Page 1 of 1

heterogeneity in redeemed vouchers and other issues

Posted: Thu Sep 05, 2019 2:38 pm
by Luca Tiberti
Dear team,

thanks for your interesting proposal!

- Very good introduction and discussion on possible impact pathways. However, the empirical analysis does not seem to try to address these points. Please develop further.
- You claim that the subsidy input usage can vary because people do not redeem the full vouchers or because they sell/buy them. Such a variation does not look being endogenous. This might be an issue for identification
- I suggest that your preferred estimation is the one controlling for individual and temporal fixed effects.
- Panel data: you are silent on the potential attrition across waves. Please clarify.
- Instrumental variable: good choice, but a few issues:
o Variation in rainfall can actually be correlated with some possible “mechanisms variables”
o Rainfall is observed at the community level and may affect the distribution of FISP. However, the real number of redeemed vouchers is at the household level and, given the decreasing trend, I am not anymore sure that rainfall might be a convincing instrument.
- The voucher is provided to the household head: what are the implications if the household head is a man (which, I guess, is the majority of households). The final effect might depend on internal to household sharing rule. I would controlled for a distributive factor and interact it with the FISP variable.
- In the presentation, you added an interaction term between the FISP and female. Does female capture whether the grantee is a woman (i.e., the household head is a woman)?
- FISP variable: is there a specific question on FISP in your dataset?

Thanks,
Luca