Perceived health status and environmental quality in the assessment of external cost of waste disposal facilities. An empirical investigation

Waste Management & Research 30(8) 864 –870

© The Author(s) 2012 Reprints and permission:

sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0734242X12445654 wmr.sagepub.com

Introduction

Taxation for urban waste management has been reformed in Italy by the introduction of the environmental law in 2006. It updated the Ronchi Act that, in 1997, implemented the EC directives 91/156, 91/689 and 94/62. A goal of allocative efficiency inspires the new tariff system based on a polluter-pays principle. However, most of the externalities generated in the life cycle of waste man-agement facilities remain unaccounted for, with a consequent distortion for prices (Basili et al., 2006). Accordingly either the siting decisions of disposal infrastructures or the programming of new plant should be supported by feasibility studies to assess the socio-economic convenience of projects.

Despite some monetary values of externalities of infrastruc-tures have been made available (as the Externe project, 1998), cases of modern typologies of facilities, such as new incinerators or compost plants, still remain uncovered or poorly represented. A stated choice approach is suggested to obtain monetary estimates of perceived external costs of waste disposal facilities. Preferences of citizens living in urban areas potentially hosting new waste dis-posal facilities are here analysed, in order to monetarily assess disamenity effects consistently with random utility approach. The first section of the paper describes the study background; the

second reports the experimental design that is based on an exer-cise of choice among residential locations, while the third dis-cusses results and the fourth concludes.

The study background

The economic evaluation of human health, landscape and other environmental externalities has been treated in a number of empirical studies (for a comprehensive review see Eshet et al. 2005a, 2005b, 2006). Beyond the more extended group of hedonic pricing evaluations, the present study focused on the recent and restrained stream of literature based on choice experi-ment (CE) surveys (Caplan et al., 2007; Garrod and Willis, 1998; Sasao, 2004). Although some monetary estimates can be adopted as a reference, and are suitable for benefit transfer practices, it

Perceived health status and environmental

quality in the assessment of external costs

of waste disposal facilities. An empirical

investigation

Sergio Giaccaria and Vito Frontuto

Abstract

Taxation for urban waste management has been reformed in Italy by the introduction of an environmental law in 2006. In the planning phase of waste management, externalities generated by new facilities remain widely unaccounted, with a consequent distortion for prices, often raising local conflicts. The paper presents a survey based on the choice modelling methodology, aimed to evaluate on a monetary scale the disamenity effect perceived by incinerator and landfills in an Italian urban context: the city of Turin. In a random utility framework the behaviour of respondents, whose choices are found to be driven by the endowment of information about technological options, socio-economic characteristics as income, education, family composition, and also by their health status was modelled. Furthermore, empirical evidence that the behaviour in residential location choices is affected by different aspects of the respondent life and in particular by the health status was found. Distinct estimates of willingness to accept compensation for disamenity effects of incinerator (€2670) and landfill (€3816) are elicited. The effect of health status of the respondents, their level of information about the waste disposal infrastructure, the presence of a subjective strong aversion (NIMBY) and the actual endowment and concentration of infrastructures are demonstrated to be significant factors determining the choice behaviour, but differentiated and specific for incinerators and landfills.

Keywords

Beliefs, choice modelling, externalities, landfill, incinerator, nimby effect, health

Department of Economics, University of Turin, Turin, Italy

Corresponding author:

Vito Frontuto, Department of Economics, University of Turin, Via Po 53, I-10124 Turin, Italy

Email: vito.frontuto@unito.it

appears that a relevant component of the preference drivers is associated with the sphere of attitudes and perceptions that should be intended as site-specific and not stable over time (Kiel and McClain, 1995). Other stated preference analyses are Basili et al. (2006) and Baccheschi et al. (2008), but are based on contingent valuation formats for the elicitation of willingness to pay (WTP) indicators. The case study presented focuses on a peculiar mix of context conditions and evaluates, on a monetary scale, the dis-amenity generated by incinerators and landfills in a specific urban context. The results suggest that CE surveys may be a source of strategic information in a bottom-up sense to inform future siting policies of waste disposal facilities, improving the component of public involvement in the decision process. The asymmetry between losers and winners (community members hosting and non-hosting facilities) has been properly considered by Caplan et al. (2007), referring to an ex-post compensation test. Herein the focus is instead on the set of preference of potential

hosting community members.

