Response letter to the Editor for the article entitled as "Gendered Economic Policy Making: The Case of Public Expenditures on Family Allowances"

1. The main concern is the arbitrariness of the 30% critical mass threshold and the absence of a formal test of this a priori unknown threshold. Looking at your reply to referee 2, my view is that the first solution you propose is the most sensible, and should be incorporated into the analysis in the paper, possibly in a separate section that precedes the critical mass analysis. Tables I and II in your reply to referee 2 could easily be replicated in the main analysis, though you should widen the scope a little, and also explore higher ranges of the possible threshold (in the 40%). This would address some of the concern of Referee 2, and also partly address Referee 1's comments. I am not convinced that your proposed solutions #2 and #3 are as suitable as your solution #1, #2 is descriptive, and as to #3, it seems highly implausible that the point of discontinuity is exogenous, which is required for an RDD design to be meaningful. Having said that, both reports as well as your replies will remain on the website, which means that you can refer interested readers to those additional results without having to necessarily replicate them in the paper.


The main concern is the arbitrariness of the 30% critical mass threshold and the absence of a formal test of this -a priori unknown -threshold. Looking at your reply to referee 2, my view is that the first solution you propose is the most sensible, and should be incorporated into the analysis in the paper, possibly in a separate section that precedes the critical mass analysis. Tables I and II in your reply to referee 2 could easily be replicated in the main analysis, though you should widen the scope a little, and also explore higher ranges of the possible threshold (in the 40%). This would address some
of the concern of Referee 2, and also partly address Referee 1's comments. I am not convinced that your proposed solutions #2 and #3 are as suitable as your solution #1, #2 is descriptive, and as to #3, it seems highly implausible that the point of discontinuity is exogenous, which is required for an RDD design to be meaningful. Having said that, both reports as well as your replies will remain on the website, which means that you can refer interested readers to those additional results without having to necessarily replicate them in the paper.
As the second referee and the editor suggested, I included a separate section on structural break analysis to formally test the critical mass argument. With respect to organization of the paper, I mentioned about this analysis at page 12 which is prior to presentation of main results but the explanation of the test is incorporated as an appendix (page 18-19) since it presents theoretical information about the test and would take too much place if I had added before the main results. Unfortunately higher ranges of thresholds bigger than 33% could not be explored due to the insufficient number of observations related to percentage share of female parliamentarians since there are only 4 countries (out of all OECD countries) where the percentage share of female parliamentarians pass 33 per cent over years analyzed. To explore the higher ranges could have produced biased results. At this point, addressing also to the Referee 1's comment, I presented the share of female parliamentarians for selected years (in Table 13) to see how many countries and at what point in time the share of female representatives exceeds the 30% threshold.

Related to the first comment of Referee 1, you should include some discussion + results on lagged female participation. Again, remember that you don't need to replicate all tables contained in your response to Referee 1, as you can reference those in the final version (your response to Referee 1 will remain posted on the website).
As Referee 1 and the editor suggested, I added discussion (page 14-section of Empirical Robustness) and results (Table 7-page 26) on lagged female participation. Although I replicated all the estimations using 2 and 3 years intervals, as editor indicated I did not present all the results except the ones especially related with the significance of over 30% threshold as the main finding of the paper.

Last but not least, I also would like to see a revision that pays close attention to the comments of both referees, as well as those by the reader, on the manuscript's exposition and organization. I agree with those comments in that a more structured exposition, with
fewer typos, will greatly improve readability.
Here are some suggestions:

-If you decide to add sections to incorporate the issues laid out above, it'd be advisable to shorten your exposition elsewhere. Section 3.1. is a good candidate, since you are only using it to build up to your main argument. I think you could tighten this section considerably, without losing much.
Following both referees and the editor's suggestion, I reorganized the parts related to data description, empirical strategy, results and robustness checks. Following the comment of Referee 2, data and descriptive statistics, baseline empirical strategy as well as baseline results and robustness checks are divided into three separate sections (Section 3, 4 and 5).
Moreover, following the comment of Referee 1, the problem with respect to duplicated explanations of economic problems is solved deleting duplications. These revisions serve also for the suggestion of editor in relation to tightening the part mentioned previously in Section 3.

-You may want to include your new results on the coalition versus majority governments, as an extension of the main idea.
New discussion on this issue is added into paper at page 16 as well as related results are presented in Table 9.

-Be careful about/avoid causality claims. Your design does not allow for a strong causal interpretation. This does not imply, however, that the results are not interesting in their own right.
I avoid to do causality claims and specifically mention in page 14 that the paper does not claim any causal relationship between related variables due to the possible problems such as endogeneity.
-Have the paper proofread by someone who is a native English speaker.
It is done. ***In addition to the specific suggestions of editor mentioned above, as she also emphasized that I pay close attention to the other comments of referees and the reader as c-The sentence that take place in the Comment 7 of Referee 2 is removed from the paper since it is found unclear by the referee. d-Furthermore, considering the 3 rd comment of Referee 2, a discussion and results on controlling ideology of governing party as well as posts occupied by left wing parties are added into the paper. (Related Discussion can be found in Section 5 and the table that shows the related results is Table 10). f-Furthermore, considering the 3 rd comment of Referee 1, I added the missing information about the "country specific time trends" to the note of Table 2. g-Moreover, pooled-OLS estimations are removed from the tables as Referee 1 suggested in his second comment. In the same comment, Referee 1 also suggested reducing the size of tables by using one large sample instead of three different samples. However, I believe that the idea of using three samples is important to understand whether only traditional OECD countries, which have higher share of female parliamentarians, drive the positive relationship between female political representation and public spending on family allowances. More explanation for this issue can be found in my previous letter to the Referee 1.