Deforestation policies in the Brazilian Legal Amazon: an analysis of the PPCDAm using the triple difference method

Abstract

Purpose

This study aims to evaluate the effects of the synergy between monitoring technologies and deforestation control policies promoted by the Action Plan for the Prevention and Control of Deforestation in the Legal Amazon (PPCDAm) during its initial stage.

Design/methodology/approach

The triple difference method is employed to explore the differences between the non-metropolitan municipalities with Indigenous lands and other regional municipalities.

Findings

The findings indicate a reduction of approximately 16.1 km² per municipality between 2004 and 2007. This reduction corresponds to a decrease of 10,293 km² in the area of deforestation and a total of 498 million tons of CO₂. To ensure the robustness of the results, placebo tests, event study and flexibility in the composition of the groups were conducted. The robustness tests substantiate the findings.

Practical implications

These results emphasize the significance of remote monitoring policies for controlling deforestation in isolated regions and Indigenous lands. Additionally, such results indicate that the policy was cost-effective.

Originality/value

This study innovates by examining the causal impact of the initial phase of the PPCDAm before 2008, a period not focused on existing literature. Further, employing the triple difference method innovates methodologically to assess PPCDAm's effect on deforestation in isolated Amazon areas.

Citation

Benevit, B., da Trindade, C.S., de Melo Junior, R.B., Uhr, D.d.A.P. and Uhr, J.G.Z. (2024), "Deforestation policies in the Brazilian Legal Amazon: an analysis of the PPCDAm using the triple difference method", Forestry Economics Review, Vol. 6 No. 2, pp. 122-143. https://doi.org/10.1108/FER-02-2024-0002

1. Introduction

Most emerging countries are home to the planet's rainforests (Araujo et al., 2009; Saatchi et al., 2011), with its deforestation being primarily linked to property rights issues (Alston et al., 1999; Araujo et al., 2009; Fetzer and Marden, 2017; BenYishay et al., 2017). This phenomenon relates to land use and conflicts of interest due to primary goods production in these countries (Hargrave and Kis-Katos, 2013; Assunção et al., 2015). In this sense, Brazil is important in discussing such a theme, covering 70% of the Amazon Forest (Castro et al., 2019) and being home to various Indigenous ethnicities in its domains (Soares-Filho et al., 2010; Walker et al., 2014; BenYishay et al., 2017). The problem of deforestation in the Brazilian Amazon has regained global attention due to the rise of deforestation rates in recent years (Azevedo et al., 2021), as well as changes in laws that have been made in the last decade (Soares-Filho et al., 2014; Azevedo et al., 2021).

The deforestation in Brazil was around 13,853 km² in 2020, and more than 80% occurred in the Brazilian Legal Amazon territory (Azevedo et al., 2021). However, the annual deforestation rate in Brazil presented a drastic reduction over the 2000s, moving from a record rate of 29,059 km² in 1995 to a level of 4,571 km² in 2012 (Arima et al., 2014). The Action Plan for the Prevention and Control of Deforestation in the Legal Amazon – PPCDAm (Brasil, 2003) was responsible for this reduction (Rosa et al., 2012; Hargrave and Kis-Katos, 2013; Arima et al., 2014; Assunção et al., 2023). The PPCDAm promoted institutional changes coordinated between ministries, the private sector and civil society entities, aiming to combat deforestation in the Brazilian Legal Amazon. The PPCDAm was implemented in Brazil in 2004, aiming to monitor environmentally vulnerable areas, restrain the high rates of deforestation registered in the region and promote territorial and land planning. The plan was executed from 2004 to 2020 in four parts.

The first phase deserves particular importance because it promoted strengthening institutions responsible for monitoring and combating environmental degradation and inaugurating the policy plan. One of the central policies of the PPCDAm in its first phase was implementing the Real-Time System for the Detection of Deforestation – DETER in 2004. DETER allowed the monitoring of the Brazilian Amazon region via satellite in near real-time, facilitating the identification and punishment of environmental infractions in the Brazilian Legal Amazon. The initial phase of the PPCDAm promoted the synergy between the use of monitoring technologies and command and control policies of the institutions. This synergy observed between 2004 and 2007 supported subsequent deforestation control policies. After 2007, the favorable institutional environment fostered by PPCDAm led to the establishment of focused policies aimed at rural credit restriction, identification of priority municipalities based on deforestation rates and reforming laws addressing infractions related to environmental crimes. Thus, the analysis of the period preceding the implementation of this set of policies is crucial for comprehending the impact of institutional changes on deforestation in the Brazilian Legal Amazon, stemming from the inception of PPCDAm. In this sense, such changes are highlighted by the improvement of Amazon monitoring and contemporary environmental policies between 2004 and 2007.

This study aims to evaluate the impact of the synergy between monitoring technologies and deforestation control policies promoted by the PPCDAm during its initial stage. The implementation of directed policies started in 2008 and generated effects that superseded the effect of such synergy. Hence, the analysis considers the intervention period from 2004 to 2007. We propose using the Triple Difference method to identify the causal effect of this synergy in the Brazilian Legal Amazon deforestation, considering the isolated context of non-metropolitan municipalities with Indigenous lands concerning other municipalities in the region. The Triple Difference method makes it possible to assess the impact by exploring different variabilities between the characteristics of municipalities, further emphasizing structural heterogeneities between urbanized and isolated regions in the Amazon. Several data sources provide geographic and demographic characteristics, agricultural commodities prices and the municipality's Indigenous land proportion. As a result, we defined the increase in annual deforestation made available by the Project for Monitoring Deforestation in the Legal Amazon (PRODES) of the National Institute for Space Research (INPE) as a variable. The data are arranged at the municipal level, comprising 756 municipalities in the Brazilian Legal Amazon region. As robustness analysis, we propose the application of placebo tests for the treatment and the outcome variable, event study analysis and the flexibility of the composition of the groups.

The results demonstrate that this PPCDAm policies' synergy provided an average annual reduction in the increase in deforestation of up to 16.1 km² per non-metropolitan municipality with Indigenous lands. This result represents a total reduction in deforestation of 10,293 km² during the period from 2004 to 2007. This reduction represented an additional stock of 498 million tons of carbon dioxide (CO₂), equivalent to approximately US$ 2.5 billion. The results remain strong after performing the robustness analyses. These results complement the studies of Assunção et al. (2013, 2015), Hargrave and Kis-Katos (2013), Mello and Artaxo (2017) and Walker et al. (2014), as they indicate that the PPCDAm was effective in containing deforestation in the Brazilian Legal Amazon in the 2000s.

This work contributes to the literature in several ways. First, the article innovates by being the first work in the literature to identify the causal effect of the first phase of the PPCDAm for the pre-2008 period. Most studies in the literature focused on analyzing the policies driven by PPCDAm implemented from 2008 onwards due to their directed design. To our knowledge, no one has examined the time frame addressed in this study. Second, it innovates in methodological terms by applying the triple difference method to identify the PPCDAm's impact on deforestation in the Amazon region's isolated areas. Third, the article explores several sources of variation in the region's increase in deforestation, including the presence, number and proportion of Indigenous lands in the municipalities. Fourth, the article provides new evidence on the importance of monitoring and enforcement policies for the protection of Indigenous peoples in Brazil and highlights the importance of these mechanisms for the effectiveness of protected areas in containing environmental degradation. Finally, the article provides a cost–benefit analysis of the program for the pre-2008 period.

In addition to this introduction, the article presents a policy background in Section 2 and reviews the literature on PPCDAm's effectiveness in Section 3. In Section 4, we present the identification strategy. Section 5 presents the database used and the construction of the variables for the estimations. Section 6 presents the results. Section 7 presents the robustness analysis and its results. Finally, Section 8 presents the policy considerations and final remarks.

