pith. sign in

arxiv: 2509.06076 · v4 · submitted 2025-09-07 · 💰 econ.GN · q-fin.EC· stat.AP

DETERring more than Deforestation: Environmental Enforcement Reduces Violence in the Amazon

Rafael Araujo , Vitor Possebom , Gabriela Setti This is my paper

Pith reviewed 2026-05-18 18:56 UTC · model grok-4.3

classification 💰 econ.GN q-fin.ECstat.AP
keywords environmental enforcementdeforestationhomicidesAmazonsatellite monitoringcausal inferenceBrazilviolence
0
0 comments X

The pith

Real-time satellite monitoring of Amazon deforestation reduced homicides by 15 percent.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that an environmental enforcement program using satellite monitoring not only targets illegal clearing but also lowers violence by expanding government presence in remote areas. The authors instrument for enforcement intensity with variation in cloud cover, which blocks satellite detection and fine issuance, and recover large negative effects on local homicide rates. A sympathetic reader would care because the results point to environmental policies generating side benefits for public safety where state authority is otherwise thin. The estimates indicate the program prevented roughly 1,477 homicides annually.

Core claim

The Real-Time Deforestation Detection System (DETER) allowed the Brazilian government to detect illegal deforestation in near real time and issue fines, which expanded state presence and reduced homicides. Using cloud cover as an exogenous source of variation in monitoring effectiveness, the analysis shows that higher enforcement intensity caused a 15 percent drop in homicides, preventing approximately 1,477 killings per year.

What carries the argument

The DETER satellite monitoring system, which supplies exogenous variation in enforcement through cloud cover that limits detection and fines.

If this is right

  • Environmental enforcement expands state capacity in frontier regions and thereby deters violent conflict tied to illegal resource extraction.
  • Replicable monitoring technologies can simultaneously address deforestation and associated social harms.
  • The 15 percent homicide reduction shows that fines and real-time oversight produce measurable gains in public safety.
  • Similar programs may yield net social returns when violence externalities are counted alongside environmental benefits.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Comparable satellite systems could be piloted in other weakly governed tropical regions to test effects on both forest loss and violence.
  • Savings from fewer homicides might offset part of the fiscal cost of expanded monitoring even if deforestation reductions alone appear modest.
  • Future work could examine whether offenders shift to other crimes or locations once monitoring intensity rises.

Load-bearing premise

Cloud cover affects violence only by hindering satellite detection of deforestation and has no direct link to local conflict or economic conditions.

What would settle it

Finding that homicide rates remain unchanged across high- and low-cloud periods in areas without DETER, or that cloud cover predicts violence through channels unrelated to enforcement such as agricultural output.

Figures

Figures reproduced from arXiv: 2509.06076 by Gabriela Setti, Rafael Araujo, Vitor Possebom.

Figure 1
Figure 1. Figure 1: Homicide Rate: Municipalities in the Amazon Biome vs. Outside the Biome [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Geographical distribution of key variables in 2006 within the Amazon biome [PITH_FULL_IMAGE:figures/full_fig_p018_2.png] view at source ↗
read the original abstract

We estimate the impact of environmental law enforcement on violence in the Brazilian Amazon. The introduction of the Real-Time Deforestation Detection System (DETER), which enabled the government to monitor deforestation in real time and issue fines for illegal clearing, significantly reduced homicides in the region. To identify causal effects, we exploit exogenous variation in satellite monitoring generated by cloud cover as an instrument for enforcement intensity. Our estimates imply that the expansion of state presence through DETER prevented approximately 1,477 homicides per year, a 15\% reduction in homicides. These results show that a replicable environmental enforcement policy produces social benefits.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper claims that the introduction of the Real-Time Deforestation Detection System (DETER) significantly reduced homicides in the Brazilian Amazon. Exploiting exogenous variation in satellite monitoring due to cloud cover as an instrument for enforcement intensity, the authors estimate that DETER prevented approximately 1,477 homicides per year, a 15% reduction, and conclude that this environmental enforcement policy generates important social benefits.