Choosing hypothetical sites where to live in, they implicitly state their level of acceptance of hosting new facilities. The CE methodology allows the inference of welfare changes from the status quo to a chosen scenario, and to estimate the relative monetary values. In the present experiment, respondents are asked to face an explicit trade-off between alternative locations characterized by the existence of various disamenity sources, in the presence of monetary and non-monetary compensative measures. According to the present assumptions, willingness to accept (WTA) seems to better represent this purely hypothetical scenario in a context of scarce resources. The differences between WTP and WTA have been deeply investigated and documented (Horowitz and McConnel, 2002), supporting the use of the former. However, in the case of siting, according to Horowitz and McConnel (2002), a low ratio WTP/WTA is expected, as for private goods. The respondents believe they own a ‘property right’ to preserve and to control, suggesting the siting decision is more like a private purchase-of-goods decision.

Our application has been conducted within the metropolitan area of Turin (north Italy), where the siting process of a new big incinerator and the closure of an old landfill has been the core of a local debate (Bobbio, 2002). In this context, some attempts to smooth conflicts arising from the not-in-my-backyard (NIMBY) syndrome have been based on a project aimed to build consensus through an experimental participative process (Norese, 2006; Tipaldo, 2006).

Sampling and experiment design

The target population is composed by potential hosts of a waste disposal facility. All the citizens living in a 2 km circle from the sites selected by local authorities as possible locations for munic-ipal solid waste facilities, in particular suburban areas to the north-west of Turin were identified (see Figure 1). A random drawn sample stratified by age was obtained (Table 1).

Respondents were contacted by mail and the questionnaire, including an introductory section motivating the present research and providing households with information on the project of facility construction: quantity of municipal solid waste poten-tially burned, energy and materials flows, emissions and residu-als, were sent out. After sending the informative material and the questionnaire by mail, the household were afterwards contacted by phone, and the questionnaire interview conducted during the phone call. This hybrid procedure between mail and phone reduced the weaknesses of the mail survey, introducing a direct interaction with trained interviewers. When contacting every household by phone, in the initial part of the interview the respondent was asked to make a choice, pointing out that the results of the survey would influence the process of facilities’ localization in order to enhance response accuracy by emphasiz-ing the responsibility of the respondents.

The design of the experiment [the profiles proposed during the interviews are randomized and the combinations of the differ-ent attributes vary, determining a pseudo-random set of possible comparisons among alternatives] was driven by the objective of specifying a utility function including determinants of the envi-ronmental quality and health risks perceptions. The respondents were asked to select the residential alternative they would be willing to move in, considering the house characteristics as Figure 1. Sample distribution in Turin’s blocks.

Table 1. Sample strata.

Age cohort % of the total

population of Turin % of the reference population

25–34 14 17 35–44 17 21 45–54 13 17 55–64 13 16 65–74 13 16 More than 75 11 14

constants. The third option was a no-choice option, a sort of sta-tus quo alternative, characterized by the absence of waste facili-ties (an example of the choice task is presented in Table 2).

The interviews ended with a section devoted to the collection of information on the households’ characteristics, the perception of the environmental quality and the health risks associated with the two infrastructures for the municipal waste management. The group of option profiles in every choice task was obtained through the software Choice-Based Conjoint (CBC), by

Sawtooth®_{. The random design obtained by the CBC tool is}

bal-anced and fulfils the orthogonality condition. Choice tasks with dominant/dominated alternatives were then removed from the constructed design. Despite its popularity, the methodology employed in the design of the experiment could be further improved. Advanced criteria to assess the statistical efficiency have been increasingly investigated, and C-optimality (Street and Burgess, 2007) or D-optimality (Ferrini and Scarpa, 2007; Kanninen, 2002) could be adopted for future refinements and applications of the CE methodology within this field.

