de données existantes et des évaluations précédentes ; et

• les emplois créés par investissement productif du FEDER, le coût total par emploi créé et le montant du financement FEDER par emploi créé.

0.8.2 La base de données des coûts unitaires de références des projets infrastructurels et des caractéristiques de projet ainsi que l’outil basé sur Excel peuvent faciliter les appréciations de projets entreprises à l’avenir par la DG Régio. Cependant, le manque

25

de disponibilité des données a limité nos tentatives (i) pour procéder à l’évaluation ex post de la performance (dépassements de coûts et retards) concernant l’exécution des grands projets infrastructurels de l’échantillon, l’autre objectif principal du Volet 10, et (ii) de fournir des coûts unitaires de référence robustes.

0.8.3 Toutefois, les travaux d’analyse comparative et de collecte de données entrepris dans le Volet 10 doivent être considérés comme une première étape dans le développement d’un outil d’analyse comparative complet pour les évaluations futures de projets et d’investissements financés par les fonds de la Commission européenne (FEDER, Fonds de Cohésion, ISPA, etc.) et doivent pouvoir s’appliquer à la fois aux méthodologies établies et aux nouvelles méthodologies (par exemple, dans l’application de la méthode « reference class forecasting »).

0.8.4 À l’avenir, et pour maintenir et développer la base de données, il sera nécessaire de recueillir de nouvelles données et informations sur les projets de manière cohérente et régulière, ce qui n’a pas été le cas dans le passé. Il sera également nécessaire de mettre en place des modes de transmission permettant la communication régulière par les États membres de données de qualité et cohérentes sur les grands projets et de mettre en place des systèmes robustes de vérification des données.

0.8.5 Le désir d’améliorer l’appréciation, la conception et l’évaluation des grands projets infrastructurels n’est pas nouveau. En particulier, dans certains pays européens comme le Danemark, les Pays-Bas, la Norvège et le Royaume-Uni, les agences publiques, l’industrie et les milieux universitaires ont pris de plus en plus conscience de la nécessité de mettre au point de nouvelles méthodes permettant d’améliorer l’estimation des coûts. Cette évolution reflète le fait qu’historiquement les grands projets ont de mauvais antécédents en termes de dépassements de coûts et de retard de mise en œuvre.

0.8.6 Les travaux effectués dans cette Étude du Volet 10 paraissent suivre ces récents développements observés dans certains pays européens, ce qui peut offrir des opportunités de future collaboration entre la DG Régio et les États membres de l’UE dans ce domaine.

0.8.7 Nos recommandations sont présentées ci-dessous.

Mise au point d’une base de données communautaire des coûts des grands projets infrastructurels

0.8.8 L’outil sur tableur Excel développé dans le cadre du Volet 10 constitue un point de départ utile pour la mise au point d’une telle base de données pour les projets financés par l’UE. La base de données peut être construite au fil du temps pour devenir une précieuse source permanente de données sur les coûts de référence.

0.8.9 Nous recommandons donc que la Commission réfléchisse au meilleur moyen de veiller à ce que la base de données soit maintenue en permanence et à ce que la qualité et la quantité des données de projet s’améliorent au fil du temps. Trois autres recommandations pourraient contribuer à l’utilité de la base de données, à savoir :

• l’adoption de séries communes de définitions des coûts unitaires au niveau national et communautaire ;

• la réforme, au niveau des États membres, du suivi et de l’information concernant les grands projets ainsi que de la profondeur de l’appréciation de projet exécutée au moment de la demande de financement communautaire ; et

• la participation des institutions financières européennes, comme la BEI et la BERD, qui participent également au cofinancement des grands projets.

26

0.8.10 L’approche collaborative suggérée ci-dessus peut présenter plusieurs avantages potentiels, dont notamment le plus grand nombre de points de données et la possibilité de partager les coûts de maintenance de la base de données. Elle pourrait également faciliter l’élaboration multilatérale et donc l’usage plus généralisé de séries communes de définitions des coûts. Ceux-ci pourraient, comme nous l’avons déjà indiqué, être définis conformément à la méthode à trois niveaux (Niveau 1 à 3) que nous avons mise au point pour le Volet 10 (voir la section 5.4 de ce Rapport).

Normaliser et améliorer les évaluations des risques ex ante dans les demandes de financement

0.8.11 Nous recommandons également l’application des méthodes d’évaluation ex ante des risques de projet qui sont présentées dans le tout dernier guide de l’analyse avantage-coût de la Commission, qui représentent une amélioration considérable de la qualité de ces évaluations (qui constituent une exigence légale pour tous les grands projets). 11,12 Notre recommandation repose sur l’étude de plusieurs des dossiers de projet relativement peu nombreux sur lesquels des rapports détaillées avaient été fournis à l’appui des demandes de financement adressées au FEDER (qui ne contenaient généralement que les conclusions de ces analyses).

Normaliser et améliorer le suivi et l’information concernant les grands projets

0.8.12 À partir des réponses que nous avons reçues à nos demandes de renseignements et questionnaires, il semble que les États membres, ou leurs représentants, produisent relativement peu de rapports d’achèvement de projet ou d’avancement pour les grands projets financés par le FEDER. De plus, les quelques rapports que nous avons étudiés présentaient une utilité limitée à des fins d’évaluation de projet.

0.8.13 Nous recommandons que la Commission prennent les mesures nécessaires pour améliorer le processus d’information pour les grands projets dans l’objectif d’améliorer le processus de suivi des projets et la disponibilité des données de projet pour la planification, l’appréciation et l’évaluation des projets.

0.8.14 En particulier, les améliorations suivantes doivent être considérées :

• L’utilisation de définitions standard pour les coûts de projet comme exigence des demandes de financement, des appréciations et du suivi.

• La préparation de rapports d’avancement de projet réguliers comme condition du financement.

• La soumission de rapports d’avancement de projet réguliers (selon un format spécifié) comme condition de la poursuite des tirages de fonds.

• La soumission d’un rapport d’achèvement de projet (selon un format spécifié) comme condition du financement.

11 Commission européenne, Direction générale de la Politique régionale, Guide de l’analyse coûts/avantages des projets d’investissement. Fonds structurels, Fonds de cohésion et Instrument d’aide de préadhésion. Rapport final. 16/06/1008. 12 Manuel utile du Government Accountability Office des États-Unis : « Cost estimating and assessment guide: Best practices for developing and managing capital program costs », GAO Applied Research and Methods (2009).

27

0 ZUSAMMENFASSUNG

0.1 Einleitung

0.1.1 Dies ist der Abschlussbericht zum Arbeitspaket 10 (WP10) der Europäischen Kommission, Generaldirektion Regionalpolitik „Ex-post-Bewertung der Programme der Kohäsionspolitik 2000-2006, die vom Europäischen Fonds für regionale Entwicklung (EFRE) in Ziel-1- und Ziel-2-Regionen finanziert wurden“. Der Titel des Arbeitspakets 10 lautet „Wirtschaftlichkeit: Einzelkosten von Großprojekten“.

0.1.2 Die Hauptziele der Studie zum Arbeitspaket 10 waren: (1) eine auf Basis einer repräsentativen Stichprobe durchzuführende Ex-post–Bewertung der Leistungseffizienz (Kosten- und Zeitüberschreitungen) vom EFRE kofinanzierter großer Infrastrukturprojekte und produktiver Investitionen; (2) Aufbau einer Datenbank mit Einzelkostenvergleichsdaten und Projektmerkmalen, welche der Kommission die Beurteilung künftiger Projektfinanzierungsanträge erleichtert; und (3) der Aufbau eines auf Excel basierenden Tools, das die im Rahmen des Arbeitspakets 10 gesammelten Projektdaten enthält (und die Möglichkeit bietet, sie mit Daten anderer von der EG finanzierter Infrastrukturprojekte zu füllen), um künftig die Beurteilung und Bewertung von Projekten zu erleichtern.

0.1.3 Es folgt eine kurze Zusammenfassung der durch die WP10-Studie gewonnenen Ergebnisse sowie der Hauptbotschaften und -empfehlungen, einschließlich einer Bewertung der aus 155 vom EFRE kofinanzierten Großprojekten bestehenden Stichprobe.13

0.2 Übergeordnete Bewertungen vor und nach WP10

0.2.1 Unsere Prüfung der einschlägigen wissenschaftlichen Literatur und professioneller Studien ergab, dass es diesbezüglich bereits viel Forschung gibt, nicht nur über die Ursachen von Kosten- und Zeitüberschreitungen bei Infrastrukturprojekten, sondern auch darüber, dass diese dazu neigen, hinsichtlich Größe und Ausmaß in verschiedenen Ländern und Wirtschaftszweigen zu differieren. Insbesondere, dass bei Projektprognosen „optimistische Fehlschlüsse“ und andere Schwierigkeiten bei der Schätzung der Projektkosten auftreten, ist weithin anerkannt. Genauso fehlt es allgemein an Projektdaten guter Qualität, so dass es schwierig ist, Vergleichsdaten als Prognosegrundlage heranzuziehen.

0.2.2 Aus diesen Gründen ist die gründliche Kostenschätzung, Projektbeurteilung und Projektbewertung, wie bisher schon, auch künftig eine wichtige Verantwortung der öffentlichen Stellen, die die EU-Mittel erhalten. Diesbezüglich könnte die Verfügbarkeit zuverlässiger und konsistenter Vergleichsdaten für Projekteinzelkosten ein nützliches Instrument für das Beurteilungs- und Bewertungsverfahren darstellen. Abgesehen von einigen wenigen Ausnahmen gibt es jedoch keine einschlägigen aktuellen Datenbanken über Infrastrukturkosten.14 Zwar gab es einige übergeordnete Vergleichsdaten zu Gesamtkosten, aber es gab Probleme hinsichtlich der Definition der Einzelkosten und der allgemeinen Vergleichbarkeit der Daten. (Die Vergleichsdaten, die wir gefunden haben, sind in Anhang I zu diesem Bericht aufgeführt.)

13 Dabei ist zu beachten, dass die zur Erstellung dieses WP10-Berichts wie auch früherer WP10-Berichte durchgeführte Arbeit von der regelmäßigen Beratung durch drei externe Experten profitierte. Diese Experten waren Prof. Bent Flyvbjerg, Herr Jacques Timmermanns und Herr Nigel Grout. Jegliche Fehler sind dem Projektteam zuzuschreiben und nicht unseren externen Experten. 14 Die bedeutendste Ausnahme ist die ROCKS-Datenbank der Weltbank für Großprojekte im Straßenbau.

28

0.2.3 Eines der von der Generaldirektion Regionalpolitik für WP10 gesetzten Ziele war die Einrichtung einer Datenbank für Infrastrukturprojektkosten, die zuverlässige und harte Vergleichskostendaten zur Beurteilung und Bewertung künftiger Großprojekte liefern könnten. Wir hoffen daher, dass die von uns erstellte Datenbank wie auch die von uns durchgeführte Analyse erste nützliche Schritte auf dem Weg zum Aufbau einer solchen Datenbank darstellen und vielleicht als nützliche Grundlage für künftige Arbeiten in diesem Bereich dienen können.

0.2.4 Die Generaldirektion Regionalpolitik möchte jedoch vielleicht auch eine Zusammenarbeit mit anderen EU-Finanzinstitutionen wie der Europäischen Investitionsbank und der Europäischen Bank für Wiederaufbau und Entwicklung in Betracht ziehen, um Umfang und Qualität der Datenbank und damit wiederum die Qualität der von allen Parteien, einschließlich der Generaldirektion Regionalpolitik, vorgenommenen Projektbeurteilungen und Projektbewertungen zu verbessern.

0.2.5 In gleicher Weise könnte die Entwicklung „allgemein anerkannter“ (sektorspezifischer) Einzelkostendefinitionen für Infrastrukturgroßprojekte einen wichtigen Beitrag zur Verfügbarkeit von Vergleichsdaten leisten. Dies könnte durch die Generaldirektion Regionalpolitik in Verbindung mit diesen anderen projektfinanzierenden EU-Institutionen geschehen. Darüber hinaus könnte deren allgemeine Übernahme durch die Ministerien, Planungsbehörden und Projektumsetzungsstellen aller EU-Mitgliedstaaten wiederum zu Verbesserungen in den Bereichen der Projektplanung, Projektbeurteilung Projektüberwachung und Projektbewertung führen (sowohl in den einzelnen Mitgliedstaaten als auch auf EU-Ebene).

0.2.6 Im Rahmen von WP10 haben wir einen dreistufigen Ansatz (Stufe 1 bis 3) für die Definition der Einzelkostenindikatoren entwickelt, der die verschiedenen Grade der Kostenaufschlüsselung berücksichtigt. Damit wurde berücksichtigt, dass Einzelkostenvergleichsdaten nur insoweit nützlich sind, als sie die Durchschnittskosten der verschiedenen, mit ausreichender Genauigkeit aufzuschlüsselnden Projektkomponenten zutreffend wiedergeben. Sollte die Generaldirektion Regionalpolitik einen gemeinsamen Satz von Einzelkostenindikatoren verfolgen wollen, so könnte man diesen gemäß dem in diesem Bericht entwickelten dreistufigen Ansatz definieren.

0.3 Datenerhebung und Einschränkungen

0.3.1 Wie bereits zuvor angemerkt wurde, waren wir gehalten, im Rahmen unserer Studie eine Bewertung von 155 Großprojekten (115 Infrastrukturprojekte und 40 produktive Investitionen) vorzunehmen. Es erwies sich jedoch als schwieriger als anfangs vermutet, die für diese Bewertungen erforderlichen Informationen von den Mitgliedstaaten zu bekommen. Dies lag allerdings größtenteils daran, dass es sich bei den Informationen, die wir brauchten, um unseren Einzelkostenindikatoren Werte zuzuweisen, nicht um Informationen handelt, die die Mitgliedstaaten bei der Europäischen Kommission einreichen müssen, um Fördermittel zu erhalten oder die Ex-post-Bewertung zu ermöglichen.15

0.3.2 Die Ergebnisse unserer Datenerhebungsbemühungen sind nachstehend in Table 1 zusammengefasst. Eine detailliertere Darstellung ist in Abschnitt 4 dieses Berichts zu finden.

15 In dieser Hinsicht stellt WP10 einen ersten Schritt bezüglich dieses neuen Aspekts der Überwachung von Großprojekten dar.

29

Tabelle 1: Zusammenfassung der Datenerhebungsergebnisse

Projektzahl Gesamt Infrastruktur Produktive Investitionen

Anvisierte Stichprobengröße 155 115 40

Projekte, für die Daten angefragt wurden 173 128 45

Nicht abgeschlossene Projekte 26 16 10

Eingegangene Datenmeldungen 96 66 30

0.3.3 Die relativ schlechte Antwortquote hatte zwei direkte Auswirkungen: (i) eine kleinere

Stichprobengröße und (ii) eine Stichprobe zu Infrastrukturprojekten, die einen Schwerpunkt bei Straßen- und Schienenprojekten aufwies. Eine indirekte Auswirkung ist die sich daraus ergebende Einschränkung hinsichtlich des Umfangs und der Gründlichkeit der im Rahmen von WP10 möglichen statistischen Analyse. Wir denken jedoch, dass die statistische Analyse stärker durch Umfang und Qualität der zur Verfügung gestellten Informationen beeinträchtigt wurde als durch Sektorschwerpunkte.

0.3.4 Im Rahmen der Ziel-1- und Ziel-2-Programme, die der EFRE im Zeitraum 2000-2006 unterstützte, wurden den elf in der WP10-Studie untersuchten Mitgliedstaaten 271 Großprojekte bewilligt. Die Schätzung der für diese Projekte angefallenen Gesamtausgaben beläuft sich auf circa 33 Mrd. €, von denen etwa 15 Mrd. € vom EFRE beigesteuert wurden. Somit stellt die in diesem Bericht analysierte Stichprobe von 96 Großprojekten etwa 35 Prozent der insgesamt 271 im Untersuchungszeitraum durch EU-Strukturfonds finanzierten Projekte dar.

0.4 Analyse der Zeitüberschreitungen

0.4.1 Table 2 unten stellt die durchschnittliche Zeitüberschreitung bei der Projektdurchführung nach Projektphase und Sektor dar (der Prozentsatz entspricht dem Verhältnis der Zeitüberschreitung zu der für jede Projektphase geschätzten Zeitdauer).

Tabelle 2: Prozentuale Zeitüberschreitung nach Projektphase und Sektor

Projektphase Schiene

(%)

Straße

(%)

Stadt (%)

Wasser

(%)

Energie

(%)

Planung 36,4 19,2 37,7 37,0 14,2

Finanzierung 115,4 0,0 60,1 73,9 71,8

Genehmigungen 31,8 3,0 7,7 33,1 22,5

Baustellenvorbereitung 47,3 27,7 18,4 153,2 21,2

Bau 51,6 22,0 13,4 37,9 11,0

0.4.2 Table 2 zeigt, dass es in jeder Phase der Projektentwicklung zu erheblichen Verzögerungen kommen kann, wobei Verzögerungen bei der Finanzierung in den

30

meisten Sektoren erhebliche Zeitüberschreitungen verursachen. Anders ist es bei Straßenbauprojekten, wo Verzögerungen bei Baustellenvorbereitung und Bau am wahrscheinlichsten sind.

0.5 Analyse der Einzelkosten für Infrastrukturprojekte

0.5.1 Eine Analyse der Einzelkosten für Infrastrukturprojekte war uns wegen der begrenzten Stichprobengrößen und der geringen Detailangaben zu den einzelnen Projekten nur eingeschränkt möglich (zu den Gründen für diese Einschränkungen vergleiche die Abschnitte 3 und 4 dieses Berichts). Ungeachtet dieser Einschränkungen gehen wir davon aus, dass unsere Datenbank Vergleichskostendaten liefern wird, die für die Beurteilung künftiger Anträge auf Fördermittel für EFRE-Projekte nützlich sein werden.

0.5.2 Die Datenbank enthält außerdem Informationen über einzelne Projektmerkmale, zum Beispiel Angaben dazu, ob sich das Projekt im städtischen oder ländlichen Raum befindet, zum geografischen Terrain und zur Projektkomplexität. Allerdings ist es uns nicht gelungen, genügend Daten über Projektmerkmale zu sammeln, um eine statistische Analyse der möglichen Beziehung zwischen bestimmten Projektmerkmalen und den Projektkosten durchzuführen (um zum Beispiel angemessen aussagekräftige Feststellungen über die erwartete Auswirkung der Projektaufteilung auf städtische und ländliche Gebiete auf die Gesamtkosten treffen zu können).

0.5.3 Mit der von uns gesammelten Datenmenge war es uns möglich, nützliche Einzelkostenvergleichsdaten für die 1. Stufe (sowie in einigen Fällen für die 2. Stufe) zu erzeugen. So zeigt zum Beispiel die nachstehende Abbildung 1 unsere Vergleichskostendaten für Straßenprojekte, die – zumindest in der Erstbeurteilungsphase – für künftige Beurteilungen nützlich sein dürfte.

Abbildung 1: Tatsächliche Einzelkostendaten für Straßenprojekte

0.5.4 Wird die Datenbank, wie von uns nachstehend empfohlen, angemessen gepflegt und laufend aktualisiert, so wird sie eine zunehmend nützlichere Informationsquelle für

0

10

20

30

40

50

60

70

80

DE02 - Pro

ject A

DE02 - Pro

ject B

DE03 - Pro

ject A

ES04 - Pro

ject A

GR03 - Egnati

a

GR05 - Kifis

sos

GR03 - Pro

ject A

GR04 - Attik

i

GR05 - Pro

ject A

GR05 - Chris

oupoli

GR04 - Pro

ject A

GR05 - Siat

ista

IE02 - Hurle

rs

IE02 - Wate

rgras

shill

IE02 - Ash

ford

IE05 - Ball

yshan

non

IE02 - Ball

incollig

IT06 - Orie

ntale

IT03 - Carl

o Felice

PL05 - Pulaw

y

PL05 - Elbag

Einz

elko

sten

(€m

/km

)

LandSteuernEventualkostenWeiche KostenBaukosten

0

10

20

30

40

50

60

70

80

DE02 - Pro

ject A

DE02 - Pro

ject B

DE03 - Pro

ject A

ES04 - Pro

ject A

GR03 - Egnati

a

GR05 - Kifis

sos

GR03 - Pro

ject A

GR04 - Attik

i

GR05 - Pro

ject A

GR05 - Chris

oupoli

GR04 - Pro

ject A

GR05 - Siat

ista

IE02 - Hurle

rs

IE02 - Wate

rgras

shill

IE02 - Ash

ford

IE05 - Ball

yshan

non

IE02 - Ball

incollig

IT06 - Orie

ntale

IT03 - Carl

o Felice

PL05 - Pulaw

y

PL05 - Elbag

Einz

elko

sten

(€m

/km

)

LandSteuernEventualkostenWeiche KostenBaukosten

31

Großprojekte darstellen. Sie wird auch neue Möglichkeiten für die statistische Analyse eröffnen, zum Beispiel zur Analyse der Beziehungen zwischen verschiedenen Projektmerkmalen und Kosten- bzw. Zeitüberschreitungen. Sie könnte auch die Nutzung der Datenbank für einen Ansatz der Referenzklassenprognose für Benchmarking und Kostenschätzung ermöglichen.16

0.5.5 Unsere Analyse der Kosten- und Zeitüberschreitungen litt unter denselben Einschränkungen hinsichtlich der Daten, die auch unserer Analyse der Einzelkosten Grenzen setzte. Wir haben jedoch die Informationen, die erhältlich waren, zusammengetragen, was uns eine deskriptive Analyse ermöglichte. Table 3 und Table 4 enthalten eine anschauliche Zusammenfassung dieser Analyse. Insbesondere enthält Table 3 eine Zusammenfassung der durchschnittlichen prozentualen Kostenüberschreitungen für alle Projekte in unserer Stichprobe, aufgeschlüsselt nach Sektoren und Ländern. Positive Werte stehen für Kostenüberschreitungen, negative Werte für Kostenersparnisse.

Tabelle 3: Zusammenfassung der Kostenüberschreitungen nach Ländern und Sektoren – durchschnittliche prozentuale Unterschiede zwischen den geschätzten und den tatsächlichen Kosten

(%) / (Projektzahl) Schiene Straße Städt.

Transport Wasser Energie

Gewichte-ter Durch-

schnitt nach

Sektoren

Deutschland -4,3 % (6) -10,0 % (3) -6,2 %

Spanien 12,8 % (6) 30,7 % (1) 17,4 % (2) 15,8 % Frankreich 32,9 % (1) 32,9 % Groß-britannien 110,7 %

(1) 110,7 %

Griechen-land 74,3 % (2) 19,7 % (8) 20,1 % (2) 0,0 % (1) 26,6 %

Irland 2,1 % (5) 74,1 % (1) 14,1 % Italien 62,4 % (5) -5,0 % (2) -0,9 % (1) 37,6 % Polen 19,7 % (2) 80,9 % (2) 50,3 % Portugal 9,0 % (1) 3,3 % (4) 4,4 % Gewichteter Durchschnitt nach Mitglied-staaten

26,9 % 9,4 % 45,4 % 11,3 % 20,7 % 21,2 %

0.5.6 Table 4 enthält eine entsprechende Zusammenfassung der prozentualen Zeitüberschreitungen (berechnet als Verhältnis der tatsächlichen Fertigstellungszeit zur geschätzten Fertigstellungszeit). Auch hier stehen die positiven Werte wieder für tatsächliche Zeitüberschreitungen, während negative Werte zeigen, dass die tatsächliche durchschnittliche Fertigstellungszeit kürzer war als erwartet.

16 Die Referenzklassenprognose ist eine Methode, bei der ein Projektergebnis auf Basis tatsächlicher Ergebnisse einer Referenzklasse vergleichbarer Projekte prognostiziert wird. Die Methode ist zum Beispiel von Prof. Bent Flyvberg im Zusammenhang mit Infrastrukturgroßprojekten angewendet worden, vgl. zum Beispiel Flyvbjerg (2007), „Eliminating bias through reference class forecasting and good governance“.

32

Tabelle 4: Zusammenfassung der Zeitüberschreitungen nach Ländern und Sektoren – durchschnittliche prozentuale Unterschiede zwischen den geschätzten und tatsächlichen Fertigstellungszeiten

(%) / (Projektzahl) Schiene Straße Städt.

Transport Wasser Energie Gewichteter Durchschnitt

nach Sektoren

Deutschland 40,2 % (6)

4,7 % (3) 28,4 %

Spanien 15,3 % (6)

27,3 % (1) 55,9 %

(2) 25,7 %

Frankreich 4,9 % (1) 4,9 % Großbritannien 0,0 % (1) 0,0 %

Griechenland 24,4 % (2)

17,8 % (7) 13,2 % (2) 12,6 %

(1) 17,7 %

Irland 9,0 % (5) 52,2 % (1) 16,2 %

Italien 88,4 % (1) 88,4 %

Polen 5,9 % (1) 2,7 % (2) 3,8 %

Portugal 258,3 % (1) 41,5 %

(4) 84,9 %

Gewichteter Durchschnitt nach Mitglied-staaten

25,8 % 13,2 % 49,6 % 66,7 % 29,8 % 26,2 %

0.5.7 Die in den Tabellen 2 und 3 angegebenen Werte sind Durchschnittswerte, hinter denen sich zwangsläufig eine hohe Varianz der Ergebnisse, die wir für die einzelnen Projekte erhielten, verbirgt. Wir weisen jedoch darauf hin, dass die meisten Projekte in unserer Datenbank nicht pünktlich und nicht ohne Kostenüberschreitungen fertig gestellt wurden.

0.5.8 Diese ersten Ergebnisse scheinen den Nutzen von Analysen dieser Art zu bestätigen. Sofern die Datenbank gepflegt und laufend mit weiteren Daten zur Leistungseffizienz der Projekte aktualisiert wird, könnte sie für die im Zuge der Beurteilung von Großprojekten vorzunehmende Risikoeinschätzung von Nutzen sein.

0.6 Analyse produktiver Investitionen

0.6.1 WP10 befasst sich auch mit der Ergebniseffizienz von Projekten im Bereich produktiver Investitionen (dies sind Projekte, bei denen Unternehmen direkt unterstützt werden), wobei die Ergebniseffizienz an den „Kosten je (durch diese Investitionen geschaffenen) Arbeitsplatz“ gemessen wird.

0.6.2 Abgesehen davon, dass es schwierig ist, die durch den Strukturfonds erzielten Beschäftigungseffekte (insbesondere Schaffung und Erhaltung von Arbeitsplätzen) genau zu messen, erscheint es angesichts der verschiedenen Arten produktiver Tätigkeit, die sich durch die unterschiedlichen Kapital-Arbeitskraft-Verhältnisse in den verschiedenen Sektoren ergeben, zweifelhaft, ob „Arbeitsplatzschaffung“ und somit die „Kosten je geschaffenem Arbeitsplatz“ als sowohl absolute und relative Indikatoren für die Ergebniseffizienz produktiver Investitionen angemessen sind.

0.6.3 Darüber hinaus basieren Entscheidungen über die Gewährung von Finanzmitteln für produktive Investitionen (anders als direkte Beschäftigungsmaßnahmen) auf einer breiten Kosten-Nutzen-Analyse und nicht allein auf der Kosteneffektivität der

33

Arbeitsplatzschaffung. Vergleiche zwischen der geschätzten Arbeitsplatzschaffung und den geschätzten „Kosten je geschaffenem Arbeitsplatz“ mit den tatsächlichen Ergebnissen für diese Indikatoren mögen für einzelne Projekte nützlich sein. Projektübergreifende Vergleiche dieser Indikatoren dürften jedoch wenig aufschlussreich sein, wenn man diese nicht den besonderen Bedingungen in den verschiedenen Wirtschaftszweigen anpasst.

0.6.4 Unserer Ansicht nach wäre es angemessener, diese Projekte anhand eines breiteren Kosten-Nutzen-Ansatzes zu bewerten und dann die relative Effizienz anhand der Kosten-Nutzen-Verhältnisse zu vergleichen.

0.7 Das Datenbank-Tool für Benchmarking

0.7.1 Als Teil dieses Berichts haben wir auf Kalkulationstabellen beruhende Datenbank-Tools für die Projekteinzelkosten entwickelt und aufgebaut, welche die im Laufe unserer Analyse gesammelten Informationen enthalten und den leichten Zugriff auf projektspezifische Daten ermöglichen. Auch Daten zu neuen Projekten könnten, sobald diese vorliegen, dort aufbewahrt werden. Zu Beginn unserer Studie versuchten wir, auftragsgemäß, Datenbank-Tools sowohl für Infrastrukturprojekte als auch für produktive Investitionen zu entwickeln. Unsere Analyse hat jedoch gezeigt, dass ein Benchmarking produktiver Investitionen auf Grundlage der „Kosten je geschaffenem Arbeitsplatz“ ohne entsprechende Anpassungen hinsichtlich der unterschiedlichen Gegebenheiten in den verschiedenen Wirtschaftszweigen wahrscheinlich nicht aussagekräftig wäre. Aus diesem Grunde wurde letzten Endes nur das Datenbank-Tool für Infrastrukturprojekte für die allgemeine Verwendung entwickelt.

0.7.2 Das auf einer Kalkulationstabelle basierende Tool ist in Input- und Output-Blätter gegliedert. In die Input-Blätter können die Dienststellen der Kommission die rohen Projektdaten eingeben. Die Daten durchlaufen dann verschiedene „Normalisierungsblätter“, in denen die Rohdaten unter Berücksichtigung von Wechselkursen und Inflation in vergleichbare Einzelkosten umgerechnet werden. Mit den Output-Blättern können die Benutzer (zum Beispiel Projektbeurteiler) nach Projekten einer vergleichbaren Klasse suchen sowie Zusammenfassungen einsehen, die ihnen einen Überblick über alle Projekte in einem bestimmten Sektor geben.

0.8 Schlussfolgerungen und Schlussbemerkungen

0.8.1 Trotz unserer Schwierigkeiten bei der Datensammlung und obwohl wir nicht genügend Informationen hatten, um statistisch begründete Schlussfolgerungen zu ziehen, haben wir versucht, einen umfassenden Überblick über die Daten zu geben, die über vom EFRE kofinanzierte Infrastrukturprojekte und produktive Investitionen vorliegen, u.a.:

• die geschätzten und tatsächlichen Gesamt- und Einzelkosten je EFRE-Infrastrukturprojekt;

• die Vergleichsdaten für Einzelkosten, die in bestehenden Datenbanken und von früheren Bewertungen vorliegen; sowie

• die Stellen, die durch produktive Investitionen des EFRE geschaffen wurden, mit den Gesamtkosten je geschaffenem Arbeitsplatz und dem Betrag der EFRE-Finanzhilfe je geschaffenem Arbeitsplatz.

0.8.2 Durch die Datenbank, die Vergleichsdaten zu den Einzelkosten von Infrastrukturprojekten und Angaben zu den Projektmerkmalen enthält, und das auf Excel basierende Tool kann die Generaldirektion Regionalpolitik ihre

34

Projektbeurteilungen künftig leichter durchführen. Allerdings gab es zu wenig verfügbare Daten, was unserem Versuch, (i) den anderen Hauptzweck von WP10, nämlich eine Ex-post-Bewertung der Leistungseffizienz (Kosten- und Zeitüberschreitungen) bei der Durchführung der in der Stichprobe enthaltenen großen Infrastrukturprojekte, zu erreichen, sowie (ii) harte Einzelkostenvergleichsdaten zu liefern, Grenzen setzte.

0.8.3 Das Benchmarking und die Datenerhebung, die im Rahmen WP10 erfolgten, sind jedoch als erster Schritt auf dem Weg zur Entwicklung eines umfassendes Benchmarking-Tools für die künftige Beurteilung mit Mitteln der Europäischen Kommission (EFRE, Kohäsionsfonds, ISPA usw.) finanzierter Projekte und Investitionen zu sehen; beides lässt sich sowohl für bereits etablierte als auch für neue Methoden anwenden (zum Beispiel bei der Anwendung der Referenzklassenprognose).

0.8.4 Künftig wird man, auch zur Pflege und zum Ausbau der Datenbank, neue Projektdaten und Informationen sammeln müssen, und zwar – anders als bisher – regelmäßig und in einheitlicher Weise. Außerdem wird man nicht nur Kanäle aufbauen müssen, über welche die Mitgliedstaaten die Daten für Großprojekte regelmäßig und in einheitlicher, guter Qualität liefern, sondern auch gründliche Systeme für die Datenüberprüfung.

0.8.5 Der Wunsch nach einer besseren Beurteilung, Gestaltung und Bewertung großer Infrastrukturprojekte ist nicht neu. Insbesondere ist den Behörden, der Industrie und den Wissenschaftlern in einigen europäischen Ländern wie Dänemark, den Niederlanden, Norwegen und Großbritannien in den letzten Jahren zunehmend deutlich geworden, dass die Entwicklung neuer Methoden zur besseren Kostenschätzung erforderlich ist. Darin spiegelt sich die Tatsache wider, dass es bei der Durchführung von Großprojekten in der Vergangenheit häufig Kosten- und Zeitüberschreitungen gegeben hat.

0.8.6 Die für diese WP10-Studie durchgeführten Arbeiten scheinen diesen jüngeren Entwicklungen in einigen europäischen Ländern zu folgen, woraus sich künftig Möglichkeiten für eine Zusammenarbeit zwischen der Generaldirektion Regionalpolitik und den EU-Mitgliedstaaten in diesem Bereich ergeben könnten.

0.8.7 Unsere Empfehlungen sind nachstehend niedergelegt.

Aufbau einer EU-weiten Datenbank für die Kosten großer Infrastrukturprojekte

0.8.8 Das im Zuge von WP10 entwickelte Tool auf Basis einer Excel-Kalkulationstabelle ist ein nützlicher Ausgangspunkt für den Aufbau einer solchen Datenbank der von der EU finanzierten Projekte. Die Datenbank kann im Laufe der Zeit aufgebaut werden und als wertvolle aktuelle Quelle für Vergleichskostendaten dienen.

0.8.9 Wir empfehlen der Kommission deshalb, Überlegungen anzustellen, wie bestmöglich sichergestellt werden kann, dass die Datenbank laufend aktualisiert wird, um Qualität und Quantität der Projektdaten im Laufe der Zeit zu verbessern. Drei weitere Empfehlungen könnten zur Nützlichkeit der Datenbank beitragen, nämlich: • die Übernahme gemeinsamer Einzelkostendefinitionen auf nationaler wie auch auf

EU-Ebene;

• die Reform der Überwachung und Berichterstattung über Großprojekte sowie eine gründlichere Beurteilung beantragter EU-Fördermittel auf der Ebene der Mitgliedstaaten; und

• die Einbeziehung von EU-Finanzinstitutionen wie der Europäischen Investitionsbank und der Europäischen Bank für Wiederaufbau und Entwicklung, die ebenfalls an der Kofinanzierung von Großprojekten beteiligt sind.

35

0.8.10 Der oben vorgeschlagene kollaborative Ansatz könnte zahlreiche Vorzüge bieten, u.a. mehr Datenpunkte und die Möglichkeit, die Unterhaltskosten für die Datenbank zu teilen. Er könnte auch die multilaterale Entwicklung gemeinsamer Kostendefinitionen und in der Folge die breitere Anwendung dieser Definitionen erleichtern. Diese könnten, wie bereits erwähnt, gemäß dem von uns für WP10 entwickelten dreistufigen Ansatz definiert werden (vgl. Abschnitt 5.4 dieses Berichts).

Standardisierung und Verbesserung der Ex-ante-Risikoeinschätzungen bei Finanzierungsanträgen

0.8.11 Außerdem meinen wir, dass die Anwendung der Methoden für die Ex-ante-Einschätzung der Projektrisiken, die in der jüngsten Anleitung der Kommission zur Kosten-Nutzen-Analyse niedergelegt sind, die Qualität dieser Einschätzungen (die für alle Großprojekte vorgeschrieben sind) erheblich verbessern würde. 17 , 18 Unsere Empfehlung basiert auf einer Prüfung etlicher der relativ wenigen Projektakten, die detaillierte Berichte zur Begründung der EFRE-Anträge enthielten (welche im Allgemeinen nur die Schlussfolgerungen solcher Prüfungen enthielten).

Standardisierung und Verbesserung der Projektüberwachung und Berichterstattung bei Großprojekten

0.8.12 Auf Grundlage der Antworten, die wir auf unsere Anfragen und Fragebogen erhielten, scheint es, dass von den Mitgliedstaaten bzw. in deren Auftrag relativ wenig Projektabschluss- oder -fortschrittsberichte für mit EFRE-Mitteln finanzierte Großprojekte erstellt werden. Darüber hinaus waren die wenigen Berichte, die wir prüften, von nur eingeschränktem Nutzen für die Zwecke der Projektbewertung.

