World Development Vol. xx, pp.

xxx–xxx, 2017
0305-750X/Ó 2017 Elsevier Ltd. All rights reserved.

www.elsevier.com/locate/worlddev
http://dx.doi.org/10.1016/j.worlddev.2017.03.014

Community Forest Management: An Assessment and Explanation

of its Performance Through QCA
BAS ARTS and JESSICA DE KONING*
Wageningen University, The Netherlands
Summary. — Community Forest Management (CFM)—ranging from community-based to co-management regimes—has become an
influential approach in the management of forests around the world in the last couple of decades. In response to some of the adverse
effects of state forestry and commercial timber production, CFM claims to improve local livelihoods and conserve forests. Many inter-
national organizations, donors, NGOs, and governments therefore advocate CFM. However, a vast body of literature reveals that the
overall results are mixed. This paper contributes to this literature in two ways. By building upon neo-institutionalism in CFM studies, the
paper uses a practice-based approach as a theoretical lens to better understand how and why CFM institutions are successful or not. In
addition, the paper applies a Qualitative Comparative Analysis (QCA) methodology to conduct a systematic cross-case comparison,
while allowing for some generalization. By analyzing a decade of CFM research at the Forest and Nature Conservation Policy
(FNP) group from Wageningen University in the Netherlands, this paper compares and synthesizes ten CFM cases from Africa, Asia,
and Latin America. It concludes that: (1) CFM does indeed present mixed results; (2) CFM performs similarly on social and ecological
parameters; (3) overall, community-based organizations are strongly engaged in CFM; (4) such strong engagement though is not suffi-
cient for CFM to perform; and (5) in particular, the presence of a ‘‘Community of Practice” that links local people to external forest
professionals for mutual learning, based on respect and trust, makes a positive difference in terms of livelihoods and forest conditions.
Ó 2017 Elsevier Ltd. All rights reserved.

Key words — Community Forest Management (CFM), practice-based approach (PBA), Qualitative Comparative Analysis (QCA),
Community of Practice (CoP)

