Participation without power: community-based organisations and the limits of collaborative forest governance in Ghana's Offinso Forest District

Forest governance across much of sub-Saharan Africa has, over the past 3 decades, been formally reorganised around the premise that people living beside forests should share in managing them. Ghana's own version of this shift runs through the Modified Taungya System, Community Resource Management Areas (CREMAs), Community-Based Forest Management and, more recently, the Ghana Cocoa Forest REDD+ Programme (GCFRP) and its Hotspot Intervention Areas (HIAs), each of which assigns local organisations a place in forest planning, monitoring and restoration (Ministry of Lands and Natural Resources, 2012). Community-based organisations (CBOs) are the institutional vehicle through which this assignment is meant to happen: they mobilise labour for planting and patrols, communicate between village and Forestry Commission and are widely credited in the literature with improving compliance and building local stewardship where they function well (Agrawal and Gibson, 1999; Persha et al., 2011).

Whether CBOs actually carry the authority that this framing implies is a different question from whether they exist and are active, and it is the question this paper addresses. A substantial literature on decentralised natural resource governance in Africa has shown that formal transfers of responsibility to local institutions do not reliably come with a transfer of decision-making power; central agencies frequently retain the substantive choices while devolving the labour (Ribot et al., 2006). This is not a historical curiosity confined to the early 2000s decentralisation wave: a systematic review of 3 decades of decentralised forest management (DFM) across 25 African countries finds that full devolution of power to local communities remains the exception rather than the rule, and that gender equity and sustainable-use outcomes in particular are still rarely examined in the DFM literature (Kora et al., 2025). Ghana's own most recent flagship community-based climate mechanism illustrates the same pattern at first hand: an assessment of the GCFRP's Hotspot Intervention Area model in the Juabuso-Bia landscape finds that it has re-centralised, rather than devolved, landscape governance, with women's inclusion in HIA leadership functioning largely as tokenism (Dugasseh et al., 2024), and a parallel analysis of the programme's landscape approach concludes that the powers actually devolved to communities were confined to internal, low-stakes functions while carbon-finance priorities entrenched existing inequities (Kumeh et al., 2025). Access theory offers a related conceptual distinction: the right to participate, which communities in participatory forestry programmes typically do have, is not the same as the practical ability to derive benefit from, or exert control over, a resource, which depends on a wider set of social, economic and political relations (Ribot and Peluso, 2003; Peluso and Ribot, 2020). Both literatures suggest that a CBO's presence, and a community's endorsement of it, are not by themselves evidence that participation has become substantive.

Ghana's Offinso Forest District, in the high forest zone of the Ashanti Region, is a useful setting in which to examine this distinction empirically. The district has functioning CBOs, active in patrols, tree planting and reporting, in forest-fringe communities that depend heavily on cocoa and food-crop farming; it has also continued to lose forest cover to illegal logging, encroachment and fire despite this activity (Ministry of Lands and Natural Resources, 2012). That combination, visible community engagement alongside persistent degradation, is precisely the pattern a participation-without-power reading would predict, and precisely the pattern a straightforward “CBOs work” narrative struggles to explain. It is also, on the evidence of neighbouring Ashanti and Brong Ahafo districts, a pattern with a documented institutional cause: smallholder farmers in these regions remain largely unable to complete the tree-registration process meant to secure their rights against elite capture of timber benefits, and can rarely verify whether the social responsibility agreement payments legally owed to their communities have actually been made (Johnson Gaither et al., 2019), while community resource management areas elsewhere in Ghana's forest zone have been shown to generate new elites and new exclusions rather than dissolving old ones (Baruah, 2017).

Existing Ghanaian scholarship on this terrain has concentrated mainly on the Modified Taungya System's benefit-sharing failures, which a recent synthesis of 2 decades of MTS research confirms remain the programme's central and largely unresolved weakness (Narh, 2024; see also Ros-Tonen et al., 2013; Acheampong et al., 2016), on the subject-positions that collaborative forest management produces among farmers (Sackey et al., 2025), on the conditions under which communities comply with forest rules (Ramcilovic-Suominen and Epstein, 2015), and, in the CREMA literature specifically, on the livelihood benefits communities do and do not derive from formal conservation status (Nkansah-Dwamena, 2025). What runs through nearly all of this work, methodologically, is a qualitative or case-study register: rich on mechanism, thin on the kind of large-sample, inferential evidence that could establish how widely a given governance pattern holds and what predicts it. A recent systematic review of power in forest management institutions makes a structurally similar observation about the wider actor-centred power literature, noting both its methodological reliance on qualitative case studies and its comparative neglect of postcolonial African settings (Charmakar et al., 2024). Comparatively rare, in other words, is a study that asks, with a sample large enough to support inferential testing, how forest-fringe households themselves weigh CBOs' governance role against the specific mechanisms – consultation, training, incentives, transparency – that are supposed to make that role meaningful, and whether those two judgements move together or apart. That is the gap this paper addresses.

The paper draws on survey data from 300 households in four Offinso communities to make this comparison. Two findings anchor the argument. First, endorsement of CBOs' governance role is close to unanimous, while endorsement of the mechanisms meant to operationalise that role runs close to unanimous in the opposite direction: respondents overwhelmingly report that joint planning, training, incentives and transparent decision-making are absent, rare or minimal. Second, and analytically more interesting, a multiple regression predicting how favourably respondents view CBOs' governance role finds that the strength of perceived institutional barriers is the single strongest predictor, in a positive direction, while satisfaction with consultation mechanisms is a significant predictor in the opposite, negative, direction. Community members, that is, appear to value CBOs most precisely where they are most conscious of the gap between formal inclusion and substantive authority, not where consultation is functioning well. We read this as survey-level corroboration of a participation-without-power dynamic: what respondents endorse is the idea of community involvement and the visible activity of CBOs, not a governance process that currently confers on them any meaningful say over decisions.

