Les méthodes d'estimation non-paramétrique et semi-paramétrique ont suscité beaucoup d'intérêt ces dernières années. En économétrie, elles permettent une plus grande flexibilité dans le choix des modèles de régression. Opposées dans un premier temps à l'économétrie classique, ces techniques se sont finalement avérées lui être fortement complémentaires, pouvant légitimer le choix d'un modèle paramétrique. Cet ouvrage présente une introduction accessible à ces méthodes (noyau, polynômes locaux, splines, ondelettes, modèles de mélanges). Conçu comme un ouvrage pédagogique illustré de plusieurs applications, l'accent est mis sur la compréhension des méthodes, les intuitions sous-jacentes et la mise en œuvre en pratique. Il s'adresse aux étudiants en sciences économiques et en gestion, aux élèves des écoles d'ingénieurs et de commerce, ainsi qu'aux professionnels et praticiens de l'économétrie.

Surveys are sometimes viewed with suspicion when used to provide economic values, since they are sensitive to framing effects. However, the extent to which those effects may vary between individuals has received little attention. Are some individuals less sensitive to framing effects than others? We use the theory of social representation to assign to each individual a new variable to serve as a proxy for the individual's sensitivity to framing effects. This allows to gather new and relevant information to limit the impact of framing effects. We examine two framing effects, starting-point bias and willingness-to-pay and willingness-to-accept divergence.

A random sample drawn from a population would appear to offer an ideal opportunity to use the bootstrap in order to perform accurate inference, since the observations of the sample are IID. In this paper, Monte Carlo results suggest that bootstrapping a commonly used index of inequality leads to inference that is not accurate even in very large samples. Bootstrapping a poverty measure, on the other hand, gives accurate inference in small samples. We investigate the reasons for the poor performance of the bootstrap, and find that the major cause is the extreme sensitivity of many inequality indices to the exact nature of the upper tail of the income distribution. Consequently, a bootstrap sample in which nothing is resampled from the tail can have properties very different from those of the population. This leads us to study two non-standard bootstraps, the m out of n bootstrap, which is valid in some situations where the standard bootstrap fails, and a bootstrap in which the upper tail is modelled parametrically. Monte Carlo results suggest that accurate inference can be achieved with this last method in moderately large samples.

We examine the statistical performance of inequality indices in the presence of extreme values in the data and show that these indices are very sensitive to the properties of the income distribution. Estimation and inference can be dramatically affected, especially when the tail of the income distribution is heavy, even when standard bootstrap methods are employed. However, use of appropriate semiparametric methods for modelling the upper tail can greatly improve the performance of even those inequality indices that are normally considered particularly sensitive to extreme values.