0.8.13 Wir empfehlen der Kommission, die erforderlichen Maßnahmen zu treffen, um das Projektberichterstattungsverfahren für Großprojekte zu verbessern, mit dem Ziel, die Projektüberwachung zu verbessern und mehr Projektdaten für die Planung, Projektbeurteilung und Projektbewertung zur Verfügung zu haben.

0.8.14 Insbesondere die folgenden Verbesserungen sollten in Erwägung gezogen werden:

• auf Antragsformularen sowie bei der Beurteilung und Überwachung von Projekten auf die Benutzung von Standarddefinitionen der Projektkosten zu bestehen;

• die Finanzierung davon abhängig zu machen, dass regelmäßig Bericht über den Projektfortschritt erstattet wird;

• den weiteren Abruf von Finanzmitteln davon abhängig zu machen, dass regelmäßig Projektfortschrittsberichte (in vorgegebenem Format) vorgelegt werden;

• die Finanzierung davon abhängig zu machen, dass ein Projektabschlussbericht (im vorgegebenen Format) vorgelegt wird.

17 Europäische Kommission, Generaldirektion Regionalpolitik, Anleitung zur Kosten-Nutzen-Analyse von Investitionsprojekten, Strukturfonds, Kohäsionsfonds und ISPA 16/06/2008 18 Ein nützliches Handbuch des US Government Accountability Office ist „Cost estimating and assessment guide: Best practices for developing and managing capital program costs“, GAO Applied Research and Methods (2009).

36

1 SECTOR SNAPSHOTS

1.1.1 The following sector snapshots provide an overview of our findings for rail, road, urban transport, waste and wastewater and energy projects. We provide an indication of typical level 1 and Level 2 costs and an indication of sample size. A summary of the main reasons found for time and cost over-runs is also provided.

1.1.2 Further details are provided in chapters 5 and 6.

Sector Snapshot - Rail Figure A: Average level 1 costs by category

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

Estimated Actual

Uni

t cos

t (EU

Rm

/km

)

Land

Taxes

Contingency

Soft

Build

Figure B: Average level 2 costs by investment component

0.0

10.0

20.0

30.0

40.0

50.0

60.0

Trackw

ork

Rolling S

tock (

/km)

Rolling S

tock (

/nr)

Stations

(/m2)

Stations

(/nr)

Bridges

(/m2)

Bridges

(/nr)

Tunnels Lan

dOther

Uni

t cos

t (EU

Rm

/km

)

Estimated MeanActual Mean

37

Sample

The sample size was 24 projects. Most of the projects in the sample were twin track projects. Seven Member States were represented: Germany, Greece, Ireland, Italy, Poland, Portugal and Spain.

The average length of rail projects in the sample is 127 km. However, there is a significant variance: the shortest project is 8.9 km, the longest 1,435 km.

Average unit costs

Level 1

Actual Level 1 unit costs (10 EURm/km) were, on average, 27% higher than estimated (8 EURm/km) (see figure A). Actual Level 1 unit costs for rail projects ranged from 0.3 to 49 EURm/km.

However, rail projects in the sample had a high degree of variation in Level 1 unit costs partly due to the varying project design features (see Figure A).

Level 2

Limited data was available for Level 2 costs. However, it indicated costs broadly in line with estimates (see figure B).

A key factor in the cost of tunnels is whether the tunnel is in an urban or rural area – urban tunnels costing significantly more than rural tunnels.

Benchmark ranges

Level 1 unit cost benchmarks derived from our literature review range between 2.9 and 4.4 EURm/km for single track; and between 2.8 and 8.6 EURm/km for twin track projects. However, these benchmarks correspond to rail projects that are not necessarily directly comparable with those in the sample (project scopes vary significantly).

Main reasons for cost overruns

‘Project specific’ factors was the most important generic category of cost overruns in rail projects. Within this category, the most frequently cited reasons for cost overruns were ‘environmental impact’, ‘delays by statutory authorities’ and ‘work suspensions’.

Main reasons for delays

On average, rail projects were completed with a delay of 50% of the initial estimated completion time. ‘Funding’ and ‘Construction’ phases experienced the greatest average delays (115% and 51% respectively).

‘Project specific’ factors was the most important broad category of delays. Within this category, ‘Delays by statutory authorities’ and ‘Late Commencement’ were the most important factors. In the general ‘procurement’ category, ‘design changes’ were also an important factor contributing to delay.

38

Sector Snapshot - Roads

Figure A: Average level 1 costs by category

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

18.0

Estimated Actual

Uni

t cos

t (EU

Rm

/km

)

LandTaxesContingencySoftBuild


0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

Pavement Other Tunnels Land Bridges (/m2) Bridges (/nr)

Uni

t cos

t (EU

Rm

/km

)

EstimatedActual

Sample

The sample size was 22 projects. Most of the projects in the sample are two lane roads projects. 6 Member States are represented: Germany, Greece, Ireland, Italy, Poland and Spain.

The average length of road projects in the sample is approximately 33 km. However, there is significant level of variance: the shortest project is 5.3 km, the longest 144.2 km.

39

Average unit costs

Level 1

Actual 1 unit costs of 15 EURm/km were, on average, 21% higher than estimated (see figure A). Actual Level 1 unit costs for road projects varied widely - from 1.03 to 49.80 EURm/km (see Figure 23). .

Higher unit costs tend to be associated with complex projects and difficult environment conditions (e.g., wetlands, mountainous terrain).

Level 2

Actual Level 2 unit costs indicates that actual costs for bridges, pavement and tunnels were 37%, 22% and 15% higher than estimated (see figure B).

The limited amount of data (due to the small sample size) indicated that urban roads suffer from significantly higher pavement unit costs, presumably due to the complexity of re-paving urban areas compared with rural ones (eg associated utility works).

Unit cost benchmarks

Level 1 unit cost benchmarks range between 0.7 and 2.0 EURm/km for a single carriageway, two-lane road; and between 2.7 and 8.7 EURm/km for a double carriageway, two-lane road. The unit cost benchmark range for a double carriageway, three-lane road is between 36.4 and 91.3 EURm/km. However, these benchmark ranges, which we have calculated on the basis of the values we obtained from the literature review, correspond to road projects that are not directly comparable with those in the sample.

Reasons for cost overruns

‘Project specific’ factors were the most important cause of cost overruns in road projects compared with ‘project environment’, ‘client specific’, and ‘procurement’ or ‘external factors’ (see section 5.7). Among these ‘project specific’ factors, the most frequently cited reasons for cost overruns were ‘design complexity’, ‘environmental impact’ and ‘delays by statutory authorities’. ‘Project environment’ factors were also important for road projects cost overruns.

Reasons for time delays

‘On average, across all construction phases, road projects tend to accumulate a delay of approximately 17% over the initial estimated completion time. Site preparation and construction phases suffered from the highest average delays (28% and 27% respectively) (see Table 29).

‘Project specific’ factors were the most important generic cause for delays in road projects. Within this general category, ‘Site access’, ‘delays by statutory authorities’ and ‘construction period’ were the most significant factors.

40

Sector Snapshot – Urban Transport


0.0

5.0

10.0

15.0

20.0

25.0

Estimated Actual

Uni

t cos

t (EU

Rm

/km

)



0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

Trac

kwor

k

Rolling

stoc

k (/km

)

Rolling

stoc

k (/nr

)

Station

s (/m

2)

Station

s (/nr

)

Bridge

s (/m

2)

Bridge

s (/nr

)

Tunn

elsLa

ndOth

er

Unit

cost

(EUR

m/k

m)

EstimatedActual

Sample

The sample size was 8 projects. Five Member States are represented, namely France, Greece, Ireland, Poland and Portugal.

The average length of urban transport projects in the sample is approximately 14.8 km. However, there is a significant level of variance. The shortest project is only 3.9 km long, while the longest is 21.9 km long.

41

Average unit costs

Level 1

Actual Level 1 unit costs were, on average, 37% higher than estimated (see Figure A). The estimated (actual) Level 1 unit costs for urban transport projects range from a minimum of 3.61 (4.93) EURm/km to maximum of 74.40 (118.32) EURm/km.

Two projects (both in Poland) had exceptionally high unit costs attributed to high degrees of complexity (see Figure 26).

Level 2

Level 2 unit costs for track work were 22% higher than estimated, but actual costs for bridges and rolling stock were 2% and 12% lower than estimates (see figure B).

Although limited, the evidence shows that unit cost (per m2) of urban transport stations is strongly affected by the specific characteristics of the project – for example an underground station in Athens (with archaeological issues) being particularly high cost.


The range of Level 1 unit cost benchmarks for Metro projects is very wide: between 27.3 and 482.9 EURm/km. This is because unit costs are strongly dependent on the physical characteristics of each project, and, specifically, on the amount of new tunnelling required. For example, at the top end of the benchmark cost range is the Jubilee Line Extension project, on the London Underground (not funded by ERDF). This project required four river crossing tunnels, which contributed to raising the project’s total cost. Unit costs for tram projects range between 2.5 and 64.3 EURm/km. As for metro projects, multiple differences in project characteristics make it difficult to use simple unit cost benchmarks for comparative analysis.


‘Project environment’ factors were the most important cause of cost overruns in urban transport projects (see section 5.7). Within this category, gaining ‘permits/consents’ is the most frequently cited cause of cost overrun (in six out of eight projects). Procurement factors were also important for urban transport cost overruns.


On average, across all construction phases, urban transport projects tended to accumulate a delay of 29% over the initial estimated completion time. ‘Funding’, followed by ‘Planning’, were the phases with the higher average delay (60% and 37% respectively) (see Table 25).

Factors related to ‘project environment’, followed by factors related to procurement issues were the most important generic cause for delays in urban transport projects.

42

Sector Snapshot Water and Wastewater


0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Estimated Actual

Uni

t cos

t (EU

Rm

/km

)


Figure B: Average level 2 costs by project type

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Water supply Sewage network Other Waste treatment Land

Unit

cost

(EURm

/uni

t)

EstimatedActual

Sample

The analysis was based on a very small sample size of 4 water and wastewater projects. Only two Member States were represented: Italy and Spain.

For three out of four projects, a measure of project length was provided. This ranged from 85 km for the ‘Desaladora’ project to 94 km for the ‘Conduccion Jucar’ project, both in Spain.

43

Average unit costs

Level 1

In the very small sample, actual average Level 1 unit costs were 17% higher than estimated (see figure A). The actual Level 1 unit costs for water and wastewater projects ranged from a minimum of 0.75 to 2.11 EURm/km.

The water projects in the sample have broadly similar unit perhaps reflecting similar conditions.

Level 2

Limited and incomplete evidence on Level 2 unit costs indicates that actual Level 2 costs for water supply projects were about 10% higher than estimated ones (see figure B). Actual land costs, on the other hand, appear to be 13% lower than estimated ones.


While Level 1 unit cost benchmarks are available for water projects (water treatment and water supply), as identified in the literature review, their values are much higher than the unit costs in our database, suggesting that these projects may not be directly comparable. These benchmarks are for cost relative to capacity, measured in m3/day. For water treatment they range between 98 and 3800 EUR/m3/day and for water supply between 370 and 6800 EUR/m3/day.


‘Procurement’ factors were perceived as the most important cause of cost overruns in water and wastewater projects (see section 5.7). Among these ‘procurement’ factors, the most cited reason for cost overruns was ‘design changes’.


On average, across all construction phases, water and wastewater projects were subject to a 44% delay over the estimated time completion time Site preparation and funding are the project phases with the highest average delays (153% and 74% respectively) (Table 26).

‘Project environment’ factors were to be the most important cause of time delays. Within this category ‘Gaining ‘permits/consents’ was the most common factor.

44

Sector Snapshot - Energy Because of the very different characteristics of different types of energy projects, it is difficult to

find a measurement unit that is appropriate for all energy projects. Therefore, in our analysis we have identified three measurement units:

• Length in Km (e.g. for pipeline projects)

• Installed capacity in MW (e.g. for power plants)

• Number of installations (in all cases where the previous two units are not appropriate).

For most projects in our sample, a length measure was provided. This has allowed us to calculated average unit costs per km, as shown in Figure A below. This chart summarises the unit cost for all project for which information about their length was available. In the current sample, however, no information about installed capacity or number of installations was available. Therefore, we did not calculate average values using these measurement units.

Table A: Average level 1 costs by category

2.1

2.2

2.3

2.4

Estimated Actual

Uni

t cos

t (EU

Rm

/km

)


Sample

The sample size was 8 projects. Most of the projects in the sample are wind farms developments. Three Member States are represented: Greece, Portugal and UK.

Average unit costs

Level 1

Using km as measurement unit, the evidence collected shows that, on average, actual Level 1 unit costs are 20% higher than estimated ones (see Figure A).

Level 2

Only one measure, unit costs of wind farm turbines, can be meaningfully compared across projects. The evidence shows that, on average, the actual unit costs of turbines were only 1.3% higher than estimated.


In the literature review, we have considered possible benchmarks for energy projects. However,

45

the range of values we obtained was too wide to be used as a meaningful reference point for the projects in our sample.


‘Project specific’ factors were the most important cause of cost overruns (see section 5.7). Among the ‘project specific’ factors, the most common cause of cost overruns was ‘environmental impact’.


On average, across all construction phases, energy projects tended to accumulate a 23% delay over their development. ‘Funding’ is the project phase with the higher average delay (approximately 72% over and above the estimated completion time) (see Table 27).

Project specific’ factors were the most important generic cause for delays in energy projects. Within this category, ‘Delays by statutory authorities’ was the most common factor.

46

2 INTRODUCTION

2.1 Background

2.1.1 This is the Final Report on Work Package 10 (WP10) of the European Commission, Directorate-General for Regional Policy’s ex post evaluation of Cohesion Policy programmes 2000-2006 financed by the European Regional Development Fund (ERDF) in Objective 1 and 2 regions. The title of Work Package 10 is ‘Efficiency – Unit costs of major projects’.

2.1.2 The notion of efficiency under WP10 refers to the achievement of the desired or projected benefits at a reasonable cost. The concern is with ‘major projects’, which includes large infrastructure and productive (or business support) investments.19 The Commission distinguishes between two aspects of evaluating efficiency: (i) output efficiency: unit costs and completion times of major infrastructure projects, including an analysis of cost and time overruns; and (ii) result efficiency: the cost per job created as a result of productive investments.20

2.1.3 The main objectives of the WP10 Study were three: (1) to undertake an ex-post evaluation of the performance (cost overruns and time delays) of major infrastructure projects and so-called productive investments co-financed by ERDF based on a representative sample; (2) to develop a database of unit cost benchmarks and project characteristics to assist the Commission in the appraisal of future project financing requests; and (3) to develop an Excel-based tool that includes the project data collected through the WP10 exercise (with the potential to be populated with data from other infrastructure projects funded by the EC) to facilitate project appraisal in the future.

2.1.4 The WP10 Tender Specifications set out a requirement for these efficiency evaluations to be undertaken for a sample of 115 major infrastructure projects and 40 productive investments. This sample of 155 major projects is, in turn, derived from a population of 271 projects from the 11 EU Member States that are the subject of the WP10 study. The total eligible expenditure involved in these projects amounts to €33 billion, of which approximately €15 billion is the ERDF contribution.

2.2 Purpose of this Report

2.2.1 This Final Report serves as the sixth (and final) deliverable of the WP10 Study and, in carrying it out we were required to report on all of the Tasks that were specified as being part of the Study in the Tender Specifications.

2.2.2 These Tasks and the corresponding sections of the report are listed below in Table 5. This is followed by a more detailed summary of the contents of the report and what was involved in the WP10 Tasks.

19 In order for projects to be considered ‘major’, they must ‘comprise an economically indivisible series of works fulfilling a precise technical function and which have clearly identified aims’. They must also be projects ‘whose total cost taken into account in determining the contribution of the Funds exceeds EUR 50 million’. See Article 25 of Council Regulation 1260/1999 laying down general provisions on the Structural Funds. 20 However, as well as evaluating the efficiency of major projects undertaken in the 2000-2006 period, it is the Commission’s intention that WP10 will also develop and test a methodology for investigating unit costs which can later be applied in future evaluations, such as the planned ex post evaluations of the Cohesion Fund and ISPA.

47

Table 5: How this report addresses the tasks (as per the Tender Specifications for WP10) Task Title Section

1.0 Review of existing literature and evaluations Sections 2, 3

2.1 Infrastructure – Definitions of unit costs Section 4

2.2 Infrastructure – Calculation of estimated unit costs and completion times

Section 3, 5

2.3 Infrastructure – Calculation of actual unit costs and completion times

Section 3, 5

2.4 Infrastructure – Analysis of cost overruns and time delays Section 5

2.5 Infrastructure – The role of ex ante risk assessment Section 5

3.1 Productive investments – definitions of employment effects and analysis of investments

Section 3, 6

3.2 Productive investments – calculation of estimated costs per job created

Section 3, 6

3.3 Productive investments – calculation of actual costs per job created Section 3, 6

4.0 Development of a spreadsheet of unit costs Section 7

2.3 Structure of this Report

2.3.1 We begin, in section 2, with a shortened version of our review of the relevant literature.21 We included, in our review, the relevant academic literature, guidance and published reports on previous evaluations carried out by (and for) DG Regio and guidance and evaluation reports published by the Member States and by relevant agencies in the Member States.

2.3.2 Section 3 describes the information gathering exercise. This had a number of elements, as follows:

• The first, outlined in section 4.2, was the search for benchmark databases of infrastructure project costs, which yielded, with little exception, data that was not fit-for-purpose. (The data we did find is presented in Annex I.)

• The second, the subject of section 4.3, was a review of the ex post evaluation activities in the Member States. This is potentially significant, particularly in respect of the potential benefits that can arise from the complementary nature of this work and the work of DG Regio.

• The third, the subject of section 4.4, was the gathering of data about the major projects in the WP10 sample. The latter proved less straightforward than expected and, so, the section also seeks to describe the potential reasons why information on the projects in our sample (both quantitative and qualitative) was difficult to acquire.

21 Shortened, that is, relative to our First Interim Report on the WP10 Study. See http://ec.europa.eu/regional_policy/sources/docgener/evaluation/pdf/expost2006/wp10_interim_report.pdf .

48

2.3.3 Sections 5 and 6 are concerned with major infrastructure projects. Section 4 addresses the following issues:

• The definition and measurement of unit cost indicators, including the categories of cost to include in the numerator of unit cost indicators (section 5.2) and the disaggregation of the physical components of projects (section 5.3).

• The unit cost indicators that we defined for the specific purposes of the WP10 Study (section 5.4).

• The information sought on the attributes (both general and sector-specific) of major infrastructure projects (section 5.5).

• The methods used to measure project completion times and project durations and, in turn, project delays and cost overruns (section 5.6);

• The methods used to identify the causes of cost overruns and project delays (section 5.7).

• The methods used to achieve data comparability across projects and across countries (section 5.8).

2.3.4 Section 5 provides our analysis of the WP10 sample of major infrastructure projects, including the comparison between estimated and actual costs and the analysis of the most prominent cost-determining factors for these projects. We also provide an analysis of project delays and cost overruns, identifying how their causes differ between sectors. Section 5 also examines the role of ex ante risk assessment in the project appraisal phase and whether and how robust such assessments were for the WP10 sample of major projects.

2.3.5 Section 6 presents our analysis of major productive investments. The section is presented with the following structure:

• Section 7.2 considers the methods used to estimate the employment effects at the appraisal stage of the WP10 sample of productive investments.

• Sections 7.3, 7.4 and 7.5 examine, respectively, the total cost of the productive investment projects in our sample, the numbers of jobs created as a result of those projects and a comparison between the estimated and actual cost per job created.

• Section 7.6 examines the role of delays in causing discrepancies between the estimated and actual total cost of projects as well as between estimated and actual cost per job created.

• Sections 7.7 and 7.8 consider, respectively, the different types of funding that make up productive investments and the amount of Structural Funds per job created involved in the projects.

2.3.6 Section 7 describes the manner of development of the spreadsheet tool of unit costs. Section 8 draws out a summary and our conclusions from the WP10 Study. Finally, Section 9 makes some recommendations, based on these conclusions.

2.3.7 Annex I provides a graphical presentation of our external benchmarking results. Annex II shows the questionnaires that we developed and sent to the Member States for the purposes of gathering information about the projects in our sample. Questionnaires for each of the infrastructure sectors are included, as well as that used for productive investments. Annex III provides accompanying notes to the project monitoring questionnaires.

49

3 IMPORTANT LESSONS FROM THE LITERATURE

3.1 Introduction

3.1.1 In this section, we provide a summary of the important lessons learned from our review of the literature on major infrastructure project evaluation, as well as on the evaluation of investments in job creation.

3.1.2 The section is structured as follows:

• Section 3.2 summarises the main issues involved in formulating robust and comparable unit cost definitions and measures;

• Section 3.3 examines the determinants of project costs and the implications for the analysis of differences between projects within sectors and both within and between countries;

• Section 3.4 summarises the key issues involved in the identification and measurement of project delays and cost overruns.

• Section 3.5 provides a review of the (relatively sparse) literature on the measurement of employment effects and the assessment of the result efficiency of productive investments.

• Section 3.6 draws some brief conclusions.

3.2 Unit cost definitions and measurement for infrastructure projects

Introduction

3.2.1 The evaluation of output efficiency of major infrastructure projects involves, in the current context, making comparisons between the costs of those projects, i.e., cost benchmarking. However, because no two projects can be expected to be of equal size or specification, it would be meaningless to compare the total costs of projects. Costs for each project must, rather, be expressed on a common ‘per unit’ basis. For example, it is more meaningful to compare the average cost per km of a 20km road project with the average cost per km of a 15km road project than it is to compare the total cost of each road.

3.2.2 For that reason, we are concerned with the concept of a ‘unit cost definition’, such as, for example, the ‘total cost per km of road’ introduced in the previous paragraph. The robustness of the results will depend on the consistent application of the unit cost definitions that are adopted as the basis for comparison. They must, therefore, be widely applicable and must facilitate comparability.

3.2.3 A number of different unit cost definitions can be considered for different types of project. For example, for transport projects, costs per km of rail or road are useful. For buildings, costs per square metre are often used. For energy or water projects, the capacity of the project is often used, for example, cost per KwH of generating capacity or millions of cubic metres of water treated.

3.2.4 There are a number of important dimensions to the formulation of useable unit cost definitions. Each of these is explored in greater detail in the following subsections. They are:

• the types of costs to be included in the numerator;

• the level of disaggregation of each project’s components; and,

50

• The methodologies that should be used to ensure data comparability across heterogeneous projects.

The types of cost to be included

3.2.5 The issue here concerns which project costs to be included in the numerator of the unit cost measure. The unit of output to be used in the denominator is usually a relatively straightforward matter. For example, the World Bank’s ROCKS framework for road projects suggests the use of road length (giving a cost per kilometre) or road surface area (giving a cost per square metre).22

3.2.6 The literature suggests that similar projects can use very different unit cost definitions. A couple of examples are illustrated as follows.

• Blanc-Brude et al. (2006) use project costs that incorporate design, engineering, construction and supervision, but that excludes land purchases, technical and price contingencies, taxes, start-up costs and fees, and the interest payments during the construction phase. The project cost, so defined, is divided by the length of the road that was constructed, measured in km.23

• The World Bank ROCKS framework uses ‘civil work’ costs, which incorporate items such as mobilisation (pavement-drainage), major structures (including line markings), contingencies and taxes, but excludes other agency costs such as design, land acquisition, resettlement and supervision.

3.2.7 The 1998 European Commission DGXVI (now DG REGIO) guidance on the cost-determining factors of infrastructure development was designed to provide desk officers with a basic understanding of the process by which project cost estimates are made so that they would be better able to review, with the project sponsors, the reasons for actual or anticipated cost and time overruns.24 The study provided indicative shares of 5 cost categories for seven types of infrastructure project. The results were presented in Figure 1 of our First Interim Report and are re-presented in Figure 2 below.

22 ROCKS stands for Road Costs Knowledge System. 23 See Blanc-Brude, Frédéric, Hugh Goldsmith and Timo Välilä (2006), ‘Ex Ante Construction Costs in the European Road Sector: A Comparison of Public-Private Partnerships and Traditional Public Procurement’, European Investment Bank, Economic and Financial Report 2006/01. 24 European Commission DG XVI (1998), ‘Understanding and Monitoring the Cost-Determining Factors of Infrastructure Projects’, A User’s Guide, Brussels.

51

Figure 2: Indicative shares of 5 cost categories for infrastructure types

Source: European Commission DG XVI (1998), ‘Understanding and Monitoring the Cost-Determining Factors of Infrastructure Projects’, A User’s Guide, Brussels

3.2.8 The table in Figure 2 is useful in providing an indication of the relative significance of different elements of cost. However, one must remember that diversities in the circumstances and conditions under which projects are implemented must be borne in mind when using it as part of an evaluation framework.

3.2.9 While all approaches reviewed include construction, they are at odds over whether items like contingencies and taxes should be included. Prof. Flyvbjerg (during the course of the expert meetings that were part of the WP10 Study) asserted, for example, that if VAT is a genuine cost to the project, then it should be included in the calculation of unit costs.

3.2.10 Flexibility in the choice of unit cost definition will, of course, often be constrained by the availability of data. A common difficulty is the absence of cost breakdowns that would facilitate the disaggregation of costs required to develop comparable unit cost measures. Moreover, it is often unclear which items have been included in an aggregate cost measure.25

3.2.11 These difficulties were, as might be expected and as outlined in section 5.2 below, encountered with all but a few of the projects in the WP10 sample.

25 See Flyvbjerg, Bent, Mette Skamris Holm and Soren Buhl (2003), ‘How common and how large are cost overruns in transport infrastructure projects?’ Transport Reviews, Vol. 23 No. 1

52

Elements of projects to include in unit cost definitions

3.2.12 As noted by the European Commission itself and in the literature, no two infrastructure projects will cost the same no matter how similar they are.26 For that reason, unit cost benchmarks will only be useful when they accurately reflect the average cost of sufficiently disaggregated components of projects.27 For example, large sections of tunnel or bridge on a road or rail project would be expected to increase the total cost of the project and, therefore, comparing such projects with analogous projects that do not involve these elements would be meaningless.

3.2.13 The Nichols Report (2007) asserts that, at a minimum, cost breakdowns for significant fixed link components (bridges and tunnels) are required.28 Flyvbjerg and COWI (2004) found that fixed link components were statistically distinct from ‘normal’ road components, both in terms of their unit cost and the risk of cost overruns.29 This, the authors claim, provides a strong argument for treating fixed-link components as a separate category of project.

3.2.14 Stalder (2000) produced a study that was part of a wider project conducted by UIC for benchmarking the cost of railway infrastructure.30 In the process, Stalder adhered to the principles of (i) never comparing projects with each other but with benchmarks that adjust for the specific complexities and aspects of the environment; (ii) taking into account all infrastructure elements except stations (buildings and platforms); (iii) the application of precise, consistent definitions for each infrastructure element and its cost; and (iv) using cost per metre of track as the basis for comparison across project dimensions.

3.2.15 For urban transport projects:

• Pickrell (1985) distinguished stations as they represent a significant component of urban rail projects.31 The author also found that tunneling and the elevation of stations were significant drivers of cost.

• Flyvbjerg et al (2008)32 established the proportions of cost involved in the various components of urban transport projects in the US. They found that stations could constitute anywhere between 20 and 35 per cent of the total cost of these projects. Engineering, project management and testing were also found to be significant.

• BB&J Consult (2000) produced similar cost disaggregations for urban transport projects in Latin America.33

3.2.16 Literature on the relevant aspects of environmental infrastructure projects is much sparser. However, in the same way that ‘normal’ stretches of road should be distinguished from fixed-link components and railway stations from track, it seems

26 See footnote 24. 27 Flyvbjerg et al (2003) noted that acquiring disaggregated data to a level sufficient to facilitate meaningful comparisons is essential, but can be time-consuming (and, therefore, expensive) or even impossible. See footnote 25 for reference. 28 Nichols, Mike (2007), ‘Review of Highways Agency’s Major Roads Programme’, Report to the Secretary of State for Transport, London, March. 29 See Flyvbjerg, Bent in association with COWI (2004), ‘Procedures for Dealing with Optimism Bias in Transport Planning: Guidance Document’, for the British Department for Transport, London, June. 30 See Stalder, Oskar (2000), ‘International Benchmarking of Track Costs’, part of the International Union of Railways benchmarking project. 31 See Pickrell, Don H. (1990), ‘Urban Rail Transit Projects: Forecast versus Actual Ridership and Costs’, Report prepared for the US Department of Transportation, Washington D.C. 32 Flyvbjerg et al. (2008) studied cost per kilometre variations across urban rail projects with a sample that included 17 European projects, 6 projects from the US, 6 from non-US/Europe and 4 from UITP (Athens, Cairo, Frankfurt and Lisbon). The latter were later rejected by the authors. See Flyvbjerg, Bent et al (2008), ‘Comparison of Capital Costs per Route-Kilometre in Urban Rail’, EJTIR, 8, no.1. 33 See BB&J Consult, S.A. (2000), ‘Implementation of Rapid Transit’, Urban Transport Strategy Review, World Bank, Washington D.C.

53

logical to distinguish between plants/installations and pipeline (that makes up the water/sewerage network). In the 2005 ex post evaluation of a sample of projects co-financed by the Cohesion Fund in the period 1993-2000, ECORYS Transport examined (i) water supply in terms of the cost per inhabitant and (ii) water networks in terms of cost per metre.34 Similarly, wastewater treatment was measured in terms of (i) cost per cubic metre or cost per inhabitant and (ii) in terms of cost per metre for networks. Solid waste management was measured in terms of (i) cost per cubic metre or (ii) cost per tonne.

3.2.17 The English and Welsh water services regulator Ofwat has rigorously sought effective and accurate methods of predicting future capital investment costs in the water sector, which are vital in its task of ensuring that consumer prices remain at an acceptable level.

3.2.18 Ofwat conducts five-yearly reviews of water companies’ business plans, as part of which each company must submit information on the typical costs they incur per area of operation or project type. The result is an Ofwat cost base against which the relative efficiency of programme delivery and projected levels of capital expenditure are evaluated. It is, therefore, in its own right a good example of a current benchmarking application based on unit costing.

3.2.19 Unit cost definitions and related issues receive a less than extensive treatment in the energy sector. However, it seems sensible to measure the construction cost of electricity plants against their total productive capacity, which would enable cost comparisons between different production technologies and sizes of plant. The UK Energy White Paper (2003) provides an indication of the extent to which the construction costs of different electricity producing technologies vary.35

3.2.20 We have used these lessons to guide our approach to the disaggregation of the physical components of the projects in the WP10 sample. That approach is outlined in section 5.3 below. Furthermore, the unit cost indicators that we have developed based on the lessons learned from this and the previous subsections (and based on the available data) is outlined in section 5.4 below.

Achieving comparability

3.2.21 Once the feasible set of unit cost definitions has been determined, it is necessary to consider what other measures are required to normalise the data and achieve comparability. This is particularly the case in the current context where: (i) projects from a range of different countries are being compared; and (ii) the projects being compared have started and finished at different times.

3.2.22 Where projects from a range of different countries are being compared, normalisation is required to achieve denomination of unit cost benchmarks in a common currency through the use of appropriate exchange rates. This may not, as outlined in Section 5.8 below, be an issue in the current circumstances as most of the countries from which projects were chosen for the sample had already adopted the single European currency. However, expression in a common currency may not fully take into account differences in the purchasing power of a unit of the common currency between Member States. For this reason, the literature suggests Purchasing Power Parity

34 The Study for DG REGIO employed a sample of 200 projects that were co-financed by the Cohesion Fund in the period 1993-2002, of which 60 transport and environmental projects from CF4 (Greece, Ireland, Portugal and Spain) were selected for detailed evaluation. See ECORYS Transport (2005), ‘Ex post evaluation of a sample of projects co-financed by the Cohesion Fund (1993-2002)’, Report to the European Commission, DG Regio, Brussels. 35 See UK Department for Trade and Industry (now BERR) (2003), ‘Energy White Paper: Our energy future – creating a low carbon economy’, The Stationary Office, London.

54

(PPP) adjustments but, as also argued in Section 5.8, we believe such adjustments might not be appropriate in the current context.

3.2.23 Where projects that started and finished at different times are being compared, normalisation is required to achieve denomination of unit cost benchmarks in a common price level, which, in turn, requires use of the appropriate sectoral and geographical price indices.

3.2.24 There are many examples of data harmonisation techniques in the literature. For example, Stalder (2000) undertook the following steps: (i) harmonisation for a common currency36; (ii) harmonisation to a common price level; and (iii) compensation for general wage and price differences which were not reflected in the exchange rate, involving the weighting of cost data with OECD purchasing-power parity indices for 1998.

3.2.25 Flyvbjerg et al (2008) undertook three steps to sanitise their data before comparisons were undertaken, as follows: (i) costs were compared for similar systems i.e. urban rail, European projects and at least partly underground projects, to ensure homogeneity in the data; (ii) costs were expressed in constant (real) prices, using construction cost indices to discount costs to the same level (year)37; and (iii) costs calculated in different currencies were converted into the same currency, typically US$ or €, by applying the appropriate exchange rates.38

3.2.26 Another important element for WP10 is the identification and assessment of the reasons for cost overruns (which occur when project cost outturns exceed the corresponding cost estimates). The magnitude of observed cost overruns will depend on the unit cost definitions and calculation methods adopted, as well as on the quality and comparability of cost estimates and outturns. Consistency between them is essential to ensure that differences can legitimately be called cost overruns.

3.2.27 The practice in the literature is to make a decision about the specific point in the project life-cycle for which cost estimates should be drawn for all projects in the sample being analysed. There is a variety of views expressed in the literature about this. For example, Merrow (1988) chose a ‘commencement of detailed engineering’ milestone39, while Flyvbjerg et al (2003) noted that it should be possible to identify a specific point in a given project schedule when the formal decision to build was made and that a cost estimate is usually available at this time.40 As observed by various authors, the later the milestone chosen for cost estimates is in the project life-cycle, the more conservative the estimates of cost overruns will be.

3.2.28 In assessing differences in the levels of cost overruns within and between sectors and countries, the milestone itself is arguably less important than the need to choose a common milestone across all projects. The cost estimates used in the current study and presented later in this report were taken from the individual ERDF project application forms. There is a question over whether these application forms represent: (i) an appropriate milestone in the project life-cycle from which to draw cost estimates; and (ii) a common milestone for all projects. While the ERDF applications generally state the year in which the estimates were made, it was not possible to determine the stage in the project life-cycle which that year represented. If the forms represent different milestones for different projects, then smaller observed cost overruns might simply be due simply to more robust forecasting at later stages in the project life-cycle.

36 See footnote 30. 37 The construction cost indices mentioned in the second step above were from OECD (1997), Construction Price Index - Sources and Methods. The base year and currency used in the study were US$ 2002 prices. 38 See footnote 32. 39 See Merrow, Edward W. (1988), ‘Understanding the Outcomes of Megaprojects: A Quantitative Analysis of Very Large Civilian Projects’, Prepared for the private sector sponsors program, the Rand Corporation, California. 40 See footnote 25.

55

3.2.29 Other important technical observations relating to the definition and measurement of unit costs from the literature are:

• That it is standard industry practice to include in project cost estimates escalation factors that inflate estimated costs to the forecast midpoint of construction.

• That costs expended in each year of the project should be converted to a common price base.

3.2.30 On the first, while it appears that this is the standard practice in tendering for construction contracts, it does not appear to be the case in the preparation of ERDF applications. Cost estimates in the applications appear always to be stated in terms of prevailing current prices. On the second, where the stream of expenditures across the project life-cycle is not provided, Merrow (1988) suggested the use of ‘an empirically derived cosine curve’ to spread expenditures across the duration of the project.41

3.2.31 Section 5.8 below provides an analysis of the methods for achieving data comparability that we considered to be appropriate for the WP10 Study. The discussion is focused around the need for purchasing power parity (PPP) adjustments, but we also set out our approach to normalizing for the time value of money.

3.3 The determinants of infrastructure project costs

Introduction

3.3.1 The European Commission itself noted that no two infrastructure projects will cost the same no matter how similar they are. This sentiment was aptly reflected in the following statement from the 1998 guidance provide by DGXVI (now DG REGIO):42

‘Apart from basic technical factors, the wide range of economic and institutional conditions in different Member States will itself always lead to variations. Nevertheless, the fundamental project costs are based on the actual cost of the land, materials, equipment and labour in the region where the project is being procured. These basic costs will vary depending upon a number of factors…’

3.3.2 This part of the Report examines the determinants of infrastructure project costs and analyses the reasons for differences between projects within each of the sectors being examined, both within and between countries. There are several categories of project cost determinants (and, therefore, of efficiency differences between projects). These categories are examined in the following subsections of this part of the report.