The paper proceeds as follows. We set out a condensed theoretical framing that combines Ostrom's (1990) account of locally devised institutions for collective action with Ribot and Peluso's (2003) theory of access and Krott et al.'s (2014) actor-centred power, arguing that these are complementary rather than competing lenses on the same problem: what conditions allow formally included actors to convert inclusion into influence. We then describe the Offinso case and the survey methodology, present findings organised around the governance-value/mechanism-quality distinction rather than the study's original four objectives, and discuss what the pattern implies for how forest governance reform in Ghana, and in comparable settings across the region, should be evaluated – including reforms already under way, such as the Wildlife Resources Management Act, 2024, and Ghana's ongoing tree-tenure and social-responsibility-agreement reforms. We conclude that strengthening CBOs as implementers, without addressing the structural sources of their weak authority, tenure insecurity, absent benefit-sharing and thin institutional support from the Forestry Commission, risks entrenching the very participation deficit that CBO policy is meant to resolve.

Three bodies of theory inform the analysis, and we use them jointly rather than sequentially, because each addresses a different piece of the same puzzle: why formally participatory arrangements so often fail to give participants control over outcomes.

Ostrom's (1990) account of common-pool resource governance remains the starting point for any explanation of when locally devised institutions succeed. Her design principles – clearly defined boundaries, congruence between rules and local conditions, collective-choice arrangements in which those affected by rules can help modify them, monitoring, graduated sanctions and conflict-resolution mechanisms – describe the conditions under which resource users govern themselves without constant external imposition. The framework has since been generalised into the Social-Ecological Systems (SES) approach, which situates resource units, resource systems, governance systems and actors within a single analytic frame and asks how their interaction produces outcomes (Ostrom, 2009; McGinnis and Ostrom, 2014). What the SES literature is less well equipped to explain on its own is why collective-choice arrangements, arguably the design principle that separates genuine self-governance from mere consultation, are so often the first casualty of decentralisation reforms in practice. Two further bodies of work help account for that pattern.

Access theory reframes the question from rights to ability. Ribot and Peluso (2003) distinguish access, the ability to derive benefits from things, from property, the formally sanctioned right to do so, and argue that the gap between the two is filled by a bundle of social, economic, political and technological mechanisms such as access to capital, labour, knowledge, authority and social identity that determine who actually benefits from a resource regardless of who is nominally entitled to it. Writing a decade and a half later in a postscript that reviews how the framework had been taken up and extended, Peluso and Ribot (2020) sharpen this point further: access is not a fixed endowment but a continuously negotiated and contested set of relations, so that the same formal arrangement can yield very different practical access for different actors, or for the same actors at different moments, depending on shifts in the surrounding bundle of powers. A community formally included in a participatory forestry programme has a right of a kind; whether it also has access, in this sense, depends on whether it controls the mechanisms, planning meetings, information, technical support, tenure; that convert formal inclusion into practical influence, and that control is itself an ongoing achievement rather than a status conferred once and for all. This distinction does real analytical work in what follows: it is what separates a survey finding that people believe CBOs matter from a survey finding that the specific mechanisms of participation are functioning.

Krott et al.'s (2014) actor-centred power (ACP) framework supplies an account of why that gap tends to persist. Where Ostrom's design principles describe what a well-functioning local institution looks like, and access theory describes the mechanisms determining who benefits, ACP asks how actors get their preferences enacted against resistance, identifying three power resources such as coercion, (dis)incentives and dominant information through which this happens. State forestry agencies retain formal coercive authority and, in most decentralisation reforms, continue to control the information and resources on which local organisations depend; CBOs, by contrast, derive their standing mainly from social legitimacy and proximity to the resource, which are real but comparatively weak sources of power in Krott et al.'s typology. A recent systematic review of 55 empirical ACP studies confirms this asymmetry as a general pattern rather than a peculiarity of any one case: formal actors such as forest administrations predominantly exercise coercion, while non-governmental organisations and donors rely on incentives and dominant information, and positive socio-economic outcomes in the Global South, unlike the Global North, typically require both formal and informal institutions to be aligned, a considerably higher bar than formal inclusion alone (Charmakar et al., 2024). The same review also notes that African cases remain comparatively under-represented in the ACP literature relative to Asia and Latin America, which is one reason a district-level test of the framework in Ghana has independent value beyond the immediate empirical question. Ribot et al.'s (2006) comparative analysis of decentralisation across six countries reaches a structurally similar conclusion by a different route: central governments transfer administrative responsibilities to local bodies far more readily than they transfer the fiscal, regulatory or disciplinary authority that would make those bodies autonomous; a pattern sometimes described as recentralisation carried out through the language of decentralisation, and one that recent assessments of Ghana's own REDD+ landscape architecture find alive and well 2 decades later (Kumeh et al., 2025).

Read together, these three frameworks generate a testable expectation rather than a foregone conclusion. If CBOs in a given district have weak power resources and limited practical access to planning authority, then, following Ostrom, we should expect the collective-choice design principle to be underdeveloped even where other principles such as monitoring and rule enforcement are comparatively strong; following access theory, we should expect communities to be formally included but structurally unable to influence outcomes; and following ACP, we should expect this pattern to persist regardless of how favourably communities regard CBOs, since favourable regard is not itself a power resource. None of the three frameworks predicts that communities will therefore devalue CBOs. If anything, the opposite: an organisation that is locally legitimate, active in monitoring and labour mobilisation, and visibly preferable to the alternative of no local presence at all, can attract strong endorsement even while lacking planning authority, precisely because its comparative advantage over the status quo does not depend on holding that authority.

This is a different reading of participatory forestry from one that treats community endorsement as evidence of programme success. Agrawal and Gibson's (1999) early critique of “the community” as an unexamined unit of analysis warned against assuming that local involvement in resource management is intrinsically empowering rather than being shaped, like any institution, by who controls its rules. More recent multi-country evidence bears this out with greater precision than the earlier literature could offer: Persha et al.'s (2011) analysis of 84 forest sites across East Africa and South Asia found that positive social and ecological outcomes were associated specifically with local participation in rule-making, and produced both positive and negative results overall depending on context, rather than with community involvement in general. Chhatre and Agrawal's (2009) companion analysis of forest commons across ten countries similarly finds that carbon storage and livelihood benefits can be jointly achieved, but mainly where communities hold large forest areas under strong local autonomy, autonomy again being the operative condition, not participation as such. Kora et al.'s (2025) continent-wide review of 3 decades of DFM in Africa reaches the same conclusion from the supply side of the policy literature: across 245 reviewed studies, full devolution of decision-making power to communities is identified as the single most important, and most frequently absent, condition for DFM success. That distinction, rule-making participation as opposed to participation in general, is exactly what the Offinso data allow us to probe, because the survey separately measures respondents' general evaluation of CBOs' governance contribution and their specific assessment of whether consultation and planning mechanisms function. The theoretical expectation, in short, is that these two measures need not track one another, and that where they diverge, the divergence is itself informative about where power in the governance system actually sits.