Model and results

The aim of the analysis is to define a functional form for the indi-rect utility function modelling a formalized decisional rule as close as possible to that used by the respondents in their decision process (see Appendix 1 for the econometric specification). The set of variables representing the risks’ perception, the level of information on the localization process of an incinerator in Turin, the presence of strong aversion to the infrastructure (NIMBY syndrome) are inserted and the analysis is completed using some individual characteristics. Following the popular conditional logit structure (as further described in Appendix 1), the probabil-ity of choice of a site is function of: (i) the attributes of the sites, (ii) other household’s characteristics, (iii) an error component. All the variables but the monetary compensation is introduced in the form of interaction with alternative (infrastructure) specific constants. This partitioned structure of the deterministic compo-nent makes it possible to test whether the magnitude of the utility parameters tends to be shared across infrastructure typologies, or if it tends to vary between sites with incinerators or landfills. The model takes the form:

V incinerdist dist

dist inciner landfill dist landfill incin = + + β β β eer odour inciner landfill odour landfill inciner a odour + odour + β β iir inciner landfill air landfill inciner green in

air +β air +β green cciner landfill green landfill comp inciner iend green comp i +β +β +β eend iend health inciner landfill iend landfill inciner health +β +β iinciner landfill health landfill inciner income health incom + + β β ee income hrisk inciner landfill income landfill inciner hrisk + + β β iinciner landfill hrisk landfill inciner nimby inc hrisk nimby +β +β iiner landfill nimby landfill inciner knowledge nimby knowled +β +β gge knowledge inciner landfill knowledge landfill +β (1) The subset of attributes of the choice tasks options (comp,

dist, air, odour, green) are described in detail in Table 3, and the

respondent-specific characteristics (iend, health, income, hrisk,

nimby and knowledge) are reported in Table 4.

Each of the 168 respondents faced two choice exercises, for a total of 336 observations. The estimates, obtained by NLOGIT, are presented in Table 5.

Most of the expectations on the behaviour of the covariates are confirmed. In particular the variables indicating beliefs, atti-tudes and perception with respect to the incinerator show a high statistical significance. The main damage component is obtained through the variables indicating the proximity to the facilities (distinciner and distlandfill ). In the utility function these variables

measure the transition from a state in which the infrastructure is absent or very far from the house (from 2000 to 10 000 m), to a condition in which the infrastructure is very close (no more than 600 m). These variables are highly statistically significant and both the coefficients present a negative sign. Sites closer to the facility imply higher compensation. The marginal willingness to accept for the presence of the two infrastructures, keeping con-stant the other variables, is €1901 for the incinerator and €1710 for the landfill. The monetary compensation coefficient, consist-ently to the expectations, has a positive sign and its estimate results were statistically significant at the 0.01 level.

The attribute relative to the presence of odours differs explic-itly between incinerator and landfill. While the number of days in which there are odours does not have a significant effect for Table 2. Choice experiment design.

Which alternative you would choose? AÐ BÐ CÐ

Waste disposal infrastructure Incinerator Landfill None of the previous Distance from the infrastructure 600 m 2000 m

Air pollution level (days per year in which PM10

concentration overcomes the EU limits) Low (15 days per year) Low (15 days per year) Disamenity derived from the presence of bad

odours (days per year in which you fell bad odours)

Low (15 days per year) Medium (90 days per year) Yearly compensation (reduction in the cost of

life for your family) 0€ 2000€

incinerators, it is particularly relevant for landfills. Respondents potentially living near a landfill require a compensation of €32 for each additional day in which bad odours are perceived.

