Andrea Debernardi 1 , Raffaele Grimaldi 2 , Paolo Beria 2 1 Polinomia Srl, Milan (Italy)

1. Cost benefit analysis to assess modular investment: the case of the New Turin-Lyon Railway Andrea Debernardi1, Raffaele Grimaldi2, Paolo Beria2 1 Polinomia Srl, Milan (Italy) 2 Department of Architecture and Planning, Politecnico di Milano (Italy)

2. CONTENTS: Introduction and Objectives A theoretical model The case of the New Turin-Lyon Railway and the “FARE” proposal Final considerations

3. Introduction and Objectives Assessment of infrastructure investments: risk of demand levels below expectations (on average -51.4% for railways; Flyvbjerg et al., 2003)  problem of overinvestment The issue is well studied in the literature: Literature on inaccuracy in trafficforecasting(Flyvbjerget al., 2003;Flyvbjerg, 2008) Transportliterature on investment timing(e.g. Szymanski, 1991 and Chu & Polzin, 1998 and 2000) Economicliterature on investment under uncertainty (e.g. McDonald & Siegel, 1986 and Dixit & Pindyck, 1994)

4. Introduction and Objectives Literature proposes tools to accurately manage the problem: Reference Class Forecasting (Flyvbjerg, 2008) Actual performance in a reference class of comparable projects already carried out Option Value assessment (Dixit & Pindyck, 1994) There is a value in waiting to invest when it allows to adopt a better decision on the basis of more information Unfortunately they are quite complex to develop in practice and they often require a large amount of data seldom really available.

5. Introduction and Objectives Our idea is to consider the project as split into smaller functional sections and bind the construction of a further section to the compliance of a pre-determined “switching rule”. A similar approach has been already used in practice, for example in the case of the Swiss Lötschberg base tunnel: the financing of the second tube was bound to the reaching of a certain demand level on the first tube.(Schreyer, Sutter & Maibach, 2009) We use CBA to assess this “split” strategy and compare it with traditional timing. This could help decision makers in understanding when postponing some decisions to the following running phase gives better value.

6. CONTENTS: Introduction and Objectives A theoretical model The case of the New Turin-Lyon Railway and the “FARE” proposal Final considerations

7. A theoretical model We study a theoretical model in order to better understand the issue θ contruction time α traffic yearly growth rate q0 first year traffic T analysis horizon

8. A theoretical model: “build together now” option

9. A theoretical model: “build together now” option We write the of this approach: Investment costs Maintenancecosts (net) benefits

10. A theoretical model: “build separate” option

11. A theoretical model • Moreover introduce a “switching rule”: • we build now the first tube; • we check actual traffic on it; • we decide if and whether to build the second tube on the basis of actual traffic. This will cost 2 ∙ I in absolute terms, which is higher than the 2 ∙ s ∙ I cost of the build together strategy. We write the of the two approaches.

12. A theoretical model Because of the r social discount rate we obtain that: if (if and r > 0) Quite obviously, a “switching rule” performs better when the actual value of the postponed second phase cost is lower than the extra-cost associated to phasing.

13. A theoretical model However, the most important contribution of planning an infrastructure in sequential functional sections on the basis of a “switching rule” is related to uncertainty. If traffic is below expectations, this approach allows to shift the decision of building the second section of the infrastructure when it is really needed, allowing to save costs in actualised terms.

14. CONTENTS: Introduction and Objectives A theoretical model The case of the New Turin-Lyon Railway and the “FARE” proposal Final considerations

15. The case: New Lyon-Turin Line (NLTL) A new mixed use HSL between Italy and France. STUDYAREA (own elaboration)

16. The case: NLTL traffic forecast Official traffic forecasts Since 1997,traffic is decreasing on the railway and steady on the road. Actual rail traffic and official forecasts (own elaboration)

17. The case: NLTL phasing On the French side, the NLTL will be built in five stages (both serial and parallel), planned from 2012 to 2035 and over. On the contrary, in the Italian side only one functional section is planned, to be completed in 2023.

18. The case: “FARE” proposal FARE is an alternative strategy, proposed by local authorities. It follows the clear evidence that sections will be saturated progressively, from Turin to the cross-border section. Therefore, four functional stages have been proposed. The four bulding stages of the “FARE” proposal(own elaboration)

19. The case: “FARE” proposal and “switching rule” Moreover, FARE proposal involve a “switching rule” strategy with respect to capacity constraints: Each section is assumed to be built when and only when traffic forecasts for previous section will be reached. In our study, this kind of rule is applied to the five French side sections, too.

20. The case: New Turin-Lyon Railway & “FARE” proposal This way, if the traffic forecasts are correct, the new infrastructure will be built as a whole, in steps, without introducing any capacity constraint, but solving them as soon as they appear. This way, if forecasts are optimistic, only the most effective parts of the scheme will be built. We show the preliminary draft results of an independent simplified analysis, with the purpose of showing the benefits of phasing infrastructures upon a “switching rule”, in a case of high uncertainties on the future demand.

21. The case: New Turin-Lyon Railway & “FARE” proposal 1. Our analysis suggests that, even if official traffic forecasting would actually occur, the FARE strategy would provide a better (less negative) socio-economical performance for nearly 1.4 billion Euro. Detailed calculations will be available on a Working Paper.

22. The case: New Turin-Lyon Railway & “FARE” proposal • 2. • Sensitivity analysis with respect to: • traffic in 25th year; • time shifting of the forecasted first year traffic. The flexible FARE strategy performs better than the rigid NLTL, as far as traffic is +15÷55% higher than forecasts.

23. CONTENTS: Introduction and Objectives A theoretical model The case of the New Turin-Lyon Railway and the “FARE” proposal Final considerations

24. Final considerations Our approach tried to solve a common problem faced by analysts making a CBA under uncertain conditions. We think that, in some conditions, introducing in the decision-making process some phasing upon “switching rules” on demand levels might be a good way to manage “on the road” the uncertainty of forecasts. Ideally, it is like reducing year by year the variance (the uncertainty) of demand forecasts by observing the actual demand and taking decisions on new expansions consequently. CBA in the usual form reveals to be perfectly adapt to manage such rules in a reasonably simple way.

25. Thank you for your attention! andrea.debernardi@polinomia.it raffaele.grimaldi@polimi.it paolo.beria@polimi.it