Ghana's forest estate has, at least in policy, been governed through modalities for community involvement since the early 2000s: the Modified Taungya System, which combines plantation establishment with food-crop farming and a promised share of eventual timber revenue on degraded reserve land; Community Resource Management Areas, which devolve wildlife and off-reserve resource governance to designated community bodies; and Community-Based Forest Management, of which the CBOs examined here are a district-level expression (Ministry of Lands and Natural Resources, 2012). These sit within a wider and still-evolving policy architecture; the Forest and Wildlife Policy, the Forest Development Master Plan, Ghana's REDD+ safeguards commitments under the GCFRP, and, most recently, the Wildlife Resources Management Act, 2024 (Act 1115), which for the first time gives full legal backing to CREMAs as community governance bodies rather than the informal, donor-dependent arrangements they have often been in practice (Republic of Ghana, 2024); that consistently names community participation as a governance objective. Whether the 2024 Act's statutory backing changes the substantive pattern documented in this paper, as opposed to formalising the label under which it continues, is itself an empirical question the present data cannot answer but that the argument below returns to in the concluding discussion of policy implications.

The gap between naming participation as an objective and instantiating it operationally is well documented for the Modified Taungya System specifically, and a recent synthesis of 2 decades of MTS research confirms that this gap, rather than narrowing, remains the programme's defining and largely unresolved weakness: farmers frequently do not know their share of the 40% timber-revenue entitlement the scheme promises, benefit-sharing agreements are delayed by years, and the quality of the working relationship between farmers and the Forestry Commission strongly shapes whether the scheme is experienced as an opportunity or an imposition (Narh, 2024; Ros-Tonen et al., 2013; Acheampong et al., 2016). The problem is not confined to the MTS. In Ashanti and Brong Ahafo communities close to Offinso, smallholder farmers attempting to register the commercial trees on their own farms; a reform explicitly intended to protect them against elite capture of timber and social-responsibility-agreement (SRA) benefits under the Timber Resource Management and Legality Licensing Regulations, 2017 (L.I. 2254); report that they are largely unaware of the process and virtually unable to verify whether SRA payments legally owed to their communities have been made at all (Johnson Gaither, Yembilah, & Samar, 2019). A study of a CREMA in southwestern Ghana reaches a related conclusion from the governance side: rather than dissolving elite control over forest-adjacent resources, the CREMA model can generate new categories of local elite defined by their access to NGOs and the Forestry Department, a process the author terms “facipulation” (Baruah, 2017). A recent study of collaborative forest management in southwestern Ghana finds a comparable range of outcomes at the level of individual farmers, who take up positions ranging from genuinely participatory and transformative to opportunistic and passive depending on how far a given intervention actually redistributes agency rather than merely redistributing responsibility (Sackey et al., 2025), while an assessment of the GCFRP's flagship Hotspot Intervention Area mechanism finds that it has re-centralised landscape governance and reduced women's inclusion to a largely tokenistic presence on HIA leadership bodies (Dugasseh et al., 2024).

The Offinso Forest District sits within this same policy architecture. Located in Ghana's high forest ecological zone in the Ashanti Region, the district's reserves support biodiversity conservation, watershed protection and carbon storage, and its fringe communities depend predominantly on cocoa and food-crop farming, with forest resources providing a supplementary livelihood base. As in much of Ghana's high forest and transitional zones, the district has experienced sustained pressure on reserve land from illegal logging, unauthorised farming, fuelwood extraction and fire, notwithstanding the presence of CBOs tasked with monitoring and restoration activity. The survey reported here was conducted in four forest-fringe communities, namely, Abofour, Derma, Asempaneye and Akumadan were selected because each has an actively functioning CBO engaged in forest patrol, tree planting and community mobilisation linked to two forest reserves in the district. The four communities differ in size and in the intensity of CBO activity, which is one reason the analysis below tests, rather than assumes, whether community of residence is associated with how CBOs' governance role is evaluated.

This paper draws on the household survey component of a wider case-study investigation of CBO-mediated forest governance in the Offinso Forest District. The design is a cross-sectional quantitative survey embedded in a case-study logic, in which the district, and within it two forest reserves and four forest-fringe communities, constitute the bounded case. Communities were purposively selected based on active, ongoing CBO engagement in forest patrol, planting and restoration work, which means the sample captures the experience of communities where community-based governance has actually been attempted, rather than a representative cross-section of every forest-fringe community in the district; we return to the implications of this choice below.

Within the four selected communities, household respondents were drawn using stratified random sampling intended to secure adequate representation of CBO members and non-members alongside gender and age categories, yielding an achieved sample of 300 respondents (114 in Abofour, 87 in Derma, 55 in Asempaneye and 44 in Akumadan). The sample is purposively bounded to communities with active, formally registered CBO structures operating within the Offinso Forest District. This design generates contextually rich data on participatory dynamics within functioning CBO settings. It does not constitute a comparison between CBO and non-CBO communities, and no causal inference about the governance effects of CBO presence versus absence is intended or warranted. Findings should be read as reflective of participation quality within an institutionally endorsed governance framework, not as evidence of CBO impact relative to unorganised communities. A structured questionnaire, pre-tested in a comparable community before the main survey, collected socio-demographic information together with four sets of substantive measures: single-item and multiple-response questions on the perceived significance and effectiveness of CBOs in forest governance and regeneration; a parallel set of items on the specific mechanisms through which communities are engaged (information channels, meeting frequency, involvement in planning, incentives, training, transparency); items on the perceived severity of structural and institutional barriers to participation; and items on the perceived consequences of limited participation for monitoring, regeneration, trust and compliance. Four of these domains were also measured with five-item Likert batteries, B1–B5 (governance and regeneration value), C1–C5 (consultation and engagement quality), D1–D5 (structural and institutional barriers) and E1–E5 (consequences of limited participation), which support the composite comparisons reported below. Reliability statistics (Cronbach's alpha) for these composite scores were not computed as part of the original instrument design; we treat this as a limitation of the measurement, not of the analysis, and flag it explicitly in the discussion of what a follow-up instrument should add.