3.3.3 Before proceeding, however, it is useful to consider the Commission’s own thinking on the issue. DG XVI’s guidance provides a framework for the analysis of the key determinants of infrastructure project costs. The diagrammatic form of that framework is re-presented in Figure 3 below.

41 See footnote 39. 42 See footnote 24.

56

Figure 3: DG XVI’s key determinants of infrastructure project costs

Source: European Commission DG XVI (1998), ‘Understanding and Monitoring the Cost-Determining Factors of Infrastructure Projects’, A User’s Guide, Brussels.

Technical characteristics of projects and their impact on costs

3.3.4 It is clear that the more complex a project is, the more it is likely to cost. Project complexity is determined by the set of technical characteristics that define the project. We analyse a range of technical characteristics in the following subsections.

3.3.5 Project specification defines the physical characteristics of the project. For example, with road projects, projections of future traffic will be used to derive a specification of the required length, depth and width of the road pavement, the material to be used in surfacing, the number of carriageways, lanes, bridges, junctions etc. In general, the larger the project, the more detailed the specification needs, and the more expensive the project is likely, to be.

3.3.6 Expectations should, however, be balanced with a consideration of the influence of economies of scale, that is, whether, for example, a 20 km road costs less on a unit cost per km basis than a 10km road. This would depend on the extent to which costs are fixed regardless of the length of the road. Likewise, it is relevant to consider whether cost-saving advances in technology have been employed in construction / rehabilitation in some projects, but not in others.

3.3.7 It is important to distinguish between new capacity, renewals and maintenance also because new build is normally more expensive than improvements to existing infrastructure.43 Some projects might involve a mixture of the two which could distort comparisons with dedicated new build projects or dedicated renewals/maintenance projects.44

3.3.8 For rail and urban transport, similar technical characteristics will be relevant but will include things like the number of stations, station spacing and the type of rolling stock

43 This is partly due to the fact that ‘non-building costs’ (such as land purchase, foundations, services provision etc.) do not feature when upgrading existing structures. However, costs borne in the implementation phases of these projects can also be expected to be lower (due to the need for fewer inputs). 44 This was observed for at least two ERDF projects in the WP10 sample.

57

to be employed. Another important issue might be vertical positioning, that is, whether track and stations are at ground-level, under ground or elevated. For example, there are significant costs in the tunneling and the diversion of utilities required to support an underground metro system.

3.3.9 For road, rail and urban transport projects, site characteristics, such as soil and drainage conditions or difficulties in accessing the site can be expected to influence the costs of remedial work before the ‘build’ stage commences. If there is uncertainty about soil and drainage conditions, accurate project costing requires soil surveys.

3.3.10 For road and rail projects, the distinction between urban and rural locations and between flat and mountainous terrain will also be relevant. Climate and weather conditions can also have an impact on design standards and on the standard and cost of materials required to meet those standards.

3.3.11 Technical issues found to be important in driving cost differences between urban transport projects included whether the project involved the use of: (i) Earth Pressure Boring Machines (EPBMs); (ii) strong geotechnical supervision monitoring; (iii) standardised station design concepts; (iv) a phased approach to power supply; (v) appropriate signalling and communications technology; and (vi) whether steel-wheel was chosen instead of rubber-tyre superstructures. The interface with existing lines (including the number of connection points) was also found to be relevant.45

Economic determinants of project costs

3.3.12 This category of determinants of infrastructure project costs incorporates inflation, location effects, financial characteristics of contractual arrangements and the competitiveness of the procurement process.

3.3.13 The effects of inflation occur when slippage occurs in timescales. Generally, the longer a project takes, the greater the project costs will be. Project timelines are dependent on the specification and the larger a project is the longer it is likely to take. The longer the expected construction period, the more account will need to be taken of expected inflationary input price increases over time. Initial cost estimates need to allow for the amount expected to be paid at the time the project goes ahead.46 Higher inflation rates would be expected in the accession countries.

3.3.14 There are also economic effects associated with location, which has an impact on construction and materials costs due to varying distances from suppliers. Moreover, land costs and design standards (and costs of meeting them), all of which can vary widely across countries (even within the EU).

3.3.15 Procurement and contracting arrangements can alter the estimated cost of projects. The greater the number of allowed tenderers in the procurement process, the more competitive the tendering process is likely to be. Project costs can, therefore, also be indirectly influenced by the effectiveness of national and EU competition authorities in detecting cartel behaviour. For example, the Hungarian competition authority (Gazdasági Versenyhivatal, GVH), as recently as 2007, upheld in the Hungarian Court of Appeal the fines it had imposed on five undertakings for rigging their bids for motorway construction contracts in 2002.47

45 See footnote 33. 46 We note the standard practice of including in project cost estimates escalation factors that inflate the estimates to the forecast midpoint of construction referred to in paragraph 3.2.29 above. 47 See Hetényi, Kinga and Franz Urlesberger (2008), ‘Hungary: The amendment is pending’, available from the International Financial Law Review website or directly from http://www.iflr.com/Article/2025674/Hungary-The-amendment-is-pending.html .

58

3.3.16 A World Bank working paper on the procurement efficiency for infrastructure discusses competitive bidding in relation to economies of scale.48 While it is generally the case that larger infrastructure projects will lead to savings in unit costs, their scope will also limit the number of contractors who are able to make a bid. Only a limited number of companies will have the necessary resources to complete projects of such extensive scale.

3.3.17 The paper provides statistical evidence suggesting that particularly transport and water projects in particular benefit from increased competition. This seems to remain true at least up until 7 bidders, after which the effect wanes. The authors found that the electricity sector is, however, less dependent on competitive bidding for affordable offers. Nevertheless, the analysis demonstrates the apparent existence of a trade-off between economies of scale and competitive bidding practices.

3.3.18 The policy on contracting has been shown to lead to cost savings through, for example, lump sum contracts (fixed or target), but these savings tended to be marginal relative to total project costs. Projects that used traditional procurement methods were, however, characterised by attractive, low price contractor bids, which turned out to be significantly underestimated over the course of the building phase of the project. This was in contrast to Design-Build-Finance-Operate (DBFO) methods. DBFO contracts seek to transfer the risk of cost overruns to the contractor, which may also result in savings. Other types of contract include progressive payment (according to tasks completed or according to human resources expended) or re-measure.49

3.3.19 The use of early contractor involvement (where the final contract cost is only laid out once a detailed design has been completed) was advocated as potentially useful in diminishing cost risk at the planning stage.

3.3.20 Likewise, contracting arrangements might involve the use of sophisticated systems of incentives for contractors to deliver on time and on budget. In some cases, even if contracts are awarded to the lowest bidder, the absence of contractual arrangements that fairly and effectively distribute risk between contractors and contracting authorities can result in significant unexpected increases in costs.

3.3.21 Other economic and financial determinants of costs include labour and materials costs, the distance from and difficulty in getting to suppliers and the competitiveness of markets like, for example, those for the supply and/or rental of plant and machinery. The independent experts that were hired to advise the WP10 project team also suggested a number of other factors, including the length of the tender period and the type of contractual dispute resolution procedures, for example, arbitration, dispute boards, expert determination or adjudication.

Institutional determinants of project costs and the stage of country development

3.3.22 Institutional arrangements refer to the allocation of responsibility and risk and the system of rewarding success and penalising failure through performance monitoring and accountability processes. Institutional arrangements reflected in the stage of country development can have a significant influence on project costs.

3.3.23 Indicators of institutional arrangements include the perceived level of political commitment, the strength and quality of leadership, design standards and the approach to procurement decisions. For example, contract procurement decisions

48 See Estache, Antonio and Atsushi Iimi (2008), ‘Procurement Efficiency for Infrastructure Development and Financial Needs Reassessed’, Policy Research Working Paper 4662, World Bank, Washington DC. 49 See UK National Audit Office (2007), ‘Estimating and Monitoring the Cost of Building Roads in England’, UK Stationary Office.

59

might be based on a sound rationale related to technical/experience reasons, or they might simply be based on the cheapest bid.

3.3.24 Another determinant of infrastructure project costs that has received some attention is the ownership structure and, more specifically, whether private finance (through, for example, public-private partnerships) or public finance can be expected to lead to more efficient outcomes. We described the conflicting evidence that emerged from our examination of this issue in our First Interim Report (Sn. 3.4). While public-private partnerships performed better in an Australian study (Allen Consulting Group, 2007), several reasons for a worse performance were identified in the European road sector (Blanc-Brude et al).50

3.3.25 The Australian study compared public-private partnerships with traditionally procured infrastructure projects and found that they perform better in terms of both cost and time efficiency than traditionally procured projects.51 Moreover, the timeliness of projects (under traditional procurement) was shown to suffer as the project size increased, whereas the timeliness of public-private partnership projects was found not to be affected by project size. The study also observed that public-private partnership projects were more transparent than traditionally procured projects, as measured by the availability of public data for the study.

3.3.26 The European study suggested a number of reasons why public-private partnerships could be expected to exhibit higher costs than traditionally procured infrastructure projects.52 These reasons were: (i) the fact that the bundling of construction and operations contracts in public-private partnerships gave the private partner incentives to make investments in the construction phase that could lower subsequent operation and maintenance costs; and (ii) the fact that the transfer of construction risk to the private partner can be expected to be explicitly priced in a public-private partnership contract.

3.3.27 Flyvbjerg et al (2003) suggested that the expectation that privately financed projects perform better than traditional publicly procured projects is an oversimplification; rather that the type of accountability structures mattered more in his sample than the type of ownership.53 These sentiments were echoed in Flyvbjerg et al (2008), which concluded that the role of ownership in causing efficiency differences between projects involving private and public finance required further research.54

3.3.28 It might also be relevant to consider whether projects involve any element of external borrowing. The use of external debt to finance projects could, in principle, impose a hard budget constraint on project managers in order to facilitate speedier debt repayments.

3.3.29 Wiser and Kahn (1996) examined the importance of the capital structure and financing arrangements for energy infrastructure (specifically generating plant).55 The rate of

50 See Allen Consulting Group (2007), ‘Performance of PPPs and traditional procurement in Australia’, Report to Infrastructure Partnerships Australia. See footnote 23 for the full Blanc-Brude reference. 51 The findings of the study were based on the publicly available data for a sample of 21 public-private partnership projects and 33 traditionally procured projects from New South Wales (19), Victoria (26) and Queensland (9). Projects were grouped into social (24), transport (23), water (3) and IT (4) infrastructure sectors. The projects in the sample were selected according to five criteria (including data availability). 52 The authors use data on ex ante construction costs (the best estimate of what the project should have cost to build at the point at which the winning bidder was awarded the contract for the project) of road projects in the EU-15 countries plus Norway financed by the European Investment Bank between 1990 and 2005 and covering 6,400 kilometres of road. They found public-private partnership roads to be, on average, 24 per cent more expensive than traditionally procured roads, all other things being equal. 53 See footnote 25. 54 See footnote 32. 55 See Wiser, Ryan and Edward Kahn (1996), ‘Alternative Windpower Ownership Structures: Financing Terms and Project Costs’, University of California, Berkeley.

60

interest on debt as well as the debt-equity ratio was found to impact on how capital costs are to be recovered. The authors state that:

‘the primary benefits associated with public ownership and finance come from the lack of project-specific minimum debt service coverage ratio (DSCR) requirements, which allows an increased level of debt in the capital structure, reduced debt costs, and an increased debt amortization period. The benefits of IOU ownership and finance come primarily from debt and equity cost reductions, longer debt amortization, and the lack of project-specific DSCR requirements’. 56

3.3.30 The same study observed that financing terms are particularly influential on the cost of wind power in the US because the associated risks are seen to be relatively high due to the early development stage of the sector.

3.3.31 Finally, institutional factors, such as difficulties in obtaining consents and requirements for public consultation, can also have a significant impact on the costs of projects. For example, where major projects are likely to be strongly opposed on environmental grounds, more cost may have to be allowed for consultation and environmental mitigation measures.

Quality of project planning and management

3.3.32 Project (and programme) governance arrangements that are of a poor standard can result in poor performance on the project. Poor quality project planning and management might, in turn, be due to poor quality people, little direct presence of top management/officials in the field or simply to inexperience with all or certain types of infrastructure project. For example, small European countries are beginning to deliver urban rail projects for the very first time.

3.3.33 Poor project preparation and management cause delays due to a range of factors including, for example, satisfying the requirements of the planning process and meeting environmental requirements. This will most likely result in rising costs due to the effects of input price inflation.

3.3.34 Procurement that involves early contractor involvement in the design phase has the potential to provide valuable information like, for example, early indications of cost problems, a basis for monitoring contractors’ costs as projects develop and potential areas that require value management/engineering.57, 58

3.3.35 Effective project management might be expected to include the employment of substantial additional resources in order to accelerate project implementation. However, if the phasing of a project is dependent on other linked projects (that are part of the same programme or part of a larger project), thereby leading to interruptions, the project can be more expensive due to the cost of re-mobilising plant and contractors.

56 The authors distinguish between four types of ownership structure: (i) Private Ownership, Project-Finance: private renewable energy companies that develop and finance the project and sell electricity on to utility companies; (ii) IOU (Investor-Owned electric Utility) Ownership involving corporate financing arrangements: (iii) Investor Owned Utility companies that develop and finance the project themselves; and (iv) Public Utility Ownership involving internal financing arrangements. Project finance differs from internal finance in that debt is claimed against revenues from the project, rather than from the revenues of the firm. 57 Value engineering or management refers to systematic methods to improve ‘value’ by examining costs relative to functionality. Value can be increased by improving functionality or reducing costs. In most cases, it will involve eliminating unnecessary expenditures. Alternatively, it might involve increases in initial capital expenditure in order to decrease, for example, longer term maintenance expenditures through a whole life costing approach. 58 See Nichols (2007) and Flyvbjerg (2008). The full references are contained in footnotes 28 and 32, respectively. The matter will also be considered by ARUP, as part of the Project Cost Forecasting and Management study being undertaken for JASPERS (Joint Assistance to Support Projects in European Regions, part of the European Investment Bank). See http://www.jaspers.europa.eu/ .

61

Conclusions

3.3.36 We used these insights on the determinants of infrastructure project costs to develop lists of general and sector-specific project attributes on which to seek information about the WP10 projects. These are set out in section 0.1 below. Section 5 presents the results of our analysis of the data provided for those projects.

3.4 Understanding time delays and cost overruns

Introduction

3.4.1 Cost overruns and delays occur because of inaccurate estimates leading to higher actual costs, and longer actual timescales, than anticipated. In the literature, we found a systematic tendency for large infrastructure projects to have overruns and delays.

3.4.2 Notably, Flyvbjerg et al (2003) found that 9 out of 10 projects, from a sample of 258, across 20 countries and 5 continents, were subject to cost overruns.59 Following Flyvbjerg (2005), we have divided explanations for time delays and cost overruns into technical, psychological and political-economic factors. We examine these in turn in the following subsections.60

Technical explanations

3.4.3 Technical explanations for delays and overruns are extensively examined in previous evaluations. They include imperfect forecasting, inadequate data and honest mistakes due, for example, to a lack of experience with infrastructure cost forecasting or with forecasting costs for certain types of infrastructure. They might also include inherent problems in predicting the future.

3.4.4 Technical explanations for cost overruns/delays in transport infrastructure are also the subject of a growing academic literature inspired by Flyvbjerg et al (2004) and continuing with Flyvbjerg (2008), the latter looking specifically at urban transport projects.61 Conventional wisdom suggests that the same principles apply to projects in other sectors, including the energy and water/wastewater sectors, so the lessons drawn from the available literature have wide applicability for WP10.

3.4.5 DG REGIO’s Guide provides a set of key determinants of delays and cost overruns. These are re-presented from the Guide in Figure 4 below.

59 See footnote 25. 60 See Flyvbjerg, Bent (2005), ‘Policy and Planning for Large Infrastructure Projects: Problems, Causes, Cures’, World Bank Policy Research Working Paper 3781, World Bank, Washington D.C. 61 See Flyvbjerg, Bent et al (2004), ‘What Causes Cost Overrun in Transport Infrastructure Projects?’ Transport Reviews, Volume 24, No. 1, January. For Flyvbjerg et al (2008), see footnote 32.

62

Figure 4: Key determinants of infrastructure project delays and cost overruns


3.4.6 The Commission identified a lack of effective project management as the main cause of cost overruns. Other important factors were design changes, and input price inflation due to delays. Less important factors were land acquisition problems, unexpected ground conditions and difficulties with contractors. The Guide’s summary of the relative importance of each of the factors for different categories of project cost is re-presented from the Guide in Figure 5 below. A large circle indicates a major effect and a small circle indicates a minor effect.

Figure 5: Common causes of cost overruns and project time delays


63

3.4.7 ECORYS Transport’s (2005) study of 60 projects co-financed by the Cohesion Fund, in the period 1993-2000, observed similar factors to the DG REGIO Guide in explaining cost overruns.62 Their report notes that the predominant causes of cost overruns, namely design changes, inflation impacts due to time delays and site contractor issues, all point to inadequate project preparation. The report also considers local community involvement in projects to be important.

3.4.8 The ‘Ex Post Evaluation of Objective 1 1994-1999’ by ECOTEC (2003) compares water infrastructure projects with similar large-scale transport projects. They use data from France, Germany, Greece, Ireland, Italy, Portugal, Spain and the UK. They found that water projects were more susceptible to delay and cost overrun – 70% of water projects go over budget, compared to 60% for transport projects. Furthermore, water projects were found to be subject to delays of in excess of 12 months, greater than the delays for the transport projects.

3.4.9 The Nichols Report (2007) on 13 of the UK Highways Agency’s largest road projects found that inflation, inaccurate cost estimation and inadequate project definition were the most important causes of cost escalation.63 While the results are not entirely representative, however, because only the 13 largest schemes were examined, they do provide a good indication of the variety of aspects that need to be accounted for to get more accurate cost estimates.

3.4.10 The UK National Audit Office (NAO, 2007) reported on a more extensive sample of road projects under the Targeted Programme for Improvement (TPI) and Local Transport Plan schemes.64 The study supports the findings of the Commission, namely that poor project management, design changes and inflation are the most significant causes of cost overruns.

3.4.11 Flyvbjerg et al (2004) conducted a statistical analysis of cost overruns and found that they could be associated with the length of the project implementation phase. They also found that the influence was not significantly different for rail, fixed-link (bridges and tunnels) or road projects. The authors also found project size to be significant for bridges and tunnels, but not for road or rail projects. Weak accountability measures for delays and overruns were also a significant influence and probably more significant than the effect of ownership incentives.

3.4.12 Bordat et al (2004) made the distinction between excusable delays (due to force majeure) and non-excusable delays (attributable to contracting agencies and contractors, and therefore preventable).65, 66 Contracting agency errors included planning and design deficiencies such as incorrect estimates of work quantities. Contractor errors included unnecessary work, work that did not follow the design plans,

62 See footnote 34. 63 See footnote 28. This review was commissioned by the UK Secretary of State for Transport and was prompted by a series of increases in cost estimates for individual road schemes (by up to 300 per cent) since entry into the Targeted Programme for Improvement (TPI) and in the TPI as a whole (over 18 per cent) over a period of 15 months. 64 See footnote 49. The study views two sets of English project groups: (i) the Targeted Programme for Improvement (TPI), managed on the national level by the Highways Agency. The TPI consists of 103 ventures set to develop or build new trunk roads and motorways, 36 of which had been finalised by September 2006. The TPI projects run in the period 1998-2021; and (ii) Local Transport Plan road schemes, managed by the Local Authorities. These 81 projects, of which 20 had been completed by July 2006, are part of a general objective to improve regional transportation links. 65 See Bordat, C. et al (2004), ‘An Analysis of Cost Overruns and Time Delays of INDOT Projects’, Joint Transportation Research Programme, Indiana Department of Transportation. 66 This study analysed the extent of and reasons for cost overruns, time delays and change orders in Indiana Department of Transport (INDOT) projects completed in the period 1996-2001. They include road and bridge construction and rehabilitation projects, maintenance projects (with road maintenance and resurfacing contracts), and traffic and traffic maintenance contracts. Detailed weather data over the relevant period and location were also considered, along with data on the frequency and extent of cost overruns from 11 other US State departments in order to assess Indiana’s relative performance.

64

and work that did not meet contract specifications. Excusable delays due to unforeseen circumstances included site conditions that differed from those described in contract documents. Majid and McCaffer (1998), upon whose work Bordat et al (2004) is built, identified 12 main causes of project time delays. They pinpoint materials, equipment, and labour-related delays as the major cause of contractors’ performance delays.

Psychological explanations

3.4.13 Psychological explanations derive from the ‘planning fallacy’, the systematic tendency to be overly optimistic about the outcomes of planned actions. In infrastructure project planning, the planning fallacy leads to ‘optimism bias’, that is, the systematic tendency to underestimate costs, completion times and risks and to overestimate the benefits of planned actions.

3.4.14 This category of explanation was inspired by the work of Kahneman (1979, 2003) and Kahneman and Lovallo (1993) inspired this type of explanation for overruns and delays.67 They relate the planning fallacy to biases at the cognitive level, i.e. ‘errors’ in the processing of information that result from ‘delusional optimism’. By examining the technical and political-economic explanations, one can, according to the authors, better understand how to resolve or, rather, correct for optimism bias.

3.4.15 Kahneman et al’s work on decision-making under uncertainty provided the foundation for the concept of Reference Class Forecasting, which is based on the presumption that past projects tend to be more similar to planned projects than normally assumed and are, therefore, a means of increasing the accuracy of forecasting. Rather than focusing only on the specific constituents of the planned projects (the ‘inside view’), there should be equal if not more focus on the outcomes of similar projects that have already been completed (an ‘outside view’).

3.4.16 Flyvbjerg and Cowi (2004) developed an applied method of Reference Class Forecasting for the UK Treasury.68 Their report advocates the use of explicit, empirically based optimism bias uplifts to produce more realistic forecasts of individual project capital expenditures. The study generated benchmark uplifts by developing probability (or frequency) distributions of cost overruns for the various classes of project under consideration.

3.4.17 Planners and promoters can choose the uplift that corresponds with the level of risk of cost overrun that they are willing to accept. So, the authors state, they should adopt the 50th percentile uplift if they were willing to accept a high risk of cost overrun and the 80th percentile if such a level of risk was unacceptable. (Note that the percentiles referred to here can be thought of as 1 minus the level of risk that the project planner is willing to accept. Therefore, the 80th percentile corresponds with an acceptable risk of

67 See Kahneman, D. (1973), Attention and effort, Englewood Cliffs, NJ: Prentice-Hall. See also Kahneman, D. (2003), ‘Maps of bounded rationality: A perspective on intuitive judgment and choice’, in T. Frangsmyr (Ed.), Les Prix Nobel 2002 [Nobel Prizes 2002]. Stockholm, Sweden: Almquist & Wiksell International. See also Kahneman, D. and D. Lovallo (1993) ‘Timid choices and bold forecasts: A cognitive perspective on risk-taking’, Management Science, 39, 17-31. 68 See footnote 29. The study explored the underlying causes and institutional context for optimism bias in British transport projects and some possibilities for reducing optimism bias in project preparation and decision-making were identified. From the Flyvbjerg database of 258 projects completed between 1927 and 1998, the following were used in the study: (i) 128 UK and 44 non-UK (Denmark, Sweden and the US) trunk road and motorway projects; (ii) 46 rail projects (including urban rail, conventional inter-city rail and high-speed rail), 3 of which are from the UK, and the others from Canada, France, Germany, the Netherlands, Norway, Sweden, and the US; and (iii) 34 fixed link (bridges and tunnels) projects, of which 4 are from the UK, and the others in Denmark, France, Germany, and the US. Data for building and IT projects were taken from the Mott MacDonald July 2002 study for HM Treasury, Review of Large Public Procurement in the UK.

65

cost overrun of 20 per cent. The lower the level of risk of cost overrun that is acceptable, the higher the optimism bias uplift needs to be.)

3.4.18 The authors found that, for road projects, optimism uplifts range from 15% at the 50th percentile to 32% at the 80th percentile. For rail projects (including urban transport), the uplifts range from 40% at the 50th percentile to 57% at the 80th percentile. Fixed link uplifts range from 23% at the 50th percentile to 55% at the 80th percentile.

3.4.19 The report also asserts that few project planners have a direct interest in avoiding cost overruns. As a result, they recommended that the UK Department for Transport apply optimism bias uplifts that are supported by:69

• establishing realistic budgeting as an ideal, and de-legitimising over-optimistic budgeting as a routine;

• the introduction of fiscal incentives to avoid cost overruns e.g. through requiring local co-financing of project cost escalation where possible;

• formalised requirements for high quality cost and risk assessment at the business case stage; and

• The introduction of independent appraisal.

Political economy explanations

3.4.20 This is most recent category of explanation for overruns and delays and addresses the possibility that delays and cost overruns may be the result of deliberate misinformation. In other words, that project planners and promoters misrepresent timescales and cost/benefit projections in order to win favour for the project and get it started. Important work in this area includes Flyvbjerg et al (2004), which deal with the cost side, and Flyvbjerg et al (2005), which deals with the demand side (demand being the factor driving expectations about future benefits).70

3.4.21 Flyvbjerg et al (2003), in their study of 258 transport infrastructure projects completed between 1927 and 1998, observed that ‘no learning appears to be taking place’ and that cost underestimation and subsequent cost overruns ‘are allowed to continue unchecked decade after decade’.

3.4.22 The authors state that we may be in an equilibrium in which ‘strong incentives and weak disincentives for cost underestimation and related escalation may have taught project promoters that cost underestimation pays off.’ Flyvbjerg et al (2004) tested this explanation of cost overruns and, as reported in Flyvbjerg et al (2003), found that cost underestimation is ‘used strategically to make projects appear less expensive than they really are in order to gain approval from decision-makers to build projects’.

Measuring cost overruns and delays in practice

3.4.23 Procuring data on cost escalation in infrastructure projects is, as pointed out in the literature, not without its difficulties. For instance, Flyvbjerg et al (2003) claim that it is very time-consuming and often impossible because:71

• funding and accounting procedures for public sector projects are typically unfit for keeping track of multiple and complex changes that occur in total costs as the projects develop;

69 The method is now a requirement for all large transport infrastructure projects seeking government funding in the UK and has also been endorsed by the American Planning Association. 70 See footnotes 60 and 61 for the full paper references. 71 See footnote 25.

66

• data for private projects are often classified to keep them from the hands of competitors; and

• Data may, in general, be held back by project owners because cost overruns are normally considered somewhat of an embarrassment.

3.4.24 Moreover, Flyvbjerg et al (2003) argued that calculations of cost overruns are likely to be conservative for a number of reasons:

• projects that are well-managed with respect to data availability may also be managed well in other respects, resulting in better-than-average (i.e., non-representative) performance;

• the very existence of data that facilitates performance evaluations may contribute to improved performance when such data are used by project management to monitor projects; and

• Project managers may have ‘wiggle-room’ to choose and give out data that present their project in a favourable light, for example, by choosing forecast costs that best fit actual costs from the several forecasts that were done.

3.4.25 Ideally, original project files should be acquired, but where this is not possible (and it often isn’t), the second-best methodology of surveys must be used according to Flyvbjerg et al. (2003). The Allen Consulting Group (2007) claimed, on the other hand, that by only using publicly available data they avoided the ‘criticism that data had been modified because it originated from a source with an interest in either of the procurement methods (public-private partnership or traditional) under question.’72

3.4.26 In Flyvbjerg and COWI (2004), projects for the sample were selected on the basis of data availability and quality, which may, according to the authors, almost certainly have biased the data for several reasons, the most important of which were that:

• managers of projects with particularly bad track records regarding cost escalation have an interest in not making cost data available (and conversely, managers of good projects may have an interest in making this public); and

• Even where managers have made cost data available, they may have chosen to give out data that present their projects in as favourable a light as possible (e.g., by choosing the forecast of costs that best fits the actual costs from the several forecasts that were done).

3.4.27 The approach adopted to the measurement of delays and cost overruns for the WP10 Study is outlined in sections 0 and 5.7 below.

3.5 Defining and measuring ‘cost per job created’

Introduction

3.5.1 WP10 is also concerned with the ‘result’ efficiency of productive investment projects. Productive investments are defined loosely as projects that involve direct support to enterprises. ERDF provides investment aid to private enterprises with the primary aim of improving competitiveness through various instruments such as, for example R&D, technological transfer, and investment in more environmentally friendly technologies.

72 See footnote 50.

67

3.5.2 When the ERDF was established, Article 4(1)(a) of Regulation 724/75 envisaged, according to Evans (1999), ERDF support for ‘investments in industry, handicraft, or service activities which were economically sound and benefited from state regional aid.’73, 74 The minimum requirement, at that time (1975), was for 10 new jobs to be created or existing jobs to be maintained and, in the latter case, ‘investments had to fall within the framework of a conversion or restructuring plan to ensure that the undertaking concerned became competitive.’75

3.5.3 Evans (1999) outlined some of the difficulties associated with this type of assistance. Knowledge of these difficulties is, we found, useful in enhancing the general understanding of productive investments. They include, as outlined by Evans, the fact that:

• applicants can make exaggerated claims about the job-creating potential of investments which might not, therefore, be realised;

• assistance can be used to reduce the workforce and / or finance restructurings that lead to redundancies, while ‘saving’ only the remaining jobs;

• job creation forecasts are difficult to make because of cyclical factors and delays;76 and

• Net jobs created by the investments might differ considerably from the gross figure because, by reinforcing the competitiveness of companies in receipt of assistance, job losses at other companies might result from implementation of the investment.

3.5.4 Evans (1999) also described the reasons why it proved difficult to ensure that ERDF assistance to productive investments was optimal. Productive investments could, by reducing the cost of capital for new investments, artificially boost their capital intensity, thus encouraging substitution away from labour towards capital. ‘This distortion of factor costs could lead to a misallocation of resources generating, for example, a capital-intensive pattern of development in regions with an abundance of labour.’

Measuring ‘cost per job created’

3.5.5 The ‘result’ efficiency of productive investments is measured according to the ‘cost per job created’ as a result of these investments.

3.5.6 The box in Figure 6 below is re-presented from DG REGIO’s Working Document No. 6.77 It sets out the reasoning for calculating costs per job created and the options (or methods) for calculating them, as well as some of important measurement issues to consider when doing so.

73 See Evans, Andrew (1999), The EU Structural Funds, Oxford University Press. 74 See Regulation (EEC) No. 724/75 of the Council of 18 March 1975 establishing a European Regional Development Fund, OJ 1975 L73/1. 75 However, the requirement for 10 jobs created or maintained was difficult for SMEs to meet and the requirement was, therefore, dropped in Regulation 1787/84 (OJ 1984 L169/1). 76 Evans (1999) refers to the 1989 Court of Auditors’ finding that job creation forecasts were fully achieved in less than half (48 per cent) the investments supported by ERDF. 77 See European Commission, DG Regio (2007), ‘Working Document No. 6: Measuring Structural Fund Employment Effects,’ Brussels, March.

68

Figure 6: Reasoning for measuring cost per job created

Source: DG REGIO (2007), ‘Working Document No. 6: Measuring Structural Funds Employment Effects’, Brussels, March.

3.5.7 Cost per job created is also a unit cost measure, analogous to those discussed for infrastructure projects above. However, for productive investments, it is the denominator of the unit cost measure that poses the greatest difficulty in measurement. Unlike infrastructure, all the focus is on the ‘outputs’ (jobs created) for a given level of investment.78

Definition and measurement of employment effects

3.5.8 The creation of a job through investment in a private enterprise is a form of employment effect. The accurate definition and measurement of employment effects poses, perhaps, the greatest challenge in evaluating, on both an individual and a comparative basis, the economic efficiency of productive investments. These issues are, however, beyond the scope of WP10.

3.5.9 The number of jobs created is one of four ‘core’ indicators of Structural Fund employment effects recommended in Annex I of Working Document No. 6. Jobs created are defined as new jobs created directly by Structural Fund intervention within three years of the completion of works.

3.5.10 Jobs created are alternatively labeled ‘green’ jobs and measuring these in terms of full-time equivalents (FTEs) should be a priority for ERDF programmes, also according to Working Document 6. It is important, however, to distinguish between jobs created and jobs maintained or ‘safeguarded’. The latter are jobs that are at risk and that would be lost without Structural Fund intervention.

3.5.11 Other important distinctions include:

• Permanent and temporary jobs. Temporary jobs last more than six person-months but end with the period of assistance, while permanent jobs must last for two years beyond the programming period (2000-2006 for current purposes) to be recorded as such. Distinctions between temporary and permanent jobs created were not,

78 For infrastructure projects, we were concerned with patterns in the levels of investment for given levels of output.

69

however, made by the respondents to our questionnaires on the productive investment projects in our WP10 sample.

• Direct and indirect effects. Direct employment effects are ‘outputs’ and should have a clear first-order causal relationship with the project, for example, a new manufacturing plant that provides employment for an additional 500 employees. Indirect effects are secondary (or ‘knock-on’) effects and are counted in the ‘results’ of a programme. An example is if the 500 employees of the new plant spend income in their local region, thereby boosting demand and encouraging firms to increase output and generate new jobs.79, 80 For WP10, we were only concerned with direct effects.

• Gross and net effects. The calculation of net employment effects is a complex area, which we touched upon briefly in paragraph 3.5.3 above. However, we refrain from exploring this issue any further here as it is also beyond the scope of WP10.

Previous evaluations of job creation programmes

3.5.12 The most relevant report on EU Structural Fund employment effects was prepared by CSES (2006).81 They summarised existing studies on cost per job created from enterprise investments, as summarised in Table 6 below.

Table 6: Cost per job created found by CSES (2006)

Type of intervention Average cost per job created

Range

All types €14,800 €7,000 to €23,700

Physical infrastructure (temporary construction jobs)

€74,000 €25,000 to €123,000

SME support measures

€19,320 €2,000 to €65,000

Average €36,000 n/a

3.5.13 We also reviewed evaluations of job creation programmes from Canada and the United States. The Canadian studies are over ten years old and may be outdated for our purposes. The main study from the US by Glasmeier (2002) could be useful in providing a (very limited) set of highly indicative cost per job estimates. However, in order to make true comparisons, it would be necessary to adjust for differing circumstances, not only for geographical but also time and the economic cycle.

79 This is also referred to as an income multiplier effect. 80 Please note that the references here to ‘outputs’ and ‘results’ correspond with the language used in the European Commission’s broader evaluation framework. See DG BUDG (2004), ‘Evaluating EU Activities: A practical guide for the Commission Services’. 81 CSES (2006), ‘Study on Measuring Employment Effects: Final Report’, Kent, England, June.

70

Achieving comparable ‘cost per job created’ data

3.5.14 Evans (1999) referred to a 1986 report, which set out observed wide disparities in the cost of creating new jobs, ranging from ECU 235,000 in Belgium to ECU 30,100 in Denmark for example.82 The finding of such wide ranges is consistent with our own findings for WP10.

3.5.15 However, we observed similarly wide variations even within the same country. As we conclude later in this Report, it was a lack of information on the different purposes of these projects and, more particularly, on the capital-labour ratios of the ERDF recipient firms that prevented us from achieving comparable ‘cost per job created’ numbers for the productive investments in the WP10 sample.

3.6 Summary and conclusions

3.6.1 The review of the literature revealed an extensive body of research into the causes of infrastructure cost overruns and project delays and of the tendencies for them to differ in magnitude and scale across countries and sectors. Our key findings were that:

• optimism bias and other difficulties in the estimation of project costs are widely acknowledged; and

• The lack of good quality project data is widespread, which makes it difficult to use benchmark data as a basis for forecasting.

3.6.2 For these reasons, robust cost estimation, project appraisal and evaluation will remain a critical area of responsibility for public agencies that are recipients of EU funds. In that respect, the availability of reliable and consistent benchmark data on project unit costs would represent a significant improvement.

3.6.3 The review of the rather limited literature on productive investments and measuring the costs of job creation revealed issues similar to those experienced in the course of the WP10 Study, namely (i) the wide ranges of observed ‘cost per job created’ data and (ii) the difficulties in establishing comparable data across projects due, in particular, to the lack of data on the recipient firms’ capital-labour ratios.

82 Evans (1999) refers, in this regard, to the Report of the Committee on Regional Policy and Regional Planning on the efficiency of national regional policy instruments – conclusions regarding new regional policy, EP Doc A2-66/86.

71

4 INFORMATION GATHERING AND WP10 METHODOLOGY

4.1 Introduction

4.1.1 We describe in this section our data gathering efforts, first in respect of our search for benchmark databases, second, in respect of the ex post evaluation activities in the individual Member States and, third, in respect of information about the projects in our WP10 sample.

4.2 The search for benchmark databases

Introduction

4.2.1 Cost benchmarks for various infrastructure projects have been compiled from a range of sources, including:

• The World Bank Road Cost Knowledge System (ROCKS) database.