2. Policy background

2.1 Brazilian legal Amazon

Brazil is essential for discussing environmental issues in the Amazon as it contains 70% of its forest, corresponding to 28% of the South American subcontinent (Castro et al., 2019). The Brazilian Amazon represents 58.9% of the country's territory and contains the Amazon biomes and part of the Cerrado and Pantanal biomes. The Brazilian Legal Amazon was made official in 1953 by the Brazilian federal government (Brasil, 1953) through law n° 1.806/1953 [1], which also established the Superintendence of the Economic Valorization Plan for the Amazon (SPVEA) to develop the region economically through agricultural, mineral and industrial production. Since 1977, the Brazilian Legal Amazon began to comprehend the regions that currently correspond to the states of Acre, Amapá, Amazonas, Mato Grosso, Pará, Rondônia and Roraima and part of the state of Maranhão (west region of the 44° meridian) [2]. The region has a population of 21.6 million people, with most of its population living in less than 1% of its territory (Silva et al., 2017) and 73% of its inhabitants in urban areas (IBGE, 2011).

The Brazilian Legal Amazon presents significant heterogeneity regarding physiographic attributes, infrastructure, population characteristics, socioeconomic organization and age of the frontier settlement (Garcia et al., 2007). The colonization of the region occurred both spontaneously and in a planned manner. It caused a rapid growth of its urban population since the 1960s (Brown et al., 2016), increasing its proportion from 37% in 1970 to 73% in 2010 (IBGE, 2011). However, the objective of the Brazilian federal government to develop the Amazon region economically with the SPVEA did not promote harmonious development between social, environmental, political and economic aspects (Brown et al., 2016; Mello and Artaxo, 2017). Broadly, the region's municipalities may be categorized as urbanized or rural non-urbanized. The urbanized municipalities are mainly located along the main channel of the Amazon River, or 1970s legacy roads infrastructure (Richards and VanWey, 2015). They have higher population density, educational levels, better sanitation, electricity, water and road infrastructure (Garcia et al., 2007) and clear property rights (Alston et al., 1999). The rural municipalities comprehend the regions deep into the frontier, with lower infrastructure, a more dispersed population and conflicts related to property rights (Alston et al., 1999; Araujo et al., 2009).

The property rights arrangements influence its inhabitants' incentives for production and political influence. Political and bureaucratic actors determine the establishment of such arrangements, facing conflicting political agendas and private interests in civil society. Thus, political bargaining and personal conflicts can compromise an effective property rights institutional arrangement (Alston et al., 1999). The literature on property rights demonstrates that land use in the Brazilian Legal Amazon historically does not have well-defined property rights, a common phenomenon in regions on the frontier of society (Alston et al., 1999; Fetzer and Marden, 2017; Mueller, 2022). The share of landholders exploited by squatters or farmers without the legal title of land property is historically high in the region, representing about 30% of landholdings in 1995 (Araujo et al., 2009). The Brazilian legal framework does not provide adequate protection for landholders' rights, resulting in a large number of landholders without legal titles and increasing the cost of enforcing property rights (Araujo et al., 2009), especially in remote areas (Alston et al., 1999; Araujo et al., 2009).

As tensions and disputes over land tenure due to extractive economic incentives advance in the Brazilian Amazon (Ricketts et al., 2010; Soares-Filho et al., 2010; Assunção et al., 2019), Indigenous people depend on decisions by the Brazilian government for the recognition and guarantee of the right to property in their historic territories. The Indigenous lands are the responsibility of Fundação Nacional do Índio (FUNAI), the Brazilian government agency responsible for the Indigenous people. Law establishes recognizing and demarcating Indigenous lands n° 1,755/96. This process begins with an anthropological survey, and it is based on a presidential decree for its validation and transfer of ownership to the appropriate tribes. However, the bureaucratic and political liability have enabled the prevention and delay of the establishment of Indigenous lands in the Brazilian Legal Amazon due to conflicting political agenda and landowner's interests (Hutchison et al., 2006; Fetzer and Marden, 2017; Assunção et al., 2019).

Similarly to the whole Brazilian Amazon, there is heterogeneity across the Indigenous territories. Some are isolated from the nearest town and not incorporated into the urban society. In contrast, others are clustered within the more urbanized and metropolitan areas (Mueller, 2022). Indigenous lands in isolated Amazon regions (Walker et al., 2014) tend to be vulnerable to illegal extractivist actors (Walker et al., 2014; BenYishay et al., 2017; Begotti and Peres, 2020). Additionally, municipalities with Indigenous lands also indicate heterogeneity in representativeness, varying in the numbers of Indigenous lands, tribes and the size of land territory within their perimeter. The lack of scrutiny in isolated regions tends to be an aggravating factor in threatening Indigenous lands' property rights and environmental integrity (Aldrich et al., 2020).

Brazil's land reform policies also maintain the risk of losing property rights, even for title landholders. In such a context, deforestation becomes part of a risk management strategy, where agents seek short-term returns in extractivist activities or establish property rights through agricultural activities (Alston et al., 1999; Araujo et al., 2009; Silva et al., 2017). Consequently, establishing institutions and policies that guarantee property rights is fundamental for contingency conflicts and deforestation related to land use in the Amazon (Alston et al., 1999; Fetzer and Marden, 2017; Assunção et al., 2019; Mueller, 2022).

2.2 First phase of the PPCDAm

The deforested area in the Brazilian Legal Amazon already corresponded to 837,000 km² in 2001 (Malhi et al., 2008; Mello and Artaxo, 2017; Soares-Filho et al., 2006). According to data from PRODES/INPE, the annual deforestation in the Amazon region grew from 17,383 km² in 1998 to 25,396 km² in 2003 (Fundo Amazônia, 2012). Faced with such a scenario, the Presidential Decree of July 3, 2003, was signed, establishing a Permanent Interministerial Working Group (WG) to establish measures and coordinate actions to reduce deforestation rates in the Legal Amazon [3] (Brasil, 2003). The Interministerial WG planned a set of actions by various public authorities to curb the deforestation of the Brazilian Amazon rainforest, resulting in the creation of PPCDAm in 2004. The PPCDAm was a strategic initiative of the Brazilian government that established guidelines and priorities under the Sustainable Development Plan for the Amazon – PAS.

The PPCDAm was conceived by integrating several ministries [4]. The Ministry of the Civil House of the Presidency of the Republic coordinated the Interministerial WG that formulated the plan. The Interministerial WG on Amazon Deforestation supported the coordination between different spheres of the public sector and subsidized the establishment of the four working subgroups responsible for elaborating strategic proposals in their respective areas of activity. This way, specialized institutions such as the Federal Police, the Federal Highway Police, the Brazilian Army and INPE could work together to combat deforestation. The working subgroups were divided into (1) Territorial Land Ordinance, working on territorial planning instruments with a focus on land policy, conservation units and sustainable development strategies; (2) Monitoring and Control, acting in instruments for monitoring, licensing and inspection of deforestation, burning and logging; (3) Promotion of Sustainable activities, operating in rural credit and tax incentives, technical assistance and rural extension and scientific and technological research; and (4) Infrastructure, working on infrastructure policies, focusing on the transport and energy sectors [5].

