Appendix 3 Algarve railway in Portugal

Transcription

1 March 2011 Frontier Economics, Atkins, ITS Introduction Project overview Location The Lisbon Algarve railway project covers a significant part of the overall modernisation of the Algarve railway line (covering the Linha do Sul and Linha do Algarve). The part included in the current study covers the segment from Coina (near Lisbon) to Faro, including a branch to Porto de Sines. The project is part of the TEN-T Priority Project 8 (Multimodal Axis Portugal/Spain Rest of Europe). Figure 1. Algarve line from Lisbon to Faro Lisbon Pinhal Novo Coina PK 94 Sines Ermidas Funcheira Tunes Faro Source: Openstreetmap.org Description During the period 2000 to 2006, the Cohesion Fund contributed to the funding the following subprojects of the Algarve railway line: Subproject 2000PT16CPT003: This subproject covered two sections:

2 2 Frontier Economics, Atkins, ITS March 2011 Pinhal Novo Poceirão Pinheiro Ermidas (excluding section from Pinheiro to Km 94). Work on this section included the complete renewal of 35 kilometres of track, the construction of sidings for another 30.5 kilometres and the installation of the automatic control system. Linha de Sines (Ermidas to Porto de Sines). This involved mostly minor works as well as the partial renewal of 5 kilometres of track Subproject 2000PT16CPT012: Coina Pinhal Novo. This subproject involved the construction of a 800-metre tunnel, a 475-metre viaduct, and three new stations and 15 kilometres of track duplication. Subproject 2001PT16CPT001: Pinheiro to Km 94. This subproject involved the modernisation of 35 kilometres of track and the installation of the automatic control system. Subproject 2001PT16CPT003: This subproject covered two sections: Ermidas Faro. This involved the redoubling of 20 kilometres of track and the renewal of over 138 kilometres of track (complete renewal for 41 kilometres and partial renewal for 97 kilometres). It also involved the renewal of twelve stations, and the construction of a viaduct, a bridge and a station. Funcheira Ourique. This involved the complete electrification of the railway section and the installation of an automatic control system. The modernisation work started in 1997 and, according to the ex post capital expenditure data provided by REFER, it was mostly completed by However, based on the ex post passenger data, we believe that the line has been open to traffic since 2003, even without reaching maximum capacity. This maximum capacity on the line should coincide with the end of 2006 and beginning of 2007, when capital expenditures in the line stopped. Final reports show some delays in the completion of the projects. These time delays range between one and two years. Further evidence of delays happening in the project is the time indicated by the ex ante CBA regarding when the capital investments were supposed to take place and the actual time when these capital investments happened. The total aggregate value for the project was 419,026,183, while the total amount eligible for funding was 405,043,302 or 96 per cent of the total. The total Cohesion Fund contribution in the period was 323,486,946, equal to about 80% of the eligible project costs.

3 March 2011 Frontier Economics, Atkins, ITS Context Socio-economic context The line connects by rail Lisbon and the Algarve region in Portugal, and allows for future improvements in the railway connections between Portugal and the south of Spain, in Andalucía. It is worth noting that before the modernisation of the line started, there was no direct rail connection between Lisbon and the Algarve region, the renovation of the existing track was required and the line had important speed limitations due to the track and the obsolete signalling equipment. The part included in the current study covers the segment from Coina (near Lisbon) to Faro, including a branch to Porto de Sines. The project involved, among other works, the duplication of the track in some sections of the line, the electrification of parts of the line which were previously non-electrified, the construction of various new stations, and the installation of a new automatic control system to increase maximum speed in various sections of the line. Strategic policy context At the time when the ex ante cost benefit analysis was carried out, and the applications for funding submitted, the stated objectives of the project were: to increase the maximum speed on the line to 220km/h; to reduce the journey time between Lisbon and Faro to 2h 45min; to allow for better railway connections between the south and the north of Portugal; to reduce the operating costs of the line; to improve safety conditions and the quality of service on the line; and to reduce transportation costs and increase productivity levels in the area covered by the railway line. As mentioned above, the project is part of the TEN-T Priority Project 8 (Multimodal Axis Portugal/Spain Rest of Europe). This axis is expected to reinforce multimodal corridors linking Portugal and Spain, to contribute to the improvement of links between the centre of the EU and its peripheral regions, and to strengthen the Iberian Peninsula s position as a western European gateway. In relation to the objectives of the TEN-T Priority Project 8, the project contributes to the development in Portugal of a service platform that articulates the long-run transport between Europe on one side and America and Africa on

4 4 Frontier Economics, Atkins, ITS March 2011 the other side. This is one of the objectives stated by the Portuguese government in its transport policy strategy for the period Sources To carry out the analysis we have relied on a variety of different sources of information provided by DG REGIO and by stakeholders in Portugal. This allowed us to review the ex ante analysis undertaken to apply for EU funds. We have obtained Funding Applications and Funding Decisions for each of the four subprojects listed above from DG REGIO. Each application provided a detailed description of each subproject, its objectives and the costs and benefits that it was expected to generate. DG REGIO also provided additional supporting documentation, such as the initial overall cost-benefit analysis and final reports for each of the four subprojects to check the section have been completed. In addition, REFER also made available the spreadsheet containing the calculation underpinning the initial comprehensive cost-benefit analysis of the project The complete list of documents obtained, mainly related with the review of the ex ante analysis, is provided in Table 1.