• www.roadtraffic-technology.com – a website offering information on the road traffic industry.

• www.railway-technology.com – a website offering information on the rail industry.

• www.water-technology.net – a website offering information on water infrastructure facilities.

• www.power-technology.com – a website offering information on the power industry.

• www.wsdot.wa.gov – the Washington State Department of Transport website.

• www.lightrailnow.org – a website supporting efforts to develop light rail systems world wide.

• http://lrt.daxack.ca – The Toronto LRT information page.

• www.welfi.info – a wind energy local financing project website

• www.aitricity.com – the website of a company which develops and operates wind farms.

• www.vattenfall.com – a German energy company website.

• UIC – International Union of Railways

• Bent Flyvbjerg et al., Comparison of Capital Costs per Route-Kilometre in Urban Rail, 2008

• Transek Consultants, Comparison of Costs between Bus, PRT, LRT and Metro/rail, 2003

• Data provided by Faber Maunsell – Consultant Engineers

• Data provided by Prof. Bent Flyvbjerg – An expert advisor to this project.

4.2.2 However, our benchmark data should be used with caution for a number of reasons, as follows:

• Project cost definitions vary and it was not possible to be sure that consistent definitions were used (e.g. some projects include taxes, others do not).

• Source data is expressed in a number of different currencies and was incurred in different years. We have converted all costs to constant 2007 Euros using a range of different indices and exchange rates.

72

4.2.3 Unit construction costs do vary significantly with projects, reflecting a number of different factors. So whilst comparing one project’s costs against others can be a useful exercise, it cannot be relied on as a basis for testing the accuracy of a particular project cost estimate.

Existing benchmarks for infrastructure project costs

4.2.4 We have sought to identify and gather data from relevant cost benchmark databases, investigating a wide range of potential sources including government ministries, industry/market organisations and multi-lateral agencies. In most cases, however, where cost databases are maintained, they are for disaggregated input costs, such as concrete, steel or labour rates. These databases are used to prepare bottom-up cost estimates but do not provide any total or component project costs suitable for benchmarking for WP10.

4.2.5 The World Bank Road Costs Knowledge System (ROCKS) database is probably the only relatively comprehensive database available for any type of transport infrastructure. The main objective of the ROCKS system was to develop an international knowledge system on road work costs to establish an institutional memory, and obtain average and range unit costs based on historical data that could ultimately improve the reliability of new cost estimates and reduce the risks of cost overruns.

4.2.6 The database was created with data collected primarily from World Bank financed projects and has 2,043 records. The earliest record dates back to September 1984, and the most recent record is from June 2007. Sources of information include World Bank Implementation Completion Reports, project appraisal documents, civil works contracts, as well as from project supervision reports, pavement management information systems and procurement and disbursements reports. There are 837 records representing average costs for maintenance, rehabilitation, or improvement programs and 1,206 records for cost of works on individual sections, with a good distribution between the numbers of cost benchmarks derived from each of estimates (884), contracts (635) and actuals (524).

4.2.7 The existing database has data from 89 developing countries. For the majority of countries only a few lines of data are available, while for others such as Brazil, Chile, Russia, Poland, Ghana, Uganda, India, Thailand, Philippines and Bangladesh, there is a large set of data. Costs are classified by a cost date to capture the representative date of the expenditures.

4.2.8 The ROCKS framework separates road projects into two main categories, namely preservation and new development, and further classifies these categories by type of road work and by predominant work activities. Each of the classifications and corresponding set of work activities for preservation and new build were presented in Figures 2 and 3 of out First Interim Report.

4.2.9 Because this was the only meaningful database available for the purposes of WP10, we compiled a survey of individual project costs from a range of other sources. This exercise, whilst limited, has, for some cost categories, yielded useful ‘indicative benchmark ranges’ against which costs of the sample projects can usefully be compared.

Benchmarking results

4.2.10 The project costs included, along with the measurement units used, in this exercise include:

73

• Level 1 total road construction costs (‘all in’ cost measured in Euro million per kilometre)

• Level 2 road construction costs

o Bridges (construction cost measured in Euro per square metre)

o Tunnels (construction cost measured in Euro per square metre)

o Carriageways (construction cost measured in Euro million per kilometre)

• Level 1 total railway project costs (‘all in’ Euro cost per kilometre)

• Level 2 railway track construction costs (construction cost measured in Euro million per kilometre)

• Level 1 total metro/underground project costs (‘all in’ cost measured in Euro million per kilometre)

• Level 1 total light rail project costs (‘all in’ cost measured in Euro million per kilometre)

• Level 1 total wind farm project costs (‘all in’ cost measured in Euro per kilowatt)

• Level 1 total water treatment and supply project costs (‘all in’ cost measured in Euro per cubic metre per day).

4.2.11 The results for road projects are summarised as follows:

• European road projects indicate a range for new dual two-lane carriageways of between €5 million and €15 million per kilometre.

• American road projects indicate a range for two-lane road widenings of between €2 million and €8 million per kilometre.

• European accession country road projects indicate a range for new two-lane carriageways of between €0.5 million and €2.5 million per kilometre.

• UK, Poland and China road projects indicate a range for new 4-lane carriageways of between €1 million and €5 million per kilometre.

• European fixed link projects indicate a range of between €60,000 and €180,000 per square metre for tunnels and between €5,000 and €25,000 per square metre for bridges.

4.2.12 The results for urban transport projects are outlined as follows:

• European urban transport projects indicate a range of between €50 million and €150 million per kilometre of metro. However, costs of up to €300 million per kilometre were also observed, particularly in the US.

• European and US urban transport projects indicate a range of between €10 million and €30 million per kilometre of light rail/tramways.

4.2.13 Data on rail projects was much less abundant, but the results are outlined as follows:

• High-speed twin-track projects were found to lie in an indicative range of between €20 million and €90 million per kilometre.

• Conventional twin-track rail projects were found to lie in an indicative range of between €3 million and €12 million per kilometre.

4.2.14 We also found some data on energy and environmental projects, the results of which are summarised as follows:

74

• European wind farm construction costs indicate a range of between €1,000 and €2,000 per kW.83

• European, American and Russian construction costs for wastewater treatment facilities indicate a range of between €500 and €2,500 per cubic metre per day.

• American, Australian, Asian and European water treatment and desalination plant construction costs lie in a range of between €400 and €1,000 per cubic metre per day.

4.2.15 Our benchmarking results, originally presented as part of our First Interim Report are re-presented graphically in Annex A.

4.3 Review of ex post evaluation activities at the Member State level

Introduction

4.3.1 The WP10 study, as noted in the Executive Summary, arose from DG Regio’s own objective to establish an infrastructure project cost database, one that might provide reliable and robust cost benchmark data for the purposes of appraising and evaluating major projects in the future. The database that is the output of WP10 can, in turn, be viewed as a first step towards the achievement of this objective.

4.3.2 However, the desire for improved project appraisal, design and evaluation tools and techniques is not new. In particular, in some European countries, public sector agencies, industry and academia have, over recent years, become increasingly aware of the need to develop new methods to improve cost estimation. Consequently, more structured processes for the ex post evaluation of projects are beginning to appear at the Member State level.

4.3.3 This is potentially significant, particularly in respect of the potential benefits that can arise from the complementary nature of the work. For example, the outputs of Member State evaluations could be incorporated as part of DG Regio’s project cost database. Likewise, DG Regio’s database might be useful for the Member States in carrying out their ex post evaluation processes (as well as in the ex ante appraisal of future projects).

4.3.4 The mere existence of ex post evaluation processes might well be also useful in establishing channels for the more regular provision by the Member States of quality and consistent data on major projects that involve some proportion of EU funding.

4.3.5 In the following paragraphs, we examine the ex post evaluation activities of three European countries, namely France, Norway and the United Kingdom.84

France85

4.3.6 The French Internal Transport Act of 1982 (Loi d’Orientation des Transports Intérieurs) systematised and standardised, according to Chapulut et al (2005), the practice of economic appraisal of projects that had been widespread for nearly twenty years. The

83 The data also indicates some relationship between the capacity to be delivered by projects and the observed unit costs, that is, economies of scale. However, there were also exceptions. 84 There are a number of references for ex post evaluation work in Denmark, but the material is in Danish. The references are (i) Danish Ministry of Transport (2006), ‘Aktstykke om nye budgetteringsprincipper (Act on New Principles for Buudgeting’, Aktstykke nr. 16, Finansudvalget, Folketinget, Copenhagen, October 24 and (ii) Danish Ministry of Transport (2008), ‘Ny anlægsbudgettering på Transportministeriets område, herunder om økonomistyringsmodel og risikohåndtering for anlægsprojekter’, Copenhagen, November 18. 85 See Chapulut, Jean Noel, Jean Pierre Taroux and Emilie Mange (2005), ‘The New Ex Post Evaluation Methods for Large Projects in France’. Paper presented at the European Transport Conference, Strasbourg, France.

75

legislation includes a requirement to produce a report of the projects’ economic and social performance five years after its opening.

4.3.7 The manner of application of the principles established in the 1982 Act was established through a Statutory Instrument (in this case a decree) dated July 1984. This set out the requirements for the appraisal and ex post evaluation of all transport infrastructures and ‘major technological options’ (such as, for example, railway signalling systems) with a value in excess of 500 million Francs (Euro 82 million). The deadline for producing the ex post evaluation reports was set at 3 to 5 years from opening. However, due to the long lead times associated with large projects, the first reports could only be produced about 15 years after the Decree of 1984.

4.3.8 By the beginning of the 2000s, therefore, only two motorway scheme-ex post reports had been produced and there were significant comparability issues between the ex post results and the ex ante projections. This required the latter’s recalibration according to contemporary models, methods and appraisal parameter values, for example, updated methods used to monetise effects. Then, in 2001, a working group, set up with the task of ensuring a more consistent delivery of these reports, established 62 projects on which ex post reports had to be performed by 2005. However, only 10 reports had been produced by the end of that year. A simplified methodology has, therefore, been prepared in order to make up lost time.

4.3.9 The two main objectives of the ex post reports are to (i) report to the public about the performance of the projects for which assessments had been presented to them before decision-making and (ii) to enhance decision-making methods for transport infrastructures, especially ex ante appraisals through experience feedback.

4.3.10 The ten or so ex post reports that are available (mainly concerning high-speed rail projects) shows that the variance between the ‘all-in’ ex ante cost projection and the outturn cost lies between 0 and 25 per cent. Wide variations were, as one would expect, attributed to various site and environmental factors, urban or rural surroundings and other differences in physical constraints.

Norway86

4.3.11 The Norwegian Ministry of Transport and Communications, in 2005, authorised the Norwegian Public Roads Administration (NPRA) to continually carry out ‘post opening’ cost benefit studies of implemented projects. This was in response to questions raised by government auditors and decision-makers as to whether the expected outcomes and predicted impacts of major investments are actually achieved. Consequently, in the period 2006-2007, NPRA performed post opening cost benefit analyses (CBAs) of eight road schemes, with each assessment being carried out five years after the road opened for traffic.

4.3.12 In the past, road projects in Norway were systematically audited only in respect of road investment costs. The purpose of the new post opening assessment methodology is to (i) determine whether the NPRA is achieving its objectives (that is, the benefits of its work programme) and (ii) analyse reasons for deviations between actual and projected costs and benefits of projects. The applied framework for the post opening assessments was developed by NPRA and proceeds by selecting approximately 5 projects per annum, each with an investment cost greater than 200 million Norwegian Kroner.

86 See Kjerkreit, Anne, James Odeck and Kjell Ottar Sandvik (2008), ‘Post Opening Evaluation of Road Investment Projects in Norway: How correct are the estimated future benefits?’ Paper presented at the European Transport Conference, Netherlands.

76

4.3.13 The evaluation framework involves comparing ex ante estimates (re-calibrated according to contemporary methods and parameter values) with the results of the post opening assessments. While detailed conclusions on exact impacts are not expected, the objective is to conclude whether the benefits are plausibly the same as, bigger or less than forecasted. A number of causes of significant deviation were identified and examined, including:

• changes in the project specification;

• higher than expected construction costs (partly explained by the latter);

• traffic volumes and growth rates;

• numbers of accidents; and

• Travel speed and time.

4.3.14 Interestingly, costs were overestimated for five of the eight projects that had been studied by up to 22 per cent. The largest cost overrun (an underestimation of costs) was 43 per cent. Interestingly, this project had a lower NPV than projected. All other projects had a higher NPV than projected, which was mainly due to an under estimation of traffic flows (an important input into determining the benefits of such projects).

United Kingdom

4.3.15 The HM Treasury Green Book sets out the requirement for ex post evaluation of UK Government policies, programmes and projects.87 The objective, as in the other countries outlined above, is to examine outturns against what was expected and to ensure that the lessons learned are fed back into the decision-making process, thus ensuring that government action is continually refined to reflect what best achieves objectives and promotes the public interest.

4.3.16 Ex post evaluation comprises robust analysis, conducted in the same manner as an economic appraisal, and to which almost identical procedures apply. In other words, the task is to conduct a cost-benefit analysis, in the knowledge of what actually occurred rather than what is forecast to happen.

4.3.17 The application of the procedures and requirements espoused by the Green Book can, as an example, be clearly seen in the work of the UK Highways Agency. Under the Agency’s Programme of Major Schemes, road projects are subjected to Post Opening Project Evaluations (POPE) one year after, as well as five years after, the road in question is opened. The evaluation involves an assessment of:

• whether the objectives of the project have been achieved;

• whether the projected impacts of the project have been realised, including environmental, safety, economy, accessibility and integration; and

• What lessons can be learned from the project being evaluated for the appraisal and design of future projects.

4.3.18 Most, if not all, of these reports provide economic summary tables, which include comparisons between estimated and outturn ‘all-in’ project costs as well as progress in the achievement of the projected benefits. Useful examples are provided by the POPE

87 HM Treasury, ‘The Green Book: Appraisal and Evaluation in Central Government’, Treasury Guidance, London: TSO.

77

Summary Reports for the A1 (M) Ferrybridge to Hook Moor or the A21 Lamberhurst Bypass.88

Conclusion

4.3.19 While it is clear, therefore, that more structured processes for the ex post evaluation of major investment projects are, in some Member States, well-established or, in others, in the process of being established, none of them have culminated in the development of an infrastructure project cost database, especially one that could provide reliable and robust cost benchmark data for the purposes of appraising future major projects.

4.3.20 The objective of DG Regio to establish such a database is, therefore, all the more justified. Not only might the database provide a powerful tool for the Member States to be used in carrying out future project appraisal and evaluation, but it could be supplemented with the data produced by the Member States from their ex post evaluation processes.

4.4 Gathering data on the WP10 sample of major projects

Data gathering strategy

4.4.1 We commenced the data gathering exercise in early September 2008. However, it was only in November that we began to have success in contacting the relevant people from the appropriate agencies in each of the 11 Member States.

4.4.2 Our data collection strategy involved the following:

• making telephone or email contact with the relevant parties listed on the ERDF project application forms, beginning with the contact listed for the organisation responsible for project implementation (usually local government or other independent government agencies in the case of infrastructure projects) and/or the organisation empowered to issue certificates;

• where this was a private firm (as in the case of productive investments) or a contractor (as was the case with certain infrastructure projects) and the contact could not be located (either due to staff turnover or dissolution/acquisition of the firm itself), we endeavoured to contact a relevant government department, ministry or agency; and

• Otherwise, where necessary, we elevated our information requests to a relevant national government department or agency, which have, in general, been notified in any case when the local government authority or private firms are reluctant, or do not have authority, to provide the required information.

4.4.3 Having made contact, we proceeded to describe the objectives of the project and the kinds of information that we were seeking.

4.4.4 In approaching the Member States, we started with requests for official project completion reports and/or progress reports. However, the responses we received to these requests suggested that the Member States do not have in place the kinds of official reporting structures or outputs that could provide us with the level of detail of information we were seeking.89

88 The former is available at http://www.highways.gov.uk/roads/documents/POPE_A1_M__Ferrybridge.pdf , while the latter is available at http://www.highways.gov.uk/roads/documents/POPE_A21_Lamberhurst_OYA.pdf . 89 On the other hand, in one country (Ireland), it was indicated that the authority for urban transport projects had assembled information on the project that we are interested in ‘in response to a number of ongoing financial audits’

78

4.4.5 In the small number of cases where ‘completion’ reports were provided (namely for German road projects), they provided an insufficient amount and level of detail of information to be counted as useful.

4.4.6 In late September, we changed our approach to the data gathering and developed questionnaires to serve as a more detailed guide to our data requirements. We based our questionnaires on the spreadsheet data templates that were presented in our Inception Report.90 These templates reflected our ambitions in the sense of achieving a level of detail required to facilitate a robust and comprehensive analysis.91

4.4.7 Sample questionnaires for the different sectors (road, rail, urban transport, water and wastewater and productive investments) are provided in Annex II. Questionnaires were sent to Member States once we had established contact with someone willing and able to assist.

Data gathering results

4.4.8 We began the data gathering exercise with a total sample of 161 projects. These had been chosen in cooperation with the Commission from the information provided at the beginning of the study. This number fell initially to below the required 155 projects because, as we discovered, some projects belonged to sectors beyond the scope of the WP10 study, namely communications, ports and airports.

4.4.9 Moreover, once we began to establish contact on the remaining projects, it also became apparent that many had not yet been finished. This had the effect of further reducing our sample size below the required number.

4.4.10 In order to ensure that we had at least a chance of achieving the required 155 projects, we searched the information that had been provided by the Commission for other projects, particularly those for which there was at least some details that might provide, or lead us to acquire, information about them. These were added to the sample. Also, in another block of information sent by the Commission in November 2008, in response to our requests for missing ERDF applications (see below), the Commission sent us information on a number of projects that weren’t in the original data.

4.4.11 Having removed the projects from sectors beyond WP10’s scope (approximately 10) and added these ‘new’ projects, we had, by the end of November, 173 projects (128 infrastructure and 45 productive investments) on which we have actively sought data since September 2008. However, 16 of the 128 infrastructure projects and 10 of the 45 productive investments were reported as unfinished (26 in total). These primarily included Greek road and rail projects, one Italian and several Spanish water projects.

4.4.12 By mid-February 2009, we had received a total of 66 data returns (49 infrastructure projects and 17 productive investments). Following the presentation of our Second Interim Report to DG Regio (and the wider ex post evaluation Steering Group), the Evaluation Unit of DG Regio undertook to work with the desk officers from their Geographical Units in order to put pressure on the Member States to respond to our information requests.

4.4.13 By end March 2009, we had received information relating to 62 of the target 115 infrastructure projects and relating to 20 of the target 40 productive investments. This

and that this information deals with contract overruns for the utility diversion and main infrastructure contracts. The authority also indicated that it was preparing a lessons learned document and that it was considering unitising budget and actual costs on a per kilometre basis for utility diversions, structures, track and track bed. 90 Available at http://ec.europa.eu/regional_policy/sources/docgener/evaluation/pdf/expost2006/wp10report.pdf . 91 Note that, for non-English speaking Member States, there was the additional task for the data gathering team of translating the questionnaires.

79

was an increase in the size of the sample of major projects of 16, a direct result of the data gathering efforts of the DG Regio desk officers.

4.4.14 By end April 2009, following another push by our own data gathering team, we had received a total of 66 data returns on infrastructure projects and 30 on productive investments. The total sample size is, therefore, 96 projects. Some data was received during May 2009, but this was limited to information that was missing on the projects for which we had already received data returns.

4.4.15 The results of the data gathering effort to date are summarised in Table 7 below.

Table 7: Summary of our data gathering results





Data returns received 96 66 30 Source: RGL Forensics

Reasons for the poor response rate

4.4.16 There are several possible reasons for the poor response rate to our requests for project data. Primarily, perhaps, there was no obligation on Member States to monitor major projects individually for the 2000-2006 funding programme (rather monitoring was required at the programme level). They were not, therefore, legally obliged to collect or provide the information about the individual projects that we have been requesting. Moreover, the information required to put values to our unit cost indicators is not information that the European Commission has ever asked for or has ever asked to be monitored by the Member States. In this respect, WP10 represents a first step in this new aspect of monitoring of major projects.

4.4.17 The absence of any legal obligation to collect data in this manner was, perhaps, reflected in the fact that we encountered a number of cases where the project, as defined for the ERDF application, appeared not to correspond with what was monitored in practice. For example, Deutsche Bahn (DB) informed us that most of its projects defined for ERDF funding, were part of wider national projects or investment programmes. As a result, the impression given to us by DB was that, in order to provide us with the level of detail of information we sought, it was necessary to undertake a cost allocation exercise, in order to achieve correspondence between actual project costs and the estimates contained in the ERDF applications.

4.4.18 In the case of one Greek road project, we were informed that it was not possible to complete our questionnaire because the purpose of the project was to fill gaps and omissions from the other main road contracts. This project appears, therefore, to constitute small parts of several projects, and was not separately monitored as a stand-alone project.

80

4.4.19 For several projects, we did not receive ERDF applications from the Commission at the beginning of WP10. In some cases such as, for example, a lot of Spanish and some French projects, we were able to establish contact nonetheless. This was achieved through research based on the names of projects. In other cases, such as other projects in France and projects in Portugal and Slovakia, we were not able to establish contact independently.

4.4.20 In a number of cases where ERDF applications were available, we encountered difficulties in establishing contact with the right people from the relevant agencies. In the case of Spanish and French projects, for example, changes that had taken place in the national telephone numbering system since the preparation of the applications caused immediate setbacks for the data gathering team.

4.4.21 There were other cases where, despite having established contact with the relevant agency, we were unable to find persons able and willing to assist us. Whether this was due to people not wanting to get involved, fearing the outcome of questions, whether they just couldn’t be bothered or they had other priorities is unknown to us.

4.4.22 For projects on which we have had no means of establishing relevant contacts in the Member State (either due to a missing application or some other reason), the Commission sent us contact details, which the team has actively pursued.

4.4.23 The task of finding the agency or person who might hold the data we have been seeking has not been as straightforward as one might expect. For some projects, particularly road projects (in Germany, Greece, Ireland and Spain), there have been significant delays while bureaucratic responsibilities are clarified in the Member States. However, in several of these cases we have now received the information we requested.

Implications of the poor response rate

4.4.24 The direct implications of the relatively poor response rate are (i) a reduced sample size and (ii) a sample structure that is heavily skewed towards the road and rail sectors. The latter is illustrated in the following paragraphs and tables.

4.4.25 A more indirect implication is the limiting effect on the scope and depth the statistical analysis that was possible for WP10. However, this was more severely affected by the depth of the information that was received.

4.4.26 Table 4 shows the structure of the WP10 sample of 155 major projects that was intended for WP10, as outlined by DG Regio in Annex 3 of the tender specifications.92

92 See ‘Call for tenders by open procedure no. 2008.CE.16.0.AT.019…’ Document reference REGIO.C.2./JS D(2008) 680043, published on 22nd February, 2008.

81

Table 8: Structure of the intended WP10 sample of 155 major projects

Country # of projectsRoad Rail Urban

transport Water supply Wastewater treatment Energy Total

infrastructureBusiness support

Germany 25 8 6 14 11Greece 23 10 7 3 3 23Spain 39 12 8 5 1 1 27 12France 16 3 2 1 2 8 8Ireland 8 5 3 8Italy 20 6 7 3 1 1 18 2Austria 1 1Poland 5 3 1 1 5Portugal 15 1 3 2 3 9 6Slovakia 2 2 2UK 1 1 1Totals 155 50 37 10 6 3 9 115 40

Infrastructure projects

4.4.27 Table 9 shows the structure of the actual WP10 sample of 96 projects that form the basis of our analysis of major projects in later sections. As can be seen, the sample of infrastructure projects is, at present, heavily skewed towards the road and rail sectors.

Table 9: Actual structure of the WP10 sample of 96 major projects

Country # of projectsRoad Rail Urban

transport Water supply Wastewater treatment Energy Total

infrastructureBusiness support

Germany 15 3 6 9 6Greece 17 8 3 2 4 17Spain 25 1 6 3 10 15France 3 1 1 2Ireland 8 5 2 1 8Italy 8 2 5 1 8Austria 1 1Poland 6 2 1 3 6Portugal 11 1 1 4 6 5Slovakia 0UK 2 1 1 1Totals 96 21 24 8 4 0 9 66 30

Infrastructure projects

4.5 Conclusions

4.5.1 Our search for benchmark databases revealed that, with only a couple of exceptions, there are no relevant up-to-date databases of infrastructure costs.93 Some high level total cost benchmark data was available but there were problems with the definition of unit costs and with data comparability in general.

4.5.2 The European Commission may wish to consider, therefore, the possibility of setting up an infrastructure project cost database to provide reliable and robust cost benchmark data for the purposes of appraising and evaluating major projects in the future. Moreover, cooperation between other European public funding agencies (such as EIB, EBRD) could greatly improve the scope and quality of such a database and the quality of project appraisal and evaluation, as well as reducing the costs of data collection for DG Regio.

4.5.3 The exercise of gathering information about the projects in the WP10 sample was also, unfortunately, less straightforward than expected. Data on actual costs and outputs are not kept by the Commission and has been difficult to obtain from the Member States in some cases.

93 Namely, the World Bank ROCKS database

82

4.5.4 We identified at an early stage that there was very little formal project monitoring by the Member States in a form that would provide the necessary information. However, importantly, there is no legal obligation to monitor projects individually for the 2000-2006 funding programme (rather monitoring was required at the programme level). The Member States were not, therefore, legally obliged to collect or provide the information about individual projects.

4.5.5 Having commenced the data gathering effort with requests for official project completion reports and/or progress reports, it soon became apparent that such reports did not exist or, where they did, that they provided an insufficient amount and level of detail of information to be counted as useful. We changed our approach, therefore, by developing questionnaires to serve as a more detailed guide to our data requirements.

4.5.6 In Section 5 below, we explore the structure of the questionnaires developed and the implications of the poor response rate for DG Regio’s future intentions in the area of benchmarking the unit costs of major projects. We also discuss the kinds of information that Member States might be required or requested to record and provide as a condition attached to project funding.

83

5 UNIT COST DEFINITION AND MEASUREMENT FOR WP10 INFRASTRUCTURE PROJECTS

5.1 Introduction

5.1.1 The definition of unit cost indicators has three main dimensions, specifically:

• the types of costs to be included in the numerator of the unit cost indicator;

• the level of disaggregation of each project’s components; and

• The methodologies that could be used to ensure data comparability across heterogeneous projects.

5.1.2 Subsections 4.2 to 4.4 describe our approach to the definition of unit cost indicators, which takes account of the requirements of Tasks 2.2 and 2.3, regarding the calculation of estimated and actual unit costs for the sample major projects. We note at this stage, however, that the choice of unit cost indicators relies heavily on the level of detail of the available data.

5.1.3 The availability of data and its level of detail also have implications for the analysis of cost overruns and time delays (requested under Task 2.4) since the likelihood of finding statistically robust links between project characteristics and their final performance depends on the richness of the available data sample. Similar considerations apply to any analysis of the importance of effective ex-ante risk assessments in forecasting the costs of future developments (Task 2.5).

5.1.4 Throughout the analysis, we have attempted to use the data available to fulfill the tasks mentioned above. However, as described in section 3, the lack of available data in some circumstances hindered our ability to carry out a comprehensive statistical analysis. Our analysis is therefore mainly descriptive in nature. Nonetheless, whenever possible, we have tried to identify the key drivers that explain the differences between projects. We have attempted to do so both when comparing across projects, and when considering the differences between estimated and actual project costs.

5.1.5 This section of the report is structured as follows:

• in subsection 5.2, we present the categories of cost that have been included in the numerator of our calculations of the unit cost indicators;

• in subsection 5.3, we discuss the disaggregation of the projects physical components (and the associated costs);

• in subsection 5.4, we define the unit cost indicators that we used for the specific purposes of WP10. The number of unit cost indicators is constrained by the available information;

• in subsection 4.5, we present the attributes of projects on which we sought information, both general and sector-specific;

• in subsection 4.6, we describe our approach to the measurement of project delays;

• in subsection 5.7, we describe our approach to the measurement of cost overruns; and

• In subsection 5.8, we examine the types of cost adjustments that are necessary to make direct comparisons between projects from different countries, recorded at different points in time and using different currencies.

84

5.2 Categories of cost to be included

5.2.1 The unit costs presented in Section 6 are based on the project cost data available in the major project dossiers. In most cases, we found the ERDF application forms to contain useful information. The dossiers also included details on the Commission’s decision granting the requested funds and various reports, mostly on the cost-benefit analysis supporting the application.

5.2.2 However, in general, these various reports did not add any relevant information on costs beyond what was already available in the application forms. Figure 7 illustrates the relevant section within the ERDF application form for a sample major infrastructure project, the LUAS urban rail project in Dublin. We have adapted it for presentational purposes. This example illustrates the type of cost breakdown information that one can expect from a comprehensive application form.

Figure 7: Cost categories provided in ERDF application forms

6. COSTS OF PROJECT 6.1 Cost breakdown Eligibility date for expenditure: 1 January 2000 (date of receipt of application) Euro x 1000 (1999 prices)

‘ALL-IN’ PROJECT COST € MILLION

INELIGIBLE EXPENDITURE

ELIGIBLE COSTS** € MILLION

Enabling works 45.637 Nil 45.637 Utilities 18.509 Nil 18.509 Construction + infrastructure

173.302 Nil 173.302

Rolling Stock 36.049 36.049 Nil Property 39.914 39.914 Nil Project Management during implementation

20.102 2.402 17.700

Technical assistance Nil Publicity Included in

PM costs Nil

Contingency 3.472 Nil 3.472 Sub-TOTAL

Nil

Tax (VAT) Nil TOTAL

336.985 78.365 258.620

5.2.3 We discussed the cost categorisation in the First Interim Report (Sn. 2.5) where we agreed to work with the following definitions:

• Construction or ‘build’ costs. The costs associated with all aspects of project implementation. In the example above, this category of cost is further broken down between enabling works, utility diversions as well as the infrastructure ‘build’.

85

• Soft costs. Project management during implementation, technical assistance and publicity in the example in Figure 7 fall into this category. Other common examples are planning and design costs and costs associated with stakeholder consultation.

• Contingencies. This is explicitly allowed for in the example above and constitutes 1.4 per cent of total eligible project expenditure.

• Taxes. In the example above, it was noted that

‘VAT is included as it is a cost to the Project. The Project is classified with VAT ‘Exempt’ status and no input VAT is reclaimable from the Revenue.’

As another example, in the case of UK road projects, non-recoverable VAT applies to works costs only; other project costs such as preparation, supervision and land are all expected to be VAT recoverable.

• Land acquisition costs. For the Luas urban rail project, €40 million of projected expenditure on property acquisition was considered ineligible under the rules of the fund.

5.2.4 Unfortunately, many of the project applications did not provide the level of detail summarised above, which constrained the scope of unit cost indicators for which project cost estimates could be provided.

5.2.5 We also note that the data available raised the following issues:

• The treatment of indirect taxes levied by contractors on contracting authorities varies across Member States. In most cases, no detailed information on the actual treatment of VAT and other taxes was available, making it difficult to apply any correction to specific projects.

• There is a lack of consistent and comparable information on estimated contingency allowances. In some cases, the information is detailed, as in the example above. In other instances, the documentation provides only an indication of the contingency percentage built into the cost estimates. There are also project files that do not provide any information at all on contingency allowances.

5.2.6 While, according to WP10’s specifications, cost estimates should be based on the information contained in the project dossiers, we made the following request to Member States when seeking additional information:

‘We have extracted the estimated cost information available from the ERDF project application forms supplied by the Commission. Where it is possible to provide more detailed breakdowns (for estimates and actuals), please do so.’

5.2.7 Therefore, whenever the cost estimates included in the application could be broken down further, we have included this information in the dataset and in our analysis. However, this applies to only a small number of projects.

5.2.8 We based our questionnaires on the cost categories outlined above, as shown in Table 10 below. Table 1 is an extract from one of our early road project questionnaires which was sent to Member States. (Annex B provides comprehensive examples of the questionnaires).

86

Table 10: Extract from questionnaire for road projects sent to Member States Cost category Estimate (Euro million) Actual (Euro million)

Total cost

Planning / design cost

Land acquisition cost

Pavement construction cost:

- one carriageway

- two carriageway

Pavement rehabilitation cost:

- one carriageway

- two carriageway

Bridges

Tunnels

Other

- project management

- publicity

- technical assistance

- contingency

- other (please specify)

5.3 Disaggregating the physical components of projects

5.3.1 We adopted different ‘levels’ of cost indicator to reflect various degrees of cost disaggregation. We defined three main ‘levels’ of disaggregation, from general to specific, as follows:

• Level 1: indicators that reflect the ‘all in’ costs of a project, including all appropriate categories of cost outlined in the previous section and all project components.

• Level 2: indicators that reflect the ‘build’ cost of individual key components of projects such as, for example, bridges and tunnels in road and rail projects.

• Level 3: indicators that further distinguish between different types of key components, such as different types of bridges and tunnels.

5.3.2 The First Interim Report contained the most important identification of cost components (see Table 3). For ease of reference, we reproduce it as Table 11 below.

5.3.3 Taking roads as an example, Table 11 shows that Level 1 indicators will include the costs of all of the components of the project, including the pavement itself, bridges, tunnels, and other key components. Level 2 indicators provide more detailed information, as they separate out the ‘build’ costs of those key components. Finally, Level 3 indicators would further identify, for example, different grades of pavement or the different types of bridges and tunnels that are built, focusing on the features that may cause costs to vary across projects.

87

Table 11: Project disaggregation and corresponding Levels 1-3 unit cost indicators

Level 1 costs (‘all in’)

Level 2 costs (key components)

Level 3 costs (different types of key components)

ROADS 1 carriageway / 2 carriageway 2 lanes / 3 lanes / 4 lanes or greater

Pavement construction Bridges Tunnels

Grade of pavement Types of bridge Types of tunnel

RAIL Single track Twin track

Track construction Stations Bridges Tunnels Rolling stock

At grade Elevated In tunnel Types of bridge Types of tunnel

URBAN TRANSPORT Metro Tramway Buses / taxis

Network (track, road) Stations / stops Bridges / tunnels Rolling stock

At grade Elevated In tunnel Types of bridge Types of tunnel

ENERGY Electricity / gas Nuclear / wind

Generation Networks (transmission / distribution) Supply

ENVIRONMENTAL Water Wastewater

Extraction / treatment Distribution Supply

Gravity or rising mains Pipes or culverts

5.3.4 Given the difficulties that we encountered in the data gathering exercise (as discussed in Section 4) we were able to calculate only Level 1 estimated unit costs (Task 2.2) and actual unit costs (Task 2.3) across all projects. Whenever possible, we calculated Level 2 and Level 3 indicators, but this information was not consistently available for all projects.

5.4 Defining unit cost indicators for Work Package 10

5.4.1 We were able to gather comparable cost information for infrastructure projects in five sectors: Rail, Road, Urban Transport, and, to a lesser extent, Energy and Water.

5.4.2 After normalising the data, we calculated unit costs to compare projects of different magnitudes. We discuss these adjustments in further detail in subsection 4.5. We calculated unit cost ratios for the various levels of aggregation, depending on the available data. As noted above, we could calculate Level 1 unit costs across all projects. Whenever possible, we also calculated Level 2 unit costs. However, the necessary information was not always available. Due to data limitations, we could calculate Level 3 indicators only for a couple of projects.

5.4.3 For each of the five sectors, we have grouped projects according to some of their infrastructure network features. For example, for Rail, in addition to calculating overall unit costs for all projects, we also distinguish between ‘Single track’ projects and ‘Twin track’ projects.

88

5.4.4 For Level 1 indicators, when possible, we attempted to break down total costs into the different categories included in the project cost estimates and outturns. This could provide some insight into the reasons for cost differences between projects. However, the lack of consistent data prevented us from undertaking a more comprehensive exercise. Specifically, we broke down Level 1 costs into Build, Soft, Contingency, Taxes and Land costs. Soft costs group all costs related to Project Management, Planning, Publicity, Supervision, Technical Assistance, External Controls, Works of Art and any undefined ‘other’ costs.

5.4.5 Tables 3 - 7 summarise the unit cost indicators which we calculated for each sector. We present the results of our analysis in Section 6 below, where we also compare estimated unit costs with actual unit costs. Given that the available data is sparse, it is difficult to do unit cost comparisons at the level of detail contained in these tables.

5.4.6 For our Level 2 and Level 3 indicators, we referred to ‘build’ costs only. When respondents to our questionnaires did not provide a breakdown of the other categories of cost involved in the project (such as soft costs, for example), we tried to isolate the ‘build’ component to achieve comparability. However, this was not always possible.