1. INTRODUCTION Over the years a vast body of scholarly literature on CFM

In the last couple of decades Community Forest Manage-
ment (CFM) has become an influential approach in the man-
agement of tropical forests around the world (Arnold, 2001;
FAO, 1978; Wiersum, 2009). About 15% of tropical forests fall
under such a management regime today (RRI, 2014). For
example, in the early 1990s India, Nepal, Mexico, Bolivia,
and Tanzania pioneered different forms of CFM, and many
countries, from Indonesia and Ethiopia to Congo Democratic
Republic, followed later (Baynes, Herbohn, Smith, Fisher, &
Bray, 2015; Charnley & Poe, 2007). As a response to colonial
state forestry, commercial tropical timber production, and
coercive conservation, the CFM approach aims to fulfill both
local livelihoods and forest conservation, while building upon
customary traditions and social forestry initiatives (Agrawal,
2001; Dressler et al., 2010; Umans, 1993). In general, CFM
can be defined as the use, management, and conservation of
forests by communities. Communities can have full, partial,
or no ownership of such forests, and their management is
often practiced in various degrees of collaboration with state
forest agencies, donor organizations, knowledge institutions
and/or companies. At one end of the spectrum, forest manage-
ment is fully community-based and the forests are 100%
owned by the community. Whereas at the other communities
just participate in some of the state forest management prac-
tices in public lands. Because of this variation, several termi-
nologies are used to refer to these practices (community
forestry, community-based forest management, community-
managed forests, collaborative forest management, participa-
tory forest management, joint forest management, and forest
co-management). We prefer the acronym CFM because,
compared to other terminologies, it is the one used most in
the literature (based on a Google Scholar search).
1
Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
has emerged, and one intriguing research question has been
the performance of these initiatives. Does CFM deliver its pro-
mises on livelihoods improvement and forest conservation?
And, what factors might explain its successes and failures?
Based on many publications, review papers and research pro-
grams, the current consensus is that—overall—the results of
CFM are mixed (Baynes et al., 2015; Charnley & Poe, 2007).
Many projects have been reported as rather successful, while
others have been considered failures (Persha, Agrawal, &
Chhatre, 2011). Moreover, forests have generally benefitted
more from CFM than people (Bowler et al., 2012), and the rel-
atively well-off have often gained more from these projects
than the poor (Kumar, 2002). In understanding why CFM
might work, scholars have found many factors particularly rel-
evant: (1) Biophysical factors, like micro-climate and land-
scape morphology; (2) Demographic factors, such as
community size and population growth; (3) Economic factors,
like the generation of additional income for communities
through CFM; (4) Institutional factors, such as clear rules
and rights related to forest ownership, access, use and manage-
ment; (5) Socio-political factors, such as the presence or
absence of cultural, political and social capital; and (6)
* The authors would like to thank the scholars who executed the CFM
studies that form the basis of this paper and contributed to the calibration
of the QCA scores in this synthesis: Alemayehu Negassa Ayana, Charlotte
Benneker, Purabi Bose, Josiah Katani, Sailaja Nandigama, Kim Phung
Dang Thi and Yurdi Yasmi. We would also like to thank Innocent Babili
for his comments and suggestions on the draft paper, Tabitha Muriuki for
assisting us in the literature search, Hilde Thoonen for her advice on QCA
methodology, and Alrik Thiem for his advice on the use of R software.
Final revision accepted: March 9, 2017.
2 WORLD DEVELOPMENT
External recognition, like support by governments, donors,
universities, etc. (Agrawal & Chhatre, 2006; Baynes et al.,
2015; Blomley et al., 2008; Charnley & Poe, 2007; IFRI,
2015; Mustalahti & Lund, 2009; Ostrom, 1990; Pagdee,
Kim, & Daugherty, 2006; Poteete & Ostrom, 2004).
The above findings have mainly been produced by either
qualitative case studies (i.e., Mustalahti & Lund, 2009;
Pagdee et al., 2006), or by quantitative data analysis (i.e.,
Agrawal & Chhatre, 2006; IFRI, 2015; Persha et al., 2011).
Although both methodological approaches have been very
helpful in assessing and understanding the performance of
CFM they both have their shortcomings. Comparison of qual-
itative case studies is difficult to make given their context
dependencies, whereas quantitative analyses have to overlook
local complexities. By applying a Qualitative Comparative
Analysis (QCA) this paper employs a third alternative, using
both qualitative data and quantitative logics to allow for both
cross-case comparison and some generalization, as well as to
identify causal pathways through a statistical method that
goes beyond multi-case-study approaches. This paper applies
a QCA to ten case studies in six countries. The case studies
have been selected from a research program at the Forest
and Nature Conservation Policy (FNP) group of Wageningen
University in the Netherlands. Over the last decade, this group
has produced several PhD theses, and research papers on
CFM cases from tropical countries all over the world (Bolivia,
Ecuador, Ethiopia, India, Indonesia, Tanzania and Vietnam).
In understanding a group of cases in depth, QCA attempts to
unravel the relationship between conditions and outcomes
over a range of cases, thus assessing the degree to which speci-
fic configurations best explain the results. We will apply a
fuzzy-set QCA (fsQCA) in this paper, details of which will
be explained in the methodological section.
In addition to the QCA methodological approach this paper
will also put a theoretical lens central stage that differs from
most CFM literature. Whereas the latter particularly builds
upon neo-institutionalism (Agrawal, 2001; Ostrom, Burger,
Field, Norgaard, & Policansky, 1999; Quinn, Huby, Kiwasila,
& Lovett, 2007; Wollenberg, Merino, Agrawal, & Ostrom,
2007), this paper—although definitely indebted to neo-
institutionalism too (Forsyth & Johnson, 2014)—prefers prac-
tice theory. Whereas neo-institutionalism emphasizes the
design of a robust institutional arrangement as a precondition
for CFM to succeed, practice theory assumes that CFM works
best when it can align itself with socially-embedded logics that
predates the CFM initiative, for example through engagement
with local practices and social learning (Arts, Behagel, van
Bommel, de Koning, & Turnhout, 2013). Three practice-
based factors are hypothesized as being crucial for CFM perfor-
mance in this paper (to be justified in the theoretical section
below): (1) active engagement of community-based organiza-
tions (CBOs); (2) the practicability of CFM rules and regula-
tions for forest users; and (3) the emergence of a ‘‘Community
of Practice” through which information is shared, trust built
and practices learnt (here the term ‘‘community” is used
broadly; it involves relevant stakeholders inside and outside
the villages and forest lands; this includes local people, state for-
est agencies, donors, NGOs, etc., jointly constituting a learning
network; hence, not just a ‘‘community” stricto sensu).
The paper is structured as follows. First, the theoretical
foundation of the paper is justified, particularly the choice of
practice theory. Some crucial conditions for CFM to perform
are deduced from this theory and further operationalized for
the QCA. The latter is elaborated upon in the section that fol-
lows. We explain why we chose this methodological approach
and which version and technique we apply (fuzzy-set QCA; R
Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
software). In the results section we present our findings.
Finally, our results are discussed in light of the broader
CFM literature.
2. THEORETICAL FRAMEWORK: THE PRACTICE-
BASED APPROACH (PBA)
(a) Neo-institutionalism and practice theory
Much academic research and literature on CFM is built
upon neo-institutionalism to analyze, understand and explain
the success and failure of such initiatives (Agrawal, 2001;
Ostrom et al., 1999; Quinn et al., 2007; Wollenberg et al.,
2007). Neo-institutionalism in the CFM literature can be char-
acterized by (at least) three features. Firstly, it assumes, like
any institutional theory, that human behavior is guided by
rules, norms, incentives and sanctions (or ‘‘institutions”),
and that, as a consequence, human behavior can also be re-
directed and changed by introducing new institutions
(Giddens, 1984; Schmidt, 2008). In other words, humans are
depicted as rule-followers. Secondly, neo-institutionalism puts
much emphasis on the robustness of institutions (Anderies,
Janssen, & Ostrom, 2004; Ostrom, 2009a). Institutions need
to be well-established and well-embedded in societies and com-
munities in order to have the desired effects. In CFM litera-
ture, this second characteristic has led to research and
debates on which conditions CFM institutions are likely to
succeed under (Agrawal, 2001). As a consequence, authors
produced lists of relevant ‘‘design principles” for institution
building, such as clear demarcation of the forest, rules that
fit local conditions, participatory decision making, sanctioning
for non-compliance, and conflict resolution among others (see
for example Dressler et al., 2016; Ostrom, 1990, 2009b).
Thirdly, the question of how to measure the effects of CFM
institutions has also dominated the literature (Agrawal,
2001). Much of the research in the early 1990s was qualitative
in nature and focused on single case studies. Since then ever
more quantitative, comparative large-N studies on CFM have
been published (IFRI, 2015; Persha et al., 2011; Poteete &
Ostrom, 2008; Wollenberg et al., 2007).
Recently scholars have begun to emphasize new topics, such
as institutional flexibility and the diverse routes available to
achieve the effective management of forests and natural
resources. For example, Ostrom et al. (2002) state that flexibil-
ity can be an important asset in the construction of new insti-
tutions for the commons, as these institutions would be better
able to adapt to changing circumstances. In a similar vein, Van
Laerhoven and Ostrom (2007) have begun to highlight the
issues of complexity, uncertainty and dynamics in the regula-
tion of common properties. They underline the need to
develop new approaches to deal with these challenges. Such
approaches should move away from rather static and narrow
views of institutions and build more dynamic frameworks,
which include theoretical insights from various disciplines
(Bardhan & Ray, 2008; Forsyth & Johnson, 2014; Ostrom,
2009b).
Two relatively new approaches in the field of CFM that
claim to do so are critical institutionalism (Cleaver, 2002;
Cleaver & de Koning, 2015; De Koning, 2011) and the
practice-based approach (Arts, Behagel, Turnhout, de
Koning, & van Bommel, 2014; Arts et al., 2013; Ayana,
Vandenabeele, & Arts, 2015; Van der Arend & Behagel,
2011). The core concept of critical institutionalism is
‘‘institutional bricolage”. The assumption is that humans do
not truly follow rules, but improvise upon them according to
 COMMUNITY FOREST MANAGEMENT: AN ASSESSMENT AND EXPLANATION OF ITS PERFORMANCE THROUGH QCA
the situation at hand. People acting like bricoleurs combine
various institutional arrangements (socially embedded ones,
bureaucratic ones, and legal ones) into a patchwork that serves
their situation best—thus mobilizing some institutions and
neglecting others. The consequence of this approach is that
the three characteristics of neo-institutionalism need to be
fine-tuned: humans are not so much rule-followers but rule-
improvisers; the robustness of an institution gives no guaran-
tee that people will follow it; and the effects of institutions can-
not only be traced by large-N studies. In-depth qualitative
research is also needed to understand how and why people
adopt, adapt or reject newly introduced CFM rules and regu-
lations (De Koning, 2011). A second, related school of
thought is the so-called practice-based approach (PBA; Arts
et al., 2013, 2014; Ayana et al., 2015; Van der Arend &
Behagel, 2011). Like critical institutionalism it refines the insti-
tutional view on human behavior, in that it assumes that peo-
ple follow rules, norms and incentives only when these appear
legitimate, relevant and practical to them (March & Olsen,
1989). In this context Bourdieu (1977, 1990), one of the found-
ing fathers of current practice theory, points at the so-called
‘‘logic of practice” in social life. This logic is based on the daily
flow of activities that have historically and culturally emerged,
patterned and routinized in social fields. People act upon these
socially-embedded logics rather than upon (new) rules, norms
and incentives of institutions, unless both worlds (institutional
and practical logics) can be meaningfully aligned.
This paper builds upon the PBA of Arts and colleagues, in
particular. Inspired by Bourdieu (1977, 1990), Giddens (1984),
Schatzki and Knorr Cetina (2001), Murray Li (2007), Schatzki
(2010) and Shove, Pantzar, and Watson (2012) and adopting
elements of critical institutionalism (Cleaver, 2002), they
designed a conceptual framework to analyze issues of forest
and nature governance, including CFM (see for example
Ayana et al., 2015). Arts et al. (2013) define a practice as: an
ensemble of doings, sayings and things in a specific field of activ-
ity (be it a restaurant, a laboratory, a forestry department or a
community-based organization). The doings refer to social
interactions, the practical skills that people employ, and the
patterns and routines that they follow; the sayings refer to
people, their discursive interactions, and the explicit conven-
tions and knowledge that they utter; while things refer to
materials, artifacts, and society–nature interactions (like
rocks, man-made technologies, and nature in general).
Together these elements constitute a specific field of activity.
Accordingly, the definition emphasizes the importance of the
social and material settings in which these doings, sayings,
and things are situated, and through which they are brought
into being.
Based on the above definition, the PBA discerns three core
‘‘sensitizing concepts”: one at the level of agency (situated
agency), one at the level of structure (logic of practice), and
one at the level of knowledge and competencies (performativ-
ity) (Arts et al., 2013). Firstly, situated agency, distances itself
from the individualist, rational-strategic model of human
agency in many social science approaches (homo economico).
Instead, humans are conceptualized as social beings (homo
sociologico). This theoretical position is still shared by both
neo-institutionalism and practice theory. But the latter distin-
guishes itself from the former in terms of theorizing structure.
Whereas practice theory believes human behavior to be
shaped by socially embedded logics in particular, neo-
institutionalism assumes rules, norms, and incentives do so
(see the discussion above). Since the PBA considers the latter
position as problematic, it adopts Bourdieu’s concept of logic
of practice. Thirdly, performativity is about how knowledges
Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
3
and competencies, as learned in practice, make people perform
in certain ways and not others.
(b) From ‘‘sensitizing concepts” to ‘‘conditions in practice”
The above entails a (meta) theoretical positioning of the
PBA. However, to make the approach useful for our CFM
synthesis and QCA, the various concepts need to be defined
and operationalized into tangible actors, factors, conditions
and configurations relevant for the topic at hand. Note that
we define concepts in terms of thresholds (‘‘. . .a high degree
of. . .”—see Table 1 below). We do so as it is prescribed by
the QCA methodology (which will be explained in the method-
ology section below).
As for the first concept (situated agency), we will focus on
social organization, in this case community organizations
since they are considered crucial for CFM to work in much
of the relevant literature (Charnley & Poe, 2007; IFRI, 2015;
Poteete & Ostrom, 2004). However, particularly large-N stud-
ies can only assess their existence, size or design, and not so
much what they actually do and say in practice, or how they
change forest management practices, for example through
meetings and activities. In line with the PBA, we therefore
focus on the active engagement of community organizations
in CFM in our comparison of case studies below. While doing
so we hypothesize that such engagement is a precondition for
positive outcomes (in terms of livelihoods, income and forest
conditions; see above).
The second concept of logic of practice is operationalized as
‘‘practicability of rules”. The starting point of our argument is
the simple observation that CFM rules have been introduced
in all our case studies in one way or another. Subsequently,
the question is: do they work? It follows from the PBA that
such rules can only become effective once they are, or can
be, closely aligned with the daily realities of local people
(Arts & Babili, 2013; Ayana et al., 2015). Hence, CFM rules
should be ‘‘practical” for users, thus allowing users to work
with these in the field, and adopt related norms, standards
and requirements for forest management as a consequence.
Of course nothing will change if new rules do not challenge
established practices, but if these rules are too far removed
from daily life nothing will happen either. This is because—
as hypothesized in the PBA—people tend to follow established
practices rather than newly introduced incentives, norms and
rules (De Koning, 2011).
The third sensitizing concept, performativity, relates to the
role of knowledges and competencies in the performance of
practices. In CFM too, knowledge is considered crucial for
its performance (Agrawal & Chhatre, 2006; Fernandez-
Gimenez, Ballard, & Sturtevant, 2008). Often, though, schol-
ars refer to external professional and scientific knowledge as
being ‘‘injected” in local communities to improve forest man-
agement (De Koning et al., 2014), whereas the PBA highlights
the practical knowledge of people already present to deal with
forests in their daily practices (although not necessarily in
compliance with Western ideals of sustainable forest manage-
ment). Other scholars, like Agrawal (1995), reject the strong
distinction between Western science and indigenous knowl-
edge and instead talk of ‘‘multiple domains and types of
knowledges”. We agree with this latter statement, and in order
to express this conceptually, we decided to adopt Wenger’s
notion of Community of Practice (Wenger, 2000; Wenger &
Snyder, 2000). Wenger’s notion has emerged from practice
theory, and presents an active form of knowing-while-doing
as well as of social learning in shared domains and networks,
whatever type or form of knowledge is involved. Communities
4 WORLD DEVELOPMENT

