Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

World Development 128 (2020) 104837

Contents lists available at ScienceDirect

World Development
journal homepage: www.elsevier.com/locate/worlddev

Is collective titling enough to protect forests? Evidence from


Afro-descendant communities in the Colombian Pacific region
Maria Alejandra Vélez a,⇑, Juan Robalino b, Juan Camilo Cardenas c, Andrea Paz d, Eduardo Pacay e
a
School of Economics and Center for the Study of Security and Drugs (CESED), Universidad de Los Andes, Bogotá, Colombia
b
Escuela de Economía e Instituto de Investigaciones en Ciencias Económicas, Universidad de Costa Rica and CATIE, Costa Rica
c
School of Economics, Universidad de Los Andes, Bogotá, Colombia
d
Biology Program, City University of New York and Graduate Center, City University of New York, United States
e
EfD Central America, CATIE, Costa Rica

a r t i c l e i n f o a b s t r a c t

Article history: During the mid-1990s, one of the most ambitious land reforms in recent decades took place in Colombia.
Available online 31 December 2019 The reform recognized collective land rights of almost 6 million hectares to Afro-Colombian communi-
ties, with the dual goals of improving livelihoods and preserving valuable ecosystems. We estimate the
Keywords: impact of this collective land titling program on forest cover using panel data and a difference-in-
Afro-descendants difference empirical strategy. We find that overall, collective titling significantly reduces deforestation
Collective property rates, but the effect varies substantially by sub-region. We observe that the larger effects are in Nariño
Colombia
and Valle del Cauca. Our qualitative analysis suggest that this might be the result of local community-
Land titling
Forest cover
based organization defining the rules for community use of natural resources and the expulsion of private
Impact evaluation companies dedicated to timber exploitation and oil palm plantations. We conclude that under the ade-
quate conditions, collective titling can lead to forest conservation.
Ó 2019 Elsevier Ltd. All rights reserved.

1. Introduction 2006; Robinson et al., 2014; Blackman, Corral, Lima, & Asner,
2017; Blackman & Veit, 2018).
The conservation of forests for the provision of ecosystem ser- Using satellite imagery and panel data, our goal in this article is
vices continues to be one of the most important and difficult chal- to analyze whether the assignment of collective property rights to
lenges in current times. The short-run rents driving extractive groups of ancestral Afro-descendant communities in the Pacific
activities seeking timber and minerals from forested lands, and region of Colombia has changed deforestation rates in areas of high
pressures from expanding agriculture and livestock, threaten the biodiversity and high levels of poverty (Peña, Vélez, Cárdenas,
capacity of forests to provide valuable ecosystem services such as Perdomo, & Matajira, 2017).
sustaining local food, climate regulation, and supporting biodiver- The process of titling land to Afro-Colombian communities in
sity (Geist and Lambin, 2001, 2002; Lambin et al., 2001). the Pacific Region began after the 1991 Colombian Constitution
Multiple strategies for conserving forests have emerged. These (Velez, 2011). This process has led to the titling of almost six mil-
include the declaration of protected areas through national or lion hectares over the last two decades. Created as a social policy
regional parks, the privatization of land for conservation by indi- while conserving valuable ecosystems, collective land titling was
viduals and NGOs, and the assignment of collective property rights based on the premise that Afro-Colombian communities occupying
to local communities (Robinson, Holland, & Naughton-Treves, territories with high ecological value could consolidate sustainable
2014). Today, 27% of forest in the developing world is communal management of their natural resources (Ng’weno, 2001; Offen,
property or used by communities through joint access arrange- 2003; Plant & Hvalkof, 2001; World Bank, 2005; Sánchez &
ments (Agrawal, 2007 citing White and Martin, 2002). However, Roldan, 2002; Sánchez & García, 2006).
there is mixed evidence on how effective these strategies are in We estimate the impact of collective land-titling to Afro-
stopping deforestation (see for example Ostrom & Nagendra, descendent communities in Colombia on forest cover using point-
level panel data along with difference-in-difference models. We
use forest data from the 1990–2000, 2000–2005 and 2005–2010
⇑ Corresponding author.
periods. We control for community and period fixed effects, imply-
E-mail addresses: mavelez@uniandes.edu.co (M.A. Vélez), robalino@catie.ac.cr
(J. Robalino), jccarden@uniandes.edu.co (J.C. Cardenas), epacay@catie.ac.cr ing that fixed differences in communities and periods would not
(E. Pacay). bias our estimations of the effects of titling on deforestation. Our

https://doi.org/10.1016/j.worlddev.2019.104837
0305-750X/Ó 2019 Elsevier Ltd. All rights reserved.
2 M.A. Vélez et al. / World Development 128 (2020) 104837

identification assumption is the inexistence of unobservable effects Our results complement previous studies that find that the
that vary both over time and by communities, and that are simulta- titling process had positive impacts on certain socio-economic out-
neously associated with titling. Therefore, we implicitly assume comes. Peña et al. (2017) estimated the impact of collective land
that if titling would not have occurred from one period to the other, titling on the well-being of Afro-descendent communities in
the changes in newly titled communities would have been the same Colombia. Their findings suggest that collective land titling creates
as those in communities were titling status did not change. We also more secure property rights and a longer time horizon for house-
implicitly assume that the effect of titling is constant over time. holds in collective territories, which leads to investment in their
We find that, overall, titling in the Pacific Region has decreased private physical and human capital. Our results suggest that house-
deforestation by more than a one percentage point. These effects holds and communities might also be investing in natural capital.
represent more than 27% reductions of what would have been The rest of the paper is organized as follows. In the following sec-
the deforestation rates without titling. These results remain robust tion, we discuss the literature related to property rights and forest
with the inclusion of control variables that capture market dynam- conservation and propose a theory of change as a theoretical frame-
ics, such as distance to navigable rivers, distance to roads, distance work. Then, we describe the specific context of the Pacific region and
to forest frontiers and the presence of illicit crops. When we sepa- the scope of Law 70. In the third section, we explain the data used
rate the analysis by sub-regions, our results show that collective and our empirical strategy. The fourth section presents and dis-
titling did not increase deforestation in any of the regions we stud- cusses our results based on both our qualitative and quantitative
ied. In fact, we find important and statistically significant effects data. Finally, our conclusions are included in the final section.
reducing deforestation in two of the four sub-regions studied:
Nariño and Valle del Cauca.
We also conducted 54 interviews with community leaders in 2. Background
each of the sub-regions included in our study to understand the
titling process, the drivers of deforestation and the channels through 2.1. Relevant literature: empirical evidence on collective titling
which collective property might protect forests. Our interviews pro-
vide evidence that the reduction of deforestation in titled communi- The relevance of evaluating the impact of conservation policies
ties is explained mainly by two forces: i) local community-based is given by the magnitude and greater amount of resources spent
organization defining the rules for community use of natural (James, Gaston, & Balmford, 1999; James, Gaston, & Balmford,
resources, mainly in Valle as also reported in Velez, 2011; and ii) 2001; Pearce, 2005). However, research focuses on the effects of
the expulsion of private companies dedicated to timber exploitation restricting land use by using command and control policies such
in Valle, and mainly oil palm in Nariño, both activities identified as as protected areas on deforestation outcomes (see Andam,
important drivers of deforestation in the region (Armenteras, Ferraro, Pfaff, Sanchez-Azofeifa, & Robalino, 2008; Joppa & Pfaff,
Cabrera, Rodríguez, & Retana, 2013). In the absence of collective 2010; Sims, 2010; Ferraro, Hanauer, & Sims, 2011; Pfaff,
titling and community organization triggered by this collective Robalino, Sanchez-Azofeifa, Andam, & Ferraro, 2009; Pfaff,
titling process, the same trend of resource extraction, and agricul- Robalino, Lima, Sandoval, & Herrera, 2013; Pfaff, Robalino,
tural expansion from previous decades would have continued. Herrera, & Sandoval, 2015; Robalino, Sandoval, Barton, Chacon, &
The importance of evaluating the impact of collective titling on Pfaff, 2015) and the effects of incentive policies such as payments
forest cover is immense. First, the Pacific region is inhabited by a for ecosystem services (Alix-Garcia, Shapiro, & Sims, 2012;
historically marginalized Afro-descendent population (Galvis- Arriagada, Ferraro, Sills, Pattanayak, & Cordero-Sancho, 2012;
Aponte, Moyano, & Alba-Fajardo, 2016) representing about 4% of Robalino & Pfaff, 2012).
the country’s population and 61.5% of the population in the Pacific1 Evaluating the impact of titling (mainly collective) programs on
(DANE, 2005). Second, the Pacific is considered one of the most environmental outcomes and in particular, forest cover, has been
diverse regions on the planet, due to its high levels of endemic species less common. In fact, we have identified few cases, all in Latin
and species variety (WWF, 2002). The natural wealth of the territories America, that aim to establish causality by controlling for possible
entitled to Afro-Colombian communities, apart from their biodiver- confounding factors (See for example Liscow, 2013 in Nicaragua
sity and 5% of the country’s natural forest, includes mineral resources and Buntaine, Hamilton, & Millones, 2015 in Ecuador). Similarly
that have historically been part of the region’s economic culture, as to our study, Blackman et al. (2017) evaluate the impact of titling
well as part of its conflict. In addition, it has valuable ecosystems collective lands to indigenous communities in the Peruvian Ama-
due to the water cycle and consequently for the provision of environ- zon using community-level longitudinal data derived from high-
mental services in the region and beyond. Third, the amount of land resolution satellite images and found that tiling reduces both clear-
that has been titled is quite substantial and some argue that this ing and forest disturbance. Blackman and Veit (2018) extend this
effort might be one of the most important land reforms in Latin study to other countries in the Amazon region to evaluate the
America (Offen, 2003). Similar collective titling processes are taking impact of indigenous community management in Bolivia, Brazil,
place in territories occupied by Afro-descendants in Ecuador, Ecuador, and Colombia. Using propensity score matching to ana-
Panama, and Brazil; thus, our results could be useful to promote lyze the effects on 2001–2013 deforestation and forest carbon
and design effective programs. Fourth, our study is relevant for a vari- emissions, they find that indigenous community management
ety of branches in academic literature discussing the role of property reduces both deforestation and forest carbon emissions in Bolivia,
rights on conservation (Baland & Platteau, 2003; Schlager & Ostrom, Brazil and Colombia. They did not find an effect for Ecuador. Also,
1992). Finally, at country level, Law 70, supporting the collective for Colombia, Bonilla and Higuera (2016) evaluate the impact of
titling process, is currently under public discussion and further devel- protected areas, including collective lands owned by Afro-
opment in the context of the implementation of REDD+ projects. Sci- Colombian communities in the Pacific Region using non-
entific data will provide crucial information for this discussion to protected lands as controls. Additionally, Romero and Saavedra
design and implement conservation projects in the region. (2018) focus on titling and use different types of control such as
national parks, indigenous lands, forest reserves, and undeclared
1
lands. Our analysis, however, differs in two important points. First,
Afro-descendants in the country make up about 10.31% of the total population
according to official records (de Cultura, 2010). In fact, after Brazil and Venezuela,
our control observations only involve Afro-Colombian communi-
Colombia is the third country in Latin America with the biggest population of Afro- ties without titles. It is important to note that choices of control
descendants (Freire, Diaz-Bonilla, Schwartz Orellana, Soler Lopez, & Carbonari, 2018). observations change the counterfactual and, therefore, the
M.A. Vélez et al. / World Development 128 (2020) 104837 3