5 March 2011 Frontier Economics, Atkins, ITS 5 Table 1. Summary of project-related documentation Documents Funding Applications All subprojects Funding Decisions All subprojects Ex ante CBA Intraplant Consult study (1996) Ferbritas study (1997) Pinhal Novo to Faro by REFER, covering 3 subprojects Ex ante CBA Spreadsheet calculation Pinhal Novo to Faro, covering 3 subprojects EIB Initial and Final reaction Subproject 2000PT16CPT003 Evaluation study (POVT) Subproject 2001PT16CPT001 Ex post evaluation, Projecto Lisboa Algarve Slide pack presentation of the completed project Environmental Impact reports All subprojects Final Payment declarations All subprojects Ex ante CBA Variante da Trofa (2010) Current ex ante CBA of a different rail link Obtained from DG REGIO DG REGIO DG REGIO and REFER REFER DG REGIO DG REGIO REFER DG REGIO DG REGIO REFER To carry out an ex post analysis of the project, REFER and CP (the train operating company) made available significant amounts of data, including outturn costs (capex and opex) and passenger numbers and updated values for other parameters (such as Value of Time). We provide a list of the data provided in Table 2.

6 6 Frontier Economics, Atkins, ITS March 2011 Table 2. Summary of Primary & Secondary Data Availability Data Source Issues Passenger and Freight numbers Infrastructure investment costs Infrastructure operating costs Train operating costs Revenues Value of time REFER REFER REFER REFER and CP REFER and CP REFER REFER and CP provided average operating costs for electric rolling stock CP provided average fares per passenger Based on two studies: For passengers: Calculating Transport Congestion and Scarcity Costs Final Report of the Expert Advisors to the High Level Group on Infrastructure charging (Working Group 2) May 1999 ; and For freight: Developing Harmonised European Approaches for Transporting Costing and Project Assessment Annex A An International Meta-Analysis of Values of Time March 2000 Vehicle operating costs REFER Based on 2000 and 2004 INFRAS/IWW studies Marginal costs of road travel and congestion Number of trains and train capacity REFER Based on 2000 and 2004 INFRAS/IWW studies REFER We had a meeting with REFER, CP, and POVT (Programa Operacional Temático Valorização do Território, the government programme for the implementation of infrastructure projects) with two key objectives: To clarify our understanding of the calculations and assumptions underpinning the ex ante CBA; and, To discuss the usefulness of the CBA in the decision-making process as experienced in the pilot project experience and in Portugal. Throughout this pilot project study we have also received regular support and advice (via and telephone) from POVT, REFER and CP in Portugal and from representatives of the DG REGIO Portuguese Geographic Unit Overview of any overall issue and challenges with the evaluation. REFER carried out two separate ex ante CBAs. The first covers three sections of the project, between Pinhal Novo and Faro. The second CBA, for the section

7 March 2011 Frontier Economics, Atkins, ITS 7 Coina-Pinhal Novo, covers a longer suburban route between Chelas and Pinhal Novo. It was not possible to extract the ex ante results applicable just to the Coina-Pinhal Novo section. For this reason, we have carried out an ex post CBA only for the section Pinhal Novo-Faro, which is directly comparable to the ex ante CBA. 3.2 Ex post cost-benefit analysis This section presents the results of the ex post evaluation of the section Pinhal Novo Faro Headline results from the analysis This section describes the key results obtained from the analysis we have carried out to assess the ex post outcome of this project, using historic data until Further details could be found in the Annexes Economic analysis We have identified two scenarios, a pessimistic one (low case) and an optimistic one (high case), which define a range of possible outcomes for the ex post CBA. To define the pessimistic scenario (Low case), we have considered the low outcome for benefits (driven by traffic volume). We have used the opposite assumption (i.e. high outcome for benefits) to define the optimistic scenario (High case). Table 3 summarises the results of the ex post analysis for each scenario. Table 3. Summary of ex post economic analysis (2008 prices) Low case High case Net Present Value ( m) Economic IRR (%) 6.7% 7.5% Benefit-cost ratio Source: Own calculation According to our analysis, the ex post NPV of the project, assessed for the period 1997 to 2004 and discounted at 5.5%, ranges between 48.5m and 79.5m in 2008 prices. The economic IRR ranges between 6.7% and 7.5%. Finally, the benefit-cost ratio ranges between 1.15 and 1.24, implying that the economic benefits of the project exceed its costs. The detailed results of the analysis are provided in Annexe 1.