Table 1. Summary of practice-based conditions for CFM to perform

Conditions
Community Active Engagement (CAE) High degree of community organization involvement
Practicality of Rules (POR)
Community of Practice (COP)
Definitions (as ‘‘thresholds”)
in CFM—indicated by membership, meetings, and
forest management activities.
High degree of connecting to and working with CFM
rules by local forest users—indicated by the actual
application of new norms, standards, and
requirements in the forest.
High degree of networking among internal and
external stakeholders based on common concerns,
social learning and mutual respect—indicated by
social interaction, joint activities, and co-production
of knowledge.
Sources
Arts et al. (2013), Bevir (2012), IFRI (2015),
Charnley and Poe (2007), Poteete and Ostrom
(2004)
Arts et al. (2013), Ayana et al. (2015), Bourdieu
(1977, 1990), Cleaver (2002), De Koning (2011)
Agrawal and Chhatre (2006), Baynes et al. (2015),
Charnley and Poe (2007), Fernandez-Gimenez
et al. (2008), Mustalahti and Lund (2009), Wenger
(2000, 2006) and Wenger and Snyder (2000)

of Practice (COPs) are ‘‘groups of people who share a concern
or a passion for something they do and learn how to do it bet-
ter as they interact regularly” (Wenger, 2006, p. 1). For CFM
too building a COP among internal and external stakehold-
ers—local people, forest officials, donors, and experts—seems
crucial for incrementally adopting new or revised forms of for-
est management, and changing established practices—while
keeping local people on board. However, a COP is often diffi-
cult to realize, because it presupposes mutual respect, trust,
and equality. Such is not easy to accomplish, particularly
because forest officials generally have problems working with
locals on an equal level, and engaging in ‘‘true” social learning
(Dang, 2014). Yet our hypothesis is that the existence of such a
COP is a precondition for CFM to perform.
3. METHODOLOGY
(a) Qualitative Comparative Analysis (QCA)
Like critical institutionalism, the PBA prescribes in-depth
qualitative research in order to understand the situated-ness
of relevant agencies, the logics of practice that those agencies
pursue, and the role of scientific, professional and practical
knowledges in the performance of CFM (Arts et al., 2013).
Therefore, we, along with our colleagues, have conducted sev-
eral field studies in the past (see Table 2). However, we need to
go beyond individual cases in order to be able to draw more
general explanatory factors and lessons from such practices
(Agrawal, 2001). Therefore, after collating the case studies,
we searched for a methodology to do so and adopted the
Qualitative Comparative Analysis (QCA) as a methodological
tool in this paper. QCA is both a research strategy and a
method to compare multiple cases (Rihoux, 2006; Rihoux &
Ragin, 2009). As a strategy, it is particularly useful to over-
come the challenges of generalizing findings from low- and
middle-N qualitative case-study research. As a method,
QCA exists as a specific collection of comparative tech-
niques—manual and software—that are both qualitative and
quantitative in nature (Ragin, 1987; Rihoux, 2006; Rihoux
& Ragin, 2009). The number of case studies involved in a
QCA generally ranges from 10 to 50 (see for example
Basurto, 2013; Dressler et al., 2016; Pahl-Wostl & Knieper,
2014).
Unlike mainstream statistical approaches, QCA does not
require large-N sample sizes to demonstrate causality, since
the basis of causal explanation is set theory, not correlational
theory (Ragin, 2000). Set theory assumes causality not on the
basis of co-variation of two variables in numerous measure-
ments, but on the basis of overlapping memberships of cases
to particular sets of things. Therefore, in QCA the terminology
of conditions—and of configurations of conditions—is pre-
ferred over ‘‘independent variables” and the terminology of
outcomes is preferred over ‘‘dependent variables” (Rihoux,
2006; Schneider & Wagemann, 2006). Moreover, conditions
and outcomes are not individually scored along an interval
scale, but cases as a whole are assessed whether they are mem-
bers to these conditions and outcomes, or not. Such an assess-
ment is executed on the basis of thresholds: if a case meets a
certain threshold, for example ‘‘a high degree of community
active engagement” (see Table 1), then that case is considered
a member of that set. If the threshold is not met, then it is not
a member. Scores are subsequently based on Boolean algebra:
‘‘true” = being a member = 1, and ‘‘false” = non-
member = 0. Once all Boolean scores of conditions and out-
comes belonging to a group of cases are determined the ana-
lyst assesses whether the outcomes are a subset of the
conditions, or the other way around. If so, causality is
assumed.
However, set theory adopts contingency: similar outcomes
may be produced by different configurations and similar con-
figurations may produce different outcomes, depending on the
context. Hence, several possible causal pathways for produc-
ing similar outcomes should be considered in QCA, and all
possible combinations of conditions should be tried out. Since
our synthesis includes three conditions, eight possible configu-
rations are implied (23 = 8). For example: presence of all three
conditions (CAE, POR, COP), absence of the first and pres-
ence of the other two (cae, POR, COP), absence of the first
two and presence of the last (cae, por, COP), absence of all
three (cae, por, cop), etc. (in QCA language, presence of a con-
dition is written in capitals and absence of a condition in lower
case). Subsequently, one should check whether these various
configurations produce the outcome, or not (through estab-
lishing set relationships, see above). This causal theory and
approach is also called ‘‘conjunctural causality” or ‘‘configura-
tional causality” (Thiem, Baumgartner, & Bol, 2015;
Wagemann & Schneider, 2010).
In set theory, the distinction between necessity and suffi-
ciency is important. ‘‘Whenever a causal condition is neces-
sary, (. . .) instances of the outcome will form a subset of
instances of the causal condition” (Ragin, 2000: 213). In other
words, for outcome X to occur, condition A needs to be pre-
sent, but A could also produce another outcome than X
(Ragin et al., 2008; Sehring, Korhonen-Kurki, & Brockhaus,
2013). For sufficient conditions, the opposite is valid—hence,
instances of a cause are a subset of the outcome (Schneider
& Wagemann, 2006; Sehring et al., 2013). Thus condition B

Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
 COMMUNITY FOREST MANAGEMENT: AN ASSESSMENT AND EXPLANATION OF ITS PERFORMANCE THROUGH QCA 5

