Page 1 of 1

a few queries

Posted: Thu May 30, 2019 2:32 pm
by Luca Tiberti
Dear Kibrom and team,

thanks for your presentation!

- in order to improve identification, I suggest adding a few socio-economic EA level pre-2011 control variables
- I am a bit puzzled about the important negative variations in most of the outcomes (not only the real welfare but also the labor market ones) between 2011 and 2013. I would check again the variables.
- I am thinking whether you should – at least as an additional definition – normalize the average NL variable by the variance within the EA. This would allow to better capture the within EA heterogeneity. This would be particularly indicated as you are not able to use HH level GPS coordinates.