8 8 Frontier Economics, Atkins, ITS March 2011 Financial analysis We have also carried a financial analysis for the Low case and High case scenarios. For each scenario, we have calculated the Financial Net Present Value and Financial Rate of Return, for both the investment FNPV(C) and FRR(C) and the capital employed FNPV(K) and FRR (K). Finally, by taking into account the EU and national contribution, we have also undertaken a financial sustainability analysis. Table 4 summarises the results of the financial analysis for each scenario. The analysis shows that, under each scenario, the project is not revenue-generating as the net present value of the costs over the assessment period exceeds the value of revenues. Table 4. Summary of ex post financial analysis (2008 prices) Low case High case Net Present Value Investment ( m) Financial IRR Investment (%) -3.6% -3.6% Net Present Value Capital ( m) Financial IRR Capital (%) 0.17% 0.17% Source: Own calculation The detailed results of the analysis are provided in Annexe 1. Wider socio-economic impacts As this project involved the upgrading of an existing railway line, rather than the construction of a new route, most of its benefits would come from time and maintenance savings, as well as from positive externalities such as the reduction in noise and pollution. These effects are already largely captured by the core CBA. Moreover, the lack of any form of impact monitoring since the opening, makes identifying a causal link between the project and any wider effect very problematic. In this kind of situations, it is difficult to separate the impacts of the project from the impact of other ongoing changes in the economy. Nonetheless, there is some evidence that per capita income might have improved in the regions connected by the railway line. It is however problematic to attribute these changes just to this project.

9 EUR (Millions) March 2011 Frontier Economics, Atkins, ITS Costs Costs are divided into two different categories. One-off costs According to REFER s historic data, the first expenditure on the Pinhal Novo- Faro section was made in 1997, on the sub-section Pinhal Novo-Poceirão- Ermidas, while the last substantial one was made in 2007, on the sub-section Ermidas-Faro. According to the cost data we have had access to, construction of the suburban section Coina-Pinhal Novo began in 2000 and was completed in Figure 2 shows the evolution of capital expenditure over time between 1997 and Figure 2. Capital expenditure 1997 to 2009 (, 2008 prices) Source: REFER We assume that REFER will not incur any further capital costs for both sections of the project as they are both complete. We further assume that, had REFER not undertaken this project, it would have not incurred any capital cost. 1 1 We recognise that this assumption is very conservative. It is likely that REFER would have incurred capital costs even in the absence of the intervention, mainly for renewals. However, as the ex ante CBA does not provide values for the do-nothing scenario, we have not been able to obtain an appropriate estimate for capital costs in the counterfactual. Assuming that REFER would have not incurred additional costs in the do-nothing scenario implies an overestimation of capital costs in the ex post evaluation

10 10 Frontier Economics, Atkins, ITS March 2011 Unit cost calculation REFER to provided disaggregated cost information by components and to provide unit costs for the various components. Table 5 presents, both in nondiscounted and in NPV terms (discounted at 5.5%), the unit costs of different investments (all in 2008 prices) we have been able to calculate using the information we received from REFER. Table 5. Summary of ex post unit costs (2008 prices) Non-discounted NPV (@ 5.5%) Level 1 All-in ( /km) 1,723,149 1,286,975 Track ( /km) 947, ,120 Stations ( /station) 4,084,850 2,995,137 Level 2 Stations ( /m2) Bridges ( /bridge) 3,053,717 2,237,135 Bridges ( /m2) 1,860 1,362 Source: Own calculation based on REFER data Ongoing costs. Operation costs REFER also provided data on historic operating costs for the period For the Pinhal Novo-Faro, operating costs start at about 8m per annum in 2004 and increase to 15m per annum in 2009 (with a peak of 16m in 2008).

11 EUR (Millions) March 2011 Frontier Economics, Atkins, ITS 11 Figure 3. Operating expenditure 2004 to 2009 (, 2008 prices) Source: REFER Because we were not provided with historic total operating cost information for CP, the train operator, it was calculated the average operating cost for electric rolling stock ( per train/km) reported by REFER and CP together with utilisation data described below. Also, to approximate the incremental operating costs in the ex post analysis, we assume that operating costs were 20% higher than what they would have been in a do-nothing scenario. Results are shown in Table 6. We note that, due to the uncertainty surrounding the assumptions that underpin this calculation, these values should be treated only as an indicative estimate of historic ex post costs.

12 12 Frontier Economics, Atkins, ITS March 2011 Table 6. Calculation of CP s operating costs (2008 prices) Passengers (000s) Passenger kms (millions)* Train kms (millions) Operating costs ( millions) Source: Own calculation using REFER and CP parameters (*) Assuming average travel length of 150 km Foregone government revenues The modal shift from road into rail implies that there will be a reduction in government revenues from fuel tax. REFER do not estimate this cost in the 1997 analysis. However, we believe that the ex post analysis should consider this effect to fully capture the costs of the project. Table 7 shows our estimate of the amount of foregone fuel tax revenues for the period 2003 to Table 7. Foregone revenue from fuel tax (2008 prices) million Foregone revenues Source: Our calculation, based on data from the OECD/EEA database on instruments used for environmental policy and natural resources management ( The future foregone fuel tax revenues are driven by the number of passenger kms on the line and they have been obtained using the average consumption for cars and light goods vehicles based on a small sample of popular vehicles 2. 2 For cars, we considered the extra-urban consumption of a Volkswagen Golf, a Seat Ibiza and a Renault Megane. For light goods vehicles, we considered the extra-urban consumption of a Renault Kangoo and a Volkswagen Caddy.