Table 2. Overview of case studies

Researcher Country, region Case study Source
Ayana (ETH) Ethiopia, Gimbo Performance of CFM Ayana et al. (2015)
– qualitative case study (1 village)
– CFM since mid-1990s
Babili (TAN) Tanzania, Babati Performance of CFM Arts and Babili (2013)
– comparative case study (4 villages)
– CFM since early 1990s
Benneker (BOL) Bolivia, Lowland Performance of Community Forest Enterprises (CFE) Benneker (2008)
– qualitative case study (1 CFE)
– quantitative study (N = 50)
– CFM since late 1990s
Bose (IND) India, Rajasthan The micro-politics of CFM Bose (2012) and Bose,
– comparative case studies (5 villages) Arts, and van Dijk (2012)
– CFM since mid-1990s
Dang (VIET) Vietnam, Provinces Tay Ninh, The governance capacity of Forest Land Dang (2014)
Dak Lak and Lao Cai Allocation (FLA) policy
– comparative case study (3 regions)
– CFM since mid-1990s
De Koning (BOL) Bolivia, Beni CFM practices and performances De Koning (2011)
– qualitative case study (1 village)
– CFM since late 1990s
Katani (TANZ) Tanzania, Ukerewe Community management of micro-spring forests Katani (2010)
– comparative case study (6 villages)
– CFM since early 1990s
Nandigama (IND) India, Andhra Pr. CFM practices and performances Nandigama (2013)
– qualitative case study (1 village)
– CFM since late 1980s
Woldeamanuel (ETH) Ethiopia, Borana Community management of gum/resin trees Woldeamanuel (2011)
– qualitative case study (1 region)
– Gada system, centuries old
– Cooperatives since late-2000s
Yasmi (INDO) Indonesia, East Kalimantan Decentralized forest management and logging practices Yasmi (2003)
– comparative case study (2 villages)
– CFM since late 1990s

will always lead to outcome X, but there might be other con-
ditions or configurations leading to outcome X (Ragin et al.,
2008; Sehring et al., 2013).
(b) Fuzzy set QCA (fsQCA)
Fuzzy set QCA (fsQCA) is one of the variants of QCA
(other variants include crisp set QCA, multi value QCA, and
two step QCA; see Sehring et al., 2013). fsQCA was developed
as a response to crisp set QCA. Crisp set is based on the Boo-
lean binary approach that works with truth-values of 1 (true,
membership) and 0 (false, non-member). In contrast, fsQCA
allows researchers to use gradual variations in the membership
assessment of configurations and outcomes. A common stan-
dardized scale used in fsQCA is the following: 1 = member-
ship; 0.67 = more a member than a non-member;
0.33 = more a non-member than a member; 0 = non-
member. This scale is of particular value when dealing with
conditions and outcomes where it is hard to assess whether
they are true or false, such as the ones identified in this article.
The three identified preconditions for CFM (Community
Active Engagement, Practicality of Rules, and Community
of Practice) in our view do not lend themselves well to a binary
approach. The engagement of a community is better described
when using a degree of engagement and not whether it is pre-
sent or absent (see Table 1). The same goes for practicality of
rules. Certain rules and certain aspects of rules can be more
practical than others, and a Community of Practice can be
more or less integrated—with more/less network relations,
trust and learning involved; all of which are gradual processes.
Finally, fsQCA is also of added value if one wants to maintain
some of the case complexities in one’s synthesis—this is partic-
ularly relevant when following a PBA.
Overall, the researcher’s decisions on membership (thresh-
olds are met) and causality (subset relationships of outcomes
and configurations are present) are based on theoretical
knowledge, expert judgment, and empirical evidence. It is
therefore important in fsQCA that: (1) the conditions chosen
and their respective configurations and thresholds can be
linked to the empirical data, and (2) the data lend themselves
to be categorized according to the theory chosen (Rihoux,
2006). Moreover, it is vital that the researcher continuously
re-assesses and also re-questions these so-called boundaries
among theory and data (Sehring et al., 2013). Doing so also
helps the researcher to compare and contrast all possible con-
figurations of conditions with the outcomes. Finally, Boolean
algebra permits minimization of configurations, known as the
most parsimonious solution. If two configurations leading to
the same outcome show one different condition, this condition
is regarded as irrelevant.
(c) Cases and data
For this article, we used data from ten different case studies on
CFM, all of which were located in tropical countries (see
Table 2). This selection of cases was based on CFM research
executed at the Forest and Nature Conservation Policy group
(FNP) of Wageningen University, the Netherlands, during the

Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
6 WORLD DEVELOPMENT
last decade (from 2005 until 2015). All of these case studies have
been long term and in-depth. Most of the cases have been part
of PhD research. For this paper we use the information pub-
lished in PhD theses, individual chapters, and journal papers,
which were produced as a direct result of these case studies.
All case studies address the CFM approach in one or more
villages or regions, and all have been in place for at least two
decades. They all include in-depth information on the three
preconditions for CFM (as identified above), as well as on
its performance. Yet, the context, details and the geographical
regions differ per case study. In addition, the case studies used
different indicators for livelihoods and forest conditions per-
formances, for example household income surveys and quali-
tative interview data for livelihoods; and both field data and
people’s perceptions on species diversity, tree density and for-
est cover for forest conditions. QCA allows for harmonizing
these indicators through the use of thresholds—high or low
degrees of performances that define (non)membership—so
that the cases can be compared. Furthermore, we dealt with
the diversity in locations, scopes, methods and indicators by
actively engaging the original researchers of the case studies
with our synthesis. In doing so, we followed a number of steps.
Firstly, we (the co-authors) re-read the existing reports on the
case studies, then we assessed the three conditions and perfor-
mances of CFM from these studies. Secondly, we discussed
our initial fsQCA scores, as well as the relationships between
the conditions and outcomes we initially found, with the orig-
inal researchers. Thirdly, the original researchers also filled in
a questionnaire specifically designed for this article. From the
questionnaire, we then—in a fourth step—constructed a gen-
eral description of each case that focused on the conditions
and outcomes relevant to the QCA. This was done in order
to deal with the variation in cases and to mutually calibrate
their fuzzy-set values, thus ensuring internal validity of the
synthesis. From this step, we finally created the so-called
‘‘raw data” on conditions and performances of CFM (see
Tables 3 and 4 below). The values in these tables refer to the
average fsQCA scores on conditions and performance of the
group of communities included in each case study. Again,
these values were decided upon by us, but in close interaction
with the original researchers.
Although ten case studies is generally considered a low
amount for fsQCA, it does not render an analysis impossible,
particularly since we are working with a limited number of
conditions (3) and configurations (8) (this is also referred to
as small-N research; see for example Basurto, 2013; Ryan &
Table 3. Performance of cases
Cases LLH FC PERF
Benneker (BOL) 0.67 0.67 0.67
Dang (VIETN) 1 0.33 0.67
De Koning (BOL) 0.67 0.67 0.67
Nandigama (INDIA) 0.67 0.67 0.67
Ayana (ETH) 0.33 0.33 0.33
Babili (TANZ) 0 0.67 0.33
Bose (INDIA) 0 0.33 0.33
Katani (TANZ) 0.33 0.33 0.33
Yasmi (INDON) 0.67 0 0.33
Woldeamanuel (ETH) 0 0 0 4. RESULTS
LLH: Degree of Livelihood Improvement; FC: Degree of Forest Condi-
tion Enhancement; PERF: Overall Degree of Performance (1 = mem-
bership, so the case concerned is fully meeting the threshold of high levels
of livelihood improvement, forest condition enhancement and/or overall
performance; 0.67 = more a member than a non-member; 0.33 = more a
non-member than a member; 0 = non-member).
Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
Smith, 2012). The use of fuzzy-sets in small-N research is dis-
cussed in the QCA literature. Some authors critique it and pre-
fer a binary approach in such small-sample research (Sehring
et al., 2013; Rihoux, 2006). Others, like Schneider and
Wagemann (2010), argue that the theoretical background
and quality of the data should be the deciding factor for the
application of fuzzy or crisp set QCA. In our synthesis, given
the gradual nature of the three preconditions for CFM perfor-
mance and the calibration of the data with the group of
researchers in several rounds, we believe that we are on safe
grounds for using fsQCA in our small-N research. We chose
a four-value fsQCA that allows us to choose between 0 (false),
0.33 (more false than true), 0.67 (more true than false), and 1
(true). The reason for this choice is that we required more vari-
ety than just 0 and 1 (see the discussion above), yet at the same
time we needed to have somewhat larger categories in order to
deal with the variation in cases and their somewhat different
research approaches.
Once the ‘‘raw data” were determined, the causal analysis
could be executed. For this article, we used the QCA Pro-
Package (Version 1.0–0) for the R Environment of Statistical
Computing and Graphics (Thiem, 2016). In earlier stages of
our analysis, we also applied a manual analysis (compare
fsQCA Sehring et al., 2013) and another software tool (fsQCA
software 2.0; Ragin, Drass, & Davey, 2006). As analysis in R
Software allows for more precise results and is less error-prone
than manual analysis, we decided to only present the results
from the analysis in R software for this paper, and in so doing
we used the most recent package (Thiem, 2016). Nonetheless,
by comparing different methods, both manual and software
based, trying out different value-sets, and by discussing our
results-in-progress with experts, we were able to triangulate
our findings from the analysis in R.
It is also possible to test for necessity and sufficiency in R
software. Technically, necessary conditions need to be identi-
fied first and then removed from the database before the suf-
ficiency analysis can take place (Ragin, 2009). However, a
software run is not necessarily based on all ‘‘raw data” from
a set of cases, since the technique posits certain requirements
in order to find causal pathways: (1) requirements for member-
ship to configurations; and (2) requirements for consistency in
configurations. By default, the causal analysis only includes
those configurations in which at least one—but preferably
more—cases show a membership above 0.5 (so 0.67 or 1.0,
indicating that these cases are a member to those configura-
tions, or more a member than a non-member). In addition,
before the analysis is run, a consistency threshold for the out-
come needs to be set (also known as inclusion cut-off). Consis-
tency refers to the proportion of cases that display the
outcome (Ragin et al., 2008, p. 46). Conceptually, it resembles
the notion of significance in statistics (Schneider &
Wagemann, 2010). The higher this threshold is set, the fewer
cases that are included in the analysis. We therefore chose
an inclusion cut-off of 0.75, as this would still include the
majority of cases, while consistency is still high enough to
identify possible causal pathways (Ragin et al., 2008). We
made sure the QCA software ran all possible solutions to
finally derive the most parsimonious one.
(a) Qualitative insights
Table 3 summarizes the performance of the different CFM
initiatives on improving livelihoods (LLH), enhancing forest
 COMMUNITY FOREST MANAGEMENT: AN ASSESSMENT AND EXPLANATION OF ITS PERFORMANCE THROUGH QCA 7

