An anatomy of household income volatility in European countries1

This paper offers an exploratory analysis of household income volatility in the 1990s in fourteen EU countries and two future member states, namely Hungary and Poland, using simple summary statistics for average income changes as advocated in Fields and Ok (JET 1996, Economica 1999). The evidence is derived from the newly generated data of the Consortium of Household Panels for European Socio-Economic Research (CHER) that contain harmonised data from the European Community Household Panel and from a series of independent panel surveys. Going a step ahead to overcome the obvious restriction of looking only at population averages as Fields and Ok suggest, both the overall distribution of individual income variations, and the variations in levels of income volatility for different starting income levels are also examined. The analysis can be viewed as looking at the primitives of income mobility at the individual level, as opposed to many of the existing analyses that assess, using other more sophisticated concepts, the aggregate outcome (like inequality of long-term income) resulting from these individual income variations. The issue of accounting for the observed cross-national differences is addressed. Important determinants of household income, and of its variability over time, such as labour market flexibility and family formation habits vary across countries and may explain cross-national differences in income volatility. This paper attempts to evaluate how much of these crossnational differences can be accounted for by differences in the socio-demographic structure of the populations, as well as in cross-national variations in the dynamics of labour market and household formation. In particular, volatility levels are compared after controlling jointly for cross-national differences (i) in the prevalence of female-headed households, (ii) in household composition (by size, number of children, and age of household head), (iii) in the frequency of household composition changes, and (iv) in the frequency of changes in the household labour market attachment. To this aim, non-parametric (or semi-parametric) methods are derived from those developed by DiNardo, Fortin and Lemieux (Econometrica, 1996) in the context of intertemporal income distribution comparisons.