Going Green? Ex-post Valuation of a Multipurpose Water Infrastructure in Northern Italy

A contingent valuation approach is used to estimate how households value different multipurpose infrastructures (conventional or green) for managing flood risk and water pollution. As a case study we consider the Gorla Maggiore water park located in the Lombardy Region, in Northern Italy. The park is a neo-ecosystem including an infrastructure to treat waste water and store excess rain water, built in 2011 on the shore of the Olona River in an area previously used for poplar plantation. This park is the first one of this type built in Italy. A novel aspect of our research is that it not only considers the values people hold for different water ecosystem services (pollution removal, recreative use, wildlife support, flood risk reduction), but also their preferences for how those outcomes are achieved (through conventional or green infrastructures). The results indicate that the type of infrastructure delivering the ecosystem services does have an impact on individuals’ preferEmail addresses: arnaud.reynaud@inra.fr (Arnaud Reynaud), lanzanov@uni-bonn.de (Denis Lanzanova), Camino.Liquete@ec.europa.eu (Camino Liquete), Bruna.Grizzetti@ec.europa.eu (Bruna Grizzetti) Toulouse School of Economics, INRA, University of Toulouse Capitole, Toulouse, France Center for Development Research (ZEF), University of Bonn, Bonn, Germany European Commission, Joint Research Centre (JRC), Directorate D Sustainable Resources, Ispra, Italy Preprint submitted to Elsevier July 31, 2017 ences for freshwater ecosystem services. Households are willing to pay from 6.3 to 7.1 euros per year for a green infrastructure (compared to a conventional one), with a premium up to 16.5 euros for a surrounding made of a park. By considering the type of infrastructure within the choice model, we gain a richer understanding of the relationship between social welfare and freshwater ecosystem services.

The green infrastructure consists in a set of constructed wetlands, sur-95 rounded by a park (Fig 1). All together the constructed wetlands and the 96 surrounding park form the Gorla water park. This neo-ecosystem was built 97 on the shore of the Olona River in an area previously used for poplar planta-  Our CV approach is not standard in two aspects. First, in our work, 154 respondents will be asked to answer sequentially four CV questions. In each 155 case, they will have to compare the ecosystem services provided in a refer-156 ence scenario (the situation which used to prevail before the construction of 157 the water park) with those derived from an alternative infrastructure which 158 was feasible when the park was built. Second, each infrastructure will be 159 described by a set of attributes. This allows us to examine the tradeoffs 160 that people are willing to make between ecosystem services provided by the 161 different infrastructures. But rather than varying the attribute levels across 162 infrastructures according to a specific design (as it is usual done when using 163 discrete choice experiment), in our case the combination of attributes for a 164 given infrastructure is chosen to represent a feasible infrastructure that was 165 really considered by policy-makers at the time at which the water park has 166 been built. We have implemented an attribute-based CV approach, in the

Alternative projects' choice
Imagine that the Gorla Maggiore Park is not built and in the site you still find the previous situation: a private poplar plantation. With respect to this reference situation you are asked to choose the best project to prevent the sewage from Gorla Maggiore to pollute the Olona River. To do so, you should value each one of the 4 projects proposed against the poplar plantation.

Reference situation: Poplar plantation
The poplar plantation is a private parcel of land where poplars are grown for the production of wood. This ecosystem produce timber but does not provide a lot of ecological services.

Alternative 2: Green infrastructure with poplar plantation
The infrastructure is a set of constructed wetlands with a wet retention pond and a private poplar plantation (not accessible) around it.

Alternative 1: Green infrastructure with park
This is the present situation. This infrastructure consists of a set of constructed wetlands with a wet retention pond, both surrounded by a recreational park with trees.

Alternative 3: Conventional infrastructure with park
The infrastructure consists of a flush tank (buried and covered by grass) and a dry retention pond, both surrounded by a recreational park with trees.

