Policy Research Working Paper 10700 Indigenous Peoples, Land and Conflict in Mindanao, Philippines Lucia Madrigal Jose Cuesta Sergio Somerville Social Sustainability and Inclusion A verified reproducibility package for this paper is Global Practice available at http://reproducibility.worldbank.org, February 2024 click here for direct access. Policy Research Working Paper 10700 Abstract This article explores the links between conflict, land and including land- and resource-related conflicts, are gener- indigenous peoples in several regions of Mindano, the ally less likely in districts (barangays) with higher shares Philippines, notorious for their levels of poverty and con- of indigenous peoples. Ancestral domain areas also have a flict. The analysis takes advantage of the unprecedented lower likelihood for general conflict but a higher likelihood concurrence of data from the most recent, 2020, census; for land-related conflict. Ancestral domains titling does not an independent conflict data monitor for Mindanao; and automatically solve land-related conflicts. When adminis- administrative sources on ancestral land titling for indige- trative delays take place (from cumbersome bureaucratic nous peoples in the Philippines. While evidence elsewhere processes, insufficient resources and weak institutional compellingly links land titling with conflict reduction, a capacity), titling processes may lead to sustained, rather more nuanced story emerges in the Philippines. Conflicts, than decreased, conflict. This paper is a product of the Social Sustainability and Inclusion Global Practice.. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at jcuesta@worldbank.org. A verified reproducibility package for this paper is available at http://reproducibility. worldbank.org, click here for direct access. RESEA CY LI R CH PO TRANSPARENT ANALYSIS S W R R E O KI P NG PA The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Indigenous peoples, land and conflict in Mindanao, Philippines Lucia Madrigal, Jose Cuesta, Sergio Somerville 1 World Bank Social Sustainability and Inclusion Global Practice Key words: Land, Conflict, Indigenous Peoples, Titling, Philippines JEL classification : D74, Q15 1. Introduction Poverty and conflict are intimately related in Mindanao. The poverty rate on the island – located in the southern region of Philippines and comprising Caraga, Davao, and the Bangsamoro Autonomous Region in Muslim Mindanao (BARMM) – is notably higher than the national average. In Caraga and BARMM, it is roughly double (Philippine Statistics Authority (PSA), 2021). Moreover, Mindanao has experienced protracted conflict between the state and insurgent groups since the late 1960s, as well as intermittent periods of violence between dozens of militia units, political groupings, communities, and clans (Adriano and Parks, 2013; Herbert, 2019). A wide range of evidence suggests that these two challenges are inextricably linked: Mindanao’s conflict zones have higher poverty levels and lower levels of economic growth than anywhere else in the Philippines. Land is also often at the heart of these challenges. Across developing countries, weak land governance is often a key driver of poverty (World Bank, 2017). Land governance in the 1 This article does not necessarily represent the views of the World Bank not its Board of Directors. Authors would like to thank Paul Adolfo, Fermin Adriano, Annabelle Corpuz, Giorgi Englebrecht, Ditte Fallesen, Craig Hammer, Elaine Chee En Hui, Ravi Kumar, Francisco Lara, Greg Larson, Carlos Perez-Brito, Mary Ann Pollisco-Botengan, Divyanshi Wadhwa for comments and feedback on previous drafts and Nadia Belghith and Sharon Piza for sharing data without which this analysis could have not been possible. Remainig errors are the sole responsibility of the authors. Philippines is weak, but even weaker in Mindanao (LGI, 2016). Disputes over land and resources are often the primary trigger of conflict in Mindanao, and conflict also weakens land governance (World Bank, 2023a). As in other developing countries, this creates a vicious cycle leading to a range of challenges, including weak job creation and underemployment (Ravanera, 2018) – which in turn exacerbate conflict, as landless people without work opportunities are more likely to join armed groups (World Bank, 2023a). Unsurprisingly, Indigenous Peoples (IPs) are particularly vulnerable to these challenges. Across the Philippines, IPs disproportionately reside in areas of high poverty and conflict (World Bank, 2023b and Reyes, Mina, and Asis, 2017). This is hardly surprising, given that IPs are disproportionately exposed to poverty and conflict around the world (UN, 2010; 2022). Recognizing these challenges in the Philippines, the Indigenous Peoples Rights Act (IPRA) gives IPs the right to manage their ancestral domains – which, in Mindanao, amounts to an estimated one-third of all land on the island. Despite this achievement, however, the process of obtaining Certificates of Ancestral Domain Titles (CADTs) is long, arduous, and complex, and many IPs are still struggling to gain ownership of their lands. While scarce data has prevented rigorous analysis of how these issues interact, new opportunities are emerging. At the global level, there is a rich empirical literature on how conflict and weak land governance can exacerbate poverty, primarily for IPs, but only small number of studies have analyzed these issues in Mindanao. The recent release of new geo-referenced data on violent incidents, however – made available by International Alert Philippines’ Conflict Alert (CA) – makes it possible to explore these linkages for the first time. Of course, more detailed and higher-quality data would enable more rigorous analysis. Likewise, the new data on conflict are currently only available for Mindanao; as data collection expands, similar analysis could be conducted for other regions. To conduct our analysis, we construct a novel dataset by merging the new data on conflict with census data and existing CADT records. We then use mapping and a suite of simple quantitative tools to analyze the patterns and interactions between violent incidents, CADTs, and the share of IPs in a given barangay (the smallest administrative unit available). Our analysis has three key findings: (i) in general, conflicts are less likely in barangays with higher shares of IPs; (ii) Land- and resource-related conflicts are also less likely in barangays with higher shares of IPs; and (iii) overall, CADT areas have a lower likelihood for general conflict but a higher likelihood for land- related conflict, particularly in CADTs that experience processing delays. These findings add depth and complexity to the current understanding of conflict, poverty, land, and IPs in Mindanao, in particular, which can be relevant to other parts of the world characterized by weak land governance. They offer policymakers a more nuanced view of these issues, while highlighting key data gaps and underscoring the need for more research. The rest of the article is structured as follows: Section 2 summarizes the context and reviews the relevant literature. Section 3 describes the evidence we gathered and our methodology. Section 4 describes our key results. The final section concludes. 2 2. Context and literature review Social conflict, violence, and poverty are persistent challenges in Mindanao. Since the late 1960s, the island has witnessed widespread conflict between the state and insurgent groups 2. Additionally, dozens of militia units, political groupings, communities, and clans are often engaged in varying levels of armed conflict (Adriano and Parks, 2013; Herbert, 2019). The island also has relatively high levels of poverty. While the poverty rate in Davao (11.9 percent) is lower than the national average of 13.2 percent, the poverty rates in Caraga (25.9 percent) and BARMM (29.8 percent) are well above that average according to the 2021 Family Income and Expenditure Survey (FIES) (Philippine Statistics Authority (PSA), 2021). The existing literature suggests that conflict and poverty are intimately related in Mindanao, especially in rural areas. Banzon (2005), Edillon (2005), Capuno (2019) and other studies have assessed the role of poverty, income redistribution, and the government’s peace-building efforts in perpetuating conflict in the Philippines. These studies find that resource deprivation— particularly access to water, which is closely associated with poverty levels—is often a major driver of conflict. Likewise, as discussed in chapters 1 and 2, Mindanao has shown highly uneven patterns of development. The areas most affected by conflict also have the highest levels of poverty and the lowest levels of human development (Adriano and Parks, 2013). Land issues are key drivers of conflict and poverty, creating a vicious cycle. Malayang III (2001) and World Bank (2017) report intense, wide-ranging conflict over lands and resources within ancestral domains in Mindanao, leading to loss of life, the displacement of communities, and increased poverty. Across developing countries, such conflicts are frequently due to weak land governance, which also exacerbates poverty. Insecure property rights discourage investments, undermine the government’s ability to collect land taxes, and deprive the poor of a critical asset base (World Bank, 2017). This constitutes a vicious cycle, as insecure property rights lead to weak job creation, food insecurity, limited access to essential services, low productivity, and underemployment – and poor individuals without land or work opportunities are driven to join armed groups to survive or to seek social status (Ravanera, 2018; World Bank, 2023a). Land governance is already weak in the Philippines, but it is even weaker in Mindanao. An independent evaluation by LGI (2016) found that that Mindanao fares worse than the national average across a range of land governance indicators, including: land rights recognition, rights to forest, common land and rural land use; public land management; transfer of large tracks of land to investors; public provision of land information, registry and cadaster; and dispute resolution. 