The Effect of Access to Information and Communication Technology on Household Labor Income: Evidence from One Laptop Per Child in Uruguay

Round  1

Reviewer 1 Report

In my opinion, the paper is interesting and provide useful and important results.

However, the paper could be improved, at least, in two major points:

(i) I would like to see some theoretical background for the hypothesis under study. In the introduction, the author affirms "thus, economic theory and the literature suggest that  a positive impact on household labor income can be expected." However, the "economic theory" is not provided.

(ii) The final section of the paper could provide a deeper discussion on the main implications of the paper.

To sum up, in my perspective the paper is intersting but could be improved along the lines presented above.

Author Response

We tried our best to incorporate the comments/suggestions of you and other anonymous referees. We are grateful to you and other reviewers for the insights to strengthen our paper.

I. Brief overview of the significant changes made in this revised submission:

a.       As suggested by one of the anonymous referees, we have expanded the theoretical background in order to make a stronger case for the proposed earnings function used in our empirical analysis.

b.      To avoid unnecessary confusion, our current Figure 2 displays monthly average per capita income in urban households in current pesos for both departments.

c.       The selection of households from the 2006 data set has been substituted by Propensity Score Matching (PSM). Households in the 2009 data set were matched to a selection of households in the 2006 data set based on the treatment variable (laptop) and four other critical covariates: average education, average age, rural and the number of household members. We used the teffects psmatch Stata command. Accordingly, our sample sizes are slightly different from our initial submission.

d.      As one of the anonymous referees suggested, we have also employed a quantile regression model to strengthen our empirics. This enabled us to capture the effects at different quantiles of household income distribution and further constitutes a robustness check for our definition of “low income”.

II. Our specific responses to your comments:

1.      We have expanded the theoretical background in order to make a stronger case for the proposed earnings function used in our empirical analysis.

-          Please refer to our expanded methodology section.

2.      We have deepened the discussion on the relevance of our findings in relation to the growing importance of technology in economic activities and the fact that other countries in South America have begun to implement similar programs.

-          Please refer to our revamped conclusion and policy implications section.

Author Response File: Author Response.docx

Reviewer 2 Report

I think this paper is well written. However, I have one question. In this paper, lower-income households are defined as the households below median income. Is it robust to the results? For example, if you change the definition of lower-income households to lower 10% household, does this result hold? I recommend to do the robustness check. 

Author Response

We tried our best to incorporate the comments/suggestions of you and other anonymous referees. We are grateful to you and other reviewers for the insights to strengthen our paper.

I. Brief overview of the significant changes made in this revised submission:

a.       As suggested by one of the anonymous referees, we have expanded the theoretical background in order to make a stronger case for the proposed earnings function used in our empirical analysis.

b.      To avoid unnecessary confusion, our current Figure 2 displays monthly average per capita income in urban households in current pesos for both departments.

c.       The selection of households from the 2006 data set has been substituted by Propensity Score Matching (PSM). Households in the 2009 data set were matched to a selection of households in the 2006 data set based on the treatment variable (laptop) and four other critical covariates: average education, average age, rural and the number of household members. We used the teffects psmatch Stata command. Accordingly, our sample sizes are slightly different from our initial submission.

d.      As one of the anonymous referees suggested, we have also employed a quantile regression model to strengthen our empirics. This enabled us to capture the effects at different quantiles of household income distribution and further constitutes a robustness check for our definition of “low income”.

II. Our specific response to your comments:

We have employed a quantile regression model, which contributes to a robustness check for our definition of “low income” (households below median income).

-          Please refer to the new set of results presented in Table 3 (based on a quantile regression model) and the additional results presented in Appendix B.

Author Response File: Author Response.docx

Reviewer 3 Report

- Figure 2: It is difficult to compare the performances of the gdp in the two regions in absolute terms because of the different size of the two economies. It would be better to refer to per capita values or growth rates.

- The Household Survey is continuos: why don't use the data of all the years and then perform a panel analysis? 

- In order to analyze the effect of "laptop" on median income, the sample is artificially selected (only household under the median income). Why don't perform quantile regressions on all the data? By this way, the effect on low incomes should arise spontaneously.

Author Response

We tried our best to incorporate the comments/suggestions of you and other anonymous referees. We are grateful to you and other reviewers for the insights to strengthen our paper.

I. Brief overview of the significant changes made in this revised submission:

a.       As suggested by one of the anonymous referees, we have expanded the theoretical background in order to make a stronger case for the proposed earnings function used in our empirical analysis.

b.      To avoid unnecessary confusion, our current Figure 2 displays monthly average per capita income in urban households in current pesos for both departments.

c.       The selection of households from the 2006 data set has been substituted by Propensity Score Matching (PSM). Households in the 2009 data set were matched to a selection of households in the 2006 data set based on the treatment variable (laptop) and four other critical covariates: average education, average age, rural and the number of household members. We used the teffects psmatch Stata command. Accordingly, our sample sizes are slightly different from our initial submission.

d.      As one of the anonymous referees suggested, we have also employed a quantile regression model to strengthen our empirics. This enabled us to capture the effects at different quantiles of household income distribution and further constitutes a robustness check for our definition of “low income”.

II. Our specific response to your comments:

1.      To avoid unnecessary confusion, our current Figure 2 displays monthly average per capita income in urban households in current pesos for both departments.

2.      We have also employed a quantile regression model to strengthen our empirics, which, as you suggested, enabled us to capture the effects at different quantiles of household income distribution.

-          Please refer to the new set of results presented in Table 3 (based on a quantile regression model) and the additional results presented in Appendix B.