Alternative 4: Conventional infrastructure with poplar plantation
The infrastructure consists of a flush tank (buried and covered by grass), a dry retention pond and a private poplar plantation (not accessible) around it.
Low pollution control: during heavy rains part of the Gorla sewage is not treated and the runoff takes a large amount of pollutants to the river.
Low recreational level: people cannot access the private area.
Low biodiversity: only one plant species is allowed, usually fostered with chemical products. Low flood control: the area does not provide a retention area for an excessive river flow. Here below we present the 4 alternative projects to the reference situation and we ask for your personal valuation. For the following questions (no.6-9), it is very important that you reflect your real intention. Imagine what the proposed reallocation of public budget means in terms of reduction of public good and services for your household (less money for public schools for example) and what types of benefits you will get from each project. High pollution control: macrophytes (wetland plants) neutralize pollutants from the sewage before the water is discharged in the Olona River.
High recreation level: the pond provides nice aesthetic views and fishing opportunities; the area has been restored with riparian trees and is suitable for outdoor activities.
High biodiversity: the area increases habitat availability for birds, dragonflies, amphibians including several endangered species. High flood control: the pond and surrounding area can store water and contribute to decrease the flood impact downstream.
High pollution control: macrophytes (wetland plants) neutralize pollutants from the sewage before the water is discharged in the Olona River. Medium recreational level: the pond provides nice aesthetic views and fishing opportunities; but the poplar area is not accessible.
High biodiversity: the area increases habitat availability for birds, dragonflies, amphibians including several endangered species. High flood control: the pond and surrounding area can store water and contribute to decrease the flood impact downstream.
High pollution control: the flush tank stores temporally the sewage running from Gorla during storms. Then this water must be pumped to the wastewater treatment plant of Olgiate Olona. Medium recreational level: the surrounding area has been restored with riparian trees and is suitable for outdoor activities. Low biodiversity: the variety of the living organisms is relatively limited High flood control: the flush tank and retention pond can store water and contribute to decrease the flood impact downstream.
High pollution control: the flush tank stores temporally the sewage running from Gorla during storms. Then this water must be pumped to the wastewater treatment plant of Olgiate Olona.
Low recreational level: the infrastructure covered with grass can be muddy after flooding; the poplar area is not accessible.
Low biodiversity: the variety of the living organisms is relatively limited. High flood control: the flush tank and retention pond can store water and contribute to decrease the flood impact downstream.
To get your opinion about the value of these projects, imagine their construction requires a reallocation of public money that would not be used to finance other public services.

6.
What is the maximum amount of your local taxes that you would be willing to allocate to construct this project instead of keeping the poplar plantation? (in euros per household per year for the next 20 years)

7.
What is the maximum amount of your local taxes that you would be willing to allocate to construct this project instead of keeping the poplar plantation? (in euros per household per year for the next 20 years)

8.
What is the maximum amount of your local taxes that you would be willing to allocate to construct this project instead of keeping the poplar plantation? (in euros per household per year for the next 20 years)

9.
What is the maximum amount of your local taxes that you would be willing to allocate to construct this project instead of keeping the poplar plantation? (in euros per household per year for the next 20 years)

11.
What is your level of concern about the environment? Reference scenario. We first describe a reference situation in which the whole Maggiore municipality. Respondents have been asked to evaluate these sce-238 narios in comparison to the reference scenario (private poplar plantation). 239 We have used the following script:  To make people more clearly understand the meaning of these four scenarios,

283
Hypothetical bias of the CV scenarios. Hypothetical bias and consequential-284 ity are a concern for any CV study. It may be an issue in our case since  Although we recognize that this payment vehicle is not fully satisfactory 341 from an incentive point of view, it is the second-best option in our setting.

342
The script used for explicating the payment vehicle is presented in Figure 2.     Now we move to the answers given by the respondents to the four con-418 tingent valuation scenarios P1, P2, P3 and P4 described above.
419 Table 3 gives some statistics on the maximum amount of money each 420 respondent is ready to allocate to each contingent valuation scenario (in euro 421 per year and per household for the following twenty years). We interpret this amount of money as a WTP for the corresponding scenario.

