COVID-19 Lockdown and Implications for Household Food Security in Zambia: Quality of Diet or Economic Vulnerability?
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
I would like to congratulate the authors on a well-done paper, which I read with great interest.
My overall assessment of the paper is that it does a very good job of linking two food security measures to the observed characteristics of households. The authors' innovative use of the coarsened exact matching between the surveys allows them to infer a greater level of causality in their results.
Here are my key questions that came up in my mind after reading the paper that I would like to encourage the authors to consider:
(1) Diversity of the diet HDDS is improving through the pandemic, and the authors express their surprise at this finding as anyone would. What if the improved diversity is the results of the households having to be more creative to survive, essentially changing their diets in the face of difficult macroeconomic times and food availability. I am not convinced that this is the case based on the results that you show, but perhaps you could dig a bit deeper to try to explain this surprising result
(2) How about households producing their own food? For them the food share is not as meaningful to measure their vulnerability--what happened to those households? Could own production be part of the improved diversity of the diets?
(3) For k-to-k match, the use of source of income can be questionable as it could have changed, especially since it is obvious that incomes declined as food shares increased between the periods. Is there no other variable that could be less impacted by the dramatic change that would allow you to match households? Maybe there are some assets that are hard to sell/buy that could give you a measure of the household's income/asset index
(4) In table 4, complete loss of income appears to increase the probability to fall in low-food insecurity. Really, why is that?
Smaller points:
(1) "The results show that in 2015, the probability of being very food insecure in female headed households was 3.1% lower than in male headed households." Is it really 3.1 percent of 3.1 percentage points? If so, please fix everywhere
(2) You write: Prior to the 35 COVID-19 pandemic, the 2019 Global Hunger Index (GHI), which assesses undernourishment, child wasting, child stunting, and child mortality ranked Zambia among the six countries with alarming levels of hunger, with a score of 38.1. This score was higher than those of countries in conflict or post-conflict situations such as Yemen (39.7), Chad (45.4), Central African Republic (53.7), and only better than Madagascar (38.0)--I think 38.1 is not higher than 39.7--perhaps you can talk about "worse" score?
Author Response
Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions in track changes in the re-submitted files.
Comment 1: Diversity of the diet HDDS is improving through the pandemic, and the authors express their surprise at this finding as anyone would. What if the improved diversity is the results of the households having to be more creative to survive, essentially changing their diets in the face of difficult macroeconomic times and food availability. I am not convinced that this is the case based on the results that you show, but perhaps you could dig a bit deeper to try to explain this surprising result
Response 1: We agree this is a plausible explanation and have now elaborated on it. Households may have adapted to shortages or price hikes by diversifying their diets with cheaper, locally available food groups.
Comment 2: How about households producing their own food? For them the food share is not as meaningful to measure their vulnerability--what happened to those households? Could own production be part of the improved diversity of the diets?
Response 2: We have included this concern as a limitation of the study as the disaggregated analysis has not been undertaken due to data limitations.
Comment 3: For k-to-k match, the use of source of income can be questionable as it could have changed, especially since it is obvious that incomes declined as food shares increased between the periods. Is there no other variable that could be less impacted by the dramatic change that would allow you to match households? Maybe there are some assets that are hard to sell/buy that could give you a measure of the household's income/asset index
Response 3: As suggested, we have used asset-based indicators as proxies for income and made the necessary adjustments in the paper.
Comment 4: In table 4, complete loss of income appears to increase the probability to fall in low-food insecurity. Really, why is that?
Response 4: We have reviewed the section and provided plausible explanations.
Comment 5: "The results show that in 2015, the probability of being very food insecure in female headed households was 3.1% lower than in male headed households." Is it really 3.1 percent of 3.1 percentage points? If so, please fix everywhere
Response 5: Agreed, we have corrected all to “percentage points”
Comment 6: Prior to the 35 COVID-19 pandemic, the 2019 Global Hunger Index (GHI), which assesses undernourishment, child wasting, child stunting, and child mortality ranked Zambia among the six countries with alarming levels of hunger, with a score of 38.1. This score was higher than those of countries in conflict or post-conflict situations such as Yemen (39.7), Chad (45.4), Central African Republic (53.7), and only better than Madagascar (38.0)--I think 38.1 is not higher than 39.7--perhaps you can talk about "worse" score?
Response 6: Thank you. We have revised the sentence to reflect this appropriately.
Reviewer 2 Report
Comments and Suggestions for Authors
Abstract. The abstract mentions two food security indicators but does not explain why these specific metrics were chosen. A brief justification for their relevance is missing. Add a short explanation of why household expenditure share and dietary diversity score are critical for assessing food security.
