Abstract 1

This paper investigates the impacts of conservation policies on deforestation, […] explicitly [taking] into account indirect general equilibrium effects. Intuitively, protecting land in certain locations affects local land prices, wages, and agricultural output prices that may lead to deforestation elsewhere, given that regional markets are interconnected – this is known in the literature as ‘leakage.’ Though widely acknowledged as conceptually important, the existing empirical works have not yet explicitly dealt with this issue, except via ‘reduced form’ spillovers, focusing on regions that are adjacent to protected areas (PA). This ‘traditional’ approach, though informative, is not fully satisfying as spillovers may affect non-adjacent regions as well. As I see it, this is a crucial gap in the literature that this paper aims to fill. That is an important contribution.
To do so, the authors develop a quantitative spatial equilibrium model of the Brazilian economy. They find that targeting protection in the regions with highest deforestation levels can be effective, despite the fact that leakage increases significantly in a longer time-horizon. Specifically, after one year, 2-3% of the deforestation reductions are outdone by leakage, which is pretty small. Although this number increases to 10% after 10 years, it is still sufficiently small in my estimation to justify protecting the land.

Summary Measures

We asked evaluators to give some overall assessments, in addition to ratings across a range of criteria. See the evaluation summary “metrics” for a more detailed breakdown of this. See these ratings in the context of all Unjournal ratings, with some analysis, in our data presentation here.2

Overall assessment: We asked evaluators to rank this paper “heuristically” as a percentile “relative to all serious research in the same area that you have encountered in the last three years.” We requested they “consider all aspects of quality, credibility, importance to knowledge production, and importance to practice.”

Journal rank tier, normative rating (0-5): “On a ‘scale of journals’, what ‘quality of journal’ should this be published in? (See ranking tiers discussed here)” Note: 0= lowest/none, 5= highest/best”.

See here for the full evaluator guidelines, including further explanation of the requested ratings.

Written report3

This paper investigates the impacts of conservation policies on deforestation, […] explicitly [taking] into account indirect general equilibrium effects. Intuitively, protecting land in certain locations affects local land prices, wages, and agricultural output prices that may lead to deforestation elsewhere, given that regional markets are interconnected – this is known in the literature as ‘leakage.’ Though widely acknowledged as conceptually important, the existing empirical works have not yet explicitly dealt with this issue, except via ‘reduced form’ spillovers, focusing on regions that are adjacent to protected areas (PA). This ‘traditional’ approach, though informative, is not fully satisfying as spillovers may affect non-adjacent regions as well. As I see it, this is a crucial gap in the literature that this paper aims to fill. That is an important contribution.

To do so, the authors develop a quantitative spatial equilibrium model of the Brazilian economy. They find that targeting protection in the regions with highest deforestation levels can be effective, despite the fact that leakage increases significantly in a longer time-horizon. Specifically, after one year, 2-3% of the deforestation reductions are outdone by leakage, which is pretty small. Although this number increases to 10% after 10 years, it is still sufficiently small in my estimation to justify protecting the land.

Main comments

My overall assessment of the paper is positive. It addresses an important problem, it uses appropriate data and modeling techniques, and is well written. I believe it should be published in a good journal. My goal here is to help the authors increase the chances of publication.

My main comment is that the paper could be much more ambitious. It could investigate further implications of preservation policies that would be important for policy makers.

1. The paper could convert the counterfactual deforestation figures into avoided carbon emissions, linking its main results to the opening paragraph of the article. There are alternative carbon maps available that could be used to that end.

2. Armed with the carbon emissions numbers, they could calculate the social benefits of PAs [protected areas using the social cost of carbon, which is also available in the climate economics literature.

3. In addition, given the complete general equilibrium model used, the authors can calculate the welfare impacts everywhere in Brazil. Quantifying Brazilian social welfare should be interesting in itself, and could be used to compare the local benefits/costs of conservation with the international gains from reduced carbon emissions. Is it the case that PAs benefit the world at the expense of Brazilian farmers, workers and consumers? And if so, how much the world could compensate Brazil for its environmental policies?

These are important questions that matter for conservation policies and international agreements; that can be addressed by the paper; and that can make the paper much stronger and influential. Further, the authors can mention that their methodology can also be used to investigate environmental policies in other important contexts, such as in Indonesia and Congo.

Specific comments

1. The introduction misses a discussion of the main quantitative findings of the paper.

2. Section 2 misses a description of the data sources and the main variables used in the paper. It should also present some summary statistics to give the reader a sense of magnitudes. The unit of analysis (microregions) and the sample period (2003–2019?) should also be clear early on.

3. I struggle to see the importance of Sections 2.2. (Econometric Specification) and 2.3 (Results). Though they provide some useful information (covering event-studies for the Priority List, and regression discontinuity for protected areas), they are not central to understanding the results of the spatial general equilibrium model. In fact, the equilibrium model points to the failure of these empirical strategies to identify the causal impacts of these environmental policies. As such, I do not see good reasons to spend so much time on these strategies.

A good rule of thumb is to maintain in the main paper only the information that is crucial to understand its main findings. Everything else either can go to the Appendix or can be cut entirely from the article. I recommend moving Sections 2.2 and 2.3 to the Appendix and refer to them briefly in the main text.

4. Section 2 should include, and expand, the discussion of the stylized facts (current on page 19) that the equilibrium model will try to incorporate.