The federal and state governments jointly coordinated the strategic planning of infrastructure works to promote infrastructure development in the Amazon region along with socio-environmental responsibility. This joint action mitigated the environmental degradation caused by the construction of highways as in previous decades. It fostered the planning and execution of preventive, mitigating and compensatory measures to be carried out in the works (PPCDAm, 2004). However, the infrastructure axis migrated to PAS in 2004, concentrating the PPCDAm's activities on activities directly related to illegal deforestation in the Amazon, with emphasis on productive activities linked to forest management, extractivist, recovery of degraded areas and productive intensification of open areas (PPCDAm, 2009). So far, the plan has had four phases: Phase I (2004–2008), Phase II (2009–2011), Phase III (2012–2015) and Phase IV (2016–2020). Several policies and institutional changes have been implemented during its phases. Table 1 presents the most notable PPCDAm policies implemented from its conception until the end of Phase I.

DETER was one of the main instruments for improvement in the monitoring and deforestation control areas (Assunção et al., 2023). The system was designed by INPE and enabled faster and more effective monitoring of the forest cover of the Brazilian Legal Amazon via satellite through the georeferenced MODIS [6] (Moderate-Resolution Imaging Spectroradiometer) images generated every two weeks. DETER allowed the identification of shallow forest cuts, forest degradations in preparation for future deforestation and scars of forest fires with a minimum size of 25 hectares (Azevedo et al., 2021). The region's monitoring relied on voluntary reports indicating each area's condition within the Legal Amazon before the implementation of DETER, imposing limitations on IBAMA's prompt legal repression against environment infractors. The innovation of this system allowed near real-time monitoring and detection of deforestation events through satellite images. This georeferenced data provided the federal government and states with subsidies for identifying new deforestation occurrences and mapping critical areas to guide inspection actions and contain environmental crimes. These changes made it possible to overlay labor, environmental, fiscal, tax and land ownership information to track activities linked to illegal logging. The federal government's actions aimed to combat public land grabbing, create new conservation units (sustainable use or integral protection) and demarcate and approve Indigenous lands. It should be noted that this axis prioritized the regions of the “Arc of Deforestation,” especially in the vicinity of the BR-163 (Santarém-Cuiabá Highway). The federal government acted with the state governments (Pará, Mato Grosso, Rondônia and Acre) and civil society entities to execute the ecological-economic zoning (ZEE) [7] along the Deforestation Arch and the area of influence of BR-163. Further, the government encouraged the expansion of sustainable activities in deforested areas, as indicated by the ZEE (PPCDAm, 2004).

Several results in its axes of action were achieved during the first phase of the PPCDAm due to a governmental environment favorable to institutional changes committed to combating deforestation (PPCDAm, 2009). The edition of Ordinance MDA/INCRA n° 10 in December 2004 determined the re-registration of rural properties in municipalities in the Amazon whose declaration of legal status was characterized by possession [8] by simple occupation, and more than 60,000 rural property titles were inhibited (PPCDAm, 2009). Law 11,132 was sanctioned [9] in July 2005 to amend the National System of Nature Conservation Units (SNUC) Law. It established the instrument of “provisional administrative limitation” of areas to perform studies for the creation of UCs in conflict zones. Conservation Units in the Brazilian Legal Amazon increased by more than 520,000 km² n between 2004 and 2009, and about 43% of the Legal Amazon area was considered a protected area in 2010 (Assunção et al., 2015). The Brazilian government homologated 48 new Indigenous Lands during the first phase of the PPCDAm period (CIMI, 2009), occupying more than 10 million hectares. Brazil had 624 Indigenous lands officially recognized by the government in 2007, representing about 13% of the territory (CIMI, 2007). More than 25 million hectares of conservation units strategically located near the Arc of Deforestation were created between 2004 and 2008. In the same period, Deforestation in the Integral Protection Conservation Units reduced from 499 km² to 119 km², from 1,277 km² to 435 km² in the Sustainable Use Conservation Units and from 567 km² to 398 km² in Indigenous Land.

The monitoring and control axis promoted several technological and institutional advances. The DETER creation and PRODES' improvement provided subsidies for strategic actions to combat deforestation strategically and quickly. Decree 6,321 of December 2007 established priority municipalities with high deforestation rates based on three criteria: (1) total deforested area, (2) total deforested area in the last three years and (3) an increase in the deforestation rate in at least three of the last five years. Priority municipalities suffered indirect consequences, such as the refusal of slaughterhouses to purchase cattle from legally irregular farms, more significant restrictions on obtaining rural credit and the requirement for greater effort in more sustainable production. In addition, Decree 6,514 of July 2008 established more detailed and objective infractions and administrative sanctions related to environmental crimes, providing federal administrative processes to investigate such infractions and the appropriate measures. Thus, IBAMA started to adopt new inspection methodologies in the Brazilian Legal Amazon, working with planning operations in priority areas and jointly with the Army, the Federal Police and the Federal Highway Police. It incurred greater effectiveness in the seizure of illegal wood, application of fines and fight against corruption, resulting in the arrest of more than 600 public servants who committed crimes against the environment and public order (PPCDAm, 2009). The National Monetary Council – CMN implemented Resolution No. 3,545 in 2008, intending to mitigate the effects of the expansion of agribusiness on deforestation in the region (Assunção et al., 2013, 2015; Hargrave and Kis-Katos, 2013). This institutional change imposed restrictions on rural credit in the Amazon by requiring proof of compliance with environmental legislation on the credit takers.

The promotion of the sustainable activities axis during the first phase of the PPCDAm attained the institution of the Public Forest Management Law (Law 11,284/06) [10], inducing greater transparency in identifying public forests and facilitating the forest concession process. The federal agency Brazilian Forest Service became responsible for managing public forests in Brazil and the National Forestry Development Fund – FNDF. It promoted the development of sustainable forest-based activities in Brazil and technological innovations in the sector. The Green Protocol [11] was improved with its implementation by public and private banks responsible for complying with environmental laws in their credit operations. Additionally, the federal government provided labor training focused on intensive agriculture (in areas already deforested) and forest management, acting in partnership with state governments, civil society and the business sector (PPCDAm, 2004). Further, the implementation of Law 11,284/06 resulted in the first public bidding for a forest concession in Flona Jamari (RO) and the Sustainable Forest District of BR-163 (PPCDAm, 2009) was created.

According to the PPCDAm report (2004), the total budget provisioned in 2004 for actions to combat illegal deforestation was R$ 394 million, of which R$ 244.3 million (62%) were allocated to land and territory, R$ 82.7 million (21%) to the monitoring and control axis and R$ 67 million (17%) to the promotion of sustainable activities. Compared to the amount allocated to the monitoring and control axis, R$ 4.7 million (1.2%) was allocated to improving monitoring systems and financing the planning, development and installation of DETER. Between 2004 and 2007, INPE, responsible for developing and using DETER, had its annual Budget with Costing and Capital (OCC) increase from R$ 41.8 million in 2004 to R$ 116.8 million in 2007. Furthermore, the annual budget provisioned for IBAMA also increased from R$570 million in 2004 to more than R$1.1 billion in 2007. Converting to dollars, the two institutes spent more than U$ 1.8 billion [12] in this period.

According to the 2012 Amazon Fund's Annual Activity Report, deforestation in the Legal Amazon decreased substantially from the second half of the 2000s. This behavior can be seen in Figure 1. Several studies consider the introduction of the PPCDAm as the main reason for this change of course (Arima et al., 2014; Sills et al., 2015; Mello and Artaxo, 2017; Assunção and Rocha, 2019; Assunção et al., 2023). Thus, the empirical literature indicates that the policies to combat deforestation implemented by the PPCDAm helped contain deforestation and reduce environmental devastation in the region from the second half of the 2000s onwards.