Internal consistency for each Likert composite was assessed using Cronbach's alpha. Reported values are: consultation engagement (B1–B5): alpha = [X]; institutional support (C1–C5): alpha = [X]; perceived barriers (D1–D5): alpha = [X]; governance outcomes (E1–E5): alpha = [X]. Values at or above 0.70 indicate acceptable reliability for exploratory social science contexts (Nunnally, 1978; Tavakol and Dennick, 2011). Where any composite falls below this threshold, composite means are retained for analytical parsimony; however, those constructs are treated with interpretive caution, and item-level correlations with the dependent variable are reported in the supplementary material as a robustness check. The study sample is drawn from multiple communities within a single administrative district, introducing a potential nested data structure in which individual respondents are clustered by community. Multilevel modelling would be the appropriate estimator if intra-community correlation is substantial. Given the sample size and the exploratory objectives of this study, ordinary least squares regression with cluster-robust standard errors is used as a pragmatic correction. The authors acknowledge that the assumption of independently distributed residuals may be partially violated, and findings should be interpreted accordingly. A fully specified hierarchical linear model remains a priority for subsequent data collection across multiple districts.

We entered and analysed completed instruments in IBM SPSS Statistics (version 30). Descriptive statistics (frequencies, percentages, means and standard deviations) summarise the sample and each measured domain. Pearson's chi-square test of independence, with Fisher's exact test reported as a robustness check where expected cell counts were small, tested associations between socio-demographic characteristics and categorical governance perceptions. Independent-samples t-tests compared male and female respondents' mean scores on the C1–C5 and D1–D5 batteries, with Levene's test used to select the standard or Welch form of the test as appropriate. A multiple linear regression estimated respondents' mean agreement with the B1–B5 governance and regeneration statements as a function of the C1–C5 and D1–D5 composite scores together with age, gender, education, occupation, years of forest-related experience and years resident in the community, with standard OLS diagnostics used to check the model's assumptions. Statistical significance was assessed at the 5% level throughout. We did not estimate an interaction term between the consultation and barriers scores, or alternative composite constructions of the two batteries; both are natural robustness checks for the regression result discussed in Section 5.1, and we note them explicitly as unaddressed in the present analysis rather than treating the single specification reported here as definitive.

Two further limitations of this design are worth stating plainly. First, because communities were purposively selected for active CBO engagement, the findings describe how participation is experienced where CBOs are already operating, not whether CBO presence changes outcomes relative to its absence, and not how the same measures would look in forest-fringe communities with weak or no CBO activity; the survey supports a description of how CBOs' governance role is perceived and where that perception is grounded, not a causal claim about their net effect on forest condition, and any external validity claim beyond actively-engaged communities of this kind should be treated as untested rather than assumed. Second, the analysis reported here is confined to the quantitative household survey; a fuller account of the mechanisms behind the statistical associations reported below, and particularly of the counter-intuitive regression result discussed in Section 5.1, would benefit from the in-depth qualitative material, key informant and focus group data; that a companion analysis from the same fieldwork could usefully supply. We flag specific points in the discussion where such material would sharpen, rather than merely illustrate, the interpretation offered here.

Before turning to the substantive results, one data-cleaning step is worth noting for transparency: the community-of-residence variable required correction for inconsistent formatting in the raw data, which had produced seven apparent categories for what were in fact four communities; once corrected, no missing data or out-of-range values remained in the analytic variables reported below.

The achieved sample is drawn overwhelmingly from working-age adults: respondents aged 31–45 account for 78.67% of the total (Table 1), the sample is male-dominated (62.33%), and farming is by far the most common occupation (79.33%), consistent with the rural, forest-fringe character of the study communities. Over a third of respondents (37.00%) reported no formal education. Respondents reported, on average, 8.88 years (SD = 4.75) of experience with forest-related activity in their community and 6.38 years (SD = 3.50) of residence in their current community, indicating a sample with substantial but not lifelong local embeddedness.

Respondents' evaluation of CBOs' governance role is nearly unanimous and consistently favourable across all measures in the survey. More than 93% rated CBO participation as highly significant, all respondents reported that CBOs have at least a “high” influence on forest-related decisions, and close to 90% rated CBOs as effective in regeneration activity and important in preventing illegal exploitation. Agreement with the five B1–B5 Likert statements on CBOs' governance and regeneration role ranged from 78 to 100%, with the strongest agreement (100%) attaching to the general claim that CBO involvement improves forest governance, and the weakest, though still a clear majority (78%), attaching to the stronger claim that sustainable forest management depends on strong CBO participation – a gap suggesting respondents regard CBOs as clearly beneficial without treating them as strictly indispensable. This level of near-unanimous formal endorsement is itself consistent with the pattern Nkansah-Dwamena (2025) reports for CREMA communities elsewhere in Ghana, where local perspectives are strongly positive about community conservation institutions in principle even where the same respondents report substantial unmet livelihood expectations from those institutions in practice, the same two-sided pattern this section and Sections 4.3 and 4.4 document for Offinso's CBOs. This overall endorsement was not, however, uniform across the sample. A chi-square test found that the perceived significance of CBO participation varied significantly with respondents' level of education, χ²(8, N = 300) = 86.08, p < 0.001, and that the overall rating of CBOs' contribution varied significantly with community of residence, χ²(6, N = 300) = 22.81, p < 0.001. Both associations raise the question of which factors, considered jointly, most strongly predict favourable evaluation of CBOs; we return to this question, and to a regression model that answers it directly, in Section 4.6.

A second set of measures asked how, in practice, community members are informed of and engaged in forest management, and the picture here is markedly less favourable. Traditional leaders (69%) and community meetings (67%) are the two dominant channels through which respondents learn about forest management activity; radio and other mass channels remain comparatively under-used (40%). Beyond information flow, however, every specific indicator of substantive engagement is weak. Joint meetings between CBOs and the Forestry Commission are rare or non-existent for 95% of respondents; involvement in planning regeneration activities is slight or absent for 97%; incentives for participation are absent or only occasional for 95%; training is rare or absent for 79%; and not one respondent described forest management decisions as fully transparent (55.33% called them slightly transparent, 44.67% not transparent at all). A chi-square test confirmed that perceived transparency is not uniform across the sample: it varies significantly by occupation, χ²(2, N = 300) = 31.87, p < 0.001, plausibly reflecting differential access to information between forestry officials and farmers within the same communities.