Table 4. Raw dataset of conditions (CAE, POR, COP) and outcomes (PERF)
Case CAE POR COP PERF
Benneker (BOL) 0.67 0.33 0.67 0.67
Dang (VIETN) 0.33 0.33 0.67 0.67
De Koning (BOL) 1 0.67 0.67 0.67
Nandigama (INDIA) 1 0.33 0.67 0.67
Ayana (ETH) 0.67 0.33 0.33 0.33
Babili (TANZ) 0.67 0.33 0.67 0.33
Bose (INDIA) 0.67 0.33 0.33 0.33
Katani (TANZ) 0.67 0.33 0.67 0.33
Yasmi (INDON) 0.67 0.33 0 0.33
Woldeamanuel (ETH) 0.33 0 0 0
CAE: Degree of Community Active Engagement; POR: Degree of Practicality of Rules; COP: Degree of Community of Practice; PERF: Overall Degree of
Performance (1 = membership, so the case concerned fully meets the threshold of high levels of CAE, POR, COM and/or PERF; 0.67 = more a member
than a non-member; 0.33 = more a non-member than a member; 0 = non-member).

conditions (FC) and provides one overall average performance
indicator (PERF), using the calibrated, four-value, fuzzy-set
scale. The results are definitely mixed. The best performers
(PERF) are in the top of the table, the least in the bottom.
Four studies show relatively high performance, five some
and one none. In addition, the scores on livelihoods and forest
conditions also vary with the cases. In six case studies, LLH
and FC score equally (win–win and lose–lose) and in four as
trade-offs (win–lose and lose–win). When the ten cases are
taken together, CFM performs rather similarly on livelihoods
and forest conditions (which goes against the general literature
in which it is claimed that—overall—CFM benefits forests
more than people; see the discussion section below).
To validate Table 3, we now present examples of cases with
high, some and no performance, respectively. Beginning with a
high-performance case, the study of De Koning (2011) covers
CFM in a Bolivian community in the Amazon region, and
analyzes how a community collectively managed their forest
resources. As the communal forest area was already substan-
tial in size and rich in biodiversity the new CFM regime did
not lead to an increase in forest size, or an improvement of
forest conditions. However, it did lead to deforestation being
avoided, as it offered the community a way to enhance their
livelihoods from standing forests—particularly through the
collection of Brazil nuts and the production of timber. More-
over, land titles of the communal forest areas were formalized
through CFM, providing the community with the necessary
stability of access to forest resources, additional income and
even to medical services (in which part of the forest revenue
was collectively invested).
Arts and Babili (2013) provides a case study where some
performance is demonstrated. In this study, the forest area
of four North-Tanzanian neighboring villages under the
CFM regime was assessed through satellite images, focus
groups and field observations. The study indicated that the
forest area expanded throughout the 2000s (by about 0.3%
each year due to replanting and natural regeneration). In addi-
tion, forest quality, particularly tree density and species diver-
sity, improved over time. Yet, according to a household survey
and respondents’ individual perceptions neither livelihoods
nor income was gained from CFM over the years.
A last example from Table 3 showing no performance, is the
one of Woldeamanuel (2011). This study analyzes the use and
management of gum and resin trees in the Borana region of
Southern Ethiopia. Overall, CFM—in the form of the tradi-
tional Gada system and modern cooperatives—does not lead
to an improvement of forest conditions and livelihoods. Most
communities tap the gum and resin without any active
management of the trees. The communities sell the products
to individual traders for very low prices without cooperatives
interfering to get a higher income for small producers. More-
over, gum and resin trees are regularly cleared to maintain the
open rangelands for pastoralism, in accordance with Gada
rules. While Gada rules do prescribe protection and active
management of certain trees, gum and resin species do not fall
under this regime (so far).
Table 4 presents the raw dataset of conditions (CAE, POR,
COP) and outcomes (PERF) for all ten studies. The outcome
PERF has been copied from Table 3. For the remaining con-
ditions, we again use the four-value approach. A first observa-
tion at face value shows that, overall, CAE scores relatively
high, POR relatively low and COP somewhere in the middle.
Therefore, taking the mixed performance of the cases as a
starting point (PERF), the active engagement of community-
based organizations seems an enabling condition for CFM to
work (CAE, high scores), whereas a lack of practical rules
seems a constraining one (POR, low scores). Of course, these
statements need to be verified in the QCA below, but before
doing so we first articulate them by recalling our three exam-
ples from above.
In the Bolivian case study of De Koning, CAE scores very
high: a well-functioning and active community organization
managed the collective forest areas effectively and without
many problems. POR also scored high, as national legislation
provided the community with formalized land titles, which
enhanced one of its most important forest practices: Brazil
nut collection. In other words, the CFM legislation matched
the practical needs of the community very well. However,
POR did not receive the highest score as the legislation was
complex, the process for applying for land titles was cumber-
some, and free access to timber became restricted through the
CFM regime. Finally, COP scored high in this particular com-
munity because of the presence of a local NGO. This NGO
facilitated the implementation of CFM, raised awareness
and actively engaged with local people to build high levels of
trust, knowledge exchange arrangements and mutual relation-
ships (De Koning, 2011).
The case study of Babili, shows relatively high scores for
CAE and COP, and a relatively low one for POR. Indeed,
all four Tanzanian villages under investigation had strong vil-
lage forest committees and leaders, who were also supported
by an extensive external network of professionals (NGOs,
donors, forest officials, and university scientists). Although
mutual trust and learning was quite limited initially, particu-
larly at the beginning of the CFM initiative in the 1990s, these
phenomena strongly improved over time. Yet the villagers

Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
8 WORLD DEVELOPMENT
remained dissatisfied with some of the CFM rules, particularly
the ban on grazing in the forests—such rules did not match
their practices well (Arts & Babili, 2013).
Finally, the case of Woldeamanuel shows low scores for all
three conditions. Whereas some engagement of Gada institu-
tions and cooperatives exists with regard to the gum and resin
use systems, the rules do not facilitate more sustainable and
economically favorable practices. Moreover, a shared Com-
munity of Practice to reverse this situation has so far not
emerged (Woldeamanuel, 2011).
(b) Software solutions
The R Software analysis revealed that there are no necessary
conditions, or configurations of conditions, that have to be
present for the outcome—a high degree of CFM perfor-
mance—to occur. In other words, the analysis could not iden-
tify one configuration that is always necessary to lead to the
outcome. Still, the outcome can be caused by various sufficient
configurations, these are not necessary for the outcome to
occur, but if present, and the outcome has not yet been pro-
duced, it will follow. We therefore focused on the sufficiency
analysis in the next step. This analysis offered the following
solution (the best explaining pathway from a sufficiency per-
spective): presence of CFM performance (PERF) is caused
by a combination of absence of Community Active Engage-
ment (cae; written in lower case) and presence of a Community
of Practice (COP; written in capitals):
cae þ COP $ PERF
Table 5 includes the so-called ‘‘parameters of fit” of this
solution. The higher these are, the stronger the causality claim.
Parameters of fit consist of the inclusion (incl.) and the cover-
age scores (both cov.r and cov.u). Incl. stands for the suffi-
ciency inclusion score, i.e., the degree to which the
conditions/configurations are subsets of the outcome (Ragin
et al., 2008). Table 5 shows that these are relatively high for
both cae (0.79) and COP (0.86). Cov.r and cov.u. stand for
raw coverage and unique coverage respectively. Coverage
shows how much the conditions/configurations cover the out-
come within the subset relationship. Unique coverage refers to
the individual conditions, raw coverage to their combination
(Ragin et al., 2008). Table 5 shows that, in particular, the cov-
erage measurements for COP are high.
As a final check, we looked at whether the negations of the
two conditions found (a presence of CAE and an absence of
cop) are also sufficient for the inverse outcome. In so doing,
one constructs a counterfactual of the analysis to check
whether the conditions behave similarly in an inverse situa-
tion. In this final check, CAE did not produce the expected
inverse result, whereas cop did. We therefore conclude that
the presence of a Community of Practice (COP) is a sufficient
causal pathway for CFM to perform well. Alternative manual
and software analyses we have run previously have also
pointed to COP as the most important condition, thus con-
firming this result. Yet, other (unknown) conditions and con-
figurations may still produce this outcome (since COP is a
sufficient condition, not a necessary one).
Table 5. Parameters of fit in the QCA-R software analysis
Incl. cov.r cov.u
1. cae 0.798 0.612 0.076
2. COP 0.855 0.924 0.388
Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
5. DISCUSSION AND CONCLUSION
Many review papers on CFM, or related initiatives, show
that field results are generally mixed. Both successes and fail-
ures in improving local livelihoods and forest conditions are to
be observed in this domain (Agrawal, 2005; Baynes et al.,
2015; Charnley & Poe, 2007; IFRI, 2015; Mustalahti &
Lund, 2009; Pagdee et al., 2006). Our findings support this
conclusion, half of the studies show medium to high perfor-
mance, the other half some or no performance. However,
the general notion that CFM appears to have more ecological
benefits than socio-economic ones is not confirmed by our
study. Besides this win-lose trade-off, the reversal is found in
our sample, as well as win–win and lose–lose combinations.
Overall, though, both objectives score similarly. As we are
working with a rather limited sample of case studies we
should, of course, be very careful in making generalizing state-
ments. We will come back to this issue of generalization at the
end of this discussion section. However, first, we will discuss
the explanatory conditions and configurations for CFM per-
formance in light of our findings and the literature.
Local community-based organizations (CBOs) have been
identified in the literature as being instrumental to the success-
ful management of common pool resources (see for example
Charnley & Poe, 2007; IFRI, 2015; Poteete & Ostrom,
2004). However, different roles have been assigned to these
community organizations, varying from a more passive gate-
keeper role (Mustalahti & Lund, 2009) to more active roles
for rule making and the monitoring of rule compliance for
example (Charnley & Poe, 2007; IFRI, 2015; Pagdee et al.,
2006; Poteete & Ostrom, 2004). Regarding the latter, current
debates and research on community biodiversity monitoring,
e.g., under REDD+, also allocate an important, active role
to communities (see for example Pratihast, Herold, De Sy,
Murdiyarso, & Skutsch, 2013). Consequently, the establish-
ment of community organizations has become a common
practice in participatory forest management and conservation
(including all ten case studies in this paper). However, our
findings show that a high degree of CAE—‘‘community active
engagement”, as we have called it—is not sufficient for CFM
to perform well. At best, from our raw data we identified it
as a potential enabling condition, but it was not confirmed
as such by the software analysis.
In the theoretical section we have already referred to much
academic research and literature that emphasizes the relevance
of robust institutions, i.e., well-embedded local rules of the
game to enhance local livelihoods and forest conditions
(Agrawal, 2001; Agrawal & Chhatre, 2006; Baynes et al.,
2015; Ostrom et al., 1999; Poteete & Ostrom, 2004; Quinn
et al., 2007; Wollenberg et al., 2007). Theoretically and empir-
ically, while being inspired by critical institutionalism
(Cleaver, 2002; Cleaver & De Koning, 2015), practice theory
(Schatzki & Knorr Cetina, 2001; Shove et al., 2012) and
practice-based field research (Ayana et al., 2015; De Koning,
2011), we challenged this position by emphasizing the rele-
vance of the practicability of local institutions, instead of their
robustness, and developed the condition of POR (practicabil-
ity of rules). However, although our raw data seem to point to
POR as being important for CFM to perform, this was not
confirmed by our software analysis (against our expectations;
see Arts et al., 2013). At the same time, neo-institutionalists
also recognize that the actual performance and effectiveness
of institutions in common property resources regimes remain
complex and uncertain (Van Laerhoven & Ostrom, 2007).
Hence, practicability is not a crucial explanatory factor for
 COMMUNITY FOREST MANAGEMENT: AN ASSESSMENT AND EXPLANATION OF ITS PERFORMANCE THROUGH QCA
CFM performance in our sample, but more research is needed
to carefully verify or falsify this position.
The condition COP (‘‘Community of Practice”) does how-
ever emerge from our QCA analysis as a crucial precondition
for CFM performance. Although the concept of COP is
hardly addressed in the CFM literature, its constitutive ele-
ments (networks, knowledge, learning and trust) are regularly
covered. For example, several scholars address the relevance
of both professional and local knowledge for CFM
(Agrawal, 2005; Agrawal & Chhatre, 2006; Fernandez-
Gimenez et al., 2008). Others conceive of social learning as
a necessary part of CFM (Berkes, 2009; Dang, 2014). Some
scholars also cover the importance of cooperative networks
that stretch beyond the local level—particularly forest agen-
cies, NGOs, donors, universities and companies (Baynes
et al., 2015; Charnley & Poe, 2007; Mustalahti & Lund,
2009; Wiersum, Humphries, & Van Bommel, 2013). A final
group emphasizes the relevance of mutual trust among those
involved in CFM (Baynes et al., 2015; Berkes, 2009). The lat-
ter is particularly difficult to establish, root and maintain,
given the different interests, positions, cultures and discourses
of all the agencies involved (Mustalahti & Lund, 2009).
Governments, while expressing support for CFM, do little
in the way of action, financial support by NGOs is often only
temporary, and forest professionals generally find it hard to
work with locals on an equal level (Dang, 2014), and such
proceedings easily undermine trust. Despite the scarcity of
research on COP in this field, our findings thus do find some
confirmation in the literature.
Of course, our findings should be put into perspective.
Firstly, we have only addressed a small fraction of all the fac-
tors that could possibly drive CFM performance. In the intro-
duction of the paper we have already referred to biophysical,
demographic, economic, institutional, socio-political, and
external factors that are derived from an extensive body of
REFERENCES
Agrawal, A. (1995). Dismantling the divide between indigenous and
scientific knowledge. Development and Change, 26(3), 413–439.
Agrawal, A. (2001). Common property institutions and sustainable
governance of resources. World Development, 29(10), 1649–1672.
Agrawal, A. (2005). Environmentality: Technologies of government and the
making of subjects. Durham, NC: Duke University Press.
Agrawal, A., & Chhatre, A. (2006). Explaining success on the commons:
Community forest governance in the Indian Himalaya. World Devel-
opment, 34(1), 149–166.
Anderies, J., Janssen, M., & Ostrom, E. (2004). Framework to analyze the
robustness of social-ecological systems from an institutional perspec-
tive. Ecology and Society, 9(1), 18.
Arnold, J. (2001). Forests and people: 25 years of community forestry.
FAO.
Arnold, J. E. M. (2001). Forests and people: 25 years of community
forestry. Rome: FAO.
Arts, B., & Babili, I (2013). Global forest governance: Multiple practices of
policy performance Forest and Nature Governance. Springer, pp. 111-
132.
Arts, B., Behagel, J., Turnhout, E., de Koning, J., & van Bommel, S.
(2014). A practice based approach to forest governance. Forest Policy
and Economics, 49, 4–11.
Arts, B., Behagel, J., van Bommel, S., de Koning, J., & Turnhout, E.
(2013). Forest and nature governance. A practice based approach.
Springer.
Ayana, A. N., Vandenabeele, N., & Arts, B. (2015). Performance of
participatory forest management in Ethiopia: Institutional arrange-
ment versus local practices. Critical Policy Studies, 1–20.
Bardhan, P. K., & Ray, I. (Eds.) (2008). The contested commons:
Conversations between economists and anthropologists. Oxford: Black-
well.
Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
9
literature. In this study we have, in particular, addressed con-
ditions related to the last three—institutional, socio-political,
external factors—and added to the literature, we believe, the
relevance of ‘‘Communities of Practice” for CFM. On the
other hand, any explanatory study by definition works in a
field of complex causality, including single case studies that
embrace many factors or large-N studies that identify only a
few variables. A full causal picture is impossible—whatever
theory or methodology one applies. Secondly, by applying
the PBA and QCA at the same time, we have created a tension
between in-depth situated practice analysis as preferred by the
PBA, and inherently more general and superfluous statements
generated through QCA. Although we do not have a solution
for this tension, we still think it is worthwhile acknowledging
both in order to go beyond single-case findings yet maintain at
least some case complexity in the analysis. This is also the rea-