3 Such factors have contributed to the longstanding grievances over land dispossession in Mindanao that are key drivers of conflict. Moreover, conflict worsens such challenges by undermining whatever rule of law does exist, displacing residents from their land and making property rights even more difficult to secure (World Bank, 2023a). In Mindanao, for example, 28 percent of the households that returned home after being displaced by conflict could not retrieve their farmlands. 2 These include the Moro National Liberation Front (MNLF); the Moro Islamic Liberation Front (MILF); Abu Sayyaf; and the New People's Army (NPA), the military wing of the Communist Party of the Philippines (CPP). 3 As reported in World Bank (2023) using a stakeholder-driven assessment of selected land governance indicators developed by LGI. 3 Figure 1 Poverty incidence (%) in Mindanao, 2021 Source: World Bank staff estimates The prevailing evidence suggests that IPs are especially vulnerable to poverty, conflict and land issues. Across the Philippines, the country’s estimated 9.4 million IPs disproportionately reside in areas of social conflict and poverty (World Bank, 2023b as well as Reyes, Mina, and Asis, 2017). This should perhaps not be surprising, since IPs constitute a disproportionately high number of the world's rural poor, and they are often caught in the middle of violent conflicts related to land or resource disputes (UN, 2010; 2022). In response to these concerns, 1997’s Indigenous Peoples Rights Act (IPRA) sought to recognize the right of IPs in the Philippines to manage their ancestral domains. On paper, IPRA offers IPs a path towards more secure land and resource rights by providing Certificates of Ancestral Domain Titles (CADTs), which formally recognize IPs’ possession and ownership over their ancestral domains (Drbohlav and Hejkrlik, 2017; Caballero, 2004). The law defines ancestral domains as the land, waters and resources that a given IP community has traditionally used as their area of domicile and basis for economic and social life (Malayang III, 2001). Despite the IPRA, however, many IPs in the Philippines are still struggling to obtain secure access to land. Identifying and delineating ancestral domains has proved challenging, leading to conflict with non-indigenous groups as well as within and between IP communities (Prill-Brett, 2007). As documented by Drbohlav and Hejkrlik (2017) and Caballero (2014), the overlapping and conflicting mandates of government bodies tasked with IPRA implementation, as well as a lack of resources and capacity, have contributed to the slow issuance of CADTs. Provincial and regional offices of the National Commission on Indigenous Peoples (NCIP), for instance, remain severely underfunded and understaffed. (Cortez et al., 2018; Rutten, 2016). Moreover, the CADT process involves cumbersome bureaucratic requirements; Figure 2 details the six steps 4 and 21 4These steps are preparatory work for delineation; conflict resolution and CADT application deliberation; survey; approval of survey and evaluation of recognition book; approval of CADT; and awarding of CADT. 4 sub-activities required to legally obtain full CADT award (Ancestral Domain Office, 2023). The result is a lengthy and complex process that often takes several years to complete. In addition to being long and arduous, many IPs perceive a lack of ownership over the CADT process. Molintas (2004), for instance, notes that the IPRA has mainly benefited a select few IPs (and non-IPs), especially people in power who seek political advantages from the process. As a result, many IPs – who often have concepts of ownership that contrast with formal laws like IPRA – feel like the state's development policies are not designed to work for them. The CADT application process, for example, involves unwieldy and complicated technical documents, sometimes written only in English, which tend to be intimidating and difficult for many IPs to understand (Molintas, 2004). The application instructions are also often confusing or vague, especially regarding the various government office(s) to which applicants must submit their documents (Erasga, 2008). Figure 2: Activities under the stages of the CADT application Source: NCIP, Ancestral Domain Office staff Note: Certificate of Ancestral Domain Title (CADT), Certificate of Ancestral Land Title (CALT), CADT/CALT Application Record Book (CARB), Provincial Delineation Team (PDT), Community Delineation Team (CDT), Information Education and Consultation (IEC), Commission En Banc (CEB), Register of Deeds (RD), Regional office (RO), Ancestral Domain Office (ADO), Land Registration Authority (LRA), Department of Environment and Natural Resources (DENR), Department of Agrarian Reform (DAR). 5 As a result, IPs face hurdles to develop CADT lands that could be a source of income, jobs and wealth. As reported by World Bank (2023b), Mindanao comprises more than three quarters of all ancestral lands covered by CADTs in the Philippines, representing about 3.37 million hectares – about a third of Mindanao’s entire land area and half of its forestlands. These lands have tremendous potential for development and job creation (World Bank, 2023a). Beyond the difficulties of the CADT process, many IP communities lack sustainable development protection plans to support their economic development and land management efforts. Moreover, IP communities’ weak capacities for assessing development proposals are compounded by distrust of private sector investors, given past experience of inequitable contracts (World Bank, 2023a). Ultimately, IP communities do not fully benefit from their lands. These factors complicate the prevailing view of how conflict, poverty, land, and IP issues interact in Mindanao. A small but notable number of studies have already shown that establishing laws to protect ancestral domains does not automatically resolve conflict – much less promote social justice or sustainable resource use (Prill-Brett, 2007). Lara and Franco (2022) note that conflicts in Mindanao often persist after land titles are awarded, highlighting a case study from Lianga and Lanuza (Surigao del Sur, located in Caraga region). Likewise, using a case study of the Higaonon Tribe in Opol (Misamis Oriental, Mindanao, Drbohlav and Hejkrlik (2017) conclude that the mere existence of a legal framework for IPs’ rights does not ensure the security of their land tenure. Adding to this scarce literature, our analysis finds that conflict is generally less likely in CADT areas and areas with high shares of IPs – but that CADT processing delays can actually increase the probability of violence. We use a range of simple quantitative methods to unpack the links between IP presence, CADTs, and the incidence (as well as magnitude and type of) conflict in a given barangay. We then go a step further and assess the various stages of CADT processing (i.e. awarded titles, approved but not fully-awarded titles, and absence of titles). We find an association between increased conflict and CADTs that are not fully awarded. To the best of our knowledge, no other study has deliberately analyzed the relationship between IPs, conflict, and the specifics of the CADT process. 5 3. Latest evidence and trends: conflict, IPs, and CADTs As noted in the introduction, we use data from three sources to explore the relationship between conflict, IPs, and CADTs: i) 2020 census data, which indicates the share of IPs in each barangay; ii) new geo-referenced data from Conflict Alert (CA), indicating the number of violent incidents in each barangay from 2011 to 2020; and iii) geo-referenced CADT information from the National Commission on Indigenous Peoples (NCIP). We overlap these data to determine where violent incidents occurred – specifically, whether they occurred inside and outside of ancestral domains. 5 These categories in the Philippines are formally known as “registered” and “for registration,” respectively. 6 3.1. Violent incidents The CA database indicates that 51,026 violent incidents occurred in Mindanao between 2011 and 2020. (The conflict data from BARMM is available from 2011 to 2020, but only 2011 to 2015 for Caraga and Davao. 