This thesis is concerned with (i) how to estimate the perceived mortality risk, (ii) how to calculate the welfare change of mortality risk reduction and (iii) whether ambiguity aversion influences subjects' treatment decision. This study is an important topic in environmental and resource economics, and the attempt to introduce ambiguity preference into the models might shed light on future research in nonmarket valuation. In this study, I estimate the economic value of reducing mortality risk relating to arsenic in drinking water employing contingent valuation in U.S. arsenic hot spots. Re-cent studies have shown that perceived risk is a more reliable variable than scientific assessments of risk when applied to interpret and predict individual's averting behavior. I am also interested in the confidence level of perceived risk, which was elicited and treated as the degree of risk ambiguity in this paper. I develop a formal parametric model to calculate the mean willingness to pay (WTP) for mortality risk reduction, and find weak evidence of ambiguity aversion.

This thesis presents an analysis of subjective risk information and participation in the National Flood Insurance Program (NFIP). Data are taken from a survey of residents in flood-prone coastal regions in the southeastern U.S. Regression models are constructed to better understand factors affecting individuals’ perceived risk ambiguity related to flood risk and the role of risk preferences, risk perceptions, and especially risk ambiguity, on the decision to purchase flood insurance. This is the first study not only of the influence of risk ambiguity on NFIP participation, but also of the impact of using different risk perception measures. Results indicate that NFIP participation is significantly affected by mean perceived risk, but the influence of range/variance of perceived risk, which presents one’s perceived ambiguity, is mixed.

The probability of a cycling crash is much higher at intersections as along the road. A number of reasons contribute to this difference, for example car drivers overlooking cyclists when taking a turn. There have been attempts to quantify the risk at prototypical, un-signaled intersections featuring different levels of cycling infrastructure, as well as cyclists' perception of risk of these intersections. However, these attempts are limited to regular, four-arm intersections, although irregular intersections featuring both a higher and a lower nwnber of anns as weil as odd angles are likely to pose additional challenges for cyclists. There appears tobe little research on the question how the complexity and layout of such intersection affects cyclists perception of risk, as weil as their allocation of attention towards the different arms of an intersection. In, we presented a first approach to taclde this issue in a virtua1 reality (VR) based setup. We found evidence that tbe type oftum affected the subjective risk (e.g. with. a higher risk associated with situations requiring a sharp turn or to continue to an offset road), but no effects of the general position of an intersection arm in relation to the cyclist' traveling trajectory. However, the repeated exposure to the same intersection in this stu.dy limits the conclusiveness of the findings. We thus developed a more flexible virtual environment allowing us to investigate the attention allocation and risk. perception at various types of intersections.

Cette thèse a pour but d'étudier les méthodes d'évaluation des ressources environnementales : la méthode des Coûts de Transport (CT) à préférences révélées et la méthode d'évaluation contingente (EC) à préférences déclarées afin de proposer les contributions suivantes. Dans le chapitre 1, nous montrons qu'il est possible si les données sur les deux méthodes sont disponibles pour les mêmes observations, d'obtenir une meilleure mesure de la disposition à payer (DAP) par la combinaison des deux méthodes en utilisant la technique du maximum de vraisemblance simulé. Dans le chapitre 2, nous montrons qu'une nouvelle approche: le "special regressor" pourrait être une solution pour traiter les problèmes d'endogénéité en EC. En utilisant des données sur la DAP pour réduire les risques subjectifs de mortalité due à la présence d' Arsenic dans l'eau potable , nous montrons que le problème d'endogénéité du niveau subjectif de risque de mortalité peut être réglé efficacement. Enfin dans le chapitre 3, en utilisant une nouvelle enquête sur l'électrification rurale au Rwanda, nous proposons un nouveau design de la méthode d'EC en permettant aux personnes interrogées de choisir entre une contribution en temps ou en argent. Ainsi, en plus de mesurer une DAP classique, nous obtenons aussi une disposition à contribuer du temps mesurée en jours, qui est une mesure aussi voire même plus pertinente que la DAP dans le contexte d'un pays en développement. / This dissertation aims at investigating the methods of the environmental resources valuation: revealed preferences Travel Cost (TC) method and stated preferences Contingent Valuation (CV) method in order to propose the following contributions. In chapter 1, we show that it is possible if both CV and TC data are available for the same observations, to obtain a better measure of willingness to pay (WTP) by combining the two methods using Simulated maximum Likelihood technique. In chapter 2, we show that the new special regressor approach could be a solution to treat endogeneity issues in CV. Using data on WTP for reducing subjective mortality risks due to arsenic in drinking water, we show that the endogeneity of the subjective mortality risk level can be treated effectively. Finally in chapter 3, using a new survey about rural electrification in Rwanda, we propose a new design for the CV method by allowing people to choose between a contribution in time or in money. Thus, in addition to measure a conventional WTP, we also obtain a willingness to contribute time measure which is as or even more relevant than WTP in the context of a developing country.