13 March 2011 Frontier Economics, Atkins, ITS Direct benefits To calculate the benefits of the project, we use the same approach, which REFER used for the 1999 cost-benefit analysis. However, REFER has since updated the parameters it uses to estimate the monetary value of time savings and of any reduction in road traffic. Given that these values are those currently used for assessing Portuguese rail projects, we believe it is appropriate to use them for this analysis. Traffic volumes For the section Pinhal Novo-Faro, REFER/CP provided us with the historic passenger numbers from 2003 the year of opening to Table 8 summarises the historic passenger numbers, and our assumptions relative to the breakdown into different sources of traffic. Table 8. Historic number of passengers (000s) (Pinhal Novo Faro) Total Existing New Of which: Induced (49%) Modal shift from cars (46%) Modal shift from collective vehicles (5%) Source: REFER/CP and our calculation based on initial REFER assumptions on sources of new traffic Because no forecast passenger numbers were available, we have decided to use a range of possible values for passenger numbers between 2009 and 2024, based around the assumptions of the ex ante analysis. Also we assume that the breakdown between induced traffic and modal shift remains constant over time. Regarding freight volumes, REFER provided data on total tonnes transported on the line between 2004 and 2009, as shown in Table 9.

14 14 Frontier Economics, Atkins, ITS March 2011 Table 9. Historic freight volumes (Pinhal Novo Faro) Million tonnes Freight Source: Our calculation based on CP data and initial REFER assumptions on sources of new traffic Time savings REFER now assumes a value of (in 2008 prices) per minute for each passenger. This value is based on the study Calculating Transport Congestion and Scarcity Costs Final Report of the Expert Advisors to the High Level Group on Infrastructure charging (Working Group 2) May In addition to considering time saving for passengers, REFER now also considers time savings for freight. Specifically, it assumes a benefit of 1.22 per hour per tonne of freight, based on the study Developing Harmonised European Approaches for Transporting Costing and Project Assessment Annex A An International Meta-Analysis of Values of Time March Finally, based on the ex post data provided, we assume that the average time saving per journey is 75 minutes. To calculate the benefits originating from modal shift, we assume that the average length of the journeys that are shifted away from road to rail is 150 km. This is the same assumption that REFER used in the ex ante CBA. In the absence of more detailed information, we believe it is appropriate to use the same assumption. Resource cost savings REFER has also updated the values it uses to estimate the resource cost savings associated with moving traffic away from road. These savings are due to a reduction in vehicle operating costs and road maintenance costs. REFER now assumes the cost for cars to be per kilometre and that each car carries, on average, 1.8 passengers. This implies a cost per passenger/km for cars of about For buses and goods vehicles, REFER assumes that the cost is both per passenger/km and per tonne/km. Revenues In the ex ante analysis, REFER considers only its expected incremental revenues under Option MAX, compared with the do-nothing scenario. However, for an ex ante CBA to capture the entire set of benefits, all revenues, including those accruing to the train operator, should be included. Table 10 shows the evolution of estimated incremental revenues from 2003 to 2009 using average fares per passenger km and passengers traffic distribution by

15 March 2011 Frontier Economics, Atkins, ITS 15 train type provided by CP. The assumption that total existing passengers do not change over time implies that revenues in the do-nothing scenario do not change over time. We have made this assumption in the absence of more specific information. Table 10. Estimated historic CP s incremental revenues (2008 prices) millions Total ex post revenues Do-nothing revenues Incremental revenues Source: Our calculation based on CP data and initial REFER assumptions on sources of new traffic Finally, we note that we have not been able to obtain sufficient information to estimate incremental revenues associated to the increase in freight volumes. However, as this amount is likely to be relatively small compared to total benefits, we do not consider its exclusion to have a significant impact on the overall ex post CBA. Terminal value In the ex ante analysis, REFER depreciates capital expenditure over 40 years, using a linear depreciation approach. For the ex post analysis, we apply the same approach to the historic capital expenditure that REFER incurred between 1997 and Externalities To calculate the total annual benefit of the project arising from reduction in pollution, climate change, noise, accidents and congestion, we multiply the unitary values shown below by the number of passengers assumed to travel on the line after re-opening that previously were travelling using road transport. Therefore these benefits only apply to modal shifters. In fact, they are the cost reductions achieved thanks to removing traffic from the roads. Pollution, climate change, noise and accidents Savings from reduction in pollution, climate change (mundança de clima), noise and accidents have been estimated. REFER has updated these values on the basis of two INFRAS/IWW studies. Specifically, these are External Costs of Transport Accident, Environmental and Congestion Costs in Western Europe March 2000 and

16 16 Frontier Economics, Atkins, ITS March 2011 External Costs of Transport Update Study, Final Report October The following tables summarise these marginal cost values, for both passengers and freight. We have used these values for our assessment of the ex post benefits of the project. Table 11. Marginal costs of road travel passengers ( per passenger/km) (2008 prices) Car Bus Light Goods Vehicle Heavy Goods Vehicle Pollution Climate change Noise Accidents Source: REFER, based on 2000 and 2004 INFRAS/IWW studies Congestion Savings from reduction in congestion have also been calculated. In the ex ante analysis, REFER did not have any reliable estimate for these benefits. For this reason, the ex ante CBA does not quantify the monetary value of the reduction in congestion associated with modal shift. However, REFER now includes an estimate of this cost in its cost-benefit analyses. Therefore, we have included a quantification of these benefits in our ex post analysis. The marginal values that REFER uses are shown in the following tables. Table 12. Marginal costs of congestion passengers ( per passenger/km) (2008 prices) Car Bus Light Goods Vehicle Heavy Goods Vehicle Congestion Source: REFER, based on 2000 and 2004 INFRAS/IWW studies Wider socio-economic impacts We have discussed the wider socio-economic impact of this project with stakeholders. It should be noted that this project involved the upgrading of an existing railway line, rather than the construction of a new route. Therefore, most of the benefits from this project would come from time and maintenance