6) CA tracks the incidence, causes, and human costs of violent conflict and violent crime, based on documentation and reports by the Philippines National Police (PNP) and print media records. However, the data has certain limitations. First, CA does not report incidents for all three regions across the full 2011-2020 period. 7 Second, the data do not report the ethnicity of the perpetrator and the victim. As such, we roughly approximate the likelihood of IP involvement in a given incident, based on whether the area where it took place can be described as having a high (above average), low (below average) or no presence of IPs. Additionally, incidents in the CA data can be simultaneously recorded as multiple types of violence, such as robbery as well as firearm use. To avoid double counting, we only use the first categorization of each incident (which we assume to be the dominant feature of the incident). Third, CA assigns a large proportion of the incidents as “undetermined” in nature – that is, cannot be classified, which impacts the precision of some of our results. Nonetheless, some key trends emerge in the data. Violent conflict has shown a downward trend, with most incidents concentrated in Davao. As shown in Figure 3), violence declined briefly in 2012, which can mostly be attributed to declines in the BARMM after the signing of the Framework Agreement on the Bangsamoro (CA 2014). Despite three subsequent years of increased violence between 2013 to 2015, there was a sharp decline starting in 2016. This might be associated with a switch to the new national administration’s “iron fist” approach (as well as the partial coverage in the dataset of Davao and Caraga after 2015). Figure 6 shows that both land and non-land-related incidents (as well as those with undetermined cause) have declined over time. Of the 51,026 incidents reported in the dataset, 44 percent took place in Davao del Norte, Davao de Oro, and Davao del Sur provinces (see Figure 4). Most violent incidents appear unrelated to land or resources, though many have undetermined causes. The CA database identifies five categories of conflict: violent incidents related to the shadow economy (20 percent of the observations), common crimes (18 percent), identity issues (12 percent), political issues (5 percent), resource issues, including related to land (3 percent), governance issues (2 percent), and “undetermined” (40 percent). Figure 6 groups these into three categories: land (part of resource issues), non-land, and undetermined. 8 Notably, non-land 6 We used the entire conflict dataset for the three regions across time, and we control our estimations by year and province. This ensures that unobservable differences between time periods are taken into account in our statistical analysis – which is relevant because, as indicated, CA did not collect information on incidents for all three regions across all years. 7 There are multi-stakeholder validation groups (MSVGs) that conduct discussions to filter and further probe the reported violent incidents. 8 Shadow economy-related incidents include cattle rustling, illicit cross-border trade, human trafficking, illegal drugs, and illegal gambling, among others. Common crimes include: alcohol intoxication, damage to properties, robbery, child abuse, and other common crimes. Political issues involve rebel groups, political repression, rebellion, elections, and violent extremism, among others. Identity issues are incidents related to gender, inter and intra-gang rivalries, personal grudges, clan feuds, and religious conflict, among others. Resource issues touch on natural resources, land conflict, water conflict, urban resources, and resource predation, among others. Governance issues 7 violent incidents comprise some 59 percent of all reported incidents, with only 1 percent being definitively related to land. However, the fact that 40 percent of incidents are categorized as having an undetermined cause makes it difficult to draw precise conclusions. Figure 3. Distribution of violent incidents by Figure 5. Distribution of violent incidents by year (number of incidents), 2011-2020 land, non-land categories (% of total for period 10,000 2011-2020) 9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0 2011201220132014201520162017201820192020 Source: CA, 2011-2020 from BARMM, Caraga and Davao. Source: CA, 2011-2020 from BARMM, Caraga and Davao Figure 6. Land and non-land-related incidents Figure 4. Distribution of violent incidents by over time (number of incidents), 2011-2020 Province (% of total for period 2011-2020) (Land on left axis, Non-land on right axis) Davao… Davao… 100 6,000 Davao… 5,000 Maguind… 80 Davao… 4,000 Lanao… 60 3,000 Sulu Basilan 40 2,000 Davao… Surigao… 20 1,000 Agusan… Surigao… 0 0 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Tawi-Tawi Agusan… Dinagat… Land Non-land 0 5 10 15 20 Source: CA, 2011-2020 from BARMM, Caraga and Davao. Source: CA, 2011-2020 from BARMM, Caraga and Davao. deal with corruption of public office, implementation of government projects, executive and judicial decisions, executive and judicial positions, and Covid-19 issues, among others. 8 3.2. Indigenous peoples The IP data used in our analysis, obtained from the 2020 Population and Housing Census, refers to total population counts and population counts by ethnicity at the barangay level. 9 This data was provided by the Philippine Statistics Authority (PSA) exclusively for this study. The definition of IPs was revised by PSA for the 2020 census to include people who identify as IPs as well as members of Muslim tribes that are categorized as IPs. Overall, this data indicates that approximately 9.4 million IPs live in the Philippines’ 17 regions. In Mindanao, IPs mainly live in Davao del Sur, Davao Oriental, Agusan del Sur, and Basilan provinces. Figure 7 shows their distribution within the BARMM, Caraga, and Davao regions. Figure 7. Percentage of IPs Population by Province, 2020 Source: 2020 Population and Housing Census, BARMM, Caraga, and Davao. 3.3. Certificates of Ancestral Domain Titles (CADT) A CADT refers to a title formally recognizing IPs’ rights of possession and ownership over their ancestral domains, as identified and delineated by the IPRA law. 10 It denotes the land and resources that IPs have traditionally used as their area of domicile and the basis of their economic and social life (Malayang III, 2001). As of March 31, 2021, the NCIP has approved 257 CADTs with a total land area of 5,971,334 hectares across the Philippines (NCIP Ancestral 9 Ideally, the inclusion of other variables to help characterize IPs, such as economic and social indicators from the census, would allow us to add further relevant controls to our analysis. While this study did not benefit from such information among IPs, geographical, economic and social data on the barangays where IPs reside was provided (as explained in section 3). 10 Sec. 3, Chapter II, 1997 IPRA. 9 Domain Office, 2023, special tabulations). There are another 205 CADTs in process, representing 3,719,176 million hectares, as well as 486 identified CADTs that are not subject to any application process, covering 3,756,151 million hectares. The CADT area data used in our analysis was georeferenced and provided by NCIP. Figure 8 shows the distribution of CADT areas in Caraga, Davao, and the BARMM, illustrating that the majority of CADTs are not yet fully awarded. Our analysis includes 72 CADT areas processed from 2002 to 2022 (one in the BARMM, 33 in Caraga, and 38 in Davao). Critically for our analysis, the figure indicates whether a given CADT area has a fully awarded title (shadowed in green) or approved but not yet fully awarded (shadowed in red). Of the 33 CADTs in Caraga, three are awarded and 30 have been approved but are not yet awarded. Of the 38 CADTs in Davao available for this study, 9 are awarded and 29 are approved but not yet awarded. In the BARMM, this information is not available for the one CADT in that region. Figure 8. Distribution of CADTs in the BARMM, Caraga, and Davao regions (as of 2022) BARMM Davao Caraga Source: CADT, NCIP, 2022 Note: Green CADT areas equals awarded, and red CADT areas approved but not fully-awarded, grey (in BARMM) is unknown 10 4. Linking conflict, IPs and CADTs 4.1. A simple analytical approach As noted in the introduction, after cleaning our three datasets, we merged them into single data file suitable for empirical analysis. 