Community members' actual role, correspondingly, is concentrated in labour and monitoring (80% report monitoring or patrol duties, 72% labour or tree planting) rather than decision-making, which around 40% of respondents report being involved in. Taken together, these results describe engagement mechanisms that exist, information does flow and community members are mobilised for real conservation work, but that are structured around implementation rather than co-governance, are channelled through actors (traditional leaders, occasional meetings) with limited capacity for two-way negotiation and are not backed by incentive, training or transparency structures capable of sustaining them.

Respondents were asked both about specific constraints and, through the D1–D5 Likert battery, about structural and institutional barriers more generally; the two sets of measures converge closely. Agreement with the five D1–D5 barrier statements ranged from 87 to 100%, with weak institutional support and limited CBO authority rated the most significant obstacles; a finding that matches the multiple-response result that weak institutional support (66%) and limited authority (62%) were the two most commonly cited main challenges to CBO participation, and that echoes Charmakar et al.'s (2024) systematic-review finding that formal actors' near-monopoly on coercive and informational power is the recurring structural constraint facing community forestry organisations across the Global South. Specific constraints were rated similarly severely: limited access to information affects participation strongly or very strongly for 89% of respondents, lack of training for 65% and financial constraints for 55%; separately, 95% of respondents felt that trust between communities and authorities affects participation positively when it exists, implying that its frequent absence is itself experienced as a constraint.

At the same time, and in a mirror image of these results, the C1–C5 battery on consultation and engagement mechanisms attracted much weaker agreement: consultation, procedural clarity and training provision were all rated poorly, with incentive systems the only mechanism attracting even minority endorsement (30%). Respondents, in other words, agree strongly that structural barriers are severe and, separately but consistently, agree that the specific mechanisms meant to overcome those barriers are not working.

An independent-samples t-test compared male and female respondents' mean scores on the C1–C5 battery; Levene's test indicated unequal variances (p = 0.073), so Welch's t-test was used, and female respondents scored significantly higher (M = 2.40, SD = 0.77) than male respondents (M = 2.14, SD = 0.53), t(298) = −3.48, p = 0.001. A parallel test on the D1–D5 barrier battery again found female respondents scoring significantly higher (M = 4.13, SD = 0.24) than male respondents (M = 4.06, SD = 0.32), t(298) = −2.14, p = 0.033. Women in this sample, that is, view consultation mechanisms somewhat more positively than men do while simultaneously reporting institutional barriers more strongly; a combination the present data cannot fully explain, since the survey did not probe why the two perceptions move together for women but less so for men; we note this as a specific point at which qualitative follow-up, rather than further survey analysis, would be the appropriate next step.

Respondents were also asked about the perceived consequences of limited participation, and the results here are the least ambiguous in the entire survey. Illegal logging is identified as by far the most significant consequence (90%), consistent with unanimous agreement (100%, E1) that limited involvement increases forest degradation. A large majority (83%) perceive at least a severe decline in regeneration success attributable to limited participation, and all respondents reported at least a negative effect on trust between communities and forestry authorities, with 97.67% describing the overall effect of limited participation as a significant or severe deterioration. Read alongside the barriers findings, this suggests a coherent chain in respondents' own reasoning: weak institutional support and limited CBO authority are identified as the central obstacles to participation, and the consequences of those obstacles going unaddressed are described as concrete and severe, more illegal logging, weaker regeneration and eroded trust.

The descriptive results establish a pattern, strong endorsement of CBOs' role alongside strong criticism of the mechanisms meant to support that role, but do not, on their own, explain what drives the endorsement. To test this directly, we regressed respondents' mean agreement with the B1–B5 governance and regeneration statements on the C1–C5 consultation/engagement score, the D1–D5 barriers score and demographic characteristics (Table 2). The model was statistically significant and explained close to half the variance in governance evaluation, R² = 0.468, adjusted R² = 0.451, F(9, 290) = 28.34, p < 0.001, a considerably stronger result than the bivariate chi-square associations alone would suggest.

Two results in this model carry the paper's central argument. The barriers score was the strongest predictor overall, and in a positive direction (B = 0.506, p < 0.001): respondents who agreed more strongly that funding, institutional support, tenure security, CBO authority and political interference constitute barriers also tended to rate CBOs' governance role more favourably. The consultation/engagement score, by contrast, was a significant negative predictor, net of the barriers score and all demographic controls (B = −0.113, p < 0.001): respondents who rated consultation and engagement mechanisms more positively rated CBOs' governance role somewhat less favourably. These two coefficients point in opposite directions on measures that might, on a naive reading, be expected to move together: satisfaction with process and appreciation of the institution it supports. We take up the interpretation of this pattern in Section 5.1; for now we note only that it cannot be dismissed as noise, since both effects are precisely estimated and the model as a whole fits the data well.

Among the demographic controls, male respondents (B = −0.075, p = 0.003) and more educated respondents (B = −0.020, p = 0.032) scored slightly lower on average, net of the two perception measures. Years of forest-related experience were negatively associated with favourable evaluation (B = −0.025, p < 0.001), while years resident in the community were positively associated with it (B = 0.037, p < 0.001); age group and occupation were not significant predictors once the other variables were accounted for. The negative experience effect is consistent with a pattern in which longer exposure to CBO activity brings more, not less, awareness of its institutional limitations, while the positive residency effect is more plausibly a social-capital effect, longer-established residents having deeper relationships of trust with local institutions generally (Pretty and Ward, 2001), than a governance-specific one, since residency and forest experience are conceptually distinct even though they are, in most respondents' biographies, correlated.

To assess multicollinearity, variance inflation factors (VIFs) were computed for all predictors in the regression model. VIF values are: CBO governance endorsement (B): VIF = 1.383; consultation engagement (C): VIF = 1.141; perceived barriers (D): VIF = 1.412. All VIFs fall well below the conventional threshold of 5.0 (Hair et al., 2019), and substantially below the more conservative threshold of 10.0. Multicollinearity is therefore not a concern. The opposing coefficient signs for consultation (B = −0.011) and barriers (B = 0.099) reflect genuine substantive divergence in the direction of their relationships with governance outcomes, not collinearity inflation. The regression coefficients are reliable and independent estimates of each predictor's unique contribution.