son why we chose to use the fuzzy set QCA and not the crisp
set QCA, since the former brings more nuance to the analysis
than the latter. Thirdly, our sample of case studies is of course
small (N = 10). This should make us very modest in externally
generalizing our findings. Yet, the QCA claims high internal
validity, given its strict procedures. This also applies to us,
we believe, since we paid much attention to concept opera-
tionalization, data calibration, and method triangulation in
this paper. In addition, like in any N = 1 or small-N case-
study research, one is always allowed to externally generalize
in a theoretical sense and thus formulate research hypotheses
(Yin, 1994). So in conclusion: we claim that a Community
of Practice—defined as ‘‘a high degree of networking among
internal and external stakeholders based on common con-
cerns, social learning and mutual respect”—is the key driving
force for success in our sample, and, moreover, hypothesize
that it is one of the relevant conditions for Community Forest
Management to perform well in terms of improving liveli-
hoods and forest conditions in general.
Basurto, X. (2013). Linking multi-level governance to local common-pool
resource theory using fuzzy-set qualitative comparative analysis:
Insights from twenty years of biodiversity conservation in Costa Rica.
Global Environmental Change, 23(3), 573–587.
Baynes, J., Herbohn, J., Smith, C., Fisher, R., & Bray, D. (2015). Key
factors which influence the success of community forestry in develop-
ing countries. Global Environmental Change, 35, 226–238.
Benneker, C. (2008). Dealing with the state, the market and NGOs. The
impact of institutions on the constitution and performance of community
forest enterprises in the lowlands of Bolivia PhD thesis. Netherlands:
Wageningen University.
Berkes, F. (2009). Evolution of co-management: Role of knowledge
generation, bridging organizations and social learning. Journal of
Environmental Management, 90(5), 1692–1702.
Bevir, M. (2012). New Labour: A critique. Routledge.
Blomley, T., Pfliegner, K., Isango, J., Zahabu, E., Ahrends, A., & Burgess,
N. (2008). Seeing the wood for the trees: An assessment of the impact
of participatory forest management on forest condition in Tanzania.
Oryx, 42(03), 380–391.
Bose, P (2012). Forest rights: The micro-politics of decentralisation and
forest tenure reform in tribal India PhD Thesis. Netherlands: Wagenin-
gen University.
Bose, P., Arts, B., & van Dijk, H. (2012). ‘Forest governmentality’: A
genealogy of subject-making of forest-dependent ‘scheduled tribes’ in
India. Land Use Policy, 29(3), 664–673.
Bourdieu, P. (1977). Outline of a theory of practice (Vol. 16). Cambridge
University Press.
Bourdieu, P. (1990). The logic of practice. Cambridge: Polity Press.
Bowler, D. E., Buyung-Ali, L. M., Healey, J. R., Jones, J. P., Knight, T.
M., & Pullin, A. S. (2012). Does community forest management
10 WORLD DEVELOPMENT
provide global environmental benefits and improve local welfare?.
Frontiers in Ecology and the Environment, 10(1), 29–36.
Charnley, S., & Poe, M. R. (2007). Community forestry in theory and
practice: Where are we now?. Annual Review of Anthropology, 36(1),
301.
Cleaver, F. (2002). Reinventing institutions: Bricolage and the social
embeddedness of natural resource management. The European Journal
of Development Research, 14(2), 11–30.
Cleaver, F., & de Koning, J. (2015). Furthering critical institutionalism.
International Journal of the Commons, 9(1), 1–18.
Dang, T (2014). The governance capacity of forest land allocation policy in
Vietnam PhD thesis. Netherlands: Wageningen University.
De Koning, J. (2011). Reshaping institutions: bricolage processes in
smallholder forestry in the Amazon PhD thesis. Netherlands: Wagenin-
gen University.
De Koning, J., Turnhout, E., Winkel, G., Blondet, M., Borras, L.,
Ferranti, F., ... Jump, A. (2014). Managing climate change in
conservation practice: An exploration of the science–management
interface in beech forest management. Biodiversity and conservation, 23
(14), 3657–3671.
Dressler, W., Büscher, B., Schoon, M., Brockington, D., Hayes, T., Kull,
C. A., & Shrestha, K. (2010). From hope to crisis and back again? A
critical history of the global CBNRM narrative. Environmental
Conservation, 37(01), 5–15.
Dressler, W. H., Wilson, D., Clendenning, J., Cramb, R., Keenan, R.,
Mahanty, S., ... Lasco, R. D. (2016). The impact of swidden decline on
livelihoods and ecosystem services in Southeast Asia: A review of the
evidence from 1990 to 2015. Ambio, 1–20.
Fernandez-Gimenez, M. E., Ballard, H. L., & Sturtevant, V. E. (2008).
Adaptive management and social learning in collaborative and com-
munity-based monitoring: A study of five community-based forestry
organizations in the western USA. Ecology and Society, 13(2), 4.
FAO (1978). Forestry for local community development. Rome: FAO.
Forsyth, T., & Johnson, G. (2014). Elinor Ostrom’s legacy: Governing the
commons and the rational choice controversy. Development and
Change, 45(5), 1093–1110.
Giddens, A. (1984). The constitution of society Outline of the theory of
structuration. Univ of California Press.
IFRI (2015). Documenting lessons from research by the IFRI network on
developing policies and programs on community forestry. Michigan:
University of Michigan.
Katani, J. Z. (2010). The role of multiple institutions in the management of
micro spring forests in Ukerewe, Tanzania PhD thesis. Netherlands:
Wageningen University.
Kumar, S. (2002). Does ‘‘participation” in common pool resource
management help the poor? A social cost–benefit analysis of joint
forest management in Jharkhand, India. World Development, 30(5),
763–782.
March, J. G., & Olsen, J. P. (1989). Rediscovering institutions: The
organizational basis of politics. New York: The Free Press.
Murray Li, T. (2007). Practices of assemblage and community forest
management. Economy and Society, 36(2), 263–293.
Mustalahti, I., & Lund, J. F. (2009). Where and how can participatory
forest management succeed? Learning from Tanzania, Mozambique,
and Laos. Society & Natural Resources, 23(1), 31–44.
Nandigama, S. (2013). Invited spaces and informal practices in participatory
community forest management in India Forest and Nature Governance,
A Practice Based Approach (pp. 89–110). Heidelberg: Springer.
Ostrom, E. (1990). Governing the commons: The evolution of institutions for
collective action. Cambridge: University Press.
Ostrom, E. (2009b). A general framework for analyzing sustainability of
social-ecological systems. Science (New York, NY), 325(5939),
419–422.
Ostrom, E., Burger, J., Field, C. B., Norgaard, R. B., & Policansky, D.
(1999). Revisiting the commons: Local lessons, global challenges.
Science, 284(5412), 278–282.
Ostrom, E., Dietz, T., Dolsak, N., Stern, P. C., Stonich, S., & Weber, E.
U. (2002). The drama of the commons. National Academies Press.
Ostrom, E. (2009a). Design principles of robust property-right institu-
tions: What have we learned?. In K. Ingram, & Y. Hong (Eds.),
Property rights and land policies. Cambridge, MA: Lincoln Institute of
Land Policy.
Pagdee, A., Kim, Y.-S., & Daugherty, P. (2006). What makes community
forest management successful: A meta-study from community forests
throughout the world. Society and Natural Resources, 19(1), 33–52.
Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014
Pahl-Wostl, C., & Knieper, C. (2014). The capacity of water governance to
deal with the climate change adaptation challenge: Using fuzzy set
qualitative comparative analysis to distinguish between polycentric,
fragmented and centralized regimes. Global Environmental Change, 29,
139–154.
Persha, L., Agrawal, A., & Chhatre, A. (2011). Social and ecological
synergy: Local rulemaking, forest livelihoods, and biodiversity con-
servation. Science, 331(2011), 1606–1608.
Poteete, A. R., & Ostrom, E. (2004). Heterogeneity, group size and
collective action: The role of institutions in forest management.
Development and Change, 35(3), 435–461.
Poteete, A. R., & Ostrom, E. (2008). Fifteen years of empirical research on
collective action in natural resource management: struggling to build
large-N databases based on qualitative research. World Development,
36(1), 176–195.
Pratihast, A. K., Herold, M., De Sy, V., Murdiyarso, D., & Skutsch, M.
(2013). Linking community-based and national REDD+ monitoring:
A review of the potential. Carbon Management, 4(1), 91–104. http://dx.
doi.org/10.4155/cmt.12.75.
Quinn, C. H., Huby, M., Kiwasila, H., & Lovett, J. C. (2007). Design
principles and common pool resource management: An institutional
approach to evaluating community management in semi-arid Tanza-
nia. Journal of Environmental Management, 84(1), 100–113.
Ragin, C. C. (1987). The comparative method: Moving beyond qualitative
and quantitative strategies The comparative method. Moving beyond
qualitative and quantitative strategies. University of California Press.
Ragin, C. C. (2000). Fuzzy-set social science. University of Chicago Press.
Ragin, C. C., Drass, K. A., & Davey, S. (2006). Fuzzy-set/qualitative
comparative analysis 2.0. Tucson, Arizona: Department of Sociology,
University of Arizona.
Ragin, C. C. (2009). Qualitative comparative analysis using fuzzy sets
(fsQCA). In B. Rihoux, & C. C. Ragin (Eds.), Configurational
comparative methods: Qualitative comparative analysis (QCA) and
related techniques (pp. 87–121). London: Sage.
Ragin, C. C., Rubinson, C., Schaefer, D., Anderson, S., Williams, E., &
Giesel, H. (2008). User’s guide to fuzzy- set/qualitative comparative
analysis (Vol. 87). University of Arizona.
Rihoux, B. (2006). Qualitative Comparative Analysis (QCA) and related
systematic comparative methods recent advances and remaining
challenges for social science research. International Sociology, 21(5),
679–706.
Rihoux, B., & Ragin, C. C. (2009). Configurational comparative methods:
Qualitative comparative analysis (QCA) and related techniques. Sage.
RRI (Rights & Resources Initiative) (2014). What future for reform?
Progress and slowdown of forest tenure reform since 2002. Washington
DC: RRI.
Ryan, M., & Smith, G. (2012). Towards a comparative analysis of
democratic innovations: Lessons from a small-n fsQCA of participatory
budgeting. Revista Internacional de Sociologı́a (RiS), 70(2), 89–120.
Schatzki, T. R. (2010). Site of the social: A philosophical account of the
constitution of social life and change. Penn State Press.
Schatzki, T. R., & Knorr Cetina, K. (2001). The practice turn in
contemporary theory. London: Routledge.
Schmidt, V. A. (2008). Discursive institutionalism: The explanatory power
of ideas and discourse. Annual Review of Political Science, 11, 303–326.
Schneider, C. Q., & Wagemann, C. (2006). Reducing complexity in
Qualitative Comparative Analysis (QCA): Remote and proximate
factors and the consolidation of democracy. European Journal of
Political Research, 45(5), 751–786.
Schneider, C. Q., & Wagemann, C. (2010). Standards of good practice in
qualitative comparative analysis (QCA) and fuzzy-sets. Comparative
Sociology, 9(3), 397–418.
Sehring, J., Korhonen-Kurki, K., & Brockhaus, M. (2013). Qualitative
Comparative Analysis (QCA): An application to compare national
REDD+ policy processes. CIFOR.
Shove, E., Pantzar, M., & Watson, M. (2012). The dynamics of social
practice: everyday life and how it changes. Sage Publications.
Thiem, A.. 2016. Professional functionality for performing and evaluating
qualitative comparative analysis. R Package Version 1.0-0. URL:
http://cran.r-project.org/package=QCApro.
Thiem, A., Baumgartner, M., & Bol, D. (2015). Still lost in translation!
Three misunderstandings between configurational comparativists and
regesssional analysts. Comparative Political Studies, 1–33.
Umans, L. (1993). A discourse on forestry science. Agriculture and Human
Values, 10(4), 26–40.
 COMMUNITY FOREST MANAGEMENT: AN ASSESSMENT AND EXPLANATION OF ITS PERFORMANCE THROUGH QCA 11