11 This combined data allowed us to discern, visualize, and analyze which violent incidents occurred inside or outside ancestral domains. We narrowed our data to incidents for which have all information of interest (i.e., whether an incident is land, non- land related, or undetermined; whether it occurred in an area with high, low, or no presence of IPs; and the CADT status of where the incident occurred). The resulting dataset included 41,380 observations (individual incidents) across 2,854 barangays. We then analyzed this dataset using maps as well as statistical analysis to unpack their patterns and interactions. We also ran a series of tests to check the robustness of our results. Additionally, we performed a Linear Probability Model (LPM) to estimate the likelihood of violent conflict and the extent to which the presence of IP populations and other factors drove that prediction, which allowed us to test our results while controlling for additional factors and unpacking them type of conflict. We used this model because it has the advantage of being easy to interpret 12 – positive coefficients associated with a given area indicate that conflict is more likely, with the reverse being true for negative coefficients. The empirical strategy to construct the LPM and the summary statistics of the variables used in the regression analysis is explanined next. The unit of observation was each incident X. The analysis estimates the probability of an incident Xi occurring at a barangay g and year t based on three broad groups of factors: (1) violence-related factors (I), (2) geographically-varying variables (G), (3) risk factors, (4) time-varying factors (T): Pr(Incident Xj,t)= f (V, G, R, S, T) (1) For violence-related factors (V), we used a dummy variable (inside CADT) that equals 1 if incident X took place inside an ancestral domain and 0 otherwise. Additionally, a set of dummy variables were used to account for V factors in the likelihood of incident Xi happening. Awarded takes the value 1 if the territory for incident X was an area with awarded CADT and 0 otherwise. Approved takes the value 1 for incidents taking place in approved but not fully-awarded CADT areas and 0 otherwise. Together, these dummy variables 11 Once each dataset was cleaned, incident data was merged with the ancestral domain data using the barangay codes. In the case of Davao, the barangay codes were not available and had to be constructed using artificial an intelligence (AI) recursive algorithm. The resulting dataset was merged with the 2020 Census population data at the barangay level following the revised classification provided by PSA, based on the Philippine Standard Geographic Code (PSGC) as of March 2022. 12 LPM is a binary regression model where the variable of interest for each observation takes values 0 or 1. In this case, a positive coefficient indicates that when one variable (e.g. share of IPs) increases, the mean of the other variable (e.g. violent incidents) also tends to increase. All econometric results were then replicated using a Probit model, which served as a robustness check; our main findings did not change as a result of these checks. Results are available upon request. 11 control for potentially different impacts on the likelihood of conflict for awarded and approved CADT areas. Our specifications also contain several variables that control geographic factors (G). Urban is a dummy variable that indicates where the population is larger than 5,000 people (which is the population threshold used by PSA to determine urban contexts). Mining is a dummy variable that equals 1 if the barangay contains a mining site and 0 otherwise. Poverty is the incidence of poverty at provincial level in 2021. Active4Ps is a variable that measures the number of beneficiaries of active development projects happening in the barangay. Central is equal to 1 if the Baranguay is a poblacion/central district or part of the población/central district; Highway is equal to 1 if the barangay is accessible to the national highway; Risk factors (R) are Precarious is equal to 1 if the household reside in a precarious location; Relocation is equal to 1 if there are a temporary relocation area in the barangay; Move-in is equal to 1 if there was a large or significant number of households who moved in or transferred to their barangay in the last five years due to environmental or peace and order reasons; Move-out is equal to 1 if there was a large or significant number of households who moved out or transferred due to environmental or peace and order reasons in the last 5 years; Services factors (S) are Svs_index is equal to 1 if the barangay has any of these services such as hospital, puericulture center, fire station, post office, landline, cellular phone signal, public street sweeper; Edu_index is equal to 1 if the barangay has elementary school, high school or college; Eco_index is equal to 1 if the Baranagay has any of the following: wholesale and/or retail trade establishment, recreational establishment, manufacturing establishment, accommodation and food service establishment, financial establishment, establishment offering repair services, establishment offering personal services, other establishments. See Appendix 2 for a more detailed description of the variables. Time factors (T) are accounted for using year-fixed effects for each incident Xi. Fixed effects is a method of controlling for all variables, whether they are observed or not, as long as they stay constant within some larger category, in this case province and year. 13 This is convenient in econometric terms because it avoids biases resulting from omitted variables not included in the model,provided they are not subject to change in the period considered. Table 1 shows the summary statistics of the variables used in the regression analysis. Table 1. Summary statistics Variable Obs Mean Std. Dev. Min Max land 41,380 0.007129 0.084133 0 1 noland 41,380 0.57639 0.494136 0 1 insidecadt 41,380 0.316651 0.465176 0 1 Ips 41,380 0.205234 0.254886 0 1 awarded 41,380 0.109715 0.312538 0 1 approved 41,380 0.168874 0.374645 0 1 poverty 41,380 20.12209 12.50888 7.2 51 urban 41,380 0.546278 0.49786 0 1 13 Estimates cluster errors by barangay. Given that the analysis spans for several years across barangays, this will naturally lead to within-barangay and across-year correlations. We would also expect some extent of correlations from the unobservables, given that the same barangay in different years is probably more alike than two randomly selected barangays, and so clustering by barangay would be appropriate. 12 active4Ps2020 41,380 416 351 0 1831 mining 41,380 0.167351 0.373294 0 1 central 41,380 0.425882 0.494482 0 1 highway 41,380 0.919913 0.271431 0 1 precarious 41,380 0.717883 0.450035 0 1 relocation 41,380 0.113992 0.317806 0 1 moved_in 41,380 0.307274 0.46137 0 1 moved_out 41,380 0.162204 0.368642 0 1 edu_index 41,380 0.978081 0.14642 0 1 svs_index 41,380 0.993596 0.07977 0 1 eco_index 41,380 0.85319 0.353921 0 1 Source: Authors’ database Land is a dummy variable that takes the value of 1 if the conflict is land-related and 0 otherwise. Noland is a dummy variable that takes the value of 1 if the conflict is non-land-related and 0 otherwise. IPs is the percentage of indigenous population. The summary statistics show the disproportion between land and non-land incidents. Land incidents represent only 1 percent of all incidents in our data set, while non-land ones make up 58 percent. 14 On average, IPs represent 20 percent of the population in the barangays where incidents happened. 29 percent of the incidents occurred inside a CADT. Urban barangays make up 55 percent of all Barangays, and the mining sites cover 17 percent of the territory. There are an average of 416 beneficiaries of active development projects per barangay. 43 percent of the barangays are a poblacion/central district or part of the población/central district and 92 percent are accessible to the national highway. 72 percent of the households are in precarious locations, and 11 percent of the barangays have a a temporary relocation area. Given to environmental and peace and order reasons, 31 percent of the people thought it was a large or significant number of households who moved in or transferred to their barangay in the last five years, and that move- out in 16 percent. 98 percent of barangays either have a primary, secondary or college, almost all barangays (99 percent) have at least one service and 85 percent have at least one establishment (See Table A1 above). 4.2. Key results The results below are organized around the three key findings of our analysis: (i) In general, conflicts are less likely in barangays with higher shares of IPs. Barangays with higher shares of IPs are associated with fewer incidents of violence, according to our analysis of the combined 2011-2020 dataset. Across Mindanao, higher levels of conflict occurred in areas with fewer IPs as a share of the local population. The scatterplot in Figure 9 14 In Figure 3, we reported 59 percent of non-land incidents. That number corresponds to the average before merging the CA violent incident data with CADT and the IP population. In that process we lost a few observations: 2,119 incidents.. 