Consumers’ Willingness to Pay for Attributes of Sustainability, Origin and Production Process in Raicilla

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study estimates the willingness to pay for agave distillate through a choice experiment and discusses the possibility of creating new value by adding sustainability factors, such as eco-labeling, as attributes to the basic purchase factors.

However, this study raises significant concerns about the significance of this research and the analytical methodology.

First, why is it necessary to study raicilla? How will the results of the study benefit society? If this research only provides us with the values of the purchasing factors of raicilla, including the labels and the strategy to increase the sales of raicilla, then the companies that sell raicilla can do their own marketing research on their own. I don't see why this research is necessary as an academic study that is in the public interest and is not mentioned in the text. Also, in the case of hard liquor, an increase in sales would lead to an increase in the number of alcoholics. Given that, wouldn't this study be more harmful to society?

Second, does the interaction model fit the purpose of this study? There is a lack of variables in the result of the interaction model, which is not worth comparing with the main effect model, and they do not show the standard deviation of random parameters even though they used a mixed logit model. In addition, they did not calculate MWTPs (and confidence intervals) which provides more intuitive look for understanding the result of this study.

 

Introduction.

I am not so familiar with raicilla and how important this liquor is to the Mexican people. Describing the background of raicilla consumption and the popularity/importance of raicilla to the Mexican people helps this study focus on raicilla rather than other liquors.

 

L159-161

Which did you use, a linear model, or a quadratic model to optimize the design? If you used a linear model, it is for the main effect model and makes less sense to use an interaction model, or the interaction model may be biased.

Also, did you optimize for D-efficiency or A-efficiency? 

 

L202, 

Table 4 should be Table 5.

 

Table 5.

In the interaction model, why is the biodiversity management label (main effect) missing? It will not be comparable to the main effect model.

Isn't the log likelihood negative? Was it positive or a typo?

 

Why do you need to include the interaction model? what is the purpose of this study? If this DCE is designed to be a linear model in optimization, the interaction model is not appropriate to use. I also find less meaningful results from the interaction model. The choice of variables also seems arbitrary. Why don't you run a latent class model if you want to find out the demographic aspect?

Also, if you ran a mixed logit model, please show which parameters you chose as random parameters. I think the estimation gives you standard deviations of the random parameters and the statistical significances. The information would also be helpful for you to understand which parameters have significant heterogeneous aspects.

 

Why not compute marginal willingness to pay and confidence intervals? The estimates are not too bad as long as the constant is negative WTP.

 

One thing I am concerned about is how often women consume Raicilla. If few females consume this alcohol, there will be a big bias in this estimate. Please clarify this point.

 

Author Response

Thank you very much for taking the time to review this manuscript. Below you will find detailed responses and corresponding revisions/corrections highlighted in the forwarded files.

Comments 1: First of all, why is it necessary to study raicilla? How will the results of the study benefit society? If this research only provides us with the values ​​of the purchasing factors of raicilla, including labels and the strategy to increase sales of raicilla, then companies selling raicilla can do their own market research on their own. I don't see why this research is necessary as an academic study that is in the public interest and is not mentioned in the text. Also, in the case of hard liquor, an increase in sales would lead to an increase in the number of alcoholics. Considering that, wouldn't this study be more harmful to society?

Answer 1: We agree with this comment, so in the introduction we have added information about the importance of the Raicilla study (L27-41). This study does not aim to intensify large-scale sales of Raicilla or agave distillates, on the contrary, it aims to evaluate market niches with consumers who are responsible with regard to consumption and the environment, so the type of consumer it is intended to target was specified in the document (L47-48; L315-316).

Comments 2: Second, does the interaction model fit the purpose of this study? There is a lack of variables in the result of the interaction model, which is not worth comparing with the main effects model, and they do not show the standard deviation of the random parameters even though they used a mixed logit model. In addition, they did not calculate the MWTPs (nor confidence intervals), which provides a more intuitive view to understand the result of this study.

Response 2: We agree with this comment, so we decided to remove the interaction model in the study. The marginal willingness to pay (L201-206; L243-256; L350-367) and confidence intervals (L235-236) were added.

Comments 3: Introduction. I am not very familiar with raicilla or the importance of this liquor to the Mexican people. Describing the background of raicilla consumption and its popularity/importance to the Mexican people helps this study focus on raicilla rather than other liquors.

Answer 3: Information has been added to the introduction about the importance of Raicilla (L27-41).

