Impact of Agricultural Drought Resilience on the Welfare of Smallholder Livestock Farming Households in the Northern Cape Province of South Africa

Reviewers 1:

1

On what basis you chose a sample size or target respondents/farmers as 207?

Thank you for the comment- the detailed explanation of chose a sample size explained in detail. Please refer to page 6 from line 207-217

2

You mentioned that you used compensation variation, resilience index and linear regression in analyzing the data, why you employed these tools to analyze your data?

We have provided a justification for the use of these methods in the revised version. Page 7-8  from lines 260-271 and  282-291

3

Please reduce the amount of text and number of references. I think the message in the manuscript is clear “drought relief support to the smallholder farmers improved their welfare, and the impact was more in males as compared to the females of the target community”.

This has been addressed

Reviewers 2

In general it is normal for a paper to have a ‘sign posting’ section at the end of the introduction setting out the structure of the rest of the paper.

The structure of the paper included at the end of the Introduction. Please refer to page 4 from lines 148 to 150. 

Conclusions section should be developed to highlight the unique contributions of the paper, limitations of the research and some future research directions.

This has been incorporated. Please refer page 17 lines  to 555-561

References must follow the style of the Journal

The references follow the journal style. Please refer reference list section.

Some references seem to have incomplete details such as the missing volume, issue and page numbers.

References completed by incorporating volume issue and the page number. Please refer reference list section. 

Reviewers 3

1

Lines 73- 118: Sub-section named 2.1 Theoretical framework

This sub-section is unclear and the theoretical framework adopted is confusing. The theoretical framework adopted should be deepened, proposed in detail and framed in the debate of the scientific literature.

In the current version of the document, the structure of the presentation follows an opposite direction. The section begins with the Sustainable Livelihood Framework (SLF) of the Department for International Development (DFID) summarized in Figure 1, ends with "(...) however, in this study, we classify agricultural capital for drought resilience into four capitals of resilience: human, social, natural and economic capital (...) ".

Furthermore, it should be noted that Figure 1 is not consistent with the text. The Capital Index for Agricultural Resilience to Drought (ADRI) is derived from the five types of capital and not vice versa. Also, if ADRI is a (synthetic) index it should come from other elementary indices, what are these indices? How are they measured? (and more, …)

This section has been revised. We have revised the structure and content of the framework in a consistent manner.

Refer to section 2.1

2

Lines 119 -132: Sub-section named 2.2 Data and sampling

Lines 120 -121 (and lines 130 – 132): “Primary data (…) using a structured questionnaire.”

The structure of the questionnaire must be explained. Special attention must be paid to the description of the variables linked to each kind of resilience capital. In the current version of the paper, the variables and their descriptive statistics are completely missing. Variables show a wide range of variability, so median values are more robust than their mean values. Median values should be added into a basic descriptive statistics table.

We have included the structure of the questionnaire and the content of the different subsections in the questionnaire and have highlighted examples of variables used in defining each of the capitals. Refer to section 2.2 of page 6, lines 205-230.

We have provided indicators used in computing the capitals.

3

Lines 121 -130: “The survey was conducted using a multi-stage sampling technique (…)” The steps and criteria adopted during the multi-stage sampling must be discussed in details. In the current version of the paper the adopted choices are missing. The descriptive statistics of the sample must be discussed in this sub-section.

The population is made by 868 farmers receiving governmental assistance. What is known about the population of all 868 farmers?

Is the governmental assistance equal to the drought relief? How is the drought relief calculated?

We have discussed the different steps followed in the multi-stage sampling.

Refer to section 2.2 lines 205-230.

We have revised this section. The sample frame consisting of 868 smallholder farmers identified by the government in 2015/2016 was used. We did not prepare another sample frame because of this existing framework. Please refer to section 2.2 lines 205-230.

We did not calculate drought relief; we only solicited information on the amount received by the smallholder farmers. 

4

Lines 135 -162: Sub-section named 2.4. Calculation of smallholder farmer household welfare

This sub-section is very unclear. Author/authors are invited to riconsider the global structure of the sub-section.