research questions. In particular, we address how titling affects two challenges but the mechanisms may differ depending on
Afro-Colombian communities. This implies that our research ques- whether the property rights are assigned individually or collec-
tion is related to how individuals in these communities change tively. Individual property rights, adequately enforced, may reduce
behavior by estimating what they would have done in the absence the collective action problem by excluding any other potential for-
of titling. Using other forest areas outside Afro-Colombian commu- est user and leave all possible streams of costs and benefits in the
nities includes within the counterfactual how other regimes would hands of the owner. When the property rights over the forested
have behaved, and the question is related to other regimes. Second, land are assigned to a group, its members have incentives to
we explore regional differences within the Pacific to capture the exploit the forest, capture the private benefits and spread the costs
heterogeneous organizational dynamics and deforestation drivers over the rest of the group by overexploiting the forest resources
in each department. (Hardin, 1968). However, as discussed by Ostrom (1990) and
As Blackman et al. (2017) reviewed, the rest of the literature Ostrom and Nagendra (2006), the institutional arrangements of
exploring the relation between collective titling or community- the group could eventually deal with this problem. As for the case
based management and forest cover, involves more than 100 pub- of the inter-temporal challenge, both individual and collective
lished papers without aiming at establishing causality but giving ownership will be subject to the behavioral problem of impatience,
important insights regarding the mechanisms behind land titling namely that higher discount rates will induce exploiting the forest
as a conservation tool. Many of the articles also compare the effec- resources in the short term, threatening the possibility of sustain-
tiveness of collective lands versus other forms of protected lands. ing the forest capacity to provide goods and ecosystem services
Focusing on Latin America, mainly Mexico, several studies explore over longer periods. Further, the community’s institutional
the variables determining conservation outcomes (See for example arrangements may include mechanisms that could alter such dis-
Alix-Garcia, 2007; Barsimantov & Kendall, 2012; Ellis & Porter- count rates through culture, norms, and regulations.
Bolland, 2008; Rodriguez Solorzano & Fleischman, 2018; Tucker, There is broad-ranging agreement that the lack of property
1999; Vergara-Asenjo & Potvin, 2014; Holland et al., 2014; rights becomes an obstacle to economic development and the effi-
Holland et al., 2017). cient use of resources. Collective rights over a shared resource cre-
Aggregate results show overall favorable outcomes for decen- ate incentives for community members to stay vigilant and impede
tralized forest management although it really depends on the local invasion by intruders, as well as to invest in resources for their
institutional design. Larson (2011) reports a review from 152 for- future benefit (Baland & Platteau, 2003; Schlager & Ostrom,
ests in nine countries showing that ‘‘forest degradation is inversely 1992). In theory, collective titling constitutes the first step towards
related to strong collective action and rule enforcement at the local a sustainable use of resources (Ostrom, 1990). It is quite likely that
level” (p.542). Porter-Bolland et al. (2012) present a meta-analysis assigning more secure property rights to these communities will
of 40 protected areas and 33 community-managed forests in the have a political and cultural impact on allowing the preservation
tropics and evaluate deforestation rates. They find that on average, of traditions and stopping outsiders from threatening their liveli-
community managed forests present lower and less variable rates hoods and the natural base. However, it is not straightforward
of deforestation than protected areas. Cronkleton and Larson what such titling of lands could lead to regarding the use of the
(2015), report a study across 21 indigenous and mestizo communi- natural base (See Blackman et al., 2017 for a discussion of these
ties in Napo, Ecuador, and Huánuco and Ucayali, Peru, concluding mechanisms). Internal forces emerging from the now more auton-
that even though formal recognition of rights is an ideal situation, omous organizations can generate different processes of land use
it is not always a sufficient factor to provide security. Robinson change, and interestingly, this can work in either direction. Decen-
et al. (2014) reviewed a large set of empirical studies in order to tralizing forest management and empowering communities could
identify the relationship between the form of tenure, tenure secu- lead to greater autonomy in establishing local enterprises based
rity, and deforestation outcomes. They report that results seem to on cash crops and timber extraction, producing much needed mon-
be context dependent. Referring to communal land, the authors etary resources, especially if poverty is widespread among its
report positive conservation outcomes for Central America but less members. However, this would entail reducing the possibility of
so for Africa, given regional conflicts and/or weak governance2. higher levels of governance to impose national agendas, such as,
Based on the reviewed literature and focusing on the effect of conservation. One could envision organizations craving fresh cash
collective tilting, we next offer a theory of change for explaining resources signing agreements, even within the law, to exploit tim-
the mechanisms through which the assignment of collective prop- ber products with external companies if, for example, the commu-
erty rights to the Afro-Colombian groups could have led to a reduc- nity organizations were to participate in value chains active in the
tion in the rates of deforestation observed, compared to those region, leading to higher deforestation rates. Alternatively, such
lands where no collective titling occurred. devolution of power could mean the strengthening of a territorial
occupation model that could align, with the right local institutions,
2.2. Theory of change: a multi-level approach of incentives and the goals of forest conservation with cultural and economic sys-
institutions tems based on a more sustainable use of forests. Collective prop-
erty rights over the territories inhabited by these communities
Groups that share a common pool resource such as forested can become crucial institutional determinants of the decisions
land will have to solve two main challenges. One, to sustain the made at the level of the community organizations in charge of
capacity of such an asset to produce a stream of net positive ben- defining yearly action plans in various ways. First, these rights over
efits for the community over time, and second, the collective action the land can extend the time horizon for which the community
problem among group members. The first challenge involves a would evaluate the net present value of the stream of benefits
problem of inter-temporal preferences and the second one implies and costs between conservation and exploitation. Secondly, secure
a dilemma between individual and group incentives. The assign- property rights could deter the arrival of external forces, such as
ment of property rights to the members of the group as suggested legal but also illegal sawmills and illegal mining, both of which
by the empirical evidence discussed above, is key to solving these drive deforestation. Third, under a more formal property rights
regime, communities could access financial mechanisms such as
2
See also Robinson et al. (2018), who present a framework that conceptualizes the
the Program on Reducing Emissions from Deforestation and Forest
relationship between land tenure, conservation outcomes and land management Degradation in Developing Countries (REDD+) by generating car-
decisions. bon credits from conservation.
4 M.A. Vélez et al. / World Development 128 (2020) 104837

Our theory of change is based on three layers or scales (micro, of social norms and self-governed institutions to solve the collective
meso, macro) at which these mechanisms interact, and draws action dilemma among the beneficiaries of the land title.
directly from the Institutional Analysis and Development (IAD)
framework (McGinnis, 2011; Ostrom, 2005) and the multi-layer
2.3. Collective titling in the Pacific region
framework developed in Cárdenas and Ostrom (2004).
At the micro-behavioral level, group members make decisions
In Colombia, indigenous groups gained recognition of their col-
about using the shared resources from the forested land pursuing
lective occupation and governance of the territory since 1890 with
their well-being, based on the incentives to exploit or preserve the
Law 89 (Plant & Hvalkof, 2001). Later, during the 1980s, indigenous
forested land. This is the local action arena in the IAD model in which
communities received collective titles of barren lands in the Amazon
interactions between community members and their physical envi-
region under the legal form of ‘‘resguardos” (reserves) (Zabala,
ronment lead to individual and group outcomes. At the meso level, a
2017), but these ethnic organizations were not fully recognized until
number of formal and informal community arrangements will affect
19913. In contrast, Afro-Colombian communities, despite occupying
those incentives for the group members. Social norms, intra-group
the Pacific region since the seventeenth century, did not receive this
regulations, conventions and traditions can all be part of these insti-
formal and legal recognition up until the new Constitution in 1991.
tutional mechanisms that can deter or enhance certain conservation
In fact, until 1991, Afro-Colombian communities inhabited the Pacific
or exploitation activities. The operating and collective choice rules
region, on lands that lacked formal arrangements, regarding individ-
in the IAD framework and the identity and group-context layers in
ual or communal property rights. For the Colombian Government this
the Cardenas & Ostrom model are in operation here. Finally, at
entire region, ranging from Ecuador to Panamá and adjacent to the
macro level, certain legal structures (IAD Constitutional Rules) will
western Andes to the East, which includes mainly four subregions
constrain the meso and micro level decision making depending on
(Chocó, Valle del Cauca, Cauca and Nariño4) was considered a massive
the capacity of the state to enforce such rules.
forest reserve of ‘‘unoccupied lands” with no recognition of the ances-
Let us now put these in operation for the case of these collective
tral presence of Afro- Colombian communities (Plant & Hvalkof, 2001).
titling of lands to the Afro-Colombian communities in case.
Afro-Colombian communities occupied the territory since the
Through the recognition of these rights, several mechanisms enter
colonial period under rather low densities in forested lands with
into operation, altering the use of forests. First of all, the titling
informal arrangements and ‘de facto’ private holdings with no legal
reduces the costs for the group members to signal to outsiders that
titles to support them, but social norms and conventions that
the access to these resources is now limited to the members of the
evolved over time (Velez, 2011). Contrary to indigenous organiza-
group, and increases the probability that authorities intervene on
tions, these Afro-descendent communities did not have hierarchical
behalf of the community assembly to expel or restrain intruders.
structures for governing their internal affairs, although leaders
Secondly, with the assignment of the collective title, the time hori-
played an important role in organizing various community projects.
zon is now extended for the members of the community, in the
Before 1991, the Pacific region was a ‘de facto’ open access land
sense that those interested in staying in this land can be assured
with no resource governance or rules imposed by Afro-Colombian
that the rights of access will remain for a much longer time. In
communities (Velez, 2011) and which was at the mercy of several
our particular case, the legal arrangement, expressed in the consti-
external actors trying to extract resources from the territory. In this
tution itself, precludes the fractioning or selling parcels of the col-
context, lessons from the indigenous movement that had raised
lective title by group members. With longer time horizons, the net
similar concerns and had achieved success in getting their collec-
present value of conservation against exploitation increases
tive rights recognized, led these new Afro- movements to petition
(Baland & Platteau, 1996; De Janvry & Sadoulet, 2001). For shorter
for the recognition of their self-governing structures and their his-
periods and under insecure or undefined property rights, the net
torical occupation of these lands.
present value of extraction of the resources will outweigh the
The new Colombian Constitution included a transitory article
costs, including the opportunity cost of foregone ecosystem ser-
(55), which acknowledged the ancestral occupation of these terri-
vices from the forest. As the timeframe extends, the net present
tories by Afro-Colombian communities and mandated that the
value of those ecosystem goods and services will win over the
State was obliged to recognize and formalize their right to these
opportunity cost of the foregone income from not exploiting the
lands as ethnic groups. In 1993, a new law (Ley 70/1993) was
material resources in the short run.
issued, establishing the path for the titling of these lands to com-
The meso and macro levels will play a fundamental role in deter-
munity organizations that became the authority over communal
mining, first legally, the possibility of extending this time horizon,
land. Law 70 triggered the formation of Community Councils, the
and culturally, through social norms in building agreements among
new political local entity in charge of managing the collective
the community members that a local livelihood based on the
lands. The newly formed Community Councils gained the right to
forested land is possible. A great deal of the initial collective titling
exploit the natural resources in their delimited territory and the
cases (See Appendix 2 for the timeline of historical events in each of
right to exclude outsiders (Velez, 2011). Using the typology by
the four subregions) includes motivations by the original commu-
Schlager and Ostrom (1992) to classify the heterogeneous bundle
nity associations to build a self-determined model of livelihood
of territorial rights, these Afro-Colombian communities gained
based on the Afro-Colombian rural identity. Also, the legal frame-
access, extraction, management, and exclusion rights and only
work will facilitate the exclusion of outsiders competing for extract-
lacked ‘‘alienation rights” for the land5, impeding these communi-
ing short-run rents from these forested lands. In fact, as one can see
in the timeline of these four subregions (See Figs. A5–A8 of Appen-
3
dix 2), several initial collective titling processes were motivated by See an evolution of indigenous rights at http://www.banrepcultural.org/bib-
lioteca-virtual/credencial-historia/numero-146/estado-y-pueblos-indigenas-en-el-
the fear that mining and timber companies were interested in these
siglo-xix.
extraction processes, along with more efficient technologies, threat- 4
In the pacific region, the Community Councils of Afro-Colombian communities
ening the possibilities of establishing a more sustainable livelihood are distributed (by number of councils), according to the shapefiles obtained from the
system for the groups occupying these territories. Agencia Nacional de Tierras (ANT) and the Departamento Administrativo Nacional de
Estadística (DANE), as follows: 34.29% in Choco, 28.57% in Nariño, 22.86% in Valle del
In sum, the collective titling process reduces the cost of excluding
Cauca (mainly in Buenaventura municipality), 10.29% in Cauca, 2.86% in Antioquia
outsiders interested in extractive activities, extends the time hori- and 1.14% in Risaralda.
zon improving the net present value of conservation benefits against 5
Also, they do not have access to the sub-soil that belongs to the state but have
the short-run rents from extraction, and encourages the generation priority to access a mining concession in case one is requested by a third party.
M.A. Vélez et al. / World Development 128 (2020) 104837 5