The introduction of the “New Forest Code” in 2012 through Federal Law n° 12.651/2012 [13] 2012 established changes in the requirements that characterize PPAs and LRs, providing conditions for amnesty for illegal deforestation committed by small rural properties [14] until July 2008. Soares-Filho et al. (2014) highlight that these changes qualified 90% of rural producers for amnesty, resulting in the forgiveness of 58% of Brazil's “environmental debt.” According to the authors, leniency with environmental crimes provided by the 2012 Forest Code represented an institutional risk. Deforestation remained stable between 2013 and 2018 (Azevedo et al., 2021). The 2020 Annual Report on Deforestation in Brazil (2021) indicates that from 2018 onwards, environmental degradation in the Amazon region increased again, with a 30% increase in the number of alerts issued by DETER in the Amazon region between 2018 and 2019. Deforestation rates rose again in 2020 compared to the previous year, reaching the mark of 13,853 km², consolidating the annual rate of deforestation in the Brazilian Amazon at a level three times higher than the 4,571 km² recorded in 2012 (Azevedo et al., 2021).

3. Literature review

The introduction of the PPCDAm provided a series of institutional changes that enforced the property rights of the Indigenous people by inhibiting conflicts due to illegal extractivist activities through command and control strategies (Soares-Filho et al., 2010; Arima et al., 2014; Assunção et al., 2015, 2023). Considering the geographically isolated location of the Indigenous tribes in the Amazon (Walker et al., 2014; BenYishay et al., 2017), the DETER was a great tool in providing greater state surveillance for Indigenous lands. DETER offered an important monitoring instrument for isolated regions with little integration with the rest of society (Walker et al., 2014), subsidizing greater inspection and potentially increasing the scrutiny of the rest of society regarding conflicts. They were related to illegal deforestation in such regions (Aldrich et al., 2020). Indigenous people were vulnerable before implementing the PPCDAm due to the lack of prompt monitoring (Walker et al., 2014) and regulation around Indigenous lands (BenYishay et al., 2017). Additionally, Börner et al. (2015) highlight the importance of enforcement that guarantees Indigenous land ownership for Indigenous peoples. The authors also reinforce that property vulnerability is associated with the deforestation of these lands and neighboring municipalities and that the isolation of Indigenous peoples can make Indigenous Lands more susceptible to extractive threats. Ricketts et al. (2010) report that the probability of deforestation within Indigenous lands or protected areas (TIAP) is between 7 and 11 times lower concerning areas around them. The authors also emphasize that the TIAPs established between 2003 and 2007 in the Amazon Brazilian law can prevent deforestation of up to 272,000 km² by 2050, equivalent to one-third of CO₂ in the world.

Regarding the economic drivers of deforestation, several studies have analyzed the relationship between agricultural production and deforestation. Hargrave and Kis-Katos (2013) addressed the relationship between the expected profitability of agricultural production, the environmental policies of the PPCDAm and deforestation between 2002 and 2007. The authors assess how land use methods are affected by variations in beef, soy and wood prices, as well as by the actions of the environmental police (IBAMA) and by the flow of rural credit. It is identified that the greater availability of agricultural credit and the increase in soy and cattle prices are associated with higher deforestation rates. On the other hand, environmental policing was effective in reducing deforestation. The authors identify that the percentage increase in the intensity of fines resulted in a reduction of about 0.5% in deforestation. Assunção et al. (2013) article verifies the impact of resolution n° 3,545/2008 on rural credit and employs the difference-in-differences (DID) strategy. The results show that the institutional change caused a reduction in the granting of rural credit in the Amazon biome. Counterfactual simulations indicate that the deforestation of 2,700 km² between 2009 and 2011 was avoided due to such reduction.

Arima et al. (2014) study assesses the impact of Ordinance MMA n° 28/2008 on deforestation in the Brazilian Legal Amazon using DID and propensity score methods. It compares priority municipalities to municipalities outside the list. The results indicate that increased inspections in the list's municipalities reduced deforestation between 2,304 and 10,653 km² between 2009 and 2011, equivalent to a stock of 110 million to 528 million tons of carbon. Assunção and Rocha (2019) also checked the effectiveness of the MMA Ordinance n° 28/2008 with the DID method, controlling for agricultural prices and the share of protected areas in each municipality. The study demonstrates that the deforestation of 11,396 km² in priority municipalities was avoided between 2008 and 2011. The main mechanisms that motivated this drop were advances in monitoring and the law's applicability.

The literature indicates that it is advantageous to interact institutional changes favorable to environmental conservation with other determinants, such as monitoring capacity and economic sanctions (Ricketts et al., 2010; Assunção et al., 2013, 2023; Pfaff et al., 2015). Assunção et al. (2015) report that the first phase of the PPCDAm contributed to a significant containment of deforestation. The advent of DETER, Presidential Decrees n° 6,321/2007 and n° 6,514/2008 and CMN Resolution n° 3,545/2008 contributed to such reduction. According to the authors, deforestation would have been 56% higher between 2005 and 2009 without the policies, equivalent to the deforestation of 73,000 km² during the period. Assunção et al. (2023) also assess the impact of the combination of increased monitoring of deforestation resulting from DETER and the total number of fines for environmental infractions applied to result from the implementation of resolution n° 3.545/2008 of the CMN. The study applies two-stage estimates for municipalities in the Amazon biome between 2007 and 2011. Its results show that the increase in the fines applied in a given year significantly reduces deforestation in the following year for the same municipality and prevents the deforestation of 122,700 km² of the Amazon biome, reducing the emission of 900 million tCO₂ annually.

4. Identification strategy

The first phase of the PPCDAm (from 2004 to 2008) affected the whole Brazilian Legal Amazon. It implemented a system for almost real-time remote monitoring of deforestation in the Brazilian Legal Amazon, in addition to promoting advances in inspection, land ordinance and delimitation of environmental protection areas and Indigenous lands. Further, a series of directed policies were implemented in the first phase last year, such as Ordinance MMA n° 28/2008, CMN Resolution n° 3,545/2008 and Presidential Decrees n° 6,321/2007 (December) and n° 6,514/2008. These specific policies have been widely analyzed by other studies due to their directed policy design, enabling the employment quasi-experimental methods of Propensity Score Matching and DID, mostly the latter (Assunção et al., 2013; Arima et al., 2014; Assunção and Rocha, 2019). Nonetheless, the impact of the synergy between monitoring technologies and deforestation control policies promoted by the PPCDAm during its initial stage between 2004 and 2007 represents a gap in the literature related to this meta-policy.

The municipalities of the Legal Amazon present differences in socioeconomic, cultural and infrastructure terms. Therefore, we can explore the variability of certain areas of the region, disaggregating these areas into two groups of municipalities. Those municipalities represent the first group with higher population density, higher income, better infrastructure and close to better quality roads. The second group is municipalities with opposite characteristics, being more isolated and difficult to access regions. An additional source of variability is whether or not Indigenous lands exist within the municipalities of these groups. Indigenous lands in Brazil have specific rules regarding the severity of environmental crimes in their domains. Non-metropolitan municipalities and Indigenous land areas are more sensitive to the innovations brought by DETER. We can assume that the ratio of characteristics between these groups is maintained over time. Thus, non-metropolitan municipalities with Indigenous lands are potentially more affected by PPCDAm than those without Indigenous lands and/or metropolitan.