13 visualizes this correlation, with the downward-sloping red line showing the general trend: the higher the share of IPs (vertical axis), the lower the incidence of violence (horizontal axis). Figure 9. Presence of IPs and violent incidents per barangay, 2021-2020 Scatterplot %IPs vs %Conflict incidents 1 .8 .6 % IPs .4 .2 0 0 1 2 3 4 % incidents Source: Authors’ database. Note: The pairwise correlation of these two variables is -0.0342 and is significant at the 1% level. (ii) Land- and resource-related conflicts are also less likely in barangays with higher shares of IPs. More specifically, our analysis also suggests that barangays with higher shares of IPs are associated with fewer incidents of land- and resource-related violence. Figure 10 shows that fewer land- and resource-related violent incidents occur in areas with high IP presence (gray bar) compared to areas with low IP presence (orange bar). (As noted from above that the CA data categorizes land-related conflict as a subset of resource-related conflict, which also includes other conflict over water and other natural resources.) However, the fewest incidents of land- and resource-related conflict actually occurred in areas with no IP presence (blue bar), which tend to be urban areas. While this fact alone would seem to suggest a positive correlation between IP presence and violence, the more important factor is the decrease in conflict when moving from low-IP areas (orange bar) to high-IP areas (gray bar). Given the large number of incidents in the dataset, this decrease (between the orange and blue bars) drive the statistical correlation. This is called a “non-monotonic” or nonlinear relationship, and our analysis finds that it is robust to alternative ways of looking at the areas where incidents take place. 15 Similar trends are observed for non-land- and non-resource-related conflict, as well as violent incidents with an undetermined cause. As with figure 10, figure 11 shows that the highest level 15 In effect, similar nonlinear relationships were observed quartiles of barangays ordered by their concentration of violent conflict: those results are available upon request. 14 of non-land-related, non-resource-related, and undetermined violent incidents occurred in areas with low IP presence. The finding about undetermined incidents is particularly notable, given that this category presumably includes both land-related and non-land-related conflict). Note that the results in figure 11 are also non-monotonic, but this does not affect the direction of the correlation for the same reasons mentioned above. Figure 10. Distribution of violent incidents by land-related and resource-related categories, 2011- 2020 Source: Authors’ database Figure 11. Distribution of violent incidents by non-land-related and non-resource-related, and undetermined categories, 2011-2020 20,000 15,000 10,000 5,000 0 Non-resource related Non-land related Undetermined No IPs presence Low Ips presence High Ips Presence Source: Authors’ database (iii) Overall, CADT areas have a lower likelihood for general conflict but a higher likelihood for land-related conflict, particularly in CADTs that experience processing delays. Our analysis further suggests that conflicts, in general, are more likely to happen outside CADT areas. The geo-referenced data allows us to map the distribution of violent incidents in Mindanao in order to visualize whether they occurred inside or outside a CADT. Figure 12 presents three such maps: one for CADT areas in the peninsular region, one for Caraga, and one for Davao. The light green shading captures the geographical boundaries of CADTs. The red dots indicate violent incidents that occurred inside a CADT area while the dark green dots represent incidents 15 that occurred outside CADT areas. All the maps consistently show that conflicts are more likely to happen outside CADT areas. Figure 12. Conflict distribution by province, 2011-2020 BARMM Caraga Davao Source: Author's database 16 When unpacking conflict by category, however, land-related conflicts are more likely to occur inside CADTs. 16 We obtained this finding by conducting a simple regression analysis that controlled for other factors that might affect the correlation. 17 Table 2 presents the regression results, which shows that land conflicts are positively correlated with CADTs to a 95% confidence level. Likewise, the table also shows that non-land-related conflicts are negatively correlated with CADTs, although this relationship is not statistically significant. Together, these results suggest that non-land conflicts are generally less likely to occur within CADTs but land- related conflicts are more likely to occur within CADTs. These results should not be interpreted as causal, but they provide rigorous support for the accuracy and strength of our findings. 18 Table 2. LPM estimation results for the likelihood of experiencing land and non-land conflict inside CADTs areas (2011-2020) (1) (2) VARIABLES Land No Land Violent incidents inside CADT 0.003** -0.023 [0.001] [0.014] Fixed effects Yes Yes Controls Yes Yes Barangay-clustered errors Yes Yes Observations 41,380 41,380 R-squared 0.006 0.043 *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimates However, focusing on a CADT’s processing stage complicates these findings, and suggests that the likelihood for land-related conflicts are only higher in CADTs that are “approved by not fully awarded.” The findings described above refer analysis of all CADTs, whether they are “fully awarded” or experiencing processing delays – “approved but not fully awarded.” When dividing the data into these two categories, the correlations change slightly. In fact, contrary to Table 2 above, we find that fully awarded CADTs are actually associated with less conflict, whether land-related to not. On the other hand, we find that approved but not fully awarded CADTs are associated with less non-land-related conflict but more land-related conflict. Thus, the pair of finding presented above – that CADTs have a lower likelihood of general conflicts but a higher likelihood of land-related conflicts – appears to only hold for approved but not fully awarded CADTs. Table 3 presents the regression results of this level of the analysis. 19 In other words, the correlation between land-related conflicts and CADTs appears to be entirely driven by those CADTs that are experiencing processing delays. Fully awarded CADT areas reduce the 16 There are 177 land incidents that happened inside CADTs and 118 that happened outside CADT areas. There are 6,779 non- land-related incidents that happened inside CADTs and 17,072 non-land-related incidents that happened outside CADT areas. 17 Controls included in the analysis are defined at the barangay level and refer to its demographics, poverty levels, the extent to which benefits from 4Ps social transfer program, inventory of public services provided, and several risk factors for conflict such as presence of mining activities or refugee centers. 18 All specifications in Table 1 include controls and year fixed effects. See Appendix 1 for full regression results, Appendix 2 for a list of variables used. Results using Probit, for robustness checks, are available upon request to the authors. 19 These results are weaker (in statistical terms), partly due to few CADT areas being fully awarded, but they are still suggestive of a distinctive effect on conflict based on whether the process of granting CADT status is completed or not. These results hold when we use other models such as Probit. Results available upon request. 17 probability of land conflict, but when still being processed, the presence of CADTs does not have the same effect. Table 3. LPM Estimation results for CADT status (2012-2020) (1) (2) VARIABLES Land No Land Awarded -0.003 -0.016 [0.003] [0.028] Approved 0.001 -0.026 [0.001] [0.016] Fixed effects Yes Yes Controls Yes Yes Barangay-clustered errors Yes Yes Observations 41,380 41,380 R-squared 0.006 0.043 *** p<0.01, ** p<0.05, * p<0.1 Source: Authors’ estimates These results are consistent with the literature on CADTs’ inability to reduce conflict on their own. As reviewed earlier in this chapter, Prill-Brett (2007), Lara and Franco (2022), and Drbohlav and Hejkrlik (2017) all suggest that awarding CADT status does not necessarily mean that conflict disappears, and World Bank (2023a) provides insights of why this is the case. It summarizes several flaws of CADTs, for instance: even when a CADT is being awarded, multiple and conflicting land titles may be issued by different government bodies (e.g., Certificates of Land Ownership Awards provided by the Department of Agrarian Reform). Likewise, required procedures to secure indigenous communities’ “free, prior, and informed consent” over land issues can be tedious and time-consuming, resulting in processing delays. Frequently, IP communities’ rights are also inadequately recognized by other parties, resulting in disregard for CADTs that are still being processed. Addressing these challenges will likely require strengthening land governance in Mindanao. NCIP estimates that at the current pace, it would take 50 years to complete land titling in Mindanao. The major reasons for slow CADT implementation are well known – e.g., delays in completion of cadastral surveys, missing or erroneous survey records, high costs for titling, and a range of other cumbersome and costly procedures. But these difficulties can lead to other challenges. The longer titling is delayed after completion of cadastral surveys, for instance, the more complex adjudication becomes – since transfers, death of original claimant, subdivisions, and consolidations usually occur (NCIP Ancestral Domain Office, 2023). Addressing these issues will require a spectrum of efforts, including broad improvements in the ecosystem of land governance institutions across Mindanao. Such efforts should include enhancing dispute resolution, particularly in cases of overlapping titles. The current legal framework for dispute resolution is overly complex and potentially creates more disputes than it solves. A key challenge is overlapping titles. As noted above, when multiple titles are issued in areas governed by different land laws, possession of one title is often 18 not enough to secure property rights (World Bank, 2021). Various land agencies frequently create their own independent records and maps without sufficient coordination, results in overlapping titles that become the subject of litigation. Prior efforts to address this challenge have not been effective. 