1991 1993 1994 1996 1998 2010 2019

Colombian Constitution Law 70 - Titling Credit by the First Decree 1320 - End of More than
Transitory Article 55: of collective World Bank to collective Right to prior period of 170 CC with
Announce creation of lands start titling titling in Consultation study. Up to collective
new Law to recognize process Chocó 141 CC with titles
right to collective collective
property titles

Fig. 1. Collective Titling Process to Afro Colombian Communities.

ties from fracturing, dividing or selling their lands, nor use it as col- For each interview, we use the deforestation maps drawn up for
lateral for credit purposes. These characteristics implied a clear def- each sub-region for the periods of 1990, 2000, 2005 and 2010 and
inition of the property rights that prevented a future contra-reform asked the interviewees to explain, according to their perceptions
or pressures to sell their lands. and local views, the drivers of deforestation in each period. Each
The process that led to this collective titling reform included a interview lasted between 45 and 75 min. We conducted 13 inter-
decade of social organization supported by Catholic missionaries views in Chocó, 13 in Valle del Cauca, 10 in Cauca, 14 in Nariño,
(e.g. Claretianos) and indigenous organizations that inspired an and 4 with national experts6.
Afro-rural movement (e.g. Movimiento Cimarrón and the PCN - It is important to note that these leaders were not chosen ran-
Proceso de Comunidades Negras) aiming to empower the commu- domly. We relied upon our previously existing network of local
nities in the region (see Escobar, Pedrosa, Cárdenas, & Correa, contacts to identify leaders with knowledge of the territory and
1996; Grueso, Rosero, & Escobar, 1998). Such social movements deforestation process. We identified leaders that were interested
emerging throughout the Chocó region (northern sub-region of in discussing the challenges and problems associated to deforesta-
the Pacific coast) and Valle del Cauca (mainly in the city of Bue- tion in the region. Also, and given the security concerns present
naventura), were also responding to an increase in pressures from (some drivers of deforestation give rise to security concerns), we
industrial logging and mining projects, in many cases authorized only interviewed leaders recommended by our contacts to avoid
by environmental and governmental authorities. The tension exposing them or the researchers to significant risks.
between these external forces for extracting resources and the
internal ones associated with the subsistence economies of these 3.2. Quantitative data
ethnic communities set the stage for this land reform.
With the mandate from Law 70 in 1993, the national government Our area of analysis involves a total of 141 Afro-Colombian com-
and its recently created Ministry of the Environment requested munities that received titles between 1996 and 2010 (See Fig. 2). We
financial support from the World Bank in 1994 for the execution obtained information of official forest cover data from 1990, 2000,
of this law and titling the first 2.3 million hectares (Sánchez & 2005 and 2010 and forest cover change information for 1990–
Roldan, 2002). The first collective title, over half a million hectares, 2000, 2000–2005 and 2005–2010 developed and provided by the
was issued in 1996, to the peasant organization ACIA, newly orga- Instituto de Hidrología, Meteorología y Estudios Ambientales
nized as a ‘‘Community Council” under the name COCOMACIA. Dur- (IDEAM). Both the forest cover and the forest cover change are devel-
ing the following two decades, this process led to the collective oped by IDEAM based on freely available Landsat imagery at a 30 m
titling of more than 170 Community Councils accounting for almost resolution (Cabrera, Galindo, & Vargas, 2011). In a given time period,
six million hectares (see Fig. 1 for a summary of the process). a pixel can only go from one state to another (e.g, Forest to Non-
The process of issuing these collective land titles, along with the Forest) since only the first and last year of the period are available,
local rules and structures required from these community organi- however, a deforested pixel can indeed be reforested in the next time
zations, created an entirely new form of local political governance period. We used official communal titles limits from 2012 from the
that increased the visibility and bargaining power of the region Instituto Geografico Agustín Codazzi (IGAC). Within the study peri-
(Velez, 2011). Further, and influenced by the ILO 169 (Indigenous ods, we know that 50, 64 and 27 communities were titled during the
and Tribal Peoples Convention, 1989 (No. 169)), a later Decree 1990–2000, 2000–2005 and 2005–2010 periods respectively.
(1320 in 1998) extended the autonomy and recognition of these We intersected the IDEAM forest covers with the official collec-
communities by requiring prior consultation to implement any tive territories provided by IGAC to obtain forest cover both for the
project that could potentially affect their territories. entire region and by community. Within these communities, we
drew 100,000 random points in an area of 53,087 km27. We
3. Methods focused on 49,193 points that were covered by forest in 1990. This

3.1. Qualitative data 6


We follow a standard IRB process. We submitted our protocol to the ethical
committee at the School of Management of University of Los Andes. The protocol was
To explore the mechanisms that could have operated through classified as minimal risk and approved (Memo No. 36, 2018). For the interviews we
used an informed verbal consent. We explained participants that answers were
these titling processes and evaluate the impacts on deforestation,
anonymous, that the activity did not imply any risk and that they could leave at any
we undertook a twofold strategy that included a quantitative panel time without any justification. Interviewed was recorded when authorized.
analysis of satellite imagery and a qualitative analysis of interviews 7
Every point can thus fall within a pixel of ~30m resolution (that is ~30m*30m).
conducted with 54 community leaders in each of the sub-regions The full dataset contains 136,171,688 pixels at about 30m resolution. This number
included in our study. Interviews aimed to understand the titling makes it computationally impossible to use the full dataset in computing distances
and performing statistical analyses. Furthermore, using a full spatial dataset
process, the drivers of deforestation, the channels through which generates problems of spatial autocorrelation where neighboring pixels are more
collective property might protect forests and the perceived impact alike than random because of space and thus cannot be considered as independent
of the collective titling. observations.
6 M.A. Vélez et al. / World Development 128 (2020) 104837

non-forest areas, which reflects proximity to forest frontiers10. We


can see that the different characteristics between titled communities
in different periods are significant (see columns D, E, and F). This is
important as these variables are highly relevant in explaining the
forces behind deforestation. Those communities not only differ in
terms of their titling status but also in terms of other characteristics
that affect deforestation.
Furthermore, Table 1, shows that the fraction of titled commu-
nities in each period varies considerably among departments. For
instance, during the 1990–2000 period, most titled communities
were in Chocó (47%), while the rest of the states reach, at most,
23% in Nariño. However, during the 2005–2010 period, most of
the titled communities were located in Nariño (39%) and Valle
del Cauca (39%).

3.3. Econometric specification

We estimate the impact of titling on deforestation by compar-


ing deforestation rates in communities that have received their
title with a credible estimate of what would have happened with-
out titling. In order to adequately estimate this alternative scenar-
io, we use a difference in difference (DID) strategy. We compare
changes in deforestation before and after a community receives a
title, to changes in deforestation in communities where no title
has been granted. Because the titling process has taken place over
the last two decades, we take advantage of this expansion of the
process over time when conducting our analysis (see Fig. 3).
In Fig. 3, we describe the identification strategy in detail. Peri-
ods are defined by the moments where forest cover pictures were
gathered. As mentioned previously, we have four forest pictures:
1990, 2000, 2005 and 2010, with which, we define three periods:
Fig. 2. Communities per titling period and titling years. Note: The label ‘‘not- Period 1 goes from 1990 to 2000, Period 2 goes from 2000 to
included” in Fig. 2 refers to those communities titled after 2010. We did not include
2005, and Period 3 goes from 2005 to 2010. We then divide the
them in the analysis as we do not have forest cover after that.
communities into three groups: those that were titled in the Period
means that we have, on average, one sampled point every 0.5 square 1, (g 1 ), those that were titled in Period 2 (g 2 ), and those that were
kilometers of forest area. For each forest point, we obtained informa- titled in Period 3 (g 3 ). We drop from the analysis all communities
tion on deforestation during each of the periods. We have 17,835; in the period when they were being titled (in treatment in Fig. 3).
23,760; and 7,598 forest points titled in 1990–2000, 2000–2005 We do so because, for instance, those observations in Group 1, have
and 2005–2010, respectively (see Table 1). both treatment status’ within Period 1. Therefore, assigning these
In Table 1, we also describe the dependent variable. We can see communities to either treatment status would cause contamina-
that deforestation rates (e.g. percentage of deforested points) are tion and, in turn, bias in the estimation. Therefore, we only use
different between those communities that received titles in differ- those quadrants with the label pre-treatment and post-treatment
ent periods, even among communities with the same treatment when observations do not change status within periods.
status. For instance, during the 1990–2000 period, deforestation We use fixed-effect panel models. Empirically, the Eq. (1) esti-
rates are very different for those communities that received titles mated is the following:
during the 2000–2005 period and for those that received titles dur-
X
S
ing the 2005–2010 period (Column D). Y it ¼ sdit þ ac þ dt þ bs Is xi þ uit ð1Þ
For each point, we calculated distance to protected areas, rivers, s¼1
and roads as well as land slope and precipitation using the infor-
where Y it takes the value of 1 if a forest point i at the beginning of
mation obtained from the SIG-OT database from IGAC8. We also
period t was deforested by the end of period t, and a value of 0 if it
have information about the presence of coca plantations in 2002
was not deforested by the end of period t; s is the effect of titling
per point, using information provided by the United Nations Office
on deforestation, dit is an indicator variable that takes a value of 1
on Drugs and Crime (UNODC). Moreover, we also use a violence
index, which is built from eight different types of crimes (homicide, if pixel i is located in a community that has a title in period t, and
kidnapping, threat, sexual assault, enforced disappearance, displace- 0 otherwise; dt is the period effect; and ac is the cross-sectional fixed
ment, torture, and terrorism). This index is a global measure of vio- effect. We use two different types of fixed effects in different speci-
lence in Colombia at municipal level that considers both the fications, one, at the treatment-status by department level and, two,
frequency and severity of crimes (based on the penalties for each at community level. The treatment-status by department fixed
type of crime) (Martinez, 2018)9. We further compute distances to effects control for time-invariant unobservable variables belonging
to group j in department k. Community fixed effects control for time
invariant unobservable variables at the community level. By control-
8
See information at http://sigotn.igac.gov.co. ling by fixed effects, we make sure that unobservable time invariant
9
We did not have information regarding illegal gold mining that could also be variables at community level would not bias our estimates.
related to deforestation outcomes. However, other studies have reported that an
increase in gold mining caused an increase in violence measures as homicide rates
10
and victims of massacres which we included in our violence index (Idrobo, Mejia, & From each point, we calculate the distance to the closest area that did not have
Tribin, 2014). forest. These could have been a past deforested area or the coast.
M.A. Vélez et al. / World Development 128 (2020) 104837 7