We propose identifying the causal effect of the synergy between policies implemented by PPCDAm from 2004 to 2007 on deforestation through the Triple Difference (DDD) method (Gruber, 1994). The DDD estimator enables exploring the synergy effect induced by PPCDAm through the relative difference between different types of municipalities: metropolitan vs non-metropolitan and with vs without Indigenous lands. Further, the DDD estimator enables controlling for unobservable economic characteristics between municipalities of the same metropolitan type (metropolitan or non-metropolitan) and institutional and property rights framework within municipalities with or without Indigenous lands (Olden and Møen, 2022). Thus, we verify the effect on non-metropolitan municipalities with Indigenous lands, comparing how the surge of policy synergy resulting from PPCDAm affected more isolated and institutionally vulnerable regions compared to more urbanized and scrutinized municipalities in the Brazilian Legal Amazon. The central hypothesis of the DDD method is the existence of parallel trends in the outcome variable between the groups of municipalities compared to the pre-intervention period [15]. The method allows the addition of covariates, making estimating the causal effect more accurate.

The year 2002 marked the election of President Luis Inácio Lula da Silva, who implemented the PPCDAm in 2004. Concurrently, 2007 is the last year preceding the implementation of a series of directed policies, which can potentially generate effects that supersede the synergy between DETER and the strengthening of environmental institutions. Thus, the analysis encompasses the period from 2002 to 2007.

In terms of econometric specification, the relationship analyzed has the following form:

(1)Yit=β0+β1*NMi+β2*ILi+β3NMi*ILi+λ1*dt+λ2dt*NMi+λ3dt*ILi+λ4dt*NMi*ILi+θit*Xit+μi+πt+εit

where Yi represents the result variable “increase in the area of deforestation” for the municipality “i” of the Brazilian Legal Amazon. The variable NMi represents a dummy variable identifying the non-metropolitan municipality. The variable ILi represents a binary variable identifying if the municipality has Indigenous lands (lands greater than 0 km²). The variable dt identifies the period after the implementation of the environmental policy (2004–2007). The interaction between the environmental policy treatment variables captures the causal effect of the environmental policy change on the outcome variable. The vector Xi represents the covariates for the municipality. The parameters μi and πt represent the municipal and time-fixed effects, respectively. The magnitudes of the effects are captured by the parameters β, λ and θ. The parameter εit represents the error term.

We estimated four different models concerning the vector of covariates considered. The first model considers only municipal fixed effects and time-fixed effects. The second model considers the geographic control covariates (AreaKm2, NoForest, Hydrography, Population, UF, DistanceCapUF, DistanceCapUFSqrd, DistanceCapProx and DistanceCapProxSqrd). The third model considers the geographic control covariates and the agricultural and livestock covariates of the municipalities (price indices of agricultural products and TemporaryTilArea). Finally, the fourth model considers all the previous covariates and adds the covariates referring to the characteristics of Indigenous lands in the municipalities (ILProportion and ILNumber). All models consider cluster-robust standard errors by the municipality.

5. Data

We utilized seven municipal-level databases for the period from 2002 to 2007. The first database comes from PRODES/INPE and provides annual deforestation rates of municipalities in the Brazilian Legal Amazon. The second database is the Registry of Metropolitan Regions, Urban Agglomerations and Integrated Development Regions (RMRUA) for 2010 of the Brazilian Institute of Geography and Statistics (IBGE). The third database is related to the demarcation of Indigenous territories and was prospected through the Terras Indígenas no Brasil website [16]. The fourth database is IBGE's Population Estimates (EstimaPop), which provides annual data on population estimates for each municipality. The fifth database was collected from the Secretary of Agriculture and Supply of the State of Paraná (SEAB-PR) of the Department of Rural Economy (DERAL) and provided the prices of agricultural commodities. The sixth database used in the analysis is IBGE's Municipal Agricultural Production (PAM). Finally, the seventh database was extracted from INPE's TerraBrasilis portal, enabling the generation of geographic coordinates for the municipalities. Table 2 summarizes the names, definitions and data sources of the variables employed in this study.

To measure the effectiveness of the synergy between policies implemented by PPCDAm during its initial stage on deforestation, we used the result variable of deforestation increment (DeforestationIncrement) from the PRODES/INPE database, which provides an estimate of the annual variation of deforestation [17] in km² for 760 municipalities in the Brazilian Legal Amazon. We considered the period from 2002 to 2007 to identify the effect of such synergy from 2004 onwards. A binary variable was created to identify the treatment, assuming a value of one for 2004 to 2007 and zero for the previous years. We estimated the effect of the intervention by the relative differences in DeforestationIncrement between non-metropolitan municipalities with Indigenous Lands and the other municipalities for the post-intervention period. For this purpose, we established the interaction between two binary variables. The first binary variable refers to the region of the municipality (NMetropMunicipality), assuming a value of one when the municipality is located in a non-metropolitan region and zero if the municipality is located in a metropolitan region. The criterion for identifying the metropolitan region was based on the RMRUA for 2010 of the IBGE. The second binary variable was collected through data from the Terras Indígenas no Brasil (“Indigenous Lands in Brazil”) Website. It captures the existence of Indigenous lands in the municipality approved until 2007 (ILMunicipality), assuming a value of one when the municipality has a non-null area of Indigenous Lands (km²) in its domains and zero otherwise [18]. From the interaction of these two variables, we identified non-metropolitan municipalities with Indigenous Lands (NMetropMunicipalityWIL), assuming value one when both variables NMetropMunicipality and ILMunicipality are equal to one and zero otherwise.

The set of covariates considers geographic and population characteristics, prices of agricultural products and the predominance of lands and Indigenous peoples. The demographic data enable the control of heterogeneous demography and geographic aspects among the municipalities in the sample. Therefore, considered the municipalities (1) total area in km² (AreaKm2), (2) non-forest area in km² (NoForest), (3) hydrographic area in km² (Hydrography), (4) population (Population) and (5) federative unit (UF). We prospected the geographic coordinates of the municipalities through the TerraBrasilis/INPE database to control the municipalities' distance to large urban centers, which can be correlated with transport modals and interfere with the incentives for deforestation (Richards and VanWey, 2015). We established the municipality centroids and constructed the covariate of the linear distance between the centroid and the capital of its federative unit (DistanceCapUF) and its quadratic term (DistanceCapUFSqrd). Furthermore, we constructed the covariate of the linear distance between the municipality and the centroid of the closest capital independent of the state (DistanceCapProx) and its quadratic term (DistanceCapProxSqrd).

The agricultural commodities prices also represent a potential incentive for deforestation (Assunção et al., 2015). We used the Secretary of Agriculture and Supply of the State of Paraná (SEAB-PR) data to compute the price indices deflated for 2000 of the same agricultural products as conducted in Assunção et al. (2013). Thus, we considered the price indices of rice, sugarcane, live cattle, cassava, corn and soybeans. Additionally, to conduct a placebo test as a robustness check, we considered the data of municipalities' area of permanent farming in hectares (PermanentFarm) from Municipal Agricultural Production (PAM) data.

Municipalities with Indigenous lands in the Brazilian Legal Amazon present differences from municipalities without Indigenous lands and within their own category. Thus, municipalities with Indigenous lands differ in terms of the heterogeneity of their lands and prevalence within municipal territories. To account for this heterogeneity within municipalities with Indigenous lands, we have also created the covariates indicating the number of Indigenous Lands in the municipality (ILNumber) and the proportion of the sum of the Indigenous Lands area in the municipality in square kilometers (ILProportion).

The final sample considers 756 municipalities [19], comprising 160 non-metropolitan municipalities with Indigenous lands, 554 non-metropolitan municipalities without Indigenous lands and 42 metropolitan municipalities. The descriptive statistics of the variables used in this study are shown in Table 3, containing their mean, standard deviation and minimum and maximum statistics.

