Page 1 of 1


Posted: Fri May 31, 2019 10:13 am
by Filipe Lage de Sousa
It was a pleasure reading such interesting paper and I have only a few comments aiming to improve paper analysis.

1) Instead of using over, under or matched, it might be good to look at matched and unmatched, since matched is positive and under&over are negative.

2) Explore more results of the different measures of matched an unmatched. And when perhaps uses a range of effects rather than a single number. For policy, it is advisable to use an unique number, but for the paper itself, it might be good to use an interval.

3) In the paper, there are two scales one from zero to two (0,1,2) and another from 1 to 3 (1,2,3)

4) Instrument: in the paper, the authors state that passing the statistical tests says that the intrument is valid. Actually, not passing the text rulled out the idea that the instrument is valid and you can not say that your instrument is valid just because it passes the test

5) The authors might explore more the reasons of why mismatch increases with age.

6) Descriptive statstistics tables does not show the number of observations

7) Authors might explore more information provided in the descriptive stats, such as does women suffer more? Do those with college degree suffer more? Region? And others.... To do that, you may just interact with a dummy for the category that you are aiming to investigate.

If you have any query on the suggestions made, please do not hesitate to contact me by email (