The availability of green areas, expressed on an ordinal scale, plays a positive role in the case of the incinerator, although it is not significant for the landfill. The increase in the availability of public green areas tends to mitigate the aversion to incinerators. Residential choices also depend on the characteristics of the place where a respondent lives at the instant of the experiment. It seems that people living in areas already heavily influenced by the pres-ence of infrastructures are more likely to move out from these sites. The coefficient of the attribute air, which was both positive and statistically significant, does not follow the prior expectation; this result is discussed further in the next paragraph.

The variable iend identifies the endowment (number of facilities present in the current respondent’s residence), includ-ing: landfills, incinerators, water depurators, power lines,

air-ports, railways, highways. The coefficients (distinciner and

distlandfill ) are both statistically significant and the presence of

an infrastructure reduces the compensation required: €1672 for

the incinerator and €981 for the landfill. The variable relative to the household income (income) has negative sign meaning that richer people ask for higher compensation.

The rest of this section is devoted to comment the variables referred to attitudes, beliefs and health risks’ perception. Among the different protocols used to create synthetic indicators of health status we have chosen the approach of ‘Healthy Days’, used by the US Center for Disease Control and Prevention, whose validity is tested in the literature from a large number of empiri-cal studies (Andresen et al., 2001a, 2001b; Beatty et al., 1996; Brzenchek et al., 2001, among others). The number of days in the last 30 days for which the respondent claimed to have been ill, both emotionally and physically, is the index used in our analysis. We derived a dichotomous variable (health) equal 1 if the respondent has had a number of days of well-being above aver-age, and 0 otherwise. The variable is statistically significant for the incinerator with a marginal effect of €1947 and it is not sig-nificant for landfill. Healthy people want to receive more com-pensation for the risks connected to the presence of an incinerator. The following lines are devoted to factors having an effect on the Table 3. Attribute variables.

Code Name Description Levels

comp Compensation Monetary compensation proposed for each alternative. 0€ 500€ 1500€ 2500€ 3500€

dist Proximity Identifies the situation of high proximity to the facility, built as a dichotomous variable assuming value 1 when the infrastructure is at 600 m or less and 0 when the distance is 2000 m or more.

1 ( 600 m) 0 ( 600 m)

air Air pollution level Number of days in which exceeds in the limits, imposed by European

directives, of particulate emissions occur 15 year90 year 150 year

odour Odour pollution level Number of days per year in which are perceptible bad odours. 15 year 90 year 150 year

green Green areas Presence of green areas close to the house. It can assume three different levels: absence, small urban garden, park or big urban garden.

Absent (−1)

Little green area (0) Urban park (+1)

Table 4. Respondent characteristics.

Code Description

iend Number of infrastructures source of potential disamenity in the actual residential location of respondents.

hrisk In the survey we ask the respondents to express their beliefs in term of the dangerousness of the presence of an incinerator close to their home. They could indicate a value in a scale from 1 to 6. We built a dichotomous dummy to indicate the respondents most worried about the presence of the incinerator (more than 3).

nimby The respondent declares that in any case they would not be willing to host an incinerator close to their home. Regardless of the goodness of the project, the management and the public utility of the infrastructure they are strongly averse. We can interpret this behavior as an expression of the so called NIMBY effect.

health This variable is an indicator of the health status of the respondents. It is modelled as dummy variable assuming value 1 if the interviewed has declared to suffer for physical or psychological disease, in terms of days per month of illness lower than the average value.

knowledge This dummy variable assumes value 1 if the respondents declare to know the project for the realization of an incinerator in Turin, 0 otherwise.

WTA for the site with the incinerator, but no statistical signifi-cance for the landfill. The variable INFO is a proxy of respond-ents’ endowment of information about the starting of the building process of the incinerator in Turin. The result suggests that better informed people tend to reduce their aversion to the facility. The

nimby variable is defined as a dummy variable identifying those

respondents who would absolutely refuse the construction of an incinerator close to their house despite the public utility of this infrastructure for the entire province. The marginal value of the variable nimby is by far the highest (€6447) indicating the strong effect on preferences of nimby syndrome.