6. Results

The hypothesis of identification of the triple difference method assumes the existence of parallel trends between the analyzed groups. To verify the assumption's validity, we examined if the mean difference of trends for the pre-treatment period between groups differed from zero. The result presents an F statistic of 1.91 with a probability of approximately 20% [20], indicating that the parallel trends assumption is not rejected. Thus, we estimate four different triple-difference models. Table 4 presents the main analysis of the effects of the synergy of the policies promoted by PPCDAm between 2004 and 2007 on the increase in deforestation in the Brazilian Legal Amazon. Columns 1–4 present the results in the same order described in Section 4.

In Table 4, all the models present at least 5% confidence significance. Column 1 results indicate that the PPCDAm's synergy implied a reduced deforestation increment of 15.3 km² (p-value <0.05) per municipality on average. After adding covariates, the results in Columns 2, 3 and 4 indicate a reduction in deforestation increment of approximately 16.1 km² (p-value <0.05). This result represents an average annual reduction in deforestation between 2,452 and 2,573 km² in the Brazilian Legal Amazon per municipality. It corresponds to a total reduction between 9,807 and 10,293 km² from 2004 to 2007 [21].

Assunção et al. (2015) report a 56% reduction in deforestation in the Amazon between 2005 and 2009. Also, in line with these results, Soares-Filho et al. (2010) identify that expanding protected areas, especially Indigenous lands, led to a decline in deforestation from 1997 to 2008. Arima et al. (2014) indicate that the increase in inspections in priority municipalities through the Ordinance MMA n° 28/2008 resulted in an average annual reduction in deforestation between 2,304 and 10,653 km² between 2009 and 2011. In another study, Assunção and Rocha (2019) identified that the MMA Ordinance n° 28/2008 avoided the deforestation of 11,396 km² in the priority municipalities between 2008 and 2011. In this sense, our results are within the range of those of Arima et al. (2014) and Assunção and Rocha (2019). They indicate that the synergy between technological and command and control policies promoted by PPCDAm during this period succeeded in the environmental protection of municipalities with Indigenous lands.

7. Robustness analyses

7.1 Placebo tests

We conducted placebo tests to check the robustness of the results. The placebo test checks whether the estimated treatment effect occurred by chance. Thus, the results of these tests should be statistically non-significant. This test checks if the results are related to the treatment variable or the time trajectory. Thus, we performed the placebo test for the treatment period and the outcome variable.

Table 5 reports the results of the placebo tests. Columns 1 and 2 report the results of the placebo test on the variable identifying the treatment period, adding the year 2003 into the treatment period. There should have been no effect during the previous period, as the law was implemented in 2004. Columns 3 and 4 show the results of the placebo test for the outcome variable. For that, the effect on the area of the permanent farming (PermanentFarm) variable was analyzed instead of the deforestation increment. The objective is to test whether implementing policies to combat deforestation impacted other outcome variables unrelated to treatment. We expect no effects, considering that this variable is unrelated to deforestation [22]. These tests verify that the effects found were not “type I errors.” The results of the placebo tests shown in Table 5 were not statistically significant, indicating that the effects identified in the main analysis (Table 4) did not occur by chance.

7.2 Event study

We want to verify the effects of the PPCDAm policies' synergy for each year after its application. Thus, we perform an event study design following the procedure performed in Assunção and Rocha (2019) to evaluate the policies applied from 2004 onwards. This analysis aims to identify the effect of PPCDAm on deforestation for each year after its implementation. All estimated values are within the 95% confidence interval, and we used cluster-robust standard errors [23]. Table 6 presents the significant heterogeneous temporal effects. Columns 1, 2 and 3 present the policy effects of 2005, 2006 and 2007, respectively. We identified a reduction of approximately 13.2 km² (p-value <0.1) in 2005. In 2006 and 2007, more significant effects were identified (p-value <0.01), resulting in reductions in the deforestation increment of 24.9 km² and 23.8 km², respectively. A possible explanation for the lack of effect in 2004 would be the agents' adaptation period for the assimilation of the policy due to the DETER's remote aspect.

7.3 Flexibility in the composition of groups

We want to test the robustness of the treatment variable municipalities for different proportions of Indigenous lands in its territorial composition (ILProportion). Thus, we propose to limit the group of non-metropolitan municipalities with Indigenous Lands according to the percentile of the variable ILProportion. In addition to the original model (independent of ILProportion), three new DDD models were established considering non-metropolitan municipalities with Indigenous Lands with ILProportion (1) equal to or above 25%, (2) equal to or above 50% and (3) equal to or above 75%. Additionally, we performed the Wald test to verify the null hypothesis of equality between the coefficients associated with the DDD effect of the four models.

The estimations are presented in Table 7. Column 1 presents the result for the model with the treatment variable independent of the ILProportion (main analysis estimates), Column 2 presents the result of the treatment variable considering the ILProportion equal to or above 25% and Column 3 presents the result for the treatment variable considering the ILProportion equal to or above 50%. Column 4 presents the result for the treatment variable considering the ILProportion equal to or above 75%. Wald test indicates whether the treatment effect considering different compositions of ILProportion are equal. Columns 1 to 3 show effects with a significance of up to 5% and Column 4 shows an effect of up to 10%. The Wald test presents a χ² statistic of 0.55 with 3 degrees of freedom. The p-value of 0.907 demonstrates the non-rejection of the null hypothesis and indicates statistical equality between the treatment effect coefficients of the four regressions.

8. Final remarks and policy considerations

The debate about climate change, biodiversity and Indigenous integrity gained prominence again with the recent advance in the Brazilian Legal Amazon's deforestation rate. However, the Brazilian Amazon region experienced reduced deforestation rates between 2004 and 2012. Therefore, understanding how public policies have affected deforestation in the region is essential for preserving the Amazon rainforest.

This study aimed to analyze the effects of deforestation in the Brazilian Legal Amazon caused by the synergy between technological and command and control policies fostered by PPCDAm during its first three years of implementation. We explored the variability between the regional municipalities' characteristics through the triple difference method by comparing the annual increase in deforestation between non-metropolitan municipalities with Indigenous lands and other municipalities in the Brazilian Legal Amazon. The results indicate that the group of isolated municipalities was more vulnerable before the PPCDAm implementation, exhibiting greater deforestation reductions than the more urbanized comparison group. The interpretation of this evidence is twofold. First, it can be directly attributed to comparative advantages related to improving monitoring and combat and control policies in such areas. Second, it suggests that PPCDAm reinforced property rights in Indigenous territories through its broad institutional changes.

This article innovated the literature in several ways. First, the article innovates by identifying the causal effect of PPCDAm between 2004 and 2007. Furthermore, it innovates methodologically by using the triple difference method to assess the effect of PPCDAm on deforestation. The article also presented new evidence on the effectiveness of initiating the PPCDAm in safeguarding the forest in municipalities with Indigenous lands in the Amazon. It explored sources of variation, such as the number of Indigenous lands and the proportion of Indigenous land area in the municipalities. The results showed a reduction of potentially 10,293 km² in deforestation in non-metropolitan municipalities with Indigenous lands between 2004 and 2007. Additionally, the analysis of heterogeneous effects indicated that the policy became more effective after 2005.

Regarding carbon stock, we calculate the volume of tons of carbon dioxide emissions per square kilometer (tCO₂/km²) considering our estimates for the amount of deforestation avoided in the Brazilian Legal Amazon. Utilizing the Amazon Fund (2012) reference value of 48,473 tCO₂/km², we estimate that the synergy of PPCDAm's policies provided an additional stock of 498 million tCO_2, representing a value of approximately U$ 2.5 billion [24]. We evaluate the PPCDAm cost-effectiveness considering two annual cost sources. First, using the PPCDAm's (2004) budget for actions to combat deforestation for 2004 (R$ 394 million) [25] as a reference results in an average annual cost of 0.97 U$/tCO2. Second, the main instruments to combat deforestation during the period, which belongs to the monitoring and control axis, are represented mainly by IBAMA's and DETER's (INPE) budget. According to IBAMA's total annual budget and INPE's Cost and Capital Budget (OCC), we estimate an average annual cost of 3.54 U$/tCO₂. Thus, PPCDAm generated a potential surplus between U$ 737 million and U$ 2 billion during the 2004 to 2007 period.