Table 1
Descriptive Statistics

Variable Total sample Titling year Diff. (C)–(B) Diff. (B)–(A) Diff. (C)–(A)
1990–2000 2000–2005 2005–2010
(A) (B) (C) (D) (E) (F)
Communities
Number of communities 141 51 64 26
Points
Number of points 49193 17835 23760 7598
Dependent Variable
Deforestation rate 1990–2000 4.10 4.35 3.58 5.17 1.59*** 0.77*** 0.82***
Deforestation rate 2000–2005 3.71 3.99 3.29 4.80 1.51*** 0.70*** 0.81***
Deforestation rate 2005–2010 3.05 2.55 3.27 3.34 0.07 0.72*** 0.79***
Community Covariates
Area 29681 30571 33380 18830 14550 2809 11741*
Pixel Covariates
Dist. to National Parks 2012 (km) 45.99 40.87 48.77 49.36 0.59* 7.9*** 8.49***
Dist. to small rivers (km) 0.41 0.53 0.34 0.37 0.034*** 0.19*** 0.16***
Dist. to navigable rivers (km) 3.13 3.46 2.86 3.25 0.394*** 0.6*** 0.21***
Dist. to main roads (km) 73.86 92.90 60.97 69.49 8.52*** 31.93*** 23.41***
Coca zone 2002 (dichotomous) 0.27 0.20 0.21 0.65 0.44*** 0.01 0.45***
Distance to non-forest 2000 (km) 0.76 0.73 0.69 1.03 0.33*** 0.04*** 0.3***
Slope (%) 1.43 1.27 1.54 1.44 0.1*** 0.27*** 0.17***
Rainfall (mm) 5341 4754 5915 4928 987*** 1161*** 174***
Municipality
Violence index 2.99 3.45 2.49 3.48 0.99*** 0.97*** 0.03*
% of points by Departments
Cauca 0.14 0.11 0.17 0.15 0.02*** 0.06*** 0.04***
Chocó 0.47 0.44 0.61 0.09 0.52*** 0.17*** 0.35***
Nariño 0.23 0.30 0.14 0.38 0.24*** 0.16*** 0.08***
Valle del Cauca 0.15 0.16 0.08 0.38 0.3*** 0.08*** 0.22***

Note: Descriptive statistics for pixel covariates are calculated from forests in 1990. ***p < 0.01, **p < 0.05, *p < 0.1

Group 3 (g3) 100


26 communities Pre-treatment Pre-treatment
titled in Period 3 In treatment
90
Percentage of tled communies

80

Group 2 (g2) 70
64 communities Pre-reatment In treatment Post-treatment
titled in Period 2 60
50
40

Group 1 (g1)
30
51 communities In treatment Post-treatment Post-treatment
titled in Period 1 20
10
0
1999

2010
1990
1991
1992
1993
1994
1995
1996
1997
1998

2000
2001
2002
2003
2004
2005
2006
2007
2008
2009

Period 1 Period 2 Period 3

Note: Group 1, Group 2 and Group 3 contain communities titled during 1990-2000, 2001-2004 and 2005
and 2010 respectively.

Fig. 3. Evolution of titled communities over time and the data structure. Note: Group 1, Group 2 and Group 3 contain communities titled during 1990–2000, 2001–2004
and 2005 and 2010 respectively.

Finally, xi represents a vector of characteristics that affects pixel (Andam et al., 2008; Pfaff, 1995; Robalino et al., 2015). We also
deforestation and that might be associated with treatment. The allow for the effects of explanatory variables to vary by depart-
vector of characteristics includes variables such as community ment. So, that Is is an indicator variable for department s and bs
area, slope, rainfall, distance to small rivers, distance to navigable is the vector of coefficients that capture the effects of the charac-
rivers, distance to main roads and distance to non-forest area teristics in vector x in department s. This is relevant as deforesta-
(see Table 1, pixel covariates), where all the variables do not vary tion patterns and rates vary among these departments (see maps
over time except distance to non-forest areas. There is ample evi- by departments in the appendix), as well as the presence of
dence that these variables are highly correlated with deforestation environmental authorities and other regional characteristics as
8 M.A. Vélez et al. / World Development 128 (2020) 104837

discussed in the next section. The effect of one explanatory vari- 4. Results
able might be very important in one department and not in
another. By controlling for these variables, we are able to avoid We first present the results from the qualitative analysis based
any potential bias that might have been caused due to pixel char- on the interviews conducted with community leaders and regional
acteristics that might be correlated with deforestation and titling experts. We discussed the formation of community councils, the
within departments. drivers of deforestation, and the conservation efforts in each region
Moreover, to avoid biasing the effects of titling by including including the role of regional environmental authorities. These
variables that might be affected by titling and affect deforestation analyses offer a cohesive story for the understanding of the econo-
(post treatment controls), we use pixel variables measured before metric specification that follows, on the estimated effects of titling
1990 or pixel variables fixed over time only. For instance, if titling on deforestation. Readers interested in skipping to the quantitative
affects infrastructure investment, and then infrastructure invest- analysis can go to Section 4.4.
ment affects deforestation rates, when controlling for changes in
infrastructure, the estimated effects will ignore the effects of titling
4.1. Formation of Community Councils
through infrastructure. As robustness test, however, we include
two variables that might present this problem to test whether
The key governance feature from Law 70 was the requirement
results change: area used for coca crops and violence variables
to form a Community Council (CC) to gain access to the collective
measured in 2002 and 2000 respectively.
title and act as the new political and territorial management orga-
Additionally, one challenge we have is that the first period is
nization. A General Assembly constituted by all inhabitants of the
longer than periods 2 and 3. This is due to the lack of data available
collective territory elects the board of the CC for a three-year term.
for 1995. To some extent, the ‘period dummies’ control for the fact
There is variation in the mechanisms of participation within this
that the first period is longer. We are using only pre-treatment
structure. However, all key decisions need to be deliberated in con-
observations for that period (see Fig. 3). Thus, as a robustness test,
sultation with the General Assembly and the representatives from
we also use Analysis of Covariance (ANCOVA) following Peña et al.
villages within the CC (Velez, 2011).
(2017). For this model, we only use periods 2 and 3 when the peri-
Different logics regarding the territory followed the creation of
ods cover the same time span. The empirical specification (Eq. 2) is
CCs. In the Chocó sub-region, these Councils (Consejos Mayores)
the following:
were formed mostly out of river and watershed interactions, merg-
ing settlements along rivers (See Peña et al., 2017 for more details).
X
S
Y it ¼ sdit þ dt þ bs Is xi þ qyt1;c þ uit ð2Þ In Cauca, accordingly with the interviews conducted for this study,
s¼1 Community Councils (CCs) were also mainly defined along rivers
and watersheds, but some rivers were divided especially where
where yt1;c is the deforestation rate at community level in the pre- communities were historically separated due to different economic
vious period. activities. In the Valle del Cauca sub-region, mostly around the
In sum, with these specifications, we are able to control for fixed rural area of Buenaventura, two types of CCs were formed: CC by
observable factors and unobservable factors that do not change watershed and CC based on pre-existing settlements or villages.
over time. It is important to acknowledge that we are not able to Organizing by watersheds maintained social, cultural and produc-
control for unobservable factors that vary over time. Our identifica- tive ties along the rivers while organizing by existing villages, used
tion assumption lies in the fact that after controlling for observable the existing community and local political divisions and associa-
factors, the unobservable factors that vary over time are not related tions known as ‘‘Juntas de Acción Comunal” (Community Action
to the timing of treatment and deforestation. We show that results Boards) (See Velez, 2011 for more details).
are highly robust to different specifications, so we believe that In Nariño, based on our interviews, Community Councils were
these factors do not generate relevant biases. Furthermore, we per- created under the umbrella of three groups (e.g. ASOCOESTNAR,
form placebo tests within Group 3 to test whether communities RECONPAS, and COPDICON) that supported different political and
titled in later years (2008, 2009 and 2010) have different trends organizational views in the region. Within each group, each commu-
to those titled in earlier years (2005, 2006 and 2007). This was con- nity was autonomous in their organization to form the Council. As in
ducted in periods 1 and 2, before any of those communities were the rest of the regions, many were defined along rivers or water-
titled (see Fig. 3). No evidence of the presence of statistically signif- sheds, but others were defined by historical productive activities
icant differences between those trends was found (see Section 4.4 or ideological differences. Others determined their organization
for details). based on the attachment to or family ties to one specific territory.
To complement our identification strategy, we use matching
analysis as a robustness test to preselect a comparable sample that 4.2. Deforestation drivers
will be used with our identification strategy. Given that we use
only similar observations, we expect our identification assumption Each sub region or department (Chocó, Valle, Cauca and Nariño)
to be more plausible too. As the characteristics of treated observa- included in this analysis, relates to regional administrative and
tions vary as new communities get titled, we choose to focus on political divisions, which imply, among other things, different
those that are similar to the ones titled in the 1990–2000 period. actors and governing structures for environmental regulation. It
Therefore, before we conduct our empirical strategies, we select also implies different dynamics regarding deforestation drivers.
communities that were titled in the 2000–2005 period that were We include a graphical summary by department in Appendix 2
similar to those titled in the 1990–2000 period. We then select with differences by sub region. However, based on our field work,
communities that were titled in the 2005–2010 period that were and according to the interviews with local leaders, some drivers
similar to those titled also in the 1990–2000 period. This is an esti- are common to the whole region, and this is consistent with the
mate of the average treatment effect on those treated in 1990– information reported in other studies (See for example, Gonzalez
2000. We used neighbor and caliper matching methods to choose et al., 2018; Leal & Restrepo, 2003; Molano, 2017; Suárez et al.,
observations with similar characteristics11. 2018; UNODC, 2017; IIAP, 2008).
Since the colonial period, various subsistence and commercial
11
We applied propensity-score matching (PSM) with caliper 1% and neighbor=1. activities such as agriculture, fishing, hunting, gold mining, and tim-
M.A. Vélez et al. / World Development 128 (2020) 104837 9

