Income, consumption and wealth - Experimental statistics

Income, consumption and wealth

Why do we produce statistics on the joint distribution of income,
consumption and wealth?    Experimental statistics logo

Disparities in income and wealth are increasingly scrutinised, not only by the academic world but also by the public. The joint distribution of income, consumption and wealth data provides links between the three economic dimensions. These data help to describe more thoroughly material well-being and households' economic vulnerability. They also help to explain the dynamics of wealth inequalities.

Why are these data published as experimental statistics?

These statistics are based on the assumption that the link between income, consumption and wealth could be completely explained by the other variables such as age, household structure, etc. that all data sources have in common. While this assumption is well documented in academic literature, it can easily be challenged.

How are these statistics produced?

The dataset underlying these statistics is based on information from different data sources.

To obtain statistics on the joint distribution of income, consumption and wealth requires that information on these three dimensions is collected/available for the same household. This can be very cumbersome and almost impossible, in particular if surveys are used.
To overcome this, the European Statistical System has proposed several options (see in particular the Vienna Memorandum). The only option that could be implemented immediately consists of merging information coming from different datasets/data sources. This merged dataset, constructed by means of statistical matching, contains information on the three dimensions for each household.

> read more

Access the statistics


To help Eurostat improve these experimental statistics, users and researchers are kindly invited to give us their feedback:

  • How useful are such indicators? Would you recommend the publication of other indicators, which ones?
  • Would you recommend having access also to data on uncertainty? In case yes, would you like the statistics to come along with information on the range of plausible values for the indicator?

Here is a specific question for researchers who can be granted access to microdata for scientific purpose: Would you consider it useful to have access to scripts that performed the matching, provided that you have access to the necessary microdata?