Page 1 of 1

a few suggestions

Posted: Thu May 30, 2019 3:06 pm
by Luca Tiberti
Dear Nad├Ęge and team,

many thanks for your presentation and important work! Here below a few suggestions:

- you provide three different methodological approaches to address potential endogeneity biases. At some point, I would make sure which approach is the preferred one, and relegate the two others for robustness checking;
- from descriptive statistics, you find no significant difference in sales between credit and non-credit takers (they are practically the same). Then, you find some strongly significant results once you correct for endogeneity. These two results combined together, I am guessing that you have important heterogeneity effects, e.g. by credit level, in your population. I would investigate this aspect;
- it is important to explain the direction of the (downward) bias (OLS VS "corrected" estimates)