Le résumé en français n'a pas été communiqué par l'auteur. / This dissertation consists of four essays on health and demographic economics. In the first chapter, I explore the implementation of the theory of equality of opportunity (EOp) developed by Roemer (1998) to health in a joint work with Guillem Lopez and Frederic Udina. A common impediment to the achievement of EOp applications with given resources constraints is that it is unlikely that public policies can fully compensate for existing unfair inequalities. This scenario is particularly relevant in the case of health policy, where public spending coexists with a large private spending component. We argue that if social justice is not attainable, social deliberation should not only focus on choosing the circumstances that ought to be compensated but also reflect on which groups suffering unfair inequalities should be prioritized. The second chapter examines the impact of income-related reporting heterogeneity on the measurement of health inequality. While most studies of health inequality rely on self-reported measures of health, recent research has studied the possibility that part of the existing differences in self-reported health could be due to systematic differences in reporting across socioeconomic groups. The concern is that part of the existing inequalities may not be founded on differences in the “true” health status of individuals. In particular, some studies have concluded that reliance on self-reported health might have resulted in an overstatement on the degree of health inequality of some countries. I study the income-related reporting heterogeneity hypothesis in the 2006 wave of the Catalan Survey of Health and I find that the main contributor to health inequality is the disproportionate concentration of the prevalence of reported conditions in lower income groups. The third chapter, joint with Hippolyte d'Albis and Loesse Jacques Esso, studies the trends in mortality convergence across developed countries from 1960 to 2008. While the epidemiological transition has provided a theory behind the expectation of convergence in mortality patterns, our results reject the convergence hypothesis for a sample of industrialized countries. We study the disparities across the mortality distributions of the countries and our sample and find no evidence of convergence towards a common mortality distribution.The fourth and final chapter of this dissertation examines the relationship between unemployment and fertility. I offer a possible explanation for the apparent contradiction between the empirical work that finds a negative relationship between unemployment and fertility and the theoretical work that emphasizes the lower opportunity cost of childbearing while unemployed. I reconcile these perspectives by distinguishing two forms of unemployment. The first form is structural unemployment while the second is cyclical unemployment, a less permanent component of unemployment that is linked to the economic cycle. I study both effects over the life cycle using cohort data on a panel of developed countries. I find that while structural unemployment has an unambiguous negative effect on fertility, reactions to cyclical unemployment depend on the age at which it is experienced.

Cette thèse est la compilation d’essais qui appliquent les méthodes expérimentales et d’enquêtes à deux sujets portant sur la prise de décisions en groupe. La prise de décision en groupe est un type de processus participatif dans lequel plusieurs individus agissent collectivement, analysent des problèmes ou des situations, examinent et évaluent les solutions alternatives d’action, et choisissent parmi les différentes alternatives, une ou plusieurs solutions (Van den Ven et Delbeq, 1974). Une des caractéristiques parmi les plus importantes dans la prise de décisions en groupe est que les individus ont une responsabilité individuelle (Katzenbach et Smith, 1993). Par conséquent, cette thèse s’intéresse à la fois aux décisions effectivement prises au niveau du groupe, mais également aux opinions individuelles.Le premier sujet s’inscrit dans le domaine du bien-être et du choix social et porte sur l’égalité des chances (désignée EOP par la suite). Nous testons d’abord les fondements intuitifs de l’EOP au moyen d’une enquête, pendant laquelle les individus sont "des spectateurs impartiaux" sans aucun gain personnel. Cette enquête nous permet d’illustrer la notion de justice impartiale en absence d’intérêt. Nous testons ensuite les fondements au cours d’une expérience non coopérative, où les décisions prises ont des conséquences sur les gains des individus. L’expérience utilisée dans ce sujet s’intéresse aux préférences révélées en matière de justice par les preneurs de décisions. Le deuxième sujet porte sur les systèmes de vote à majorité pondérée (désignés WMV par la suite). Nous réalisons d’abord une enquête auprès de différents groupes d’étudiants pour découvrir que les individus ne tiennent pas compte de la relation non monotone entre le nombre de votes et le pouvoir de vote associé. Nous menons ensuite plusieurs expériences de laboratoire afin de tester si cette relation non monotone peut être apprise dans le cadre d’un jeu répété. Ainsi, la première partie de cette thèse comprend deux chapitres à propos de l’EOP, tandis que la deuxième comprend deux chapitres à propos des WMV. Le premier chapitre est une