5.4.7 Finally, we note that unit costs have been calculated without differentiating between fixed costs and variable costs. This fixed/variable cost split represents another variable which can explain the differences between unit costs of different projects. Also, it may mean that unit costs from one project may not be particularly relevant for other projects of a very difference scale. Identifying benchmark projects of a similar scale as well as other attributes is therefore likely to help identify the most relevant benchmarks against which to compare a particular sample project.

Table 12: Unit cost indicators for rail projects

RAIL Indicators Units

Level 1

‘All in’ unit cost EURm/km

Level 2

Unit cost of land EURm/ha

Unit ‘build’ cost of trackwork EURm/km

Unit ‘build’ cost of stations EURm/nr

EURm/nr Unit ‘build’ cost of bridges

EUR/m2

Unit ‘build’ cost of tunnels EURm/km

Level 3

Unit ‘build’ cost of single track - at grade * EURm/km

Unit ‘build’ cost of single track – elevated EURm/km

Unit ‘build’ cost of single track – in tunnel EURm/km

Unit ‘build’ cost of single track - bank EURm/km

89

RAIL Indicators Units

Unit ‘build’ cost of twin track - at grade EURm/km

Unit ‘build’ cost of twin track - elevated EURm/km

Unit ‘build’ cost of twin track - in tunnel EURm/km

EURm/nr Unit ‘build’ cost of stations - ground

EUR/m2

EURm/nr Unit ‘build’ cost of stations - underground

EUR/m2

EURm/nr Unit ‘build’ cost of beam bridges ^

EUR/m2

EURm/nr Unit ‘build’ cost of cantilever bridges §

EUR/m2

EURm/nr Unit ‘build’ cost of arch bridges

EUR/m2

EURm/nr Unit ‘build’ cost of suspension bridges

EUR/m2

EURm/nr Unit ‘build’ cost of cable stay bridges

EUR/m2

EURm/nr Unit ‘build’ cost of truss bridges

EUR/m2

Notes:

* Track ‘at grade’ means that the trackwork is at ground level, rather than elevated or in a tunnel.

^ ‘Beam’ bridges are the simplest kind of bridge that exist today and consist of a single horizontal beam with 2 supports, usually one on either end.

§ ‘Cantilever’ bridges are bridges built using cantilevers, structures that project horizontally into space, supported on only one end. For small footbridges, the cantilevers may be simple beams. However, large cantilever bridges designed to handle road or rail traffic use trusses built from structural steel or concrete box girders.

Table 13: Unit cost indicators for road projects

ROAD Indicators Unit

Level 1


90


Level 2


Unit ‘build’ cost of pavement work EURm/km

Unit ‘build’ cost of bridges EUR/m2


Level 3

Unit ‘build’ cost of pavement – one carriageway one lane EURm/km

Unit ‘build’ cost of pavement – one carriageway two lane EURm/km

Unit ‘build’ cost of pavement – one carriageway three lane EURm/km

Unit ‘build’ cost of pavement – one carriageway four lane EURm/km

Unit ‘build’ cost of pavement – one carriageway shoulder EURm/km

Unit ‘build’ cost of pavement – two carriageway two lane EURm/km

Unit ‘build’ cost of pavement – two carriageway three lane EURm/km

Unit ‘build’ cost of pavement – two carriageway four lane EURm/km

Unit ‘build’ cost of pavement – two carriageway six lane EURm/km

Unit ‘build’ cost of pavement – two carriageway shoulder EURm/km

Unit ‘build’ cost of tunnels – bored * EURm/km

Unit ‘build’ cost of tunnels - cut and cover ^ EURm/km

Unit ‘build’ cost of tunnels – other EURm/km

EURm/nr Unit ‘build’ cost of beam bridges

EUR/m2

EURm/nr Unit ‘build’ cost of cantilever bridges

EUR/m2


EUR/m2


EUR/m2


EUR/m2

91



EUR/m2

EURm/nr Unit ‘build’ cost of in-situ bridges

EUR/m2

EURm/nr Unit ‘build’ cost of launching bridges

EUR/m2

Notes:

* ‘Bored’ tunnels are what most people understand as a tunnel. It is normally constructed by one of two methods: (i) a tunnel boring machine (TBM), which is a large drill-like machine which bores through the rock using cutting teeth on the front face, usually resulting in circular tunnels; and (ii) the New Austrian Tunneling Method (NATM), during which a hole (which may be circular or more commonly oval) is cut through the rock using a combination of excavators, manual labour and occasionally explosives. Concrete is then sprayed on the rock to hold it in place until rings (usually of concrete) are inserted inside the tunnel and grout placed on the space between.

^ ‘Cut and cover’ tunnels are a simple method of construction for shallow tunnels, where a trench is excavated and roofed over. A strong overhead support system is required to carry the load of the covering material. These tunnels are usually square in shape.

Table 14: List of unit cost for Urban Transport projects

URBAN TRANSPORT Indicators Unit

Level 1


Level 2


Unit ‘build’ cost of trackwork EURm/km

EURm/nr Unit ‘build’ cost of stations

EUR/m2

Unit ‘build’ cost of bridges EUR/m2


Level 3

Unit ‘build’ cost of metro single track - at grade EURm/km

Unit ‘build’ cost of metro single track – elevated EURm/km

Unit ‘build’ cost of metro single track – in tunnel EURm/km

92


Unit ‘build’ cost of metro twin track - at grade EURm/km

Unit ‘build’ cost of metro twin track - elevated EURm/km

Unit ‘build’ cost of metro twin track - in tunnel EURm/km

Unit ‘build’ cost of tram single track - at grade EURm/km

Unit ‘build’ cost of tram single track – elevated EURm/km

Unit ‘build’ cost of tram single track – in tunnel EURm/km

Unit ‘build’ cost of tram twin track - at grade EURm/km

Unit ‘build’ cost of tram twin track - elevated EURm/km

Unit ‘build’ cost of tram twin track - in tunnel EURm/km

Unit ‘build’ cost of bus single track - at grade EURm/km

Unit ‘build’ cost of bus single track – elevated EURm/km

Unit ‘build’ cost of bus single track – in tunnel EURm/km

Unit ‘build’ cost of bus twin track - at grade EURm/km

Unit ‘build’ cost of bus twin track - elevated EURm/km

Unit ‘build’ cost of bus twin track - in tunnel EURm/km

EURm/nr Unit ‘build’ cost of stations - ground

EUR/m2

EURm/nr Unit ‘build’ cost of stations - underground

EUR/m2

EURm/nr Unit ‘build’ cost of tram stops

EUR/m2

EURm/nr Unit ‘build’ cost of bus stops

EUR/m2

EURm/nr Unit ‘build’ cost of beam bridges

EUR/m2

EURm/nr Unit ‘build’ cost of cantilever bridges

EUR/m2


EUR/m2

93



EUR/m2


EUR/m2


EUR/m2

Table 15: List of unit cost for Water projects

WATER Indicators Unit *

Level 1


Level 2


EURm/km Unit ‘build’ cost - water supply

EURm/inhabitant

EURm/km Unit ‘build’ cost – treatment

EURm/inhabitant

EURm/m3 Unit ‘build’ cost – sewage network

EURm/inhabitant

Level 3

Unit ‘build’ cost of water supply – gravity - pipes EURm/km

Unit ‘build’ cost of water supply – gravity - culverts EURm/km

Unit ‘build’ cost of water supply – gravity – tunnel EURm/km

Unit ‘build’ cost of water supply – gravity – construction EURm/km

Unit ‘build’ cost of water supply – gravity - machinery EURm/km

Unit ‘build’ cost of water supply – rising - pipes EURm/km

Unit ‘build’ cost of water supply – rising - culverts EURm/km

Unit ‘build’ cost of water supply – pressure - pipes EURm/km

Unit ‘build’ cost of water supply – pressure - culverts EURm/km

94

WATER Indicators Unit *

Unit ‘build’ cost of water network – gravity - pipes EURm/km

Unit ‘build’ cost of water network – gravity - culverts EURm/km

Unit ‘build’ cost of water network – rising - pipes EURm/km

Unit ‘build’ cost of water network – rising - culverts EURm/km

Unit ‘build’ cost of water network – pressure - pipes EURm/km

Unit ‘build’ cost of water network – pressure - culverts EURm/km

Notes:

* Project length was made available for most projects in the water sector. Therefore, we have chosen it to allow the comparison of as many projects as possible. However, in some cases, such as projects focused on sewage treatment, volumetric measures appear to be a more suitable unit for the comparison.

Table 16: List of unit cost for Energy projects

ENERGY Indicators Unit *

Level 1


Level 2


Unit ‘build’ cost - gas supply EURm/km

Unit ‘build’ cost - turbines EURm/nr

Level 3

Unit ‘build’ cost of gas supply - pipes EURm/km

Unit ‘build‘ cost - turbines EURm/nr

Unit ‘build’ cost - earthworks EURm/km

Unit ‘build’ cost - foundation EURm/km

Unit ‘build’ cost - substations EURm/nr

Unit ‘build’ cost - cabling EURm/km

Notes:

* The variety of energy project type makes it difficult to choose a common unit of measurement. Therefore, in order to make a comparison between as many projects as possible, we use the number of installations as a common measurement unit. We recognize that this unit may not be appropriate for all types of project. Specifically, for example, we used the number of kilometres as the unit of measurement for pipeline projects.

95

This, however, rendered it inappropriate to compare their ‘all-in’ Level 1 unit costs with those of other energy projects.

5.5 Information sought on the characteristics of projects

5.5.1 With a view to establishing sound bases for comparisons between projects, across sectors and countries, we also sought data on the characteristics of projects. The general attributes sought for all infrastructure projects is shown in Table 17 below.

Table 17: General attribute information sought on infrastructure projects

General attributes

Country

Contract type

Project complexity

Procuring agency

Pricing structure e.g., lump-sum or re-measure

Number of tenderers

Tender period Weeks / months

Proportion of new build / refurbishment

Conditions of contract e.g., FIDIC, ad hoc

Design responsibility Contractor, contracting authority or both

Contractual dispute resolution procedures e.g., arbitration, dispute boards, expert determination, adjudication

Funding structure % ERDF

Predominant locality

Predominant terrain

Predominant geology

Predominant environment

5.5.2 We also sought sector-specific attributes. The specific attributes sought for rail and urban transport projects are shown in Table 18 below.

Table 18: Sector-specific project attributes

Sector-specific attributes – Rail

Length of track km

Design capacity Passenger hours (‘000), Average speed (km / h)

Proportion of track in fixed link %

High-speed or normal rail

Average station spacing

Electrical / diesel type

Type of rolling stock type

96

Sector-specific attributes – Rail

Stops number

Controlled junctions number

Depots number

Park-and-ride facilities number

Track gauge mm

Power supply

Platform length m

Platform width m

Railway signalling

Sector-specific attributes – Urban transport

Length of track

Design capacity

Proportion of track in fixed link

Number of stations

Electrical / diesel

Type of rolling stock

Proportion of elevated track

Proportion of underground track

Proportion of underground stations

Proportion of surface stations

5.6 Measuring project completion times

5.6.1 In addition to measuring project unit costs, we also carried out an analysis of both estimated and actual project completion times. For each project, we collected information on completion times at each construction phase. We then calculated the average estimated and actual completion times across all projects. We summarise this information using the template shown in Table 17 below.

Table 19: Project completion times template

Project phase Estimated

completion time

(months / km)

Actual completion

time (months / km)

Absolute difference

(months / km)

Percentage difference

(%)

Planning

Funding

Permissions

Site preparation

Construction

TOTAL

97

5.6.2 This analysis allowed us to develop some insights on projects and sectors which tend to suffer from delays and on project development phases where the largest delays tend to occur.

5.7 Indentifying the causes of cost overruns and time delays

5.7.1 In addition to gathering information about estimated and actual costs and time delays, we also set out to identify the main causes of these discrepancies.

5.7.2 To do so, we adapted the framework of Mott MacDonald (2002), which subdivided the reasons for cost overruns and time delays into categories (which we call Type 1) and sub-categories (Type 2) to better capture the granularity of the possible explanations for delays and cost overruns. As mentioned in the First Interim Report, we only focus on technical reasons since they are most frequently offered by project managers.

5.7.3 Table 20 shows the cost overrun and time delay categories used in the project questionnaires.

Table 20: Type 1 and Type 2 cost overrun and time delay categories

Type 1 Type 2

Procurement Issues Complexity of Contract Structure

Design Changes

Contractor specific difficulties

Disputes with suppliers and subcontractors

Poor Planning/Methodological errors

Project specific Design Complexity

Degree of Innovation

Environmental Impact

Site access difficulties

Suspension of works

Delays by statutory authorities and/or contractors

Late commencement of work

Construction period

Client Specific Inadequacy of the Business Case

Large Number of Stakeholders

Funding Availability/Problems

Project Management Team

Project environment Public Relations

Site Characteristics

Permits/Consents/Approvals

External issues Political

Economic

Changes in Legislation/Regulations

Technology

98

Type 1 Type 2

Inflation

Exchange Rates

Force Majeure

Other (specify)

5.7.4 In our analysis, we asked respondents to score, for each project, each of the reasons for cost overruns and time delays on a scale of zero to three, with zero representing ‘no or insignificant cause of delay’, one a ‘minor factor’ (less than 20%), two a major factor (20-50%) and three a ‘very significant factor’ (greater than 50%).

5.7.5 We then aggregated the responses received to obtain average scores for each Type 1 category, namely procurement issues, project-specific issues, client-specific issues, project environment issues and external issues. This allowed us rank the categories with respect to their relative impact on the projects’ cost overrun and delays. Unfortunately, the small data sample did not allow a more granular discussion of the relative impact of Type 2 factors.

5.7.6 The analysis of cost overruns and time delays is mainly descriptive. This is because the small sample size did not allow the determination of statistically robust relationships between cost overruns / delays and specific Type 1 and 2 factors.

5.8 Methods for achieving data comparability

5.8.1 In principle, several adjustments are necessary to compare projects carried out in different jurisdictions and in different periods as we highlighted in the literature review. In the First Interim Report, we identified a variety of these approaches.

5.8.2 In this case, however, we believe the need for adjustment to be small. This is because the sample of projects we have collected originates from 9 Member States, of which 8 belong to the EU-15 group. We would expect the differences between these countries in terms of institutional arrangements, procedures and regulations to be relatively small. This allows for a more direct comparison of data.

5.8.3 Therefore, we believe only two adjustments are necessary. First, we adjusted all cost data for inflation, converting them to a common price base (2007 prices). To do so, we used Eurostat’s country-specific construction price inflation indices, as shown in Table 21 below.

5.8.4 Then we converted all monetary values into Euros, in order to have a common comparable currency base. For most countries, the data provided were already expressed in Euros. Therefore, out of all the entries in the sample, we only needed to convert cost data for Poland and the UK. To do so, we used the exchange rates shown in Table 22 below.

Table 21: Price indices used for inflation adjustment

Country 2000 2001 2002 2003 2004 2005 2006 2007

France 0.765 0.786 0.811 0.838 0.887 0.907 0.956 1

Germany 0.897 0.903 0.910 0.920 0.942 0.951 0.971 1

99

Country 2000 2001 2002 2003 2004 2005 2006 2007

Greece 0.802 0.824 0.843 0.865 0.892 0.920 0.957 1

Ireland 0.546 0.638 0.664 0.731 0.825 0.896 0.982 1

Italy 0.787 0.804 0.836 0.861 0.897 0.933 0.962 1

Poland 0.859 0.886 0.884 0.877 0.900 0.925 0.938 1

Portugal 0.825 0.834 0.861 0.876 0.909 0.932 0.965 1

Spain 0.762 0.783 0.796 0.813 0.851 0.891 0.952 1

United Kingdom 0.562 0.612 0.688 0.781 0.875 0.919 0.962 1

Source: Eurostat

Table 22: Exchange rates for non-Euro Member States

Country FX 2000 2001 2002 2003 2004 2005 2006 2007

Poland PLN/€ 4.008 3.672 3.857 4.400 4.527 4.023 3.896 3.784

United Kingdom GBP/€ 0.609 0.621 0,628 0.691 0.678 0.683 0.681 0.684

Source: European Central Bank

5.8.5 We believe that converting all monetary values into a single currency and applying the appropriate inflation adjustments should capture most of the intrinsic differences between Member States, especially within the EU-15 group.

5.8.6 However, it could be argued that some intrinsic differences in costs may still remain. The common currency eliminates the price level differences, but may not address possible volume differences. For example, it may not take fully into account differences in the purchasing power between Member States. To address this issue, adjustments for Purchasing Power Parity (PPP) could be made, in some contexts, to adjust for price and volume level differences between countries.

5.8.7 However, we note that Eurostat and the OECD warn against using PPP indices for specific goods and services. Specifically, they state that ‘PPPs are statistical constructs rather than precise measures. While they provide the best available estimate of the size of a country’s economy, of the economic well-being of its residents and of its general price level in relation to the other countries in the comparison, they are, like all statistics, point estimates lying within a range of estimates – the ‘error margin’ – that includes the true value. The error margins surrounding PPPs depend on the reliability of the expenditure weights and the price data as well as on the extent to which the particular goods and services selected for pricing by participating countries actually represent the price levels in each country. As with national accounts data

100

generally, it is not possible to calculate precise error margins for PPPs or for the real final expenditure levels and comparative price levels derived from them.’94

5.8.8 While PPP estimates are commonly used to compare national economic aggregates such as GDP per capita, they become less reliable for lower levels of aggregation and certainly for the comparison of individual cost items. Again, according to the OECD, ‘below the level of the main aggregates, error margins are compounded by differences in the national classifications used by participating countries in their national accounts. Because the margins of error increase as the level of aggregation gets lower, neither Eurostat nor the OECD publish results of their comparisons below a certain level of detail’.95 This implies that PPP indices specific to the industries under analysis in this case are not readily available, and that using off-the-shelf PPP adjustment factors may not be appropriate.

5.8.9 We decided not to apply any PPP correction to the cost data we collected. We believe that, based on the evidence we found in the literature, applying non-specific PPP corrections would not add significant insights to our analysis. Moreover, any adjustment based on PPP indices, which are normally intended to be used only with national aggregates, could be a source of unobservable bias and would reduce the robustness of the results.

94 Eurostat – OECD (2006), Methodological Manual on Purchasing Power Parities, European Commission ISBN 92-79-01868-X 95 Eurostat – OECD, ibid.

101

6 INFRASTRUCTURE: ESTIMATED AND ACTUAL UNIT COSTS AND COMPLETION TIMES

6.1 Introduction

6.1.1 This section sets out the actual and estimated costs for major infrastructure projects funded by ERDF in the 2000-2006 programming period. The results for each of the sectors (Rail, Road, Urban Transport, Water and Energy) are presented separately.

6.1.2 We start with Level 1 ‘all in’ unit costs. We then focus on the more specific Level 2 costs. Unfortunately, as noted above, the data is sparse. This has allowed us to calculate only a subset of Level 2 costs.

6.1.3 Whenever possible, we compare the project unit costs with cost benchmarks. However, given the diversity in the underlying conditions surrounding each project and the benchmark values, the comparability of these benchmarks with projects unit costs is limited. We report unit cost benchmarks, therefore, as ranges, and for reference purposes only. They should not be treated as specific cost targets for the projects discussed in this section.

6.1.4 We follow the presentation of project unit costs with a comparison of estimated and actual project completion times. The aim of this analysis is to assess the extent of average delays for each project phase, following the template shown in Table 19 (see section 4.6 above). In addition to measuring project unit costs, we also carried out an analysis of both estimated and actual project completion times. For each project, we collected information on completion times at each construction phase. We then calculated the average estimated and actual completion times across all projects.

6.1.5 After analysing each type of project separately, we bring together the results on cost overruns and delays to carry out a comparison across project types and countries. While this analysis is mainly descriptive due to the small size of the sample, it helps us gather insights on possible specific data patterns and illustrates the type of analysis that is possible with the spreadsheet tool. For example, it helps show whether cost overruns for a particular Member State generally tend to be higher or lower than the average. As more data is input into the spreadsheet, the statistical significance of the analysis produced from it will improve.

6.1.6 This chapter concludes with some observations on the role of ex-ante risk assessment for the determination of potential causes of cost overruns and delays. We based it on a review of the scope of the risk assessment done in a small sample of ERDF application forms. We briefly present the findings of that review.

6.1.7 In the charts of this section, we have identified projects by using custom short names. These names identify the project year and the project country. Whenever possible we have also included an indication of the project locality. For reference, Annex I provides a complete list of all projects we have analysed and their respective project code.

102

6.2 Rail 6.2.1 This sector has the largest number of projects with available data at the aggregate

level. It features 24 infrastructure projects out of 66 across all sectors or about 37 per cent. Overall, the data is fairly complete. However, for four projects out of 24, key information, such as project length in kilometres was not available.

6.2.2 At a higher level of detail, the lack of a consistent set of data across most projects prevents full comparison of Level 2 unit cost categories.

Level 1

6.2.3 Figure 8 shows Level 1 ‘all-in’ estimated unit costs of all rail projects, while Figure 9 compares actual unit costs. Whenever the data allows, the chart breaks down total costs into its key components, this information is only available for a small subset of projects.

Figure 8: Level 1 Rail: ‘All in’ estimated unit costs for all projects

0

10

20

30

40

50

60

70

DE03 - P

aderb

orn

DE05 - H

albers

tadt

DE03 - S

uedan

bindung

DE05 - R

ostock

DE02 - B

erlin

DE03 - C

ity tu

nnel

ES05 - P

roject C

ES05 - P

roject B

ES02 - C

oncurso

Abierto

ES05 - P

roject A

ES06 - P

roject A

ES05 - N

uevo acc

eso

GR05 - P

roject A

GR05 - S

ignal

IT03 - P

alerm

o

IT06 - R

oma

IT03 - C

asert

a

IT04 - B

ari

IT04 - D

ecim

omannu

Uni

t cos

t (€m

/km

)


103

Figure 9: Level 1 Rail: ‘All in’ actual unit costs for all projects

0

10

20

30

40

50

60

70

DE03 - P

aderb

orn

DE05 - H

albers

tadt

DE03 - S

uedanbindung

DE05 - R

ostock

DE02 - B

erlin

DE03 - C

ity tu

nnel

ES05 - P

roject C

ES05 - P

roject B

ES02 - C

oncurso

Abierto

ES05 - P

roject A

ES06 - P

roject A

ES05 - N

uevo acc

eso

GR05 - P

roject A

GR05 - S

ignal

IT03 - P

alerm

o

IT06 - R

oma

IT03 - C

asert

a

IT04 - B

ari

IT04 - D

ecim

omannu

Uni

t cos

t (€m

/km

)


6.2.4 The two charts show a very high degree of variation in Level 1 unit costs across all

projects. This is not unexpected given the wide variation in types of project and differences in key characteristics, such as numbers of stations, tunnels and bridges and different terrain.

6.2.5 The charts also group different classes of project. We have grouped together single-track projects (the first and third on the left-hand side, namely DE03 Suedanbindung and DE03 Paderborn) with twin-track projects. However, this does not appear to be a clear determinant of cost differences, as the two single-track projects appear to be more expensive than some twin-track projects.

6.2.6 Some projects stand out with very high unit costs. A possible reason for high unit costs can be found in each project’s environmental characteristics. For example, the two Spanish projects (ES05 – Project C and ES05 – Project B) are characterised by high unit costs, both estimated and actual. This may be explained by the fact that these projects have been carried out in 100% mountainous terrain, characterized by a 100% hard geology. Similarly, the Italian projects (in particular IT03 – Palermo and IT06 – Roma) have been carried out in hilly terrain. Information on the projects with the lowest unit costs was not available.

6.2.7 From the literature review, we also identified some benchmarks for Level 1 rail unit costs. For single track projects, we found that unit costs would range between 2.9 and 4.4 EURm/km. For twin-track projects, we identified a wider range of costs, between 2.8 and 8.6 EURm/km. We note, however, that these benchmarks are based on a set of projects which are not directly comparable with those in our analysis. Therefore, they should not be treated as necessarily relevant.

Level 2

6.2.8 As noted above, Level 2 unit costs are only available for a subset of indicators.

104

6.2.9 The following charts compare Level 2 build unit costs for rail projects. Each chart summarises all the information available for that specific Level 2 unit costs. The same level of information is not always available for all projects. Some projects have either estimated data or actual data, but not both. Where this is the case, we leave the missing data column blank.

6.2.10 Figure 10 provides the estimated and actual unit build costs for trackwork for all projects. Also for Level 2 unit costs, we note a degree of variation similar to that observed at Level 1, suggesting that varying project characteristics can lead to wide variations in cost. Also in this case, the two Spanish projects show higher unit costs due to their difficult location, characterised by mountainous terrain.

Figure 10: Level 2 Rail: Estimated and actual unit ‘build’ cost of trackwork

0

10

20

30

40

50

60

70

DE03 - S

uedan

bindung

DE03 - P

aderb

orn

DE05 - H

albers

tadt

DE02 - B

erlin

DE05 - R

ostock

DE03 - C

ity tu

nnel

ES05 - P

rojec

t C

ES05 - P

rojec

t B

ES02 - C

oncurso

Abier

to

ES05 - P

rojec

t A

ES06 - P

rojec

t A

ES05 - N

uevo ac

ceso

GR05 - P

rojec

t A

GR05 - S

ignal

IT06 - R

oma

IT03 - P

alerm

o

IT03 - C

asert

a

IT04 - D

ecim

omannu

IT04 - B

ari

Uni

t cos

t (€m

/km

)

Estimated Actual

6.2.11 The following three charts compare Level 2 unit costs for stations, bridges and tunnels respectively. Due to the small sample size, it is difficult to draw conclusions. Nonetheless, some salient project characteristics appear to play a role. For example, tunnels built in urban areas (such as project DE03 – City tunnel) are likely to cost significantly more than comparable tunnels in rural areas.

105

Figure 11: Type 2 Rail: Estimated and actual unit ‘build’ cost of stations

0

5

10

15

20

25

30

35U

nit c

ost (

€m/n

r)Estimated Actual

6.2.12 The impact of the terrain on project unit costs is also evident. The three Spanish projects and the Greek project are characterized by a mountainous terrain. This has an implication for unit cost of bridges. On the other hand, the Italian project was built on relatively flat terrain..

Figure 12: Level 2 Rail: Estimated and actual unit ‘build’ cost of bridges

0.0

0.5

1.0

1.5

2.0

2.5

ES05 - Project A ES02 - ConcursoAbierto

ES05 - Nuevo acceso GR05 - Project A IT04 - Bari

Uni

t cos

t (€0

00s/

m2)

Estimated Actual

106

Figure 13: Level 2 Rail: Estimated and actual unit ‘build’ cost of tunnels

0

20

40

60

80

100

120

140

160

DE03 - City tunnel ES02 - ConcursoAbierto

GR05 - Project A IT03 - Palermo IT03 - Caserta IT04 - Bari IT06 - Roma

Uni

t cos

t (€m

/km

)Estimated Actual

6.2.13 Figure 13 provides another example of how project characteristics can have a marked

impact on unit costs. Urban tunnels are significantly more expensive to build than tunnels carried out in predominantly rural locations.

Cost overruns and delays

6.2.14 To explore the causes of cost overruns and rank them according to their relative importance, we used the framework described in Chapter 4 (see section 4.6). We calculated the average score for each Type 1 factor, based on the individual scores of the respective Type 2 factors. These are shown in Figure 14.

6.2.15 ‘Project specific factors’ were the most important Type 1 source of cost overruns. Within these factors, the most important Type 2 was the project’s ‘environmental impact’ which scored slightly above one, indicating a minor factor (<20%). ‘Authority delays’ and ‘work suspensions’ closely followed. Procurement issues due to ‘design changes’ were also a major cause of cost overruns.

107

Figure 14: Type 1 Rail: Average cost overrun scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Project specific Project environment Client specific Procurement issues External issues

Cos

t ove

rrun

sco

re

6.2.16 We also considered the average completion times for rail projects. The results of the analysis are shown in Table 23. The table presents completion times for typical phases of a project and total completion times. The completion times are expressed in months per km, in order to make the results comparable across projects of different size.

6.2.17 Table 23 shows the average estimated and actual completion times for each phase of a typical rail projects. We note that, while some of the project phases are sequential, there is a degree of overlapping. Therefore, the total project completion time is not given by the sum of the completion times for each phase. In order to make the data comparable, we have standardized the measures of delay using the length of projects. The information is broken down by project phase to facilitate the identification of the project phases that tend to show the longest delay. While the actual delay for each phase depends on the actual project characteristics, the focus of this analysis is on the difference between estimated completion times.

6.2.18 In order to make different projects comparable, we have normalised completion times using project length in kilometres. However, we note that we have constructed this index purely for comparison purposes. These measures should not be extrapolated to estimate the likely completion times for similar projects. Specifically, completion times are likely to be made up of a fixed time component that does not change with project length in kilometres, and a variable time component that increases with it. This tends to make the times in months per kilometre less than the average for projects that cover a longer distance.

6.2.19 Of the various project development phases, funding and construction tend to accumulate the highest level of delay in percentage terms, 115.4% and 51.6% respectively. The project phases with the lowest amount of delay are planning and permissions.

108

Table 23: Rail: Estimated completion times, actual completion times, absolute delay and percentage delay

6.2.20 Using the same framework that we employed for cost overruns, we attempted to score and rank the main causes of project delay. Also in this case, ‘Project specific factors’ were the most important cause of time delays. The lack of data prevents us from exploring all the Type 2 factors underpinning these results. Nonetheless, we found that within ‘Authority Delays’ and ‘Late Commencement’ appear to be the most relevant factors. Regarding procurement Issues, ‘design changes’ were also a major cause of time delays.

Figure 15: Type 1 Rail: Average time delay scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Project specific Project environment Client specific Procurement issues External issues

Tim

e de

lay

scor

e

Project phase Estimated

completion time (months per kilometre)

Actual completion time

(months per kilometre)

Absolute delay (months per kilometre)

Percentage delay

(%)

Planning 1.2 1.6 0.4 36.4

Funding 0.5 1.0 0.5 115.4

Permissions 1.0 1.3 0.3 31.8

Site preparation 0.4 0.6 0.2 47.3

Construction 1.2 1.9 0.6 51.6

109

6.2.21 Using the same framework that we employed for cost overruns, we attempted to score and rank the main causes of delay. Also in this case, ‘Project specific factors’ were the most important Type 1 classification of time delays. We found that ‘Authority Delays’ and ‘Late Commencement’ appear to be the most relevant factors. Regarding procurement issues, ‘design changes’ were also a major cause of time delays.

6.2.22 Finally, Figure 16 provides a summary of the average score of all Type 2 factors for cost overruns and delay.

Figure 16: Type 2: Average score of cost overrun and time delay categories

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Design c

hanges

Legisl

ation c

hanges

Enviro

nmental

impac

t

Authorit

y dela

ys

Work su

spen

sions

Politic

al

Permits

/Con

sents

Site ch

arac

terist

ics

Late co

mmence

ment

Constr

uctio

n peri

od

Funding pro

blems

Stakeh

older n

umbe

rs

Techn

ology

Innovatio

n degree

Contra

ctor s

pecific

Design c

omple

xity

Contra

ct co

mplexity

Projec

t man

agem

ent

Inflatio

n

Site ac

cess

Public

relat

ions

Force m

ajure

Poor p

lannin

g

Supplie

r disp

utes

Busine

ss in

adeq

uacyOth

er

Economic

Exchan

ge rate

s

Scor

e

Cost overrun score Time delay score

110

6.3 Roads

6.3.1 The road sector has the second highest number of projects with data. It features 22 infrastructure projects out of 66 across all sectors, about 33 percent. This allows us to present a range of unit cost breakdowns in some detail.

6.3.2 For Level 1, project length in kilometres was missing for one project, preventing the calculation of the corresponding unit costs. Although there is a higher percentage of data in place than for rail, it is still far short of what would be required to present a full comparison of Level 1 and Level 2 unit costs.

Level 1

6.3.3 Figure 17 compares Level 1 ‘all-in’ estimated unit costs of all road projects, and Figure 18 compares actual unit costs. Where the data allowed it, the chart breaks down the cost categories making up the total cost of the project.

Figure 17: Level 1 Road: ‘All in’ estimated unit costs for all projects

0

10

20

30

40

50

60

70

80

DE02 - P

rojec

t A

DE02 - P

rojec

t B

DE03 - P

rojec

t A

ES04 - P

rojec

t A

GR03 - E

gnatia

GR05 - K

ifisso

s

GR03 - P

rojec

t A

GR04 - A

ttiki

GR05 - P

rojec

t A

GR05 - C

hrisoupoli

GR04 - P

rojec

t A

GR05 - S

iatist

a

IE02 - H

urlers

IE02 - W

aterg

rassh

ill

IE02 - A

shford

IE05 - B

allys

hannon

IE02 - B

allinco

llig

IT06 - O

riental

e

IT03 - C

arlo Feli

ce

PL05 - P

ulawy

PL05 - E

lbag

Uni

t cos

t (€m

/km

)


111

Figure 18: Level 1 Road: ‘All in’ actual unit costs for all projects

0

10

20

30

40

50

60

70

80

DE02 - P

rojec

t A

DE02 - P

rojec

t B

DE03 - P

rojec

t A

ES04 - P

rojec

t A

GR03 - E

gnatia

GR05 - K

ifisso

s

GR03 - P

rojec

t A

GR04 - A

ttiki

GR05 - P

rojec

t A

GR05 - C

hrisoupoli

GR04 - P

rojec

t A

GR05 - S

iatist

a

IE02 - H

urlers

IE02 - W

aterg

rassh

ill

IE02 - A

shford

IE05 - B

allys

hannon

IE02 - B

allinco

llig

IT06 - O

riental

e

IT03 - C

arlo Feli

ce

PL05 - P

ulawy

PL05 - E

lbag

Uni

t cos

t (€m

/km


6.3.4 As in the case of rail projects, the Level 1 unit costs for roads show a high degree of variation.

6.3.5 There are multiple reasons for higher unit costs. In our database, we found that mountainous terrain and, more generally, the complexity of a project seem to lead to high unit costs. However, other factors may be at play. For example, in one case part of the project was in wetland and protected reserves, featuring ‘sensitive soil and sand’. By contrast, low unit costs tend to be associated with less complex projects, often developed on easy terrain with an accommodating geology.

6.3.6 Very different project characteristics can lead to high unit costs. For example, project GR03 – Egnatia shows very high costs due to the largely mountainous characteristics of the terrain. On the other hand, project GR05 – Kifissos owes the high unit costs to the fact that it was developed in 100 per cent urban territory.

6.3.7 In the literature review, we have identified a set of unit cost benchmarks. As in the case of rail, we do not believe that these benchmarks are necessarily comparable with the projects in our database. This is because of the multiple factors that affect the unit costs of these projects, resulting in relatively large unit cost ranges. For a single carriageway two-lane road we found that the benchmarks range between 0.7 EURm/km to 2.0 EURm/km. For a double carriageway two-lane road, the benchmark values vary from between 2.7 EURm/km to 8.7 EURm/km. For a double carriageway three-lane road, we found the unit costs benchmarks range between 36.4 EURm/km and 91.3 EURm/km.

Level 2

6.3.8 The following charts compare Level 2 build unit costs for all road projects. Each chart summarises all the information available for its specific Level 2 unit costs. Not all information is always available for all projects. Some projects have either estimated data or actual data, but not both, for example. Where this is the case, we leave the missing data column blank.

112

6.3.9 Across all available Level 2 unit costs we observe a degree of variability similar to that of the Level 1 unit costs. As noted above, this variability is due to a variety of factors. For example, the fact that project GR05 – Kifissos is entirely urban leads to significantly higher pavement unit costs, probably due to the complexity of re-paving urban areas compared with rural ones. However, due to the small sample size, we are unable to identify specific causes for Level 2 unit cost differences, particularly for bridges and tunnels.

Figure 19: Level 2 Road: Estimated and actual unit ‘build’ cost of pavement

0

5

10

15

20

25

30

35

40

DE02 - P

rojec

t A

DE02 - P

rojec

t B

DE03 - P

rojec

t A

ES04 - P

rojec

t A

GR05 - K

ifisso

s

GR04 - A

ttiki

GR03 - P

rojec

t A

GR03 - E

gnatia

GR05 - C

hrisoupoli

GR04 - P

rojec

t A

GR05 - P

rojec

t A

IE02 - H

urlers

IE02 - W

aterg

rassh

ill

IE02 - A

shford

IE05 - B

allys

hannon

IT06 - O

riental

e

IT03 - C

arlo Feli

ce

PL05 - P

ulawy

PL05 - E

lbag

Uni

t cos

t (€m

/km

)

Estimated Actual

Figure 20: Level 2 Road: Estimated and actual unit ‘build’ cost of bridges

0.0

1.0

2.0

3.0

4.0

5.0

6.0

ES04 - ProjectA

GR03 - Egnatia GR05 - Siatista GR03 - ProjectA

GR05 - Kifissos GR05 - ProjectA

IE02 -Watergrasshill

IE02 -Watergrasshill

IT03 - CarloFelice

IT06 - Orientale

Uni

t cos

t (€0

00s/

m2)

Estimated Actual

113

Figure 21: Level 2 Road: Estimated and actual unit ‘build’ cost of tunnels

0

5

10

15

20

25

30

35

40

45

50

DE02 - Project A GR03 - Egnatia GR03 - ProjectA

GR05 -Chrisoupoli

GR04 - Ioannina IT06 - Orientale PL05 - Pulawy

Uni

t cos

t (€m

/km

)Estimated Actual


6.3.10 Using the framework described above, we ranked the Type 1 factors according to their perceived impact on costs overruns. Figure 18 presents the average score, across all road projects, given to different Type 1 explanations of cost overruns.