Conclusions

The paper presents results of a survey based on stated preferences of households, choosing hypothetical urban sites to live in. The choice behaviour is analysed through a discrete choice model implementing a monetary willingness to accept compensation for the disamenities produced by the proximity to waste disposal facilities: a landfill and an incinerator. The valuation exercise built on the specific context of Turin, in north-western Italy, is aimed to obtain estimates of the citizens’ perceptions of the exter-nalities caused by the two waste disposal facilities. According to the structure of elicited preferences, landfills generate disamenity impacts that are stronger than incinerator plants: the compensa-tion for the presence of the incinerator is around €2670 and for the landfill it is around €3816 for each household.

Empirical evidence was also found that the choice behaviour is affected by different aspects of the respondents’ life and their health status. Furthermore, people who are well-informed about the facilities features are more likely to accept the presence of the facilities. Other impacts embedded in the estimated utility func-tion enter into the choice process and refer to air pollufunc-tion, bad odours, distance from the facility and the presence of green areas. The evaluated magnitude of the willingness to accept was assessed on the Italian context and turned out to be interestingly higher than the values obtained in a previous similar UK experi-ment (Garrod and Willis, 1998). The site-specificity may be a plausible explanation for these differences in the scale of WTA estimates. This divergence of results is also plausible due to the different reference points in the lifetime of the infrastructure (a more adverse perspective with highest external costs in a pre-rumour phase, that gradually shrinks in the construction and operational life (Kiel and McClain, 1995)).

Moreover, the increase of green area endowment, as a form of non-monetary compensation, is a strong instrument to balance the perception of disamenities generated by waste disposal facili-ties, suggesting a reference point for the policy-makers employed in the management of local conflicts. Furthermore, if the case of the incinerator is considered, the aversion to this facility is well explained in terms of ‘NIMBY’ syndrome, stated by respondents in the interview and implemented in the model.

The air pollution level, expressed by the yearly number of days with excessive levels of particulate matters is present in the Table 5. Results.

Log-likelihood: −256.8825 Number of observations: 336 McFadden pseudo R2_{: .27973}

Variable Coefficient Standard error b/St.Er. P-value

comp 0.00043801 0.00013764 3.182 0.0015 dist inciner −0.83307117 0.31778317 −2.622 0.0088 dist landfill −0.74884995 0.45740706 −1.637 0.1016 air_{inciner 0.00623116 0.00276943 −2.250 0.0245} air landfill 0.00840975 0.0036016 −2.335 0.0195 odour inciner −0.00170903 0.00275859 −0.620 0.5356 odour landfill −0.01394858 0.00377371 −3.696 0.0002 green_{inciner 0.62506139 0.20818482 3.002 0.0027} green landfill −0.07459341 0.26186267 −0.285 0.7758 iend inciner 0.73269743 0.1324709 5.531 0.0000 iend landfill 0.43010840 0.15718767 2.736 0.0062 health_{inciner −0.8529001 0.03751865 −2.273 0.0230} health landfill −0.6252674 0.04427932 −1.412 0.1579 income

inciner −0.127823e-7 0.402135e-08 −3.179 0.0015

income

landfill −0.120801e-7 0.605427e-08 −1.995 0.0460

hrisk_{inciner −1.67376851 0.35388708 −4.730 0.0000} hrisk landfill −0.64663058 0.46378390 −1.394 0.1632 nimby inciner −2.82390021 0.75682678 −3.731 0.0002 nimby landfill −0.9247582 0.55531021 −0.167 0.8677 knowledge_{inciner 1.03018623 0.34493815 2.987 0.0028} knowledge_{landfill 0.35582851 0.42066209 0.846 0.3976}

alternative site profiles with values ranging from 30 to 90 days, with the real number measured from 2007 to 2010 being greater than 120 days. A possible interpretation for this positive sign is that the variation in utility does not represent a variation relative to the status quo alternative presented in the choice task, but a perceived change with respect to the real status quo, that is much worse than the designed one. Further investigation must address this point either on the design side, or in the use of more sophis-ticated econometric strategies, such as mixed logit approaches.