Significance. If the central causal claim holds, the results highlight substantial co-benefits of environmental monitoring policies for reducing violence and improving social outcomes in frontier regions. The creative use of cloud cover variation for identification in a setting with limited state capacity is a methodological strength, and the focus on a replicable policy intervention adds policy relevance.

major comments (3)
  1. [Empirical Strategy] Empirical Strategy section: The manuscript does not report first-stage regression coefficients, F-statistics, or other diagnostics for the cloud cover instrument. This information is essential to assess instrument strength and support the causal interpretation of the 15% homicide reduction.
  2. [Identification discussion] Identification discussion: The exclusion restriction is asserted but not tested or robustly defended against direct channels; cloud cover is mechanically correlated with precipitation and dry-season timing, which can affect agricultural output, logging feasibility, rural mobility, and resource disputes that influence homicide rates independently of enforcement intensity.
  3. [Main results (Table 3)] Main results (Table 3): The headline estimate of 1,477 prevented homicides per year lacks reported robustness checks that include rainfall controls, seasonal fixed effects, or placebo tests that would help evaluate sensitivity to potential exclusion restriction violations.
minor comments (2)
  1. [Abstract] Abstract: The 15% reduction figure would benefit from explicit mention of the baseline annual homicide count in the sample for better context.
  2. [Introduction] Introduction: Additional citations to recent literature on weather instruments and exclusion restriction concerns in conflict or crime settings would improve positioning.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their constructive and insightful comments, which have helped us improve the clarity and robustness of the manuscript. We appreciate the recognition of the paper's methodological contribution and policy relevance. Below we respond to each major comment in turn, indicating where revisions have been made.

read point-by-point responses

  1. Referee: Empirical Strategy section: The manuscript does not report first-stage regression coefficients, F-statistics, or other diagnostics for the cloud cover instrument. This information is essential to assess instrument strength and support the causal interpretation of the 15% homicide reduction.

    Authors: We agree that first-stage diagnostics are essential for evaluating instrument strength. In the revised manuscript we have added the first-stage regression results, including the coefficient on the cloud cover instrument, the Kleibergen-Paap F-statistic (which exceeds 20), and Sanderson-Windmeijer statistics, both in the Empirical Strategy section and in the notes to Table 3. These confirm that the instrument is strong. revision: yes

  2. Referee: Identification discussion: The exclusion restriction is asserted but not tested or robustly defended against direct channels; cloud cover is mechanically correlated with precipitation and dry-season timing, which can affect agricultural output, logging feasibility, rural mobility, and resource disputes that influence homicide rates independently of enforcement intensity.

    Authors: We acknowledge the referee's concern about potential direct effects operating through weather and seasonal channels. We have substantially expanded the Identification discussion to explicitly address these issues, explaining why we view the exclusion restriction as plausible: any direct weather effects on violence would likely be negative (reducing mobility and disputes) and thus bias our estimates toward zero. We have also added rainfall and dry-season controls to the main specifications as a robustness check. Full statistical tests of the exclusion restriction remain infeasible without additional exogenous variation, but the stability of results under these controls supports our interpretation. revision: partial

  3. Referee: Main results (Table 3): The headline estimate of 1,477 prevented homicides per year lacks reported robustness checks that include rainfall controls, seasonal fixed effects, or placebo tests that would help evaluate sensitivity to potential exclusion restriction violations.

    Authors: We have incorporated the suggested robustness checks into the revised Table 3 and accompanying text. These include specifications that add rainfall controls, more granular seasonal fixed effects, and placebo tests using pre-DETER years and non-Amazon municipalities. The estimated effect remains stable in sign, magnitude, and statistical significance across these checks, reinforcing the headline result of approximately 1,477 prevented homicides per year. revision: yes

Circularity Check

0 steps flagged

No significant circularity; identification uses external cloud-cover instrument

full rationale

The paper identifies causal effects via an instrumental-variables design that exploits exogenous variation in satellite monitoring due to cloud cover as an instrument for DETER enforcement intensity. The headline estimate (approximately 1,477 homicides prevented annually) is obtained from this IV regression applied to homicide data. No step reduces by construction to a fitted parameter, self-defined quantity, or self-citation chain; the instrument is presented as external to local conflict dynamics and the authors' prior fitted values. The derivation chain is therefore self-contained against external benchmarks and receives a score of 0.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard econometric assumptions for instrumental variables rather than new free parameters or invented entities.

axioms (1)
  • domain assumption Cloud cover is independent of violence determinants except through its effect on enforcement intensity.
    Invoked to justify the instrument for enforcement.

pith-pipeline@v0.9.0 · 5634 in / 1052 out tokens · 34893 ms · 2026-05-18T18:56:51.168613+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    We exploit exogenous variation in satellite monitoring generated by cloud cover as an instrument for enforcement intensity... One additional deforestation-related fine is associated with a decrease of approximately 0.73 homicides per 100,000 inhabitants

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

24 extracted references · 24 canonical work pages

  1. [1]

    Johnson, and J

    Acemoglu, D., S. Johnson, and J. A. Robinson (2005). Institutions as a fundamental cause of long-run growth.Handbook of economic growth 1, 385–472. Aldrich, S., R. Walker, C. Simmons, M. Caldas, and S. Perz (2012). Contentious land change in the amazon’s arc of deforestation.Annals of the Association of American Geographers 102(1), 103–128. 36 Alston, L. ...

