Who Is Afraid of Biotic Threats? An Econometric Analysis of Veneto Wine Grape Farmers’ Propensity to Insure

Round 1

Reviewer 1 Report

Referee Report on “Who’s Afraid of Biotic Threats? An Econometric Analysis of Veneto Wine Grapes Farmers’ Propensity to Insure”_agriculture-863291-peer-review-v1

The paper aims at understanding what affect farmers’ choice to ensure by using a probit model with endogenous variables with ML routine. They show the farmers’ socio-economic characteristics are treated as endogenous variables, which exist/are pre-determined before the choice to insure (or not) against biotic threats. They also explain the reluctance of farmers to swap to organic agricultural practices, as a policy suggestion based on the economic and econometrics results. However, I have the following specific concerns.

Major Concerns and Comments:

1. The questionnaire used by the author should be carefully discussed and detail how the variables are measured. Authors should carefully provide descriptive statistics for the main variables. It is suggested that the author can supplement the explanation.
2. The most important variable in this study is risk attitude, and the author should clearly explain how this variable is measured.
3. The author believes that risk attitudes are endogenous and provide empirical results. However, I suggest that the author should compare whether the risk attitude is endogenous and the risk attitude is exogenous, whether the two models have different effects on insurance choices.
4. It is recommended to include the calculation of marginal effect in Table1 to understand the impact of risk attitude on the probability of insurance behavior.

Minor Concerns

1. English spelling and grammar need to be checked. For example, on line 35 and line 89, the periods are missing.

Evaluation:

For the above reasons, I suggest that this article can be published in this journal after making some appropriate corrections.

Author Response

We thank the referee for the insightful points, which have helped us to improve the previous version.

We have addressed all points. In particular: 

1. We have totally rewritten the data section, with bibliographic references to the dataset and performance of descriptive statistics. Please see section 2.3 of the revised version of the paper
2. In the data section, please see the table 2 descriptive statistics
3. We believe that risk attitude is endogenous.  The rationale of our modelling strategy, is a “technical” requirement, and implies treating risk aversion as endogenous. it is not a choice to be tested, but “driven” by the content and nature of such variable. As explained in the paper, we assume that age, experience and socio-economic covariates are predetermined with regard to the choice to insure and are predetermined with regard to the risk profile of the farmer (that may change across time). Therefore, it is technically proper  to treat them differently. Typically, this line of reasoning is applied in the models that explain wage formation. Education affects wage level. Research tells us that the worker’s parents/grandparents education affects his current education level, and therefore wage level. However, those variables have to be treated very differently, since one is endogenous and the other is exogenous (see Verbeek 2000).
4. With the computation of marginal effects, we may agree with the reviewer if the risk attitude would be expressed as a continuous variable. In this case risk aversion is a qualitative variable (0/1) and the marginal effect cannot be interpreted correctly. In other words, we cannot say that the X% of extra risk aversion implies X% of increased probability to insure, having a dummy variable. The estimated marginal effects have no sense.
5. The paper has been partially rewritten and English grammar and language has been checked. 

Reviewer 2 Report

The subject of the article is interesting and it is linked to the objectives of the journal, however, there are a number of issues that have to be reconsidered.

For a better visibility on databases, the authors are asked not to repeat among keyword the words/concepts included on the title of the article.

How the sample of 1187 wine grapes’ farmers was selected and why it is representativ for the enrire population cand be deeper explained.

As there are a lor of information, the REsults part can be enlarged and compared with similar research.

Author Response

We thank the referee for the insightful points, which have helped us to improve the previous version.

We have addressed all points. In particular: 

1. We have totally rewritten the data section, with bibliographic references to the dataset and performance of descriptive statistics. Please see section 2.3 of the revised version of the paper
2. We have explained how the sample was selected and why it is representative for the enrire population. See section 2.3.
3. The results part was enlarged and compared with similar research.

Round 2

Reviewer 1 Report

Referee Report on “Who’s Afraid of Biotic Threats? An Econometric Analysis of Veneto Wine Grapes Farmers’ Propensity to Insure”_agriculture-863291-peer-review-v2

Thanks to the author for the modification of this article, especially the descriptive statistics of the relevant variables in the questionnaire. However, I have the following specific concerns.

Major Concerns and Comments:

1. The author pointed out that this study is a semi-structured questionnaire and contains 22 questions. I suggest just discussing the questions related to this research and its descriptive statistics. The questions which is not directly related to this research should be deleted.
2. The author used the methodology based on observed economic behavior of farmers (Iyer et al, 2019). In table 1, the author pointed out that they have interpreted such sample of respondents as the sample of risk adverse population, whose risk preference is characterized by a concave Bernouilli utility function. As I mentioned in the first review report, the most important variable in this study is the measurement of risk attitude, and the author should clearly explain how this variable is measured among the items which were selected from Iyer et al. (2019). The author put the interpretation in table 1 is not suitable, I strongly recommend the author rewrite this section carefully in the main text.
3. The author pointed out, in this case risk aversion is a qualitative variable (0/1) and the marginal effect cannot be interpreted correctly. In other words, we cannot say that the X% of extra risk aversion implies X% of increased probability to insure, having a dummy variable. The estimated marginal effects have no sense. However, comparing the marginal effects among the four regression models is meaningful to show which threats are the most concerned of the farmers. Otherwise, the Multivariate Probit Model should be used in this article.

Author Response

We thank again for the points, and take the occasion to better clear up and explain.  

1. The questions, operationalized as variables in the descriptive statistics Table, are the ones related to the research. We have explained in the text that we have only selected some replies from the surveys. However, the presented variables and descriptive statistics represent the whole sample of variables we have used to test the model. We only present selected results (in Table 2), over a broad variety of checks and attempts that have involved the use of all the variables described in the Table
2. We have better explained how the risk aversion variable was created and put in the main text. We have referenced literature and authors that have used similar approaches.
3. In the paper we test the micro theory and check whether risk aversion incentivize to insure (against biotic threats in our application). The impact of socio-economic variables on risk aversion profiles was selected and measured preliminary, through the use of a probit model that we did not present in previous versions of the paper and have now included in a final appendix.  For the sake of comparability, given our previous reasoning on marginal effects, we think that the signs, the magnitude of the estimated coefficients and the statistical diagnostics are solid indicators for comparability. This position is justified by the fact that the sample is the same for all the regressions proposed and consequently the hypothetical Marginal Effects at the Mean are calculated over the same mean value of risk aversion. Being the fact that the comparison of signs and magnitude of estimated coefficients gives the same information of the marginal effect and that their calculation is not rigorous for dummy variables, we still prefer to avoid to present marginals of the estimates.

Author Response File: Author Response.docx