Comments 4: Which did you use, a linear model or a quadratic model to optimize the design? If you used a linear model, it is for the main effect model and it makes less sense to use an interaction model, or the interaction model may be biased. Also, did you optimize for D-efficiency or A-efficiency?

Answer 4: We agree with the comment made, because the optimization model was linear (L180), so the interaction model was eliminated.

Comments 5: Table 4 should be Table 5.

Answer 5: Your assessment is correct, the correction was made to the text (L237).

Comments 6: Table 5. In the interaction model, why is the biodiversity management label missing (main effect)? It will not be comparable with the main effect model.

Answer 6: The interactions model was removed due to the comments and suggestions mentioned above.

Comments 7: The log-likelihood is not negative? Was it positive or was that a typo?

Answer 7: The log likelihood was positive in the model.

Comments 8: Why is it necessary to include the interaction model? What is the purpose of this study? If this DCE is designed to be a linear model in optimization, the interaction model is not suitable for use. I also find less significant results from the interaction model. The choice of variables also seems arbitrary. Why don't you run a latent class model if you want to find out the demographic aspect?

Answer 8: We agree with the comment, so the interactions model has been removed.

Comments 9: Also, if you ran a mixed logit model, please indicate which parameters you chose as random. I believe the estimation gives you the standard deviations of the random parameters and the statistical significances. The information will also be useful for you to understand which parameters have significant heterogeneity.

Answer 9: The random parameter was the price (L162).

Comments 10: Why are marginal willingness to pay and confidence intervals not calculated? The estimates are not that bad if the constant is a negative willingness to pay.

Answer 10: We agree with this comment, which is why the marginal provision to pay (L201-206; L243-256; L350-367) was included.

Comments 11: One thing that concerns me is the frequency with which women consume raicilla. If few women consume this alcohol, there will be a large bias in this estimate. Please clarify this point.

Answer 11: This data was not presented in the manuscript, the frequency of consumption included was for both sexes.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Following are my comments provided for authors regarding sustainability-3145038.

(1) In Introduction, it is recommended that the author supplement the analysis and review of such issues to highlight the innovativeness of this article.

(2) The structure description of this paper in Section 1 should be added in the form of a separated paragraph.

(3) In Section 2.1, authors stated that Participants were recruited through intentional and reasoned sampling with predetermined criteria [14], using maximum variance, with a reliability of 95% and a margin of error of 7% [15]. I wonder the relation between Refs [14-15] and this paper. Obviously, the author list is quite different.

(4) In Table 1, the content in line ‘Average consumption of Raicilla per month’ misses.

(5) How do authors ensure that the number of attribute levels for each area in Table 2 is reasonable.

(6) Fractional factor analysis was used on different attributes through the orthogonal design. What is the optimal parameter combination obtained by the SPSS?

(7) Authors are advised to provide a fair weakness and limitation of their work, and how it can be improved.

(8) To better suit this journal, some related studies published in Sustainability can be reviewed and referenced.

(9) The linguistic should be improved carefully throughout the manuscripts.

Comments on the Quality of English Language

Minor editing.

Author Response

Comments 1: In Introduction, it is recommended that the author supplement the analysis and review of such issues to highlight the innovativeness of this article.

Response 1: We agree with this comment. Therefore, we have added the information in the introduction (L27-41).

Comments 2: The structure description of this paper in Section 1 should be added in the form of a separated paragraph.

Response 2: It was not possible to address this comment in its entirety, as it was not clear to us which specific section it referred to.

Comments 3: In Section 2.1, authors stated that Participants were recruited through intentional and reasoned sampling with predetermined criteria [14], using maximum variance, with a reliability of 95% and a margin of error of 7% [15]. I wonder the relation between Refs [14-15] and this paper. Obviously, the author list is quite different.

Response 3: Regarding this comment, a change was made to the wording for greater clarity (88-89), because Reference 14 (now 15) proposes the predetermined criteria and Reference 15 (now 16) refers to the maximum variance formula.

Comments 4: In Table 1, the content in line ‘Average consumption of Raicilla per month’ misses.

Response 4: This information had not been included because it is a quantitative variable, however, for the purposes of better understanding the table, it was added as indicated in the comment (L99-100).

Comments 5: How do authors ensure that the number of attribute levels for each area in Table 2 is reasonable.

Response 5: This comment was implemented into the manuscript at L104-107.

Comments 6: Fractional factor analysis was used on different attributes through the orthogonal design. What is the optimal parameter combination obtained by the SPSS?

Response 6: The optimal combination was the one used in the study (L172-177; L184-185).

