However, these new families of densities make the restrictive hypothesis that the income distribution has a single
mode, so they cannot detect properly heterogeneity in the sample. A finite mixture of distributions provides a flexible
parametric framework for statistical modelling allowing both for flexibility and for the treatment of tail behaviour. The
choice of the members of the mixture can be of importance. For instance, Flachaire and Nunez (2002) make use of a finite
mixture of normal distributions for modelling log income. Chotikapanich and Griffiths (2008) use a finite mixture of Gamma
distributions while Paap and van Dijk (1998) use a finite mixture formed by a Weibull and a truncated normal to model
GDP per capita for 120 countries.