17 March 2011 Frontier Economics, Atkins, ITS 17 savings, as well as from positive externalities such as the reduction in noise and pollution. These effects are already largely captured by the core CBA. Moreover, the lack of any form of impact monitoring since the opening, makes identifying a causal link between the project and any wider effect very problematic. In this kind of situations, it is difficult to separate the impacts of the project from the impact of other ongoing changes in the economy. Socio-economic impacts Information from Statistics Portugal (Instituto Nacional de Estatística) provided by REFER shows that those counties (concelhos) that have a direct relation with the Lisbon Algarve railway have increased their IPC (Indicador per Capita) with the exception of Setúbal. This index compares the purchasing power of a county with the average purchasing power of the whole country. Table 13 presents the figures. Table 13. IPC of Portuguese counties (Portugal = 100) County Palmela 92,77 90,03 103,96 Setúbal 119,94 114,46 113,03 Alcácer do Sal 52,47 58,58 80,96 Grândola 72,40 72,02 94,16 Santiago do Cacém 76,60 76,90 94,55 Ourique 49,27 56,44 64,59 Odemira 52,39 53,17 68,92 Silves 68,05 64,27 80,30 Source: Instituto Nacional de Estatística Environmental impacts REFER provided a list of studies related with environmental impact of the railway. These are listed in Table 14. All studies were undertaken either during construction or immediately after opening.

18 18 Frontier Economics, Atkins, ITS March 2011 Table 14. List of studies on environmental impact Section Date Author Traffic data? Estimation of Emissions Estimation of Noise effect Pinhal Novo / Setubal Jun Gibb Portugal No No Yes Km 94 / Ermidas Jan Febritas / Seia Yes Yes Yes Ermidas / Funcheira Sept Febritas / Seia Yes Yes Yes Pinheiro / Km 94 May 2003 Febritas /Ecossistema Yes Yes Yes Source: REFER According to these studies, there were no significant actions to undertake, apart from those already anticipated to mitigate any environmental impact during construction or afterwards. The only additional action was the implementation of a noise barrier in Funcheira, according to legal requirements. Additional issues According to stakeholders, the success of the project may have been limited by the capacity on the 25 de April bridge for medium and long distance trains in and out of Lisbon. Because of this constraint, according to REFER, journey times to Setúbal, Évora, Beja and Faro are normally higher than for road transport. Other projects RAVE (Rede Ferroviária de Alta Velocidade), is the Portuguese company whose mission is to develop and co-ordinate the necessary studies and projects to facilitate decisions on the planning, construction, financing, provision and operation of a high-speed rail network to be created in mainland Portugal and its connection with a similar Spanish network. Three routes are being considered: Lisbon Porto, Porto Vigo (Spain), and Lisbon Madrid. The latter will use part of the Lisbon Algarve railway track, specifically the Lisbon Évora section. According to RAVE, the Lisbon Porto service will be launched in 2015 and will reduce journey time from 2h35 to 1h15. The one hour Porto Vigo (Spain) service will be finished in Finally the Lisbon Madrid high-speed train will

19 March 2011 Frontier Economics, Atkins, ITS 19 start functioning in 2013 and will reduce travel time from 9 hours to less than 3 hours. Utilisation rates As requested by the TORs, we have considered the evolution of the utilisation rate of the project since opening. We have calculated this by considering, for each year, the ratio between the actual total number of passengers and the total potential number of passengers that could have travelled on the line given the installed capacity. Due to lack of appropriate data we consider only long distance trains ( longo curso ) and, based on the information received by REFER, we have estimated the average capacity to be 230 passengers per train. Table 15 shows the total number of trains between Pinhal Novo and Faro for each year in the period The table also shows the total maximum capacity available, the actual number of long distance passengers that travelled on the line during the same period and the utilisation rate. Table 15. Utilisation rates on the Pinhal Novo Faro route ( ) Number of trains 4,869 4,980 4,483 Maximum number of passengers (000s) Actual number of longdistance passengers (000s) 1,124 1,150 1, Utilisation rate 58% 61% 63% Source: Own calculation based on REFER data Based on our calculation, we have found that utilisation rates, on long distance trains only, ranges between 58% and 63%, with an average of about 61%. However, we note that, due the assumptions underpinning this calculation and the fact that it considers only one type of train service, this result should be treated as indicative Uncertainty and sensitivity analysis Despite the provision of generally comprehensive historic information we have had to assume certain parameters to be able to estimate some of the project impacts. This makes the results of the ex post evaluation dependent on these assumptions thereby increasing uncertainty.