ber exploitation have remained the main economic activities in the the technical, financial and administrative capacities . . . There are no
region (Grueso et al., 1998). Private companies, both national and instruments to implement what the regulations say . . . The absence of
foreign, have always been interested in accessing the rich natural instruments is causing great deforestation problems . . . the lack of
resources of the region, including forest and mining resources. Over instruments, that gap, is the cause for the arrival of all those actors that
the years, deforestation has been mostly driven by timber exploita- drive deforestation in the territory”.
tion (from mangrove and tropical forest) and gold mining by locals
and outsiders. These activities have evolved since the 17th century
4.3.2. Valle del Cauca
with the arrival of mechanized exploitation technologies. In addi-
Before Law 70, some territories such as Bahía Málaga and San
tion, some commercial plantations such as coconut, plantain, cocoa,
Cipriano, had community action boards that exercised a role of
rice, sugar cane and specially oil palm (mainly in Chocó and Nariño)
environmental control. For example, during the ’70s, a forest
have been identified as the drivers of deforestation in the region.
reserve was created in San Cipriano as the result of joint work
Cattle ranching in the north of Chocó and shrimp farming in Nariño
between the community action board and the State. Still, it wasn’t
have also been identified as important causes of land use change.
until the enactment of Law 70 that the organizations began to
Since the 1990s, but especially after 2000, illegal coca crops
speak loudly about environmental protection. According to some
expanded in the region, mainly in Nariño, Cauca, and Chocó. Over
leaders, ‘‘Since the council was created, people began to see and
the years, peaks in production of timber, gold, and coca have
respect the territory as a source of work and subsistence for the com-
responded to market demand and global commodity prices. Differ-
munities that live there, and so they automatically have to take care of
ent actors have entered the region promoting these productive
it”. Others mentioned that ‘‘the formation of community councils had
activities. However, since the approval of Law 70 in 1993, legal
a huge impact. During this period, the deforestation processes were
enterprises that did not belong to Afro-Colombian communities
greatly reduced. The community was recognized, and reflected in a
had more restrictions to entering the region, and massive forest per-
government authority to control the territory and the use of the
mits were no longer issued to private companies, which according to
resources”. In this context, the leaders argue that it was because
locals ‘‘generated the greatest deforestation in this region” (e.g., Cartón
of Law 70, that the communities were able to expel the different
de Colombia, Maderas del Darién, Triplex Pizano). The same did not
actors that indiscriminately extracted natural resources from the
happen with illegal enterprises, as often, these involve illegal armed
territory. Some leaders also recognized the resistance of some
actors that threaten local government structures. Thus, we expect
community councils against the introduction of illegal crops. One
that titling will have differential effects in places where legal enter-
leader affirmed that ‘‘the fact that we had been organized in councils,
prises played a more important role in deforestation.
allows the community to effect some kind of resistance”; another
affirmed, ‘‘because of our organizational process, the armed groups
4.3. Conservation efforts and role of regional environmental
stopped cutting down more forest; they stayed in the places that
authorities
had already been cut down. It stopped there”.
After Law 70, more actors became involved in the region’s con-
Conservation efforts and the presence of enforcement by regio-
servation efforts. Following the enactment of Law 70, the CVC (Cor-
nal governmental environmental authorities in the Pacific vary
poración Autónoma del Valle del Cauca) funded communities’
across sub regions and hence the need to describe each case below.
territorial management plans and regulated timber exploitation.
It also enforced the national regulation prohibiting mangrove for-
4.3.1. Chocó
est exploitation. In fact, the CVC is recognized as the strongest
During the ’80s, the community of the middle Atrato watershed,
authority in the pacific region with some level of articulation with
supported by the diocese of Quibdó, decided to organize itself, and
community councils. Also, WWF Colombia, other environmental
created the Asociación Campesina Integral del Atrato (ACIA) in a
NGOs, and the PCN supported CCs in specific projects focusing on
reaction against the exploitation permit requested by the Triplex
community organization, environmental management, and sus-
Pizano Company. Following this, the Ministry of Agriculture sus-
tainable productive practices.
pended the expedition of the permit to the company and, in
1988, the government granted 800 thousand hectares to the com-
munity to implement a management plan to protect the forest, set- 4.3.3. Cauca
ting an example in the region and a preamble for Law 70. Before 1990, the environmental authorities in the territory were
After Law 70, Codechocó (Corporación Autónoma Regional del the CVC, and INDERENA. However, the interviewees mentioned that
Chocó), the regional environmental authority, suspended the pri- these governmental organizations did not exercise any environmen-
vate licenses for forest exploitation throughout the territory. How- tal control in the territory. On the other hand, the interviews
ever, Codechocó has been immersed in several scandals regarding reported that community elders already practiced conservation by
the issuance of illegal forest exploitation permits and the presence simple wisdom. One leader argued that ‘‘the black and the indigenous
of systematic illegal mining exploitation without any enforcement. people have ancestrally managed the territory through wisdom. Our
Community Councils in Chocó, supported in many cases by the efforts in development have always been sustainable and included cul-
Claretian organizations, have played an important role in the envi- tural practices, such as letting the territory rest”.
ronmental protection of the territory. A leader mentioned that ‘‘once By 1990, the CRC (Corporación Autónoma Regional del Cauca)
the CC appeared, they did reduce the deforestation process a bit, because became the environmental authority in the region. However, it
they (the community) began to delimit their territory. And in that delim- was not able to fully monitor the territory. It was perceived as
itation of the territory, forest extraction no longer occurs in all the sce- weak and locals did not really know what it did. During the last
narios, there are some reserve zones, some exploitation zones, and other period of study, interviews reported that some CCs began to work
zones for conservation”. However, most of the interviewees argue towards conservation. In this respect, one interviewee affirmed
that collective titling is not enough, and that due to the lack of speci- that ‘‘the CCs have been very important actors, because despite all
fic instruments or programs some communities have been forced to their limitations, they began to generate a community and cultural
engage in economic activities that produce a great impact in terms of process to highlight the importance of the forest. Although there are
deforestation. According to a leader, ‘‘. . . the law told the communities, all these actors and the extractive economy, the communities have
’You can exploit, you are the owners of the land, you are the owners of started to implement some interesting conservation processes”. How-
the forest’. But it only made them believe that it was going to provide ever, leaders reported that despite the locals’ efforts to resist, it is a
10 M.A. Vélez et al. / World Development 128 (2020) 104837

difficult task when their lives are in danger. Further, in central each of the periods. The impact is still negative and highly signifi-
Cauca today, there are still many councils without formal titling. cant. In Column 3, we also include the pixel characteristics men-
tioned in Table 1 and the results are still negative and significant.
4.3.4. Nariño We then control for the presence of illegal crops (coca) and violence
During the ’90s, Corponariño (Corporación Autónoma Regional and the results are still negative and significant. Finally, we include
de Nariño) became the environmental authority in the region. Municipality fixed effects (Column 5) and community council fixed
According to the interviewees, although the environmental institu- effects (Column 6) and results are still negative and significant.
tion attempted to implement training and small conservation pro- When we control for community fixed effects, variables that do
jects, some leaders affirm that it is locally known for issuing forest not vary in time and across communities are not included in the
permits without the support of the Community Councils. A leader analysis since communities fixed effects control at a more
mentioned, ‘‘In our concept, it is a total disaster. Most of its officials disaggregated level. All estimates are negative and above one per-
are more concerned about the money they receive from forest permits centage point in reduction in deforestation rates.
than about conserving forests . . . there has not been a harmonious In Table 3, we show the estimated effects of titling using differ-
joining of the CCs and Corponariño”. ent methodological approaches. In Column 1, for comparisons pur-
However, Community Councils recognized that with Law 70, poses, we include the estimates of the difference in difference
conservation began to gain importance in the territory. A leader approach with community fixed effects and all the control vari-
mentioned, ‘‘companies no longer entered like they used to, like the ables (see Table 2 column 6). Following Peña et al. (2017), we also
own the place, unless they had an alliance with the council”. Yet, include ANCOVA estimates (Table 3, column 2), which show a
others mentioned that due to the Community Council’s delays in higher impact of titling reducing deforestation. Finally, we use
titling, some companies accelerated their exploitation processes matching analysis as a pre-analysis sample selection strategy to
while they waited for formalization. make sure that observations were similar to those that were trea-
In 2000, management plans were implemented in several CCs. ted in the first period of titling as discussed in the methodology
According to some leaders, ‘‘with the creation of the CCs, we began section. With the matched sample, we then run the difference in
to talk about zones of conservation within the same territories . . . then
we preserved more . . . and with the internal regulations, we started
Table 3
talking about specific measures needed to cut a forest, we started to Estimates of the Titling Effects on Deforestation using various empirical strategies.
become more aware”. Others mentioned that ‘‘the councils raise an
awareness among the community, which would otherwise not exist”. DID ANCOVA DID with Matching
1 2 3
In 2008, several agricultural projects were implemented in the
area to try to substitute illicit crops with limited success. In the Effect 0.0123** 0.0141*** 0.0141**
Standard errors [0.005] - [0.003] [0.007]
words of the leaders, ‘‘the CCs have protected the territory, although
# obs. 108,224 76,866 58,781
not as we would have liked . . . We know that there are still many legal # treated obs. 69,270 69,270 37,256
representatives who do not protect the territory, sometimes we allow # untreated obs. 38,954 7,596 21,525
projects and foreigners to enter. Coca permeated the region because Controls
many made it easy for foreigners to cultivate large extensions”. Period Fixed effects+ Yes Yes Yes
Forest cover % per Yes Yes Yes
community+++
4.4. Estimation of the overall and department titling effects Pixel characteristics+++ Yes Yes Yes
Community Fixed effects Yes No Yes
In Table 2, we present the effects of titling on deforestation rates. Lagged deforestation at No Yes No
community level
In Column 1, we use a simple specification to capture titling effects Sample All All Matched
while controlling for period fixed effects and fixed effects of the
group of communities according to the period that they received *** p < 0.01, ** p < 0.05, * p < 0.1 +++ We allowed variables to have different effects
by department by making them interact with department dummies. DID is differ-
the title by department (treatment-status by department fixed
ence in difference. ANCOVA is the analysis of covariance. DID with matching uses
effects). According to this point estimate, receiving a collective title the DID analysis with a matched sample. The matched sample was selected using
reduces the rate of deforestation by more than 1 percentage point. only the first period (before treatment) with propensity score with the closest
We then control the percentage of forest cover per community in neighbor and a caliper of 0.01.