Special attention must be accorded to describe of the following topics unclear in the current version of the paper:

·         the problem (change in smallholder farmer household welfare by drought effects),

·         variation in the level of welfare and adopted estimation procedure,

·         hypothesis of the regression model.

This section has been revised. We have elaborated further on the section and sticked to consistent use of welfare gains. The hypothesis has been clarified as well. Refer to the revised text under section 2.4. in general and from line 260-271; 282-291 in particular

4

Lines 203 -162: Sub-section named 3.1. Socio-economic characteristics of the respondents

Although the paper is focused on the  smallholder livestock farming households, in this sub-section nothing is reported about the composition (adults, children),  employment of household members  on the farm (full-time family members) or employment of household members  outside the farm (part-time members), and level of the global extra-farm revenue of the household.

Household size and the number of household members that can assist on the farm are included in the computation of the human capital index. We have included results on main and secondary employment in Table 1 highlighting whether the respondent is full-time or part-time as well as a secondary occupation. The result section is amended accordingly. Please refer pages 8-9, from lines  303-305, and Table 1 from lines 310-311.

5

Lines 350 - 385: Empirical results of the impact of agricultural drought resilience on smallholder farmer welfare

Comments reported in the previous paragraphs (2.3 and 2.4) do not allow evaluation of the results. However, 2 general remarks can be made. The first observation is that the regression model using the  pooled sample (n = 207) includes two models, so the regression model for the two subsamples is redundant. The second remark refers to the regression models for the subsamples (males / females). It should be noted that in the Females regression model, the sign of the drought variable is negative, an unexpected  result which goes against policy recommendations made by the authors in the Discussions section.

.

We disagree with the first observation that the regression with the pooled data contains two models. Results were presented for a single model. On the second remark yes we observed that the drought relief variable for the females model was negative and this is a finding of interest and we have elaborated on that in the discussion section.

Reviewers 3

Rebuttal

1

Dear Authors, thank you for your efforts, unfortunately, I am dissatisfied with your revised paper. The reason is that the revised paper does not respond to any of the major critical observations related to the results of the multivariate analysis (PCA) and the econometric model. The revised version of the article does not provide the reader with the data to verify what the authors suggest. 

We appreciate the comments from Reviewer 3. However, we believe we have provided enough information on what the reviewer considers as major critical observations.

First, the argument raised about the PCA has been addressed. Specifically, in line236-240 of page 6 we indicated what the PCA was used for.

Secondly, we provided the variables used in computing the different capitals under section 3.3 (highlighted in green) so we disagree with the reviewer on the point that we have not provided data to readers for verification. In addition, we stated under the data availability section (page 17 lines 569-570) that the full data is available upon request.

To add to our response to the first point, we want to state that our paper is not a methodological paper where we teach how to perform PCA step-by-step and hence we believe the information provided is clear from our point of view.

2.

Furthermore, the redundancy of the estimated models and the incongruity of the results were not removed, as well as the incongruity between the results of the econometric model and the policy proposals. I am very sorry, but it discourages me to note that the authors completely ignore that only the model estimated on the complete sample provides statistically solid and consistent results with their analysis, thanks to the dummy variable (0; 1) associated with the gender of the interviewee.

Again as stated in the previous response, we totally disagree with this argument about the pooling and sub-samples. Our reasons are that pooled results and that of the sub-samples are all relevant. For instance in the pooled sample, one cannot precisely tell the magnitude of the effects of drought relief for instance on females. However, in the separate models, these are clearly seen and provided.

In addition, the results of the sub-samples also serve as a robustness check for the pooled results. Before we estimated the results for the sub-samples, our LR-test proved that it is feasible to examine the results for the sub-samples.

The arguments raised by the reviewer would have held if the results of the subs-samples contradict that of the pooled results. In our case, it does not contradict it but rather adds more insight to the pooled results in terms of specific magnitudes and direction for instance the negative effect observed for the females model.