Introduction. The novelty of the study is not clearly articulated. While previous research is cited, the unique contribution of this work needs emphasis. Explicitly state how this study fills gaps in existing literature.
Methodology. The Coarsened Exact Matching (CEM) method has several limitations, such as difficulties with large datasets (the 'curse of dimensionality'), data loss, result instability, and variable sensitivity. Therefore, a clearer explanation is needed as to why this method was chosen as the primary approach. How was the problem of missing data addressed? Add a subsection on the limitations of the method and the steps taken to minimize their impact on the results.
The authors note that while CEM is widely used in the literature, caution is required when using balancing covariates to avoid yielding erratic and spurious results." It is necessary to clarify how the authors avoided this in their work.
Results. Tables 3 and 4 present marginal effects, but the included variables and their interactions are unclear, complicating interpretation. Clarify in captions or text which variables were controlled for and whether interaction terms were tested.
Discussion. The paradoxical improvement in dietary diversity amid rising economic vulnerability lacks exploration. Potential explanations (e.g., reallocation of spending, aid programs) are omitted. Discuss plausible reasons and their implications.
Conclusion. Policy recommendations are overly generic (e.g., "strengthen social protection"). Targeted actions are missing. Specify measures based on findings.
References. Citations (e.g., Aaron et al., 2021) lack complete details (e.g., page numbers, DOIs). Ensure all references follow the required format and include full metadata.
Appendices. Appendix tables (e.g., A3–A6) are disconnected from the main text, leaving their purpose unclear. Reference appendices where relevant (e.g., "See Appendix A3 for matching diagnostics") and explain their role.
Comments on the Quality of English Language
The English language quality of this article is good overall, but there are areas where improvements could enhance clarity, conciseness, and academic tone.
Recommendations: shorten wordy phrases, replace ambiguous pronouns, define abbreviations consistently, reduce passive voice where possible, proofread for prepositions/articles.
Author Response
Reviewer 2
We have revised the introduction to more clearly articulate the study’s novelty and contribution to the existing literature. Specifically, we now emphasise how this study addresses gaps related to large-scale, nationally representative assessments of food security in Zambia using a matched design pre- and during the COVID-19 pandemic. We have expanded the justification for using the Coarsened Exact Matching (CEM) approach despite its known limitations.We have substantially revised the conclusion to better reflect and synthesise the study’s empirical findings.
Comment 1: Abstract. The abstract mentions two food security indicators but does not explain why these specific metrics were chosen. A brief justification for their relevance is missing. Add a short explanation of why household expenditure share and dietary diversity score are critical for assessing food security.
Response 1: We have now included a brief rationale in the abstract explaining why these indicators were chosen—economic vulnerability and diet quality represent two complementary dimensions of food security.
Comment 2: Introduction. The novelty of the study is not clearly articulated. While previous research is cited, the unique contribution of this work needs emphasis. Explicitly state how this study fills gaps in existing literature.
Response 2: We now emphasise how this study’s use of nationally representative, matched datasets and simultaneous analysis of both food quantity and quality distinguishes it from prior work.
Comment 3:Methodology. The Coarsened Exact Matching (CEM) method has several limitations, such as difficulties with large datasets (the 'curse of dimensionality'), data loss, result instability, and variable sensitivity. Therefore, a clearer explanation is needed as to why this method was chosen as the primary approach. How was the problem of missing data addressed? Add a subsection on the limitations of the method and the steps taken to minimize their impact on the results. The authors note that while CEM is widely used in the literature, caution is required when using balancing covariates to avoid yielding erratic and spurious results." It is necessary to clarify how the authors avoided this in their work.
Response 3: We’ve added a subsection on the limitations of CEM and our steps to reduce potential biases, including sensitivity checks and balance diagnostics.
Comment 4: Results. Tables 3 and 4 present marginal effects, but the included variables and their interactions are unclear, complicating interpretation. Clarify in captions or text which variables were controlled for and whether interaction terms were tested.
Response 4: We have revised the table captions and added more detail in the text about the variables included and the interaction terms.
Comment 5: Discussion. The paradoxical improvement in dietary diversity amid rising economic vulnerability lacks exploration. Potential explanations (e.g., reallocation of spending, aid programs) are omitted. Discuss plausible reasons and their implications.
Response 5: We agree this is a plausible explanation and have now elaborated on it. Households may have adapted to shortages or price hikes by diversifying their diets with cheaper, locally available food groups. This adaptive behaviour may reflect resilience through dietary creativity in the face of constrained purchasing power.