Figure 22: Type 1 Road: Average cost overrun scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Project specific Project environment Procurement issues External issues Client specific

Cos

t ove

rrun

sco

re

114

6.3.11 ‘Project specific factors’ were the most important cause of cost overruns. Among these factors, ‘design complexity’, ‘environmental impact’ and ‘authority delays’ were top-rated. ‘Project environment’ factors were also important.

6.3.12 We also considered the average completion times for each phase of a typical road projects. We note that, while some of the project phases are sequential, there is a degree of overlapping. Therefore, the total project completion time is not given by the sum of the completion times for each phase. Table 24 summarises the average estimated and actual completion times for road projects, normalised using the length of each project. Also in this case, the focus of the analysis is on the relative delay, rather than on the absolute level of time overrun.


6.3.14 The data show that, generally, site preparation and construction are the project phases which accumulate the longest delays, respectively about 28 per cent and 22 per cent of the original estimate. On the other hand, the funding phase does not seem to be affected by delay..

Table 24: Road: Estimated completion times, actual completion times, absolute delay, percentage delay

Project phase

Estimated completion

time (months per kilometre)

Actual completion


Absolute difference



(%)

Planning 2.1 2.6 0.4 19.2

Funding 0.7 0.7 0.0 0.0

Permissions 0.9 0.9 0.0 3.0



6.3.15 The table shows that in the case of roads, site preparation and construction are, in percentage terms, the project phases with the higher average delay. On average, across all construction phases, road projects tend to accumulate a delay of approximately 1.3 months per km.

6.3.16 Using the same framework that we employed for cost overruns, we attempted to score and rank the main causes of delay. Also in this case, ‘Project specific factors’ were the most important cause of time delays. Within these Type 1 categories, site access’, ‘authority delays’ and ‘construction period’ were the Type 2 factors attracting the highest scores.

115

Figure 23: Type 1 Road: Average time delay scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0


Tim

e de

lay

scor

e

6.3.17 Figure 24 provides a summary of the average score of all Type 2 factors for cost overruns and delay.

Figure 24: Type 2: Average score of cost overruns and time delays

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Site ch

arac

terist

ics

Design c

hanges

Contra

ct co

mplexity

Design c

omple

xity

Enviro

nmental

impac

t

Authorit

y dela

ys

Inflatio

n

Innovatio

n degree

Constr

uctio

n peri

od

Permits

/Con

sents

Site ac

cess

Legisl

ation c

hanges

Late co

mmence

ment

Contra

ctor s

pecific

Work su

spen

sions

Force m

ajure

Public

relat

ions

Supplie

r disp

utes

Stakeh

older n

umbe

rs

Poor p

lannin

gOth

er

Economic

Funding pro

blems

Projec

t man

agem

ent

Politic

al

Techn

ology

Busine

ss in

adeq

uacy

Exchan

ge rate

s

Scor

e


116

6.4 Urban Transport

6.4.1 The urban transport sector has only 8 projects with data, out of 66 across all sectors. We have fewer comparison points at Level 1 than for rail and road. Due to few data points, we are also unable to provide as many Level 2 graphs due to the scarcity of data.

6.4.2 The urban transport sector had 36% of key data in place such as project length in km, Level 1 cost components and environmental characteristics.

Level 1

6.4.3 Figure 25 compares Level 1 ‘all-in’ estimated unit costs of all urban transport projects, and Figure 26 compares actual unit costs. Where the data allows, the chart breaks down the cost categories making up the total cost of the project. We are missing actual data for the Irish urban transport project in the sample.

Figure 25: Level 1 Urban Transport: ‘All in’ estimated unit costs for all projects and metro and tram benchmarks

0

5

10

15

20

25

30

35

40

FR06 - T

CSP

GR03 - P

rojec

t A

GR04 - A

thens

IE01 - P

rojec

t A

PL06 - P

rojec

t A

PL05 - K

rakow

PL06 - L

odz

PT03 - P

rojec

t A

Uni

t cos

t (€m

/km

)


117

Figure 26: Level 1 Urban Transport: ‘All in’ actual unit costs for all projects and metro and tram benchmarks

0

5

10

15

20

25

30

35

40

FR06 - T

CSP

GR03 - P

rojec

t A

GR04 - A

thens

IE01 - P

rojec

t A

PL06 - P

rojec

t A

PL05 - K

rakow

PL06 - L

odz

PT03 - P

rojec

t A

Uni

t cos

t (€m

/km


6.4.4 With the exception of two outliers, most projects appear to show similar unit cost

ranges. The two projects with higher unit costs, both in Poland, were deemed to be ‘complex’. Interestingly, the Irish project had lower estimated unit costs than the Polish projects despite being described as ‘very complex’. We were unable to explain the reason for this discrepancy. One of the Polish projects (PL06 – Project A) experienced a large cost overrun, which was justified by the ‘degree of innovation’ on the project. It is, however, unclear which specific factors contributed to the large cost overrun.

6.4.5 In the literature review, we have identified a set of benchmarks also for urban transport projects, specifically for metro projects and tram projects. The range of unit costs for metro projects is very large, between 27.3 EURm/km and 482.9 EURm/km. Tram projects unit costs range between 2.5 EURm/km and 64.3 EURm/km. These very wide ranges are due to the heterogeneity of projects used to calculate the benchmarks. For example, the top end of the range is the unit cost of the Jubilee Line Extension project, on the London Underground. This project required four river crossing tunnels, which contributed to raising the project’s total cost. The multiple differences in project characteristics make it very difficult to use the unit cost benchmark values as reference points for the projects contained in our dataset.

Level 2

6.4.6 The following charts compare Level 2 unit ‘build’ costs for urban transport projects. Each chart summarises all the information available for its specific Level 2 unit costs. Estimated data are missing for some projects. Where this is the case, we leave the missing data column blank.

6.4.7 Figure 27 shows the estimated and actual unit build costs of trackwork. Most projects show relatively similar Level 2 unit costs. As in the case of Level 1 unit costs, one Polish project had much higher Level 2 unit costs than the other projects, but this project took place within a tunnel which will have increased the trackwork costs.

6.4.8 The graphs below show two Level 2 unit cost breakdowns for stations, one specified as EURm/m2 and the second as EURm/nr of stations. It can be seen that the cost per m2

118

of station is strongly affected by the type of project. For example, the project GR04 – Athens is an underground metro project, which implies much higher unit costs for each m2 of station built. Figure 31 shows the information we have for bridges, with estimated data for only one project. The information provided did not offer any explanation for the variability of unit costs.

6.4.9 The scarcity of data for these Level 2 components prevents us from drawing conclusions on the causes of unit cost variations between projects. Especially project PL06 – Project A was labeled as ‘complex’ but no further information was provided that could explain the very high unit cost of trackwork.

Figure 27: Level 2 Urban Transport: Estimated and actual unit ‘build’ cost of trackwork

0

5

10

15

20

25

FR06 - TCSP GR03 - Project A GR04 - Athens IE01 - Project A PL06 - Project A PL05 - Krakow PL06 - Lodz PT03 - Project A

Uni

t cos

t (€m

/km

)

EstimatedActual

119

Figure 28: Level 2 Urban Transport. Estimated and actual unit ‘build’ cost of stations

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

GR03 - Project A GR04 - Athens PL06 - Project A

Uni

t cos

t (€0

00s/

m2)

EstimatedActual

Figure 29: Level 2 Urban Transport. Estimated and actual unit ‘build’ cost of stations

0.0

0.1

0.1

0.2

0.2

0.3

FR06 - TCSP GR04 - Athens

Uni

t cos

t (€m

/nr)

EstimatedActual

120

Figure 30: Level 2 Urban Transport. Estimated and actual unit ‘build’ cost of bridges

0

2

4

6

8

10

12

14

16

FR06 - TCSP PT03 - Project A

Uni

t cos

t (€0

00s/

m2)

EstimatedActual


6.4.10 Also in this case, we ranked the Type 1 factors according to their perceived impact on costs overruns. Figure 31 presents the average score across all urban transport projects of Type 1 stated explanations for cost overruns.

Figure 31: Type 1 Urban Transport: Average cost overrun scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Project environment Procurement issues Project specific Client specific External issues

Cos

t ove

rrun

sco

re

6.4.11 ‘Project environment’ factors were the most important cause of cost overruns. Within

them, gaining ‘permits/consents’ appeared to be the top-rated cause of cost overruns

121

in six out of eight projects. Procurement issues also appeared to be a close second cause for urban transport cost overruns.

6.4.12 Table 23 summarises estimated and actual average completion times for a typical urban transport projects. We note that, while some of the project phases are sequential, there is a degree of overlapping. Therefore, the total project completion time is not given by the sum of the completion times for each phase. It shows absolute and percentage time delays, normalised using project length. The focus of the analysis is on relative delay, rather than absolute delays, as these would depend on the specific characteristics of the projects under analysis.

6.4.13 In order to make different projects comparable, we have normalised completion times using project length. However, we note that we have constructed this index purely for comparison purposes. These measures should not be extrapolated to estimate the likely completion times for similar projects. Specifically, completion times are likely to be made up of a fixed time component that does not change with project length in kilometres, and a variable time component that increases with it. This tends to make the times in months per kilometre less than the average for projects that cover a longer distance.

6.4.14 Funding appears to be the project phase most affected by time delays (60% more than the estimated completion time). This is followed by planning, taking approximately 38 per cent longer than originally conceived. Permissions, on the other hand, do not appear to be affected by delays significantly.

Table 25: Urban transport estimated completion times, actual completion times, absolute delay, percentage delay

Project phase



Actual completion


Absolute difference



(%)

Planning 2.3 3.1 0.9 37.7

Funding 1.1 1.7 0.7 60.1

Permissions 1.1 1.1 0.1 7.7



6.4.15 Finally, Figure 32 present the ranking of the Type 1 factors according to their perceived

impact on construction delay. Also in this case, ‘Project environment’ factors appeared to be the most important cause of cost overruns.

122

Figure 32: Type 1 Urban Transport: Average time delay scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Projectenvironment

Procurementissues

Projectspecific

Clientspecific

Externalissues

Tim

e de

lay

scor

e

6.4.16 Figure 33 provides a summary of the average score of all Type 2 factors for cost overruns and delay.

Figure 33: Type 2 Urban Transport. Average score of each overrun and delay category, descending by overrun score

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

Design c

hanges

Contra

ct co

mplexity

Permits

/Con

sents

Constr

uctio

n peri

od

Authorit

y dela

ys

Public

relat

ions

Contra

ctor s

pecific

Enviro

nmental

impac

t

Politic

al

Design c

omple

xity

Site ac

cess

Late co

mmence

ment

Busine

ss in

adeq

uacy

Innovatio

n degree

Stakeh

older n

umbe

rs

Inflatio

n

Funding pro

blems

Site ch

arac

terist

ics

Work su

spen

sions

Projec

t man

agem

ent

Poor p

lannin

g

Techn

ology

Supplie

r disp

utes

Exchan

ge rate

s

Legisl

ation c

hanges

Other

Economic

Force m

ajure

Scor

e


123

6.5 Water and Wastewater

6.5.1 We were able to collect data only on 4 water and wastewater projects out of 66 across all sectors. We have fewer comparison points at Level 1 unit cost than for rail and road and we could only calculate a small subset of Level 2 unit costs due to the lack of data. In one case we could not obtain the project length, which prevented us from calculating the project’s Level 1 unit costs.

Level 1

6.5.2 Figure 34 compares Level 1 ‘all-in’ estimated unit costs of all water and wastewater projects, while Figure 35 compares actual unit costs. Where the data allows, the chart breaks down the cost categories which constitute the total cost of the project. Data on actual costs is missing for two water and wastewater projects in Spain.

Figure 34: Level 1 Water and Wastewater: ‘All in’ estimated unit costs for all projects

0.0

0.5

1.0

1.5

2.0

2.5

3.0

ES02 - C.Jucar ES05 - Desaladora IT05 - A.Gela

Uni

t cos

t (€m

/km

)


124

Figure 35: Level 1 Water and Wastewater: ‘All in’ actual unit costs for all projects

0.0

0.5

1.0

1.5

2.0

2.5

3.0


Uni

t cos

t (€m

/km


6.5.3 Three out of four projects were described as ‘complex’, and the remaining project was

located in a ‘demanding environment’. In addition all the complex projects had some mountainous terrain, ranging from 15% to 80% of total project length. With one exception, the unit costs are broadly similar, perhaps reflecting similar challenges for each project. However, the small sample size prevents us from exploring the effect of specific factors on unit costs.

6.5.4 We have also identified some benchmarks for water projects, both for water treatment and water supply. However, the values we have identified in the literature review are much higher than the unit costs in our database, suggesting that these projects may not be directly comparable and useful as benchmarks. Moreover, these benchmarks are for cost relative to capacity (measured in m3/day) and, for this reason, not directly comparable to the unit costs identified above. For water treatment they range between 98 and 3,766 EURO/m3/day and for water supply between 370 and 6,800 EURO/m3/day.

Level 2

6.5.5 The following charts compare Level 2 build unit costs for water and wastewater projects. Each chart summarises all the information available for its specific Level 2 unit costs. Figure 36 compares estimated and actual unit costs for land.

6.5.6 Figure 37 compares estimated and actual unit ‘build’ costs for water supply. ES02 – Conduccion Jucar has high unit costs. This project is the oldest, starting in 2002. Otherwise, its features are similar to the other projects. It was complex, and involved some mountainous terrain.

125

Figure 36: Level 2 Water and Wastewater: Estimated and actual unit ‘build’ cost of land

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

ES05 - Desaladora ES02 - C.Jucar ES05 - Project A

Uni

t cos

t (€m

/ha)

EstimatedActual

Figure 37: Level 2 Water and Wastewater: Estimated and actual unit ‘build’ cost of water supply

0.0

0.5

1.0

1.5

2.0

2.5


Uni

t cos

t (€m

/km

)

EstimatedActual


6.5.7 Figure 38 presents the average score, across all water and wastewater projects, given to different Type 1 explanations for cost overruns. ‘Procurement issues’ appears to be perceived as the most important cause of cost overruns in water and wastewater projects. ‘Design changes’ were the top-rated Type 2 procurement factor.

126

Figure 38: Type 1 Water and Wastewater: Average cost overrun scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Procurement issues Project specific Project environment Client specific External issues

Cos

t ove

rrun

sco

re

6.5.8 Table 24 summarises estimated and actual average completion times for each phase of a typical water and wastewater projects. It also shows absolute and percentage time delays. We note that, while some of the project phases are sequential, there is a degree of overlapping. Therefore, the total project completion time is not given by the sum of the completion times for each phase.


6.5.10 Site preparation appears to be project phase with the highest average delay, followed by funding. Planning, permissions and construction phases took between 33% and 38% longer than estimated.

127

Table 26: Water and Wastewater: Estimated completion times, actual completion times, absolute delay, percentage delay

Project phase



Actual completion


Absolute difference



(%)

Planning 0.1 0.2 0.0 37.0

Funding 0.1 0.1 0.1 73.9

Permissions 0.2 0.2 0.1 33.1



6.5.11 Figure 39 presents the average scores on the Type 1 causes for water and wastewater project delays.

Figure 39: Type 1 Water and Wastewater: Average time delay scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Project environment Project specific Procurement issues Client specific External issues

Tim

e de

lay

scor

e

6.5.12 ‘Project environment’ factors appear to be the most important cause of time delays.

The top-rated factor was gaining ‘permits/consents’, which obtained an average score of two, indicating a major cause of delay (20-50 per cent).

6.5.13 Figure 40 provides a summary of the average score of all Type 2 factors for cost overruns and delays.

128

Figure 40: Type 2 Water and Wastewater. Average score of each overrun and delay category, descending by overrun score

0.0

0.5

1.0

1.5

2.0

2.5

Design c

hanges

Contra

ctor s

pecific

Design c

omple

xity

Innovatio

n degree

Politic

al

Techn

ology

Contra

ct co

mplexity

Permits

/Con

sents

Site ch

arac

terist

ics

Enviro

nmental

impac

t

Site ac

cess

Authorit

y dela

ys

Constr

uctio

n peri

od

Supplie

r disp

utes

Poor p

lannin

g

Stakeh

older n

umbe

rs

Legisl

ation c

hanges

Public

relat

ions

Inflatio

n

Funding pro

blems

Work su

spen

sions

Late co

mmence

ment

Busine

ss in

adeq

uacy

Projec

t man

agem

ent

Force m

ajure

Exchan

ge rate

s

Economic

Other

Scor

e


129

6.6 Energy

6.6.1 We have data on 8 projects for this sector, out of 66 across all sectors. For two projects, we were unable to obtain the project length in km, preventing us from calculating those projects’ Level 1 unit costs.

Level 1

6.6.2 Figure 41 compares Level 1 ‘all-in’ estimated unit costs of all energy projects excluding pipeline projects, and Figure 42 compares actual unit costs. Where the data allowed, the chart breaks down the cost categories which constitute the total cost of the project. Estimated cost data are missing for one of the Greek projects.

Figure 41: Level 1 Energy: ‘All in’ estimated unit costs for all projects excluding pipelines

0

1

2

3

4

5

6

7

8

9

GR05 - P

rojec

t A

PT05 - P

ampilh

osa

PT06 - E

.Gard

unha

PT06 - E

.Cara

mulo

Uni

t cos

t (€m

/nr)


130

Figure 42: Level 1 Energy: ‘All in’ actual unit costs for all projects excluding pipelines

0

1

2

3

4

5

6

7

8

9

GR05 - P

rojec

t A

PT05 - P

ampilh

osa

PT06 - E

.Gard

unha

PT06 - E

.Cara

mulo

Uni

t cos

t (€m

/nr)


6.6.3 We can explain most of the variation in unit costs by the project characteristics. For

example, two projects were ‘mostly straightforward’, three were ‘complex’, and one had ‘some complexity’. In addition, four of the projects were in 100% mountainous terrain. All the Portuguese projects, with higher unit costs, were wind farms built in 100% mountainous terrain. The Greek project, GR05 – Project A, on the other hand had 70% level terrain. This may help explain its very low unit cost.

6.6.4 The Portuguese projects are wind farm projects which show broadly similar unit costs. At this level of aggregation, it is not possible to identify the specific factors that would explain these differences.

6.6.5 Also in this case, we have analysed possible benchmarks in the literature review. However, the range of values we obtained is too wide to be used as a comparable reference point for the projects in the database.

Level 2

6.6.6 The following charts compare Level 2 build unit costs for energy projects. Each chart summarises all the information available for its specific Level 2 unit costs. Estimated cost data are missing for one project in Portugal. We leave the missing data column blank.

6.6.7 Figure 44 shows the build cost of turbines and Figure 45 shows the unit cost of land. Both British projects experienced a far higher actual unit ‘build’ cost for the gas supply than expected. This may have been due to the same reasons for Level 1 cost overrun, namely ‘delay by statutory authorities and/or contractors’ and concerns about the project’s environmental impact. The land cost is also much higher for the British project than the Portuguese projects perhaps reflecting regional price differences.

131

Figure 43: Level 2 Energy: Estimated and actual unit ‘build’ cost of turbines

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

PT05 - Pampilhosa PT06 - E.Gardunha PT06 - E.Caramulo PT06 - Pinhal

Uni

t cos

t (€m

/nr)

EstimatedActual

Figure 44: Level 2 Energy: Estimated and actual unit cost of land

0.00

0.02

0.04

0.06

0.08

0.10

0.12

GB02 - UK Pipeline PT06 - E.Gardunha PT06 - Pinhal PT06 - E.Caramulo

Uni

t cos

t (€m

/Ha)

EstimatedActual


6.6.8 Figure 46 presents the average score across all energy projects of Type 1 stated explanations for cost overruns. ‘Project specific’ factors were the most important cause of cost overruns. Within this category, the top-rated factor was the project’s ‘environmental impact’.

132

Figure 45: Type 1 Energy: Average cost overrun scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Project specific Procurement issues Project environment External issues Client specific

Cos

t ove

rrun

sco

re

6.6.9 Table 27 summarises estimated and actual average completion times for each phase of a typical energy projects. It also shows absolute and percentage time delays. We note that, while some of the project phases are sequential, there is a degree of overlapping. Therefore, the total project completion time is not given by the sum of the completion times for each phase.


6.6.11 We find that, on average, funding tends to be the phase most affected by delays with a completion time almost 72% longer than estimated. Construction, on the other hand, is the project phase with the shorted delay..

133

Table 27: Energy: Estimated completion times, actual completion times, absolute delay, percentage delay

Project phase



Actual completion


Absolute difference



(%)

Planning 0.4 0.5 0.1 14.2

Funding 0.2 0.3 0.1 71.8

Permissions 0.3 0.3 0.1 22.5



6.6.12 Figure 46 shows the ranking of the Type 1 causes of delay, whose average scores have been calculated using the Mott MacDonald framework. ‘Project specific’ factors were also the most important cause of time delays. Within this category for delays, the top-rated factor was ‘authority delays’.

Figure 46: Type 1 Energy: Average time delay scores

0.0

0.5

1.0

1.5

2.0

2.5

3.0


Tim

e de

lay

scor

e

6.6.13 Figure 47 provides a summary of the average score of all Type 2 factors for cost overruns and delays.

134

Figure 47: Type 2 Energy: Average score of each overrun and delay category, descending by overrun score

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Enviro

nmental

impac

t

Site ac

cessOth

er

Site ch

arac

terist

ics

Design c

omple

xity

Design c

hanges

Contra

ct co

mplexity

Supplie

r disp

utes

Innovatio

n degree

Authorit

y dela

ys

Late co

mmence

ment

Constr

uctio

n peri

od

Work su

spen

sions

Contra

ctor s

pecific

Funding pro

blems

Legisl

ation c

hanges

Force m

ajure

Stakeh

older n

umbe

rs

Projec

t man

agem

ent

Techn

ology

Public

relat

ions

Permits

/Con

sents

Politic

al

Inflatio

n

Exchan

ge rate

s

Poor p

lannin

g

Economic

Busine

ss in

adeq

uacy

Scor

e


6.7 Time delay and cost overrun analysis

6.7.1 After analysing each type of projects separately, we bring together the results of the analysis of cost overruns and delays to carry out a comparison across project types and countries. This analysis is mainly descriptive due to the small sample size. To be sure, the results presented here are specific to the sample of projects in the database and cannot be deemed to have general validity.

Cost overruns

6.7.2 Table 28 shows the average scores for the Type 1 causes of delay, calculated using the Mott MacDonald framework described in Section 4. As noted earlier, we asked respondents to score each category from 0 to 3. A score of 0 means ‘no or insignificant cause of delay’, 1 means a ‘minor factor’ (less than 20 per cent), 2 means a major factor (20-50 per cent) and 3 means a ‘very significant factor’ (greater than 50 per cent).

135

Table 28: Level 1 Cost overruns: Average score (0-3) by sector

Level 1 category Rail Road Urban Water Energy

Procurement issues 0.8 0.6 1.1 1.1 0.6 Project specific 1.0 0.7 1.0 0.9 0.8 Client specific 0.8 0.3 0.9 0.5 0.3 Project environment 0.9 0.6 1.1 0.8 0.5 External issues 0.5 0.4 0.4 0.4 0.4

6.7.3 The table shows that, across all groups, project specific factors tend to be the lead cause for cost overruns. For some urban transport and water project, procurement issues also appear to rank among the highest causes of cost overruns. For urban transport and rail projects, issues with the project environment appear to contribute to cost overruns more than for projects in other sectors.

Delays

6.7.4 Table 29 summarises the percentage delays by project phase and sector. On average, rail projects in our database appear to be those with the highest level of overall delay across all project phases, followed by water and wastewater projects with an average overall delay of 32 per cent. Road projects seem to accumulate the smallest delay. We note that, while some of the project phases are sequential, there is a degree of overlapping. Therefore, the total project completion time is not given by the sum of the completion times for each phase.

6.7.5 With regards to specific project phases, funding appears to be the phase with the highest level of delay (with the exception of roads). For rail projects, the construction phase is also characterised by a relatively high average delay (about 50 per cent). For water projects, the permissions phase appears to be affected by long delays, approximately 82 per cent of the estimated completion times.

Table 29: Percentage delays by project phase and sector

Project phase Rail

(%)

Road

(%)

Urban

(%)

Water

(%)

Energy

(%)

Planning 36.4 19.2 37.7 37.0 14.2

Funding 115.4 0.0 60.1 73.9 71.8

Permissions 31.8 3.0 7.7 33.1 22.5

Site preparation 47.3 27.7 18.4 153.2 21.2

Construction 51.6 22.0 13.4 37.9 11.0

6.7.6 Table 30 presents a summary of the average scores that the respondents to our questionnaires gave to Type 1 delay factors. Project specific issues appear to be

136

among the key factors explaining time delays in all sectors. Issues associated with the ‘project environment’ and ‘procurement’ also rank high among the causes that have been identified by the questionnaire respondent. Finally, client specific and external issues appear to have a less prominent role.

Table 30: Level 1 time delays: Average score (0-3) by sector

Level 1 category Rail Road Urban Water Energy

Procurement issues 0.9 0.8 1.1 1.2 0.5

Project specific 1.0 0.9 1.1 1.2 0.8

Client specific 0.9 0.4 0.7 0.8 0.3

Project environment 1.0 0.9 1.3 1.5 0.7

External issues 0.5 0.5 0.4 0.4 0.4

Interaction between cost overruns and delays

6.7.7 In this subsection we briefly explore the possible relationship between cost overruns and time delays. In order to use the largest sample possible, we grouped together all the projects in the database.

6.7.8 Figure 48 shows a scatter plot of all observations. It can be seen that most observations are scattered around the origin, showing no apparent relationship between cost overruns and delays.

6.7.9 This appears to be confirmed by the very low correlation coefficient between cost overruns and delays (0.3 per cent). Note, however, that the statistical significance of the results is low as they are based on relatively small samples. This conclusion therefore cannot rule out the existence of this relationship in a larger sample.

137

Figure 48: Cost overrun per km against absolute delay per km

Correlation between cost overruns and time delays

-5.0

-3.0

-1.0

1.0

3.0

5.0

7.0

9.0

-10 -5 0 5 10 15

Cost overruns

Tim

e de

lays

6.7.10 Finally, we checked whether the data we collected showed any differences across countries both in terms of cost overruns and delays. If the data sample was sufficiently large, this analysis would allow testing whether country-specific effects may exist. In this case, the analysis cannot go beyond its descriptive function, as no rigorous statistical analysis can be carried out, given the small sample size (in some cases only a single observation was available).

6.7.11 Table 31 shows the results of the analysis of cost overruns grouped by sector and by country. Table 32 shows the corresponding results for time delays. The table show both average cost overruns and time delays in percentage terms, to ensure comparability across countries and sectors. Positive values indicate cost or time overruns. Conversely, negative values indicate cost savings and actual completion times lower than expected. The number of observations for each country-sector combination is shown in brackets.

6.7.12 In neither case, it does not appear possible to identify patterns or regularities in the data, suggesting that no country-specific effects exist. However, we stress the fact that the small sample size does not allow us to treat these results as statistically significant.

138

Table 31: Cost overrun summary by country and sector – average percentage differences between estimated and actual costs

(%)/(number of projects) Rail Road Urban

Transport Water Energy Weighted

average by sector

Germany -4.3% (6) -10.0% (3) -6.2%

Spain 12.8% (6) 30.7% (1) 17.4% (2) 15.8%

France 32.9% (1) 32.9%

Great Britain 110.7% (1) 110.7%

Greece 74.3% (2) 19.7% (8) 20.1% (2) 0.0% (1) 26.6%

Ireland 2.1% (5) 74.1% (1) 14.1%

Italy 62.4% (5) -5.0% (2) -0.9% (1) 37.6%

Poland 19.7% (2) 80.9% (2) 50.3%

Portugal 9.0% (1) 3.3% (4) 4.4%


26.9% 9.4% 45.4% 11.3% 20.7% 21.2%


(%)/ (number of projects)

Rail Road Urban Transport Water Energy

Weighted average by

sector

Germany 40.2% (6) 4.7% (3) 28.4%

Spain 15.3% (6) 27.3% (1) 55.9% (2) 25.7%

France 4.9% (1) 4.9%


Greece 24.4% (2) 17.8% (7) 13.2% (2) 12.6% (1) 17.7%

Ireland 9.0% (5) 52.2% (1) 16.2%

Italy 88.4% (1) 88.4%

Poland 5.9% (1) 2.7% (2) 3.8%

Portugal 258.3% (1) 41.5% (4) 84.9%


25.8% 13.2% 49.6% 66.7% 29.8% 26.2%

139

6.7.13 As noted previously, Flyvbjerg et al (2003) found that 9 out of 10 projects, from a sample of 258, across 20 countries and 5 continents, were subject to cost overruns.96 97 Our analysis found that, for those projects were the data was available, 75% of projects were subject to some delay, and 51% were subject to cost overrun. However, it is important to note that this was not a representative sample, and the results for all projects could be significantly different.

6.8 The role of ex-ante risk assessment

6.8.1 Cost overruns and delay may also be due to the lack of a thorough ex-ante risk analysis. This analysis, if carried out appropriately, could identify the key risk factors for a project. Risk mitigation strategies could then be devised to minimise any impact on project costs and completion times.

6.8.2 To gather some insights on whether and to what extend ex-ante risk assessment is carried out in the projects under analysis, we have reviewed a small sample of applications across four of the five infrastructure sectors we considered (excluding energy).

6.8.3 Our high-level review found that, in most cases, some form of ex-ante risk analysis is carried out. Generally, the analysis focuses on the effects of some changing some key variables on the economic and financial outcomes of the projects. These are normally measured using the Internal Rate of Return (IRR), the Net Present value (NPV) and the Benefit Cost ratio (BCR).

6.8.4 We found that the analysis tends to be simple and focused on some key variables. These include traffic levels (in transport projects), demand growth and some running costs such as maintenance. In some cases, variables like input prices, tariffs and revenues have also been considered in the risk analysis. In most cases this analysis also explore the extent to which the key variable need to change to alter the outcome of the project (e.g. to turn the NPV negative).

6.8.5 The majority of the documents we reviewed include a quantification of the effect of the different variables considered on the IRR, NPV and B/C. However, while most documents present the result of the sensitivity analysis, the majority fails to report the ranges used for the key variables. Only in a small subset of cases these have been made explicit.

6.8.6 In most cases the results of the sensitivity analysis are presented separately for each variable. We could not find any evidence of covariance analysis in most of the projects we reviewed. Only in a very small subset of cases, a section of the analysis is dedicated to exploring the joint impact of some combinations of key parameters.

6.8.7 While some form of ex-ante risk assessment or sensitivity analysis appear to be present in most cases, we could not find any clear evidence of these results having been used to devise a set of risk-mitigation strategies to reduce the likelihood of cost overruns and delays.

96 See footnote 25. 97 Paragraph 3.4.2

140

7 PRODUCTIVE INVESTMENTS: COST PER JOB CREATED

7.1 Introduction

7.1.1 In this section, we present the results of our examination of productive investment projects. We analysed the cost and job creation estimates contained in the ERDF applications and compared them with outturn data that was provided about the projects by the Member States.

7.1.2 We sought data on 45 projects and had a target sample size of 40. However, due to a poor response from the Member States (albeit less poor than in the case of infrastructure projects), we have received data on only 30 projects. The data in this section is presented for projects in Austria, Germany, Spain, France, Portugal and the UK.

7.1.3 For consistency, the data presented from the applications has been adjusted for inflation to a common 2007 price base. We did this in a similar manner for project cost estimates and actuals by assuming that costs would be incurred at the midpoint between the (estimated or actual) start and finish dates for the project.

7.1.4 The sample is dominated by manufacturing projects. The majority involves expansion of existing manufacturing facilities, but there are also several ‘new build’ projects. We also received data relating to one R&D-related investment and one investment fund project from the UK.

7.1.5 This Section is structured as follows:

• Subsection 7.2 considers the definition and methods used to measure employment effects for our sample projects

• Subsection 7.3 examines the estimated total cost of projects from the information provided in ERDF applications and compares these estimates with actual total costs.

• Subsection 7.4 examines the estimated number of jobs created that the projects were expected to yield and compares these with the actual figures.

• Subsection 7.5 illustrates and describes estimates and actual cost per job created figures that are implied from the information presented in subsections 7.3 and 7.4.

• Subsection 7.6 examines the role of project delays and durations in causing discrepancies between cost estimates and outturns.

• Subsection 7.7 assesses the potential role that different types of funding have in explaining the levels of expected costs.

• Subsection 7.8 examines the amounts of ERDF funding per job created.

7.2 Methods used to estimate employment effects of productive investments

7.2.1 The ERDF application forms typically provided forecasts of the number of full-time equivalent (FTE) jobs either created and / or safeguarded as a result of the investment and, because the NACE sector codes tend not be provided, we have only been able to establish the nature of the investment, for example, whether the investment is in manufacturing or R&D.

141

An example from the UK

7.2.2 Table 33 shows the projected employment effects detailed in section 8 of the ERDF application form for the UK productive investment Doncaster-Sheffield Airport and Business Zone.

Table 33: Employment effects projected for the Doncaster-Sheffield Airport productive investment

Number of jobs directly created 2430

- of which during operational phase 2430

- average duration Permanent

Number of jobs indirectly created 35

- of which during project implementation 35 *

Number of jobs safeguarded 1620 ** * Construction jobs based on the contractor’s estimate of 1500 weeks of employment ** There were already a significant number of businesses on the site, many of which could have been lost if the airfield had been put to an alternative use.

7.2.3 The method used to arrive at these projections was outlined in a supporting cost-benefit analysis report. A summary of that method is provided in the following paragraphs.

7.2.4 The starting point was a projection of annual air transport movements (ATMs) and passenger throughput at the airport. These were 57,000 ATMs and 2.33 million passengers per annum by 2014.

7.2.5 They then considered low-end and high-end estimates of the job impacts and noted that the impacts projected by opponents of the project at Public Inquiry all fell within this range, which could, therefore, be considered conservative. The low-end and high-end values were calculated using an industry standard employment density ‘rule-of-thumb’, that is, 1,000 direct operational jobs per 1 million passengers per annum (mppa). However, this was adjusted upwards to 1,100 to take account of the findings of an Airports Council International report specifically for UK airports.

7.2.6 For the low-end estimate, they adopted an average productivity growth assumption of 3 per cent per annum, which reduced the 1,100 jobs in 1998 to 675 per 1 mppa by 2014. For the high-end estimate, they adopted an average productivity growth assumption of 1 per cent per annum, which reduced the 1,100 jobs in 1998 to 935 per 1 mppa by 2014.

7.2.7 The multiplier used was a combined direct and induced employment multiplier of 0.3 for South Yorkshire. A deliberately conservative approach was adopted because leakages from a sub-regional economy could be significant and because the risk of double-counting would be minimised. In other words, indirect employment effects were estimated according to the multiplier approach as well as in part by including them in the direct airport-related employment figures.

7.2.8 They assumed that, by 2014, only 3 per cent of the airport’s traffic would be displaced from Sheffield City Airport, since very little of this traffic would otherwise use Sheffield. However, they assumed in the high-end estimate that all of the alternative use employment is displaced activity from elsewhere in South Yorkshire. That is, all the jobs could be accommodated elsewhere in South Yorkshire if they were not at the DSA

142

site. The low-end estimate assumed that only 50 per cent of these jobs are displaced from the rest of South Yorkshire.

Examples from Italy and Spain

7.2.9 Table 34 shows equivalent employment effect projections for an Italian productive investment.

Table 34: Employment effects projected for an Italian productive investment

Number of jobs directly created 351

- of which during operational phase 351

- average duration Permanent

Number of jobs indirectly created 150

- of which during project implementation 150

Number of jobs safeguarded 530 * * Before the investment (30-Jun-00) there were already 895 employees. 530 of these were on fixed term contracts which, because of the investment became permanent jobs.

7.2.10 TDIT already had 895 people working at the factory, of which 530 had fixed term contracts which should become permanent with the investment. The new investment itself will create 351 new permanent posts and totals €155.2m and this should be fully operational by 2004. These projections were based on a proposed capital/worker factor of 0.8 and the Commission had no objections to this as it would appear to lead to a higher labour productivity.

