1 The role of ambiguity in the evaluation of the net benefits of the MOSE system in the Venice lagoon Fulvio Fontini Department of Economics, University of Padua (Italy) Georg Umgiesser ISMAR-CNR, Venice (Italy) Lucia Vergano ECCET, IPTS,JRC, European Commission, Sevilla (Spain)

Structure of the paper 2 Introduction to decision making under ambiguity The Neo-additive capacities framework: introduction and graphical representation 3. The Venice lagoon case study: analysis of the problem and policy implications

Decision making under ambiguity I 3 Policy decisions impacting on the environment → Ambiguity Ecological system Scenarios whose likelihood + → can not be inferred from any Economic system probability distribution No clear definition of the problem → Incomplete states of the world (Mukerji, 1997) Use of any a priori distribution not justified (Chichilnisky, 2000)

Decision making under ambiguity II 4 Decision criteria: Maximin decision criterion (Wald, 1950) - complete ignorance (Rawls, 1971): evaluation of the worst possible consequence - Arrow-Hurwicz criterion (1971): evaluation of a linear combination of the best and worst possible consequences → strong interpretation of the Precautionary Principle (PP): only the worst possible consequence(s) is (are) to be taken into account unless it can be proved “beyond any reasonable doubt” that some other consequences will occur

Capacity as a measure of ambiguity I 5 Capacity = normalized monotone measure of ambiguity ν: E → ℜ → generalisation of the subjective utility theory as a decision criterion through the Choquet Integral where f: S → X ⊂ ℜ is a utility function → the integral w.r.t. a capacity identifies the so called Choquet Expected Utility (CEU)

Capacity as a measure of ambiguity II 6 Wakker, 2001 Concave capacity → Overweight good outcomes → Optimistic attitude towards ambiguity Convex capacity → Overweight bad outcomes → Pessimistic attitude towards ambiguity

NEO-capacity as a measure of ambiguity I 7 Chateauneuf et al., 2007: NEO-additive capacity = a specific type of capacity, additive on non-extreme outcomes, reflecting pessimism for some events and optimism for some other events where: π = finitely additive probability distribution μαN (A) = Hurwicz capacity

NEO-capacity as a measure of ambiguity II 8 8 → The CEU calculated w.r.t. a NEO-additive capacity: where: Eπ = expected value calculated w.r.t. π f = act function

Special cases of the CEU 9 Several decision criteria may be interpreted as special cases of the Choquet Expected Utility framework: 1. δ=0: the Expected Value approach 2. N={Ø}, δ>0, α=0: pure pessimism N={Ø}, δ=1, α=0: Maxmin criterion 3. N={Ø}, δ>0, α=1: pure optimism N={Ø}, δ=1, α=1: Maxmax criterion 4. N={Ø}, δ=1, α=(0,1): Hurwicz criterion The parameters δ(1-α) and δα capture the impact of pessimism and optimism

Generalization of the functional form 10 Renaming 1. λ=δ(1- α) 2. γ=δα 3. 4. the CEU function can be expressed in the following way:

Graphical representation I 11 the simplex constraints the set of the admissible ranges for γ and λ. The space of the CEU graphically corresponds to the side of the triangle shown in Fig. 1

Graphical representation II 12 Figure 1

Interpretation of the graphical representation I 13 1. γ=λ=0: the Expected Value approach 2. γ=0: pure pessimism 3. λ=0: pure optimism 4. λ+γ=1: Hurwicz criterion

Interpretation of the graphical representation II 14 Expected Value Equivalent set: (γ,λ) s.t. CEU=Eπ and the Hurwicz criterion holds → it defines the implicit values of the ambiguity attitude that a decision maker has in mind when taking the decision on the basis of the expected value only, i.e. if having an optimistic or pessimistic attitude towards it

The ‘acqua alta’ phenomenon in the Venice lagoon 15 ‘Acqua alta’ = the periodical high water event causing partial flooding of Venice, corresponding to +80 cm above the ‘Punta della Salute’ tidal datum During last decades: increased frequency and intensity (Fig. 2) Forecasts for the next century (IPCC, 2007: Fig. 3): further worsening of the phenomenon, due to the global sea level rise induced by climate change → Mitigation and prevention measures: hydraulic pumps, ‘vasche’, ‘paratie’, rising of pavements, MOSE

The ‘acqua alta’ phenomenon I 16 Figure 2

The ‘acqua alta’ phenomenon II 17 Figure 3

The ‘acqua alta’ phenomenon: the frequency 18 Figure 4 – Yearly distribution of tidal  +110 cm in Venice, 1872-2006

The IPCC forecasts 19 Figure 5 – Global sea level rise Sources: IPCC, 2007

The MOSE functioning 20 MOSE = system of mobile barriers installed on the sea floor of the inlets (‘Chioggia’, ‘Lido’ and ‘Malamocco’) → Separate from a hydraulic point of view the lagoon from the Adriatic Sea whenever the forecasted water level exceeds the safeguarding level (+100 or +110 cm above ‘Punta della Salute’)

The MOSE impact: benefits I 21 avoided damages to buildings: damages affecting the real estate stock, depending on the intensity of flooding; 2) avoided damages to individuals: damages to the flow of (touristic and personal) services depending on the frequency of flooding MOSE → flooding intensity and frequency↓ → costs saved = benefits

The MOSE impact: benefits II 21 avoided damages for buildings: costs of renovation after the highest tide experienced in 2000-2002 - plastering costs of the walls’ flooded surface - lead plate introduction costs (historical buildings)

