Losing the Plot: The Impact of Urban Agriculture on Household Food Expenditure and Dietary Diversity in Sub-Saharan African Countries

Round 1

Reviewer 1 Report

This paper intends to contribute to the existing body of knowledge by performing cross-country analyses using household-level data from nine sub-Saharan countries. The paper is in general well written and exhibits a good command of econometric modeling to provide scientific evidence of the effects of urban agriculture. However, I have a couple of comments to make to help the authors improve the readability and contribution of the paper.

Introduction

The departure of this study from the existing literature is its cross-country analysis, which is clearly stated in lines 73-75. However, the time span for the nine sub-Saharan African countries varies with countries: Burkina Faso (2014), Ethiopia (2013), Ghana (2009), Malawi (2013), Niger 2014, Nigeria (2012), Tanzania (2010) and Uganda (2013). Inconsistency in the time span for the nine countries may bias the cross-country estimates of the average treatment effect of urban agriculture due to the differences in environmental factors, for example, climate conditions and national economy performance, which can significantly affect the precision of the estimation. Therefore, I would suggest the authors to perform a robustness check by excluding Ghana (2009), Nigeria (2012) and Tanzania (2010) from the data.

Methodology

Please cite the references you followed while elaborating on the treatment-effect model.

There are two types of quantile treatment effects, one is conditional quantile treatment effect and the other one is unconditional quantile treatment effect (QTE). Please add a paragraph to state the advantages/disadvantages of the two models and why you choose to estimate the unconditional quantile treatment effect (QTE).

References

Please update your references with more recent papers which are published within five years from the present.

Minor comment

(Line 268) Please provide the full name of ATET at its first appearance.

Author Response

Comment: This paper intends to contribute to the existing body of knowledge by performing cross-country analyses using household-level data from nine sub-Saharan countries. The paper is in general well written and exhibits a good command of econometric modeling to provide scientific evidence of the effects of urban agriculture. However, I have a couple of comments to make to help the authors improve the readability and contribution of the paper.

Response: Many thanks for acknowledging the strengths of the manuscript.

Comment: The departure of this study from the existing literature is its cross-country analysis, which is clearly stated in lines 73-75. However, the time span for the nine sub-Saharan African countries varies with countries: Burkina Faso (2014), Ethiopia (2013), Ghana (2009), Malawi (2013), Niger 2014, Nigeria (2012), Tanzania (2010) and Uganda (2013). Inconsistency in the time span for the nine countries may bias the cross-country estimates of the average treatment effect of urban agriculture due to the differences in environmental factors, for example, climate conditions and national economy performance, which can significantly affect the precision of the estimation. Therefore, I would suggest the authors to perform a robustness check by excluding Ghana (2009), Nigeria (2012) and Tanzania (2010) from the data.

Response: We thank the reviewer for this suggestion. We performed a robustness check by excluding Ghana (2009), Nigeria (2012) and Tanzania (2010) from the data. Please see the table below which shows results very similar to the one we previously presented.

We did not add this table to the manuscript, but we did mention this further robustness check in a footnote, which says: ‘To check the robustness of our results, we excluded the data from Ghana (2009), Nigeria (2012) and Tanzania (2010) to make our data consistent with the time span. We ran the same model to find that the coefficients are very similar to our main results.’

Comment: Please cite the references you followed while elaborating on the treatment-effect model.

Response: We had added the relevant references in the manuscript (just before formula (3)). The reference is the following:

Manda, J., Gardebroek, C., Kuntashula, E., & Alene, A. D. (2018). Impact of improved maize varieties on food security in Eastern Zambia: A doubly robust analysis. Review of Development Economics, 22(4), 1709-1728.

Comment: There are two types of quantile treatment effects, one is conditional quantile treatment effect and the other one is unconditional quantile treatment effect (QTE). Please add a paragraph to state the advantages/disadvantages of the two models and why you choose to estimate the unconditional quantile treatment effect (QTE).

Response: The following information has been added to explain unconditional QTE and the rationale to prefer it over conditional QTE

‘Unconditional quantile treatment effect (UQTE) is a measure of the average treatment effect on a specific quantile of the outcome distribution that is not conditioned on any covariates. Unlike the conditional quantile treatment effect (CQTE) which estimates the treatment effect at a specific quantile while taking into account the values of one or more covariates, UQTEs estimate the treatment effect for the entire population. UQTEs are less sensitive to the choice of covariates used in the estimation than CQTEs. Since UQTEs do not depend on any specific set of covariates, they can provide a more robust estimate of the treatment effect. These effects are of particular interest in policy evaluations as they are simple to interpret and can be easily conveyed and summarised (Frölich & Melly, 2013).’