We identify that the institutional changes promoted by PPCDAm in favor of environmental crime monitoring in the Amazon significantly reduced deforestation in municipalities isolated from large urban centers. Regarding the limitations of this study, the databases used do not allow for the classification of the types of environmental crimes that cause the levels of deforestation observed in the municipalities. Additionally, the population size of Indigenous lands was not considered due to the lack of population data for several isolated peoples. We recommend using new methodologies to isolate the causal effect of PPCDAm on deforestation rates in Amazonian municipalities and new analyses to verify the effectiveness of policies that strengthen environmental surveillance and monitoring of Indigenous lands and isolated municipalities.

Figures

Figure 1

Annual deforestation in the Brazilian Legal Amazon (km²)

Table 1

PPCDAm’s Phase I notable policies

Policy	Purpose	Entity	Year
Presidential Decree of July 3, 2003	Creation of a Permanent Interministerial Working Group to establish measures and coordinate actions to reduce deforestation rates in the Legal Amazon	Ministry of the Civil House	2003
PPCDAm	Establish a comprehensive set of strategic measures, guidelines and priorities to be implemented and executed through integrated action among various government institutions, specialized organizations and civil society aimed at conserving the Brazilian Legal Amazon	Interministerial WG; Ministry of the Civil House	2004
DETER	System based on satellite technology that captures and processes georeferenced images of forest cover in near real-time in the Brazilian Legal Amazon	INPE	2004
Ordinance MDA/INCRA n° 10, 2004	Determines the re-registration of rural properties in municipalities in the Brazilian Legal Amazon whose declaration of legal status was characterized by simple occupation possession	MDA/INCRA	2004
Law 11,132/2005	Establish the guidelines for conducting studies for the creation of protected areas in environmentally vulnerable areas	SNUC	2005
Law 11,284/06	Establish institutional facilities and guidelines for the management of public forests and forest concession process	MMA	2006
Decree 6,321, 2007	Establish a list of priority municipalities associated with high deforestation rates in recent years and determine economic sanctions for those included in it	Civil House	2007
Decree 6,514, 2008	Specify detailed and objective environmental violations and institute administrative sanctions for such transgressions	Civil House	2008
Resolution No. 3,545, 2008	Establish the requirement of proof of compliance with environmental legislation for credit recipients, imposing the risk of rural credit restrictions	CMN	2008

Note(s): This table presents notable phase I policies fostered by PPCDAm, providing its policy/regulatory names, purposes, associated entities and year of implementation

Source(s): Table created by authors

Table 2

Variables description

Variable name	Definition	Data source
Outcome variable
DeforestationIncrement	Difference in annual deforestation between years t and t−1 measured in km2	PRODES/INPE
Municipalities types
NMetropMunicipalityWIL	Group of non-metropolitan municipalities with Indigenous lands, assuming value 1 if the municipality is non-metropolitan and has at least one Indigenous land, and 0 otherwise	Terras Indígenas no Brasil; RMRUA
NMetropMunicipalityWoIL	Group of non-metropolitan municipalities without indigenous lands, assuming value 1 if the municipality is non-metropolitan and has at least one Indigenous land, and 0 otherwise	Terras Indígenas no Brasil; RMRUA
MetropMunicipality	Group of metropolitan municipalities, assuming value 1 if the municipality metropolitan, and 0 otherwise	Terras Indígenas no Brasil; RMRUA
Covariates
Population	Population of municipality in year t	EstimaPop
AreaKm2	Total area in km2	PRODES/INPE
NoFlorest	Municipality non-forest area in km2	PRODES/INPE
Hydrography	Municipality hydrographic area in km2	PRODES/INPE
UF	Categorical binary variable indicating municipality federative unit	PRODES/INPE
DistanceCapUF	Straight line distance from municipality to its own federative unit capital	TerraBrasilis
DistanceCapUFSqrd	Squared value of the variable DistanceCapUF for the municipality	TerraBrasilis
DistanceCapProx	Straight line distance from municipality to the nearest federative unit capital	TerraBrasilis
DistanceCapProxSqrd	Squared value of the variable DistanceCapProx for the municipality	TerraBrasilis
RiceIndex2000	Rice price indices in year t deflated for 2000	SEAB-PR
SugarcaneIndex2000	Sugarcane price indices in year t deflated for 2000	SEAB-PR
CattleIndex2000	Live cattle price indices in year t deflated for 2000	SEAB-PR
CassavaIndex2000	Cassava price indices in year t deflated for 2000	SEAB-PR
CornIndex2000	Corn price indices in year t deflated for 2000	SEAB-PR
SoybeanIndex2000	Soybean price indices in year t deflated for 2000	SEAB-PR
PermanentFarm	Municipality permanent farming area in year t in hectare	PAM
ILNumber	Municipality number of Indigenous lands	Terras Indígenas no Brasil
ILProportion	Proportion of Indigenous land area in the total municipality area	Terras Indígenas no Brasil; PRODES/INPE

Note(s): This table presents the description of variables names and data source

Source(s): Table created by authors

Table 3

Descriptive statistics

Variable	Mean	S.D.	Min	Max
Outcome variable
DeforestationIncrement	28.193	80.408	0	1407.80
Municipalities types
NMetropMunicipalityWIL	0.212	0.409	0	1
NMetropMunicipalityWoIL	0.733	0.443	0	1
MetropMunicipality	0.056	0.229	0	1
Covariates
Population	28829.34	92908.65	981	1,688,524
AreaKm2	6684.467	13892.57	64	159,540
NoFlorest	1266.804	2407.819	0	19780.8
Hydrography	149.300	431.435	0	4499.9
DistanceCapUF	324.976	238.172	0	1485.384
DistanceCapUFSqrd	162322.9	246075.7	0	2,206,364
DistanceCapProx	281.3542	165.651	0	902.557
DistanceCapProxSqrd	106594.4	113,722	0	814609.1
RiceIndex2000	216.950	49.782	151.681	296.464
SugarcaneIndex2000	172.397	30.241	123.964	219.615
CattleIndex2000	135.119	12.657	111.244	151.948
CassavaIndex2000	167.483	72.805	67.912	284.257
CornIndex2000	137.555	14.410	119.748	164.541
SoybeanIndex2000	182.313	25.918	151.652	220.662
PermanentFarm	8.247	22.413	0	304.87
ILNumber	0.522	1.457	0	14
ILProportion	0.049	0.140	0	0.999

Note(s): Descriptive state statistics have been omitted for space considerations. The municipality types presented in this table are the group of non-metropolitan municipalities with Indigenous lands (NMetropMunicipalityWIL), the group of non-metropolitan municipalities without Indigenous lands (NMetropMunicipalityWoIL) and metropolitan municipalities (MetropMunicipality)

Source(s): Table created by authors

Table 4

PPCDAm synergy effect – 2002–2007

	(1)	(2)	(3)	(4)
DDD effect	−15.324** (6.709)	−16.083** (6.761)	−16.083** (6.761)	−16.069** (6.762)
Cov. Demographics	NO	YES	YES	YES
Cov. Agriculture	NO	NO	YES	YES
Cov. Indigenous lands	NO	NO	NO	YES
FE municipalities	YES	YES	YES	YES
FE time	YES	YES	YES	YES
N	4,536	4,536	4,536	4,536