    work page 2005

  2. [2]

    Azevedo-Ramos, C. and P. Moutinho (2018). No man’s land in the brazilian amazon: Could undesignated public forests slow amazon deforestation?Land use policy 73, 125–127. Barreto, P., A. Pinto, B. Brito, and S. Hayashi (2008).Quem ´ e dono da Amazˆ onia?: uma an´ alise do recadastramento de im´ oveis rurais. IMAZON, Instituto do Homem e Meio Ambiente da Am...

    work page 2018

  3. [3]

    Bragan¸ ca, A. and R. Dahis (2022). Cutting special interests by the roots: Evidence from the brazilian amazon.Journal of Public Economics 215, 104753. Brazilian Forum on Public Safety (2021). Mapping of the violence in the amazon region. Technical Report. Brazilian Ministry of Health (2011). Consolida¸ c˜ ao da base de dados do sistema de informa¸ c˜ oes...

    work page 2022

  4. [4]

    Brown, D. S., J. C. Brown, and C. Brown (2016). Land occupations and deforestation in the brazilian amazon.Land use policy 54, 331–338. Chalfin, A. and J. McCrary (2017). Criminal deterrence: A review of the literature. Journal of Economic Literature 55(1), pp. 5–48. Chimeli, A. B. and R. R. Soares (2017). The use of violence in illegal markets: Evidence ...

    work page 2016

  5. [5]

    Rio de Janeiro: Instituto Brasileiro de Geografia e Estat´ ıstica

    IBGE (2013).Caracter´ ısticas ´ etnico-raciais da popula¸ c˜ ao: classifica¸ c˜ oes e identidades. Rio de Janeiro: Instituto Brasileiro de Geografia e Estat´ ıstica. Acessed in September

    work page 2013

  6. [6]

    Mej´ ıa, and A

    Idrobo, N., D. Mej´ ıa, and A. M. Tribin (2014). Illegal gold mining and violence in colombia.Peace Economics, Peace Science and Public Policy 20(1), 83–111. INPE (2018). Real-time deforestation detection system (deter) – modis data (2004–2017). Accessed in July

    work page 2014

  7. [7]

    Deter system: Methodology for real-time deforestation detection based on modis sensor data

    INPE (2022). Deter system: Methodology for real-time deforestation detection based on modis sensor data. Technical report, National Institute for Space Research (INPE). INPE (2025). Terrabrasilis – painel de monitoramento do desmatamento: Amazˆ onia legal. Acessed in July

    work page 2022

  8. [8]

    IPCC (2023)

    IPCC (2007).Climate Change 2007: Synthesis Report.Geneva, Switzerland: Intergov- ernmental Panel on Climate Change. IPCC (2023). Summary for policymakers. in: Climate change 2023: Synthesis report. IPCC, pp. 1–34. Ipea (2024). Mapa dos homic´ ıdios ocultos no brasil entre 1996 e

    work page 2007

  9. [9]

    Authored by Daniel R

    Discussion Paper 3015, Institute for Applied Economic Research (Ipea), Bras´ ılia, Brazil. Authored by Daniel R. C. Cerqueira and Gabriel O. A. Lins. 40 Ipea (2025). Atlas da violˆ encia

    work page 2025

  10. [10]

    Available athttps://forumseguranca

    Discussion Paper 3015, Institute for Applied Economic Research (Ipea), Bras´ ılia, Brazil. Available athttps://forumseguranca. org.br/wp-content/uploads/2025/05/atlas-violencia-2025.pdf. Jayachandran, S. (2022). How economic development influences the environment.An- nual Review of Economics 14(1), 229–252. Levitt, S. D. (2002). Using electoral cycles in ...

    work page 2025

  11. [11]

    Matsuura, K. and C. J. Willmott (2018b). Terrestrial precipitation: 1900–2017 gridded monthly time series (version 5.01). Retrieved in May

    work page 1900

  12. [12]

    Morton, D. C., R. S. DeFries, Y. E. Shimabukuro, L. O. Anderson, E. Arai, F. del Bon Espirito-Santo, R. Freitas, and J. Morisette (2006). Cropland expansion changes deforestation dynamics in the southern brazilian amazon.Proceedings of the National Academy of Sciences 103(39), 14637–14641. Moutinho, P. and C. Azevedo-Ramos (2023). Untitled public forestla...

    work page 2006

  13. [13]

    Mueller, B. (2018). Property rights implications for the brazilian forest code.Revista de Economia e Sociologia Rural 56(2), 329–346. 41 Pereira, L. and R. Pucci (2022).A Tale of Gold and Blood: The Consequences of Market Deregulation on Local Violence. Insper. Pereira, L., R. Pucci, and R. R. Soares (2024, October). Landing on Water: Air Interdiction, Dr...