7.2.11 For most of the Spanish productive investments on which we had information, most projected employment effects were made on the basis of the experience of the company with its existing operations.

Comparability of projects

7.2.12 The information on the measurement of employment effects is very limited. We found 5 or 6 productive investments where the applicant had provided a small amount of detail in the appropriate section of the ERDF application form. Even fewer had supporting cost-benefit analysis reports, where more detail about methodologies was set out.

7.2.13 Where some detail was provided, employment effect projections tended to be based on industry-specific rules of thumb. Unfortunately, in most cases, we did not have enough information about the projects to carry out a robust assessment of the comparability between them.

7.3 Total cost of productive investment projects in our sample

7.3.1 We begin this subsection by presenting the orders of magnitudes of the costs involved in the productive investments on which we have received information and that we have

143

examined. We begin by assessing ‘new build’ manufacturing projects, followed by the expansion-based ones.

‘New build’ manufacturing projects

7.3.2 The magnitudes and breakdowns of the costs involved in our nine new build manufacturing projects are illustrated in Figure 49 below. These projects have only resulted in the creation of new jobs, as opposed to the expansion-based projects which have also led to the preservation of existing jobs.

7.3.3 Five of the projects are from Germany, where total forecasted costs did not exceed €200 million. In all cases, plant and equipment costs constitute the greatest proportion of total cost.

7.3.4 Three of the projects are from Spain and, whilst two of these projects were, in relative terms, low cost, project ES04 cost in excess of €400 million. The largest cost components for this project were professional fees and building costs. This contrasts with the other Spanish and Portuguese projects, which, like the German projects, were predominantly composed of plant and equipment costs.

Figure 49: Total cost estimates and cost breakdowns for ‘new build’ projects

0

50

100

150

200

250

300

350

400

450

DE

16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

22

ES

23

PT2

7

Project reference

€ M

illio

n

OtherProfessional fees Training IT infrastructureSite preparation Land acquisition Plant & equipment Construction

7.3.5 The ‘other’ category of cost in Figure 49 includes costs associated with planning and

design and the purchase of used machines and equipment, as well as IT infrastructure, professional fees and training that could not be allocated. In ES23, these ‘other’ costs are attributed to engineering and intangible services.

7.3.6 Figure 50 below shows the net position, as measured by subtracting the total cost forecasts from the actuals for each of the new build projects. The bars above the line (or x-axis) are, therefore, what might be described as cost overruns, while those below the line show cost savings.

144

Figure 50: Cost variance for new-build projects (actual minus forecast costs)

-40

-35

-30

-25

-20

-15

-10

-5

0

5

10D

E16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

30

ES

31

PT32

Euro

mill

ions

7.3.7 Figure 51 below shows the overruns or savings in each of the components of cost,

which, when aggregated, equate to the total cost overruns or savings in Figure 50 above. We were not provided with a breakdown of the cost differences for PT27.

145

Figure 51 Cost variances for individual cost components of new-build projects

-70

-60

-50

-40

-30

-20

-10

0

10

20D

E16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

30

ES

31

PT32

Euro

mill

ions

Construction Plant & equipment Land acquisition Site preparation Professional fees Other

7.3.8 Actual total costs, expressed as a percentage of the estimated total costs, are

presented in Figure 52. This shows that five projects cost slightly more than anticipated, whilst four cost less than expected.

7.3.9 The most dramatic cost saving was experienced on project PT32, which experienced an almost 50 per cent lower cost than expected. Figure 51 shows that this project had the largest cost saving in plant and equipment costs. However, in the case of other projects, what appear to be unexpected plant and equipment costs is also a significant contributor to cost overruns.

146

Figure 52 Actual cost divided by estimated cost of new build, shown as a percentage

0%

20%

40%

60%

80%

100%

120%

DE

16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

30

ES

31

PT32

%

Expansion-based manufacturing projects

7.3.10 The total costs and breakdowns for manufacturing projects that involve expansion are presented next, in Figure 53 below.

7.3.11 First, we note the comparison between the orders of magnitude in Figure 49 and Figure 53. While thirteen of the expansion projects in Figure 53 are of the same order of magnitude as the new build projects, there are three that are of the same order of magnitude as the largest of the new build projects above, and a further one that far exceeds it. Further, one project (ES24) had total cost of only €0.2m; far less than any other project of its kind.

7.3.12 Plant and equipment costs dominate these projects. This is to be expected as the projects likely involve the expansion of existing capacity. The data might be deemed consistent with this line of argument, if one considers that the range of actual building costs involved is between €8 million and €45 million, which could be expected to be broadly consistent with the costs of manufacturing buildings (or building expansions) of different sizes and specification.98

98 It might be relevant to note, perhaps, that we have to assume that the configuration of existing capacity – in terms of buildings and fixtures/fittings that meet the needs of existing plant – is efficient. If this were not the case, we would expect the company to have commenced with a ‘new build’ instead.

147

Figure 53: Total cost estimates and breakdowns for expansion projects

0

100

200

300

400

500

600

AT0

1

FR18

DE

15

ES

10

ES

08

ES

07

ES

06

ES

02

ES

05

ES

03

PT2

8

PT2

9

PT2

6

ES

24

ES

25

ES

21

ES

09

ES

20

Project reference

€ M

illio

n

Other

Professional fees

Plant & equipment

Construction

7.3.13 In Figure 54, we present actual total costs expressed as a percentage of the estimated total costs. As above, a figure below 100 per cent indicates a cost saving for the project.

7.3.14 We note the differences in the orders of magnitude between Figure 54 and the analogous graph for new build projects in Figure 52 above. For these projects the scale of cost overruns and cost savings is generally greater, in proportional terms, than for ‘new build’ projects. In the case of project ES08, the total project cost was €114 million higher than expected, whilst for ES03, the project cost less than expected by more than €66 million. This compares with the largest cost saving for a new build of less than €34 million and the largest overrun of less than €4.5 million.

148

Figure 54 Actual cost divided by estimated cost of expansion projects, shown as a percentage

0%

20%

40%

60%

80%

100%

120%

140%AT

01

FR18

DE

15

ES

10

ES

08

ES

07

ES

06

ES

02

ES

05

ES

03

PT28

PT29

PT26

ES

24

ES

25

ES

21

ES

09

ES

20

%

7.3.15 Figure 55 below shows the overruns or savings in each of the components of cost, which, when aggregated, equate to the total cost overruns or savings that are presented in percentage terms in Figure 54 above.

7.3.16 As in the case of new build projects, considerable unexpected plant and equipment costs (or savings in one case) were the main causes of total cost overruns (or total savings).

149

Figure 55 Cost variance and cost breakdowns for expansion projects

-200

-150

-100

-50

0

50

100

150AT

01

FR18

DE

15

ES

10

ES

08

ES

07

ES

06

ES

02

ES

05

ES

03

PT28

PT29

PT26

ES

24

ES

25

ES

21

ES

09

ES

20

Euro

mill

ions

Construction

Plant & equipment

Professional fees

Other productive investment projects

7.3.17 As well as the 9 new build manufacturing and the 18 expansion-based manufacturing projects presented in Figure 49 and Figure 53, we have information on three additional projects.

7.3.18 A manufacturing project involving rehabilitation in Portugal, which is still ongoing, has an expected inflation-adjusted project cost of €66.7m with the vast majority of this devoted to plant and equipment costs. The relatively low cost compared to the new build and expansion-based manufacturing projects is consistent with this project involving rehabilitation. The total estimated inflation-adjusted cost of a Research and Development projects from France was €318.8 million. This was presented as all R&D expenditure.

7.3.19 Finally, a UK-based investment fund with total project costs of €146.1m. The objectives of the fund were to indirectly affect employment in the area in several sectors. The project experienced a slight cost overrun of €90,000. Further details are shown in Figure 56 below.

150

Figure 56: UK Merseyside Special Investment Fund Expenditure

0

10

20

30

40

50

60

Equity fund investment

Seed corn fund

Mezzanine Loan fund investment

Small firm Loan fund investment

Investment support module

Management development

Networking

Start up module

Interest rate Rebates

Consultancy

Overheads

Salaries

Specialist Costs

Travel/Subsistence

Euro

mill

ions

7.4 Job creation analysis

7.4.1 In this subsection, we present the corresponding job creation estimates for the projects we have analysed and compare them with actual job creation figures. We present the results in the same order as above.


7.4.2 Job creation estimates for the new build manufacturing projects are illustrated in Figure 57 below. As can be seen, the estimated job creation numbers range from between 31 to 350 FTEs.

7.4.3 Six of these projects were forecast to only lead to the creation of new jobs, rather than also to the safeguarding of existing jobs. The exception was ES23, for which it was estimated that 31 jobs would be created, whilst 253 jobs would be safeguarded.

151

Figure 57: Estimates of full-time equivalent (FTE) jobs created by ‘new build’ projects

0

50

100

150

200

250

300

350

DE

16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

23

Jobs

7.4.4 The differences between actual and forecasted numbers of jobs created are shown in Figure 58 below. It shows that four out of the seven projects appear to have created exactly the number of jobs projected.

7.4.5 Otherwise, two of the German projects resulted in the creation of fewer jobs than expected. The most notable of these is DE14, which created only 31 per cent of the jobs that were anticipated. Taken together with the information on the net position of this project above (i.e. a cost overrun), we can anticipate significant implications for the cost per job created.

152

Figure 58 Variance in number of jobs created by ‘new build’ projects

-250

-200

-150

-100

-50

0

50D

E16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

22

ES

23

PT27

Jobs


7.4.6 Equivalent estimates for the expansion-based manufacturing projects are presented in Figure 59 below. The range of job creation estimates is largely analogous to those for ‘new build’ projects, bar three exceptions; FR18, PT29 and PT26.

153

Figure 59: Estimates of FTE jobs created for expansion projects

0

200

400

600

800

1000

1200

1400

1600AT

01

FR18

DE

15

ES

10

ES

08

ES

07

ES

06

ES

02

ES

05

ES

03

PT28

PT29

PT26

ES

24

ES

25

ES

21

ES

09

Num

ber o

f job

s

7.4.7 All of these expansion-based projects involved the safeguarding of existing jobs. The stated numbers of jobs projected to be safeguarded were, in some cases, broadly equivalent to the job creation estimates, suggesting straightforward doublings of capacity.

7.4.8 As can be seen from Figure 59 above, there are no job creation estimates presented in the applications for three of the projects. These involved the safeguarding of jobs only.

7.4.9 Figure 60 below shows that the majority of projects resulted in the creation of exactly the numbers of jobs that were originally expected. Four projects (three Spanish and the single Austrian) actually resulted in the creation of more jobs than forecast. (Note that Figure 60 is analogous to Figure 58 above, except expressed in proportional terms).

154

Figure 60 Actual minus forecast number of jobs created by expansion projects

0%

20%

40%

60%

80%

100%

120%

140%

AT0

1

FR18

DE

15

ES

10

ES

07

ES

02

ES

05

ES

24

ES

25

ES

21

ES

09

Jobs

Other projects

7.4.10 For the additional R&D-based project (FR17), the ERDF application estimated that the project would result in the creation of 415 FTEs.

7.4.11 Job creation numbers were not provided for the Portuguese manufacturing rehabilitation project (PT30). However, the UK Merseyside Special Investment fund (GB19) projected the creation of 5,804 jobs – significantly more than any of the other projects, although only 3,015 of these were actually created. This was in addition to the safeguarding of 2,439 jobs.

7.5 Cost per job created

7.5.1 In this subsection, we present data on the estimated and actual cost per job created as a result of these investments.


7.5.2 Figure 61 shows the forecasted and actual cost per FTE job created as a result of our new build manufacturing projects.

7.5.3 The range of cost per job estimates suggested by the data is wide, between €500,000 and nearly €3.5 million per job created. However, we note that the estimates for the majority of projects presented lie in a range from €500,000 to about €2 million. The outlier is the high-cost project from Spain identified in subsection 7.3 above. This is because the total cost of the project is very high for a relatively moderate amount of new jobs created.

7.5.4 Figure 61 shows that the largest deviation from what was expected occurred on project DE14, which exceeded its expected cost per job created by €1.4 million. This is the

155

same project on which we observed a relatively small cost overrun, but a significant shortfall in the number of jobs created relative to projections.

7.5.5 Project ES04 resulted in the creation of slightly more jobs than expected and, because the project cost less than expected (a cost saving of about €34 million), the outturn cost per job created was also lower than expected.

Figure 61: Estimated versus actual cost per job created by new build projects

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

DE

16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

23

Euro

mill

ions

EstimateActual


7.5.6 In Figure 62 below, we present the forecasted and actual cost per job created by expansion-based manufacturing projects.

7.5.7 We note the smaller range of estimated cost per job created of between €200,000 and €1.4 million. While there is significant variation within that range, estimates for the majority of projects presented do, however, lie in the range of €350,000 to €800,000 per FTE created.

7.5.8 The data in Figure 62 shows that cost per job created lies predominantly below €1.5 million. There are, however, two outliers (ES21 and ES25) with costs per job in excess of €3 million.

156

Figure 62: Estimated versus actual cost per job created by expansion projects

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0A

T01

FR18

DE

15

ES

10

ES

08

ES

07

ES

06

ES

02

ES

05

ES

03

PT2

8

PT2

9

PT2

6

ES

24

ES

25

ES

21

ES

09

Euro

mill

ions

Estimate

Actual

Other productive investment projects

7.5.9 For the French R&D-based project, an estimate of €735,000 per FTE job created was found in the ERDF application.

7.5.10 The cost per job created as a result of the UK special investment fund project was €45,000, compared to an estimated cost of around €23,000. As noted above, this project experienced only a slight cost overrun but created only 3,015 jobs out of a predicted 5,804.

7.6 The role of delays in causing discrepancies

7.6.1 In this subsection, we assess whether project delays can be identified as playing a role in discrepancies, where they were observed, between the actual and forecast costs of productive investments.


7.6.2 We begin, as before, with an examination of ‘new build’ manufacturing projects, for which the projected and actual starting dates, are illustrated in Figure 63 below. Under normal circumstances, one would expect a project that begins late to cost more because, with the passage of time, comes rising input prices.

7.6.3 Two of the three projects on which cost savings were achieved also started on time (we do not have data for the third). Meanwhile, two of the projects that delivered total cost overruns also started late.

157

Figure 63: Actual and projected start dates of ‘new build’ projects

24/07/1998 06/12/1999 19/04/2001 01/09/2002 14/01/2004 28/05/2005

DE16

DE14

DE13

DE12

DE11

ES04

ES30

Proj

ect r

efer

ence

num

ber

Date

Estimate project startdate

Actual project start date

7.6.4 In Figure 64 below, we show the differences between actual and forecasted project durations. In this case, bars above the line indicate that the project has taken longer than forecast, while bars below the line indicate a project that has been completed sooner than expected.

Figure 64: ‘New build’ project durations (actual minus projected project durations)

-10.0

0.0

10.0

20.0

30.0

40.0

50.0

DE

16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

22

ES

23

€m

158

7.6.5 The relationship between project durations and cost overruns is less discernible. Two of the projects that took longer than their estimated durations experienced cost savings. Likewise, both projects that had shorter durations than estimated experienced cost overruns. Only two projects were completed according to the anticipated timetable, both of which achieved cost savings.


7.6.6 The projected and actual starting dates of our expansion-based manufacturing projects are shown in Figure 65. As can be seen, all of these projects (bar one) started on time.

Figure 65: Actual and projected start dates of expansion projects

15/06/1994 28/10/1995 11/03/1997 24/07/1998 06/12/1999 19/04/2001 01/09/2002 14/01/2004 28/05/2005 10/10/2006 22/02/2008

AT01

DE15

ES08

ES06

ES05

ES24

ES21

GB19

Proj

ect r

efer

ence

num

ber

Date

Estimated project startdateActual project start date

7.6.7 The differences between actual and forecast project durations for these projects are

illustrated in Figure 66 below. Six of the projects were of a longer duration than anticipated, mostly by between 4 and 9 months, but two by about 2 years.

7.6.8 The analysis reveals that one of these two projects with a much longer duration than anticipated also suffered a significant overrun on cost, while the other finished on budget. The relationship is less discernible for the rest of the projects that took longer than expected.

159

Figure 66: Expansion project durations (actual minus projected start dates)

0

5

10

15

20

25

30AT

01

FR18

DE1

5

ES1

0

ES0

8

ES0

7

ES0

6

ES0

2

ES0

5

ES0

3

PT28

PT29

PT26

ES2

4

ES2

5

ES2

1

ES0

9

ES2

0

Euro

mill

ions

7.7 The type of funding that make up productive investments

7.7.1 Most projects have various sources of funding, including private investment and other local, regional and state funding in the various Member States. Figure 67 below shows the proportions of three different types of funding used in all of the productive investments that we have examined.

7.7.2 Most of the projects received the majority of funding from the private sector, such as FR18, FR17 and ES08. Some projects, such as ES04 and ES25, received the majority of their funding from ERDF and other public sources.

160

Figure 67: Proportions of different types of funding used for productive investments

0

50

100

150

200

250

300

350

400

450D

E16

DE

14

DE

13

DE

12

DE

11

ES

04

ES

22

ES

23

PT2

7

AT0

1

FR18

DE

15

ES

10

ES

08

ES

07

ES

06

ES

02

ES

05

ES

03

PT2

8

PT2

9

PT2

6

ES

24

ES

25

ES

21

ES

09

ES

20

PT3

0

GB

19

FR17

Project reference

€ M

illio

n

Other public funding

ERDF funding

Private sectior funding

161

7.8 The amount of structural funds per job created

7.8.1 We conclude this Section with an illustrative comparison across all productive investments of the amount of Structural Funds per job created involved.

7.8.2 For ‘new build’ manufacturing the full range is €61,000 to €817,000 of ERDF funding per job created. However, in four of the five ‘new build’ projects that received funding, the amount did not exceed €150,000. The outlier, with an amount of ERDF funding per job created of in excess of €800,000, is the same project from Spain that was identified as an outlier in earlier parts of this section.

7.8.3 The ERDF funding contribution to the manufacturing expansion projects and the R&D projects was estimated at between €1,200 and €15,000 per job created.

Figure 68: ERDF funding per estimated job created

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

DE1

6

DE1

4

DE1

3

DE1

2

DE1

1

ES0

4

ES2

2

ES2

3

PT2

7

AT0

1

FR18

DE1

5

ES1

0

ES0

8

ES0

7

ES0

6

ES0

2

ES0

5

ES0

3

PT2

8

PT2

9

PT2

6

ES2

4

ES2

5

ES2

1

ES0

9

ES2

0

PT3

0

GB1

9

FR17

Euro

mill

ions

No ERDF funding

Job safeguarding

only

No ERDF funding

162

8 DEVELOPMENT OF THE SPREADSHEET TOOLS

8.1 Introduction

8.1.1 In line with the Terms of Reference, we set out to develop spreadsheet-based databases of unit cost for both infrastructure projects and productive investments to serve two specific functions:

• provide a summary for the information we have collected during our analysis allowing quick access to data relative to specific projects; and

• act as a repository for new project data as these become available. The databases can be maintained and expanded with new data, which can be compared with the information that have been collected during the course of this analysis.

8.1.2 At the start of our study, in line with the Terms of Reference, we set out to develop database tools for both infrastructure projects and infrastructure investments. However, as discussed in the previous Chapter, our analysis has shown that the benchmarking of productive investment on the basis of the ‘cost per job created’ is unlikely to be meaningful without appropriate adjustments for different industry circumstances. For this reason, only the infrastructure projects database tool was eventually developed for general use. The remainder of this section therefore focuses only on the database tool for infrastructure projects.

8.2 Overall structure

8.2.1 Figure 69 shows the main menu of the spreadsheet tool. From the front page, the user can access all database information, both on an individual project basis and on a sector summary basis.

Figure 69: Spreadsheet tool main menu

163

8.2.2 The spreadsheet follows a simple structure. We created input sheets that allow the Commission services to enter new project data. Normalisation sheets convert raw data into comparable unit costs, adjusting for exchange rates and inflation. Finally, output sheets allow the Commission services to examine detailed information for a particular project, and to get an overview of all the projects in the database for a particular sector.

8.2.3 At this stage, the limited number of projects in our sample restricted the benchmarks we could provide. For example, it is not clear what ‘standard’ projects should be classified into in order to provide appropriate benchmarks. For this reason, we do not provide output sheets that give overviews of projects by type or by Member State.

8.2.4 In addition, the sample was too small to provide clear evidence of the variability of actual unit costs over time, after adjusting for inflation and the exchange rate. For this reason, we do not account for changes in actual unit costs over time driven by, for example, technical change.

8.2.5 Nevertheless, the input sheets will allow the Commission services to update the spreadsheet tool as new project data become available. More data is likely to become available in the new programming period or in future evaluations.

8.2.6 We normalise the data using construction cost indices and annual average exchange rates. This allows for easy comparison of projects.

8.2.7 The output sheets provide indicative benchmarks for the Commission to use for project appraisal. In particular, users will be able to search for projects similar to that under appraisal, and compare unit costs. The spreadsheet tool also allows the consultation of summary sheets for an overview of all the projects in the database for a particular sector.

8.3 Input

8.3.1 Input sheets are data entry forms. We thought carefully about how best to allow the Commission services to enter new information on project data from the new programming period or future evaluations.

8.3.2 To avoid confusion, there is one input sheet per sector, i.e. one for Rail, Road, Urban Transport, Water and Wastewater and Energy. There are also subheadings for each section within the input sheet. We prioritised the more important information by placing it at the top of the input sheet. The user only needs to complete the template, and to save it, for the project to be included in the database.

8.3.3 We learnt from the evaluation that project managers typically do not hold information on the complete set of variables of interest. To counter this, we provide as much flexibility in the input sheets as possible. For example, if they have all the information, users can enter units for the number of bridges and stations, or their area, or both.

8.3.4 For illustration purposes, Figure 70 shows the input sheet for Rail projects

164

Figure 70: Input sheet for Rail sector projects

8.4 Normalisation

8.4.1 The normalisation sheets within the tool, one for each sector, contain the normalised data. They transform the raw cost data to Euros in 2007 prices. We followed the methodology to achieve data comparability set out in section 98.

8.4.2 The first normalisation step adjusts for inflation. We did this using a table of Eurostat construction cost indices, one for each Member State. Each sector references the same construction cost index. This allowed us to convert all costs to 2007 prices. The second step adjusts for the exchange rate. We used annual exchange rates taken from the European Central Bank to convert local currencies to Euros. This allowed us to convert all costs to Euros.

8.4.3 By combining the inflation adjustment and the exchange rate adjustment, the sheets normalise all raw data, regardless of country or year. The normalisation parameters can be adjusted to reflect changes in exchange rates and, if required, to change the price base.

8.5 Output

8.5.1 The output sheets provide two sets of information to facilitate project appraisal by the Commission services. First, they provide detailed individual project information. Second, they summarise all the projects for a particular sector, presenting sector-specific average unit costs.

165

8.5.2 There is one single project output sheet for each sector. This allows the Commission services to examine a particular project in detail. For example, for each project, we display all its project characteristics and we provide a full list of normalised Level 1, Level 2, and Level 3 unit costs. The Commission services will also be able to check graphs for Level 1, Level 2 and Level 3 unit costs. Figure 71 illustrates the single project output sheet for road projects.

Figure 71: Road sector single project output sheet

8.5.3 There is also one summary output sheet per sector. This provides summary statistics on all the projects in the database for that sector. For example, it shows the number of projects and their mean Level 1 unit cost. We intended to give the Commission an overview of the range of information available for a particular sector, and what are typical unit cost values. Figure 74 provides an example of summary output sheet, in this case for the Urban Transport sector.

166

Figure 72: Urban Transport sector summary output sheet

8.5.4 For project appraisal, we anticipate that the Commission services might decide to implement Reference Class Forecasting Method. This requires comparison of a project to similar projects. Estimates that are far away from what similar projects achieved in the past reveals the possible presence of optimism bias in cost estimates or time forecasting. The spreadsheet tool could facilitate such a comparison.

167

9 SUMMARY AND CONCLUSIONS

9.1 Review of the literature, previous evaluations and databases

9.1.1 The first task, a review of the literature, revealed an extensive body of research into the causes of infrastructure cost overruns and project delays and of the tendencies for them to differ in magnitude and scale across countries and sectors. Our key findings were that:

• Optimism bias and other difficulties in the estimation of project costs are widely acknowledged; and

• The lack of good quality project data is widespread, which makes it difficult to use benchmark data as a basis for forecasting.

9.1.2 For these reasons, robust cost estimation, project appraisal and evaluation will remain a critical area of responsibility for public agencies that are recipients of EU funds. In that respect, the availability of reliable and consistent benchmark data on project unit costs would provide a powerful tool to improve the appraisal and evaluation process.

9.1.3 Our search for benchmark databases revealed that, with only a couple of exceptions, there are no relevant up-to-date databases of infrastructure costs.99 Some high level total cost benchmark data was available but there were problems with the definition of unit costs and with data comparability in general.

9.1.4 We note, of course, that the WP10 study itself arose from a DG Regio objective to set up an infrastructure project cost database, one that might provide reliable and robust cost benchmark data for the purposes of appraising and evaluating major projects in the future. We hope, therefore, that the database we have constructed and the analysis we have conducted, once completed, will be considered to be a useful first step in establishing such a database and, perhaps, a useful foundation for future work in this area.

9.2 Definition and measurement of unit cost indicators for infrastructure

9.2.1 The literature review and the examination of the cost information on the WP10 sample projects (from the project dossiers) suggested that there are two key dimensions involved in the definition of unit cost indicators: (i) the categories of cost to be included and (ii) the level of disaggregation of the individual project components.

9.2.2 Major infrastructure project cost categories include, in general, construction (‘build’) costs, soft costs (things like project planning and management), contingencies, taxes, land acquisition costs etc. However, the treatment of these different categories of cost was widely divergent across projects. For example, some projects provided separate amounts for taxes and contingencies, while other projects included such costs in the estimates of the other cost categories. This made the calculation of comparable cost data difficult.

9.2.3 Because no two infrastructure projects are identical, unit cost benchmarks are useful if they accurately reflect the average cost of sufficiently disaggregated

99 The most notable exception is the World Bank ROCKS database for large highway projects.

168

project components. We developed a three-tier approach to the definition of unit cost indicators in order to reflect various degrees of cost disaggregation. We defined three levels of disaggregation, which are illustrated next for a road project that includes bridges and tunnels:

• Level 1: indicators that reflect the ‘all in’ costs of a project, including all appropriate categories of cost outlined in section 5.4 above and all project components.

• Level 2: indicators that reflect the ‘build’ cost of individual key components of projects. Separate indicators were defined for each of the road pavement itself, the bridges and the tunnels.

• Level 3: indicators that distinguish further between different types of key components, such as, for example, the different possible grades of road pavement or the different types of bridges and tunnels.

9.3 Methodology, data gathering and results

Overview

9.3.1 We identified at an early stage that the project monitoring undertaken by the Member States was in a form that would not provide the necessary project cost information. More importantly, there is no legal obligation to monitor individual projects for the 2000-2006 funding program. Monitoring is required at the program level. The Member States were not, therefore, legally obliged to collect or provide the information about individual projects.

9.3.2 Having commenced the data gathering effort with requests for official project completion reports and/or progress reports, it soon became apparent that such reports did not exist or, where they did, that they provided an insufficient amount and level of detail of information to be considered useful. We changed our approach, therefore, by designing questionnaires to serve as a more detailed guide to our data requirements.

9.3.3 Since early September 2008, we have actively sought data on a total of 173 major projects (128 infrastructure and 45 productive investments), which compares with the Commission’s target of 155 projects (115 infrastructure and 40 productive investments). By mid-February 2009, we had received a total of only 66 data returns, at which point the Evaluation Unit of DG Regio undertook to work with the desk officers from the Geographical Units to contact Member States and ask them to respond to our information requests.

9.3.4 This involvement of the country desk officers of the Commission, combined with additional effort from our own data gathering team, produced some positive results, increasing the sample size from 66 projects in February 2009 to 96 by the time we wrapped up the data gathering exercise in mid-May 2009. Table 1 below summarizes the results of the data gathering exercise.100

100 Projects were counted in the ‘data returns received category if the relevant ministry / agency in the Member State submitted information on actual costs and / or alternative or more up-to-date cost estimates from those in the ERDF application forms that were in our possession.

169

Table 35: Summary of the data gathering results





Data returns received 96 66 30

Possible reasons for poor response rate

9.3.5 For several projects, the ERDF applications could not be located by the Commission services. In some of these cases, it was more difficult to establish contact with relevant agencies in the Member States. In other cases, we were able to make contact through basic internet research on the projects. However, issues like changes in national telephone numbering systems also caused setbacks for the data gathering team.

Implications of poor response rate

9.3.6 The direct implications of the relatively poor response rate are (i) a reduced sample size and (ii) a sample structure that is biased in favour of road and rail projects. A more indirect implication is the limiting effect on the scope and depth of the statistical analysis that was possible for WP10. However, we believe that the statistical analysis has been more severely affected by the scope and quality of the information that was made available than by any sector bias.

9.3.7 Under the 2000-2006 Objective 1 and Objective 2 ERDF programme, 271 major projects were adopted for the 11 Member States that were the subject of the WP10 Study. The total estimates of expenditure involved in these projects were circa Euro 33 billion, of which approximately Euro 15 billion was to be the contribution of the ERDF. Therefore, the sample of 96 major projects analysed in this report represents about 35 per cent of the total of 271 projects financed with EU structural funds in the period of study.

9.4 Comparing estimated and actual unit infrastructure costs

9.4.1 Our analysis of unit costs was constrained by the limited sample sizes and level of detail available for each project, as summarised above and discussed in Sections 5 and 6 of this Report. Notwithstanding these limitations, we expect our database to provide benchmark cost data that will be useful in the appraisal of future ERDF-project financing requests.

9.4.2 Our database also contains information on project characteristics for each project which, for example, identifies the urban or rural location of the project, the geographic terrain, and the project complexity. We did not, however, succeed in gathering sufficient data on project characteristics to enable us to

170

carry out a statistical analysis of the possible relationship between certain project characteristics and costs (for example, to determine the impact of the urban/rural split of a project on overall cost).

9.4.3 The level of data we currently have has enabled us to produce useful Level 1 (and some Level 2) unit cost benchmarks. For example, Figure 1 below shows our benchmark unit cost data for road projects, which will prove a useful resource for future project appraisal, at least at the initial high-level stage of appraisal.


0

10

20

30

40

50

60

70

80

DE02 - P

rojec

t A

DE02 - P

rojec

t B

DE03 - P

rojec

t A

ES04 - P

rojec

t A

GR03 - E

gnatia

GR05 - K

ifisso

s

GR03 - P

rojec

t A

GR04 - A

ttiki

GR05 - P

rojec

t A

GR05 - C

hrisoupoli

GR04 - P

rojec

t A

GR05 - S

iatist

a

IE02 - H

urlers

IE02 - W

aterg

rassh

ill

IE02 - A

shford

IE05 - B

allys

hannon

IE02 - B

allinco

llig

IT06 - O

riental

e

IT03 - C

arlo Feli

ce

PL05 - P

ulawy

PL05 - E

lbag

Uni

t cos

t (€m

/km

)


9.4.4 If, as recommended below, the database is adequately maintained and updated over time, it will provide an increasingly useful information resource on major projects. It will also extend the boundaries of the statistical analysis that can be undertaken to enable, for example, the analysis of the relationships between different project characteristics and overall time delays. It might also enable the use of the database in a ‘reference class forecasting’ approach to benchmarking and cost estimation.101

9.5 Analysis of cost overruns and time delays

9.5.1 Our analysis of infrastructure unit costs was constrained by the limited sample sizes and level of detail available for each project, as summarised above and discussed in Sections 5 and 6 of this Report. Notwithstanding these

101 Reference class forecasting was developed in the context of major infrastructure projects by Prof. Bent Flyvbjerg. See, for example, Flyvbjerg (2007), ‘Eliminating bias through reference class forecasting and good governance’.

171

limitations, we expect our database to provide benchmark cost data that will be useful in the appraisal of future ERDF-project financing requests.

9.5.2 The database also contains information on individual project characteristics which, for example, identifies the urban or rural location of the project, the geographic terrain, and the project complexity. We did not, however, succeed in gathering sufficient data on project characteristics to enable us to carry out a statistical analysis of the possible relationship between certain project characteristics and costs (for example, to determine with any rigour the expected impact of the urban/rural split of a project on overall cost).

9.5.3 The level of data we have collected has enabled us to produce useful Level 1 (and some Level 2) unit cost benchmarks. For example, Figure 1 below shows our benchmark unit cost data for road projects, which may be useful for future project appraisal, at least at the initial appraisal stage.


0

10

20

30

40

50

60

70

80

DE02 - P

rojec

t A

DE02 - P

rojec

t B

DE03 - P

rojec

t A

ES04 - P

rojec

t A

GR03 - E

gnatia

GR05 - K

ifisso

s

GR03 - P

rojec

t A

GR04 - A

ttiki

GR05 - P

rojec

t A

GR05 - C

hrisoupoli

GR04 - P

rojec

t A

GR05 - S

iatist

a

IE02 - H

urlers

IE02 - W

aterg

rassh

ill

IE02 - A

shford

IE05 - B

allys

hanno

n

IE02 - B

allinco

llig

IT06 - O

riental

e

IT03 - C

arlo Feli

ce

PL05 - P

ulawy

PL05 - E

lbag

Uni

t cos

t (€m

/km

)


9.5.4 If, as recommended below, the database is adequately maintained and updated over time, it will provide an increasingly useful information resource on major projects. It will also extend the boundaries of the statistical analysis that can be undertaken to enable, for example, the analysis of the relationships between different project characteristics and cost overruns and/or overall time delays. It might also enable the use of the database in a ‘reference class forecasting’ approach to benchmarking and cost estimation.102

9.5.5 Our analysis of cost overruns and time delays was influenced by the same data constraints that limited our analysis of unit costs. However, we did collate information where it was available and this has allowed us to undertake a descriptive analysis. Table 36 and Table 37 provide an illustrative

102 Reference class forecasting is a method that predicts the outcome of a project based on actual outcomes in a reference class of similar projects to that being forecast. It has been applied in the context of major infrastructure projects by Prof. Bent Flyvbjerg; see, for example, Flyvbjerg (2007), ‘Eliminating bias through reference class forecasting and good governance’.

172

summary of this analysis. Specifically, Table 36 shows a summary of average percentage cost overruns for all observations in our sample, broken down by sector and by country. A positive value indicate a cost overrun, while a negative value indicates a cost saving.

Table 36: Cost overrun summary by country and sector – average percentage differences between estimated and actual costs (%)/(number of projects) Rail Road Urban


average by sector

Germany -4.3% (6) -10.0% (3) -6.2%

Spain 12.8% (6) 30.7% (1) 17.4% (2) 15.8%

France 32.9% (1) 32.9%

Great Britain 110.7% (1) 110.7%

Greece 74.3% (2) 19.7% (8) 20.1% (2) 0.0% (1) 26.6%

Ireland 2.1% (5) 74.1% (1) 14.1%

Italy 62.4% (5) -5.0% (2) -0.9% (1) 37.6%

Poland 19.7% (2) 80.9% (2) 50.3%

Portugal 9.0% (1) 3.3% (4) 4.4%


26.9% 9.4% 45.4% 11.3% 20.7% 21.2%

9.5.6 Table 37 provides an analogous summary for percentage delays (calculated as the ratio of the actual completion time and the estimated completion time). Also in this case, positive values correspond to actual delays, while negative values indicate that the actual average completion time was shorted than expected.

173


(%) /(number

of projects) Rail Road Urban


average by sector

Germany 40.2% (6) 4.7% (3) 28.4%

Spain 15.3% (6) 27.3% (1) 55.9% (2) 25.7%

France 4.9% (1) 4.9%


Greece 24.4% (2) 17.8% (7) 13.2% (2) 12.6% (1) 17.7%

Ireland 9.0% (5) 52.2% (1) 16.2%

Italy 88.4% (1) 88.4%

Poland 5.9% (1) 2.7% (2) 3.8%

Portugal 258.3% (1) 41.5% (4) 84.9%


25.8% 13.2% 49.6% 66.7% 29.8% 26.2%

9.5.7 The values presented in both tables are averages and, as such, they mask the high variance of results which we obtained for each project. However, we note that most of the projects in our database were not completed on time and without any cost overruns.

9.5.8 These initial results appear to confirm the usefulness of this type of analysis. If the database is maintained and updated with more data on project performance, it has the potential to be useful in the risk assessment which is required as part of major project appraisal.