The MOSE impact: benefits I 21 2) avoided damages for individuals: costs due to displacement problems - costs of children (27% of children in schooling age) and elderly people (10% of people aged 75-84) caring - loss of touristic expenses (daily tourist flow x average daily expenses)

The MOSE impact: benefits II 22 Table 1 – Avoided costs and expenses for category Buildings Plastering: internal walls 2.77 €/1m 1cm high external walls 5.44 €/1m 1cm high Artistic buildings Lead plate: internal walls 20.11 €/m 61.87 €/m Aged people Caring 12.37 €/hour Children in age of schooling 8.25 €/hour Tourists Tourist expense 85.96 €/day

The MOSE impact: costs I 23 → Operational and maintenance costs (11,136 thousands of €/year) → Direct costs due to the interferences with port activities (Tables 2-3): additional time to getting in and out of the lagoon (charter costs) longer period staying in a wharf/quay (mooring costs) Safeguarding level → Frequency of mobile barriers closure → Amount of the additional charter and mooring costs → The choice between a + 100 or a + 110 cm safeguarding level = example of decision under ambiguity: the likelihood of the environmental parameter (sea level rise) can not be inferred on the basis of any probability distribution

The MOSE impact: costs II 24 Table 2 – Charter and mooring costs for ship category Charter costs Mooring costs Tons €/m/Hour €/Hour Crude oil tanker x 0.03 1,333.33 Other oils tanker x>1,000 0.13 1,000>=x<1,500 0.12 x>=1,500 0.11 LNG tanker x<7,000 0.29 7,000<=x<20,000 0.19 x>=20,000 0.14 Container x<15,000 1,000.00 x>=15,000 0.02 Cargo x>=4,500 0.01 Carrier x<4,500

The MOSE impact: costs III 25 Table 3 – Charter and mooring costs for ship category Charter costs Mooring costs €/Pass/Hour €/m/Hour Passenger ship 1.60 0.17 Yacht 46.77 0.45

MOSE direct impacts economic assessment: the hydrodynamic model 26 Hydrodynamic model (Umgiesser et al., 2004): - using water level measurements at the inlets and wind over the lagoon → simulates water levels and barotropic currents inside the lagoon → computes how often the water level exceeds the safeguarding level and the time of mobile barriers closure - using ship traffic data for 2000-2002 (Venice harbour office) → computes how often and for how long the ship traffic is interrupted due to MOSE functioning (Umgiesser and Matticchio, 2006)

Direct impacts economic assessment: scenarios 27 Hydrodynamic model hypotheses: - Adriatic Sea level rise (0, +30 cm, +50 cm) - safeguarding level (+100 cm, +110 cm) - security increment (0, +10 cm) Table 4 – The twelve scenarios +100 cm +110 cm +30 cm +50 cm A E I B F J +10 cm C G K D H L

Direct impacts economic assessment: the estimates 28 Table 5 – Total costs and benefits (millions of €/year) for each scenarios Scenarios Total costs Total benefits Net benefits (A) 00_100_00 31.838 152.533 120.695 (B) 00_110_00 31.934 25.444 -6.490 (C) 00_100_10 31.669 209.828 178.159 (D) 00_110_10 31.862 120.671 (E) 30_100_00 33.588 268.182 234.593 (F) 30_110_00 32.080 230.794 198.714 (G) 30_100_10 35.485 232.696 (H) 30_110_10 33.024 231.238 198.214 (I) 50_100_00 41.383 369.152 327.770 (J) 50_110_00 36.277 306.543 270.265 (K) 50_100_10 45.084 418.373 373.290 (L) 50_110_10 39.522 369.041 329.520

Policy implications: decision variables 29 Environmental variable: Adriatic Sea level rise Decision making variables: safeguarding level security increment Safeguarding level → Frequency of mobile barriers closure → MOSE net benefits The choice between a safeguarding level of + 100 or + 110 cm = example of decision under ambiguity: the likelihood of the environmental parameter (sea level rise) can not be inferred on the basis of any probability distribution → choice based on the estimated costs associated to each scenario, weighted according to the decision criterion adopted by policy makers

Policy implications: decision criteria I 30 No ambiguity → Expected value criterion (benchmark): states of the world are equally weighted according to a subjective probability measure pi =1/12, i=A, …, L ci, pi i=A; …,L Ambiguity → application of the CEU framework under different values of the subjective parameters

Policy implications: decision criteria II 31 1. Max-Min criterion (pessimistic attitude – C1) The option corresponding to the less bad outcome (B) among the worst outcomes associated with each possible decision (B,D,A,C) is taken 2. Max-Max criterion (optimistic attitude – C2) The option corresponding to the best outcome (K) among the best outcomes associated with each possible decision (K,L,I,J) is taken 3. Hurwicz criterion The best (B) and the worst (K) outcomes are equally weighted

Implicit ambiguity attitude: λhut = 0.81: γhut = 0.19 Estimated net benefits: a comparison between different decision criteria 32 Table 6 – Estimated net benefits (millions of €/year) Decision criterion Net benefits EU 214.841 Max-Min 178.159 Max-Max 373.290 CEUp λ (-36.683) + 214.841 CEUo γ (158.448) + 214.841 CEUH γ (195.131) + 178.159 Implicit ambiguity attitude: λhut = 0.81: γhut = 0.19

Conclusions 33 Cost estimates vary substantially according to the attitude decision makers have towards ambiguity (pessimism vs optimism) 2. Even assuming a symmetric attitude of decision makers towards ambiguity, estimated costs substantially differ from the ones calculated following the expected value criterion 3. Decision makers have implicitly shown pessimism, i.e. overevaluation of the scenarios providing the lowest benefits and underevaluation of those inducing the highest ones (precautionary approach)