20 Overlapping titles often lead to the underutilization of large tracks of land, since owners’ rights (even when there are titles) remain insecure, increasing the risks for further litigation and even violent conflict (World Bank, 2023). 5. Concluding remarks This article has explored the relationship between conflict and IPs in Mindanao. Its main contribution is to provide more systematic and comprehensive evidence to complement existing qualitative work, which points to a prevalence for conflict inside CADT areas. While most of these studies base their results on case studies and qualitative evidence, our analysis focuses on a broad region (including Caraga, Davao, and the BARMM) and uses a suite of quantitative tools, such as econometric and spatial analysis. We have shown that barangays with higher shares of IPs are associated with less conflict, including land-related conflicts. These findings hold when expanding the scope of the analysis to include resource-related conflict, and even when looking at conflicts of undetermined origin. While the chapter focuses on Mindanao, it is likely that our results could apply to other IP areas in the Philippines. There are no reasons for complacency, however: our evidence also shows that CADT processing delays can increase violence. Only 31 percent of all violent incidents take place inside CADT areas – but CADTs have a higher likelihood of land-related conflict. Moreover, this positive correlation with land conflict appears to be entirely driven by CADTs that are experiencing processing delays. “Approved but not yet fully awarded” CADTs are associated with more land conflict and less non-land conflict, while “fully awarded” CADTs are associated with less conflict for both land and non-land conflict). These findings suggest that the existence of CADTs do not on their own solve land-related conflicts. Our results suggest that CADTs can reduce conflict in general, and fully awarded CADTs can reduce land-related conflict. But the problem is that too few CADTs are fully awarded, and “approved but not yet awarded” CADTs can increase violence. These administrative delays, in turn, are the result of lengthy and bureaucratic processes that are not backed up by sufficient resources and institutional capacity. Addressing CADT implementation challenges must be coupled with stronger land governance. Improvements are needed in multiple areas, such as: rights to forest; public land management; transfer of large tracks of land to investors; public provision of land information, registry and cadaster; and dispute resolution. These challenges will not be easy to address, but they can no longer be ignored. Success on these reforms promises far-reaching benefits, including increased 20 The joint administrative order approved in 2012 to prevent the overlapping in tenure issuance across different public agencies (the Department of Environment and Natural Resources, the Department of Agrarian Reform, and the Department of Land Registration) largely failed to correct previous overlapping claims or address the overlapping of legal frameworks; it simply provides a procedure for cross-validation (NCIP 2023). 19 investment, job creation, income and wealth generation for IPs, and – ultimately – less poverty and conflict. This study underscores persistent data gaps and the need for more analysis into conflict, land, and IP issues. More and higher-quality data is always needed for better policymaking, but this is especially true for the case of IP issues in the Philippines. For example, there is a strong need for more continuous collection and dissemination of IP-specific statistics. More specifically, data on IPs should include details about their socioeconomic characteristics, and data on conflict should include more explicit details about the victims and perpetrators, including their IP status. Only with more detailed and precise data will policymakers be equipped to more effectively reduce conflict, strengthen land governance, and reduce poverty – for IPs in Mindanao and for all members of Philippine society. References Adriano, F. and T. Parks. 2013. The Contested Corners of Asia: Subnational Conflict and International Development Assistance. The Case of Mindanao, Philippines. The Asia Foundation. Ancestral Domain Office, NCIP, 2023. Activities under the Stages of the CADT/CALT Application (AO 1, s. of 2020). Manila: NCIP. Ancestral Domain Office, NCIP, n.d. Special Tabulations on CADTs. Manila: NCIP. Caballero, E. 2004. Ancestral Domain Delineation and Recognition: CADT of Aytas of Bataan. Capuno, J. 2019. What drive the local incidence of crime, shadow economy and resource-related conflicts in Mindanao, Philippines? Evidence of spillover effects, Journal of The Asia Pacific Economy, VOL. 24, NO. 1, 42–65. Conflict Alert, CA, 2020. Special Database on Census Population Counts. Conflict Alert, CA, 2014. Rebellion, Political Violence and Shadow Crimes in the Bangsamoro, Understanding conflict. Building peace. The Bangsamoro Conflict Monitoring System (BCMS), 2011–2013 Cortez, I., Fajardo, J., Galang, T., Gonzales K., Lagrada, R. Rivera L. 2018. An Assessment of the Certificate Of Ancestral Domain Title (CADT) Delineation and Recognition Process of The National Commission on Indigenous Peoples (NCIP): The Cases Of Limay, Bataan And Botolan, Zambales Drbohlav, P. and J. Hejkrlik. 2017. Indigenous Peoples’ Struggle for Secure Land Tenure in the Philippines: Case Study of Higaonon Tribe in Opol, Mindanao, Asian Social Science, 13(7). Edillon, R. 2005. Ideologically Motivated Conflicts in The Philippines: In Search of Underlying Causes, Human Development Report Erasga, D. Ancestral Domain Claim: The Case of the Indigenous People in Muslim Mindanao (ARMM). Asia-Pacific Social Science Review 8:1 (2008), pp. 33-44 Hebert, S. 2019. Conflict Analysis of the Philippines. University of Birmingham Lara, F and B. Franco. 2022. Identity-Based Conflicts and the Politics of Identity in Eastern Mindanao, Philippine Journal of Public Policy: Interdisciplinary Development Perspectives LGI Consultants. 2016. “Analysis of the influence of land governance on employment creation in Mindanao.” Manila and Washington DC: World Bank. 20 Malayang III, B. 2001. Tenure rights and ancestral domains in the Philippines. A study of the roots of conflict in: Bijdragen tot de Taal-, Land- en Volkenkunde, The Philippines Historical and social studies 157, no: 3, Leiden, 661-676 Molintas, J. The Philippine Indigenous Peoples’ Struggle for Land and Life: Challenging Legal Texts, Arizona Journal of International and Comparative Law Prill-Brett, J. (2007) Contested Domains: The Indigenous Peoples Rights Act (IPRA) and Legal Pluralism in the Northern Philippines, The Journal of Legal Pluralism and Unofficial Law, 39:55, 11-36. Reyes, C., C. Mina and R. Asis, 2017. Inequality of Opportunities Among Ethnic Groups in the Philippines, Discussion Paper Series, No. 2017-42, Philippines Institute for Development Studies, Manila, Philippines. Sustainable Development Goals, 2019. Indigenous Peoples and the Sustainable Development Goals, Manila, Philippines. United Nations Development Programme, UNDP. 2013. Indigenous peoples in the Philippines. Manila, Philippines. United Nations, 2022. Truth, transitional justice and reconciliation processes, Concept Note, Santiago, Chile. United Nations, 2012. The EU-UN Partnership on Land, Natural Resources and Conflict Prevention, available at Land and conflict. United Nations, 2010. Permanent Forum on Indigenous Issues, Report on the ninth session. Wetzlmaier, M. 2012. Cultural impacts of mining in indigenous peoples’ ancestral domains in the Philippines. ASEAS – Austrian Journal of South-East Asian Studies, 5(2), 335-344 World Bank, 2023a. Mindanao Jobs Report, Part II. Agrarian Mindanao. Washington, DC. World Bank, 2023b. No Data No Story: Indigenous Peoples in the Philippines. Manila. World Bank, 2021. Mindanao Peace Lens, Volume I, Background materials. Washington, DC. World Bank, 2017. Mindanao Jobs Report, Part 1. Washington, DC. World Bank, 2015. Access to Land by Indigenous Peoples Groups in the Philippines: Using Ethnographic Methods to Inform Policy Making, mimeo, Washington, DC. World Bank and International Organization for Migration, 2019. Land: Territory, Domain, and Identity – A report submitted by the WB-IOM technical team to the Transitional Justice and Reconciliation Commission (TJRC) World Bank and International Organization for Migration. 