Apres une presentation de la construction de predicteurs par arbre de classification, nous nous interessons a l'instabilite de cette methode et proposons une methodologie dans laquelle intervient le bootstrap. Une etude empirique detaillee illustre ce travail.

CART est une des rares techniques non parametriques permettant d'etablir une hierarchie des variables explicatives. Dans cet article nous decrivons le calcul de l'importance des variables a partir d'un arbre de regression et nous analysons un exemple de hierarchie de variables explicatives...

Apres avoir empiriquement mis en evidence l'instabilite des procedures CART de classification par arbre, nous presentons des methodes d'agregation de classificateurs obtenues a l'aide d'un reechantillonage de type bootstrap. Enfin, nous mettons en oeuvre ces procedures sur un probleme de...

Nous montrons à l’aide de l’enquête Logement que les personnes d’origine africaine sont surreprésentées dans les grandes agglomérations et tendent à s’y fixer dans des proportions beaucoup plus importantes que les personnes d’origine française. Deux raisons expliquent ce...

This paper examines the normative properties of an empirically implementable dominance criterion for comparing alternative distributions of two attributes, one of which being cardinally measurable, between an arbitrary number of individuals. The criterion, which generalizes the one proposed by...

This paper examines how allowing individuals to emigrate to pay lower taxes changes the optimal non-linear income tax scheme in a Mirrleesian economy. Type-dependent participation constraints are borrowed from contract theory. An individual emigrates if his domestic utility is less than his...

Consider a developing country that has the potential for biodiversity conservation, and developed countries that benefit from biodiversity but are not in position to produce it. From the statu quo, some incremental protections of biodiversity would be harmful for the developing country but would...

inequality decreases and welfare increases as a result of a progressive transfer. We explore the implications for welfare and inequality measurement of substituting the weaker absolute differentials and deprivation quasi-orderings for the Lorenz quasi-ordering. Restricting attention to...

Under perfect observation of incomes, designing such scheme boils down to solving an optimisation program under constraints, which can be achieved with well-defined methods. In contrast, when incomes cannot be perfectly observed, the schemes are usually based on predictions of living standards...

The universal moral public good of fighting poverty is provided by both public and private transfers. Efficient public transfers do not crowd out giving because of the particular motives for it. Understanding these effects is necessary for both explaining aid and choosing policy. This analysis...