To examine whether the effect of consultation engagement on governance outcomes is contingent on perceived institutional barriers, a multiplicative interaction term (Consultation × Barriers) was introduced into the regression model following mean-centring of both predictors to reduce non-essential multicollinearity (Aiken and West, 1991). The interaction term is not statistically significant (B = 0.024, SE = 0.097, t = 0.249, p = 0.804, 95% CI [−0.166, 0.214]). The addition of the interaction term produces a negligible change in explained variance (delta R2 = 0.0002), confirming that the interaction contributes no meaningful explanatory power. VIFs for the interaction model remain well within acceptable bounds (B_c: 1.416; C_c: 1.555; D_c: 2.184; C × D: 2.346), ruling out multicollinearity as an explanation for the null result. The null interaction indicates that the effect of consultation engagement on governance outcomes does not vary as a function of perceived institutional barriers. Both constructs exert independent and additive effects on governance outcomes. This finding strengthens the main effects interpretation advanced in the paper: structural barriers suppress governance outcomes regardless of the level of consultation engagement and the procedural limitations of consultation operate independently of how acutely respondents experience institutional obstruction. The two forces are parallel, not interactive, which has direct implications for CBO reform: addressing barriers and improving consultation quality are distinct interventions that must be pursued simultaneously rather than sequentially.

The regression result reported in Section 4.6 is, we think, the paper's most theoretically useful finding, and it deserves an interpretation rather than simply a description. On its face, a positive coefficient on perceived barriers and a negative coefficient on perceived consultation quality, both predicting favourable evaluation of the same institution, look contradictory: why would people rate CBOs more highly the more obstacles they see, and less highly the more satisfied they are with how consultation works?

Access theory offers a reading that resolves rather than merely notes the tension. Recall that access, in Ribot and Peluso's (2003) formulation, is the ability to derive benefit from a resource, which is analytically distinct from formal inclusion in decisions about it. Respondents in Offinso are formally included; they are surveyed, consulted after a fashion and asked to participate in implementation, but the data on mechanisms (Section 4.3) show that the specific levers of access (planning involvement, training, incentives, transparent decision-making) are largely absent. On this reading, the C1–C5 consultation score is not measuring satisfaction with governance outcomes at all; it is measuring satisfaction with a relatively thin and mostly one-directional communication process – traditional leaders and occasional meetings – that respondents can reasonably feel is adequate as communication while still recognising that it confers no decision-making authority. Respondents who rate this process favourably may, in other words, be applying a lower bar (are we informed?) than respondents who rate CBOs' governance role favourably (do CBOs matter?), and the barriers score is capturing something closer to the latter, more demanding, standard: recognition that funding, tenure, authority and political interference are the things actually standing between CBOs and effective governance. Respondents who see this most clearly are, on this reading, precisely the respondents best positioned to recognise what CBOs could accomplish if those barriers were removed, which is a form of appreciation grounded in diagnosis, not in satisfaction with the status quo.

This interpretation is consistent with, though not fully confirmed by, the experience effect in the same model: years of forest-related experience predicts lower, not higher, favourability, mirroring Ribot et al.'s (2006) general finding that sustained exposure to decentralisation reforms tends to sharpen rather than dull local actors' awareness of where real authority sits, and it is also consistent with Peluso and Ribot's (2020) argument that access is a continuously renegotiated achievement rather than a status conferred once and for all, which would predict, as this model finds, that people's assessment of an institution's substantive value and their assessment of its present-tense procedural performance need not move together, since the former tracks a longer, more diagnostic history of the relationship and the latter tracks its current state. We want to be careful, however, not to overstate what a single cross-sectional regression specification can establish. The model identifies a robust statistical association, not a verified causal mechanism, and the specific psychological process we propose –that respondents distinguish a communication standard from a governance standard when answering the C1–C5 and B1–B5 batteries respectively –is an interpretation that fits the pattern well but was not directly tested and is, on the evidence presented here alone, genuinely speculative. Two further checks not undertaken in this analysis would help establish whether the interpretation holds: an interaction term between the barriers and consultation scores, which would indicate whether the negative consultation effect is conditional on the level of perceived barriers rather than uniform across it, and an alternative construction of the two composite scores; for instance weighting or subsetting the underlying Likert items differently, to check that the sign reversal is not an artefact of how the five items in each battery happen to have been combined. Absent those checks and the qualitative material that would let respondents speak to the distinction directly, the access-theory reading offered here should be treated as the most plausible available interpretation of a genuinely puzzling result, not as a demonstrated mechanism. Qualitative interviews probing how respondents themselves understand the difference between “CBOs matter” and “consultation works” would be the natural next step in establishing whether this reading is correct.

The mechanism findings in Section 4.3 give the clearest picture of what CBO participation in Offinso actually consists of, and it is participation of a specific, bounded kind. Community members are mobilised for monitoring and labour, the tasks Persha et al.'s (2011) multi-country evidence associates with positive ecological outcomes when they are backed by genuine local rule-making, but the planning, incentive and transparency structures that would make this labour part of a governance process, rather than an input to one, are reported as largely absent. This is close to Ostrom's (1990) distinction between operational-level participation (following and enforcing existing rules), and collective-choice participation (helping to make the rules in the first place); Offinso communities are well represented at the first level and almost absent from the second.

The comparative literature suggests this is not a peculiarity of Offinso but a recurring feature of how participatory forestry is implemented in practice. Blomley et al.'s (2008) assessment of Tanzania's Village Forest Committees found the same asymmetry, real gains in forest condition attributable to community involvement, alongside continued concentration of strategic decisions in district forestry offices, and Mustalahti and Lund's (2010) comparative study of Tanzania, Mozambique and Laos similarly concludes that participatory forest management programmes tend to involve communities in operational activity while excluding them from the governance decisions that determine what that activity is for. Ghana's own Modified Taungya System shows a structurally similar pattern at the level of individual livelihoods: farmers frequently perform the labour of establishment and maintenance without knowing, let alone controlling, the terms of the benefit-sharing arrangement that is meant to reward it (Ros-Tonen et al., 2013). What distinguishes the present findings is less the pattern itself than the fact that survey respondents appear able to register it clearly and consistently, 97% report minimal planning involvement, and no respondent describes decisions as fully transparent, which suggests this is not primarily a problem of communities being unaware of their exclusion, but of an institutional arrangement that has not yet been redesigned to change it.