2013. Land disputes in conflict affected areas of Mindanao: report of the joint World Bank – International Organization for Migration scoping mission. Manila, Philippines. 21 Appendix 1: Full econometric results (LPM) Results are divided in two sets, those comprising land-related conflicts, columns 1 to 6, in tables A2.1 and A2.2, and estimations for non-land- related conflicts, columns 7 to 12. Estimates for models in columns 1 and 7 are unconditional and only control for whether the incident took place inside or outside a CADT area for land and non-land conflicts, respectively. Estimates in columns 2 and 8 add year- and province-fixed effects to the unconditional effects of CADT location. Columns 3 and 9 add to those estimates the extent to which the barangay receives 4Ps benefits and whether it contains mining sites; columns 4 and 10 results additionally control for the proximity to highways; columns 5 and 11 add several conflict risk factors; while columns 6 and 12 provide the results for most comprehensive models where all fixed effects, barangay-clustered errors, and all possible types of controls—beneifts, mining sites, proximity to highways, conflict risk factors and levels of public service provision--are considered. Table A1.1 Full LPM results predicting violent incidents in BARMM, Caraga and Davao (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) VARIABLES Land Land Land Land Land Land No Land No Land No Land No Land No Land No Land Inside CADT? 0.364*** 0.338** 0.308* 0.224 0.252 0.287* -0.154*** -0.072*** -0.039*** -0.038*** -0.041*** -0.041*** [0.118] [0.151] [0.162] [0.163] [0.161] [0.164] [0.010] [0.011] [0.012] [0.012] [0.012] [0.012] Poverty -0.032 -0.030 -0.029 -0.031 -0.015*** -0.015*** -0.015*** -0.015*** [0.020] [0.020] [0.020] [0.020] [0.001] [0.001] [0.001] [0.001] Urban -0.835*** -0.682*** -0.713*** -0.675*** 0.103*** 0.099*** 0.103*** 0.100*** [0.167] [0.173] [0.173] [0.176] [0.011] [0.011] [0.011] [0.011] active4Ps_2020 -0.000 -0.000 0.000 0.000 -0.000*** -0.000*** -0.000*** -0.000*** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Mining -0.141 -0.215 -0.230 -0.234 -0.066*** -0.065*** -0.064*** -0.064*** [0.196] [0.192] [0.192] [0.192] [0.018] [0.018] [0.018] [0.018] Central -0.463*** -0.472*** -0.436*** 0.009 0.010 0.008 [0.150] [0.151] [0.155] [0.009] [0.009] [0.009] Highway -0.396** -0.357** -0.290 0.012 0.011 0.011 [0.169] [0.167] [0.178] [0.015] [0.015] [0.016] Precarious 0.105 0.161 -0.021** -0.024** [0.136] [0.138] [0.010] [0.010] Relocation -0.405* -0.376* -0.015 -0.016 [0.212] [0.213] [0.013] [0.013] moved_in 0.402*** 0.410*** -0.026** -0.025** 22 [0.134] [0.135] [0.010] [0.010] moved_out -0.397** -0.393** 0.037*** 0.036*** [0.176] [0.176] [0.012] [0.012] edu_index 0.342 -0.064*** [0.359] [0.025] svs_index -0.413 -0.025 [0.523] [0.044] eco_index -0.414** 0.031** [0.163] [0.013] prov = 24 -0.973*** -1.024*** -0.935*** -0.981*** -1.021*** -0.122*** -0.143*** -0.145*** -0.137*** -0.137*** [0.357] [0.356] [0.358] [0.359] [0.359] [0.015] [0.015] [0.015] [0.016] [0.016] prov = 25 0.740*** 1.114*** 1.173*** 1.165*** 1.171*** -0.092*** 0.141*** 0.140*** 0.140*** 0.139*** [0.235] [0.324] [0.324] [0.327] [0.325] [0.018] [0.023] [0.023] [0.023] [0.023] prov = 36 1.231** 0.880 0.785 0.809 0.727 -0.096** -0.019 -0.016 -0.020 -0.027 [0.563] [0.589] [0.586] [0.582] [0.593] [0.040] [0.041] [0.041] [0.041] [0.042] prov = 38 1.240*** 1.821*** 1.729*** 1.751*** 1.681*** -0.123*** 0.237*** 0.239*** 0.236*** 0.237*** [0.266] [0.478] [0.474] [0.478] [0.485] [0.020] [0.031] [0.031] [0.031] [0.032] prov = 66 -0.015 1.489 1.228 1.325 1.211 -0.069*** 0.678*** 0.684*** 0.670*** 0.671*** [0.421] [0.959] [0.955] [0.966] [0.977] [0.022] [0.058] [0.058] [0.059] [0.059] prov = 67 0.273 0.387 0.451 0.438 0.476 0.057** 0.288*** 0.286*** 0.283*** 0.280*** [0.375] [0.433] [0.432] [0.432] [0.432] [0.023] [0.027] [0.027] [0.027] [0.027] prov = 68 1.446*** 1.710*** 1.740*** 1.764*** 1.782*** 0.011 0.294*** 0.293*** 0.290*** 0.289*** [0.254] [0.379] [0.377] [0.378] [0.377] [0.022] [0.029] [0.029] [0.029] [0.029] prov = 70 -0.110 0.502 0.354 0.432 0.417 -0.042 0.283*** 0.286*** 0.289*** 0.290*** [0.616] [0.715] [0.714] [0.725] [0.725] [0.027] [0.035] [0.035] [0.035] [0.035] prov = 82 0.480** 0.844*** 0.865*** 0.827*** 0.822*** -0.064*** 0.108*** 0.107*** 0.115*** 0.115*** [0.229] [0.266] [0.268] [0.268] [0.267] [0.016] [0.018] [0.018] [0.018] [0.018] prov = 85 1.769*** 1.877*** 1.936*** 1.942*** 1.911*** 0.002 0.335*** 0.333*** 0.338*** 0.338*** [0.330] [0.463] [0.460] [0.465] [0.465] [0.038] [0.044] [0.044] [0.044] [0.044] prov = 86, omitted - - - - - - - - Year = 2012 -0.188 -0.156 -0.154 -0.144 -0.147 -0.022 -0.021 -0.021 -0.023 -0.023 [0.195] [0.195] [0.195] [0.194] [0.194] [0.018] [0.018] [0.018] [0.018] [0.018] Year = 2013 -0.299 -0.304 -0.299 -0.290 -0.293 -0.091*** -0.091*** -0.091*** -0.092*** -0.092*** [0.187] [0.186] [0.186] [0.186] [0.186] [0.016] [0.016] [0.016] [0.016] [0.016] Year = 2014 -0.393* -0.378* -0.388* -0.370* -0.373* 0.040*** 0.041*** 0.041*** 0.040*** 0.040*** 23 [0.202] [0.201] [0.201] [0.201] [0.201] [0.015] [0.015] [0.015] [0.015] [0.015] Year = 2015 -0.733*** -0.687*** -0.692*** -0.675*** -0.670*** 0.146*** 0.146*** 0.146*** 0.145*** 0.145*** [0.229] [0.225] [0.226] [0.225] [0.225] [0.015] [0.015] [0.015] [0.015] [0.015] Year = 2016 -1.150*** -1.133*** -1.127*** -1.111*** -1.116*** 0.170*** 0.170*** 0.170*** 0.170*** 0.170*** [0.427] [0.425] [0.425] [0.424] [0.424] [0.027] [0.027] [0.027] [0.027] [0.027] Year = 2017 -0.208 -0.192 -0.195 -0.176 -0.174 0.245*** 0.247*** 0.247*** 0.247*** 0.247*** [0.323] [0.320] [0.320] [0.319] [0.319] [0.026] [0.026] [0.026] [0.026] [0.026] Year = 2018 -0.685* -0.693* -0.689* -0.681* -0.668* 0.435*** 0.437*** 0.438*** 0.438*** 0.438*** [0.397] [0.395] [0.394] [0.393] [0.393] [0.024] [0.024] [0.024] [0.024] [0.024] Year = 2019 -0.642 -0.684* -0.699* -0.699* -0.694* 0.391*** 0.396*** 0.397*** 0.397*** 0.398*** [0.410] [0.409] [0.409] [0.408] [0.407] [0.025] [0.025] [0.025] [0.025] [0.025] Year = 2020 -0.309 -0.338 -0.336 -0.328 -0.330 0.347*** 0.354*** 0.354*** 0.355*** 0.355*** [0.375] [0.373] [0.372] [0.372] [0.372] [0.026] [0.026] [0.026] [0.026] [0.026] prov = 86 -0.743 -0.415*** [0.539] [0.032] Constant -5.074*** -5.220*** -4.467*** -4.092*** -4.253*** -3.957*** -0.505*** -0.531*** -0.450*** -0.462*** -0.443*** -0.379*** [0.075] [0.216] [0.298] [0.335] [0.344] [0.677] [0.005] [0.013] [0.018] [0.023] [0.024] [0.049] Observations 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 R-squared 0.000 0.004 0.005 0.006 0.006 0.006 0.007 0.039 0.042 0.042 0.043 0.043 *** p<0.01, ** p<0.05, * p<0.1 Barangay-clustered standard errors in brackets Urban is a dummy variable that indicates where the population is larger than 5,000 people. Mining is a dummy variable that equals 1 if the barangay contains a mining site and 0 otherwise. Poverty is the incidence of poverty at provincial level in 2021. Active4Ps is a variable that measures the number of beneficiaries of active development projects happening in the barangay. Central is equal to 1 if the baranguay is a poblacion/central district or part of the poblacion/central district; Highway is equal to 1 if the barangay is accessible to the national highway; Precarious is equal to 1 if the household reside in a precarious location; Relocation is equal to 1 if there are a temporary relocation area in the barangay; Move-in is equal to 1 if there was a large or significant number of households who moved in or transferred to their barangay in the last five years due to environmental or peace and order reasons; Move-out is equal to 1 if there was a large or significant number of households who moved out or transferred due to environmental or peace and order reasons in the last 5 years; Svs_index is the 1 if the barangay has all services such as hospital, puericulture center, fire station, post office, landline, cellular phone signal, public street sweeper; Edu_index is equal to 1 if the barangay has elemtary school, high school and college; Eco_index is equal to 1 if the Baranagay has a wholesale and/or retail trade establishment, recreational establishment, manufacturing establishment, accommodation and food service establishment, financial establishment, establishment offering repair services, establishment offering personal services, other establishments. See Appendix 4 for a more detailed description of the variables. 