20 20 Frontier Economics, Atkins, ITS March 2011 Historic data and current impacts Information provided by REFER/CP on historic passenger numbers from 2003 to 2009 should be treated with careful because it relies on strong assumptions: Existing passenger volumes before the implementation of the project. REFER/CP confirmed the assumption of 410,000 passengers per annum used in the original CBA. REFER has assumed that, in the do-nothing scenario, the total number of passenger would not change. New passengers. The data does not differentiate between induced traffic (that is, passengers that were not travelling before) and traffic from modal shift from road. Therefore, in the absence of specific information, we have used the same assumptions underpinning the traffic forecasts in the ex ante analysis. Forecast of impacts As mentioned above, we have identified two scenarios, a pessimistic one (low case) and an optimistic one (high case), which define a range of possible outcomes for the ex post CBA. The two scenarios differ on the traffic volume considered. Specifically, we have decided to use a range of possible values for passenger and numbers and freight volume between 2010 and 2024, based around the assumptions of the ex ante analysis. Low case: we assume that the total number of passengers reached in 2009 (824,000) remains constant until the end of the appraisal period. We also assume that the total number of tonnes of freight is 1,863,000 (subtracting 10% to the central value assumed) High case: we assume that the total number of passengers grows from the level reached in 2009 (824,000) to the value assumed in the ex ante CBA at the end of the period (830,000). We also assumed that the total number of tonnes of freight is 2,277,000 (adding 10% to the central value assumed) 3.3 Review ex ante cost-benefit analysis The first assessments of this project were carried in and These were mainly financial assessments and did not include an economic CBA. In Intraplan Consult Gmbh, Previsões de Procura e Avaliação Financeira para a Linha Lisboa/Algarve (November 1996) 4 Ferbritas/SMA, Estudo de Modernização da Linha Lisboa/Algarve (February 1997)

21 March 2011 Frontier Economics, Atkins, ITS 21 REFER extended these studies by preparing an ex ante CBA of the main implementation options. We focus our ex ante review on this analysis. 5 Given that it was not possible to extract an ex ante CBA for the Coina-Pinhal Novo section, the remainder of this review focuses on the main ex ante CBA which covers the route between Pinhal Novo and Faro Headline results from the analysis Economic analysis REFER carried out a complete ex ante CBA for two options, option MAX and option MIN. REFER assessed each option for a 28-year period, between 1997 and It assumed that capital costs would be incurred during the first three years. The line would open in 2000, with operating costs and revenues accounted for from that year onwards. Economic benefits (Beneficios Sociais) are also included from 2000 onwards. The CBA report presents the results of the calculations of benefits and costs for each year of the appraisal period. The CBA report shows the Net Present Value and the economic rate of return of each option, but it does not report the benefit-cost ratios. REFER calculates the NPV of the project using a 5% discount rate. Table 16 summarises the results of the cost-benefit analysis for both Option MAX and Option MIN. Table 16. Results of ex ante cost benefit analysis considering the period. Net Present Value (PTE million, 1997 prices) Net Present Value ( million, 2008 prices) Economic IRR (%) Option MAX 14, % Option MIN 16, % Source: REFER, Modernição da ligação ao Algarve, Estudio Económico-Financeiro, May 1999 Despite this lack of narrative, the financial and economic analyses included in the application forms for the various projects only provide the results corresponding to Option MAX. In addition, the 2000 revision of the ex ante CBA declares that 5 An updated analysis, from 2000, is also available. However, it only reports the results for one of the implementation options. The 1999 document is the last one providing the results of the CBA of all the main implementation options. 6 We have briefly reviewed the ex ante CBA covering the section between Chelas and Pinhal Novo. We believe that REFER used a similar approach to the one used in the section Pinhal Novo and Faro to carry out this CBA. Therefore, we believe that the methodological observation presented in Section 4.3 could be applied also to the Chelas to Pinhal Novo ex ante CBA, albeit at a high level.

22 22 Frontier Economics, Atkins, ITS March 2011 the option selected in the analysis coincides with Option MAX, 7 and as such only results corresponding to this option are shown. It is important to mention that the 2000 revision was recalculated with the inclusion of freight benefits that REFER omitted in the previous analysis. That version presents the results of the updated CBA, showing a benefit-cost ratio of 1.6. We note that this result refers only to three out of four subprojects, as it excludes the assessment of the section Coina Pinhal Novo. Table 17. Results of ex ante cost benefit analysis updated in 2000 Net Present Value (PTE million, 1997 prices) Net Present Value ( million, 2008 prices) Economic IRR (%) Benefit- Cost Ratio Option MAX 23, % 1.6 Source: REFER, Modernição da ligação ao Algarve, Estudio Económico-Financeiro (Actualização), November 2000 For comparisons purposes, we will consider this result when considering the ex post benefit-cost ratio of the project. Financial analysis The ex ante CBA documents we have reviewed provide the results of the financial analysis, but do not discuss the methodology REFER used to carry it out. Instead, the CBA documents simply refer to the 1997 and use the results of the financial analysis as starting point for the economic analysis. Table 18 presents the results of the financial analysis, showing the results of the financial analysis with and without EU contribution. REFER calculated the financial NPV using a 5% discount rate. 7 The ex ante REFER CBA of November 2000, used in the final version of the various application forms, literally indicates that the option chosen is Scenario C. The original ex ante CBA carried out by FERBRITAS in 1997 makes explicit that Scenario C corresponds to Option MAX.