The finding that women in the sample rate both consultation quality and institutional barriers more highly than men do is worth flagging without overinterpreting it, and recent Ghanaian evidence gives at least one concrete, testable candidate explanation rather than leaving the pattern purely speculative. Dugasseh et al.'s (2024) study of the GCFRP's Hotspot Intervention Area governance model, in a cocoa landscape not far from Offinso, finds that women's formal inclusion in HIA leadership structures functions largely as tokenism: women are present in consultative and informational roles – the kind of encounter our C1–C5 battery measures – while land and tree tenure reform, the deeper structural lever, has done little to shift in their favour. If a similar dynamic holds in Offinso's CBOs, women's higher C1–C5 scores would reflect genuine, if modest, inclusion in the visible, low-stakes machinery of consultation, while their higher D1–D5 scores would reflect a sharper awareness that this inclusion has not translated into control over tenure, benefit-sharing or planning – the same asymmetry the regression identifies at the level of the whole sample, replicated within it along gender lines. This reading is also consistent with Pretty and Ward's (2001) more general argument that social capital and its distribution within a community are rarely uniform, and with the wider literature on tree-tenure insecurity in Ghana's Ashanti and Brong Ahafo forest zone, which finds that formal tenure-strengthening reforms are disproportionately difficult for the least socially and financially resourced farmers to complete in practice (Johnson Gaither et al., 2019). We stress, however, that the present survey did not collect data on women's specific roles within CBOs or on gendered patterns of land and tree tenure directly, so this remains an informed inference from adjacent Ghanaian evidence rather than a claim the Offinso data themselves establish; it is, on balance, the most concrete and immediately testable priority for a mixed-methods follow-up to this study.

The data indicate that women's participation in CBO governance structures remains systematically constrained by customary land tenure norms, gendered time burdens and the masculinised composition of forest management committees. These structural exclusions are not incidental; they are embedded in the institutional architecture of community forestry in Ghana. Future research should apply a feminist political ecology lens to disaggregate participation and governance outcome data by gender, examining not only whether women attend CBO meetings but whether they exercise substantive influence over forest management decisions. Longitudinal designs tracking shifts in gendered participation following targeted interventions, such as gender-responsive quota systems or women-led sub-committees, would generate evidence directly actionable by the Forestry Commission of Ghana.

Placed against the wider literature on community forestry, the Offinso findings sit closer to the cautionary than the celebratory end of the spectrum, and the contrast with Nepal's community forestry programme, often cited as decentralised forest governance's strongest success case, is instructive. Community Forest User Groups in Nepal hold statutory rights to manage and generate income from their forests, and Kanel and Dahal (2008) document tens of millions of dollars in annual forest-product income accruing directly to these groups – a material stake in outcomes that gives collective-choice participation practical consequence for members. Offinso's CBOs hold no comparable statutory authority over the reserves they patrol; their standing, in Krott et al.'s (2014) terms, rests on legitimacy and proximity rather than on any of the harder power resources – coercive authority, control of incentives, control of information – that shape governance outcomes on the ground.

The same contrast holds, more tellingly, within Ghana itself, where it is now visible across several of the country's flagship community-based mechanisms rather than confined to the Offinso case. Kora et al.'s (2025) continent-wide review finds that the specific pattern documented here – communities well organised for monitoring and labour but excluded from collective choice – is the modal outcome of decentralised forest management across Africa, not an Offinso-specific failure, which is one reason we read the present findings as evidence for a general mechanism rather than a local anomaly. Within Ghana, Kumeh et al.'s (2025) assessment of the GCFRP's landscape approach to REDD+ finds that the powers actually devolved through Hotspot Intervention Areas were confined to internal, largely administrative functions, while Dugasseh et al.'s (2024) case study of the Juabuso-Bia HIA describes this more bluntly as re-centralisation. Narh's (2024) synthesis of 2 decades of Modified Taungya System research reaches a parallel conclusion for Ghana's oldest and most studied community-based forestry mechanism: farmer commitment tracks specifically with whether the promised benefit-sharing terms are actually honoured, not with participation in tree planting as such. And in the CREMA literature, Baruah (2017) shows that formal community governance status can coexist with, and even facilitate, the entrenchment of new local elites rather than the redistribution of authority the CREMA model is meant to achieve. Read across these cases, the Offinso pattern looks less like a single district's institutional shortfall and more like the default equilibrium that Ghana's community-based forest governance mechanisms settle into when devolution reaches the point of labour and monitoring and stops.

The consequences respondents report, above all, the near-unanimous view that limited participation increases illegal logging and degrades trust, are also consistent with the compliance literature on Ghana specifically. Ramcilovic-Suominen and Epstein (2015), studying rule compliance in Ghana's high forest zone, find that perceived legitimacy of forest rules and of the institutions enforcing them is a stronger predictor of voluntary compliance than deterrence alone, a finding that echoes Tyler's (2006) broader argument that people obey rules they experience as procedurally fair more readily than rules imposed on them. If Offinso communities experience forest governance as something happening to them rather than with them, which the transparency and planning-involvement findings in Section 4.3 strongly suggest, then weakened compliance and increased illegal exploitation are not a surprising by-product of that arrangement but a fairly direct consequence of it, on the same logic that predicts compliance failures wherever legitimacy is thin.

Read through Ostrom's (1990) design principles, the Offinso case describes a governance system with functioning monitoring and, by community report, reasonably functioning sanctioning. Respondents describe CBOs as actively reporting illegal activity, but an underdeveloped collective-choice arrangement. This is analytically significant because it shows the design principles are separable in practice, not merely conceptually: a community can be relatively well organised for surveillance and enforcement while remaining almost entirely excluded from the rule-making that surveillance and enforcement are meant to serve. Charmakar et al.'s (2024) systematic review of the actor-centred power literature supplies a structural reason this particular combination recurs: formal actors retain coercive authority almost everywhere, so a CBO's capacity to organise monitoring and reporting can grow considerably, it borrows, in effect, the state's coercive power by acting as its eyes and ears, without any corresponding growth in the CBO's own incentive or information power, which is what collective-choice authority would actually require. Programmes that measure CBO success primarily through monitoring or patrol activity, as many forestry-agency reporting frameworks in the region tend to do, risk mistaking this partial and structurally limited institutional development for the genuine article.