5. In terms of the model presented in Section 3, I have two main substantive questions/comments:

   1. Brazil is the main worldwide exporter of soy and beef. Does the model incorporate an export sector or consider a completely closed economy? If the economy is assumed completely closed, I would recommend including an export sector into the model.

   2. The model comprises the entire Brazilian economy, but some assumptions give the impression of being motivated by the Amazon rainforest alone. For example, assuming open access to land seems reasonable for the Amazon, but much less reasonable for the rest of the country, which has a well-consolidated (and highly competitive) agricultural sector. One possibility – if feasible – is to model the Amazonian region differently than the rest of the country.

Minor comments

1. Page 2: “Since approximately three-quarters of global deforestation is driven by agriculture…” What is the data source for this statement?

2. Page 3: Where is the arc of deforestation in Figure 1? It might help to have a map of the biomes somewhere in the paper (maybe in the Appendix).

3. Page 3: What is the fraction of protected areas in the Amazon, and in Brazil in general?

4. Page 4: In the paragraph “Typically,…” the authors can mention that leakage leads to the violation of SUTVA, motivating the use of equilibrium models.

5. Page 4: “Our model considers a multi-region economy…” of Brazil? Or of the Brazilian Amazon? Please clarify this early.

6. Page 7: the paragraph “Before turning to…” is too defensive. You can convey the same information without being so defensive.

7. Page 7: The paragraph “Reduced-form evidence…” is repetitive, you can cut it out.

8. Page 8: Here is the first time I see reference to “additionality.” This is an important issue in the literature of the impacts of PA, and should appear much earlier in the paper. Some PAs in Brazil seem to be placed in regions with no immediate threat of deforestation, leading to small or zero additionality, while others were put in places in which forests were at risk, leading to substantial additionality.

9. Page 10: “The graph below shows…” Which graph?

10. Page 13: Where are the synthetic difference-in-differences results?

11. Page 19: The paragraph “Mechanisms for leakage” seems repetitive. The mechanisms are already clear at that point in the paper. I would move this discussion to the main results section.

12. Page 19: Footnote 9 misses the definition of $L_o$.

13. Page 21: In “Model features” is worth explaining which are the $K$ agricultural commodities considered in the empirical exercise.

14. Page 21-22: I would include Table 3 at the end of the model description. One cannot fully grasp it before understanding the model.

15. Page 22-23: It is worth explaining the mechanics of Figure 11. Even better is to bring Figure 16 currently in the Appendix to the main text and explain it.

16. Page 24: There must be a typo here. It claims that “there is free entry of deforesters.” The free-entry condition leads to zero profit in equilibrium, i.e., $q.Z.(I)^\delta-p.I=0$. But equation (4) reflects the FOC of the profit maximization problem.4

17. Page 24: I found the explanation of the estimation of $\delta$ and $\psi$ confusing. It would be helpful to have the data sources and variables explained earlier, as suggested above. Discussing this estimation after presenting the demand side could also help.

18. Page 25: It is worth mentioning that labor and land are complements, which explains why when the price of land goes up, the demand for labor goes down, in spite of the substitution effect.

19. Page 20: I did not understand how the “wage and occupational choice” part relates to the rest of the model. For instance, by reading just this section, it is unclear how the $z_i^A$ and $z_i^{NA}$ relate to the TFP $Z_r^{Ak}$ presented in the previous page.

20. Page 26: The economic discussion after explaining how $\alpha$ is estimated requires knowledge of the consumers’ demand. It is therefore worth postponing the discussion to [put after you present] the consumers’ part, maybe in the results section.

21. Page 27: It would be useful to have some a-priori justification of the utility function specification used. And to provide the formulae more explicitly too.

22. Page 28: Some justification for taking $\omega=9$ is warranted.

23. Page 28: Some variables missed a proper definition when introduced, as for example $X_d^g$.

24. Page 29: Presenting the migration decisions (Section 3.5) after closing the model (Section 3.4) seems weird.

25. Page 46: Implied $\delta$ is 0.26-0.29, but $\delta$ in Table 3 equals 0.5. This is confusing, please check.

26. I missed the citation of a few important papers

    1. “The Environmental Impacts of Protected Area Policy,” (2023, RSUE) by Reynaert, Souza-Rodrigues, and van Benthem.[1]

This is a recent overview of the empirical literature on the impacts of protected areas. It explicitly states that incorporating spatial general equilibrium effects is an important gap. The authors here can mention this fact to reinforce the contribution of their paper. They can also find other papers in the review that may be worth citing.

2. “Optimal Environmental Targeting in the Amazon Rainforest,” (2023, RESTUD) by Assuncao, McMillan, Murphy, and Souza-Rodrigues[2]

This paper focus on the Priority List and finds important complementarities between the list and the protected areas.

3. “An Evaluation of Protected Area Policies in the European Union” (2023, Working Paper), by Grupp, Mishra, Reynaert, and van Benthem.[3]

This paper uses state-of-the-art techniques (including the use of machine learning) to study the impacts of PAs in Europe (one of the rare papers focused on developed countries), and found negligible impacts there. Developing a general equilibrium model for Europe to capture leakage might therefore not be as important as for Brazil.

Evaluator details

1. How long have you been in this field?

   • [about 10 years]

2. How many proposals and papers have you evaluated?

   • “I review about 10-15 papers/proposals per year, on average.”