### structure, hypotheses validation and other few issues

Posted:

**Mon Jun 11, 2018 1:53 pm**Thanks Martina for your very good presentation. Here a few comments I would like to share with you and your team:

* You present some useful discussion on the expected results depending on the chosen approach (neoclassical/perfect labour markets-mobility VS imperfect labour markets-mobility). But which is the conceptual framework you adopt in the empirical setting? I would also suggest looking at the review of studies on the role of labour market frictions due to search costs and to cases where higher wages induce larger work efforts (Halmlund 2014 - enclosed).

* The results depend – as usual – on some underlying hypotheses like the functioning of the formal and informal labour markets.

* You list a series of assumption about the functioning of the labour markets (the same as in Jales 2017 for Brazil?) for which the identification approach would be valid. Some of them look strong to me, and not necessarily transferable to all context. I give two examples: (4) “no spillover are allowed: workers with latent wages above the MW are not affected by the policy”; (7) “wages of informal workers below the MW are not affected by the possible inflow of formal workers”. Could you test them? It is crucial that you validate them to your own context, and that you discuss the expected implications of such assumptions (e.g., which are the implications of not considering the spill-over effects?)

* The main probability indicators as estimated by the adopted approach are presented. I would suggest also to briefly discuss the mechanisms driving them, as induced by the MW. For example, it would be useful to add some discussion on how what you call “probability of compliance in the informal domestic workers” works (workers remaining in the informal sector, but increasing their salary to the MW level). Also, as another possible outcome, could we have the probability of those moving from the informal to the formal sector due to a rise in the MW?

* Sector-specific wage density:

- Define better “c” parameter, and its implications

- A very short intuitive discussion of the formula would be useful. This is the key formula and should be introduced better and discussed in more details

- The formula you presented seems the general one: what if this is applied to the formal sector? And what if to the informal sector?

* Qualitative survey: very interesting but should be linked better to the quantitative component, otherwise, it risks to be as two disjoint components. E.g., does it serve to validate the quantitative results? Or the quantitative assumptions?

* How did you define the percentiles? Based on per capita income (net or gross?) at the base year?

* Figure 3: I suggest to normalise all series to 2006 values (so, all series would start from 100)

* Domestic work is a job with relatively flexible contracts, with very large variance in terms of worked hours. Would it then more interesting to study the effects on the labour supply at the intensive margins, rather than at the extensive margins?

* Figure 5 on the evolution of model parameters: their trends are often nonlinear. How would you explain it? Also, Figure 5 seems in contrast with Figure 5 (panel π_d^1).

* The final and ultimate objective should be, in my opinion, to assess the welfare effects. To do that, some more analyses should be implemented. For example, I am not sure you can assert that the MW policy reduced wage inequality in the formal sector, as it crucially depends on who remains in the sector and who are the most affected by the wage increase (e.g., those working half-time or full-time?)

* Final note: all results are hyper-significant – your sample is relatively big, but I am still a bit surprised about this very strong significance everywhere.

* You present some useful discussion on the expected results depending on the chosen approach (neoclassical/perfect labour markets-mobility VS imperfect labour markets-mobility). But which is the conceptual framework you adopt in the empirical setting? I would also suggest looking at the review of studies on the role of labour market frictions due to search costs and to cases where higher wages induce larger work efforts (Halmlund 2014 - enclosed).

* The results depend – as usual – on some underlying hypotheses like the functioning of the formal and informal labour markets.

* You list a series of assumption about the functioning of the labour markets (the same as in Jales 2017 for Brazil?) for which the identification approach would be valid. Some of them look strong to me, and not necessarily transferable to all context. I give two examples: (4) “no spillover are allowed: workers with latent wages above the MW are not affected by the policy”; (7) “wages of informal workers below the MW are not affected by the possible inflow of formal workers”. Could you test them? It is crucial that you validate them to your own context, and that you discuss the expected implications of such assumptions (e.g., which are the implications of not considering the spill-over effects?)

* The main probability indicators as estimated by the adopted approach are presented. I would suggest also to briefly discuss the mechanisms driving them, as induced by the MW. For example, it would be useful to add some discussion on how what you call “probability of compliance in the informal domestic workers” works (workers remaining in the informal sector, but increasing their salary to the MW level). Also, as another possible outcome, could we have the probability of those moving from the informal to the formal sector due to a rise in the MW?

* Sector-specific wage density:

- Define better “c” parameter, and its implications

- A very short intuitive discussion of the formula would be useful. This is the key formula and should be introduced better and discussed in more details

- The formula you presented seems the general one: what if this is applied to the formal sector? And what if to the informal sector?

* Qualitative survey: very interesting but should be linked better to the quantitative component, otherwise, it risks to be as two disjoint components. E.g., does it serve to validate the quantitative results? Or the quantitative assumptions?

* How did you define the percentiles? Based on per capita income (net or gross?) at the base year?

* Figure 3: I suggest to normalise all series to 2006 values (so, all series would start from 100)

* Domestic work is a job with relatively flexible contracts, with very large variance in terms of worked hours. Would it then more interesting to study the effects on the labour supply at the intensive margins, rather than at the extensive margins?

* Figure 5 on the evolution of model parameters: their trends are often nonlinear. How would you explain it? Also, Figure 5 seems in contrast with Figure 5 (panel π_d^1).

* The final and ultimate objective should be, in my opinion, to assess the welfare effects. To do that, some more analyses should be implemented. For example, I am not sure you can assert that the MW policy reduced wage inequality in the formal sector, as it crucially depends on who remains in the sector and who are the most affected by the wage increase (e.g., those working half-time or full-time?)

* Final note: all results are hyper-significant – your sample is relatively big, but I am still a bit surprised about this very strong significance everywhere.