    work page 2018

  14. [14]

    Souza-Rodrigues, E. (2019). Deforestation in the amazon: A unified framework for estimation and policy analysis.The Review of Economic Studies 86(6), 2713–2744. Stern, D. I. (2017). The environmental kuznets curve after 25 years.Journal of Bioe- conomics 19(1), 7–28. Stock, J. H., J. H. Wright, and M. Yogo (2002). A survey of weak instruments and weak ide...

    work page 2019

  15. [15]

    We begin by discussing the economic theory of crime. Building on this theoretical foundation, we then examine the escalation of violence in the Brazilian Amazon, with a focus on its links to illegal land use, natural resource exploitation, and conflicts over territorial control. Next, we turn to patterns of deforestation, often driven by the same illicit ...

    work page 1968

  16. [16]

    However, trajectories diverged in the early 2000s, as homicide rates in Amazon municipalities began to rise more rapidly, while rates in non-Amazon areas remained relatively stable

    11 At the beginning of the period, municipalities outside the biome had higher rates of lethal violence. However, trajectories diverged in the early 2000s, as homicide rates in Amazon municipalities began to rise more rapidly, while rates in non-Amazon areas remained relatively stable. By 2016, this divergence had become substantial: average homicide rate...

    work page arXiv 2016

  17. [17]

    Homicide Rate

    Column (1) focuses on assault by firearm (ICD-10 X93–X95), Column (2) focuses on assault by sharp or blunt object (ICD-10 X99–Y00), Column (3) focuses on legal intervention (ICD-10 Y35), and Column (4) focuses on intentional self-harm (ICD-10 X60–X84). “Homicide Rate” is the number of homicides per 100,000 inhabitants. “Lagged Enforcement” refers to the t...

    work page 2006

  18. [18]

    This stability over time is in line with the broader literature documenting that lethal 62 violence in Brazil is overwhelmingly concentrated among men (BFPS and IPEA, 2025)

    The descriptive statistics reveal that the vast majority of homicide victims are men, consistently representing around 87–88% of all cases throughout the period, while women account for roughly 12% and cases of unknown sex remain negligible. This stability over time is in line with the broader literature documenting that lethal 62 violence in Brazil is ov...

    work page 2025

  19. [19]

    Homicide Rate

    Column (1) reports results for men and Column (2) for women. “Homicide Rate” is the number of homicide victims of the corresponding sex per 100,000 inhabitants. “Lagged Enforcement” refers to the total number of fines issued and serves as a proxy for law enforcement effectiveness. The set of control variables con- tains PRODES cloud coverage and non-obser...

    work page 2006

  20. [20]

    ”Homicide Rate” is the number of homicides victims in each municipality per 100,000 inhabitants

    Column (1) reports results for the white and yellow population, Column (2) for the black and mixed-race population, and Column (3) for the indigenous population. ”Homicide Rate” is the number of homicides victims in each municipality per 100,000 inhabitants. ”Lagged Enforcement” refers to the total number of fines issued and serves as a proxy for law enfo...

    work page 2006

  21. [21]

    The descriptive statistics reveal that the majority of homicide victims are 26 years or older, consistently representing around 59–63% of all cases throughout the period. 66 Table D.5: Annual Totals and Shares of Deaths by Age Group Year 0-15 16-25 26-35 36+ Unknown Total Share of Share of Share of Share of Share of 0-15 (%) 16-25 (%) 26-35 (%) 36+ (%) Un...

    work page 2006

  22. [22]

    Homicide Rate

    Column (1) reports results for the 0–15 age group, Column (2) for ages 16–25, Column (3) for ages 26–35, and Column (4) for ages 36 and above. “Homicide Rate” is the number of homicide victims of the corresponding sex per 100,000 inhabitants. “Lagged Enforcement” refers to the total number of fines issued and serves as a proxy for law enforcement effectiv...

    work page 2006

  23. [23]

    16 These conservation policy controls help ensure that other concurrent conservation initiatives do not confound the estimated effect of environmental enforcement on violence

    This policy targeted municipalities with the highest deforestation rates with intensified enforcement actions, including embar- 69 goes, fines, and credit restrictions. 16 These conservation policy controls help ensure that other concurrent conservation initiatives do not confound the estimated effect of environmental enforcement on violence. Table E.1 re...

    work page 2019

  24. [24]

    Homicide Rate

    Column (1) is the main specification; Column (2) includes two conservation policies as additional controls. “Homicide Rate” refers to the number of homicides per 100,000 inhabitants. “Lagged Enforcement” refers to the total number of fines issued and serves as a proxy for law enforcement. The set of control variables contains PRODES cloud coverage and non...

    work page 2006