Van der Arend, S., & Behagel, J. (2011). What participants do. A practice
based approach to public participation in two policy fields. Critical
Policy Studies, 5(2), 169–186.
Van Laerhoven, F., & Ostrom, E. (2007). Traditions and trends in the study
of the commons. International Journal of the Commons, 1(1), 3–28.
Wagemann, C., & Schneider, C. Q. (2010). Qualitative comparative
analysis (QCA) and fuzzy-sets: Agenda for a research approach and a
data analysis technique. Comparative Sociology, 9(3), 376–396.
Wenger, E. (2000). Communities of Practice and Social Learning Systems.
Organization, 7(2), 225–246.
Wenger, E. (2006). Communities of Practice: A brief introduction. http://
www.ewenger.com.
Wenger, E., & Snyder, W. (2000). Communities of practice: The
organizational frontier. Harvard Business Review, 2000, 139–145.
Wiersum, K. (2009). Community forestry between local autonomy and
global encapsulation: quo vadis with environmental and climate
change payments. Paper presented at the First Community Forestry
International Workshop: Thinking globally—acting locally: commu-
nity forestry in the international arena, Pokhara, Nepal.
Wiersum, K. F., Humphries, S., & Van Bommel, S. (2013). Certification of
community forestry enterprises: Experiences with incorporating com-
munity forestry in a global system for forest governance. Small-scale
Forestry, 12(1), 15–31.
Woldeamanuel, T. (2011). Dryland resources, livelihoods and institutions:
Diversity and dynamics in use and management of gum and resin trees in
Ethiopia PhD Thesis. Wageningen University Netherlands.
Wollenberg, E., Merino, L., Agrawal, A., & Ostrom, E. (2007). Fourteen
years of monitoring community-managed forests: Learning from
IFRI’s experience. International Forestry Review, 9(2), 670–684.
Yasmi, Y. (2003). Understanding conflict in the co-management of forests:
The case of Bulungan Research Forest. International Forestry Review,
5(1), 38–44.
Yin, R. (1994). Case study research: Design and methods. Thousand Oaks:
Sage.

Available online at www.sciencedirect.com

ScienceDirect

Please cite this article in press as: Arts, B., & de Koning, J. Community Forest Management: An Assessment and Explanation of its
Performance Through QCA, World Development (2017), http://dx.doi.org/10.1016/j.worlddev.2017.03.014