Comments 7: Authors are advised to provide a fair weakness and limitation of their work, and how it can be improved.

Response 7: We agree with this comment, so the limitations of the study were included in the discussion section (L368-374).

Comments 8: To better suit this journal, some related studies published in Sustainability can be reviewed and referenced.

Response 8: References from the Sustainability journal was included for better adaptation (L408-409; L485-486; L489-490; L493-494).

Comments 9: The linguistic should be improved carefully throughout the manuscripts.

Response 9: Improvements were made to the linguistic part of the manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

This study estimates the willingness to pay for agave distillate through a choice experiment and discusses the possibility of creating new value by adding sustainability factors, such as eco-labeling, as attributes to the basic purchase factors.

However, this study raises significant concerns about the significance of this research and the analytical methodology.

First, I was not convinced why this study is necessary and how the results will benefit society? If this research only provides us with the values of the purchasing factors of raicilla, including the labels and the strategy to increase the sales of raicilla, then the companies that sell raicilla can do their own marketing research on their own. I could not find why this research is necessary as an academic study that is in the public interest and is not mentioned in the text (e.g. introduction). Also, in the case of hard liquor, an increase in sales would lead to an increase in the number of alcoholics. Given that, wouldn't this study rather be more harmful to society? Especially, ecolabeling (e.g. organic label) may confuse people to understand raicilla with the label is healtheir than ordinary ones.

Second, does the interaction model fit the purpose of this study? There is a lack of variables in the result of the interaction model, which is not worth comparing with the main effect model, and they do not show the standard deviation of random parameters even though they used a mixed logit model. In addition, they did not calculate MWTPs (and confidence intervals) which provides more intuitive look for understanding the result of this study.

 

Introduction.

I am not so familiar with raicilla and how important this liquor is to the Mexican people. Describing the background of raicilla consumption and the popularity/importance of raicilla to the Mexican people helps this study focus on raicilla rather than other liquors.

 

L159-161

Which did you use, a linear model, or a quadratic model to optimize the design? If you used a linear model, it is for the main effect model and makes less sense to use an interaction model, or the interaction model may be biased.

Also, did you optimize for D-efficiency or A-efficiency? 

 

L202, 

Table 4 should be Table 5.

 

Table 5.

In the interaction model, why is the interaction model missing the biodiversity management label? It will not be comparable to the main effect model.

Also, isn't the log likelihood negative? Was it positive or a typo?

 

Why do you need to include the interaction model? what is the purpose of this study? If this DCE is designed to be a linear model in optimization, the interaction model is not appropriate to use. I also find less meaningful results from the interaction model. The choice of variables also seems arbitrary. Why don't you run a latent class model if you want to find out something with regard to the demographic aspect?

Also, if you ran a mixed logit model, please show which parameters you chose as random parameters. I think the estimation gives you standard deviations of the random parameters and the statistical significances. The information would also be helpful for you to understand which parameters have significant heterogeneous aspects.

 

Why not compute marginal willingness to pay and confidence intervals? The estimates are not too bad as long as the constant is negative WTP.

 

One thing I am concerned about is how often women consume Raicilla. If few females consume this alcohol, there will be a big bias in this estimate. Please clarify this point.

Author Response

Below you will find the detailed answers and corresponding revisions/corrections highlighted in the forwarded files.

Comments 1: First of all, I was not convinced why this study is necessary and how the results will benefit society. If this research only provides us with the values ​​of the purchasing factors of raicilla, including labels and the strategy to increase raicilla sales, then companies selling raicilla can do their own market research on their own. I could not find why this research is necessary as an academic study that is in the public interest and is not mentioned in the text (e.g. the introduction). Also, in the case of hard liquor, an increase in sales would lead to an increase in the number of alcoholics. Considering that, wouldn't this study be more harmful to society? Especially, eco-labeling (e.g. organic label) can confuse people into understanding that raicilla with the label is healthier than the regular ones.

Answer 1:In the introduction we have added more information about the importance of the Raicilla study. This study does not aim to intensify large-scale sales of Raicilla or agave distillates, on the contrary, it aims to evaluate market niches with consumers responsible with consumption and the environment, so the type of consumer it is intended to target was specified in the document (L27-56).

Comments 2: Secondly, does the interaction model fit the purpose of this study? There is a lack of variables in the result of the interaction model, which is not worth comparing with the main effects model, and they do not show the standard deviation of the random parameters even though they used a mixed logit model. Also, they did not calculate the MWTPs (nor the confidence intervals), which provides a more intuitive view to understand the result of this study.