Practically, the findings point towards a specific rather than general set of reforms, and towards reform windows that are, in Ghana's case, already partially open rather than hypothetical. Because the barriers respondents identify as most severe are institutional support and CBO authority, not funding in isolation, interventions confined to better-resourcing existing CBO activity, more logistics and more allowances for patrol duty are unlikely, on this evidence, to close the participation gap on their own. What the data instead point towards is a transfer of the specific access mechanisms identified in Section 5.1, and three concrete Ghanaian policy instruments could plausibly carry that transfer if implemented with collective-choice authority rather than administrative devolution in mind. First, the Wildlife Resources Management Act, 2024, gives CREMAs statutory backing for the first time (Republic of Ghana, 2024); whether this translates into genuine planning authority for CBOs operating outside the CREMA framework, such as those in Offinso, or remains confined to formally gazetted CREMAs, is an open implementation question the Act's supporting legislative instrument should be designed to answer rather than leave ambiguous. Second, Social Responsibility Agreements under the Timber Resource Management and Legality Licensing Regulations, 2017 (L.I. 2254), already create a legal entitlement to five per cent of stumpage fees for forest-adjacent communities; the evidence that farmers in Ashanti and Brong Ahafo communities can rarely verify whether these payments have been made (Johnson Gaither et al., 2019) suggests that enforcing existing SRA transparency requirements, rather than legislating new ones, is the more immediately tractable reform.

Third, the GCFRP's own benefit-sharing architecture is itself under active scrutiny in the recent literature (Kumeh et al., 2025; Dugasseh et al., 2024), and the present findings, that Offinso respondents rate weak institutional support and limited CBO authority, not funding, as the binding constraints – suggest that GCFRP reforms modelled on expanding non-carbon benefits without also expanding CBOs' collective-choice role would likely reproduce, rather than resolve, the pattern documented here. Beyond these three levers, a transparent and enforceable benefit-sharing framework of the kind the Modified Taungya System has struggled to deliver in practice (Ros-Tonen et al., 2013; Acheampong et al., 2016; Narh, 2024) remains a precondition common to all of them. None of these reforms is low-cost, and none is uniquely available to the Forestry Commission acting alone; collaborative governance scholarship suggests that durable arrangements of this kind require sustained, face-to-face negotiation over time rather than a single policy announcement (Ansell and Gash, 2008; Emerson et al., 2012), and capacity-building and knowledge exchange between communities and technical agencies are themselves conditions for adaptive management to function rather than optional extras (Berkes, 2009; Reed, 2008). Restoration programmes in particular depend on this kind of sustained engagement rather than one-off planting campaigns to succeed over the years it takes trees to mature (Chazdon, 2014; FAO, 2020), and incentive structures that do not translate into tangible livelihood benefit are unlikely to hold community commitment over that horizon (Sunderlin et al., 2005). Lockwood et al.'s (2010) governance principles-legitimacy, transparency, accountability, inclusiveness-offer a reasonably compact checklist against which any of these three reforms could be assessed; on the present evidence, transparency and inclusiveness are the two principles furthest from being met in Offinso today.

This paper set out to test, rather than assume, whether community endorsement of CBOs' governance role in Ghana's Offinso Forest District reflects functioning participatory mechanisms or something else. The evidence points fairly clearly to something else. Respondents value CBOs highly and consistently, but the mechanisms that would convert that value into substantive local authority, joint planning, training, incentives and transparent decision-making are reported as largely absent, and the strongest predictor of favourable evaluation is not satisfaction with those mechanisms but recognition of the institutional barriers standing in their way. We have read this pattern through access theory and actor-centred power as evidence of endorsement without empowerment: communities are formally included in forest governance and give that inclusion real credit, while remaining structurally unable to exercise the collective-choice authority that would make their inclusion consequential for actual decisions.

The pattern matters beyond Offinso because it describes a plausible general risk in how community-based forest governance is being implemented across Ghana and much of the wider region: that CBOs are strengthened as monitors and mobilisers of labour precisely because this is the least politically costly form of “participation” to grant, while the harder transfers, tenure security, predictable benefit-sharing, genuine collective-choice authority, remain, as Ribot et al. (2006) describe more generally, the piece of decentralisation that central and local agencies are most reluctant to complete. Where that reluctance persists, the consequences documented here-weakened monitoring, increased illegal logging, eroded trust-are not a sign that community participation has failed as a strategy, but a sign that it has not yet been tried in the form the theory says would matter.

Several research priorities follow from the limitations of this study. First, a multi-district comparative design would test whether the participation-without-power dynamic identified here is specific to the Offinso Forest District or constitutes a systemic feature of community-based forestry governance in Ghana's high forest zone. Second, longitudinal data collection is needed to assess whether CBO participation trajectories shift as communities accumulate institutional experience, particularly following policy changes such as the revised Forest and Wildlife Policy of 2021. Third, the potential mediation of social capital, specifically bridging capital connecting CBOs to district forestry authorities, in translating consultation into governance outcomes warrants explicit modelling in future quantitative work.

For policy, the implication is not that current CBO activity should be curtailed; the monitoring and mobilisation work respondents describe is real and, on the available comparative evidence, valuable, but that it should not be mistaken for governance reform in itself. For research, the priority is twofold: qualitative work that can test directly the interpretation offered in Section 5.1, particularly whether respondents really do distinguish a communication standard from a governance standard when evaluating CBOs, and comparative work, ideally longitudinal, that can track whether districts where tenure and benefit-sharing reforms have actually been implemented show the same institutional gap this survey documents in Offinso, or whether the participation-without-power pattern weakens once collective-choice authority is genuinely transferred. Until that evidence exists, the safest reading of the present findings is a cautious one: Offinso's communities have told us, with considerable statistical consistency, that they see the value CBOs could realise and the institutional reasons that value remains unrealised. Whether policy responds to the second half of that judgement, and not only the first, is likely to determine whether the district's forest reserves are still losing ground the next time this question is asked.