Published online by Cambridge University Press: 28 April 2015
Use of the Tobit model for estimation of adult equivalent scale (AES) parameters and expenditure equations presupposes that zero observations represent exclusively corner solutions. This paper tests for the underlying causes of zero observations and estimates accordingly the AES parameters, the household size in adult equivalents and the resulting expenditure equations and consumption profiles. The empirical application concerns Greek household “soft” and “hard” cheese consumption. The statistical results indicate that for the examined cheese categories the “double hurdle participation” model adjusted for non-normality and heteroskedasticity is the appropriate model for the estimation of AES parameters and expenditure equations.