introduction générale, le Chapitre 6 expose les conclusions.Le Chapitre 2 présente l’enquête de l’EOP, laquelle est une enquête sur les préférences utilisant de nombreux visuels pour présenter des scénarios hypothétiques. Ceci nous permet de révéler les préférences des individus sur les principes de l’EOP. À notre connaissance, notre enquête est la première tentative pour explorer pleinement le concept de l’EOP du point de vue d’un “spectateur impartial”. Nous avons analysé systématiquement cinq facteurs qui relèvent de l’EOP dans deux vignettes et plusieurs scénarios. Entre les deux vignettes : ventes et alcool, nous avons trouvé un niveau élevé de consensus sur les circonstances. Toutes les personnes ne devraient pas subir les conséquences liées aux circonstances dans toutes les situations pour au moins deux raisons : premièrement, cela serait clairement arbitraire du point de vue moral et deuxièmement parce qu’il faut ajuster la péréquation résultante. Le maigre consensus sur l’effort et la chance brute est également présenté dans nos résultats. Puisque la chance brute est définie par rapport au décroissement des risques liés à la chance de non-option, elle peut entrainer des résultats non souhaitables au regard de l’EOP. Inversement, l’effort est défini comme un reflet du comportement des individus. Ils contrôlent cet effort, lequel exerce une influence sur leur statut. Ainsi, les effets des différents niveaux d’effort pourraient être non neutralisés. / Decision making in groups largely exists in almost every aspect of daily life. In this thesis, we use survey and experimental methods to examine decision making in two different areas, equality of opportunity (EOP) and weighted majority voting (WMV). The goal of the research efforts is to understand the importance of achieved decisions for the functioning of the group, involving taking into account the needs and opinions of every group member.In the survey study of EOP, we used the attitude survey consisting of vignettes to elicit stated preferences of quasi-spectators over the conceptions of EOP problems. Impartiality in this study is achieved by considering only evaluations of individuals who have no stake in the situation they are judging. We analysed whether individuals are considered to be held responsible for outcomes by different factors: circumstances, effort, talent, brute luck and option luck. Apart from the baseline treatment, another compared treatment was designed to test context effect, with the introduction of ``need'' and social-cognitive age. In these two treatments, we found a high level of consensus on non-responsibility of circumstances, slight consensuses to responsibility of effort but non-responsibility of brute luck, and no agreement on talent and option luck. When other considerations are not involved, inequality caused by differential option luck and talent is unobjectionable. While with other considerations, people should not fully be held responsible for the outcomes resulting from option luck and talent.We then tested revealed preferences over the conceptions of EOP problems in an experiment, where subjects were stakeholders of each decision. People were recruited to form a micro society, and engaged in actual decisions about redistribution of their own money they earned in a previous phase through four factors: circumstances, effort, brute and option luck. Yet despite a lot of self-interested behaviours when participants have a stake in the redistribution, we clearly observe that participants are sensitive to the different factors through which people can earn their money. There were three treatments in our experiment: the baseline, prior reflection and talent treatments. The voting results among three treatments are very homogeneous, with only one exception on effort in the talent treatment, and present a very high level of self-serving bias.For WMV, the survey study investigates whether people have the knowledge of the relation between the number of votes and corresponding power which they have in a committee that takes decisions with a WMV system. Power indices show that actual voting power is often quite different from the nominal distribution of voting weights. Our survey observed that subjects cannot see through the non-monotonic relationship between the number of votes and voting power. In addition, we found that subjects trained in political sciences had more acquaintance on this issue.The experiment on WMV wanted to see whether people could learn this non-monotonic relation between the number of votes and the actual voting power. In the laboratory experiment, people played similar games 20 times. The first stage of games was the votes apportionment stage, where two out of four subjects in a group independently and simultaneously made decisions to determine the distribution of votes among four members. The second stage of games was the point allocation stage, where four member bargained to divide fixed amount of resources among themselves through the WMV determined in the first stage. The results of our experiments show that initially subjects tend to choose an option that gives them more votes ignoring how remaining votes are distributed among others. But, as subjects ``learn'' about the payoffs they can obtain in the points allocation stage, they start to choose the option in the first stage that could have chance to give them higher payoffs.