Table 2
Estimates of the Titling Effects on Deforestation using difference in difference by various specifications.

1 2 3 4 5 6
Effect 0.0108** 0.0183*** 0.0129*** 0.0128*** 0.0109** 0.0123**
Standard errors [0.004] [0.005] [0.005] [0.005] [0.005] [0.005]
# obs. 108,816 108,816 108,224 108,224 108,224 108,224
# treated obs. 69,604 69,604 69,270 69,270 69,270 69,270
# untreated obs. 39,212 39,212 38,954 38,954 38,954 38,954
Controls
Period Fixed effects Yes Yes Yes Yes Yes Yes
Treatment-status by department fixed effects+ Yes Yes Yes Yes Yes No
Forest cover % per community+++ No Yes Yes Yes Yes Yes
Pixel characteristics+++ No No Yes Yes Yes Yes
Presence of illegal crops (coca)+++ and violence No No No Yes No No
Municipality Fixed effects No No No No Yes No
Community Fixed effects No No No No No Yes
Sample All All All All All All

*** p < 0.01, ** p < 0.05, * p < 0.1 and +communities are divided into three groups by department, those that were titled in 1990–2000, those that were titled in 2000–2005, and
those that were titled in 2005–2010 +++ We allowed variables to affect differently by department by interacting with department dummies.
M.A. Vélez et al. / World Development 128 (2020) 104837 11

Table 4
Titling Effects by departments.

(1) (2) (3) (4) (5)


All Nariño Valle del Cauca Cauca Chocó
a
Treatment status by department Fixed Effects 0.0129*** 0.031*** 0.015* 0.004 0.009
[0.005] [0.010] [0.008] [0.008] [0.009]
b
Communities Fixed Effect 0.0123** 0.021* 0.021** 0.008 0.000
[0.005] [0.012] [0.010] [0.008] [0.009]
b
Communities Fixed Effect 0.0141** 0.026* 0.029** 0.010 0.001
Matched sample [0.005] [0.012] [0.010] [0.008] [0.009]
# obs. 108,224 23,732 15,484 16,736 52,272
# treated obs. 69,270 14,937 7,789 10,302 36,242
# untreated obs. 38,954 8,795 7,695 6,434 16,030

Standard errors in brackets. *** p < 0.01, ** p < 0.05, * p < 0.1 a: The specification is the same as (5) in Table 2. b: The specification is the same as (6) in Table 2.

difference strategy and show the results in Table 3 Column 3. We 4.5. Placebo tests
find that the effects lead to an approximately 1.4 percentage point
reduction in deforestation. Again, all estimates are negative and The validity of our identification strategy, as we explained in
above one percentage point in reduction in deforestation rates. Section 3.3, depends on the fact that observations with the same
For those observations titled between 1990 and 2000, deforesta- treatment status have similar deforestation trends. If, for example,
tion was 2.55 percentage points (see Table 1 Column A) and had they communities that receive treatment early on present a declining
not been treated, deforestation would have been of around 3.78 per- deforestation trend compared to of that communities that receive
centage points if we add up the effect found in Specification 6 in treatment later on, one could expect that the treatment effect
Table 1. This represents a 32% reduction on the deforestation rate. would capture this even in the absence of a titling effect. Therefore,
For those observations titled between 2000 and 2005 with a 3.27% we test whether this is the case within communities where treat-
rate of deforestation, the impact on the 2005–2010 period repre- ment status does not change for two periods. Within Group 1, we
sents a 27.3% reduction in the deforestation rate. test whether communities titled in later years (1999 and 2000)
In Table 4, we present the same analysis performed in Col- have different trends than those titled in earlier years (1996,
umns 3 (Groups Fixed Effects), and 6 (Communities Fixed 1997 and 1998). This was done in periods 2 and 3, after all of those
effects) in Table 2 and Column 3 (community fixed effects with communities were titled (see Fig. 3). There is no evidence of the
matched sample) in Table 3. We find that the reduction in defor- presence of statistically significant differences between those
estation due to titling is mainly present in Nariño and Valle del trends (see Table 5). Also, for Group 3, we test whether communi-
Cauca, the southern part of the country’s Pacific coast. Estimates ties titled in later years (2008, 2009 and 2010) have different
show a reduction of more than 3 percentage points in deforesta- trends than those titled in earlier years (2005, 2006 and 2007). This
tion rates in Nariño, which represents a 37% reduction in defor- was done in periods 1 and 2, before any of those communities were
estation rate (deforestation in treated observations in Nariño was titled (see Fig. 3). Again, we find no evidence of the presence of sta-
5.2 percentage points). While for Valle del Cauca, the estimated tistically significant differences between those trends (see Table 5).
reductions are above 1.5 percentage points, which represent a
55% reduction (deforestation in treated observations in Valle 5. Discussion
del Cauca was 5.2 percentage points). For Cauca and Chocó, we
do not find evidence that collective titling affects deforestation ‘‘The community council began to act and say: We are going to
rates. However, we should also mention that for these states, defend; we are going to get organized. Then, there was a change
balances of covariates are off. On average, there are no similar because not everyone could enter the territory as if they owned
Community Councils either within or outside those departments the place”.
that generate good matches that can allow us to conclude
whether titling had any type of effect. This implies that the iden- The issue of institutional design for assigning property rights to
tification of the effects relies heavily on the model specification. a group instead of individuals continues to attract attention with
However, as previously shown in Tables 2 and 3 for the general important policy implications for land conservation programs in
results, and in Table 4 for Chocó and Cauca, estimates are highly the developing world (Agrawal & Ostrom, 2001). The most classical
robust to different specifications. view against collective titling argues that as the costs and respon-
In conclusion, we do not find any statistically significant sibilities are spread around a group, the incentives for free-riders in
evidence that titling collective lands increases deforestation. We terms of monitoring, sanctioning, using and misusing natural
find that, on average, this policy has decreased deforestation, but resources can cause the so-called ‘‘tragedy of the commons”
these results are driven mainly by the departments of Nariño and (Hardin, 1968). Even if there are well-organized communities with
Valle del Cauca, in the southern half of the Pacific coast12. strong self-governing structures, they may prefer short-term gains
from extractive activities and bear the cost of future profits from
forest conservation, in some cases, because of their immediate
needs, given the levels of poverty in many of these regions.
12 Among the arguments in favor, however, is the possibility of
For Nariño and Valle del Cauca, we also explored heterogeneous effects. We split
the sample and estimate the effects far from and close to non-forest areas, far from aligning monitoring and control with the possibilities of self-
and close to roads, far from and close to navigable rivers and finally with and without governed institutions in the communities, which have better infor-
illegal crops (coca).We find some evidence that negative effects of titling on mation about the use and misuse of ecosystems. Another argument
deforestation are stronger in areas with high deforestation threats (close to the is that collective titling can make use of economies of scale as
forest frontier, to roads, and to navigable rivers) as well as in places without illegal
crops. However, these effects are not robust when we use continuous variable and
opposed to assigning private property, which is highly costly in
allow for nonlinear effects (following the specification in Sims, 2010). Further terms of transaction costs with a much larger number of individual
research should be pursued. owners. Furthermore, within an institutional framework that has
12 M.A. Vélez et al. / World Development 128 (2020) 104837

Table 5
Placebo tests: Tests of different trends in deforestation without changes in treatment status.

Observations Group 1 Group 1 Group 3 Group 3


Communities treated during 1990–2000 1990–2000 2005–2010 2005–2010
Periods included 2000–2005 and 2005–2010 2000–2005 and 2005–2010 1990–2000 and 2000–2005 1990–2000 and 2000–2005
Difference in the trend (placebo effect) 0.006 0.004 0.001 0.000
Standard errors 0.004 0.004 0.027 0.026
Controls
Period Fixed effects+ Yes Yes Yes Yes
Communities treated during 1999–2000 Yes Yes – –
Communities treated during 2008–2010 – – Yes Yes
Pixel characteristics+++ No Yes No Yes

increasingly protected the rights of indigenous and ancestral enterprises was not effective as it was in Valle and Nariño. According
groups, which is the particular case of Colombia, and growing to the interviews with local leaders, in Chocó legal companies con-
international attention to issues related to protecting the rights tinue exploiting resources without the forest permits issued by the
of these ethnic communities, it may become harder for economic government. They kept operating illegally in the territory, by buying
forces favoring extractive activities to enter into these territories land from communities, providing machinery and technical advice to
without being noticed. locals and also buying timber exploited by locals. For example, a lea-
We find that collective titling to Afro-Colombian Communities der from Chocó mentioned, ‘‘What happened is that the strategy chan-
has reduced deforestation in the Pacific by more than a one per- ged. The company . . .initially requested the permits and exploited
centage point. These effects represent more than 27% reductions directly. Then they changed the figure, the community councils exploited
of what would have been the deforestation rates without titling. the resource and the company bought it. But the company lent the com-
Back of the envelop calculations suggest that titling avoided defor- munity all the machinery and the equipment needed for the exploitation
estation in around 60 thousand hectares in the area under analy- and paid them for the timber. They changed the strategy to avoid facing
sis13. However, when conducting the analysis by region, we only the exploitation process”. Similar strategies were reported in Cauca.
find evidence that this effect is statistically significant for Nariño Despite these heterogeneous results at the sub regional level,
and Valle del Cauca: two regions with different deforestation collective titling was a first important step in the protection of
dynamics and rates. Nariño had the highest deforestation rates in the territory, but for greater effectiveness, it must be comple-
both titled and non-titled communities in the whole region. The mented with other processes, including clarity in the distribution
rates for Valle were among the lowest. of conservation costs. Currently, mechanisms such as payment
Based on the findings from the interviews conducted for this for environmental services PES or REDD+ could be part of the
study, our hypothesis is that the reduction of deforestation in titled answers to these demands, but the development of sustainable
communities is explained mainly by two forces: i) local productive enterprises in the region that improve quality of life
community-based organization that defines the rules for commu- in the communities, without putting pressure on natural resources,
nity use of natural resources mainly in Valle, and ii) the cancelation is still necessary. Although some regional authorities, such as the
of forest permits and thus the expulsion of private companies ded- CVC in Valle, have supported new local authorities, in general,
icated to timber exploitation in Valle, and mainly oil palm in the expected synergies with other levels of environmental author-
Nariño. In the absence of collective titling and community organi- ities have not occurred. For example, most of the timber exploita-
zation triggered by this collective titling process, the same trend of tion in the region continues to be illegal. Titling reform was not
resource extraction, and agricultural expansion from previous dec- accompanied in general by public funding to finance the monitor-
ades would have continued14. Also, in Valle, it is important to recog- ing needs of local authorities and the implementation of local man-
nize, that the regional environmental authority, CVC, had more agement plans.
coordinated efforts and synergies with ethnic authorities by for Methodologically, using a combination of qualitative and quan-
example financing community councils’ natural resources manage- titative methodologies allowed us to improve the estimates and
ment plans. interpretation of our results. Previous studies do not consider the
The definition of new rules to manage natural resources and the fact that deforestation processes vary across sub-regions and this
reduction of legal companies in Valle and Nariño seem to be strong could play a role in the estimated effects of titling. This insight
enough in titled communities that reduction of deforestation in the was salient within the qualitative interviews.
area is significant and probably greater than the incursion from Further studies should go beyond forest cover as an environ-
illegal activities15. The same is not occurring in Chocó and Cauca mental indicator to study the process of forest degradation,
where titling did not reduce deforestation. Even though in Chocó, water pollution and biodiversity impacts, which could be
leaders identified new rules to manage the territory and a strength affected by mostly illegal mining activities in the region and
in community organization, the expulsion of private extractive for whom forest cover is not the most appropriate indicator.
The role of local rules and social norms in managing natural
resources after the titling process should also be studied to gain
13
To estimate this, we assume that the deforestation rate would have been 4 more in depth knowledge of the micro institutional mechanisms
percentage points for a five year period (lower than any of the titled deforestation behind the effectiveness of collective titling. New research
rates in Table 1 plus 1 percentage point impact), and consider for each council the
should also explore the impact of collective titling on hindering
forest area and the years that were titled. After estimating this counterfactual
deforestation per community, we assume that titling saved 27% (the lowest estimated or allowing coca crops and illegal activities such as gold mining.
impact we find). New instruments to control illegal markets should be developed
14
Note that in some parts of Nariño, a plague also affected oil palm plantations since the definition of property rights in a context of a weak
according to some leaders. state without the monopoly of violence is insufficient to protect
15
This is a very intriguing result for Nariño given the increase of illicit crops in this
the forest. As one leader from Nariño mentioned, ‘‘the Councils
region. Suggesting that in the absence of collective titling, the spread of coca crops
would probably be even higher. For an evolution of coca crops in the region see have struggled here, but the intimidation has not allowed the
UNODC (2017). impact to be successful. . . .. people have died here for standing up
M.A. Vélez et al. / World Development 128 (2020) 104837 13