9.6 Ex-ante risk assessment of major projects

9.6.1 Our review of the ex ante risk assessments provided by Member States in project application forms based on a small sample of road projects indicated a diverse range of approaches to risk assessment, in terms of methodology and quality of analysis.

9.6.2 We note that the Commission has published a comprehensive guide to cost benefit analysis which includes a thorough guide to risk analysis.103 The application of the risk analysis outlined in the Commission’s guide would represent a significant improvement in the quality of ex ante risk assessments

103 European Commission, Directorate General Regional Policy, Guide to Cost-Benefit Analysis of investment projects Structural Funds, Cohesion Fund and Instrument for Pre-Accession Final Report 16/06/2008.

174

compared to the project application forms we reviewed. For example, the application of probability based risk analysis as set out in the Commission’s has not generally been applied, but would provide a significant improvement in risk quantification compared to the most frequently adopted approach in the projects reviewed, of applying a contingency sum to the budget.

9.7 ‘Cost per job created’ from productive investments

9.7.1 WP10 is also concerned with the ‘result’ efficiency of productive investment projects (i.e., projects that involve direct support to enterprises), where ‘result’ efficiency is measured according to the ‘cost per job created’ as a result of these investments. However, the accurate measurement of Structural Fund employment effects poses the greatest challenge in evaluating the economic efficiency of productive investments, in both absolute and relative terms. This, in turn, casts doubt over the possibility of calculating meaningful ‘cost per job created’ benchmarks for ERDF co-financed productive investments.

9.7.2 The assessment of the methods used in the project dossiers to forecast employment effects was part of the scope of WP10. In a few cases, the methods were set out in a supporting cost-benefit report although the quality of the analysis varied. In many cases, the methodological details to support employment effect projections may not have been made available to us due perhaps to either a lack of analysis or lack of adequate data. We believe that Member States should be encouraged to make better use of the Commission’s CBA guidelines to improve their data generation and maintenance regarding productive investments.

9.7.3 There is very little comparable research on the costs of job creation. The CSES (2006) Study found an overall average of €36,000 per job, albeit across a wide range of productive investments. It is not clear however, how comparable this benchmark is to jobs created in different sectors under different funding arrangements.

9.8 Development of a spreadsheet tool

9.8.1 The project has gathered a significant amount of detailed information on a wide range of projects regarding costs and project characteristics. As required by the Terms of Reference, we set out to develop database tools for both infrastructure projects and infrastructure investments. However, as discussed in the previous Chapter, our analysis has shown that the benchmarking of productive investment on the basis of the ‘cost per job created’ is unlikely to be meaningful without appropriate adjustments for different industry circumstances. For this reason, only the infrastructure projects database tool was eventually developed for general use.

9.8.2 For infrastructure projects, the data has been incorporated into an Excel spreadsheet tool, which serves two specific functions, namely to:

• Provide a summary of the information that we have collected during the analysis, allowing quick access to data relating to specific projects.

• Act as a repository for new project data as they become available. The database can be maintained and expanded with new data, which can be compared with the information that has been collected during the course

175

of WP10. New project details can easily be added to the spreadsheet and output unit benchmark costs then easily updated to include the new data.

9.8.3 The spreadsheet tool allows the user to include details on the characteristics of individual projects if these are available. We expect this tool to facilitate statistical analysis in the future. It should also be useful in the application of the reference class forecasting approach to project cost benchmarking, estimation and appraisal.

176

10 RECOMMENDATIONS

10.1 Recommendations

Development of an EU-wide database of major infrastructure project costs

10.1.1 The lack of availability of relevant up-to-date databases of infrastructure project costs makes the Excel spreadsheet tool developed in the course of this work all the more useful as a starting point in the development of such a database for EU-funded projects. This can be built up over time to provide a valuable on-going source of benchmark cost data.

10.1.2 We recommend, therefore, that the Commission considers how best to ensure that the database is maintained on an ongoing basis and that the quality and quantity of data improves over time. Three further recommendations may contribute to the usefulness of the database, namely:

• the participation of European financial institutions, such as EIB and EBRD that are also involved in major project co-financing;

• the adoption of common sets of unit cost definitions (see section 5.4 below) at national and EU level; and

• the reform, at the Member State level, of major project monitoring and reporting and in the depth of project appraisal carried for the acquisition of EU funds (see section 5 in general).

10.1.3 The collaborative approach suggested above can have a number of potential benefits, including the increased number of data points and the potential for sharing the costs of maintaining the database. It might also facilitate the multilateral development and consequent greater use of common sets of cost definitions. The common sets of unit cost definitions could be defined according to the three-tier Level 1 to 3 approach that we have developed for WP10 (see section 5.4 below).

Standardize and improve the ex ante risk assessments in funding applications

10.1.4 We also recommend that the application of the methods for the ex ante assessment of project risks that are set out in the Commission’s most recent guide to cost-benefit analysis would represent a significant improvement in the quality of such assessments (which are legally required for all major projects). 104 , 105 This we concluded from a review of several of the relatively few project dossiers on which detailed reports were available in support of the ERDF application forms (which generally provided only the conclusions of such reviews).

Standardize and improve project monitoring and reporting for major projects

10.1.5 Based on the responses we received to our enquiries and questionnaires, it appears that relatively few project completion or progress reports are

104 European Commission, Directorate General Regional Policy, Guide to Cost-Benefit Analysis of investment projects Structural Funds, Cohesion Fund and Instrument for Pre-Accession Final Report 16/06/2008 105 A useful manual from the US Government Accountability Office is ‘Cost estimating and assessment guide: Best practices for developing and managing capital program costs’, GAO Applied Research and Methods (2009).

177

received by the Member States on major projects funded by ERDF. Furthermore, the usefulness of the few reports we reviewed for project evaluation purposes was limited.

10.1.6 We recommend that the Commission considers how best to improve the project reporting process for major projects with the objectives of improving the project monitoring process and the availability of project data for planning, project appraisal and evaluation.

10.1.7 In particular, the following improvements should be considered:

• The use of standard definitions for project costs to be required in project application forms, appraisal and monitoring.

• The preparation of regular project progress reports to be a condition for funding.

• The submission of regular project progress reports (to a specified format) to be a condition for continuing drawdown of funds.

• The submission of a project completion report (to a specified format) to be a condition of funding.

10.1.8 Annex IV shows completed questionnaires for two of the projects that were in our sample. These particular questionnaire responses contain a level of detail that, if all questionnaire responses contained the same level of information, would provide a basis for a more thorough application of the WP10 methodology, as well as increasing the richness of the statistical analysis that would be possible.

10.1.9 The projects for which questionnaire responses are provided :

• A rail project from Greece with project reference 2005GR161PR011 – this was Phase 1 of the Thessalonika to Idomeni programme, Phase 1 stretching from Polycastro to Idomeni; and

• A road project from Spain with project reference 2004ES161PR023 – this was a road project stretching from, as far as we can make out de Gijon to Sevilla.

10.1.10 Improving the project monitoring and reporting processes would involve, first and foremost, communicating to the Member States the need to and importance of monitoring projects in such a way as to enable them to provide the type and level of detail of information illustrated in Annex IV. This would likely require the Member States, in turn, to communicate to their implementing ministries and agencies the need for them to agree with contractors to record information about projects in this manner.

178

ANNEX I – GRAPHICAL PRESENTATION OF OUR BENCHMARKING RESULTS

10.1.11 The following charts show the benchmark data gathered and, where relevant our ‘indicative benchmark range’ of costs against which other projects can usefully be compared.

10.1.12 Figure 75 shows total unit project costs for different types of road. Not surprisingly, these costs vary hugely and further classification of project types is needed. Figures 2 and 3 show how analysing costs for specific road types can yield more useful information.

179

Figure 75: Level 1 Road Costs

Sources: www.wsdot.wa.gov, www.roadtraffic-technology.com, Data from Faber Maunsell, Data from Prof. Bent Flyvbjerg.

0

5

10

15

20

25

30

35Ire

land

Sw

eden

USA

Fran

ce

Fran

ce

Fran

ce

Gre

ece

Irela

nd

Pol

and

Pol

and

Aust

ralia UK

Sw

eden

USA

USA

Gre

ece

Aust

ralia

Fran

ce UK

UK

USA

USA

USA

USA

USA

USA

Dual 1 lane carriageway New dual 2 lane carriageway New dual 3 lanecarriageway

New 4lane

Road Widening(1 lane)

Road Widening (2 lane) RoadWidening

2007

Pric

es (€

m/k

m)

62.9 91.3

180

Figure 76: Two-lane carriageway level 1 construction costs

0

5

10

15

20

25

30

35

40

Fran

ce

Fran

ce

Fran

ce

Gre

ece

Irela

nd

Pol

and

Pol

and

Aus

tralia UK

Sw

eden

USA

New dual 2 lane carriageway

2007

pric

es (M

illio

ns €

/km

) 62.59

Sources: www.wsdot.wa.gov, www.roadtraffic-technology.com, Data from Faber Maunsell, Data from Prof. Bent Flyvbjerg

10.1.13 Figure 76: Two-lane carriageway level 1 construction costs shows level 1

costs for two lane carriageway construction. The chart shows an indicative benchmark range for this category of cost of €5m to €15m.

Figure 77: Road widening level 1 construction costs

0

2

4

6

8

10

12

14

16

18

UK UK USA USA USA USA USA USA

Road Widening (1 lane) Road Widening (2 lane) Road Widening

2007

Pric

es (€

m/k

m)

Sources: www.wsdot.wa.gov, www.roadtraffic-technology.com, Data from Faber Maunsell, Data from Prof. Bent Flyvbjerg

181

10.1.14 Figure 77 shows unit construction costs for road widening projects. The indicative benchmark range lies between €2m and €8m per km.

10.1.15 The following charts show level 2 costs for carriageways, bridges and tunnels.

10.1.16 Figure 78 shows the carriageway costs for new constructions, reconstruction and widening projects, grouped by project type. As one would expect, the most expensive of these is the construction of wider lane roads and the least expensive is the cost of road widening.

Figure 78: Level 2 construction costs for carriageways

0

1

2

3

4

5

6

Brazil Estonia India India Albania Thailand Bolivia Turkey Croatia Estonia Cyprus Poland Poland UK China

Road Widening Reconstruction New 2 Lane Carriageway New 4 Lane Carriageway

2007

Pric

es (€

m/k

m)

Sources: World Bank ROCKS database, Data from Faber Maunsell

10.1.17 Figure 79 shows that unit tunnel construction costs vary widely depending on project specific factors.

182

Figure 79 Level 2 construction costs for tunnels

0

20000

40000

60000

80000

100000

120000

140000

160000

180000

200000

Undersea 3 lanetunnel

Twin Parallel Tubetunnel

2 deck, 6 lanetunnel (notcomplete)

Undersea Twin 2lane tunnels

2 Tunnel Tubes Twin Parallel Tubetunnel

Two cut and coversections, Two bored

sections

Norway Germany France Netherlands Switzerland UK Ireland

2007

pric

es (€

/m)

Sources: www.roadtraffic-technology.com

10.1.18 Figure 80 shows bridge construction costs. As with tunnels, these vary significantly.

Figure 80 Level 2 construction costs for bridges

0

5000

10000

15000

20000

25000

Cab

le S

taye

dBr

idge

Cab

le S

taye

dBr

idge

Cab

le S

taye

dBr

idge

Cab

le S

taye

dBr

idge

Cab

le S

taye

dBr

idge

Cab

le S

taye

dBr

idge

Cab

le S

taye

dBr

idge

4 La

ne B

ox G

irder

Brid

ge

Beam

Brid

ge

Can

tilev

er

Cly

de A

rc B

ridge

Lifti

ng B

ridge

Susp

ensi

on B

ridge

Susp

ensi

on B

ridge

France Greece Sweden USA USA USA USA Norway UK UK UK France USA UK

2007

pric

es (€

/squ

are

m)

29,098

Sources: www.roadtraffic-technology.com, Data from Faber Maunsell

183

10.1.19 Figure 81 shows unit construction costs for metro services. The indicative benchmark range lies between €50m and €150m.

Figure 81 Metro construction costs

0

50

100

150

200

250

300

350

Cop

enha

gen

Met

ro (2

1km

, 22

stop

s, 4

8% T

unne

l)

Toul

ouse

Val

Lin

e B

(15k

m, 2

0st

ops)

Mar

seill

e Li

ne 1

Ext

. (2.

5km

, 4st

ops)

Turin

Met

ro P

hase

1 (9

.6km

, 15

stop

s, 1

00%

Tun

nel)

Toul

ouse

Val

Lin

e A

Ext

. (2.

2km

,3

stop

s)

Lond

on J

ubile

e Li

ne E

xt. (

2.2k

m,

3 st

ops)

Mad

rid E

xt. (

56.3

km, 3

8 st

ops,

68%

Tun

nel)

LA N

orth

Hol

lyw

ood

Ext

. (10

.1km

,3

stop

s)

Atla

nta

Nor

t Ext

(3.7

km, 2

sto

ps)

San

Fran

cisc

o Ai

rpor

t Ext

. (14

km,

4 st

ops)

Mex

ico

City

Lin

e B(

23.7

km, 2

1st

ops,

25%

Tun

nel)

San

tiago

Lin

e 5

Ext

. (2.

8km

, 3st

ops,

100

% T

unne

l)

Shep

pard

Sub

way

, Tor

onto

(5.5

km)

Sto

ckho

lm C

ityba

nan

Met

ro (6

km,

2 st

ops,

100

% T

unne

l)

Ligh

t Rai

l (tw

in tr

ack

segr

egat

ed)

Ath

ens

Met

ro li

ght r

ail s

yste

m(7

5.4k

m, 5

5 st

ops,

66%

Tun

nel)

Am

ster

dam

Nor

th S

outh

Lin

e(9

.8km

, 8 s

tops

, 70%

Tun

nel)

Fran

kfur

t U-b

ahn

Ext

. (1.

7km

)

Kaoi

shun

g M

etro

(42.

7km

, 37

stop

s, 5

3% T

unne

l)

Denmark France France Italy France UK Spain USA USA USA Mexico Chile Canada Sweden UK Greece Netherlands Germany Taiwan

2007

pric

es (€

m/k

m)

482.9

Sources: http://lrt.daxack.ca, www.railway-technology.com, Data from Faber Maunsell, Bent Flyvbjerg et al., Comparison of Capital Costs per Route-Kilometre in Urban Rail, 2008

10.1.20 Figure 82 shows Light Rail and Tram service construction costs in millions of Euros per km. The indicative benchmark range lies between €10m and €30m per km.

10.1.21 Figure 83 is based on data provided by Prof. Bent Flyvbjerg for level 1 total urban rail project costs. Whilst this dataset lacks any detailed information, the quantity of data means this chart is of some interest. The indicative benchmark range for all metro lies between €50m and €150m per km

184

Figure 82: Light Rail and Tram construction costs

Sources: www.railway-technology.com , www.lightrailnow.org, Transek Consultants, Comparison of Costs between Bus, PRT, LRT and Metro/rail, 2003

0

10

20

30

40

50

60

70

Light rail (twin track)

street running

Montpellier (19.5km

)

LRT S

tockholm

LRT S

tockholm

Alicante Tram

- 51km,

42 stops, 19 trams

Athens Tram

- 26km, 47

stations, 35 trams

Caen Tram

way -

15.7km, 34 stations

Croydon - 28km

, 39

stations

Valenciennes - 18.8km

,

26 stops, 21 trams

Tacoma Link, S

eattle

Denver (14km

)

Kenosha (3.2km

)

San Jose (3.1km

)

Denver (2.9km

)

Cam

den (54.7km)

LA (22.01km

)

Sacram

ento (17.5km)

Sacram

ento (18.02km)

Houston (12.07km

)

Minneapolis (18.7km

)

San Jose (10.3km

)

San D

iego (5.5km)

Denver (30.9km

)

LA (9.7km

)

Central P

hoenix/ East

Valley (32.7km

)

UK FranceSwedenSweden Spain Greece France UK France USA USA USA USA USA USA USA USA USA USA USA USA USA USA USA USA

2007

Pric

es (€

m/k

m)

185

Figure 83 Metro services cost of construction from Prof. Bent Flyvbjerg’s research

0

50

100

150

200

250

300

350

Cop

enha

gen

Met

ro (2

1km

, 22

stop

s, 4

8% T

unne

l)

Toul

ouse

Val

Lin

e B

(15k

m, 2

0

stop

s)

Mar

seille

Lin

e 1

Ext.

(2.5

km, 4

stop

s)

Turin

Met

ro P

hase

1 (9

.6km

, 15

stop

s, 1

00%

Tun

nel)

Toul

ouse

Val

Lin

e A

Ext.

(2.2

km,

3 st

ops)

Lond

on J

ubile

e Li

ne E

xt. (

2.2k

m,

3 st

ops)

Mad

rid E

xt. (

56.3

km, 3

8 st

ops,

68%

Tun

nel)

LA N

orth

Hol

lyw

ood

Ext.

(10.

1km

, 3 s

tops

)

Atla

nta

Nor

t Ext

(3.7

km, 2

sto

ps)

San

Fran

cisc

o Ai

rpor

t Ext

.

(14k

m, 4

sto

ps)

Mex

ico

City

Lin

e B(

23.7

km, 2

1

stop

s, 2

5% T

unne

l)

Sant

iago

Lin

e 5

Ext.

(2.8

km, 3

stop

s, 1

00%

Tun

nel)

Shep

pard

Sub

way

, Tor

onto

(5.5

km)

Sto

ckho

lm C

ityba

nan

Met

ro

(6km

, 2 s

tops

, 100

% T

unne

l)

Ligh

t Rai

l (tw

in tr

ack

segr

egat

ed)

Athe

ns M

etro

ligh

t rai

l sys

tem

(75.

4km

, 55

stop

s, 6

6% T

unne

l)

Amst

erda

m N

orth

Sou

th L

ine

(9.8

km, 8

sto

ps, 7

0% T

unne

l)

Fran

kfur

t U-b

ahn

Ext

. (1.

7km

)

Kaoi

shun

g M

etro

(42.

7km

, 37

stop

s, 5

3% T

unne

l)

Denmark France France Italy France UK Spain USA USA USA Mexico Chile Canada Sweden UK Greece Netherlands Germany Taiwan

2007

pric

es (€

m/k

m)

482.9

Sources: Data from Bent Flyvbjerg

10.1.22 Figure 84 shows the total costs of different types of rail projects.

Figure 84 Level 1 Rail Costs

0

10

20

30

40

50

60

70

80

90

100

New twin track High speed twin track High speed twin track High speed twin track

2007

Pric

es (€

m/k

m)

Sources: Data from Faber Maunsell

10.1.23 Figure 85 shows the track costs for different types of rail projects.

187

Figure 85 Level 2 Rail Costs – Tracks

0

2

4

6

8

10

12

New single track New twin track New twin track High speed twin track High speed twin track

2007

Pric

es (€

m/k

m)

Sources: Data from Faber Maunsell

10.1.24 Figure 86 shows the cost of construction of various wind farms across Europe. The indicative benchmark range lies between €1000 per kW and €1600 per kW as this is where the majority of the data falls.

188

Figure 86 Wind farm construction costs

0

200

400

600

800

1000

1200

1400

1600

1800

2000

Onshore,500 kWRated

Onshore,660kWRated

Onshore,750kWRated

Onshore,1500kWRated

Onshore,1500kWRated

Onshore,1500kWRated

Onshore,1650 kW

Rated

Onshore,1650kWRated

Offshore,2000 kW

Rated

Offshore,2300kWRated

Onshore,2500kwRated

Offshore,3000kWRated

Germany Ireland Spain Ireland Ireland Ireland Germany Spain Denmark Denmark Ireland Netherlands

€/kW

10.1.25 Figure 87 and Figure 88 show the cost of construction of water treatment and supply facilities in various countries.

10.1.26 Although we only have a small number of benchmarks, Figure 87 and Figure 88 suggest that water projects vary significantly in costs, perhaps suggesting that project specific features have a particularly significant impact on costs.

189

Figure 87 Water treatment facilities’ construction costs

0

500

1000

1500

2000

2500

3000

3500

4000

Sewage TreatmentPlant

Sewage TreatmentPlant

Waste WaterTreatment Plant




Waste WaterTreatment Plant x 3

Norway UK Turkey UK Russia USA UK

2007

Pric

es (€

/m c

ubed

/day

)

Sources: www.water-technology.net Figure 88 Water supply facilities’ construction costs

0

200

400

600

800

1000

1200

1400

1600

1800

2000

Water Treatment Plant Seawater DesalinisationPlant

Desalinisation Plant Desalinisation Plant,SWRO

SWRO DesalinisationPlant

Water Treatment Plant

USA Australia Spain Israel Singapore UK

2007

Pric

es (€

/cub

ic m

/day

)

Sources: www.water-technology.net

ANNEX II – DATA QUESTIONNAIRES

ROADS

Project: Project Ref No:

COST – Insert total cost of the project and final year of expenditure. LAND – Estimate the area of land acquired and the estimated and the out-turn cost.

Year Estimated Cost - € Actual Cost - € Total Cost of Project

SITE CONDITIONS – Please insert percentage of project attributes most appropriate

Attribute % % % Terrain Mountainous Hilly Level Locality Urban Semi Rural Ground Conditions Hard Normal Soft Environmental Constraints Difficult Normal Easy

CARRIAGEWAY – Whether new construction or rehabilitation insert number of carriageways and lanes per carriageway. Include the shoulder in the number of lanes if the width exceeds 3m, also the length, the estimated cost and the actual cost.

i) New Construction ii) Rehabilitation

BRIDGES / TUNNELS – Insert the number of bridges and tunnels and the OTHER – Please insert any other relevant information. total area or length.

Area - Ha Estimated Cost - € Actual Cost - €

No km Est Cost - € Act Cost - € Carriageways Lanes / carriageway


Bridge Type No Deck Area – m2 Est Cost - € Act Cost - € Beam Cantilever Arch Suspension Cable Stay Truss Tunnels No Total Length Est Cost - € Act Cost - € Bored Cut and Cover

Description / Quantity Est Cost - € Act Cost - €

191

TIME – Please provide the estimated and actual periods for the PROCUREMENT – Please provide information on the Contract following stages. procurement process. Insert or delete as appropriate

COST / TIME OVER-RUNS – Provide an assessment of the impact relating to cost or time over-runs. Score 1 to 5, Minimum 1 if little impact, maximum 5 if major impact. Enter zero if not applicable.

Level 1 Level 2 Cost Time Level 1 Level 2 Cost Time Procurement Complexity of Contract Structure Project Environment Public Relations Design Changes Site Characteristics Contractor specific difficulties Permits/Consents/Approvals Disputes with suppliers and

subcontractors External Factors Changes in Legislation /

Regulations

Poor Planning/Methodology errors Political Project Specific Design Complexity Technology Degree of Innovation Inflation Environmental Impact Exchange Rates Site access difficulties Force Majure Suspension of works Other (specify) Delays by statutory authorities

and/or contractors Other

(Please specify)

Late commencement of work Construction period Client Specific Inadequacy of the Business Case Large Number of Stakeholders Funding Availability/Problems Project Management Team

Attribute Estimated Time Period - months

Actual Time Period - months

Planning Stage Securing Funding Permissions / Consents Procurement / Preparation Construction

Attribute Response Contract Type ICE, NEC, FIDIC, Bespoke etc Project Complexity Range from straightforward to complex Procuring Agency Fixed Cost / Remeasure Design responsibility Client, Designer, Contractor Funding Structure Public, PPP etc

192

RAIL






RAIL– single or twin track, elevated, at grade or in tunnel insert estimated cost, actual cost and length.

Diesel / At Grade Elevated In Tunnel Electric Est Cost - € Act Cost - € km Est Cost - € Act Cost - € km Est Cost - € Act Cost - € km Single Twin

STATIONS / ROLLING STOCK–Surface, underground or elevated, insert estimated and actual cost and size.

Rolling Stock Station Type Station Area m2 Platform Area m2 Est Cost - € Act Cost - € Type Nr Est Cost - € Act Cost - € Surface Level Underground Elevated

BRIDGES / TUNNELS – Insert the number of bridges and tunnels and the total area or length. Bridge Type No Deck Area –

m2 Est Cost - € Act Cost - € Bridge Type No Deck Area – m2 Est Cost - € Act Cost - €

Beam Suspension Cantilever Cable Stay Arch Truss


193

OTHER COSTS – Please insert any other relevant information.

TIME – Please provide the estimated and actual periods for the PROCUREMENT – Please provide information on the Contract following stages procurement process. Insert or delete as appropriate

Description / Quantity Est Cost - € Act Cost - € Utilities Power Supply Signalling Communication Park & Ride Depot Other Other Other





Tunnels No Total Length Est Cost - € Act Cost - € Bored Cut and Cover

194




Regulations



(Please specify)


195

URBAN TRANSPORT






FORM OF TRANSPORT– Whether metro, tram or guided bus, single or twin track, elevated, at grade or in tunnel insert estimated cost, actual cost and length.

Diesel / At Grade Elevated In Tunnel Electric Est Cost - € Act Cost - € km Est Cost - € Act Cost - € km Est Cost - € Act Cost - € km Metro Single Twin Tram Single Twin Guided Bus Single Twin

STATIONS / ROLLING STOCK– Whether metro, tram or guided bus, surface or underground, insert estimated and actual cost and size.

Surface Level Station / Stop Underground Station / Stop Rolling Stock m2 Est Cost - € Act Cost - € m2 Est Cost - € Act Cost - € Type Nr Est Cost - € Act Cost - € Metro Tram Guided Bus


196

BRIDGES / TUNNELS – Insert the number of bridges and tunnels and the OTHER – Please insert any other relevant information. total area or length. Bridge Type No Deck Area – m2 Est Cost - € Act Cost - € Beam Cantilever Arch Suspension Cable Stay Truss Tunnels No Total Length Est Cost - € Act Cost - € Bored Cut and Cover


Description / Quantity Est Cost - € Act Cost - € Utilities Power Supply Signalling Communication Park & Ride % Fixed Link Depot Other Other





197




Regulations


and/or sub-contractors Other

(Please specify)


198

WATER / WASTE WATER




POPULATION – Estimate the number of people benefiting from the project.

Population Served - No



WATER MAIN / SEWER – Insert the length of main or sewer provided broken down into whatever details of pipe, main or culvert size is available. If this information is not available estimate the maximum size.

Gravity main / sewer Pressure main / sewer Pipe Dia - mm Length - km Length - km

Est Cost - € Act Cost - €

Total - km

INFRASTRUCTURE – estimate the volume or area of building if any (either not both) and the estimate and out-turn cost. If roads are involved estimate the number of lanes, the length and the costs.

Buildings Type Volume – m3 Area – m2 Est Cost - € Act Cost - €

Roads Description No Lanes Length Est Cost - € Act Cost - €


199

For Mechanical & Electrical and other infrastructure insert a description and estimated and actual costs.

Mechanical & Electrical Description Quantity Est Cost - € Act Cost - €

Other Infrastructure - 1 Quantity Est Cost - € Act Cost - €

Other Infrastructure - 2 Description Quantity Est Cost - € Act Cost - €






200

ENERGY – WIND FARMS





Attribute % % % Terrain Mountainous Hilly Level Ground Conditions Hard Normal Soft Environmental Constraints Difficult Normal Easy

WIND TURBINES Type Capacity (MW) Rotor

Diameter (m)Tower Height (m)

Number Estimated cost - €

Actual cost - €

OTHER COSTS - Please insert any other relevant information Description/ Quantity Estimated cost - € Actual cost - € Electrical substations Cabling/ power lines Earthworks Turbine foundations Other Other Other

Onshore/ Offshore Area - Ha Estimated Cost - € Actual Cost - €

201





Regulations



(Please specify)






PRODUCTIVE INVESTMENTS Costs Please indicate clearly the dates for which estimated and actual cost data are stated (to facilitate adjustments to a common price base). We have extracted the estimated cost information available from the ERDF project application forms supplied by the Commission. Where it is possible to provide more detailed breakdowns (for estimates and actuals), please do so. Cost category Estimate Actual

Total project cost (€)

Private sector funding (€)

ERDF funding (€)

Other public funding (€) (please provide details)

Cost breakdown:

- construction works

- plant & equipment

- land acquisition

- site preparation

- IT infrastructure

- training

- professional fees

- other (1) provide details



FTE jobs created

FTE jobs safeguarded

Total cost per FTE job created

203

Delivery time Please provide estimated and actual delivery times in months. Project stage Estimate Actual

Start date

Finish date General attributes Attribute Description

Project name

Project reference

Country

Approval date

Project type*

Amount of ERDF funding *Please choose from:

• Manufacturing – expansion • Manufacturing – rehabilitation • Manufacturing – new build • Transport infrastructure • Training • Sports/leisure facilities • Investment fund • Other (please specify)

Cost overrun analysis Please rate the reasons for cost overruns according to the following guidelines. Scoring for cause of overruns and delaysNo or insignificant cause of over run or delay 0Minor factor contributing to overrun or delay (<20%) 1Major factor contributing to overrun or delay (20%- 50%) 2Very signif icant contributing factor (>50%) 3 Issue Score

Delays in implementation – planning

204

Delays in implementation – construction

Design changes

Material cost increases/decreases

Inflation

Funding costs

Other 1 (please specify)


Other 3 (please specify) Job creation variance analysis Please rate the reasons for deviations from expected levels of job creation. Scoring for cause of overruns and delaysNo or insignificant cause of over run or delay 0Minor factor contributing to overrun or delay (<20%) 1Major factor contributing to overrun or delay (20%- 50%) 2Very signif icant contributing factor (>50%) 3 Issue Score

Delays in implementation – planning

Delays in implementation – construction

Design changes

Changes in business plan

Demand for services lower/higher than planned

Lack of suitable staff




205

ANNEX III – TIPS FOR PROJECT MONITORING

RAIL

Pro ject:2a Construction dune nouvelle voie ferroviaire entre Thessalonika et Idomeni (Phase1) Polycastro -Idomeni

Project Ref No: 2005 GR 161PR 011

COST – Insert tota l cost o f the project and final year of expenditure. LAND – Estimate the area of land acquired and the estimated and the out-turn cost.

Year 2008 Estim ated Cost - € Actual Cost - € Total Cost of Project 54.600.000 36.690.687

SITE CONDITIONS – Please insert percentage of pro ject attributes m ost appropriate

Attribute % % % Terra in Mountainous 60 Hilly 30 Level 10 Locality Urban Semi 20 Rural 80 Ground Conditions Hard Normal 70 Soft 30 Environmental Constrain ts Difficult Normal Easy

RAIL– single or twin track, elevated, at grade or in tunnel insert estimated cost, actual cost and length.

Diesel / At G rade Elevated In Tunnel Electric Est Cost - € Act Cost - € km Est Cost - € Act Cost - € km Est Cost - € Act Cost - € km Sing le Twin * 17.455.450 11.636.967 8,48 22.500.000 15.000.000 0,92 6.750.000 4.500.000 0,39

STATIONS / ROLLING STOCK–Surface, underground or e levated, insert estimated and actua l cost and size.

Rolling Stock Station Type Station Area m 2 Platform Area m2 Est Cost - € Act Cost - € Type Nr Est Cost - € Act Cost - € Surface Level Underground Elevated

BRIDGES / TUNNELS – Insert the num ber of bridges and tunnels and the total area or length. Bridge Type No Deck Area –

m2 Est Cost - € Act Cost - € Bridge Type No Deck Area – m 2 Est Cost - € Act Cost - €

Beam Suspension Cantilever 2 12 .180 22.500.000 15.000.000 Cable Stay

Area - Ha Estimated Cost - € Actua l Cost - € 76,36 4.600.000 4.800.000

206

Arch Truss

OTHER COSTS – Please insert any other relevant information.

TIME – Please provide the estimated and actual periods for the PROCUREMENT – Please provide information on the Contract following stages procurement process. Insert or delete as appropriate

Attribute Response Contract Type Pumblic works Greek legislation

Project Complexity Complex project

Procuring Agency ERGOSE

Fixed Cost / Remeasure Remeasure Design responsibility Designer / ERGOSE

Funding Structure 75% EU Fund 25% National Fumd

Description / Quantity Est Cost - € Act Cost - € Public Utilities networks

Relocation of P.U (PPC, ΟΤΕ, ΕΥΔΑΠ)

250.000 30.000

Traction power - - -

Signalling - - -

Telecomunicatin - - -

Archaology Archeological excavation 100.000 172.396,65 Design Initial estimation: Track work

design, station design, preparation of tender documents Current status: The tender was out only for design of stations and stops , The track work design was prepared in house by ERGOSE.

2.900.000 551.324

Other



Planning Stage 30

31

Securing Funding 3 3 Permissions / Consents 36 56 Procurement / Preparation 8 6 Construction 18 23

Tunnels No Total Length Est Cost - € Act Cost - € Bored Cut and Cover 2 380 6.750.000 4.500.000

207


Level 1 Level 2 Cost Time Level 1 Level 2 Cost Time Procurement Complexity of Contract Structure 1 1 Project Environment Public Relations 1 1 Design Changes 1 1 Site Characteristics 1 1 Contractor specific difficulties 0 0 Permits/Consents/Approvals 1 1 Disputes with suppliers and

subcontractors 0 0 External Factors Changes in Legislation /

Regulations 0

0

Poor Planning/Methodology errors 0 0 Political 0

0

Project Specific Design Complexity Technology 0 0 Degree of Innovation 3 3 Inflation 0 0 Environmental Impact 1 1 Exchange Rates 0 0 Site access difficulties 0 0 Force Majure 0 0 Suspension of works 0 0 Other (specify)

5 5

Delays by statutory authorities and/or contractors

5 5 Other (Please specify)

Late commencement of work 0 0 Construction period Client Specific Inadequacy of the Business Case 0 0 Large Number of Stakeholders 0 0 Funding Availability/Problems 0 0 Project Management Team 0 0

208

ROADS

Project: Autovia de la Plata. CN-630 de Gijon a Sevilla. Tramo Fuente de Cantos Project Ref No: 2004ES161PR023


Year 2007 Estimated Cost - € Actual Cost - € Total Cost of Project 35.234.168,56 47.881.842,93


Attribute % % % Terrain Mountainous 0 Hilly 20 Level 80 Locality Urban 0 Semi 0 Rural 100 Ground Conditions Hard 19 Normal 34 Soft 47 Environmental Constraints Difficult 0 Normal 10 Easy 90

CARRIAGEWAY – Whether new construction or rehabilitation insert number of carriageways and lanes per carriageway. Include the shoulder in the number of lanes if the width exceeds 3m, also the length, the estimated cost and the actual cost.

i) New Construction ii) Rehabilitation

BRIDGES / TUNNELS – Insert the number of bridges and tunnels and the OTHER – Please insert any other relevant information. total area or length.

Area - Ha Estimated Cost - € Actual Cost - € 128,5 1.110.773,80

No km Est Cost - € Act Cost - € Carriageways 2 11,268 17.617.084,28 23.940.921,51 Lanes / carriageway 2 11,268 8.808.854,14 11.970.460,73


Bridge Type No Deck Area – m2 Est Cost - € Act Cost - € Beam 2 1.386 931.239,79 1.299.028,37Cantilever 6 3.570 1.591.378,82 2.087.353,29Arch Suspension Cable Stay Truss Tunnels No Total Length Est Cost - € Act Cost - € Bored Cut and Cover 3 123 942.351,41 1.527.755,30

Description / Quantity Est Cost - € Act Cost - €

209



Level 1 Level 2 Cost Time Level 1 Level 2 Cost Time Procurement Complexity of Contract Structure 3 3 Project Environment Public Relations 1 1 Design Changes 2 3 Site Characteristics 1 1 Contractor specific difficulties 1 1 Permits/Consents/Approvals 1 2 Disputes with suppliers and

subcontractors 1 1 External Factors Changes in Legislation /

Regulations 0 0

Poor Planning/Methodology errors 0 0 Political 0 0 Project Specific Design Complexity 2 2 Technology 3 3 Degree of Innovation 3 3 Inflation 1 1 Environmental Impact 1 1 Exchange Rates 0 0 Site access difficulties 1 1 Force Majure 0 0 Suspension of works 1 1 Other (specify) 0 0 Delays by statutory authorities

and/or contractors 0 0 Other

(Please specify)

Late commencement of work 1 1 Construction period 2 2 Client Specific Inadequacy of the Business Case 0 0 Large Number of Stakeholders 1 1 Funding Availability/Problems 0 0 Project Management Team 4 4



Planning Stage 40 80 Securing Funding 24 24 Permissions / Consents 12 12 Procurement / Preparation 3 3 Construction 42 35

Attribute Response Contract Type Project Complexity medium Procuring Agency Fomento Fixed Cost / Remeasure Fixed Cost Design responsibility Designer Funding Structure Public

EX POST EVALUATION OF COHESION POLICY PROGRAMMES...

Documents

Transcript of EX POST EVALUATION OF COHESION POLICY PROGRAMMES...