Using Conditional Cash Payments to Prevent Land-Clearing Fires: Cautionary Findings from Indonesia

Round 1

Reviewer 1 Report

Dear Authors,

I want to thank you for your submission and the opportunity to review an interesting manuscript focused on a current topic related to the analysis of conditional cash payments as a solution to prevent land-clearing fires in West Kalimantan, Indonesia.

I consider that the article addresses a current topic based on an in-depth analysis of land-clearing forest fires (sometimes deliberately) in rural areas of West Kalimantan that generate negative effects in environmental, economic, social and geopolitcal terms: greenhouse gas emissions, deforestation, habitat destruction, worsened human health, strained international relations.

 

Introduction:

The authors make a brief presentation of the subject highlithing the solution identified and implemented by  the government by granting conditional cash payments to prevent  land-clearing fires in villages in rural areas. One of the objectives formulated by the authors is to analyse the efficiency of this measure and to study an alternative approach based on a large policy experiment focused instead on the use of incentives.

Methodology: Could you specify if the questionnaire applied to the local population is formulated by the authors  or a model from the specialized literature was used?

Result section: I recommend you to rename the section: Results and Discussion

Most of the results are presented in the discussion section.

Discussion section is missing.

You should include in the Discussion subsection several information: (1) the results should be better compared with the results obtained in other studies (elaborated for Indonesia or other countries where the land-clearing fires are practiced to expand agriculture land) by reference to similarities or differences. This comparison of the results would better reflect the relevance of the study;

(2) you should make a better description of the contribution of the results to the development of the field (theoretical or methodological) emphasizing the importance of the study;

 (3) you should emphasize the main elements that are related to the originality of the study and how you obtained original results.

(4) you should present the limitation of the study and future research.

Author Response

The authors are grateful for your helpful comments.  We have done our best to incorporate your suggestions. In the few cases where we could not comply, we have indicated the reasons as to why.  Specific reactions to your suggestions are interspersed point by point in the commentary attached file.

Author Response File: Author Response.docx

Reviewer 2 Report

This manuscript aims to provide a broad-level overview of the reasons behind null results of a conditional cash payment program aimed at preventing forest fires in Indonesia. The experimental set-up was very ambitious and I applaud the authors for pursuing reasons behind the 'non-findings'. I also understand that the difficulties of Covid-related travel restrictions made follow-up fieldwork difficult. Yet, I feel like the arguments in the paper need to be supported with better and more systematic data for them to be credible explanations for the outcomes. While the methods mention a number of data collection methods to understand conditions in the treatment and control villages, much of the results focus on broader-level explanations that do not utilize the specific village and interview data to corroborate their points. In a first instance, I would utilize these interviews and focus group discussions more fully, including using quotes that illustrate broader-level findings and arguments.

Second, the statement "Among the 10 villages studied in depth, we found considerable variation in knowledge about our experiment. Since the facilitation process was virtually identical in all treatment villages, the information-transfer appeared inadequate in some communities, and this deficiency may have influenced the collective response to our incentive program." I think this issue needs to be given much more space and treated more seriously, including through additional data collection among all of the villages in the treatment group. If knowledge about the treatment is limited in villages in the treatment group, in effect the number of villages that you 'intended to treat' is not the same as the villages that you 'treated', and I would limit the analysis to villages that were fully aware of the prize offer and rules of the program, as only in those instances are the alternative explanations that you provide (small number of 'rogue actors' that cannot be controlled by the rest of villagers, feeling of no responsibility for fire events in land with unclear tenure, etc.) relevant.

I am also curious about the exclusion of the 'traditional fires' set and would like some more information on the methodology - did you only exclude fire events that were 'notified' by villagers? How do you see this approach in the context of the fact that some villages had less information about the treatment intervention, do you think you may have undercounted traditional fires in settings where villagers were unaware of the intervention? Second, did you always exclude fire events that were notified, irrespective of size? Was there the possibility of villagers strategically 'hiding' intentional fires for clearing behind the 'traditional label', or how did you verify the traditional type of fire on the ground?

Finally, in all your analyses in the document I would encourage you to be clearer about what is the independent variable (treatment/control) and what the dependent variable (fire/non-fire), even in this more qualitative setting. Statements such as "Interestingly, the average portion of village land still in forest (28 percent) was the same in both fire and non-fire villages in our study sample (i.e., treatment and control villages)." confuse these two concepts, and comparisons regarding particular correlates or antecedents of fire outcomes (e.g. land in forest, Dayak-majority villages) would need to at the very least always be evaluated in this two-way comparison (treatment/fire, treatment/non-fire, control/fire, control/non-fire), with indicative evidence of potential explanations for fire events holding in cases where the variable is similar between treatment and control but different between fire and non-fire. For instance, the statement "Interestingly, of the 21 treatment villages that had no fires, 11 were predominantly Dayak (i.e., greater than 90 percent Dayak ethnicity), whereas only four Dayak villages had fires. This result helps lay to rest one commonly held assumption that Dayaks are the primary instigators of fires in the region" is insufficient in that we first need information about overall proportions (I read this as saying that 11/21 non-fire/treatment villages are Dayak and 4/54 fire/treatment villages are Dayak, with 15/75 treatment villages being Dayak), but we also need to know about the control villages with and without fires - to what extent is a share of 20% Dayak villages representative of the overall population in the region? Also, does this statement mean that the Dayak villages responded particularly well to the treatment or that Dayaks in general do not instigate fires (in which case you really need the control village distribution as well)? 

