Engel elasticities, pseudo-maximum likelihood estimation and bootstrapped standard errors. A case study

Abstract

Estimation of standard errors of Engel elasticities within the framework of a linear structural model formulated on two-wave panel data is considered. The complete demand system is characterized by measurement errors in total expenditure and by latent preference variation. The estimation of the parameters as well as the standard errors of the estimates is based on the assumption that the variables are normally distributed. Considering a concrete case it is demonstrated that normality does not hold as a maintained assumption. In the light of this standard errors are estimated by means of bootstrapping. However, one obtains rather similar estimates of the standard errors of the Engel elasticities no matter whether one sticks to classical normal inference or perform non-parametric bootstrapping.