Metodi per tenere conto della complessità del disegno nella stima di modelli per la propensione all'acquisto

Abstract

The aim of the work is to estimate a model describing households' purchase of durable goods by using data from the italian Consumer Expenditure Survey. Since the sample design turns out to have important consequences for estimation, two approaches are used to incorporate sample weights in the model estimation: the conventional weighting, which suffers some limitations, and a method recently suggested by Skinner (1994) in order to overcome those limitations. The empirical study proves more efficient estimates are obtained by Skinner's approach. nevertheless, it could lead to different estimates of the parameters for different subgroups of the population when qualitative regressor are selected in the model.