The field of empirical literature about discrete choice now embraces a wide series of applications, from the marketing of public services to land-use planning. This study can be con-sidered as a preliminary step in the attempt to deepen the analy-sis of public preferences as a support for the location of waste facilities. As a future extension of the study, it appears that the development of econometric methodologies in the filed of econometric modelling of discrete choice, could help the diffu-sion of both methodological and empirical driven research in this field.

Acknowledgements

The authors are grateful to the Directorate of Environment of the Piedmont Region and the Office for the Evaluation of Public Investments of the Piedmont Region (NUVAL) for funding the study. They are also grateful to Silvana Dalmazzone and Ugo Colombino, scientific coordinators of the project, for their support in all the steps of the study. Alberto Martini, Marilena Locatelli and Magda Fontana gave us helpful suggestions in the questionnaire design, and Giovanna Garrone, managed the data collection.

Funding

This research was supported by the Directorate of Environment of the Piedmont Region and the Office for the Evaluation of Public Investments of the Piedmont Region (NUVAL).

References

Andresen EM, Catlin TK, Wyrwich KW, et al. (2001a) Retest reliability and validity of a surveillance measure of health-related quality of life. Quality

of Life Research 10(3): 199.

Andresen EM, Vahle VJ and Lollar D (2001b) Proxy reliability: Health-related quality of life (HRQOL) measures for people with disability.

Quality of Life Research 10: 609–619.

Baccheschi A, Bimonte S and Ferrini S (2008) Il Contingent Behaviour per la valutazione delle esternalità odorifere dell’impianto di selezione e com-postaggio di ‘Le Cortine’ in Provincia di Siena, Working Paper No. 542 Quaderni del Dipartimento di Economia Politica, Università degli Studi di Siena, Italy.

Basili M, Di Matteo M and Ferrini S (2006) Analysing demand for environ-mental quality: A willingness to pay/accept study in the province of Siena (Italy). Waste Management 26(3): 209–219.

Beatty P, Schechter S and Whitaker K (1996) Evaluating subjective health questions: cognitive and methodological investigations. Proceedings of the American Statistical Association, Section on Survey Research Methods.

Ben-Akiva M and Lerman SR (1985) Discrete choice analysis: theory and

application to travel demand. Cambridge, MA, USA: MIT Press.

Bennett J and Blamey R (2001) The choice modelling approach to

environ-mental valuation. Cheltenham, UK: Edward Elgar.

Bobbio L (2002) Smaltimento dei rifiuti e democrazia deliberativa, Working Paper no. 1. Dipartimento di Studi Politici, Università degli Studi di Torino, Italy.

Brzenchek J, Ahern K, Dominick D, et al. (2001) Self-reported health-related quality of life and health service utilization of Alzheimer’s patients.

Gerontologist 41(S1): 366–366.

Caplan A, Grijalva T and Jackson-Smith T (2007) Using choice question formats to determine compensable values: The case of a landfill-siting process. Ecological Economics 60: 834–846.

Eshet T, Ayalon O and Shechter M (2005a) A critical review of economical valuation studies of externalities from incineration and landfilling. Waste

Management & Research 23: 487–504.

Eshet T, Ayalon O and Shechter M (2005b) An inclusive comparative review of valuation for assessing environmental goods and externali-ties. The International Journal of Business Environment (IJBE) 1(2), 190–210.

Eshet T, Ayalon O and Shechter M (2006) Valuation of externalities of selected waste management alternatives: a comparative review and analy-sis. Resources, Conservation & Recycling 46(4): 335–364.

ExternE (1998), Research Project, European Commission DG Environment, 1995–1998, Brussels, Belgium: European Commission.

Ferrini S and Scarpa R (2007) Designs with a priori information for non-market valuation with choice experiments: A Monte Carlo study.