In addition, it would be good to discuss a bit more the philosophy and ethical implications behind the idea of yearly, large-scale conditional payments to villages. You mention that increasing the payment amount would have been prohibitive both for the experiment and for the scalability of the intervention, but do you think that in its current form paying over 10'000 USD per year per Indonesian village is realistic, even if it did work? What are the assumptions behind such compensatory schemes (i.e. that villagers hold the right to burn or clear unless compensated), and what would a large-scale roll-out imply?

Many thanks for the interesting work.

Author Response

The authors are grateful for your helpful comments.  We have done our best to incorporate your suggestions. In the few cases where we could not comply, we have indicated the reasons as to why.  Specific reactions to your suggestions are interspersed point by point in the attached file.

Author Response File: Author Response.docx

Reviewer 3 Report

This article aims to assess both the potential and limitations of using conditional cash payments to prevent villages from setting land-cleaning files. Using an experimental approach to compare the villages with treatment and the villages without treatment, the Author concludes that the incentive had no impact on fire outcomes. The study on the incentive mechanisms is rare in the context of fire mitigation, so this article is interesting and will significantly contribute to the debate on fire mitigation. However, the discussion is too general. —The explanation of the underlying qualitative of land-clearing fires applied both the incentive mechanisms and command control approach. Refers to the objectives of this study to assess the impact of using incentive mechanisms. I expect more specific results of this study, under what circumstances the conditional payment works and does not. For example, you mention that collective action is essential, but I did not see the comparison of social capital between treatment and control villages. Another example is whether the treatment villages have more access to machinery tools for land clearing than the control villages?.  This will be important as the basis for the policy intervention. For example, strengthening the social capital of smallholder farmers and providing them access to machinery tools for land clearing is more effective than the conditional cash payment approach. The second issue is the sample village. Are all sample villages have a similar risk to fire?. Villages with low risk will have a high probability to won the experiment. So, a stratified sample is needed to avoid bias.

There are two major constraints on the use of zero-burning practices for land clearing by farmers, these being: (i) farmers’ limited financial capability to bear the associated costs; and (ii) farmers’ limited access to the technology and machinery required to implement them. Those limitations are also the main causes of villages cannot win the experiment since the upfront money is too small to use for applying zero burning.

Minor Questions:

1.      If the traditional fire applied by Dayak people is excluded, why don’t you exclude the villages with dominant Dayak people.

 

2.      Who is rouge people?

Author Response

The authors are grateful for your helpful comments.  We have done our best to incorporate your suggestions. In the few cases where we could not comply, we have indicated the reasons as to why.  Specific reactions to your suggestions are interspersed point by point in the attached file.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

The authors have responded to my comments. However, It seems the Author is misinterpretation one of my comment. My comment is the access to machine tools for land clearing, while the authors interpreted it as extinguishing fires. Fire as a tool for land clearing is the cheapest, fastest, and easiest method. Zero burning is more expensive and requires access to machines for land clearing that are often not available for smallholder farmers. This is one explanation for why the conditional cash payment did not work well. Could the authors re-response to this issue.  Consider deleting your response on the machine for extinguishing the fire. 

Author Response

"The authors have responded to my comments. However, It seems the Author is misinterpretation one of my comment. My comment is the access to machine tools for land clearing, while the authors interpreted it as extinguishing fires. Fire as a tool for land clearing is the cheapest, fastest, and easiest method. Zero burning is more expensive and requires access to machines for land clearing that are often not available for smallholder farmers. This is one explanation for why the conditional cash payment did not work well. Could the authors re-response to this issue.  Consider deleting your response on the machine for extinguishing the fire.”                                                                               

 

RESPONSE

I agree with your fundamental point, and I apologize for my inept response in the earlier round of comments.  I had been thinking about a two step process, first, the decision to clear, and second, which “technology" to use for clearing.  Fire, as you point out, has a key role in both of those issues.  The possibility of using low-cost fire increases greatly the expected earnings from new land and thus the technology choice that results in severe environmental damage.  I have rewritten the conclusions (copied below) to make your point.

"Fundamentally, however, we believe that basic economics drove the fires, and, conversely, that the fires drove the economics.  The net present value of land is high in these villages as the consequence of a well-established oil palm industry. That high profitability, and the financial incentives it provides for illegal behavior, occurs in large part because of the potential for using low-cost fire techniques for clearing land. Once the decision has been made to clear, relative costs, and often the physical unavailability of machinery, inevitably drive poor farmers toward the use of land-clearing fires. The net result is a terrible dilemma for everyone. Oil palm, the means for higher incomes for many people in this relatively poor province, simultaneously creates large negative externalities in the form of deforestation, habitat destruction,  human health damages, and strained international relations."

 

Sincerely yours,

Walter Falcon