Comment: Please update your references with more recent papers which are published within five years from the present.

Response: The reviewer has a good point here; as urban agriculture is a fast changing phenomenon, it is recommendable to discuss newer publications along with more dated ones. We have added few more recent publications to our discussion (see the list below); please keep in mind that only 2 publications in our reference list are 20 years older or more; the rest of the publications we cite have been published in the last 20 years, with the majority in the last ten years.   

References:

Azunre, G. A., Amponsah, O., Peprah, C., Takyi, S. A., & Braimah, I. (2019). A review of the role of urban agriculture in the sustainable city discourse. Cities, 93, 104-119.

Chari, F., & Ngcamu, B. S. (2022). Climate change and its impact on urban agriculture in Sub-Saharan Africa: A literature review. Environmental & Socio-economic Studies, 10(3), 22-32.

Cilliers, E. J. (2020). Reflecting on green infrastructure and spatial planning in Africa: the complexities, perceptions, and way forward. Sustainability 37, 105–129.

de Oliveira Alves, D., & de Oliveira, L. (2022). Commercial urban agriculture: A review for sustainable development. Sustainable Cities and Society, 104185.

Jagganath, G. (2021). The Transforming City: Exploring the Potential for Smart Cities and Urban Agriculture in Africa. The Oriental Anthropologist, 0972558X211057162.

Langemeyer, J., Madrid-Lopez, C., Beltran, A. M., & Mendez, G. V. (2021). Urban agriculture—A necessary pathway towards urban resilience and global sustainability?. Landscape and Urban Planning, 210, 104055.

Tasciotti, L., & Wagner, N. (2018). How much should we trust micro-data? A comparison of the socio-demographic profile of Malawian households using census, LSMS and DHS data. The European Journal of Development Research, 30(4), 588-612.

Comment: (Line 268) Please provide the full name of ATET at its first appearance.

Response: This has been corrected and the acronym for ATET (average treatment on the treated) now appears in the methodology section.

Reviewer 2 Report

check my comments

Comments for author File: Comments.pdf

Author Response

Comment: Figure 1 should be placed below the figure.

Response: We moved the titles of the 3 figures in the manuscript and positioned them after the figures.

Comment: Explanation for the x and y-axis are missing from Figure 1.

Response: Figure 1 has been updated and it has now indicate that numbers in the x-axis are percentages and the five histograms refer to the 5 wealth quintiles.

Comment: Please give more explanation about the content of Table 3.

Response: We added the following text to the paragraph describing the statistics in Table 3: ‘Those households non engaging in UA activities tend to have higher scores for the food group and food count categories, albeit the difference being very marginal. The differ-ences between UA and non UA households in terms of food count in the 12 categories here used are very negligible.’

Comment: Define what ATET stands for.

Response: This has been corrected and the acronym for ATET (average treatment on the treated) now appears in the methodology section.

Comment: Define what IPWRA stands for.

Response: The definition of IPWRA (an inverse-probability weighted regression adjustment) is now present in the paper.

Reviewer 3 Report

The only suggestion is to provide a hypothesis: why the quantile treatment effect of the UA is so different for the countries studied (fig. 3). Does it depend on the average level of social and economic development, the size of the country, or any other factor.

Author Response

Comment: The only suggestion is to provide a hypothesis: why the quantile treatment effect of the UA is so different for the countries studied (fig. 3). Does it depend on the average level of social and economic development, the size of the country, or any other factor.

Response: The reviewer makes an intriguing observation about the difference in quantile treatment effect of the UA across different nations. Without further information, it is difficult to say what could be causing this variation, as it could be a result of many factors such as climate, culture, attitudes, economic and social development. However, it's important to note that this would be pure speculation at this stage. We did attempt to find some sort of correlation by examining other variables such as the average plot size, number of allotments, number of crops being grown, soil quality in terms of carbon and pH value, but no consistent pattern was found. Which is the reason why we offer the readers only a description of what we observe.

Round 2

Reviewer 1 Report

The authors have addressed my comments as stated in the previous round of review. I do not have any further comments to add. Congrats!