Journal of Environmental Economics and Management 53(3):

342–363.

Garrod GD and Willis K (1998) Estimating loss amenity due to landfill waste disposal. Resources, Conservation and Recycling 22: 83–95.

Horowitz JK and McConnell KE (2002) A review of WTA/WTP studies.

Journal of Environmental Economics and Management 44: 426–447.

Kanninen BJ (2002) Optimal design for multinomial choice experiments.

Journal of Marketing Research 39(2): 214–227.

Kiel KA and McClain KT (1995) House prices during siting decision stages: the case of an incinerator from rumour through operation. Journal of

Environmental Economics and Management 28(2): 241–255.

Louviere JJ, Hensher DA, Swait JD (2000) Stated Choice Methods. Analysis and Applications. Cambridge, UK: Cambridge University Press.

McFadden D (1974) Conditional logit analysis of qualitative choice b14ehaviour. In Zarembka P (ed.) Frontiers of Econometrics. New York: Academic Press, pp. 105–142.

Norese MF (2006) ELECTRE III as a support for participatory decision-making on the localisation of waste-treatment plants. Land Use Policy 23: 76–85.

Sasao T (2004) An estimation of the social cost of landfill siting using a choice experiment. Waste Management 24: 753–762.

Tipaldo G (2006) La costruzione del consenso intorno a un inceneritore di rifiuti: il caso di Torino. Proceedings of the Italian Society of Political Science, Bologna, September 12–14.

Street DJ and Burgess L (2007) The construction of optimal stated choice

experiments: theory and methods. New York, USA: Wiley.

Appendix 1

Choice experiments (CE) applied to the valuation of

environ-mental goods and disamenities effects may be considered as a generalization of the dichotomous close-ended format of con-tingent valuation (Bennett and Blamey, 2001). From an econo-metric point of view they extend to a multinomial structure the classical binomial response, as during the interview the respondents express their preferences by choosing their favour-ite option among a set of alternative scenarios. The choice prob-ability of each option is linked to the perceived level of satisfaction and to the attractiveness of each mutually exclusive

option (Ben-Akiva and Lerman, 1985; McFadden, 1974). The data analysis is based on a conditional logit model (McFadden, 1974), for which the general assumptions is rationality among agents who will maximize their utility level in the presence of a budget constraint. According to random utility models (RUMs), the utility function can be decomposed into two parts: a deter-ministic component and a stochastic component (error term) generated by the presence of unobservable characteristics and heterogeneity among individuals. The usual form of this utility function is:

Uiq=Viq( ,β Xiq)+ε iq (1)

Where U_iq is total utility, V_iq is the deterministic component that is function of Xiq a vector of observables and εiq the stochastic

com-ponent; i identifies the alternative in each choice experiment and

q is referred to individuals. If utility associated to alternative i is

grater than each other alternative j, the individual q will choose i, in this sense the probability that q will choose among a set of J alternatives is: P U U j C j i V V j C j i iq iq jq iq jq jq iq = > ∀ ∈ ≠ = − > − ∀ ∈ ≠ Pr( , ) Pr( ε ε , ) (2)

The assumption of independent extreme value distribution for random variables εjq allows the expression of probability of

choosing alternative i to be written in a closed-form solution:

P e e iq V V j J iq jq = =

∑

where λ is a scale parameter conventionally normalized to 1.

The parameters’ estimates included in V were obtained using the maximum likelihood method.

In order to derive a monetary valuation we define p the compen-sation (or in general the monetary attribute), X the vector of all the n attributes, and S the vector of characteristics of the respondent. These were introduced in interaction with alternative specific constants d_j. The indirect utility function can be written in linear form (Ben-Akiva & Lerman, 1985; Louviere, 2000):

V p X nXn pp d S n J S J ( , )=

∑

∑

∑

substitu-tion between the attribute n and the monetary compensasubstitu-tion p, and is usually employed to evaluate on a monetary scale, the weight of a marginal variation of n.