A study of income stability in the Czech Republic by finite mixtures
J. Bartosova and N. T. Longford
Abstract
Income, expenditure and similar variables in monetary units tend
to have distributions similar to log-normal. Description of such variables
after logarithmic transformation by the normal model is often not accurate
enough, especially for multivariate data. Deviations of their empirical
distributions from the theoretical lognormal distribution often require
more sophisticated analysis. Mixtures represent a very fl exible way of
reconstructing complex distributions with irregular features and are
suitable for detailed modelling. Multivariate mixture models are applied
to the Czech longitudinal survey of household income in the European Union
Statistics on Income and Living Conditions (EU-SILC) in 2005-2008.
The analysis identifi es distinct patterns of progression of income,
with a high percentage of households having steady annual increases over
the four years (three transitions). Graphical presentation of the results
is emphasised.
In Prague Economic Papers 23, 330-348 (2014)