Emmanuel Flachaire

Bootstrapping heteroskedastic regression models: wild bootstrap vs. pairs bootstrap

In regression models, appropriate bootstrap methods for inference robust to heteroskedasticity of unknown form are the wild bootstrap and the pairs bootstrap. The finite sample performance of a heteroskedastic-robust test is investigated with Monte Carlo experiments. The simulation results suggest that one specific version of the wild bootstrap outperforms the other versions of the wild bootstrap and of the pairs bootstrap. It is the only one for which the bootstrap test always gives better results than the asymptotic test.

Estimation of the income distribution and detection of subpopulations: An explanatory model

Empirical evidence, obtained from non-parametric estimation of the income distribution, exhibits strong heterogeneity in most populations of interest. It is common, therefore, to suspect that the population is composed of several homogeneous subpopulations. Such an assumption leads us to consider mixed income distributions whose components feature the distributions of the incomes of a particular homogeneous subpopulation. A model with mixing probabilities that are allowed to vary with exogenous individual variables that characterize each subpopulation is developed. This model simultaneously provides a flexible estimation of the income distribution, a breakdown into several subpopulations and an explanation of income heterogeneity.

Propriétés en échantillon fini des tests robustes à l'hétéroscédasticité de forme inconnue

Dans la pratique, la plupart des statistiques de test ont une distribution de probabilite de forme inconnue. Generalement, on utilise leur loi asymptotique comme approximation de la vraie loi. Mais, si lechantillon dont on dispose nest pas de taille suffisante cette approximation peut etre de mauvaise qualite et les tests bases dessus largement biaises. Les methodes du bootstrap permettent dobtenir une approximation de la vraie loi de la statistique en general plus precise que laloi asymptotique. Elles peuvent egalement servir aapproximer la loi dune statistique quon ne peut pas calculer analytiquement. Dans cet article, nous presentons une methodologie generale du bootstrap dans le contexte des modeles de regression.

The Role of Economic Space in Decision Making: A Comment

Over the last few years, Margaret Slade has contributed to some major improv ments in the field of industrial economics. The important question of location and spatial interaction in economic decision is one of her central interests. Her paper, prepared for a presentation at the ? Conf?rence de VADRES ? in Paris, presents the ways and the methods she developed with her coauthors to incorporate the influ ence of space location in regression model. The new attention to specifying, esti mating and testing for the presence of spatial interaction they have taken, concerns the use of semiparametric methods to allow less restrictions on the form of the spatial dependence. The paper is clearly written, without technical developments and the discussion of potential applications is very convincing on the significant role that the location can take in economic decisions. In the standard linear model, there are two ways to incorporate spatial depen dence : in the covariance matrix of the error term and/or in the parametric portion of the model as additional regressors. In this comment, I would like to explore dif ferent utilizations of semiparametric methods to treat the problem of location effect in regression model and compare them with the methods used by Margaret Slade and her coauthors. These different utilizations suggest that location effects could be incorporated in regression model at a low cost, with an easy estimation of the model and a simple interpretation of the estimators. In section 2,1 consider the specification of the dependence in the error term and propose to use semiparametric methods in order to obtain efficient estimators. In section 3,1 consider the specification of the dependence as additional regressors