against certain dynamics: coca, oil palm . . . the councils have Quintero and specially to Juliana Unda who conducted the inter-
resisted. . ., but it is not enough”. views in the field. For sharing all their thoughts and ideas to inter-
The titling reform has not been able to fully deter the pressure pret the forest cover maps developed in the project, we are
from illegal economies, coca crops and gold mining, also present in indebted to Mauricio Cabrera and Carmen Candelo at WWF-
the region and exacerbated by the violent conflict associated with Colombia, and Richard Moreno, Regional Attorney for Choco. In
emerging gangs and drug trafficking. In general, these extractive the regions, for all their support to conduct the interviews, we
activities come together with illegal armed actors infiltrating com- thank Alvaro Arroyo and Santiago Valencia in Buenaventura,
munities through violence and disrupting the local organizations’ Onécimo Gonzales in Tumaco, Victor Candelo in El Charco, and Edi-
capacity to enforce their rules and norms. son Duque in Guapi. We are grateful to all the leaders interviewed
with whom we discussed the results of this article. We assume
Acknowledgments complete responsibility for the final contents.

This work was possible thanks to the financial support from


Appendix 1. Maps of the 4 departments
Environment for Development (EfD), grant MS- 383. We acknowl-
edge the excellent research assistance of Pamela Sanabria, Daniela
Map A1–A4.

Map A1. Chocó Department.


14 M.A. Vélez et al. / World Development 128 (2020) 104837

Map A2. Valle del Cauca Department.


M.A. Vélez et al. / World Development 128 (2020) 104837 15

Map A3. Cauca Department.

Map A4. Nariño Department.


16 M.A. Vélez et al. / World Development 128 (2020) 104837

Fig. A5. Graphical Summary for Chocó.


M.A. Vélez et al. / World Development 128 (2020) 104837 17

Fig. A6. Graphical Summary for Valle del Cauca.


18 M.A. Vélez et al. / World Development 128 (2020) 104837

Fig. A7. Graphical Summary for Cauca.


M.A. Vélez et al. / World Development 128 (2020) 104837 19

Fig. A8. Graphical Summary for Nariño.

Appendix 2. Drivers of deforestation by sub-region References

The arrows in the graphs are related to the activity identified as Agrawal, A. (2007). Forests, governance, and sustainability: Common property
theory and its contributions. International Journal of the Commons, 1(1),
a driver of deforestation and represent:" Increase, ; decrease, ? 111–136. https://doi.org/10.18352/ijc.10.
Continue. Agrawal, A., & Ostrom, E. (2001). Collective action, property rights and
decentralization in resource use in India and Nepal. Politics and Society, 29(4),
485–514. https://doi.org/10.1177/0032329201029004002.
20 M.A. Vélez et al. / World Development 128 (2020) 104837