Comment 6: Conclusion. Policy recommendations are overly generic (e.g., "strengthen social protection"). Targeted actions are missing. Specify measures based on findings.
Response 6: We now recommend more targeted interventions, including supporting informal food markets, enhancing rural-urban linkages, and expanding social protection to self-producing households.
Comment 7: References. Citations (e.g., Aaron et al., 2021) lack complete details (e.g., page numbers, DOIs). Ensure all references follow the required format and include full metadata.
Response 7: All references have been checked and updated, as far as possible, to conform to MDPI style.
Comment 8: Appendices. Appendix tables (e.g., A3–A6) are disconnected from the main text, leaving their purpose unclear. Reference appendices where relevant (e.g., "See Appendix A3 for matching diagnostics") and explain their role.
Response 8: We now refer to appendices explicitly throughout the results section
Comments on the Quality of English Language
The English language quality of this article is good overall, but there are areas where improvements could enhance clarity, conciseness, and academic tone. Recommendations: shorten wordy phrases, replace ambiguous pronouns, define abbreviations consistently, reduce passive voice where possible, proofread for prepositions/articles.
Response: We have thoroughly reviewed the entire manuscript and made numerous revisions to improve clarity, conciseness, and the overall academic tone.
Round 2
Reviewer 2 Report
Comments and Suggestions for Authors
The authors did a good job, but I have a few more comments.
- The literature review is comprehensive but would benefit from incorporating more recent sources 2023–2024 studies to ensure relevance.
- It is recommended to formulate a research hypothesis and identify which cause-effect relationships are being tested.
- The description of the multinomial logistic regression is overly technical. Additional explanations for readers unfamiliar with the method would enhance accessibility.
- The handling of missing data (e.g., deletion or imputation) is not specified, which could affect the reproducibility of results.
- Tables are well structured, but some figures require further clarification (e.g., why income loss reduces the likelihood of high vulnerability).
- The observed improvement in dietary diversity despite increased economic vulnerability warrants deeper analysis, as the current discussion does not fully explore potential mechanisms.
- The findings on the diminished protective effects of education and urban residence are compelling but would benefit from a more nuanced interpretation (e.g., linking them to shifts in employment patterns).
- Policy recommendations are sound but somewhat generic. Concrete examples (e.g., specific social protection programs) would strengthen their applicability.
- Suggestions for future research—such as examining long-term pandemic effects on food security—are notably absent.
Author Response
- The literature review is comprehensive but would benefit from incorporating more recent sources 2023–2024 studies to ensure relevance.
Response: We have revised the literature review to incorporate more recent peer-reviewed sources published in 2023 and early 2024 that examine the food security impacts of COVID-19 and associated lockdowns.
- It is recommended to formulate a research hypothesis and identify which cause-effect relationships are being tested.
Response: We have revised the introduction to include the research hypotheses.
- The description of the multinomial logistic regression is overly technical. Additional explanations for readers unfamiliar with the method would enhance accessibility.
Response: We have revised Section 2.2.3 to simplify the language around the multinomial logistic regression model
- The handling of missing data (e.g., deletion or imputation) is not specified, which could affect the reproducibility of results.
Response: An explanation of the handling of missing data has now been added to Section 2.2.1.
- Tables are well structured, but some figures require further clarification (e.g., why income loss reduces the likelihood of high vulnerability).
Response: We have added further clarification regarding the unexpected finding that reduced or lost income was associated with a lower probability of high vulnerability.
- The observed improvement in dietary diversity despite increased economic vulnerability warrants deeper analysis, as the current discussion does not fully explore potential mechanisms.
Response: We have expanded our discussion to further examine potential mechanisms
- The findings on the diminished protective effects of education and urban residence are compelling but would benefit from a more nuanced interpretation (e.g., linking them to shifts in employment patterns).
Response: We have strengthened the interpretation by linking the diminished protective effects of education and urban residence to structural shifts in employment and the differential impacts of lockdowns across sectors.
- Policy recommendations are sound but somewhat generic. Concrete examples (e.g., specific social protection programs) would strengthen their applicability.
Response: We have revised Section 5 to include concrete policy examples, such as Zambia’s existing social cash transfer programme, and the potential for emergency top-ups during crises.
- Suggestions for future research—such as examining long-term pandemic effects on food security—are notably absent.
Response: We have added a paragraph in the conclusion recommending future research to explore long-term impacts of the COVID-19 pandemic on household food security, particularly in relation to chronic poverty, intergenerational nutrition outcomes, and changes in food systems resilience.