23 March 2011 Frontier Economics, Atkins, ITS 23 Table 18. Results of ex ante cost benefit analysis NPV (PTE million, 1997 prices) NPV ( million, 2008 prices) FRR (%) Option MAX without EU grant Option MAX with 80% EU grant Option MIN without EU grant Option MIN with 80% EU grant -10, % 20, % -5, % 23, % Source: REFER, Modernição da ligação ao Algarve, Estudio Económico-Financeiro, May Key aspects of ex ante CBA The ex ante CBA considers two main options for implementation. These options are not defined within the CBA itself, but they are discussed in the financial study of February The 1997 study presents a very high level overview of the reference base case and of four alternative implementation scenarios. For each option, it provides a brief description of what type of works it entails, an indication of the type of rolling stock that would be used, its freight capacity and the estimated journey time. No further details are given. After providing this brief overview, the report selects, without any apparent justification, two of these four scenarios as the alternative implementation options. From this point onwards, these options, labelled option MAX, involving more significant works on the line, and option MIN, involving smaller interventions, are the only two alternatives that the CBA considers. Traffic forecast Using the information we received from REFER and the spreadsheet that was used to carry out the analysis in 1999, we have been able to re-construct the traffic forecasts assumptions underpinning the main cost-benefit analysis, including the implicit assumption about the number of existing passengers. Table 19 summarises the results of our analysis. We note that the assumptions on new traffic generated are lower than in the 1997 study. The shares of the various sources of traffic are similar to the initial assumptions, although now air

24 24 Frontier Economics, Atkins, ITS March 2011 and bus travel are not considered any longer. 8 Instead, REFER considers modal shift from cars and light good vehicles. 9 We note that the assumption on modal shift 51% of total new traffic is particularly high. However, this may be justified on the basis of the significant improvements to the line, which might have made rail a viable alternative for a large number of travellers. The analysis does not provide any justification for this assumption. Table 19. New annual passengers (in steady state) 1999 ex ante CBA Option MAX Option MIN Total existing passengers 410, ,000 Total new passengers 420, ,000 Of which: Induced traffic 205,800 (49%) 166,600 (49%) Modal shift from cars 193,200 (46%) 156,400 (46%) Modal shift from light goods vehicles 21,000 (5%) 17,000 (5%) Source: REFER, spreadsheet underpinning 1999 CBA calculations The analysis assumes that the number of new passengers from modal shift gradually builds up over the first four years since opening. The report does not disclose this assumption neither provides a justification for it. However, we have been able to re-construct it from the spreadsheet underpinning the analysis. Table 20 shows the increase in the number of new passengers from modal shift until it stabilises, while Table 21 shows the volume of induced traffic which is implied by the assumptions on modal shift. 8 The analysis does not provide any justification for this assumption. 9 Note that REFER considers light goods vehicles for their passenger traffic component, not for freight, which is not included in the 1999 analysis.

25 March 2011 Frontier Economics, Atkins, ITS 25 Table 20. Build-up of new passengers from modal shift and % of steady state Option Year 1 Year 2 Year 3 Year 4 Year 5 MAX 130,662 (61%) 167,076 (78%) 186,354 (87%) 190,638 (89%) 214,200 (100%) MIN 117,912 (68%) 142,188 (82%) 154,326 (89%) 157,794 (91%) 173,400 (100%) Source: REFER, spreadsheet underpinning 1999 CBA calculations Table 21. Build-up of new passengers from induced traffic and % of steady state Option Year 1 Year 2 Year 3 Year 4 Year 5 MAX 125,538 (61%) 160,524 (78%) 179,046 (87%) 183,162 (89%) 205,800 (100%) MIN 113,288 (68%) 136,612 (82%) 148,274 (89%) 151,606 (91%) 166,600 (100%) Source: REFER, spreadsheet underpinning 1999 CBA calculations Overall, the assumptions on traffic forecasts are complete and they allow the calculation of the project s key benefits (time savings and vehicle operating costs) and positive externalities (for example, reduction in pollution and number of traffic accidents). However, their presentation in the CBA documents is opaque. With rare exceptions, the documents provide a clear justification for the assumptions made, both in terms of existing traffic and new passengers. With regards to freight, it is unclear whether assuming an additional two million tonnes as a result of Option MAX was appropriate. Given that rail freight has increased significantly since the early 1990s, part of this growth could have taken place also under a do-nothing scenario Quality of ex ante CBA Overall, the ex ante analysis we have reviewed (both the original 1997 analysis and the subsequent 2000) appears to be complete in its core parts. The methodology used takes into account a variety of different impacts, including several externalities such as noise, pollution and congestion.this is not common for the projects we have reviewed in this evaluation. Despite the completeness of the core analysis, we found the approach used lacking in two aspects, specifically:

26 26 Frontier Economics, Atkins, ITS March 2011 the reports accompanying the ex ante applications are often too concise and somewhat cryptic, making the reconstruction of the analysis very difficult; the analysis does not consider a variety of different options, particularly in its latest iteration, where only one option (Option MAX) is actually analysed. 3.4 Differences between ex post and ex ante analysis Table 22 compares the results of the ex ante economic analysis of Option MAX with the results of the ex post CBA for the section Pinhal Novo Faro. Table 22. Comparison of ex ante and ex post economic CBA (2008 prices) Ex ante Ex post Low case High case Net Present Value ( m) Economic IRR (%) 9.1% 6.7% 7.5% Benefit-cost ratio Source: REFER for ex ante analysis and own calculation for ex post analysis Overall, REFER estimated that the project would deliver value for money, with an NPV of about 156m and a benefit-cost ratio of about 1.6. While this result is outside the range of the ex post CBA, the two analyses are largely consistent. The ex post analysis has confirmed the economic benefits of the project exceed its costs. The main differences between the two analyses stem from the following areas: forecast and actual traffic volumes, expected and actual capital cost, approach to the benefit calculation, inclusion of additional impacts, and discount rate. We briefly address each in turn. Traffic volumes With regards to traffic volumes, the main difference between the ex ante and ex post analyses lies in the number of passengers in the first ten years of the project assessment period. Specifically, the ex ante analysis assumes that: the line would be re-opened to traffic already from 2000; and, at the time of re-opening, total traffic would already be at about 80% of the long run level and it would reach that level in about four years.

27 Passengers (000s) March 2011 Frontier Economics, Atkins, ITS 27 Looking at historic passenger data, we found that the line was instead re-opened in At re-opening, traffic was lower than estimated in the ex ante analysis. However we have noticed a faster growth rate, leading to traffic reaching a level very close to the ex ante steady state by Overall, while the traffic growth profile for the first years of operation has turned out to bee too optimistic, the long-term forecast is quite accurate. Figure 4 compares total passenger numbers in the ex ante and ex post analyses. Figure 4. Ex ante and ex post passenger numbers ( ) Ex ante assumption Ex post actual passengers Source: REFER Capital costs Capital costs are the main difference between the ex ante analysis and the ex post evaluation of the project. The nature of this difference relates to the time when the capital costs were incurred and the total amount of capital expenditures. In the ex ante analysis, REFER assumes that capital cost would be incurred in the first three years of the assessment period ( ). However, the ex post historic data shows that, while REFER started incurring costs in 1997, the bulk of the capital expenditure costs took place between 2001 and This has had the following implications: In nominal terms, REFER spent more than expected. In the ex ante analysis, it envisaged that capital expenditure would be 268m. Instead, the actual capital expenditure was 406m (about 51% higher). We could not identify the cause for this cost overrun.

28 EUR (millions) 28 Frontier Economics, Atkins, ITS March 2011 However, due to the different expenditure profile, the NPV of ex ante and ex post capital costs is not as different. Using a 5.5% discount rate, the NPV of capital expenditure in the ex ante analysis is 252m, while the NPV of the actual capital expenditure is 306m (21% higher). Ex post expenditure profile implies a higher terminal value for the project, as investment should be depreciated from the year in which it is incurred. Figure 5 shows the profiles of capital expenditure in the ex ante and ex post cases. Figure 5. Ex ante and ex post capital expenditure ( m, 2008 prices) 160 Ex post actual capital expenditure 140 Ex ante assumption Source: REFER Approach to benefit calculation With regards to benefit calculation, there are two main causes for the difference between the ex ante and ex post analyses. Different parameters. For the ex post CBA we have used updated values for all the parameters used to calculate the benefits of the project, ranging from the value of time to the reduction of vehicle operating costs and pollution thanks to modal shift. These updated values are those that

29 March 2011 Frontier Economics, Atkins, ITS 29 REFER currently uses. 10 In addition, REFER has now introduced a value to estimate the benefit from the reduction in road congestion, which was not available at the time of the ex ante analysis. The use of these updated values in the ex post analysis has had the effect of increasing the value of benefits, if compared with the ex ante analysis. Application of rule of half. We have used the rule of half for the calculation of the benefits from time savings (see Annexe 2). Based on the calculations for the ex ante analysis, REFER did not apply the rule of half. This has potentially led to an over-estimation of the project s benefits in the ex ante analysis. Accordingly, we have applied the rule of half in the ex post CBA. This has had the effect decreasing the value of benefits, if compared with the ex ante analysis. Inclusion of additional impacts The ex ante CBA does not include two elements that we believe should be included in analysis. Total revenues. REFER carried out the 1999 cost-benefit analysis (and its updated version) from the point of view of REFER itself. REFER is the network operator, while CP (and Fertagus) provide rail services. As REFER s revenues only amount to the network charges paid by the train operators, we believe that this analysis is not sufficiently comprehensive. For this reason, in the ex post CBA we have considered CP s revenues. Foregone government revenues from the fuel tax. The modal shift from road to rail implies a reduction in overall fuel consumption, which, in turn, implies lower government revenues from the fuel tax. This effect offsets the benefits of the project. REFER did not include this impact in the ex ante CBA. Discount rate Finally, we note that, in the ex ante analysis, REFER calculated the NPV of the project using a 5% discount rate, as suggested by the guidance available at the time. Currently, the EC Guide on Investment Appraisal recommends using a 5.5% discount rate. We have used this last value for the ex post evaluation. Table 23 compares the results of the ex post and ex ante economic CBA when holding constant the discount rate at 5%. The results of the ex post CBA are in this case closer to ex ante results, even though the later are still outside the range 10 We have extracted these updated values from an ex ante CBA that REFER has recently elaborated (March 2010) Variante da Trofa.