24 Table A1.2 Full LPM results predicting violent incidents in BARMM, Caraga and Davao by processing status of CADTs (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) VARIABLES Land Land Land Land Land Land No Land No Land No Land No Land No Land No Land Approved -0.001 -0.003 -0.003 -0.004 -0.003 -0.003 -0.097*** -0.055 -0.011 -0.009 -0.017 -0.016 [0.002] [0.002] [0.003] [0.003] [0.003] [0.003] [0.033] [0.039] [0.030] [0.030] [0.028] [0.028] Awarded 0.001 0.002 0.001 0.000 0.001 0.001 -0.107*** -0.038** -0.023 -0.023 -0.026 -0.026 [0.001] [0.002] [0.001] [0.002] [0.001] [0.001] [0.019] [0.018] [0.018] [0.017] [0.016] [0.016] Poverty -0.000* -0.000 -0.000 -0.000 -0.007*** -0.007*** -0.007*** -0.007*** [0.000] [0.000] [0.000] [0.000] [0.001] [0.001] [0.001] [0.001] Urban -0.006*** -0.004*** -0.004*** -0.004*** 0.060*** 0.058*** 0.060*** 0.058*** [0.001] [0.001] [0.001] [0.001] [0.014] [0.014] [0.014] [0.014] active4Ps_2020 -0.000 0.000 0.000 0.000 -0.000* -0.000* -0.000** -0.000** [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Mining 0.001 0.001 0.000 0.000 -0.036* -0.035* -0.033* -0.033* [0.002] [0.002] [0.002] [0.002] [0.020] [0.020] [0.020] [0.020] prov = 24 -0.002** -0.003*** -0.002** -0.003*** -0.003*** -0.069*** -0.081*** -0.082*** -0.078*** -0.077*** [0.001] [0.001] [0.001] [0.001] [0.001] [0.023] [0.022] [0.022] [0.021] [0.021] prov = 25 0.005** 0.006*** 0.007*** 0.007*** 0.007*** -0.052*** 0.060** 0.059** 0.059** 0.058** [0.002] [0.002] [0.002] [0.002] [0.002] [0.020] [0.027] [0.027] [0.027] [0.027] prov = 36 0.008 0.004 0.003 0.004 0.003 -0.054* -0.011 -0.009 -0.013 -0.016 [0.006] [0.006] [0.006] [0.006] [0.006] [0.031] [0.034] [0.035] [0.034] [0.034] prov = 38 0.010*** 0.011*** 0.010*** 0.010*** 0.010*** -0.084*** 0.093*** 0.095*** 0.091*** 0.092*** [0.003] [0.003] [0.003] [0.003] [0.003] [0.017] [0.032] [0.031] [0.031] [0.031] prov = 66 0.000 0.004 0.002 0.003 0.002 -0.035* 0.325*** 0.329*** 0.317*** 0.318*** [0.002] [0.004] [0.004] [0.005] [0.005] [0.020] [0.063] [0.062] [0.061] [0.061] prov = 67 0.001 0.000 0.001 0.001 0.001 0.031* 0.142*** 0.141*** 0.139*** 0.137*** [0.002] [0.002] [0.002] [0.002] [0.002] [0.019] [0.024] [0.024] [0.024] [0.024] prov = 68 0.014*** 0.013*** 0.014*** 0.014*** 0.014*** 0.006 0.143*** 0.142*** 0.140*** 0.139*** [0.004] [0.004] [0.004] [0.004] [0.004] [0.019] [0.027] [0.027] [0.027] [0.027] prov = 70 0.000 0.002 0.001 0.001 0.001 -0.016 0.140*** 0.142*** 0.142*** 0.143*** 25 [0.003] [0.003] [0.003] [0.003] [0.003] [0.026] [0.035] [0.035] [0.034] [0.034] prov = 82 0.005** 0.007*** 0.007*** 0.006*** 0.006*** -0.030 0.042 0.041 0.048* 0.047* [0.002] [0.002] [0.002] [0.002] [0.002] [0.030] [0.028] [0.028] [0.027] [0.027] prov = 85 0.021** 0.019** 0.020** 0.020** 0.020** 0.001 0.162*** 0.160*** 0.162*** 0.162*** [0.009] [0.009] [0.009] [0.009] [0.008] [0.021] [0.030] [0.031] [0.032] [0.032] prov = 86, omitted - - - - - - - - Year = 2012 -0.001 -0.001 -0.001 -0.001 -0.001 -0.011 -0.012 -0.012 -0.012 -0.012 [0.002] [0.002] [0.002] [0.002] [0.002] [0.011] [0.011] [0.011] [0.011] [0.011] Year = 2013 -0.003 -0.003 -0.002 -0.002 -0.002 -0.045*** -0.045*** -0.045*** -0.046*** -0.046*** [0.002] [0.002] [0.002] [0.002] [0.002] [0.011] [0.011] [0.011] [0.011] [0.011] Year = 2014 -0.003** -0.003** -0.003** -0.003** -0.003** 0.023* 0.023* 0.023* 0.022* 0.023* [0.002] [0.002] [0.002] [0.002] [0.002] [0.013] [0.013] [0.013] [0.013] [0.013] Year = 2015 -0.005*** -0.004*** -0.005*** -0.005*** -0.005*** 0.084*** 0.083*** 0.083*** 0.083*** 0.083*** [0.002] [0.002] [0.002] [0.002] [0.002] [0.014] [0.013] [0.013] [0.013] [0.013] Year = 2016 -0.008*** -0.008*** -0.008*** -0.008*** -0.008*** 0.098*** 0.098*** 0.098*** 0.097*** 0.098*** [0.003] [0.003] [0.003] [0.003] [0.003] [0.020] [0.020] [0.020] [0.019] [0.020] Year = 2017 -0.001 -0.000 -0.001 -0.001 -0.001 0.147*** 0.148*** 0.148*** 0.148*** 0.148*** [0.004] [0.004] [0.004] [0.004] [0.004] [0.019] [0.018] [0.018] [0.018] [0.018] Year = 2018 -0.005 -0.005 -0.005 -0.005* -0.005 0.288*** 0.289*** 0.290*** 0.290*** 0.290*** [0.003] [0.003] [0.003] [0.003] [0.003] [0.019] [0.019] [0.019] [0.019] [0.019] Year = 2019 -0.005 -0.005 -0.005 -0.005 -0.005 0.255*** 0.257*** 0.257*** 0.258*** 0.258*** [0.003] [0.003] [0.003] [0.003] [0.003] [0.019] [0.019] [0.019] [0.019] [0.019] Year = 2020 -0.002 -0.002 -0.002 -0.002 -0.002 0.219*** 0.223*** 0.223*** 0.224*** 0.224*** [0.004] [0.004] [0.004] [0.004] [0.004] [0.022] [0.022] [0.022] [0.022] [0.022] Central -0.003*** -0.003*** -0.003*** 0.006 0.006 0.005 [0.001] [0.001] [0.001] [0.012] [0.012] [0.012] Highway -0.005** -0.005* -0.004 0.009 0.009 0.008 [0.002] [0.002] [0.003] [0.015] [0.014] [0.015] Precarious 0.001 0.001 -0.014 -0.015 [0.001] [0.001] [0.010] [0.010] Relocation -0.003** -0.003** -0.008 -0.008 [0.001] [0.001] [0.016] [0.016] moved_in 0.002** 0.002** -0.015 -0.014 [0.001] [0.001] [0.013] [0.013] moved_out -0.003** -0.003** 0.023 0.022 26 [0.001] [0.001] [0.017] [0.017] edu_index 0.004 -0.039** [0.003] [0.020] svs_index -0.004 -0.014 [0.009] [0.034] eco_index -0.004** 0.018 [0.002] [0.011] prov = 86 -0.002 -0.197*** [0.002] [0.033] Constant 0.007*** 0.007*** 0.011*** 0.016*** 0.016*** 0.018* 0.605*** 0.589*** 0.625*** 0.616*** 0.628*** 0.667*** [0.001] [0.001] [0.002] [0.003] [0.003] [0.009] [0.007] [0.013] [0.020] [0.024] [0.025] [0.041] Observations 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 41,380 R-squared 0.000 0.004 0.005 0.006 0.006 0.006 0.009 0.039 0.042 0.042 0.043 0.043 *** p<0.01, ** p<0.05, * p<0.1 Barangay-clustered standard errors in brackets Urban is a dummy variable that indicates where the population is larger than 5,000 people. Mining is a dummy variable that equals 1 if the barangay contains a mining site and 0 otherwise. Poverty is the incidence of poverty at provincial level in 2021. Active4Ps is a variable that measures the number of beneficiaries of active development projects happening in the barangay. Central is equal to 1 if the baranguay is a poblacion/central district or part of the poblacion/central district; Highway is equal to 1 if the barangay is accessible to the national highway; Precarious is equal to 1 if the household reside in a precarious location; Relocation is equal to 1 if there are a temporary relocation area in the barangay; Move-in is equal to 1 if there was a large or significant number of households who moved in or transferred to their barangay in the last five years due to environmental or peace and order reasons; Move-out is equal to 1 if there was a large or significant number of households who moved out or transferred due to environmental or peace and order reasons in the last 5 years; Svs_index is the 1 if the barangay has all services such as hospital, puericulture center, fire station, post office, landline, cellular phone signal, public street sweeper; Edu_index is equal to 1 if the barangay has elemtary school, high school and college; Eco_index is equal to 1 if the Baranagay has a wholesale and/or retail trade establishment, recreational establishment, manufacturing establishment, accommodation and food service establishment, financial establishment, establishment offering repair services, establishment offering personal services, other establishments. See Appendix 4 for a more detailed description of the variables. 27 Appendix 2: List of variables Variable Definition land Equals to 1 if the conflict is land-related and 0 otherwise noland Equals to 1 if the conflict is not-land-related and 0 otherwise insidecadt Equals 1 if incident X took place inside an ancestral domain and 0 otherwise Ips Percentage of indigenous population awarded Equals 1 if the territory for incident X was an area with awarded CADT and 0 otherwise. approved Equals 1 for incidents taking place in approved but not fully-awarded CADT areas and 0 otherwise poverty Incidence of poverty at provincial level in 2021 urban Equals to 1 if population is larger than 5,000 people active4ps202 Measures the number of beneficiaries of active development projects happening in the 0 barangay mining Equals 1 if the barangay contains a mining site and 0 otherwise central Equal to 1 if the Baranguay is a poblacion/central district or part of the población/central district highway Equal to 1 if the barangay is accessible to the national highway precarious Equal to 1 if the household reside in a precarious location (along estero/cree/waterway, riverbank/shoreline, railroad, garbage dumpsite, under a bridge, along sidewalk or easement of roads and highways, cliff/cementery/others, government land without legally recognizable claim to the land, private land they do not own relocation Equal to 1 if there are a temporary relocation area in the barangay moved_in Equal to 1 if there was a large or significant number of households who moved in or transferred to their barangay in the last five years due to typhoon, other natural calamity/disaster, peace and order problem, other reasons moved_out Equal to 1 if there was a large or significant number of households who moved out or transferred to their barangay in the last five years due to typhoon, other natural calamity/disaster, peace and order problem, other reasons edu_index Equal to 1 if the barangay has elementary school, high school or college svs_index Equal to 1 if the barangay has any of these services such as hospital, puericulture center, fire station, post office, landline, cellular phone signal, public street sweeper eco_index Equal to 1 if the Baranagay has any of the following: a wholesale and/or retail trade establishment, recreational establishment, manufacturing establishment, accommodation and food service establishment, financial establishment, establishment offering repair services, establishment offering personal services, other establishments. 28