Answer 2: We agree with this comment, so we decided to remove the interaction model in the study. The marginal willingness to pay and confidence intervals were added.

Comments 3: Introduction. I am not very familiar with raicilla or the importance of this liquor to the Mexican people. Describing the background of raicilla consumption and its popularity/importance to the Mexican people helps this study focus on raicilla rather than other liquors.

Response 3: Information has been added to the introduction about the importance of Raicilla.

Comments 4: What did you use, a linear model or a quadratic model to optimize the design? If you used a linear model, it is for the main effect model and it makes less sense to use an interaction model, or the interaction model may be biased. Also, did you optimize for D-efficiency or A-efficiency?

Response 4: We agree with the comment made, because the optimization model was linear, so the interaction model was removed.

Comments 5: Table 4 should be Table 5.

Response 5: Your assessment is correct, the correction was made to the text.

Comments 6: Table 5. In the interaction model, why is the biodiversity management label (main effect) missing? It will not be comparable with the main effect model.

Reply 6: The interaction model was removed due to the comments and suggestions mentioned above.

Reply 7: The log likelihood is not negative? Was it positive or was this a typo?

Reply 7: The log likelihood was positive in the model.

Reply 8: Why is it necessary to include the interaction model? What is the purpose of this study? If this DCE is designed to be a linear model in optimization, the interaction model is not suitable for use. I also find less significant results from the interaction model. The choice of variables also seems arbitrary. Why don't you run a latent class model if you want to find out the demographic aspect?

Reply 8: We agree with the comment, so the interaction model was removed.

Comments 9: Also, if you ran a mixed logit model, please indicate which parameters you chose as random. I believe the estimation gives you the standard deviations of the random parameters and the statistical significances. The information will also be useful for you to understand which parameters have significant heterogeneous aspects.

Answer 9: The random parameter was the price.

Comment 10: Why are marginal willingness to pay and confidence intervals not calculated? The estimates are not that bad if the constant is a negative willingness to pay.

Reply 10: We agree with this comment, which is why the marginal willingness to pay was included.

Reply 11: One thing that concerns me is the frequency with which women consume raicilla. If few women consume this alcohol, there will be a large bias in this estimate. Please clarify this point.

Reply 11: This data was not presented in the manuscript, the consumption frequency included was for both sexes.

Reviewer 2 Report

Comments and Suggestions for Authors

My comments have been well addressed by authors.

Author Response

Thank you very much for your review, comments and suggestions.

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

The revised version was adequately addressed to most of my comments. However, there are still some inadequacies. I was particularly concerned about the lack of a gender ratio. In the case of alcoholic beverages, I would expect that men would account for the majority of consumption, but currently the share of sample between men and women is almost even. In other words, if men were the majority consumers, it is conceivable that the women's preference may strongly influence on the result compared with reality. This could could be a significant bias and needs to be addressed. 

 

Answer 9: The random parameter was the price.

Why is only the price the random parameter? Random parameters are usually applied to control heterogeneity of preferences. Price is understandable, but other variables such as sustainability also seems to have heterogeneous preference. What is the criterion to include/exclude random parameters? Citing a reference may be necessary.

Reply 11: This data was not presented in the manuscript, the consumption frequency included was for both sexes.

That means there may be a big bias in this estimation. It may be good to separate samples between male and female to compare how estimation results are different if the majority of consumers are skewed towards the male.

 

Author Response

Comment: Why is only the price the random parameter? Random parameters are usually applied to control heterogeneity of preferences. Price is understandable, but other variables such as sustainability also seem to have heterogeneous preferences. What is the criterion to include/exclude random parameters? Citing a reference may be necessary.

Response: We have a typographical error, which has already been corrected. All the variables used were considered as random parameters, which was indicated in the text and included in the reference (L208-212).

Comentario: Eso significa que puede haber un gran sesgo en esta estimación. Puede ser bueno separar las muestras entre hombres y mujeres para comparar cómo los resultados de la estimación son diferentes si la mayoría de los consumidores están sesgados hacia el hombre.

Respuesta: Se especificaron en el texto los porcentajes del nivel de consumo de Raicilla para los géneros femenino y masculino, con el fin de aclarar que no existe sesgo, ya que ambos sexos presentaron porcentajes similares en cada categoría de consumo (L238-244). Por otro lado, los modelos que incluían datos sociodemográficos, específicamente de género, no fueron significativos (L250-251).