423
In a contingent valuation analysis, it is important to make the distinction 424 between the "true zero bids" corresponding to respondents having indicated 425 that they are not willing to pay anything because they are truly averse or in-

437
To identify protest answers, respondents having reported zero WTP for 438 the four proposed scenarios have been asked if they agree or disagree with 439 the six following reasons: "1-I am not confident that the money will be used 440 efficiently by the municipality", "2-I am against any tax expenses", "3-I 441 prefer the money to be spent on more important things", "4-I cannot afford 442 to pay any tax", "5-I believe that the park should not be paid by me but 443 directly by a central administration" and "6-I will never go to the park". All the reason 1-, 2-or 5-, can be classified as "false zeros". In Table 3 we then 448 report some statistics on WTP per scenario first based on the full sample 449 and second on a subsample excluding "false zeros".
450 Table 3 calls for a few comments. First, whatever the sample considered where W T P * i denotes the true WTP for respondent i, X i a vector of explana-508 tory variables and i a random component following a normal distribution 509 with mean zero and standard deviation σ.

510
A standard procedure to estimate Equation (1) is to assume that the 511 true WTP is the midpoint between the highest amount to which the re-512 spondent said "yes" and the lowest amount to which she said "no" Cameron and t ui . In the right-censoring case, W T P * is greater than t li whereas the 539 left-censoring case correspond to a W T P * lower than t ui . The conditional 540 probability of observing each case for respondent i writes: where Φ is the cimulative standard normal density function. The corre-542 sponding log-likelihood function is made of three parts, which correspond to 543 interval-censoring, left-censoring and right-censoring observations.

544
Since each respondent is asking to answer several CV questions, our ap- package.

567
We present in Table 5  is a recreational park (the reference category is a private poplar plantation).

577
Since the previous analysis has suggested that there might be a premium 578 for the scenario combining the green infrastructure and a recreational park, 579 a third dummy variable has been added to account for this situation. As  old. Household income is introduced in logarithm and we also control for the 584 frequency of visits to the park.  The most interesting finding is given by the positive and highly significant 594 coefficient for the interaction between the green infrastructure and the park.

595
There is a specific premium for a project combining a green infrastructure 596 together with a recreational park. This premium is quite significant in terms 597 of amount of money since it varies from 14.7 to 16.5 euros per household 598 and per year, depending upon the model considered. Our results suggest 599 that people in Gorla Maggiore do not put any specific value on a park if it 600 associated with a conventional infrastructure. On contrary the park will be 601 highly valued if is associated with the green infrastructure. One possible in-602 terpretation of this result is that the park and the green infrastructure may 603 be perceived as two highly complementary goods. Another explanation is 604 the "endowment effect" we have discussed previously.

605
As expected from the descriptive statistics, WTP is significantly impacted 606 by respondent's income and respondent's frequency of visits to the Gorla 607 Maggiore park. The higher is the household income, the higher will be the 608 WTP. In addition, respondents reporting that they went to the park at least 609 20 times during the last 12 months have an additional WTP which varies 610 between 6.7 and 7.8 euros per household and per year.

611
Since the four alternative infrastructures are directly related to the level of 612 ecosystem services they provide (attributes "low", "medium" and "high"), 613 our estimates may directly be interpreted in terms of WTP per attribute.   conducting our CBA, we will then report the discounted net benefits for each 641 scenario for three different interest rates (2%, 3% and 4%).  these costs in Table 6, we make the distinction between infrastructure and 647 landscaping expenses since they differ across the proposed infrastructures.

648
The measure of the benefits is less straightforward, first because our WTP Results presented in Table 6 call for a few comments. First, the defini-  We argue that WTP surveys may be useful for regional planning Van-