Note(s): This table presents the results for the triple difference regressions. The symbols ^*, ^** and ^*** represent statistical significance of 10%, 5% and 1%, respectively. Column 1 presents the results for the model controlling for municipal and temporal fixed effects. Column 2 presents the results for the model controlling for fixed effects and demographic covariates. Column 3 presents the results for the model controlling for fixed effects, demographic covariates and agriculture covariates. Column 4 presents the results for the model controlling for fixed effects, demographic covariates, agropecuary covariates and characteristics of Indigenous lands covariates. The values in parentheses are standard deviations of the coefficient. Covariate coefficients were omitted for space considerations

Source(s): Table created by authors

Table 5

Placebo tests – 2002–2007

	(1)	(2)	(3)	(4)
Placebo DDD effect	0.656 (11.936)	−0.096 (22.899)	0.346 (0.860)	0.020 (0.942)
Fixed effects	YES	YES	YES	YES
Covariates	NO	YES	NO	YES
N	4,536	4,536	4,536	4,536

Note(s): The symbols ^*, ^** and ^*** represent statistical significance of 10%, 5% and 1%, respectively. Columns 1 and 2 present the placebo test results on the variable identifying the treatment period using 2003 as a placebo. Columns 3 and 4 present the placebo test results using the municipality’s permanent crop area as the result variable. Covariate coefficients were omitted for space considerations

Source(s): Table created by authors

Table 6

Heterogeneous effects analysis – 2002–2007

	(1)	(2)	(3)
DDD effect	−13.159* (6.975)	−24.926*** (8. 217)	−23.814*** (8.829)
Fixed Effects	YES	YES	YES
Covariates	YES	YES	YES
N	4,536	4,536	4,536

Note(s): The symbols ^*, ^** and ^*** represent statistical significance of 10%, 5% and 1%, respectively. Columns 1, 2 and 3 present the results for analyzing heterogeneous effects for 2005, 2006 and 2007, respectively. Covariate coefficients were omitted for space considerations. No significant effect was identified for the year 2004

Source(s): Table created by authors

Table 7

PPCDAm synergy effect conditioned to IL proportion – 2002–2007

	(1)	(2)	(3)	(4)
	ILProp. > 0%	ILProp.> 25%	ILProp. > 50%	ILProp. > 75%
DDD effect	−16.069** (6.762)	−19.211*** (6.737)	−20.398** (8.599)	−15.900* (8.766)
Fixed effects	YES	YES	YES	YES
Covariates	YES	YES	YES	YES
N	4,536	4,536	4,536	4,536
Wald test (p-value)	0.55 (0.907)

Note(s): The symbols ^*, ^** and ^*** represent statistical significance of 10%, 5% and 1%, respectively. Columns 1–4 present the results for the analysis with the treatment variable considering non-metropolitan municipalities with indigenous lands with ILProportion above 0%, 25%, 50% and 75%, respectively. The Wald test verifies the null hypothesis of equality between the coefficients of the four models. The Wald test statistic is distributed as χ². Covariate coefficients were omitted for space considerations

Source(s): Table created by authors

Notes

1. The law can be accessed at https://www.planalto.gov.br/ccivil_03/leis/1950-1969/l1806.htm.

2. This event occurred after the dismemberment of the state of Mato Grosso from Complementary Law n° 31/1977. The law can be accessed at http://www.planalto.gov.br/ccivil_03/leis/lcp/lcp31.htm.

3. The decree can be accessed at http://www.planalto.gov.br/ccivil_03/dnn/2003/dnn9922.htm.

4. The following ministries integrated the PPCDAm: Ministry of Agriculture, Livestock and Supply (MAPA), Ministry of Defense (MD), Ministry of Development, Industry and Foreign Trade (MDIC), Ministry of Environment (MMA), Ministry of Finance (MF), Ministry of Integration (MI), Ministry of Justice (MJ), Ministry of Labor and Employment (MTE), Ministry of Mines and Energy (MME), Ministry of Science, Technology and Innovation (MCT) and Ministry of Agrarian Development (MDA).

5. The bodies responsible for each work subgroup were: (1) Secretariat of Policies for Sustainable Development – SDS/MMA (coordinator), Secretariat for the Coordination of the Amazon – SCA/MMA, MDA, MI, MDIC, MD and MJ/FUNAI; (2) IBAMA/MMA (coordinator), SCA/MMA, MCT, MD, MJ, TEM and Amazon Protection System – Sipam/CasaCivil; (3) Secretariat of Biodiversity and Forests – SBF/MMA (coordinator), SCA/MMA, MDIC, MAPA, MDA, MCT, MI, TEM and MF (guest); and (4) SCA/MMA (coordinator), MT, MME, MAPA, MI and MDIC.

6. MODIS is a space instrument launched from Earth by NASA and started in 1999 and is part of the Earth Observing System (EOS). The program provides remote sensing data of high temporal and spectral resolution and moderate spatial resolution.

7. Established by Decree No. 4,297/2002, the ZEE establishes measures and standards for environmental protection in order to ensure environmental quality, water resources and soil and the conservation of biodiversity, promoting sustainable development and improving living conditions.

8. Squatter without title document, promising buyer who holds possession and holder of possession arising from a concession of use provided by the Federal, State or Municipal Government.

9. The law can be accessed at https://www.planalto.gov.br/ccivil_03/_ato2004-2006/2005/lei/l11132.htm.

10. The law can be accessed at http://www.planalto.gov.br/ccivil_03/_ato2004-2006/2006/lei/l11284.htm.

11. The Green Protocol is a letter of principles signed in 1995 by Brazilian financial institutions in favor of measures in harmony with sustainable socio-environmental development.

12. Value corrected according to the free exchange rate of the US Dollar (sale) as provided for at the address: https://www3.bcb.gov.br/sgspub/consultarvalores/consultarValoresSeries.do?method=consultarValores.

13. The law can be accessed at http://www.planalto.gov.br/ccivil_03/_ato2011-2014/2012/lei/l12651.htm.

14. The size of properties considered small varies between 20 hectares, for the Brazilian Southeast, up to 440 hectares, for the Brazilian Legal Amazon.

15. The difference-in-differences method has been adopted in other evaluations of policies to combat deforestation and also relies on parallel trends (Assunção et al., 2013; Arima et al., 2014; Assunção and Rocha, 2019).

16. The data can be accessed at the address: https://terrasindigenas.org.br.

17. These estimates are calculated by the analysis of images captured between August 1st and July 30th of the following year.

18. Only indigenous lands homologated until 2008 were considered.

19. Four municipalities that did not have agricultural data in the PAM database were excluded, resulting in a sample of 756 municipalities.

20. The test was performed with cluster-robust standard errors by the federative unit.

21. We consider the 160 non-metropolitan municipalities with indigenous lands in the sample.

22. According to Assunção et al. (2015), rice, sugarcane, cassava, corn and soybean (temporary) harvests corresponded to approximately 70% of the harvest in the region between 2002 and 2009.

23. We cluster the observations at the municipal level.

24. We consider the standard price of 5 U$/tCO₂ commonly used at the time (Fundo Amazônia, 2012; Assunção et al., 2023).

25. We assume a budget growth for the following years in the same proportion as that observed in IBAMA's budget in the same period. The dollar amounts have been corrected according to the free exchange rate of the US Dollar (sale) for each year as provided at the address: https://www3.bcb.gov.br/sgspub/consultarvalores/consultarValoresSeries.do?method=consultarValores.

Competing interests: The authors have no competing interests to declare that are relevant to the content of this article.