Alix-Garcia, J. M. (2007). A spatial analysis of common property deforestation. and Escobar, A. (Eds.), Cultures of Politics/Politics of Cultures. Re-visioning Latin
Journal of Environmental Economics and Management, 53(2), 141–157. https:// American Social Movements (pp.196–219). Westview: Boulder.
doi.org/10.1016/j.jeem.2006.09.004. Hardin, G. (1968). The tragedy of the commons. Science, 162(3859), 1243–1248.
Alix-Garcia, J. M., Shapiro, E. N., & Sims, K. R. (2012). Forest conservation and https://doi.org/10.1080/19390450903037302.
slippage: Evidence from Mexico’s national payments for ecosystem Holland, M. B., De Koning, F., Morales, M., Naughton-Treves, L., Robinson, B. E., &
services program. Land Economics, 88(4), 613–638. https://doi.org/10.3368/ Suárez, L. (2014). Complex tenure and deforestation: Implications for
le.88.4.613. conservation incentives in the Ecuadorian Amazon. World Development, 55,
Andam, K. S., Ferraro, P. J., Pfaff, A., Sanchez-Azofeifa, G. A., & Robalino, J. A. (2008). 21–36. https://doi.org/10.1016/j.worlddev.2013.01.012.
Measuring the effectiveness of protected area networks in reducing Holland, M. B., Jones, K. W., Naughton-Treves, L., Freire, J. L., Morales, M., & Suárez, L.
deforestation. Proceedings of the National Academy of Sciences, 105(42), (2017). Titling land to conserve forests: The case of Cuyabeno Reserve in
16089–16094. https://doi.org/10.1073/pnas.0800437105. Ecuador. Global Environmental Change, 44, 27–38.
Armenteras, D., Cabrera, E., Rodríguez, N., & Retana, J. (2013). National and regional Idrobo, N., Mejia, D., & Tribin, A. (2014). Illegal gold mining and violence in
determinants of tropical deforestation in Colombia. Regional Environmental Colombia. Peace Economics, Peace Science and Public Policy, 20, 83–111.
Change, 13(6), 1181–1193. https://doi.org/10.1007/s10113-013-0433-7. IIAP. (2008). Evaluación interdimensional de los daños ambientales ocasionados por
Arriagada, R. A., Ferraro, P. J., Sills, E. O., Pattanayak, S. K., & Cordero-Sancho, S. el cultivo de la palma aceitera y la ganadería extensiva en los municipios de
(2012). Do payments for environmental services affect forest cover? A farm- Carmen del Darién, Riosucio (Chocó) y Mutatá (Antioquia). Quibdó: Instituto de
level evaluation from Costa Rica. Land Economics, 88(2), 382–399. https://doi. Investigaciones Ambientales del Pacífico -IIAP-.
org/10.3368/le.88.2.382. James, A., Gaston, K. J., & Balmford, A. (1999). Balancing the earth’s accounts. Nature,
Baland, J., & Platteau, J. (1996). Halting degradation of natural resources: Is there a role 401, 323–324.
for rural communities?. New York: Oxford University Press. James, A., Gaston, K. J., & Balmford, A. (2001). Can we afford to conserve
Baland, J. M., & Platteau, J. P. (2003). Economics of common property management biodiversity? BioScience, 51(1), 43–52. https://doi.org/10.1641/0006-3568
regimes. Handbook of environmental economics (Vol. 1, pp. 127–190). Elsevier. (2001) 051[0043:CWATCB]2.0.CO;2.
Barsimantov, J., & Kendall, J. (2012). Community forestry, common property, and Joppa, L., & Pfaff, A. (2010). Reassessing the forest impacts of protection: The
deforestation in eight Mexican States. Journal of Environment and Development, challenge of nonrandom location and a corrective method. Annals of the New
21(4), 414–437. https://doi.org/10.1177/1070496512447249. York Academy of Sciences, 1185, 135–149. https://doi.org/10.1111/j.1749-
Blackman, A., Corral, L., Lima, E. S., & Asner, G. P. (2017). Titling indigenous 6632.2009.05162.x.
communities protects forests in the Peruvian Amazon. Proceedings of the Lambin, E. F., Turner, B. L., Geist, H. J., Agbola, S. B., Angelsen, A., Bruce, J. W., et al.
National Academy of Sciences, 114(16), 4123–4128. https://doi.org/10.1073/ (2001). The causes of land-use and land-cover change: Moving beyond the
pnas.1603290114. myths. Global Environmental Change, 11(4), 261–269.
Blackman, A., & Veit, P. (2018). Amazon indigenous communities cut forest carbon Larson, A. M. (2011). Forest tenure reform in the age of climate change: Lessons for
emissions. Ecological Economics, 153, 56–67. REDD+. Global Environmental Change, 21(2), 540–549. https://doi.org/10.1016/
Bonilla, L., & Higuera, I. (2019). Protected areas under weak institutions: Evidence j.gloenvcha.2010.11.008.
from Colombia. World Development, 122, 585–596. Leal, C., & Restrepo, E. (2003). Unos bosques sembrados de aserríos: Historia de la
Buntaine, M. T., Hamilton, S. E., & Millones, M. (2015). Titling community land to extracción maderera en el Pacífico colombiano. Medellin: Editorial Universidad de
prevent deforestation: An evaluation of a best-case program in Morona- Antioquia.
Santiago, Ecuador. Global Environmental Change, 33, 32–43. https://doi.org/ Liscow, Z. D. (2013). Do property rights promote investment but cause
10.1016/j.gloenvcha.2015.04.001. deforestation? Quasi-experimental evidence from Nicaragua. Journal of
Cabrera, E., Galindo, G., Vargas, D.M. (2011). Protocolo de Procesamiento Digital de Environmental Economics and Management, 65(2), 241–261. https://doi.org/
Imágenes para la Cuantificación de la Deforestación en Colombia, Nivel 10.1016/j.jeem.2012.07.001.
Nacional Escala Gruesa y Fina. Bogotá D.C. (Colombia): Instituto de Martinez, T. (2018). Cultivos de coca y violencia: El cambio después de los diálogos
Hidrología, Meteorología, y Estudios Ambientales- IDEAM-, 44. de paz. Tesis de Maestría en Economía. Facultad de Economía. Universidad de
Cárdenas, J. C., & Ostrom, E. (2004). What do people bring into the game? los Andes.
Experiments in the field about cooperation in the commons. Agricultural McGinnis, M. D. (2011). An introduction to IAD and the language of the Ostrom
Systems, 82(3), 307–326. workshop: A simple guide to a complex framework. Policy Studies Journal, 39(1),
Cronkleton, P., & Larson, A. (2015). Formalization and collective appropriation of 169–183. https://doi.org/10.1111/j.1541-0072.2010.00401.
space on forest frontiers: Comparing communal and individual property Ministerio de Cultura. (2010). Afrocolombianos, población con huellas de africanía.
systems in the Peruvian and Ecuadoran Amazon. Society and Natural Retrieved from https://www.mincultura.gov.co/areas/
Resources, 28(5), 496–512. https://doi.org/10.1080/08941920.2015.1014609. poblaciones/comunidades-negras-afrocolombianas-raizales-y-palenqueras/
DANE. (2005). Census of Population, 2005: Estimation and projection of national, Documents/Caracterizaci%C3%B3n%20comunidades%20negras%20y%
department and total municipal population by area 1985-2020. Retrieved from 20afrocolombianas.pdf.
https://www.dane.gov.co/index.php/estadisticas-por-tema/demografia-y- Molano, A. (2017). De río en río. Aguilar: Vistazo a los territorios negros.
poblacion/proyecciones-de-poblacion. Ng’weno, B. (2001). On Titling Collective Property, Participation, and Natural
De Janvry, A., Sadoulet, E. (2001). Access to land and land policy reforms. Policy brief Resource Management: Implementing Indigenous and Afro-Colombian
no. 3, UNU/WIDER. Demands. A Review of Bank Experience in Colombia. Washington DC: World
Ellis, E. A., & Porter-Bolland, L. (2008). Is community-based forest management Bank.
more effective than protected areas?: A comparison of land use/land cover Offen, K. (2003). The territorial turn: Making black territories in Pacific Colombia.
change in two neighboring study areas of the Central Yucatan Peninsula, Journal of Latin American Geography, 2(1), 43–72. https://doi.org/10.1353/
Mexico. Forest Ecology and Management, 256(11), 1971–1983. https://doi.org/ lag.2004.0010.
10.1016/j.foreco.2008.07.036. Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective
Escobar, A., Pedrosa, A., Cárdenas, M., and Correa, H. (1996). Pacífico: ¿Desarrollo o Action. Cambridge: Cambridge University Press.
diversidad? Estado, Capital y Movimientos Sociales en el Pacífico Colombiano. Ostrom, E. (2005). Understanding institutional diversity. Princeton University Press,
Cerec. Bogotá, Colombia. Princeton, New Jersey, USA.
Ferraro, P. J., Hanauer, M. M., & Sims, K. R. (2011). Conditions associated with Ostrom, E., & Nagendra, H. (2006). Insights on linking forests, trees, and people from
protected area success in conservation and poverty reduction. Proceedings of the the air, on the ground, and in the laboratory. Proceedings of the National Academy
National Academy of Sciences, 108(34), 13913–13918. https://doi.org/10.1073/ of Sciences, 103(51), 19224–19231. https://doi.org/10.1073/pnas.0607962103.
pnas.1011529108. Pearce, D. (2005). Paradoxes in biodiversity conservation. World Economics, 6(3),
Freire, G., Diaz-Bonilla, C., Schwartz Orellana, S., Soler Lopez, J., & Carbonari, F. 57–69.
(2018). Afrodescendientes en Latinoamérica: Hacia un marco de inclusión Peña, X., Vélez, M. A., Cárdenas, J. C., Perdomo, N., & Matajira, C. (2017). Collective
Retrieved from. Washington DC: World Bank https:// property leads to household investments: Lessons from Land Titling in Afro-
openknowledge.worldbank.org/handle/10986/30201. Colombian Communities. World Development, 97, 27–48. https://doi.org/
Galvis-Aponte, L. A., Moyano, L. M., & Alba-Fajardo, C. A. (2016). La persistencia de la 10.1016/j.worlddev.2017.03.025.
pobreza en el Pacífico colombiano y sus factores asociados. Documentos de Pfaff, A. S. P. (1995). The economics of deforestation: Evidence from the Brazilian
Trabajo Sobre Economía Regional y Urbana; No. 238. Amazon and New England Doctoral dissertation. Massachusetts: Institute of
Geist, H. J., Lambin, E. F. (2001). What drives tropical deforestation. LUCC Report Technology.
series, 4, 116. Pfaff, A., Robalino, J., Herrera, D., & Sandoval, C. (2015). Protected areas’ impacts on
Geist, H. J., & Lambin, E. F. (2002). Proximate causes and underlying driving forces of Brazilian Amazon Deforestation: Examining conservation-development
tropical deforestation: Tropical forests are disappearing as the result of many interactions to inform planning. PLoS One, 10(7). https://doi.org/10.1371/
pressures, both local and regional, acting in various combinations in different journal.pone.0129460.
geographical locations. BioScience, 52(2), 143–150. Pfaff, A., Robalino, J., Lima, E., Sandoval, C., & Herrera, L. D. (2013). Governance,
Gonzalez, J., Cubillos, A. Chadid, M., Cubillos, A., Arias, M., Zuñiga, E., Joubert, F., Location and Avoided Deforestation from Protected Areas: Greater restrictions
Pérez, I., & Berrío, V. (2018) Caracterización de las principales causas y agentes can have a lower impact due to differences in location. World Development, 55,
de la deforestación a nivel nacional período 2005-2015. Bogota: Instituto de 7–20. https://doi.org/10.1016/j.worlddev.2013.01.011.
Hidrología, Meteorología y Estudios Ambientales -IDEAM-, Ministerio de Pfaff, A., Robalino, J., Sanchez-Azofeifa, G. A., Andam, K., & Ferraro, P. (2009). Park
Ambiente y Desarrollo Sostenible, Programa ONU-REDD Colombia. location affects forest protection: Land characteristics cause differences in park
Grueso, L., Rosero, C., & Escobar, A. (1998). The process of black community impacts across Costa Rica. The B.E. Journal of Economic Analysis Policy, 9(2).
organizing in the Southern Pacific Coast of Colombia. In Alvarez, S., Dagnino, E., https://doi.org/10.2202/1935-1682.1990.
M.A. Vélez et al. / World Development 128 (2020) 104837 21

Plant, R., & Hvalkof, S. (2001). Land titling and indigenous peoples. Washington DC: Sims, K. (2010). Conservation and development: Evidence from Thai protected
InterAmerican Development Bank. areas. Journal of Environmental Economics and Management, 60(2), 94–114.
Porter-Bolland, L., Ellis, E. A., Guariguata, M. R., Ruiz-Mallén, I., Negrete-Yankelevich, https://doi.org/10.1016/j.jeem.2010.05.003.
S., & Reyes-García, V. (2012). Community managed forests and forest protected Suárez, J.A, González, J.J, Cubillos, A., Chadid, M.A., Cubillos, A., Arias, M., Zúñiga, E.,
areas: An assessment of their conservation effectiveness across the tropics. Joubert, F., Pérez, I., Berrío, V., Yepes, A., Ome, E., Garcia, P. & Leguía, D. (2018).
Forest Ecology and Management, 268, 6–17. https://doi.org/10.1016/ Causas y agentes de la deforestación en Territorios de comunidades negras del
j.foreco.2011.05.034. pacífico colombiano. En García et al. (Eds). Perspectiva del pueblo negro frente a
Robalino, J. A., & Pfaff, A. (2012). Contagious development: Neighbor interactions in la deforestación y degradación del territorio: un insumo para la construcción e
deforestation. Journal of Development Economics, 97(2), 427–436. https://doi. implementación de Bosques Territorios de Vida- Estrategia Integral de Control a
org/10.1016/j.jdeveco.2011.06.003. la Deforestación y Gestión de los Bosques (pp. 82-113). Bogotá D.C. (Colombia),
Robalino, J., Sandoval, C., Barton, D., Chacon, A., & Pfaff, A. (2015). Evaluating Programa ONO-REDD.
interactions of forest conservation policies on avoided deforestation. PLoS One, Tucker, C. M. (1999). Private versus common property forests: Forest conditions and
10(4). https://doi.org/10.1371/journal.pone.0124910. tenure in a honduran community. Human Ecology, 27(2), 201–230. https://doi.
Robinson, B. E., Holland, M. B., & Naughton-Treves, L. (2014). Does secure land org/10.1023/A:1018721826964.
tenure save forests? A meta-analysis of the relationship between land tenure UNODC (2017). Colombia, monitoreo de territorios afectados por cultivos ilícitos 2016.
and tropical deforestation. Global Environmental Change, 29, 281–293. https:// Bogotá: UNODC.
doi.org/10.1016/j.gloenvcha.2013.05.012. Velez, M. A. (2011). Collective titling and the process of institution building: The
Rodriguez Solorzano, C., & Fleischman, F. (2018). Institutional legacies explain the new common property Regime in the Colombian Pacific. Human Ecology, 39(2),
comparative efficacy of protected areas: Evidence from the Calakmul and Maya 117–129. https://doi.org/10.1007/s10745-011-9375-1.
Biosphere Reserves of Mexico and Guatemala. Global Environmental Change, 50, Vergara-Asenjo, G., & Potvin, C. (2014). Forest protection and tenure status: The key
278–288. role of indigenous peoples and protected areas in Panama. Global Environmental
Romero, M., Saavedra, S. (2018). Communal Property Rights and Deforestation: Change, 28, 205–215. https://doi.org/10.1016/j.gloenvcha.2014.07.002.
Evidence from Colombia. DOI:10.2139/ssrn.3179052. World Bank (2005). The Gap Matters; Poverty and Well-being of Afro-Colombian and
Sánchez, E., Roldan, R. (2002). Titulación de los Territorios Comunales Indigenous Peoples (Report No. 33014-CO). Santiago, Chile. Environmentally and
Afrocolombianos e Indígenas en la Costa Pacífica de Colombia. Banco Socially Sustainable Development Unit of Latin America and the Caribbean Region.
Mundial, No. 21711. WWF (2002). Memorias: Cartografía social para la planificación del uso y manejo de
Sánchez, E., & García, P. (2006). Los Afrocolombianos. In J. Stubbs & H. Reyes (Eds.), la Piangua (Andara tuberculosa) y otros recursos asociados al manglar en la
Más allá de los Promedios: Afrodescendentes en America Latina. Washington, DC: costa pacífica Nariñense. Fondo Mundial para la naturaleza.
International Bank for Reconstruction and Development/World Bank. Zabala, O. A. G. (2017). Constitución y ampliación de resguardos indígenas en
Schlager, E., & Ostrom, E. (1992). Property-rights regimes and natural resources: A Colombia. Una mirada al avance histórico y perspectivas de cumplimiento.
conceptual analysis. Land Economics, 68(3), 249–262. Campos en Ciencias Sociales, 5(1 y 2), 73–113.

You might also like