I have data concerning the distribution of income, at the individual level, i.e. I have for each individual income and some socio-demographic descriptors. I have this kind of data for several years, but for unrelated individuals (it is not panel data). With this data I can calculate various inequality measures (e.g. Gini coefficient, D9/D1 ratio, relative poverty threshold (EU definition : 60% of the median income)).

My question is : how can I examine the effects of the sociodemographic determinants and more particularly the changes over time of these determinants on these indicators ? A question I would like to be able to answer is "what is the influence of the increasing share of single parents/increasing female participation in the workforce/the increase of the average schooling duration?" on these indexes of inequality.

One of the difficulty lies in the fact that these indexes depend on the whole distribution, not only on means. The most convincing literature I have found yet would be using unconditional quantile regressions / RIF regressions as descriped in unconditional quantile regressions and Decomposition methods in economics, but it seems a bit "heavy".


  • Is this method actually making sense in my context ?
  • Are there any other "obvious" methods I have missed ?
  • $\begingroup$ To me, the best way to analyze these questions would be to have enough information at the individual level to track changes in subjects' status on the relevant factors and how this impacts income. So, as people go from nonparent to parent as a function of marital status or from jobless to employed by gender, etc. What will be very hard to decompose is the confounding as a function of income, poverty, education, social status, etc., which are all correlated. $\endgroup$ – Mike Hunter Nov 2 '15 at 16:00
  • $\begingroup$ Unfortunately I do not have panel data, it is different individuals each year $\endgroup$ – Anthony Martin Nov 2 '15 at 17:19
  • 2
    $\begingroup$ Just mentioning Firpo (2009) and assuming that people recognise the reference is going to cut your readership down to a very few people who work in the same field. More crucially, open and general questions here usually backfire on the OP, as people are then just unclear about what you want and/or unable or unwilling to respond to what looks like a fishing expedition. I am now confused too: title says "determinants of inequality" but the question itself seems all about the consequences of inequality. $\endgroup$ – Nick Cox Nov 2 '15 at 20:32
  • 1
    $\begingroup$ @DJohnson I have the data each year from 1996 to 2012, likely 2013. But my concern is definitely about micro-level analysis. $\endgroup$ – Anthony Martin Nov 4 '15 at 10:19
  • 1
    $\begingroup$ Came across this paper by Phillip Tobler on the neural representations of inequality... here's a link to a google search on this topic ... google.com/… $\endgroup$ – Mike Hunter Dec 1 '15 at 12:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.